Theoretical framework and hypotheses

In recent decades, the general trend of deregulation, which favours companies acting unilaterally, has limited the scope of action of the social partners and their representativeness in the labour market (Baccaro and Howell Reference Baccaro and Howell2017; Streeck Reference Streeck2010). The literature has particularly highlighted the erosion of union power (Vandaele Reference Vandaele2019).

On the other hand, theory has indicated the difficulties of collective action among employers and, consequently, questioned the ability of EAs to achieve high membership rates. The nature of companies and the capitalist system means that the decisive source of power remains with the individual firm and its strategic choices or in informal agreements between a few companies. Only strong union power or public policies detrimental to employers’ interests provide a strong incentive for high EA membership (Offe and Wiesenthal Reference Offe and Wiesenthal1980). Furthermore, the broad heterogeneity of companies and interests, and the natural competition between them, hardly favour the creation of large associations (Olson Reference Olson1965). Schmitter and Streeck (Reference Schmitter and Streeck1999) comment on the complex interaction of two logics of collective action that are, to a large extent, opposed and contradictory. There is the logic of membership, where maximising affiliation necessitates integrating a plurality of interests; and the logic of influence (with regard to the state and unions), which requires the imposition of a single stance, with the predominance of few interests over a more extensive plurality. Nevertheless, the elements that favour affiliation in terms of representation and resources available for companies have predominated, which has meant that EA membership rates have been higher than union density (Traxler Reference Traxler, Sadowski and Jacobi1991, 45).

Historically, EAs have had two essential functions from which they obtain their membership, resources, and power: collective goods related to the state and the industrial relations system, and the provision of services as selective goods. The provision of collective goods is essential as it involves the representation of companies. In fact, EAs emerged in response to the challenges posed by unions and pro-employee legislation. Representation is the original source of the organisational purpose, identity, policies, and professionalised structure of these associations (Gladstone Reference Gladstone, Windmuller and Gladstone1984, 25–27). Collective goods related to the state involve activities such as lobbying, public relations (intervention in public opinion), and participation in labour courts and the state arbitration system. In Western Europe, representation of employers has also taken the form of corporatist representation and participation in tripartite social pacts, which gained momentum in some countries in the 1990s (Mesch et al Reference Mesch, Pasztory, Zotter, Huemer, Mesch and Traxler1999, 186–187).

Collective goods related to collective bargaining involve representation in multi-employer negotiations and their coordination (Sheldon et al Reference Sheldon, Della Torre and Nacamulli2019). Collective goods are universal, enjoyed by EA members and non-members, which generates the free-rider phenomenon. To encourage new members and to retain companies that are already members, EAs provide selective goods in the form of services that are only available to members and are provided free or at a considerable discount (Olson Reference Olson1965). These selective goods include economic, legal, and industrial relations information, training, and networking (Sheldon and Thornthwaite Reference Sheldon, Thornthwaite, Gooberman and Hauptmeier2022, 142).

The processes of decentralisation and deregulation of the labour market initiated in the 1980s have had a notable impact on this centralised and sectoral representation of companies. The increased power of companies and the individualisation of industrial relations have undermined the usefulness of collective goods provided by EAs and even reduced a sense of collective identity among employers (Waddington Reference Waddington, Bamber, Lansbury and Wright2016), thereby discouraging EA membership.

Subsequently, a third function was added and EAs incorporated a portfolio of individual services as elective goods, which represents a strategic departure from Olson (Reference Olson1965). The aim was to improve EA finances, resources, and membership in order to face the challenges posed by deregulation and decentralisation of collective bargaining (Sheldon and Thornthwaite Reference Sheldon and Thornthwaite2004). These elective goods consist of expert services at commercial prices for non-members and with discounts for members. Examples are industrial relations and human resource management consultancy, firm-based training, firm-level strategy development, and customised legal advice and representation. As a component of these elective goods, EAs may also commercialise the traditional representation of employers’ labour market interests as an exclusive and on-demand service for companies, for example, assistance in company-level collective bargaining. EAs also started to offer labour law advisory services as elective goods (for non-members) and/or selective goods (for members) (Sheldon et al Reference Sheldon, Della Torre and Nacamulli2019, 19). The commercialisation of elective goods opened the door to providing a wider range of services, such as assistance with internationalisation, dealing with financial institutions, and applications for EU subsidies in the case of European countries (Sheldon et al Reference Sheldon, Nacamulli, Paleotti and Morgan2016, 180–181). This led to EAs becoming mixed organisations (Brandl and Lehr Reference Brandl and Lehr2019, 933). As a matter of fact, according to some studies, this third function of private services (elective goods) to companies helps explain the strong membership resilience of these associations, making up for their reduced capacity for coordinating multi-employer bargaining (Brandl and Lehr Reference Brandl and Lehr2019; Gooberman and Hauptmeier Reference Gooberman and Hauptmeier2022). On the contrary, related to the aforementioned deregulation processes, some national studies have noted falling EA membership in certain countries (Fanfani et al Reference Fanfani, Lucifora and Vigani2023; Gooberman et al Reference Gooberman, Hauptmeier and Heery2019b; Sánchez-Mosquera Reference Sánchez-Mosquera2022; Silvia and Schroeder Reference Silvia and Schroeder2007).

The literature has also analysed the characteristics of member companies. This makes it possible to know the dominant types of companies in the make-up of EA membership and what determines company affiliation, which, moreover, improves understanding of the evolution of EA representation. The literature that documents the diversity of company responses to the challenges posed by economic and institutional changes is not novel or groundbreaking. Logically, companies are interested in minimising or even eliminating union presence in the workplace, and in a way EAs have accompanied this aspiration. However, the literature has also revealed the importance of company-level determinants related to union membership (which is falling), something that has historically driven the ‘compensatory’ affiliation of companies to EAs (Offe and Wiesenthal Reference Offe and Wiesenthal1980; Gooberman et al Reference Gooberman, Hauptmeier and Heery2019b; Sisson Reference Sisson1987) and, above all, those related to multi-employer bargaining (Demougin et al Reference Demougin, Gooberman, Hauptmeier and Heery2019, Gooberman et al Reference Gooberman, Hauptmeier and Heery2019b, 13–14; Sheldon et al Reference Sheldon, Nacamulli, Paleotti and Morgan2016; Traxler Reference Traxler2004, 57). In this respect, the concept of ‘countervailing power’ applied to the need for employers to seek coordination to counteract union power is well established in the literature and has even been extended to pressure exerted on the state (Barry and Wilkinson Reference Barry and Wilkinson2011). However, conversely, employers have historically influenced and still influence the collective bargaining system and structure. The business profile of their organisations and the general attitude of employers towards unions and workers are decisive and, in turn, condition the behaviour of the unions (Clegg Reference Clegg1976, 99–117).

Multi-employer bargaining schemes protect individual enterprises (Demougin et al Reference Demougin, Gooberman, Hauptmeier and Heery2019; Gooberman et al Reference Gooberman, Hauptmeier and Heery2019a, Reference Gooberman, Hauptmeier and Heery2019b, 13–14; Traxler Reference Traxler2004, 57) and help keep up EA membership. Indeed, in certain institutional environments, multi-employer bargaining has been used to render wage competition between companies ineffective and control the labour factor through shared norms that neutralise conflict and remove it from the workplace (Sisson Reference Sisson1987, 23–26; Swenson Reference Swenson2002; Tolliday and Zeitlin Reference Tolliday, Zeitlin, Tolliday and Zeitlin1991). As has been theoretically expressed and empirically demonstrated in the literature, this is especially true for sectoral bargaining (Traxler Reference Traxler2004, 49–50, Reference Traxler2010). In this way, it has been emphasised that it is collective bargaining and all its coordination mechanisms, more than union power, that help sustain EAs (Gooberman et al Reference Gooberman, Hauptmeier and Heery2019a; Sheldon et al Reference Sheldon, Nacamulli, Paleotti and Morgan2016). The breakdown of coordination mechanisms and the enforced retreat of collective bargaining invariably resulted in a membership crisis and a reduction in the power of EAs, as occurred in several anglophone countries (Sheldon et al Reference Sheldon, Della Torre and Nacamulli2019; Traxler Reference Traxler2004, 57; Waddington Reference Waddington, Bamber, Lansbury and Wright2016, 44).

Continuing with the company-level determinants of EA membership, company size is also important, as small companies are less likely to join (Gooberman et al Reference Gooberman, Hauptmeier and Heery2019b, 14), as well as the fact that large companies are vital for EAs, insofar as they increase their representation of the workforce significantly (Traxler Reference Traxler2004, 56). Traxler indicated that large companies have a greater propensity to join EAs than small companies because they are more likely to have active unions in the workplace and they enjoy a more effective provision of collective goods. Small companies, meanwhile, may be more interested in selective goods since, although they do not usually have to deal with unions in the workplace, they are affected by regulations arising from labour law and, where applicable, by multi-employer (sectoral and territorial) agreements (Sheldon and Thornthwaite Reference Sheldon and Thornthwaite2004, 133).

On the other hand, Olson (Reference Olson1965) affirms that only large companies have the leadership capacity and the power to agree on collective action that is coherent for EAs, and which is subsequently supported by SMEs. In this respect, the decentralisation of collective bargaining has eroded the interest of large companies in EA membership, as they may negotiate individually, as illustrated by the case of Fiat and Confindustria in Italy. This is notably detrimental to EAs, as with a reduced affiliation with large companies they lose a great deal of power, prestige and financial resources (Sheldon et al Reference Sheldon, Nacamulli, Paleotti and Morgan2016, Reference Sheldon, Della Torre and Nacamulli2019, 19).

The literature has indicated that, in the countries studied, EAs supported the traditional institutions of the industrial relations system, above all collective bargaining. One of the explanations for this behaviour is the inability of micro and small companies, prevalent in the Southern European business fabric, to bear the costs arising from company-level collective bargaining, and also the organisational legitimisation that multi-employer bargaining offers EAs (Bulfone and Afonso Reference Bulfone and Afonso2020). On the contrary, the literature on EA membership, as indicated above, and the cases of Spain and Italy, have highlighted the organisational influence and power of large employers and sectors with a high level of EA membership, and with a long tradition of sectoral (multi-employer) bargaining and union representation in their companies (Fanfani et al Reference Fanfani, Lucifora and Vigani2023; Sánchez-Mosquera Reference Sánchez-Mosquera2022).

Traxler (Reference Traxler2000, 312–313, Reference Traxler2004, 45) indicated the higher level of EA membership of manufacturing sector companies. In fact, the growth of the services sector has even been considered as a factor of membership decline (Gooberman et al Reference Gooberman, Hauptmeier and Heery2019b, 13–14). Studies on Spain and Italy show the predominance of the industrial sector in EAs (Fanfani et al Reference Fanfani, Lucifora and Vigani2023; Sánchez-Mosquera Reference Sánchez-Mosquera2022) and this could be compatible with the decline in membership hypothesised. In this respect, the interaction of this independent variable with other variables should be taken into account, as the industrial sector has a long tradition of sectoral bargaining and a strong union presence (Demougin et al Reference Demougin, Gooberman, Hauptmeier and Heery2019, 10). This has been addressed and resolved in the following section.

Different institutional frameworks and unequally affected by unilateral institutional changes under pressure from the Troika

Evidently, the collective action of employers does not take place outside or regardless of legal and regulatory frameworks. Recent literature shows the great relevance of the institutions. This is evidenced by a wide variety of national cases analysed in relation to the response and adaptation of EAs to the widespread trends of liberalisation and deregulation of labour markets (Behrens and Helfen Reference Behrens and Helfen2019; Demougin et al Reference Demougin, Gooberman, Hauptmeier and Heery2019; Ibsen and Navrbjerg Reference Ibsen and Navrbjerg2019; Sheldon et al Reference Sheldon, Nacamulli, Paleotti and Morgan2016, Reference Sheldon, Della Torre and Nacamulli2019; Signoretti Reference Signoretti2016). This relevance of the institutional environment had already been affirmed by Traxler (Reference Traxler2004, 56).

The Southern European countries under study developed a particular variety of capitalism, usually referred to as the Mediterranean model (Amable Reference Amable2003). In sum, it is characterised by a mixed corporatist institutional architecture and a problematic labour market, with higher unemployment and the consolidation and advancement of atypical forms of employment that usually cause above-European average precarity. This variety of capitalism has been noted for its marked heterogeneity and relatively high rates of informal economy (Goergen et al Reference Goergen, Brewster, Wood and Wilkinson2012).

Starting in the 1990s, countries such as Ireland, Italy, Portugal, Spain, Finland, and, to a lesser extent, Greece concluded tripartite social pacts to adapt to the criteria of European Economic and Monetary Union (Mesch et al Reference Mesch, Pasztory, Zotter, Huemer, Mesch and Traxler1999, 186). These entailed a move towards agreed market deregulation, decentralisation, and flexibility, but without any radical changes. These pacts were generally favourable to the EAs of Southern European countries, as they made it possible to reach a deregulation and decentralisation based on competitive social pacts (Sánchez-Mosquera Reference Sánchez-Mosquera2018). This was advantageous to companies, but without losing their relevance or institutional influence in relation to employers’ labour market interests.

In these countries, employment regulation depends on labour law more than on collective bargaining, except in Italy (Hall and Thelen Reference Hall and Thelen2009, 26). Extensive collective bargaining coverage, meanwhile, benefited historically from statutory (state) extension mechanisms for agreements between unions and EAs, representative of all the companies and workers of the area of negotiation. In Italy, extensive coverage has been achieved through collective bargaining itself (Koukiadaki et al Reference Koukiadaki, Távora, Martínez Lucio, Koukiadaki, Távora and Martínez Lucio2016, 20–21), supported by the constitutional obligation to pay a ‘fair wage’. In fact, this is a functional equivalent of extension mechanisms because judicial practice refers to the reference collective agreement to determine what the level of a ‘fair wage’ is (OECD 2017).

Union density is low, again with the exception of Italy. In fact, this country is distinguished by having a tradition of more robust social partners and regulatory practices arising from sectoral bargaining. In comparative terms, this has mitigated the impact of the common trends of deregulation (re-regulation). This country has the greatest degree of territorial and sectoral coordination of collective bargaining of those analysed. In Italy, the large sectoral agreements at national level (which set minimum working conditions in the sector throughout the country) were decentralised, organised in subsectors and territorially and, where applicable, in company agreements. Company agreements had to have at least equal conditions set at intermediate and higher (national) levels. This model has certain similarities with the Nordic countries (Marginson Reference Marginson2015). The other countries analysed did not have this level of organisation.

The Southern European area in particular suffered the severity of the recent economic crisis, with serious recessions and sovereign debt crises. Governments have been subject to considerable external pressure to liberalise and decentralise industrial relations. The wave of unilateral reforms (Koukiadaki et al Reference Koukiadaki, Távora, Martínez Lucio, Koukiadaki, Távora and Martínez Lucio2016) could be considered as being disruptive, and thus a crisis of profound institutional change could be added to the economic crisis. The persistence and severity of the economic recession, rampant public deficit and debt and pressure from the Troika and the memorandums of understanding that accompanied aid programmes have been decisive factors in the process (Addison et al Reference Addison, Portugal and Vilares2017; Schulten and Müller Reference Schulten, Müller, Natali and Vanhercke2013). In Portugal (memorandum of understanding of 2012) and Greece (memorandums of understanding of 2010, 2012 and 2015) the pressure was greater than in Spain (except in relation to its banking sector) and Italy, also under pressure from the ECB for its high public debt and financing needs. In Portugal, the 2009 reform, drawn up before the crisis, already limited the principle of favourability and the durability of agreements beyond their end date. These changes were already subjecting its system to a certain stress but did not actually affect the collective bargaining coverage rates.

The radical reforms curbed the competitive social pacts and limited the effectiveness of sectoral agreements in favour of company agreements (Sánchez-Mosquera Reference Sánchez-Mosquera2022). As Table 1 shows, this produced a strong legislative impetus in favour of disorganised decentralisation. However, employers were more inclined to advocate such arrangements and even endorse having no collective bargaining (Addison et al Reference Addison, Portugal and Vilares2017, 556; Koukiadaki et al Reference Koukiadaki, Távora, Martínez Lucio, Koukiadaki, Távora and Martínez Lucio2016, 52; Marginson Reference Marginson2015, 104–106). Portugal concluded a tripartite agreement at the beginning of 2012, but without the participation of the most important union, Confederação Geral dos Trabalhadores Portugueses – Intersindical Nacional (CGTP-IN). This agreement nonetheless included the demands laid down by the Troika and even went further in some respects. In general, the discontinuation of social concertation and government agreements led the unions to reconsider their political stance and gave rise to a number of general strikes throughout Southern Europe.

Table 1. Characteristics and changes in collective bargaining in Southern Europe before and after the radical reforms (2010–2012)

X: strong; x: weak.

Decentralisation of collective bargaining in which X: organised; and x: disorganised.

EL = Greece; IT = Italy; PT = Portugal; ES = Spain.

(*) In PT, the limitation of conditions to activate the mechanisms (state mechanisms under government control) to extend national collective was imposed by Law 23/2012. But the Resolution 82/2017 reintroduced the extension of collective agreements.

(**) Time-limited continuation. EL severe (3-month) and ES (12-month) and PT (18-month) moderate.

(***) ES introduced opt-out clauses in 1994, and these clauses were extended in 2010 and 2012. In Italy, this type of clause was agreed in sectoral and inter-sectoral bargaining.

Source: Prepared by the authors on the basis of OECD/AIAS Database (2021), Koukiadaki et al (Reference Koukiadaki, Távora, Martínez Lucio, Koukiadaki, Távora and Martínez Lucio2016) and Marginson (Reference Marginson2015).

Despite everything, as Table 2 shows, no dramatic change in the unionisation trend of these countries has been noted after these reforms, this being a declining trend as in the rest of Europe (Vandaele Reference Vandaele2019).

Table 2. Average rates of union density (UD) and collective bargaining coverage (CBC) (%)

Note: (*) data from preceding or following years has been used to make up for the lack of information for the years indicated.

Source: OECD/AIAS Database (2021).

No significant alteration in the collective bargaining coverage rate has been observed either, with the relevant and notable exception of Greece. The Troika’s powerful influence over this country continued until 2018 and, subsequently, Law 4635/2019, passed by the government of ‘Nea Demokratia’, confirmed the hindering of collective action. Nevertheless, in line with the other Southern European EAs, the position of the Greek EAs was in favour of restoring the previous system in existence before the Great Recession (Voskeritsian et al Reference Voskeritsian, Veliziotis, Kapotas and Kornelakis2017). On the contrary, there has been no important change in the collective bargaining systems and prevalence of sectoral bargaining over company agreements in Italy, Portugal, and Spain (Bulfone and Afonso Reference Bulfone and Afonso2020; Colombo and Regalia Reference Colombo and Regalia2016; Cruz Villalón Reference Cruz Villalón2015).

Furthermore, the economic recovery coincided with a partial reversal of the changes referred to in Table 1, as well as there being no exploitation of these changes by the EAs. In Portugal, despite the regression experienced in 2010–2012 (Campos Lima Reference Campos Lima2016), the end of the strict conditions imposed by the Troika (2014) and a new ‘progressive’ government (2015) were fundamental in restoring the traditional parameters of union participation and collective agreements in the country. In Spain, meanwhile, the Spanish employers’ peak organisation favoured negotiated interconfederal agreements for the (multi-employer) coordination of the whole collective bargaining system from 2010 to 2020 (Sánchez-Mosquera Reference Sánchez-Mosquera2022, 1, 164).

Results: evolution of representativeness, composition, and determinants of membership

Table 3 helps respond to the first research question and points to the verification of H1. These results indicate decline in terms of the representativeness of EAs. This was especially detrimental in Greece, as the rate of EA membership was 6.5%. On the contrary, the decline was far less pronounced in the case of Portugal. Thus, the first results fit with the resilience capacity of the different systems of collective bargaining (Table 2) and also point to the verification of H2.

Table 3. EA membership density of companies with 10 or more workers as a percentage of the total number of such companies

Source: ECS (2013, 2019).

The following Tables 4 and 5 address the second research question. The general results of EA membership should be disaggregated by company size (Table 4). The analysis emphasises the role of large and medium-sized companies, which, despite not being predominant numerically, are more stable and employ a greater number of workers.

Table 4. Member companies of EAs by company size as a percentage of total number of companies

Source: ECS (2013, 2019).

Table 5. Member companies by sector as a percentage of total number of member companies of EAs in 2013 and 2019 in Southern Europe

Source: ECS 2013, 2019.

The drop in membership among large and medium-sized companies marks the two extremes of the group of countries studied. On the one hand, Portugal shows great stability in the percentage of EA membership of these types of companies, practically invariable in the case of large companies. On the other hand, the sharp drop in affiliation among large companies in Greece should be related to the significant decrease in EA membership in general (2019). In the case of Italy, the high membership rate of large companies and, despite the decline, the acceptable rate of medium-sized companies have helped maintain representativeness. Finally, there has been a notable fall in membership among large companies in Spain.

Moreover, it is evident from Table 4 that the weight of membership by company size reflects the greater organisational influence of large companies (with 250 or more employees), and although their membership rate has fallen (2019) they continue to have the highest membership rates in all southern countries. The membership rates of small companies were the lowest in all four countries in 2013 and 2019. These results point to the verification of H3. In these countries, the strong presence of small companies (including micro-enterprises) in their productive fabric is in stark contrast with the profile of EA membership.

Regarding the sectoral profile, Table 5 addresses the membership weight by sector. The industrial sector was far greater than other sectors in 2013, followed by the commerce and hospitality sectors. The industrial sector was even more outstanding in this respect in Italy. However, the 2019 wave of the ECS showed that this sector no longer occupied first place in terms of membership, but rather the second position, in Portugal and Spain. It was overtaken by commerce and hospitality, in Portugal, and other services (without public services) in Spain.

The descriptive statistics outline an organisational map of EAs in Southern Europe dominated by large companies and the industrial sector (with the nuances indicated in Spain and Portugal in 2019), and only indicate the partial verification of H4. This also points to sectoral divergence between EA membership and the productive fabric in these countries. In the Spanish case, this divergence is not so pronounced in 2019. These are all elements related to the third research question, which contributes to the debate about why the EAs of Southern Europe neither supported nor exploited the radical reforms of 2010–2012 (in the following discussion section).

In order to complete the analysis and estimate the determinants of EA membership, an analytical study was conducted. First, a study of association of the regressor variables of interest with EA membership was carried out, which determined the lack of relationship of membership with evolution of employment in the company over the previous 3 years in 2013; temporary employment in the company in 2019; and part-time employment in the company in 2013 and 2019 (Appendix D). Thus, only the variables that have association with EA membership were taken into consideration for constructing the logistic regression models (Table 6).

Table 6. Odds ratios. Logistic regression models in Southern Europe

95% Confidence Interval for Odds Ratios in parentheses.

Source: ECS (2013, 2019).

* p < .05,

** p < .01.

The models approximate the statistical link between the set of independent variables and the dependent variable (membership of EAs) and were analysed on the following base categories: Greece, employee representative not present (no union representation), no multi-employer agreement (sectoral agreements), no single-employer agreement (company agreements), company size 10–49 employees, industrial sector, positive evolution of employment in the previous 3 years (increase in number of employees in the previous 3 years), and less than 20% of company staff in temporary employment. With the exception of the industrial sector and the percentage of temporary workers, the variables that are expected to have a less intense relationship with the dependent variable, EA membership, were used as base category.

The results of the two models have been taken into consideration separately. In the 2013 model, the country variable, which captures the economic and institutional framework, shows a substantial and significant difference in favour of Italy and Portugal with respect to Greece, and without statistical significance between Spain and Greece. On this basis, multi-employer bargaining is a greater predictor of EA membership, but to a lesser extent, this is also true of the single-employer collective agreements and union representation categories, in this order, compared with the base categories where these three variables are absent. The large company category has the greatest importance in terms of determining EA membership in the model with respect to its base category, small company. Companies in the financial services sector show most membership with respect to the base category and industry, and moreover, it is prevalent with respect to the other significant sectors (commerce and hospitality and other services). There is not sufficient statistical evidence to affirm that there is a significant relationship between the rest of the variables studied and EA membership. This suggests that there is a very significant influence of the institutions and the provision of traditional collective goods to companies: representation in sectoral bargaining; and representation, advice and defence against union representation. Company size is more significant than sector. The sector has non-significant categories and the differences in probability between the significant categories are not especially high, with the highest probability corresponding to the financial services sector.

In 2019, EA membership showed significant differences in the country variable, with intensity favourable to Portugal, Spain, and Italy, in that order, compared to the base category Greece. This not only reveals the relevance of this variable but also indicates a sharp contrast between Greece and the rest of the countries analysed. This model also indicates a greater intensity of significance in the relationship with EA membership in multi-employer bargaining, union representation and single-employer agreements categories, in this order, compared with the base categories where these three variables are absent. In the size of company variable, an increasing and significant direct relationship can also be observed. With regard to the sector, there is a greater probability of affiliation in the industrial sector, but only in comparison with transport and communication and other services, the remaining sectors do not have any significant difference with the base category. Finally, the model shows a direct and significant relationship between companies that register job destruction and EA membership compared with the base category, companies that create jobs. The great influence of institutions and the provision of traditional collective goods to companies with regard to affiliation are also evident in this model. This suggests that disaffiliation stems to a greater extent from companies that do not use or have stopped using these services. Nevertheless, in this model, it is also important to emphasise the lack of statistical significance between most categories and the base in the sector variable.

The analytical study has confirmed H1 and H2. These results are not only in line with the literature but also, by virtue of the greater influence of sectoral bargaining, once again point to the importance of the reforms of 2010–2012 and the influence of the Troika. These were aimed at reducing collective bargaining, especially sectoral, a key element in the industrial relations models of these countries. This was already evident in the case of Greece in 2013. The study has also confirmed the invariable predominance of large companies (H3). Despite the role of the industrial sector with respect to other sectors in determining EA membership, financial services (a sector with high union membership) had a higher predictive effect in this respect in the 2013 model. In the 2019 model, however, the greater predictive intensity of the industrial sector prevails, but it is only significant in two of the other five sectors studied. This, therefore, once again suggests a partial verification of H4.

Discussing the representativeness of employer associations and their capacity to participate in the industrial relations system in Southern Europe

According to the OECD/AIAS database (2021), the percentage of the total workforce comprised of employees of member companies evolved as follows: from 58.4% (2008) to 52.7% (2017) in Greece; from 64.8% (2012) to 78.3% (2018) in Italy; from 65% (2008) to 51.2% (2014) in Portugal; and from 75% (2013) to 77% (2018) in Spain. In other words, it shows a great stability and even an increase in membership in some cases. In the same vein, Brand and Lehr (Reference Brandl and Lehr2019) indicated the high level of resilience in general of Western European EAs, including Italian and Spanish associations, until 2012.

Recent studies have provided more reliable figures and are more consistent with the results obtained here, for example, 44% for Greece (2008) (Katsaroumpas and Koukiadaki Reference Katsaroumpas, Koukiadaki, Müller, Vandaele and Waddington2019). With greater implications for this study, however, an important decline was registered for Italy up to 2015 and for Spain until 2019 (Fanfani et al Reference Fanfani, Lucifora and Vigani2023; Sánchez-Mosquera Reference Sánchez-Mosquera2022). Likewise, the results obtained here show that the evolution of membership has shown a general decline (extreme in Greece).

Within this general decline in membership, this study highlights the decrease in large companies membeship (Table 4). This type of company is a key element of EAs because, as they employ more workers, they are very important disseminators of sectoral agreements to the workforce as a whole. They also exercise a leadership that would be difficult to substitute, which helps give employers, subject to conflicting interests, a coherent voice (Olson Reference Olson1965), and their logic of influence (Schmitter and Streeck Reference Schmitter and Streeck1999). Furthermore, if they let their membership lapse this entails a loss of financial resources and prestige that it is difficult for EAs to compensate (Sheldon et al Reference Sheldon, Della Torre and Nacamulli2019, 19).

The percentage of EA membership of both large and medium-sized companies in Portugal remained fairly stable. Thus, Portuguese company representation did not decline significantly with respect to the country’s workforce. Despite Fiat’s withdrawal from Confindustria, Italy has also shown considerable stability in terms of the affiliation of large companies to EAs. This is the country with the highest rate of collective bargaining coverage and the greatest union density (Table 2), and also the best-coordinated collective bargaining system (Marginson Reference Marginson2015) in the region. The fall in EA membership in Italy, concentrated in its medium-sized companies, had already been documented before the Great Recession (Fanfani et al Reference Fanfani, Lucifora and Vigani2023). Consequently, the reduction experienced from 2013 to 2019 would be part of a general trend prior to the effect of the Great Recession and the accompanying legislative changes of little real impact (Table 2).

This could also be extrapolated to Portugal and, to a lesser extent, Spain. The decline in EA membership among large Spanish companies could be explained by the regulatory change of 2012 that gave preference to company agreements (less influential with regard to EA membership) over sectoral ones. The labour reform of 2021 (Royal Decree-Law 32/2021, of 28 December), agreed with the Spanish employers’ peak association, reversed this change in such an important area as wages, where the sectoral agreement once again takes precedence. It will be interesting to monitor whether this change curbs or reverses the membership trend described up to 2019. With regard to small companies, Spain has experienced the smallest drop in membership among the countries analysed, and in 2019 these were the small companies with the highest rate of representation in EAs in Southern Europe. Nevertheless, it cannot be said that these changes altered the balance of power within the Spanish employers’ peak association (Sánchez-Mosquera Reference Sánchez-Mosquera2022).

The Greek case, meanwhile, involves serious obstacles to the restoration of fundamental institutions such as sectoral extension of collective bargaining, which also requires EAs to represent companies employing over 50% of the total workforce. A vicious circle has been created where EAs do not have capacity for representation beyond their member companies and where employers do not have incentives to affiliate as the sectoral collective agreements do not affect them and union representation (and pressure) has generally decreased in the workplace (Karamessini and Giakoumatos Reference Karamessini, Giakoumatos and Vaughan-Whitehead2016, 190). Greek EAs can react by offering selective and elective goods, but the collapse of collective bargaining must have sent out shock waves and created an organisational predicament difficult to deal with. The situation is similar to the case of the United Kingdom in the 1980s and 1990s (Waddington Reference Waddington, Bamber, Lansbury and Wright2016), but the changes have been implemented with greater speed and comprehensiveness. This means that it is no longer very realistic to continue including Greece in the classification of Mediterranean capitalism of Southern Europe made by Amable (Reference Amable2003).

With regard to factors that favour membership at company level, this study coincides with the literature in that union membership (Gooberman et al Reference Gooberman, Hauptmeier and Heery2019b, 14; Traxler Reference Traxler2000, 313, Reference Traxler2004, 47) and collective bargaining (Traxler Reference Traxler2004), above all multi-employer, have a powerful and unquestionable influence, confirmed in both regression models (Table 6). This could underline the countervailing power indicated in the literature (Barry and Wilkinson Reference Barry and Wilkinson2011). However, the greater influence of multi-employer bargaining on EA membership compared to company-level negotiations, and even compared to union density, underlines the importance of the representation of collective goods for companies (the classic historical functions of this type of organisation). The fundamental factor that explains EA membership is not so much trade union power as the collective bargaining process, although this obviously includes union pressure (Gooberman et al Reference Gooberman, Hauptmeier and Heery2019a; Silvia and Schroeder Reference Silvia and Schroeder2007). Thus, as has been shown in very different institutional settings, the effects of the decentralisation of collective bargaining constitute a greater organisational threat than the decrease in union power (Sheldon et al Reference Sheldon, Nacamulli, Paleotti and Morgan2016, Reference Sheldon, Della Torre and Nacamulli2019).

This is especially relevant in industrial relations systems with relatively low union membership, such as the Southern European countries except Italy, where the important factors are high collective bargaining coverage rates and the availability of state and legal mechanisms to extend collective agreements. It was precisely the reforms of 2010–2012 that hindered the extension of sectoral collective agreements and favoured negotiation at company level and even an absence of collective bargaining. This is so radical that even the association representing large Greek employers, the Hellenic Federation of Enterprises (SEV), which welcomed the reforms and openly favoured the primacy of company agreements over sectoral ones (Katsaroumpas and Koukiadaki Reference Katsaroumpas, Koukiadaki, Müller, Vandaele and Waddington2019, 273), has understood the importance for the organisation and its members of having a collective bargaining system that has at least a minimum of territorial and sectoral coordination (Voskeritsian et al Reference Voskeritsian, Veliziotis, Kapotas and Kornelakis2017, 9).

Also in line with the literature, the results confirm the direct and predominant relationship of the industrial sector and financial services (only in the first model), with a long tradition of sectoral bargaining (and high union membership), compared with the so-called ‘other services’ (Traxler Reference Traxler2000, 312–313). This negative relationship remains invariable to the different composition of the ‘other services’ category, which included public services in 2013 but not in 2019. Nevertheless, the condition of large company is a more important factor in membership than belonging to any particular economic sector, which is also something that the literature has indicated for unions (Schnabel Reference Schnabel2013). Furthermore, larger companies usually have a greater degree of union representation (Ebbinghaus et al Reference Ebbinghaus, Gobel and Koos2011, among others) and also have an interest in sectoral bargaining as a way of regulating competition (Traxler Reference Traxler, Sadowski and Jacobi1991, 44) or even neutralising competition from smaller companies via labour costs, as in the German case (Silvia and Schroeder Reference Silvia and Schroeder2007).

The answer to third research question contributes to the literature by discussing why EAs in Southern Europe have not been in favour of these significant institutional changes (Bulfone and Afonso Reference Bulfone and Afonso2020; Colombo and Regalia Reference Colombo and Regalia2016; López-Andreu Reference López-Andreu2019; Regalia and Regini Reference Regalia and Regini2018, 73–75; Voskeritsian et al Reference Voskeritsian, Veliziotis, Kapotas and Kornelakis2017, 5). In this vein, the findings have shown that EAs in Southern Europe have had to confront two important paradoxes that have already conditioned them and will also do so in the future. The first is the notable disparity between company membership and sectoral representativeness within their organisations, on the one hand, and the productive fabric that they represent, on the other. This disparity obviously leads to contradictions. However, the ECS of 2019 qualifies this first paradox. More so in the Spanish case, as the profiles of EA members have a closer relationship with the predominant characteristics of its business fabric: small companies and the ‘other services’ sector (Tables 4 and 5). In Portugal, this is also true sectorally, as commerce and hospitality is the first sector in terms of affiliation, ahead of the industrial sector, with the former also being naturally more important in the productive fabric of this country. It would be necessary to monitor the forthcoming waves of the ECS in order to analyse whether, at least sectorally, this paradox between representation (EA membership) and productive fabric is being resolved.

With regard to this first paradox, moreover, the bias in favour of large companies increases due to the effect of the superior political influence and power of large employers in these organisations, as indicated by theory (Olson Reference Olson1965; Traxler Reference Traxler2000, Reference Traxler2004). This could help to refute the affirmation that one of the main reasons for not implementing the reforms was the preponderance of micro and small companies in the industrial fabric of these countries (above the EU average) and their enormous difficulties in concluding company agreements with their employees (Bulfone and Afonso Reference Bulfone and Afonso2020). In fact, the underrepresentation of micro and small companies makes it unlikely that their interests are taken into account.

The second paradox is that the reforms favoured the unilateral power of companies, but weakened the representativeness and power of EAs. The traditional institutions and, particularly, sectoral bargaining create a space for discussion where the power of companies is reduced. They also need to preserve the balance of the industrial relations system that they have helped maintain, and this implies relatively higher labour costs and greater union power. Favouring and taking full advantage of deregulation and the accumulated unilateral power of companies discourages membership and reduces the institutional influence and power of EAs, as the cases of certain anglophone countries have shown (Sheldon et al Reference Sheldon, Nacamulli, Paleotti and Morgan2016; Sheldon and Thornthwaite Reference Sheldon, Thornthwaite, Gooberman and Hauptmeier2022; Waddington Reference Waddington, Bamber, Lansbury and Wright2016). On the other hand, as mentioned above (p. 11), the EAs of Southern European countries had already experienced competitive social pacts from the 1990s (although less in the Greek case). These opened the door to agreements on deregulation and decentralisation that were favourable to their members (companies) without reducing their power as an organisation, and in fact, having the opposite effect. As Clegg (Reference Clegg1976, 108) indicated, the attitudes of employers and EAs are very important in designing and implementing changes in industrial relations systems (especially collective bargaining). Despite everything, the employers of the countries analysed had a solid tradition of concluding agreements (favourable to their members), which seems to have distanced them from the attitudes shown by many employers and EAs in other countries.

Finally, in the 2019 model, for one of the control variables, evolution of employment, it turned out that EA membership was notably favoured by companies where employment decreased compared with thriving companies. Could it be that economic difficulties favour membership? Brandl and Lehr (Reference Brandl and Lehr2019) examined this with aggregate macroeconomic data on growth but failed to obtain sufficient statistical evidence to reach a firm conclusion. This study does show this relationship with microdata, but in a year, 2019, economic growth in the countries analysed. As well as the possibility that economic hardship favours membership, this could suggest that more dynamic companies, those that create jobs, turn away from EAs and this, evidently, would create a trend unfavourable to affiliation. This is something that would be worth exploring in future research.