Power resources and successful trade union actions that address precarity in adverse contexts: The case of Central and Eastern Europe

Is it possible for trade unions to fight precarity in an adverse global context? Although existing research suggests this is possible, there is limited understanding of the interplay of resources that enable unions to address precarity in deregulated markets. This study employs a power resource approach to investigate how unions overcome their external constraints. It draws upon 130 in-depth interviews with key informants across nine Central and Eastern European countries to investigate successful and unsuccessful union actions in sectors with differing external resources. In each sector, unions that mobilise their internal resources have been able to reduce various precarity dimensions, such as low wages, lack of voice, and irregular working time. The results reveal that unions whose objectives are based on convincing win–win discourses can make strides, acting as drivers of change in precarity patterns even in unfavourable conditions. Moreover, the study introduces a multi-dimensional conceptualisation of union success, identifying union actions that result in measurable improvements in precarity dimensions for all worker types. To deepen understanding of the role unions play in fighting precarity in adverse contexts, future research could investigate union actions that improve a wider range of precarity dimensions for all workers.


Introduction
In deregulated global markets, trade unions (hereafter, unions) struggle to counter the increased precarity faced by workers (Doellgast et al., 2018;Keune and Pedaci, 2020). Precarity refers to 'employment that is uncertain, unpredictable, and risky from the point of view of the worker' (Kalleberg, 2009: 2). Declines in union membership and collective bargaining coverage following the 2008 crisis (Visser, 2019) have made it more challenging for unions to further workers' interests. Meanwhile, there has been an increase in the proportion of workers with low (or insecure) incomes (Grimshaw et al., 2018), irregular (or uncertain) working hours (Wood, 2016), high job insecurity (Benassi and Dorigatti, 2015), and lack of union representation (Keune and Pedaci, 2020).
Despite widespread agreement in the literature that favourable employment relations (ER) institutions empower unions to fight precarity (Baccaro and Howell, 2011;Doellgast et al., 2018), the impact of unfavourable ER institutions remains ambiguous. In such conditions, comparative research suggests that because unions draw associational power from mutual solidarity, they adopt exclusive strategies that prioritise the interests of their traditional membership base (Benassi and Dorigatti, 2015). Since institutional deterioration is forcing unions to adopt strategies that include precarious workers (Marino et al., 2019), further empirical investigation is needed to understand the success or failure of their actions in contexts where institutional support is limited or absent.
This study employs a power resource approach (Korpi, 2006) to investigate how unions that devise inclusive strategies in adverse contexts can overcome external constraints and improve working conditions for precarious workers. To accomplish this, the study examines successful and unsuccessful union actions in sectors with stronger and weaker external (i.e. institutional and structural) resources. This study addresses two interrelated issues using qualitative data from 130 in-depth interviews across nine countries (Croatia, Czechia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia) in Central and Eastern Europe (CEE) that are characterised by unfavourable, albeit varying, ER institutions (Bohle and Greskovits, 2012). First, it examines whether union actions in sectors with relatively stronger external resources (healthcare and metal) are more likely to improve precarity dimensions than those in sectors with weaker resources (retail and construction). Second, it investigates the role played by internal union power resources (Lévesque and Murray, 2010) in successful union actions.
This study expands on Doellgast et al.'s (2018) theoretical model of precarious work by adopting a cross-sectoral comparative perspective to examine the roles played by external and internal union power resources in addressing precarity under unfavourable ER institutions. The results show that weak external resources do not prevent unions from successfully tackling precarity and, at the same time, they deepen our understanding of the role played by internal resources in empowering unions. Specifically, the study posits that proactive unions with clear objectives that are based on convincing win-win discourses can overcome their external constraints by acting as drivers of change in precarity patterns. Moreover, it introduces a multi-dimensional conceptualisation of union success, capturing the entire range of union actions that improve precarity dimensions for all types of workers, not just nonstandard workers. Accordingly, the study responds to calls for 'new middle-theories focusing on actor strategies' capable of capturing resilient unions' various forms of experimentation in a context of undeniable and prevalent institutional erosion (Murray et al., 2020: 151).
The next section reviews the extant literature on how unions use their power resources to address precarity and develops research questions. The research context and methods are described in the following sections. Finally, the study findings are presented and discussed.

Conceptualising union success in addressing precarity
Traditionally, precarity concerns the implications of using nonstandard contracts to ensure external labour market flexibility for the workforce (Pulignano et al., 2016). Most nonstandard contract workers experience high job insecurity, including uncertain income and working time (Doellgast et al., 2018;Keune and Pedaci, 2020;Wood, 2016). Accordingly, union actions seek to address precarity by improving the employment status of nonstandard workers relative to that of employees on standard contracts (Keune and Pedaci, 2020).
However, since the 2008 economic crisis, employment protections for standard workers have deteriorated in both the public (Grimshaw et al., 2018;Koukiadaki et al., 2016) and private sectors, such as metal in Germany (Benassi and Dorigatti, 2015), retail in the UK (Wood, 2016), and construction (Denmark, Germany, Italy, the Netherlands, Slovakia, and Spain) (Keune and Pedaci, 2020). Consequently, the employment conditions for standard and nonstandard workers have become increasingly similar, and precarity is beginning to concern both groups ( Benassi and Dorigatti, 2015;Bernaciak and Kahancová, 2017).
As workers can experience uncertain and risky employment conditions regardless of their employment status (Kalleberg, 2009), this study develops a multi-dimensional conceptualisation of union success to capture all union actions related to precarity. Union actions are viewed as successful if they induce a measurable improvement in at least one dimension of precarity for either nonstandard or standard contract workers. The relevant dimensions of precarity include (i) insecure or low income (Grimshaw et al., 2018;Pulignano et al., 2016), (ii) unpredictable and irregular working time owing to managercontrolled flexible scheduling (Wood, 2016), (iii) high job insecurity or short tenure (Keune and Pedaci, 2020;Wood, 2016), (iv) a lack of voice or restricted access to union representation (Kalleberg, 2009;Keune and Pedaci, 2020), and (v) social dumping. The last dimension refers to semi-legal or illegal employer practices such as 'cash-in-hand' payments or use of non-contract workers to reduce labour costs (Eaton et al., 2017). Social dumping is especially important when the state capacity to enforce employment regulations is weak or where such regulations are missing (e.g. CEE) (Bernaciak and Kahancová, 2017). This multi-dimensional conceptualisation enables an empirical investigation of the success or failure of unions actions that address precarity dimensions.

Factors that influence successful union actions
Comparative studies suggest that unions derive power primarily from context-related factors (Baccaro and Howell, 2011). Unions' weakened ability to preserve worker entitlements results from a recent series of intertwined structural and institutional developments (Doellgast et al., 2018;Müller et al., 2019). Structural factors such as slow economic growth and declining manufacturing employment have contributed to a downward trend in union power (Grimshaw et al., 2018). Simultaneously, institutional changes such as market deregulation have increased employers' prerogative to set employment terms and conditions. These changes have reduced individual employment protection by making it easier for employers to hire, fire, or employ workers using temporary contracts. Collective employment rights have also been reduced by restricting the right to strike or undermining multi-employer collective bargaining, particularly post-2008 Müller et al., 2019). While there continues to be cross-country variation in union density and collective bargaining coverage (Visser, 2019), the erosion of ER institutions has contributed to a decline in union power in most countries (Müller et al., 2019).
In these unfavourable conditions, comparative studies argue that unions will be increasingly unable to protect precarious workers' interests (Baccaro and Howell, 2011;Keune and Pedaci, 2020). Doellgast et al.'s (2018) theoretical model partly aligns with this view. They show that weak institutions combined with tenuous structural resources, low worker solidarity, and exclusive union strategies lead to a process of increasing precarity referred to as a 'vicious circle' (Doellgast et al., 2018: 25). They identify two interrelated change drivers that enable unions to reverse this pattern. Consistent with Keune and Pedaci (2020), they argue that development of more supportive ER institutions can allow unions to (re)gain power. In addition, Doellgast et al. (2018) indicate that unions can activate a series of strategies to reverse the precarity pattern by mobilising unorganised workers that share similar identities (e.g. occupation, gender, and ethnicity).
Specifically, Doellgast et al.'s (2018) model suggests that unions can escape the vicious circle by relying on the associational power they derive from resources developed through collective organisation (Silver, 2003;Wright, 2000). Doellgast et al. (2018) convincingly argue that this power stems from union density and/or localised support for precarious workers. Nevertheless, their study does not specify how unions can obtain associational power in contexts where supportive ER national institutions and workers' solidarity are limited or absent.
The literature on sociological actor-centred institutionalism offers further insight into the sources of union power in adverse contexts. Scharpf (2018) explores union experiences and contextualises their social interactions in complex cultural and ideological processes, arguing that unions (and other collective actors) can alter their roles, identities, and strategies. Consistent with this research, Dufour and Hege (2010) state that analyses that primarily link unions' capacity to defend precarious workers' interests to external conditions may overlook instances of union transformation. Since unions can derive strength from their voluntary organisation status, they can also find the resources needed to achieve their objectives elsewhere (Murray et al., 2020). Several studies show that when unions mobilise their internal power resources, including a wide repertoire of ideas Vlandas, 2016), leaders' identities (Bernaciak andKahancová, 2017), and alliances with community groups (Bernaciak and Kahancová, 2017;Eaton et al., 2017), they can fight precarity even in unfavourable conditions. This study examines the role of both external and internal power resources in national contexts where unions cannot rely on supportive ER institutions. In such contexts, the external resources unions can rely on have been claimed to vary primarily depending on sector (Keune and Pedaci, 2020;Bechter et al., 2012). However, existing studies have not systematically investigated the success (and failure) of union actions to protect precarious workers' interests across sectors with varying external resources. Thus, further empirical research is needed to deepen our understanding of the relationship between the strength of sectoral resources and union capacity to address precarity dimensions in adverse contexts. Moreover, when faced with dwindling external resources, unions may seek to develop a wider variety of internal resources than those observed to date (Bernaciak and Kahancová, 2017). Responding to Murray et al.'s (2020) call for research on union capacity to formulate new strategies in contexts of institutional erosion, this study devises an integrated analytical framework to empirically investigate successful and unsuccessful actions of unions that, despite their adversities, are committed to protecting precarious workers.

The role of external resources in addressing precarity
Unions can derive their power to combat precarity from both institutional and structural resources (Doellgast et al., 2018;Gumbrell-McCormick and Hyman, 2013). ER institutions are statutory and non-statutory support for union activities and workers' rights that result from labour laws and established social compromises and range from employer support for collective bargaining to tripartite forums for peak-level consultation (Dörre and Castel, 2009). High levels of ER institutions include strong traditions of voluntary collective bargaining and statutory support for employment rights (Gumbrell-McCormick and Hyman, 2013). Encompassing ER institutions increase the likelihood of union involvement in collective bargaining and reduce the risk of employers exploring exit options during negotiations (Doellgast et al., 2018). This direct participation in shaping labour policy has an equalising effect on working conditions for nonstandard workers relative to those on standard contracts (Baccaro and Howell, 2011). ER institutions promote coordinated bargaining and constrain employers' ability to shift work to precarious employment forms, giving unions a significant role in defining and enforcing labour standards. Thus, countries with such institutions have low precarity (Doellgast et al., 2018). A rare comparative study in CEE by Mrozowicki et al. (2013) investigates union roles in confronting retail sector precarity across three countries with different institutional regimes. They found that the stronger statutory protection for equal rights and extension of sectoral agreements in Slovenia enabled unions to improve minimum wage and job security more than their counterparts in Poland and Estonia. Hence, comparative research identifies a direct relationship between encompassing ER institutions and union power (Morgan and Pulignano, 2020).
With declining ER institutions, recent cross-sectoral comparative studies credit structural resources with a greater role in unions' ability to address precarity (Keune and Pedaci, 2020). Structural resources refer to the extent to which workers have a strategic position within the labour market and production process (Gumbrell-McCormick and Hyman, 2013;Wright, 2000). Unions have the greatest capacity to advance workers' interests when (i) there is high demand and high internationalisation of products and services, as these increase capital's vulnerability to workers' direct actions; and (ii) tight labour markets produce skill shortages, as these are key to disrupting production (Silver, 2003). Via a long-term worldwide historical analysis, Silver (2003) contends that automobile sector workers are better positioned than textile workers to disrupt production. Given the automobile sector's labour-intensive production processes, workers' skills and education provide the structural power needed to inflict high cost on employers. Studies find that the capacity of unions to obtain security-enhancing collective agreements is greater in high value-added sectors than in low value-added ones (Paolucci and Galetto, 2020). Hence, cross-sectoral analysis finds a direct relationship between structural resources and unions' capacity to reduce precarity.
The most favourable contexts for unions to derive power include strong institutional and structural resources (Keune and Pedaci, 2020). Pulignano and Doerflinger (2018) investigate union actions in two sectors in Belgium and Germany-metal and chemical-with comparably strong external resources. In Belgium, which has more supportive ER institutions, unions have more capacity via collective bargaining to formulate inclusive strategies that increase employment equality for all workers. Conversely, Keune and Pedaci (2020) investigate sectors with weak external resources (construction, industrial cleaning, and temporary agency work) across countries with different ER institutions (Denmark, Germany, Italy, the Netherlands, Slovakia, and Spain). While showing that unions operating in these sectors failed to prevent the expansion of precarity, they were still able to address some of its dimensions, such as voice and wage. However, there is no analysis of whether union capacity to address such dimensions is linked to the strength of external resources because the sectors investigated had similar external resources.
The erosion of ER institutions has not prevented unions from taking actions against precarity. While existing studies highlight the importance of sectoral external resources in empowering unions (Keune and Pedaci, 2020;Bechter et al., 2012), there is still limited understanding of the relationship between the strength of external resources and union capacity to address various dimensions of precarity across sectors, particularly in adverse contexts. Further empirical research is necessary to investigate whether the success and failure of union actions is contingent on sectors. This leads to the following research question: Research Question 1: Despite the erosion of ER institutions, are union actions in sectors with stronger external resources (institutional and structural) more likely to improve precarity dimensions than those in sectors with weaker resources?

The role internal resources play in enabling unions to address precarity
Research on successful country-specific union actions for fighting precarity reveals that unions can improve precarity dimensions even with weak external resources (Bernaciak and Kahancová, 2017;Eaton et al., 2017). A Uruguayan union organised domestic workers-the most unprotected workforce worldwide-to increase their job security (Goldsmith, 2017). Similarly, a union in Cambodia gained formal employment status for women hired informally to promote beer, providing them with legal protection against sexual harassment and assaults (Evans, 2017).
These studies echo Lévesque and Murray (2010), who suggest that unions can mobilise internal resources to counter precarity when their external resources are limited. First, union officials can employ proactivity to initiate and frame their agenda by developing clear objectives and acting on them vis-a-vis employers or the government. Studies on innovative union strategies attribute union leaders with a crucial role in successfully organising nonstandard workers (Bernaciak and Kahancová, 2017). They argue that leaders possess the experience and skill to disrupt path-dependent processes and induce change through novel actions that target representation of vulnerable groups (Gumbrell-McCormick and Hyman, 2013). While they show that proactivity is needed to develop strategies that tackle the voice dimension of precarity, proactivity's role in enabling unions to address other dimensions is underexplored.
Second, unions can utilise internal mechanisms for democratic worker representation via internal democracy. This includes the 'internal democratic traditions and procedures that enable the articulation of members' stances across different levels of unions structure' (Marino et al., 2019: 116). This implies a two-way relationship, where members offer the union legitimation, and leaders provide recognition for worker demands. Internal democracy is employed to build solidarity between members with otherwise potentially heterogeneous interests. In Italy and Poland, Marino et al. (2019) reveal that workers' representation without internal procedures hinders precarious workers from shaping the decisions that concern them. However, the study's focus on the voice dimension makes it unclear whether limited internal democracy prevents unions from taking successful actions to improve other dimensions such as low wages or irregular working time.
Third, unions can coordinate both horizontally and vertically with other unions, community groups, social movements, and between different segments of workers via external links (Lévesque and Murray, 2010). These links are key resources for developing solidarity across individuals and groups with different, even contradictory, identities (Morgan and Pulignano, 2020). External links may cover any ad-hoc union interactions with employers or political actors, through which unions seek to increase their power (Eaton et al., 2017). Bernaciak and Kahancová (2017) find that unions in CEE used the support of their Western European counterparts to organise and provide individual services to migrant workers. However, there is no systematic investigation of how unions develop and use external links to address multiple precarity dimensions.
Thus, although existing comparative research theorises that internal resources might positively impact the ability of unions to take actions against precarity where external resources are weak, no prior empirical study has systematically explored the role played by proactivity, internal democracy, and external links in accounting for the success and failure of union actions in adverse contexts. This gap leads to the following research question: Research Question 2: What is the role of internal power resources in the success or failure of union actions that address precarity dimensions in adverse contexts?

Research context: Countries and sectors
This study adopts a comparative perspective to investigate how the interplay between external and internal power resources allows unions to address multiple precarity dimensions in unfavourable national contexts. The post-communist legacies of CEE countries, combined with more severe post-2008 erosion of ER institutions in countries with stronger institutions (Glassner, 2013), make them ideal for this study. Most notably, they possess weak state capacity to enforce laws and regulations (Bohle and Greskovits, 2012). The limited trust in institutions, along with misalignment between laws and social norms, has induced about 25% of the labour force to operate in the informal economy (ILO, 2018). Moreover, the relatively low wages make it likely that both employers and workers will use creative forms of non-compliance with short-term mutual gains, such as cash-in-hand, in addition to minimum wages to increase worker income and reduce payroll taxes (Trif et al., 2016). Thus, the need for both employers and workers to sidestep the rules is widely accepted, which increases precarity, particularly its social dumping dimension. Nevertheless, degrees of variation exist in the ER institutions of CEE countries (Bohle and Greskovits, 2012). They are strongest in Slovenia, followed by Czechia, Slovakia, Hungary, and Poland (Bohle and Greskovits, 2012). Although Romania and Croatia have stronger legal protections for workers and greater bargaining coverage than Lithuania and Latvia, all four countries are neo-liberal (Bohle and Greskovits, 2012). Moreover, the legal changes adopted post-2008 have further deregulated labour markets across CEE countries with stronger institutions (Glassner, 2013;Koukiadaki et al., 2016). Slovenia and Romania, with the greatest bargaining coverage in 2008, suffered the largest decline (Visser, 2019). Post-2008 legislation, adopted in response to the global financial crisis, allows employers to increase or decrease weekly working hours for standard employees and has institutionalised seasonal or annual time accounts across CEE (Bernaciak and Kahancová, 2017). As a result, the region has seen an increase in employer control over key precarity dimensions.
Although the ER institutions available to unions vary across CEE, union capacity to use them is de facto limited (Ost, 2009). The latest data show that the average union density across the nine countries is 16% (ranging from 26% in Croatia to 8% in Lithuania); bargaining coverage averages 33% (ranging from 65% Slovenia to 13% in Latvia) (EIRO, 2019); and although national tripartite bodies exist, they have little (if any) power to reduce precarity across CEE, particularly post-2008 (Bernaciak andKahancová, 2017;Trif et al., 2016). These weak national ER institutions make unions' external resources largely contingent on their sectoral characteristics (Bechter et al., 2012).

Union external resources in healthcare, metal, retail, and construction
To investigate how the strength of external resources affects union success in addressing precarity, four sectors were selected where unions have stronger and weaker external resources. First, public healthcare workers have the most favourable structural and institutional conditions for disrupting services. Demand for healthcare services has remained relatively stable despite massive budgets cuts after 2008 (Glassner, 2013). However, the labour market for doctors and nurses has tightened. The emigration of medical professionals to Western Europe has led to significant skill shortages, following the relaxed employment restrictions after gaining European Union (EU) membership. Staff shortages have led to high job security, although in most of CEE, more than 10% of healthcare staff are self-employed (Eurostat, 2017). In addition to low healthcare professional wages caused by budgetary constraints, the medical staff shortage has contributed to excessively high workloads, including unpaid overtime (Trif et al., 2016). Moreover, the scarce and rarely up-to-date evidence on sectoral ER institutions suggests that healthcare unions have stronger ER institutions than those in other sectors. While union density varied from 51% in Slovakia to 16% in Poland (Weber and Nevala, 2011) in 2010, the public sector had 100% bargaining coverage in Lithuania, Romania, and Slovenia. Moreover, the role played by tri-or bipartite institutions in setting minimum wages was greater than that in other sectors (Clarke and Kerckhofs, 2018).
Second, unions in large automobile and steel firms have the strongest structural and institutional private-sector resources (Ost, 2009); however, the capacity to disrupt production varies significantly, as the metal sector is heterogeneous in terms of product features, value-added, internationalisation level, and firm size. The 2008 crisis has stimulated global demand for cheaper cars from CEE. In 2015, the automotive industry in Slovakia became the largest producer of motor vehicles per capita. In Hungary, automobile production expanded by 50% in 2014, and highly skilled production suffered major labour shortages (Eurostat, 2017). Further, large firms have high union density and bargaining coverage (Clarke and Kerckhofs, 2018). Meanwhile, firms at the bottom of international supply chains that require semi-skilled and unskilled workers were severely affected by the crisis; many went bankrupt, and numerous jobs were lost (Trif et al., 2016). Metal sector employers' quest for lower labour costs during the recession led to increased use of agency workers. Sub-contracted workers have lower job security, wages, and voice than those employed directly. The average union density in the metal sector is 9% and bargaining coverage is 33% (Clarke and Kerckhofs, 2018), indicating that metal unions have weaker resources than those in healthcare.
Third, turning to retail, although this comprises more than 10% of the labour in each country (Eurostat, 2017), disrupting production is challenging owing to the industry's structural conditions. Given the low skill requirements, workers are easily replaceable, and investment in employee retention can be minimal. Product demand contracted by up to 25% post-2008 owing to lower customer purchasing power, which, except in Poland, reduced employment (Eurostat, 2017). While small domestic retailers suffered during the recession, large foreign retail chains gained significant market share; for example, four foreign retail chains cover more than half the market in Lithuania, and 27% of employees in Slovakia work in foreign companies (Trif et al., 2016). Moreover, growing retail internationalisation has not increased workers' capacity to disrupt production, as there is limited integration across subsidiaries and products are sold on the domestic market (Mrozowicki et al., 2013). The average retail sector union density across CEE is 6%, and bargaining coverage is 21% (Adam, 2011), both lower than those in the metal sector. In this context, most retail workers experience high job insecurity, low wages, and irregular working time.
Fourth, construction workers have weaker power resources than retail workers. On average, a 46% drop in product demand occurred between 2008 and 2015, except in Poland where the need to build infrastructure for the European Football Championship in 2012 induced a 10% increase (Eurostat, 2017). This reduction led to rising unemployment and mass migration of skilled workers to Western European countries. Moreover, the seasonal nature of construction work, long outsourcing chains, and procurement laws that favour the lowest bidder make this sector prone to social dumping (Glassner, 2013). The extensive informal work in construction reveals the weakness of institutions. The cross-country average union density in construction is approximately 5% with bargaining coverage of approximately 41%; however, some agreements solely specify a commitment to implementing laws (Trif et al., 2016).

Research methodology
This study employs a cross-sectoral comparative approach using qualitative methodology. It examines successful and unsuccessful union actions across sectors with strong (healthcare and metal) and weak (retail and construction) external resources in countries with comparably weak ER institutions. The 2014-2016 data were collected for an EU project on union responses to precarious work in the four selected sectors across nine countries (Croatia, Czechia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia). This sectoral focus responds to calls for strengthening sector-level ER comparative research (Bechter et al., 2012).
An analytical framework was developed to systematically compare data across and within sectors. First, this study considers union actions at different observation levels and evaluates their success in improving five dimensions of precarity: low wage, irregular working time, lack of voice, job insecurity (Keune and Pedaci, 2020), and social dumping (Eaton et al., 2017). This multi-dimensional conceptualisation enables investigating union actions in adverse contexts where precarity affects workers on both standard and nonstandard contracts (Eaton et al., 2017;Grimshaw et al., 2018).
Second, the study draws on western democratic capitalism to operationalise sectoral ER institutions: union density, bargaining coverage, and unions' input into tripartite or bipartite forums (Dörre and Castel, 2009;Doellgast et al., 2018). Moreover, union capacity to disrupt production that is assumed to be directly related to the demand for labour, skill levels, and products characterises the structural resources (Silver, 2003). Finally, internal union resources were operationalised by evaluating union proactivity, internal democracy, and external links (Lévesque and Murray, 2010).
For an in-depth understanding of the widespread dimensions of precarity and the power resources available to unions, the first stage involved a team in each country that collected secondary sources, including legislation, collective agreements, union policy statements, and social media reports. Thus, a set of common semi-structured interview questions was developed.
In the second stage, the nine teams of national researchers gathered primary data about post-2008 union actions against precarity in the four selected sectors. A total of 130 native-language interviews were conducted (Table 1). First, interviews were conducted with union and employer confederation officials and public officials in specialised agencies that were responsible for designing or implementing policies related to precarious work. The officials identified unions that had addressed precarity in each of the four sectors. Second, based on their organisational function, key informants in each sector with 'expert opinions' (Saunders and Townsend, 2016: 837) were identified and contacted. These senior officials of the largest union federations and union representatives were directly involved in addressing precarity. Employer representatives were also interviewed to corroborate the unions' responses and ensure consistent interpretation. In sectors with no union (or employer) federations, union activists and senior managers of large firms (where unions are more likely to be present) were interviewed. At least one union and one employer representative were interviewed in each sector by country, except in Lithuania where no healthcare, metal, or retail employers agreed to be interviewed. However, three cross-sectoral Lithuanian respondents provided information about each selected sector. The variation in the number of interviews per sector and country reflects the fragmentation of unions and employers' associations.
The nine research teams conducted a first round of data analysis to determine which union actions improved specific dimensions of precarity. The lead author then re-coded all reported union actions and produced a list of codes that were shared for review and approval. Several meetings were held to revise and discuss any discrepancies to ensure a consistent approach (Cascio et al., 2019). A shared system was developed and consistently applied to the data for a third iteration of coding, which was double-checked by the research teams for emergent issues (Bernard and Ryan, 2009). This iterative approach improved the reliability of the analysis and facilitated a participatory process (Cascio et al., 2019).

Retail (22)
Unions, 1 1 1 1 1 3 1 1 1 11 employers 2 1 1 2 0 1 2 1 1 11 Construction (24) Successfully addressing precarity may not be directly linked to the number of successful union actions (e.g. two union actions that increase low wages in two separate companies do not necessarily improve the wage dimension of precarity more than one union action that raises minimum wages within a sector). Therefore, all union actions that targeted a specific dimension in a particular sector and country were categorised as 'successful' or 'unsuccessful'. The successful (unsuccessful) code was used when at least one union action succeeded (failed) in measurably improving a dimension of precarity in a sector and country. Accordingly, a successful or unsuccessful code may cover one or multiple union actions seeking to address a specific dimension of precarity in a particular sector and country. For example, union actions that increased low wages by consulting with the government on minimum wage or collective bargaining in one or multiple companies in a sector and country were coded as successful. Nevertheless, trade-offs considered beneficial for workers (from the evidence provided by respondents), such as increased job security in exchange for irregular working time, were recorded once for the dimension that was improved (i.e. successful for job security). This systematic binary coding of all reported union actions per dimension of precarity resulted in comparable data across the four sectors and nine countries investigated.

The role of external resources in enabling unions to address precarity
This section answers Research Question 1. Table 2 presents an overview of the successful and unsuccessful union actions that addressed each precarity dimension in sectors with stronger (healthcare and metal) and weaker (retail and construction) external resources across countries. In each sector, unions improved wage in most countries, followed by voice, working time, job security, and social dumping.
Similar developments for wage exist in sectors with stronger and weaker external resources (Table 2). In the two sectors with stronger resources, unions took successful actions in eight countries (except healthcare in Slovenia and metal in Lithuania). However, unions in these sectors were also unsuccessful (i.e. the healthcare sector in Romania and Slovakia and the metal sector in Croatia and Slovenia). Similarly, in the two sectors with weaker external resources, union actions improved wages in retail and construction in eight and seven countries, respectively. In retail, unions were unable to improve low wages in Lithuania, taking an unsuccessful action that sought to stop employers from paying workers less than the minimum wage when output targets are not met. In construction, only Croatian unions took unsuccessful actions (i.e. strikes and annual campaigns against failure to pay wages and benefits and payment delays). In an interview, a construction union federation official reported that they made a list of about 9000 companies that regularly fail to pay workers on time.
Independent of the sector, similar union actions successfully addressed low wages. For example, multi-employer agreements were reached in sectors with both strong and weak external resources. Wages were improved by healthcare unions in Croatia, Latvia, and Romania and by metal unions in Romania, Slovakia, and Slovenia. Construction unions in Croatia, Czechia, and Poland sought to prevent a race to the bottom on labour costs and achieved similar results despite relying on weaker external resources. Low retail wages increased via single-employer bargaining in the largest retail chains; unions often exploited employers' fear of negative publicity, which could deter consumers from buying a company's products. Further, unions utilised government consultations to contribute to raising the national minimum wage and introduced a minimum wage for the most vulnerable groups of workers (e.g. the self-employed in Poland and student workers in Slovenia).
Sectors with weaker and stronger external resources experienced similar developments in the voice dimension. For instance, organising low paid workers was the most common successful action in large multinational companies in the metal and retail sectors. Retail union federations unionised workers in multinational retail chains in Poland, Romania, and Slovenia, while metal unions unionised agency workers in multinational automotive companies, such as Volkswagen in Slovakia and Poland and GM-Opel in Poland. The role of multi-employer bargaining deteriorated in the metal and retail sectors in Romania, Slovakia, and Slovenia, while metal unions established new bipartite forums to address precarity in Croatia and Hungary.
Healthcare and construction unions also contributed to institutionalising sectoral forums for social dialogue, despite having diverse external resources. In Lithuania, union actions in 2013 resulted in establishing a tripartite sectoral council to shape healthcare reforms. In addition, they persuaded employers and state representatives to establish sectoral collective bargaining in healthcare. Similarly, union actions in the construction sector in Croatia and Latvia contributed to establishing new bipartite forums to address issues concerning payment delays and illegal work. A prevalent union action was providing individual services to vulnerable groups, such as migrants and young workers. Overall, union success in addressing the voice dimension is not contingent on the strength of external resources. Retail unions took successful actions related to working time in more countries than healthcare and metal unions, which have stronger external resources (Table 2). In retail, the most common action was restricting shop opening hours on Sundays and bank holidays via legal changes (e.g. Czechia, Hungary, and Slovakia) or collective bargaining (e.g. Latvia, Romania, and Poland). In healthcare, despite overtime work issues, successful actions were reported in three countries. In Latvia, unions negotiated up to eight additional annual leave days and reduced weekly working hours from 40 to 38.5. Unions in Romania increased annual leave and made it easier for local unions to verify enforcement of mandatory working time regulations. Despite the 2011 mandatory employment freeze in Slovenia, unions contributed to gaining additional public funds to employ new staff (about 1700 nurses) via lobbying. In 2013, the threat of industrial action to reduce workloads in healthcare led to negotiations with the Slovenian government for new norms, which were unsuccessful. According to a healthcare union representative, more than 200,000 hours of nursing overtime work in Slovenia have remained unpaid since 2008. Healthcare unions were also unsuccessful in reducing overtime practices in Croatia, Czechia, and Slovakia.
Similarly, metal union actions had limited success improving working time provisions. In Croatia and Lithuania, unions negotiated collective agreements in a few firms to provide additional time off and shorter workweeks. However, in Romania, irregular working hours were introduced during the negotiation of the automotive sectoral collective agreement in 2010, although unions had significant ability to disrupt production owing to increasing demand for inexpensive cars post-2008. Across the nine countries, no reported union actions addressed irregular working time in construction. While weak external resources may explain their lack of action, irregular working time may be tolerated in construction, as unions focus on improving the other four dimensions of precarity (Table 2).
Job security saw no reported successful or unsuccessful union actions for retail workers. It seems unlikely this is owing to low external resources, since retail unions had successful actions that improved voice and working time in more countries than healthcare unions. Unions may have addressed other precarity dimensions in retail's context of very high voluntary labour turnover. In other sectors, external resources have no apparent role in strengthening job security. Metal and construction unions had successful actions in approximately half the countries investigated, while healthcare unions took hardly any action ( Table 2). The most common successful actions in metal and construction were trade-offs between improved job security and either increased employer control over working time or wage cuts. Both construction and metal unions in Croatia, for instance, accepted reduced wages in exchange for preserving jobs. In Poland, unions have successfully pushed employers to hire experienced agency workers on standard contracts in GM-Opel, Volkswagen, and ArcelorMittal. No successful actions were reported in healthcare, except in Croatia, where unions blocked the government plan to outsource non-core services.
Finally, regarding social dumping, unions had successful actions in two sectors with contrasting external resources. Table 2 shows that metal unions had successful (and unsuccessful) actions in several countries. The Croatian metal union federation sought to negotiate an agreement to cover minimum standards for the entire sector. Despite their failure, unions negotiated an explicit provision to cover agency workers in most company-level agreements. In Czechia and Slovakia, metal unions also lobbied the government to eliminate illegal practices, such as paying daily subsidies instead of wages, and obtained legal amendments concerning paying agency workers. Retail unions had successful and unsuccessful actions in several countries. In Croatia and Slovenia, union confederations organised successful campaigns against government proposals that sought to liberalise mini-jobs (i.e. temporary part-time jobs with pay below the minimum wage) in retail; union actions resulted in a referendum that rejected this bill. In Slovakia, retail unions improved monitoring and reporting of illegal practices. However, Croatian unions and employers' confederations have been involved in a 'Stop the work in the informal economy' campaign in approximately 20 towns. Although they organised local events to expose the adverse effects for workers, companies, and the community and provided free phones for reporting work in the informal economy, they were unsuccessful.
Social dumping is rarely an issue in healthcare; consequently, the only actions to address it were in Romania, where unions were able to reduce the share of doctors claiming state subsidies for 'fictional' patients. However, social dumping is arguably the most important dimension of precarity in construction owing to widespread illegal practices. Construction unions took actions to reduce illegal labour practices in most countries (except Czechia) and were successful in Poland and Hungary. The most common unsuccessful action was lobbying governments to modify the provision in public procurement laws that favours the lowest bidder. Nonetheless, the high demand for public works to build the infrastructure for the 2012 European Football Championship in Poland enabled unions to introduce non-price criteria for public tenders (e.g. social clauses concerning employment standards), indicating that stronger external (structural) resources can make a difference. In Hungary, a union respondent indicated that the sectoral collective agreement is 'guaranteeing that safe working conditions cannot be subject to the cost-cutting race to the bottom' (Interview, construction union federation), illustrating the limited union success in preventing social dumping.
Successful union actions in adverse contexts do not vary depending on the strength of external resources. First, union success in addressing the dimensions of wage and voice across sectors with stronger and weaker external resources show little variation. The precarity dimension that unions most consistently address across sectors is wage. Second, the cross-sectoral variation in union success in addressing the dimensions of working time, job security, and social dumping is not contingent upon the strength of external resources. Critically, unions with weak resources can improve precarity dimensions. In adverse contexts, the success of union actions varies depending on the precarity dimension investigated in each sector. For instance, while unions in retail took no actions to reduce job insecurity, they successfully addressed working time in most countries. Similarly, healthcare unions successfully dealt with working time (wages) in a few (most) countries.

The role of internal resources in enabling unions to address precarity
To answer Research Question 2, this section compares successful and unsuccessful union actions in each sector along the precarity dimension addressed in most countries. Holding the external context constant allows an examination of the role of internal resources.
In healthcare, low wage is addressed by unions in most countries. When there were shortages of healthcare professionals, which were exacerbated by job opportunities in Western Europe, unions extensively used public protests and resignation campaigns (e.g. in Czechia, Hungary, and Slovakia) to demand more wages. Unions recognised that they could gain the support of healthcare services beneficiaries via these actions.
Developments in Slovakia illustrate the success and failure of resignation campaigns in addressing low wages. The doctors' union succeeded in gaining a legislatively stipulated wage increase, while a similar action by the nurses' union failed. The doctors' union was better able to formulate clear objectives that resonated with both members and the wider community. It persuaded its members to act together (i.e. mass resignation) to prompt the government to meet their demands. Moreover, it convinced the public to support the doctors and avoid a health system collapse. Despite its similar potential, the nurses' union could not convince a critical mass to resign. Nurses did not act together owing to the lack of well-defined common objectives. For instance, it was perceived that the union demand to introduce pay scales based on seniority could deepen the inequalities between older and younger nurses. The inability to establish strong connections among members across hospitals led to low campaign participation. More than half the nurses who resigned worked in two hospitals: 382 in Prešov and 212 in Žilina. Further, doctors provided facts to demonstrate that their working conditions had a detrimental effect on the quality of healthcare services provided. Contrarily, the nurses' union appealed to emotion, highlighting that nurses were burned out and demotivated, as indicated by the following quote: 'The population is ageing . . . and we are also ageing. Nobody is prohibiting us [from working] past retirement age, but nurses are leaving . . . they are physically and emotionally exhausted' (Interview, nurses' union, Slovakia). However, the public perceived that their demands would hinder hospital operations.
In the metal sector, unions in most countries address voice. In a context of increasing the share of agency workers at metal companies, unions focused on representing the interests of agency workers, who are generally not unionised. When inferior working conditions (e.g. up to 50% lower wages in Poland) began undermining their core members' interests, unions began addressing these interests. Initially, unions relied on statutory provisions and collective bargaining to equalise pay between agency workers and direct employees. As employers found creative ways to circumvent this in most countries, unions realised it was important to unionise agency workers to secure enforcement of their rights. In Slovenia, company-level unions are crucial in implementing rules to protect agency workers. Despite statutory rights for equal pay for agency workers and direct employees that perform similar functions, this provision was implemented in few companies. As a successful example, the union in an electric equipment firm organised all agency workers, which comprised approximately 20% of its labour force. The local union leader's commitment to prioritising the interests of the agency workers made a difference, as indicated by a respondent: The trade union leader in that company has been successful in recruiting agency workers. She pays a visit to them, introduces herself, explains who she is and asks them to join the union. And she takes care of them the same [way] she cares for her workers. (Interview, metal union federation, Slovenia) Proactivity was key in guaranteeing that agency workers had the same de facto entitlements as their counterparts on standard contracts. Apart from ensuring equal pay (e.g. base pay, overtime, and annual leave payments), it helped obtain standard contracts for experienced agency workers.
Nevertheless, company-level unions are often reluctant to prioritise agency worker interests. As agency workers exert downward pressure on labour standards, union members feel threatened by them. Thus, most unions do not proactively reconcile the seemingly divergent demands of agency workers and employees. A respondent stated that: [W]e are so passive in Slovenia . . . Everyone is indifferent until the things go to the extreme. I think that's precisely what will happen. The current system will collapse, and every worker will be a precarious worker. (Interview, metal union federation official, Slovenia) While the union federation does acknowledge that neglecting agency workers will increase precarity in the sector, it lacks internal procedures to oblige company-level unions to organise agency workers. Only proactive company-level unions that organise agency workers can ensure enforcement of their statutory provisions.
In retail, unions in most countries also address voice, but for different reasons. First, the legal changes that allow employers to use nonstandard contracts and irregular working hours induced a massive increase in all dimensions of precarity. Second, severe undermining of collective employment rights risked the survival of retail unions (e.g. in Hungary and Romania). A union official indicated that: 'we had to increase the number of members to support our activities and ensure our survival' (Interview, retail union federation, Romania). In this context, proactive unions intensified their efforts to organise workers in Croatia, Lithuania, Poland, Romania, and Slovenia.
In Romania, the same retail union federation employed unsuccessful and successful actions to unionise workers. Following a leadership change in 2001, it launched a campaign to organise workers in domestic companies. In half the counties, union activists visited retail shops to unionise workers so they could reach the legal threshold to be eligible for sectoral collective bargaining. They approached workers during their lunches, cigarette breaks, and closing periods to convince them to join. A union official stated: 'We worked very hard for the three years, sleeping in poor conditions, in a van, or hostelswherever we could find very cheap accommodation-and we ate canned food' (Interview, retail federation, Romania). Despite their strong commitment, the first attempt was unsuccessful, as their objective of establishing a critical mass proved too ambitious; the union was unable to maintain regular connections with the newly recruited members and activists. This experience was key in preparing and successfully implementing the second phase. The union narrowed its objectives by focusing their efforts on large multinational retail chains and strengthened its external links with international unions. It sought support from the German service sector union (Ver.di) via training and access to relevant senior managers in German companies that operated in Romania, which in turn helped unionise workers in retail chains such as Metro, Real, and Selgros. The union also benefited from UNI Global (with which it is affiliated) in its organisation of workers in the Carrefour chain. Thus, union density increased to over 50% in four retail chains by 2011, reaching the legal threshold for company-level bargaining. Hence, unions that learn from their failures and reformulate objectives as new challenges or opportunities emerge can develop useful links with international unions and successfully unionise workers, even when external resources are weak.
In construction, wage, voice, and social dumping were addressed by unions in most countries. Despite targeting different dimensions, union actions commonly aimed to reduce illegal practices by strengthening and enforcing the regulatory framework. Unions provide individual services to all workers in Latvia and Romania to ensure that they receive their due payments on time. This assistance is needed because it is common for workers at the bottom of the outsourcing chain to experience delays in or no payment. In contrast to retail, unions strengthen the voice dimension by establishing sectoral bipartite bodies and offering individual services to all workers seeking to eliminate illegal practices (e.g. Czechia, Latvia, Hungary, Romania, Poland, and Slovakia). Moreover, to reduce the race to the bottom in labour standards, unions lobbied the governments to modify the public procurement law provision that favoured the lowest bidder. Nevertheless, they struggled to alleviate illegal practices.
Croatia illustrates the success and failure of union actions in pushing employers to address the issue of unpaid wages. By defining failure to pay wages to workers as a criminal activity, the union federation contributed to introducing severe sanctions for employers to the penal code. The union successfully persuaded the parties involved in the national social dialogue of the benefits to employers (e.g. reducing unfair competition) and governments (e.g. collecting additional payroll taxes), given a clear and narrow objective (i.e. targeting unscrupulous employers). The way it framed this action prioritised it as a national agenda. The union was, however, unsuccessful in achieving a more ambitious objective. Despite their strong proactivity and external links, they could not convince the government that a legal change was required to guarantee workers' payments when their companies go bankrupt. This example suggests that union actions are successful when they identify objectives compatible with stakeholder interests. Overall, the dynamic relationship between internal and external resources affects unions' capacity to take successful actions in sectors with stronger and weaker external resources, enabling them to mobilise their members and community groups to address the dimensions of precarity.

Discussion and conclusion
The worldwide expansion of precarity and deregulation of labour markets have highlighted the importance of understanding the factors that enable unions to successfully address precarity in adverse contexts (Doellgast et al., 2018;Eaton et al., 2017;Grimshaw et al., 2018). This study provides a qualitative investigation of the role of external and internal power resources in the success and failure of union actions that address low wages, irregular working time, job insecurity, lack of workers' voice, and social dumping in sectors with stronger and weaker resources across nine CEE countries. It primarily deepens Doellgast et al.'s (2018) theoretical model (Table 3) and specifies how unions can overcome external constraints by mobilising their internal resources (Dufour and Hege, 2010;Murray et al., 2020). How proactive unions frame objectives is crucial for creating allies among workers, employers, and the government.
Regarding Research Question 1, the findings show that unions that operate in adverse contexts cannot draw sufficient power from external resources (Doellgast et al., 2018). However, they can still address the dimensions of precarity. By echoing Korpi (2006), this study reveals that unions can articulate and represent the interests of precarious workers even when they face unfavourable conditions, as illustrated by the comparable union success in improving wage and voice across sectors in CEE. Moreover, the crosssectoral variation in union successes addressing working time, job security, and social dumping is not contingent on the strength of external resources, as unions in sectors with weaker external resources can be successful.
Consistent with Doellgast et al.'s (2018) model, this study confirms that weak external resources do not prevent successful union actions, since unions committed to addressing precarity can use their associational power to drive change. However, Doellgast et al.'s (2018) model does not explain how unions can gain this form of power in adverse contexts such as lack of supportive ER national institutions and low solidarity among unorganised workers. This study identifies new combinations of internal resources, including proactivity, external links, and internal democracy, that enable unions to gain associational power and overcome their constraints (Table 3). Accordingly, it argues that theoretical models should include a wider spectrum of internal resources to capture the various forms of experimentation by unions when institutions are eroding.
Regarding Research Question 2, the study demonstrates that unions that can formulate clear and narrow objectives based on convincing win-win discourses may successfully improve multiple precarity dimensions even in adverse contexts. While confirming that proactivity is essential for unions (Bernaciak and Kahancová, 2017;Dufour and Hege, 2010), this study shows that it is not sufficient for success. A discourse that highlights the potential benefits for all parties is necessary to strengthen internal democracy and external links, thereby gaining the support of stakeholders (e.g. members, precarious workers, wider public, social movements, and other unions); this is illustrated by the construction union in Croatia, doctors' union in Slovakia, and retail union in Romania. This study confirms Marino et al.'s (2019) finding, where a robust internal democracy, such as involving precarious workers in formulating objectives, facilitates union success. Meanwhile, formulating a win-win argument and effectively communicating it to members is necessary to persuade them to support actions that prioritise the interests of precarious workers.
Crucially, this study bridges sociological, actor-centred, and comparative ER research gaps by showing that unions can overcome their external constraints by adding new resources to their repertoire of potential actions. This interplay of resources enables unions to experiment (Murray et al., 2020) with their (adverse) context via past failures, test new strategies, and (re)frame actions through a convincing discourse. Thus, they can progressively (re)gain the support of different worker groups. This study extends Doellgast et al.'s (2018) theoretical model ( Table 3) by demonstrating that not only shifts in external resources (primarily in ER institutions) or individual workers' initiatives (workers' solidarity) but also unions themselves can be drivers for change in precarity patterns. By facilitating actions that reconcile the interests of different categories of workers with those of employers and governments, union intermediation is critical for improving the relevant dimensions of precarity. It also contributes to (re)building solidarity among various societal actors.
Furthermore, this study shows that unions are willing to address precarity based on pragmatic and moral considerations. First, they seek to address the interests of precarious workers when their working conditions undermine those of union members (e.g. metal unions addressing the interests of agency workers), thus confirming that when precarity threatens all workers, unions cannot ignore it (Benassi and Dorigatti, 2015). Second, unions are willing to address the interests of precarious workers when such actions are essential to preserving their social role (e.g. the Romanian retail union). Unions understand that expanding precarity threatens their existence, and their loss of regulatory power can be only at least partially compensated by (re)gaining internal legitimacy (Dufour and Hege, 2010). However limited, their success has an immediate impact on (precarious) workers, facilitating reconstruction of their collective identity. Thus, the increasing number of unions struggling to retain their diminishing influence (and membership) can-and should-(still) fight against precarity by altering their strategies, reshaping their roles, and strengthening their internal resources for their benefit and that of society. This practical implication extends Dufour and Hege (2010) by demonstrating that unions can act upon their own transformation in their quest to improve working conditions for precarious workers.
Finally, this study posits that a multi-dimensional conceptualisation of precarity enables a more nuanced understanding of the factors that explain union success in addressing precarity. While previous studies indicate that unions take actions primarily when external resources are strong (Baccaro and Howell, 2011;Doellgast et al., 2018), this study reveals that unions with weak external resources enhance their chances of success by tackling dimensions that can feasibly be improved within their context. For instance, given high voluntary labour turnover in retail, retail unions did not seek to strengthen job security in any country. Interestingly, job security (the main dimension of precarity in the extant literature) was among the least tackled dimensions in each sector in CEE countries. Nevertheless, union actions have improved low wages and worker voices in each sector in most countries. Accordingly, studies that focus solely on job security or equalising employment conditions for standard and nonstandard workers (e.g. Doellgast et al., 2018;Keune and Pedaci, 2020) may disregard the majority of union actions that induce benefits for precarious workers in adverse contexts. Further, consistent with Benassi and Dorigatti (2015), this study shows that reducing the gap between nonstandard and standard workers can stem from a failure to prevent deterioration in standard workers' working conditions. Thus, future research must focus on any union actions that improve at least one dimension of precarity to understand their role in addressing precarity in an adverse global context (Table 3). Longitudinal comparative research could further clarify the sustainability or degree of success of union actions that improve specific dimensions of precarity. Moreover, considering that this study solely accounts for proactivity, internal democracy, and external links, future research can attempt to understand the role (and interplay) of the wider spectrum of internal resources identified in the literature about enhancing union power (Lévesque and Murray, 2010).