Social democracy transformed? Party change and union ties

Abstract

Social democratic parties have undergone a transformative ideological and electoral shift. But how has this affected their alliance with trade unions in terms of personal ties? This article hypothesises that Social democratic parties appoint fewer union-linked ministers as they become more economically centrist and less dependent on working-class voters. However, institutions that stabilise party–union relations should moderate these effects (statutory linkages, high union density, or union involvement in policymaking). Data on 2,600 ministerial appointments in Western Europe show that there is no direct relationship between trade unionist appointments and changes in party electorates or party ideology. However, electoral change is correlated with appointment patterns when institutional stabilisers are weak. Thus, social democrats weaken their elite ties to unions in response to the middle-class shift in their electorates, but only when the institutional context is unfavourable for the party-union relationship.

Keywords:

• Social democracy
• trade unions
• party–union linkages
• ministers

Social democratic (SD) parties have been important political actors in many Western democracies since the late nineteenth century (Keman 2017). In alliance with trade unions, they have shaped the political economies of Europe (Allern and Bale 2017). In many countries, the ‘siamese twins’ (Ebbinghaus 1995) of the labour movement worked in lockstep to promote worker interests and establish new social rights (Howell 2001; Korpi 1980; Padgett and Paterson 1991; Poguntke 2002). This party–union relationship has been underpinned by a political exchange: Unions offer parties organisational, financial, and electoral support in return for a labour-friendly regulatory environment, pro-employment and pro-welfare policies (Howell 2001).

However, over the past decades, social democracy has undergone significant electoral and ideological transformations (Kitschelt et al. 1994): As de-industrialization has progressed throughout Europe, social democratic electorates have become more middle-class (Gingrich and Häusermann 2015). At the same time, some social democratic parties have changed ideologically to become more open towards globalisation, business, and the market economy (Giddens 1998; Hall 2002). Both these developments may challenge established party–union linkages. Yet, we theorise that the extent to which they will do so depends on the presence of institutional anchors (statutory ties, union density, union involvement in policy making) that have the potential to isolate party–union relationships from electoral and ideological change.

This article thus examines how ideological and electoral change have affected the linkages that bind unions and social democratic parties together, and how institutions moderate these effects. As multiple case studies suggest (Allern et al., 2007; Anthonsen et al., 2011; Aylott 2003; O’Grady 2019; Thomas 2001), the mutually beneficial bargain that once ensured party–union cooperation has weakened (see also Allern and Bale 2017). Yet it remains unclear to what extent the case-specific explanations that have been put forward in small-N studies apply more generally. Our understanding of how the electoral and ideological transformation of social democracy has affected party–union ties thus remains limited.

In order to address this problem, we combine survey and manifesto data with a novel data set on social democratic ministerial appointments to provide the most comprehensive analysis of personal linkages¹ between parties and unions to date. Our analysis examines the biographical backgrounds for around 2,600 ministerial appointments made by social democratic parties in 16 West European countries between 1960 and 2014 and identifies whether ministers had links to trade unions prior to their appointment. By establishing such personal linkages, parties provide unions with access to the policy-making process in exchange for electoral and political support.

Using personal linkages as an indicator of party–union ties has upsides and downsides. On the one hand, such linkages can be seen as informal, yet powerful organisational ties at the individual level (Allern et al. 2020: 1258) that enable unions to exert significant policy influence over policy making (Adolph 2013; Alexiadou 2015, 2016). On the other hand, ministerial appointments are not just expressions of organisational linkage, but originate in multidimensional political calculations involving demands for regional and factional representation (Ceron 2014; Ennser-Jedenastik 2013), gender balance (Goddard 2019), or the availability of subject-matter expertise (Alexiadou and Gunaydin 2019). Given their importance in providing access to the policy-making process, we nevertheless consider them highly relevant objects of analysis.

Our analysis documents a steep decline in personal linkages between social democratic parties and unions: Between the 1960s and the 2010s, the proportion of social democratic ministers with trade union ties almost halved from 30 to 16%. Yet, despite a temporal coincidence of this trend with substantial electoral and ideological change, we find no direct association between trade unionist appointments and party ideology or party electorates.

However, we find support for one of our conditional expectations: As social democrats become less reliant on working-class voters, their propensity to appoint union-linked ministers declines – but only in the absence of institutional stabilisers (union density, concertation). Yet no such interaction effect exists for the association of party ideology and the selection of trade unionists as government ministers. While ideological shifts have certainly affected party–union ties in some instances (Allern et al., 2007; Christiansen 2012; O’Grady 2019), they do not account for the general disintegration of social democratic parties and trade unions in terms of ministerial appointments.

These results significantly deepen our understanding of party–union relationships and highlight the importance of institutional context factors for their stability. The political transformations that social democrats have undergone in the post-industrial era clearly have the potential to affect these parties’ ties with trade unions. However, such changes do not play out uniformly, but are shaped by the institutional framework in which party–union relationships are embedded.

Our findings also have implications for understanding political elite recruitment and its (potential) effect on policy outputs. As social democratic parties rely less on unions as recruitment pools, they must find ministerial appointees elsewhere. As work by Alexiadou (2015, 2016) shows, this may push socio-economic policy outputs to the centre.

The following sections introduce our theoretical arguments and derive four hypotheses from this discussion. We then go on to outline our empirical strategy, before delving into the analysis. The conclusion discusses our study’s implications for understanding ties between unions and social democratic parties, highlights limitations of our approach, and points out potential avenues for future research.

Social democracy and trade unions: an exchange relationship

The relationship between trade unions and social democratic parties has typically been conceptualised as a cost–benefit model of political exchange in which both sides engage to further their goals (Allern et al., 2007; Allern and Bale 2017; Howell and Daley 1992; Pizzorno 1978; Poguntke 2006; Quinn 2002). Like any other exchange, this relationship presupposes that both actors do not value each exchange good equally and can thus obtain ‘gains from trade’: Unions seek to influence public policy in the interest of their members and wage earners in general. After all, the bargaining power of union members is inextricably linked to that of the wage-earning population as a whole (Ebbinghaus and Visser 2000; Hyman 2001). Social democratic parties, by contrast, have vote-seeking and office-seeking ambitions in addition to their policy goals (Müller and Strøm 1999; Strøm 1990). Given these varying motivations, the ‘obvious’ (though certainly not the only) mutually beneficial exchange is for parties to provide unions with access to and influence over the policy-making process, whereas unions support parties in achieving their electoral (and thus office) goals.

In order to maintain such a relationship in practice, parties may grant unions policy influence in multiple ways, ranging from informal consultations to joint policy committees, from union representation in party organs to the appointment of union representatives to public office – the empirical focus of this article. In return, unions may further social democrats’ ambitions in terms of financial resources, human capital (e.g. campaign staff), ideational, or organisational support (Allern et al. 2017; Allern et al. 2021).

Indeed, the relationship between social democratic parties and unions goes beyond a purely instrumental logic. They share deep historical roots that continue to shape their relationship today. For example, party–union ties today are still influenced by the sequence of party and union formation and the strength of Christian, Communist, and syndicalist tendencies in the labour movement (Ebbinghaus 1995). As this variation produced significant cross-country differences in party–union relations, the weakening of ties between social democrats and unions in the late twentieth century was especially notable where the two had previously been very close (Christiansen 2012; Padgett and Paterson 1991). As a consequence, the relationship between unions and social democratic parties became more instrumental in times of increased electoral competition (Howell 2001: 13).

Appointments as linkage

The personal linkages between social democratic parties and trade unions constitute only one of many potential linkage types between parties and interest groups (Poguntke 2002; Thomas 2001). At a most basic level, three forms of ties between parties and trade unions can be distinguished (Allern et al. 2020): organisational (collective membership, mutual executive representation, liaison committees, leadership overlap), material (money, staff, information), and ideological ties (policy positions and programs).

More specifically, organisational ties can be understood as ‘structured interaction’ between parties and unions. Thus, the union-linked appointments we examine should be viewed as informal organisational ties at the individual level (Allern et al. 2020: 1258), since there are typically no political or formal mandates for such appointments within unions or parties. Even so, unions’ policy influence certainly depends on the strength of their informal elite-level linkages (Adolph 2013; Alexiadou 2016, 2015).

There are at least three important caveats when examining ministerial appointments as party–union linkages: First, our premise is that these appointments are made by party actors. While it is true that the presence of nonpartisan appointees in European cabinets has increased (Vittori et al. 2023), the appointers generally remain firmly partisan. This is visible, inter alia, in the sustained recruitment of parliamentarians and party officers to ministerial positions (Almeida et al. 2003; Dowding and Dumont 2008; Kaltenegger 2023).

Second, the party actors who appoint ministers typically need to balance multiple demands and concerns, including regional and factional representation (Ceron 2014; Ennser-Jedenastik 2013), gender balance (Goddard 2019), and the availability of expertise (Alexiadou and Gunaydin 2019). In coalition governments, these demands are further complicated by the constraints of quantitative and qualitative portfolio allocation, i.e. the number and types of portfolios available to a party (Browne and Frendreis 1980; Ecker et al., 2015). Trade union representation in the executive is thus only one dimension of a multidimensional appointment game.

Third, one important pipeline leading to ministerial office is the parliamentary party group (De Winter 1991). A strong trade union presence among SD legislators could thus be an important predictor of union-linked ministerial appointments. However, cross-national data on legislators’ union presence simply does not exist. Furthermore, we assume that our theoretical argument applies to legislators no less than it does to ministers. If so, the analytical gain from including such data would actually be limited.

The scant evidence that exists from single-country studies suggests that ministerial appointments of trade unionists have become less common over the past decades (Davidsson and Bäck 2019; Ennser-Jedenastik 2017). In the following section, we theorise that this trend (which we also find in our data) is due to the electoral and ideological changes in the post-industrial era.

The transformation of social democratic electorates

From their founding days, social democratic parties relied on working-class voters as their most important electoral constituency: More workers meant more social democratic votes. Yet the working class never achieved majority status in any Western democracy – not even during the industrial period (Przeworski 1986: 23–24). Social democrats have thus always faced an ‘electoral dilemma’ in that they depended on some modicum of middle-class support to be successful (Przeworski and Sprague 1986).

While this dilemma has long been recognised (Kirchheimer 1966), it has been reinforced by the advent of post-industrial capitalism. As de-industrialization progressed, the proportion of working-class voters shrank. This has been one of the most important drivers of the electoral decline of social democratic parties (Best 2011). Across Europe, the contraction of the industrial working class has been associated with substantially lower vote shares for social democratic parties (Benedetto et al., 2020).

In addition to the decline of industrial employment and the rise of the service economy, a reconfiguration of political conflict around a second, non-economic, dimension has contributed to the change in social democrats’ voter coalition (Kriesi et al. 2006). As Kitschelt et al. (1994: 32) theorise, social democratic electorates in post-industrial societies will, as a result, include more white-collar voters, especially those working in public-sector, ‘symbol-producing’, and interpersonal jobs (see also Oesch 2006).

There is ample evidence for this transformation of social democratic electorates: As Gingrich and Häusermann (2015) document, middle-class voters have outnumbered working-class voters among (employed) centre-left voters in Europe since the 1990s. As working-class voters decline in numbers and also retreat from centre-left parties, contemporary social democratic parties rely more heavily on a cross-class coalition of voters (Rennwald 2020). Among the middle class, socio-cultural professionals, especially those employed in the public sector, have become an important stronghold of social democratic support (Abou-Chadi and Hix 2021; Oesch and Rennwald 2018).

What are the implications of these changes for party–union ties? Our argument rests on the premise that union-linked politicians are often viewed by parties and voters as representatives of the working class, broadly defined, or at least representing working-class interests. Certainly, many union members today are highly educated and work in white-collar professional occupations. However, as Arndt and Rennwald (2016) show, the electoral link between unions and social democratic parties remains strongest where historically allied blue-collar unions are still dominant.² While the class–union link has thus weakened overall, unions remain a potent ally for SD parties to attract working-class voters (Mosimann et al., 2019). Union affiliation among social democratic politicians can therefore be understood as a ‘class residual’ (Kitschelt et al. 1994: 46) that provides voters with cues about a party’s class loyalties.

In general, voters are not biased against working-class politicians (Albaugh 2021). Electorates even prefer them over wealthy ones (Campbell and Cowley 2014; Carnes and Lupu 2016; Hoyt and DeShields 2021; Vivyan et al. 2020, but see Wüest and Pontusson 2018). What is more, the class background of party candidates can affect the prevalence of class voting (Bellucci and Heath 2012; Evans and Tilley 2017; Heath 2015). For parties with a strong working-class base, it thus makes sense to appeal to this group by appointing trade unionists, since unions provide credible ‘class markers’ for parties.

Yet, the middle-class voters that social democrats increasingly rely on may view traditional party–union ties less favourably (Gumbrell-McCormick and Hyman 2013: 138–39). As Abou-Chadi and Wagner (2019: 1408) argue, the trade-off between appealing to working-class and middle-class voters becomes sharper when unions are stronger, since unions may be sceptical of shifting programmatic emphasis away from worker-oriented social consumption policies towards middle-class-friendly social investment programs. Still, Rennwald and Pontusson (2021) document that union members are less likely to abandon social democratic parties when they attract new non-worker voters.

Even so, the sharpened electoral trade-offs in the post-industrial age have led some social democratic parties who seek to appeal to new electorates to loosen their ties with trade unions (Allern et al., 2007; Christiansen 2012; Koelble 1992). As unions become less attractive electoral allies for social democrats, the transformation of social democratic electorates is likely to affect parties’ elite recruitment strategies, too. With working-class voters becoming fewer in numbers and less reliable in terms of vote choice, unions’ potential for electoral mobilisation diminishes. As a result, the necessity to reward them with ministerial appointments is declining.

Hypothesis 1. The less a social democratic party relies on working-class voters, the less likely it is to appoint trade unionists as ministers.

The transformation of social democratic ideology

Throughout their existence, social democratic parties have faced pressures to moderate and adapt ideologically. Debates over the merits of radical versus reformist approaches or about revolutionary versus parliamentary strategies characterised the party family’s early history (Keman 2017; Lipset 1983). After their accommodation to representative democracy, social democratic (and other) parties in the post-war era moderated further to become catch-all parties in response to softening class lines and increasing electoral competition (Kirchheimer 1966). They thus abandoned working-class symbols and class-based rhetoric and turned towards promoting Keynesian demand management and welfare state expansion instead of state ownership of the means of production (Padgett and Paterson 1991: 21–26).

In more recent decades, social democratic parties have not only witnessed radical changes to their electoral coalition (see above), but also adapted their programmatic profile – often in ways that antagonised their union allies. As the ideological distance between trade unions and social democrats has thus increased, party–union ties have weakened (Howell 2001).³

Beginning in the 1990s, some centre-left parties endorsed Third Way programs and policies (Giddens 1998). Under this label, social democrats became more accepting of globalisation, the market economy, and budgetary restraint. In addition, they endorsed social investment (as a partial alternative to social consumption) and supply-side reforms to labour market policy (Keman 2017). Peter Hall has identified four common themes among Third Way social democrats: a more pro-business stance, an embrace of the market, a preference for regulatory over spending policies, and a shift from viewing unemployment as a demand-side to a supply-side problem (Hall 2002). Importantly, party leaders saw union influence as antithetical to these policy goals (Larkin and Lees 2017: 66).

The most prominent examples of parties endorsing Third Way programs are the UK Labour Party and the German SPD (along with the U.S. Democratic Party). Yet it is important to note that similar tendencies were on display in smaller European democracies (Green-Pedersen et al., 2001; Green-Pedersen and van Kersbergen 2002). This produced a number of national and regional variations on the general theme of Third Way principles (Bonoli and Powell 2002). While Third Way rhetoric fell out of fashion in the aftermath of the 2008–9 financial crisis (and, importantly, after the electoral defeats of social democrats in Germany, France, and the United Kingdom), its policy prescriptions have survived to some extent (Arndt and van Kersbergen 2015; Bremer 2018).

From the perspective of trade unions, social democrats’ right-ward turn in economic ideology was not welcome (Ceron and Negri 2017), especially the supply-side oriented labour market policies and the greater tolerance for low-wage work. These developments ‘call[ed] into question policies designed to strengthen the trade unions’ (Hall 2002: 37). Yet, the Third Way diminished not only the ideological commonalities between social democratic parties and trade unions, it also went hand in hand with a decrease in working-class and unionist representation among social democratic elites (O’Grady 2019). This suggests that Third-Way social democrats view union-linked appointees as less faithful agents to carry out their socio-economic agenda. Conversely, trade union officials may be less eager to serve as ministers for centrist SD parties. Therefore we expect that ideological shifts towards the centre will be associated with fewer trade unionist appointments among social democratic ministers.

Hypothesis 2. As social democratic parties become more economically centrist, they appoint fewer trade unionists to ministerial office.

Institutions as moderators of electoral and ideological change

The two hypotheses outlined above theorise direct associations between electoral and ideological transformations and SD parties’ appointment decisions: Union-linked appointments are expected to correlate with the composition of party electorates and party ideology. Yet, it is unlikely that the impact of these two factors is independent of the institutional framework in which party–union relationships are embedded. The more stability an institutional framework provides to party–union relationships, the less likely it is that electoral and ideological change have large effects on appointment patterns.

As argued above, parties and interest groups have a long history of building linkages in multiple dimensions (organisational, material, ideological) (Allern et al. 2020; Allern and Bale 2017; Christiansen 2012), and this is especially true of social democrats and trade unions. While our empirical focus is on informal organisational linkages (elite-level appointments), formal, or statutory, organisational linkages such as collective membership or mutual representation/delegation in executive bodies tend to be much more stable in the short- and medium-term (Poguntke 2002, 2006). Whereas some social democratic parties have severed statutory ties with trade unions, others maintain such linkages. Even where ties have weakened, this was the result of long-term shifts, rather than short-term fluctuations, in cost–benefit calculations (Allern et al., 2007; Aylott 2003). Statutory linkages thus have the tendency to stabilise the relationship between parties and unions over time and can thus isolate short-term appointment patterns from changes in the composition of the electorate or party ideology:

Hypothesis 3. The association between union-linked appointments and electoral or ideological change will be weaker if statutory linkages between unions and social democratic parties are strong.

In addition, party–union ties are embedded in a set of socio-political institutions that characterise industrial relations and interest-group politics more broadly – most prominently captured as the degree of corporatism in a country (Jahn 2016; Kenworthy 2003; Siaroff 1999). Corporatist arrangements are characterised by centralised and highly concentrated peak organisations – often with a representation monopoly – for employers and employees that are embedded into a system of institutionalised policy formulation that also includes state actors (tripartism) (Jahn 2016; Molina and Rhodes 2002).

While corporatism is a multi-dimensional phenomenon, two central features for the purpose of this article are high levels of unionisation and the routine involvement of unions in policy making: Concertation – the regular involvement of corporatist peak organisations in policy-making processes – institutionalises relationships between trade unions and governments in a different manner. It provides an entry point for unions into the governing arena, familiarises union officials with the policy-making process, and thus stabilises elite-level ties between unions and the political sphere. These conditions strengthen relationships between social democratic parties and unions, and thus provide stability to party–union ties. We therefore assume that ministerial appointments will be less responsive to electoral and ideological shifts when unions are routinely involved in policy making (concertation).

Hypothesis 4. The association between union-linked appointments and electoral or ideological change will be weaker if levels of concertation are high.

High levels of union density lend bottom-up legitimacy to union demands, irrespective of labour market transformations and changes in the occupational structure of electorates. Out of electoral self-interest, SD parties are well-advised to maintain functioning relationships with large unions. In addition, union membership at the individual levels binds workers to SD parties even if their electorates become more middle-class (Rennwald and Pontusson 2021). High unionisation thus lessens the electoral trade-off for social democrats between their traditional and their new electoral constituencies. In sum, union density thus has the potential to isolate party–union ties from the effects of electoral and ideological change.

Hypothesis 5. The association between union-linked appointments and electoral or ideological change will be weaker if levels of union density are high.

Empirical strategy

Our key empirical contribution is a novel data set comprising information on the trade union background of around 1,200 social democratic ministers. The data set covers ministers from 19 parties in 16 Western European countries (EU-15 minus Greece, plus Switzerland and Norway) between 1960 and 2014 (see Table 5 in the online appendix). Note that the majority of individuals in our data is appointed to a ministerial position in more than one cabinet. Therefore, the 1,200 ministers correspond to over 2,600 appointments in total. We exclude reshuffle appointments within the same cabinet, though. Our results are robust to the exclusion of re-appointments in multiple cabinets (see Tables 8 and 9 in the online appendix).

Dependent variable: trade union background

Our dependent variable is binary and records whether a social-democrat-appointed minister has a career background in a trade union or similar labour movement organisation (1) or not (0). We define ‘career background’ as holding political positions (elected or appointed) in or being employed (e.g. as advisor, aide, lawyer, economist, …) by a trade union, trade union confederation, trade-union related organisation (e.g. educational organisations) or functionally equivalent associations (e.g. the Austrian Chamber of Labour). Union membership alone is not sufficient.⁴

In order to obtain these data, we collected and coded the biographies of around 1,200 social democratic ministers between 1960 and 2014. In a first stage we assembled biographies from national biographical dictionaries and Wikipedia. Biographies were then searched for union-related keywords (e.g. the national-language translation of ‘trade union’, ‘labour movement’ as well as names and abbreviations plus their derivatives of the most important unions and union confederations in a country). Inevitably, this search strategy returned many false positives (e.g. ministers who experienced conflict or negotiations with unions) which we eliminated by manually examining all matches to the keyword search.

In a next step, we cross-checked the trade union background of ministers against existing data (Blondel 1985) and Wikipedia’s ‘trade unionist’ categories (which exist separately for each language or nationality).⁵ To minimise the possibility of remaining false negatives we manually checked the union background for the rest of the ministers by country.

Figures 1 and 2 display the proportion of social democratic ministers with a trade union background across countries and over time. The time trend shows a continuous decline in the dependent variable between the 1960s and the 2010s. Across our period of observation, the proportion of social democratic ministers with a trade union background halves, from 30 to 16%. Most of this decline happens during the 1960s and 1970s.

Figure 1. Share of social democratic ministers with trade union background per decade.

Figure 2. Share of social democratic ministers with trade union background per country and over time. Note: Small number of cases for Ireland (34), Switzerland (35), Luxembourg (52), and the Netherlands (54). For all other countries, $N>130$.

The between-country variation by and large follows well-known patterns of union strength and corporatism across Western Europe. SD parties in the Nordic countries (Norway, Finland, Denmark) as well as those in Switzerland and Austria have the highest shares of union-linked ministerial appointments (around one in three), whereas some of the Southern countries (France, Italy, Portugal) display very low proportions of trade unionists among social democratic ministers (less than 10% each). The aggregate decline that is visible in Figure 1 is manifest in many individual countries, too. Notable exceptions are Austria, Finland, Ireland, Norway, and Sweden.

Independent variables

Working-class share among SD voters

To operationalise our first hypothesis (electoral change), we record for each observation (ministerial appointment) the share of working-class voters in the electorate of the appointing social democratic party. To obtain these data, we use cross-national and national surveys. Our main sources are Eurobarometer (1970–2001) and the European Social Survey (2002–14). To complement these data, we gather national election studies where available (see online appendix for a comprehensive list of sources). To identify party voters we combine prospective and retrospective questions – depending on which were asked in the survey (or where there are more observations available). Our working-class measure is based on respondents’ occupations and includes all employed individuals in working-class jobs.⁶ While the occupational categories vary somewhat across surveys, they are quite consistent in identifying working-class jobs: Typical categories are ‘skilled workers’, ‘manual workers’, ‘semi-skilled workers’ or ‘unskilled workers’. We aggregate them and then calculate the share of working-class voters for each relevant SD party and year and include only those observations where the total number of respondents voting for a certain party is higher than 20.

Parties’ economic left–right position

To operationalise our second hypothesis (ideological change), we calculate social democratic parties’ economic left–right positions based on data provided by the Manifesto Project (Lehmann et al. 2022). Following Lowe et al. (2011), we compute party positions as the logged proportion of right-wing minus the logged proportion of left-wing quasi-sentences (both values increased by 0.5 to avoid logging zeros and thus producing missing data points). We limit our position measure to economic categories, see Table 1.

Table 1. Manifesto categories used for ideology measure.

Left-wing categories	Right-wing categories
per403: Market Regulation	per401: Free Market Economy
per404: Economic Planning	per402: Incentives: Positive
per405: Corporatism/Mixed Economy	per407: Protectionism: Negative
per406: Protectionism: Positive	per414: Economic Orthodoxy
per409: Keynesian Demand Management	per505: Welfare State Limitation
per412: Controlled Economy	per702: Labour Groups: Negative
per413: Nationalisation
per415: Marxist Analysis
per416: Anti-Growth Economy
per504: Welfare State Expansion
per701: Labour Groups: Positive

Figure 3 shows how both of our key independent variables evolve over time. While there is quite substantial variation across parties, the average share of workers among social democratic electorates more than halves between the 1970s and the 2010s. As for ideological change, the local regression estimate shown in the right-hand panel of the figure indicates that there is a steady right-ward move up until the 1990s (the heyday of the Third Way), followed by a left-wing shift after 2000.

Figure 3. Key independent variables over time: share of workers among social democratic voters (left-hand panel) and economic left–right positions (right-hand panel). Note: In contrast to the analysis below, this figure includes data points for SD parties in opposition. The left-hand panel only uses data from Eurobarometer and the European Social Survey.

In addition to these two key independent variables, we require moderator variables to operationalise H3, H4 and H5. First, we include a binary measure of statutory linkage between social democratic parties and trade unions. This measure takes on the value 1 for country-years in which one or several unions had collective membership in the appointing party or were formally (i.e. as regulated by party statutes) represented in the party executive and 0 otherwise. We code collective membership and statutory representation based on information provided in case studies (Allern et al., 2007; Christiansen 2012; Quinn 2002), the information collected by Katz and Mair (1992), party statutes available from the Political Party Database (Scarrow et al. 2017), and digitised versions of historical party statutes obtained from a number of libraries (see online appendix for details).

Next, we include a measure of union density, that is, the proportion of wage and salary earners that are union members. We use the union density measure provided by the CPDS (Armingeon et al. 2019), as these data include some imputed data points and thus achieve greater temporal and spatial coverage than the original data collected by Visser (2019).

Finally, we use a three-level measure of concertation provided by Visser (2019) who categorises country-years into ‘no’, ‘partial’ and ‘full concertation’. We use ‘no concertation’ as the reference category and include binary indicators for ‘partial’ and ‘full concertation’ in our analyses.

Furthermore, we include the (logged) number of social democratic appointments per cabinet. This is because parties will have an easier time finding a ministerial portfolio for a trade union representative if they have more positions to fill.

We also add parties’ vote shares and vote share changes at the (lower-chamber, if applicable) legislative election preceding the appointment (taken from Lehmann et al. 2022), to control for the electoral size and performance of social democratic parties.

Next, we include unemployment rates for the country and year of the respective appointment. High unemployment weakens the bargaining power of unions, threatens collective bargaining coverage and lowers the reservation wage (Hein and Schulten 2004). Furthermore, it may produce right-ward shifts in parties’ programmatic profile (e.g. more focus on supply-side measures). Data are taken from the European Commission’s AMECO database.⁷

We also control for economic globalisation which exposes many workers in advanced economies to higher levels of competition through trade, outsourcing, or immigration (Sano and Williamson 2008: 484). Globalisation thus has the potential to weaken unions’ bargaining position vis-à-vis employers and political actors (Piazza 2001, 2005: 129–30; Scruggs and Lange 2002). To operationalise globalisation we use trade openness, the sum of a country’s imports and exports, expressed as a percentage of GDP. Data are taken from the Comparative Political Dataset (CPDS) (Armingeon et al. 2019). We log-transform this variable to eliminate skew (Table 2).

Table 2. Descriptive statistics.

	N	Mean	SD	Min	Max
Trade unionist appointment	2,627	0.21	0.41	0.00	1.00
% Workers among party voters	2,627	42.13	16.13	0.00	81.21
Economic left–right position	2,627	−1.72	1.13	−4.71	2.17
Union density	2,627	0.45	0.22	0.09	0.87
Statutory linkage	2,627	0.45	0.50	0.00	1.00
Concertation: partial	2,627	0.27	0.44	0.00	1.00
Concertation: full	2,627	0.48	0.50	0.00	1.00
Unemployment rate	2,627	6.02	3.97	0.00	20.80
Trade openness (ln)	2,627	4.09	0.35	3.28	5.64
SD appointments per cabinet	2,627	23.92	7.13	6.00	41.00
Vote share	2,627	34.24	10.76	2.96	51.03
Change in vote share	2,627	1.36	5.15	−11.24	17.89

Analysis

We first present binary relationships between our dependent variable and our two key independent variables. To that end, we calculate the percentage of union-linked ministers per quintile of the proportion of workers among SD voters and SD parties’ economic left–right positions.

As Figure 4 shows, the relationships are broadly in line with our hypotheses: The proportion of ministers rises steadily as we move from the quintile of observations with the lowest share of workers among SD voters to the top quintile. The slope is somewhat less steep for parties’ economic ideology, but there is still a notable drop in the proportion of union-linked ministers between the observations with the leftmost positions and those with the rightmost positions.

Figure 4. Share of union-linked ministers by quintiles of key independent variables.

Since our data are hierarchically structured (individuals nested in countries), we estimate two-level mixed-effects logistic regressions with trade unionist background as the dependent variable.⁸ To be sure, there is also a time dimension to our data, yet the data are not well-suited for time-series analysis, since data points within countries are clustered at certain points in time (e.g. multiple appointments happening at the start of one cabinet) and have long gaps (long absences from government, e.g. UK Labour during the Thatcher years or the German SPD during the Kohl chancellorship). We therefore attempt to account for time trends by including fixed effects for decades.

For the multivariate analysis, we first present three regression models to test the direct hypotheses (H1 and H2) in Table 3. The first model includes only the two key independent variables: the share of workers among SD voters and the economic left–right position of the appointing party. The second model adds all the control variables, and the third model adds decade-fixed effects to capture time dependencies.

Table 3. Mixed-effects Models: explaining trade unionist appointments.

	I	II	III
% Workers among party voters	0.0106**	0.00447	−0.00202
	(0.00371)	(0.00477)	(0.00630)
Economic left–right position	0.0374	0.0807	0.0805
	(0.0531)	(0.0589)	(0.0600)
Statutory linkage		0.0680	0.0599
		(0.176)	(0.183)
Union density		0.441	0.400
		(0.633)	(0.691)
Concertation: partial		0.121	0.0895
		(0.214)	(0.219)
Concertation: full		0.0305	0.0170
		(0.306)	(0.317)
Appointments per cabinet		−0.0152	−0.0113
		(0.0117)	(0.0120)
Vote share		−0.00340	−0.00567
		(0.0116)	(0.0122)
Change in vote share		−0.00630	−0.00666
		(0.0118)	(0.0120)
Unemployment rate		−0.0523*	−0.0398
		(0.0238)	(0.0278)
Trade openness (ln)		−0.116	0.0485
		(0.312)	(0.350)
Constant	−1.843***	−0.516	−0.627
	(0.227)	(1.619)	(1.675)
Variance: countries	0.397*	0.292	0.342
	(0.170)	(0.163)	(0.187)
Decade FEs	No	No	Yes
Observations	2,584	2,584	2,584
Log likelihood	−1,258	−1,253	−1,250
AIC	2,523	2,532	2,537

* $p<0.05$. ** $p<0.01$. *** $p<0.001$.

Table 4. Mixed-effects Interaction models: electorate and ideology effects conditional on institutional stabilisers.

	IV	V	VI	VII	VIII	IX
% Workers among party voters	0.00413	0.0384**	0.0196	−0.00111	−0.00156	−0.00124
	(0.00830)	(0.0141)	(0.0104)	(0.00638)	(0.00630)	(0.00649)
$\times$ Statutory linkage	−0.00939
	(0.00836)
$\times$ Union density		−0.0805**
		(0.0251)
$\times$ Concertation: partial			−0.0292*
			(0.0118)
$\times$ Concertation: full			−0.0275*
			(0.0112)
Economic left–right position	0.0861	0.0661	0.0674	0.143	0.264	−0.119
	(0.0600)	(0.0607)	(0.0623)	(0.0778)	(0.198)	(0.174)
$\times$ Statutory linkage				−0.142
				(0.112)
$\times$ Union density					−0.294
					(0.303)
$\times$ Concertation: partial						0.312
						(0.193)
$\times$ Concertation: full						0.163
						(0.193)
Statutory linkage	0.467	0.240	0.123	−0.182	0.0112	0.0424
	(0.405)	(0.191)	(0.194)	(0.265)	(0.189)	(0.191)
Union density	0.469	3.353**	0.575	0.295	0.0437	0.253
	(0.667)	(1.143)	(0.662)	(0.698)	(0.775)	(0.719)
Concertation: partial	0.148	0.142	1.269*	0.0939	0.0751	0.549
	(0.224)	(0.222)	(0.540)	(0.219)	(0.218)	(0.355)
Concertation: full	0.0756	−0.0661	1.071	0.0484	−0.0205	0.243
	(0.319)	(0.321)	(0.556)	(0.319)	(0.319)	(0.419)
Constant	−1.122	−2.021	−1.535	−0.364	−0.693	−0.879
	(1.680)	(1.719)	(1.654)	(1.699)	(1.657)	(1.677)
Variance: countries	0.308	0.341	0.281	0.351	0.327	0.361
	(0.167)	(0.190)	(0.160)	(0.191)	(0.179)	(0.195)
Decade FEs	Yes	Yes	Yes	Yes	Yes	Yes
Observations	2,584	2,584	2,584	2,584	2,584	2,584
Log likelihood	−1,250	−1,245	−1,247	−1,250	−1,250	−1,249
AIC	2,537	2,528	2,533	2,537	2,538	2,537

* $p<0.05$. **$p<0.01$. ***$p<0.001$.

In model I, the proportion of working-class voters in social democratic electorates yields a significant coefficient, thus supporting H1. However, the coefficient shrinks and turns insignificant in models II and III.

The economic ideology of social democratic parties is uncorrelated with trade unionist recruitment to ministerial office (H2). While the variable displays the expected movement over time (social democrats move towards the centre in the 1980s and especially 1990s, see Figure 3), it is not associated with changes in appointment patterns. Surprisingly, thus, SDs’ economic left–right position – at least as measured by manifesto data – appears to play no role in explaining the decline of unionist appointments. While the rise of Third Way politics coincides temporally with a decline in union-linked ministers, there is no statistical relationship between the two in our sample. The analyses reported in Table 3 thus suggest that there is no robust linear relationship between the electorates and ideological positions of SD parties and their likelihood of appointing trade unionists as ministers.

To test our conditional hypotheses (H3, H4, and H5), we estimate regression models that interact the share of workers among SD voters and parties’ economic left–right position with the binary statutory linkage indicator, union density, and the level of concertation. We include all control variables from model III (see Table 3), but we do not report them here (their coefficients are substantively identical to those reported in model III). Models IV to VI interact the share-of-workers variable with the three measures of institutional stabilisers. Two of these models yield the expected conditional effects, with negative and significant interaction terms: Thus, according to models V and VI, union-linked appointments are less likely when SD parties rely less on working-class voters, but this relationship is only present when union density is low or concertation is absent. The interaction with statutory linkage points in a substantively similar direction, yet the effect does not reach statistical significance ($p=0.21$).⁹

While there is thus good evidence for a conditional effect of electoral change, the interaction terms with ideology vary in their direction and do not reach statistical significance. Thus, the relationship between parties’ economic left–right positions and their ministerial appointments does not change substantially under different ­institutional conditions.

In order to see this more clearly, take a look at the predicted probabilities for high and low levels of the moderating variables in Figure 5. The top panels show that union-linked appointments are positively correlated with the share of workers in party electorates if union density is low or concertation is absent – thus if institutional stabilisers are not present (blue-shaded graphs). Under such circumstances, social democratic appointments respond much more strongly to the transformation of party electorates. More specifically, the probability of trade unionist appointments increases across the empirical range of the share-of-workers variable when union density is low (from 11 to 39%) and when concertation is not established (from 12 to 35%). Yet when institutional stabilisers are strong (high union density, full concertation), the association between SD electorates and union-linked appointments is actually negative, as indicated by the downward-sloping dashed red lines in the top three panels of Figure 5.

Figure 5. Predicted probability of trade unionist appointment (interaction effects). Note: Predicted probabilities and 95-percent confidence intervals; results based on Models IV–IX in Table 4; for ‘low’ and ‘high’ values, statutory linkage was held at 0 (‘no’) and 1 (‘yes’), concertation held at 0 (‘none’) and 2 (‘full’), union density held at the 10th and 90th percentile; other variables held at observed values.

By contrast, no strong differentiation according to institutional factors becomes visible for the association of economic ideology with union-linked appointments. Whether institutional stabilisers are present or not, economic ideology remains essentially uncorrelated with the appointment of trade unionists as ministers.

Discussion and conclusion

This article presents the most comprehensive analysis to date of how social democratic parties’ electoral and ideological transformations over the past decades affected their elite-level linkages with trade unions. Examining over 2,600 ministerial appointments between 1960 and 2014, we document a steep decline in the share of social democratic appointees with a trade union background. Whereas 30% of social democratic ministers had a union background in the 1960s, this proportion has declined to 16% in the recent past (Figure 1).

Yet how do the electoral middle-class shift and the ideological moderation that (some) SD parties underwent relate to this trend? We theorised that union-linked ministerial appointments become less likely as party electorates become less reliant on working-class voters and parties shift rightward on the economic left–right dimension. However, we also argue that the institutional framework in which party–union ties are embedded (statutory linkages, union density, concertation) affects how ministerial appointments respond to these changes. If the institutional environment is favourable to party–union linkages, we should see weaker effects of electoral and ideological change.

Our analysis yields very little support for our direct-effect hypotheses (H1 and H2). Electoral change and ideological shifts have no linear relationship with union-linked ministerial appointments. At least in the short term, these appointments are not dictated by electoral or party-ideological considerations. This may be due to the fact that ­ministerial positions are public rather than party offices, hence other considerations may come into play (e.g. expertise, experience, or communication skills).

By contrast, we find support for some of our interaction hypotheses. In particular, union density and concertation moderate the effect of electoral change on ministerial appointments: In the absence of these institutional stabilisers, the association between working-class voter shares and union-linked ministerial appointments is substantial. Institutional environments that are less conducive to strong party–union linkages thus display a strong correlation between the shape of party electorates and ministerial appointments. The same moderators, however, have no bearing on the relationship between party ideology and appointments.

The central conclusion from our findings is thus that the electoral and ideological transformation of social democracy in itself does not dictate the shape of party–union linkages. Economic centrism and low working-class support can co-exist with substantial personal linkages among SD and union elites. Likewise, ideological leftism and sizeable proportions of working-class voters do not guarantee union-linked ministerial appointments. As important as these factors may be, their impact, if any, is moderated by the institutional context in which party–union linkages are embedded.

One important limitation of our article is the fact that the causal relationships between key variables in our analysis are likely to be complex. For example, ministerial appointments may be a consequence of electoral and ideological change, but they may also drive these trends to some extent (Heath 2015; O’Grady 2019). In addition, the institutional moderators theorised in H3–5 can hardly be viewed as fully exogenous: Statutory linkages and levels of concertation – while much more stable in the long-term – may at times be affected by informal personal linkages (i.e. ministerial appointments).

While our observational research design thus does not allow for causal identification, we still consider it an important contribution. Earlier case studies found that party–union cooperation has changed and weakened, but remains important (Allern et al., 2007; Anthonsen et al., 2011; Aylott 2003; Thomas 2001). Hence, the absence of direct associations between union-linked appointments and ideological and electoral change of SD parties adds a new piece to the ever-changing puzzle of party–union relationship.

Overall, our findings thus contribute to a better understanding of the relationship between the political and the economic arm of the labour movement. Yet the explanations tested here are certainly not exhaustive. For example, the current article takes a party-centered perspective, yet there have been significant transformations of the trade union landscape that have, in some cases, triggered a re-evaluation by unions of their ties with political parties (Allern et al., 2007). A comprehensive explanation of the personal linkages between parties and unions would certainly need to take both actors’ calculations into account. What is more, it remains an open question to what extent the explanations offered in this article apply to party–interest group ties more broadly.

This point opens up important avenues for future research. Our article provides a first step in demonstrating the theoretical and empirical value of examining party–interest group ties through the prism of transforming party politics. Future research could apply this logic more broadly, for example by examining how changed patterns of party competition alter the recruitment strategies of political parties (Abou-Chadi and Wagner 2019; O’Grady 2019). How shifts in party–interest group relations affect political elite selection and feed into policy outputs and outcomes will thus be one of the most pertinent questions to answer for students of parties, interest groups, and policy making in post-industrial democracies.

Acknowledgements

The authors would like to thank Björn Bremer, Nicole Bolleyer, Elin Haugsgjerd Allern, and Florian Weiler for their valuable comments on earlier versions of the paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors gratefully acknowledge funding from the Anniversary Fund of the Austrian National Bank (OeNB), grant [no. 17886].

Notes on contributors

Mario T. Taschwer

Laurenz Ennser-Jedenastik is a political scientist at the University of Vienna. He studies political parties, political elites, political institutions and welfare states in Austria and in a comparative European perspective. [[email protected]]

Laurenz Ennser-Jedenastik

Mario T. Taschwer is a scholar of political science and international business and economics. He was a research associate and lecturer at the University of Vienna, Department of Government and he is interested in economic policy, social inequality and ice hockey. [[email protected]]

Verena Reidinger

Verena Reidinger is a PhD Candidate at the Department of Political Science, University of Zurich. Her research focuses on political representation, political trust and public opinion. [[email protected]]

Notes

1 The terms linkages and ties are used interchangeably in this article.

2 To examine whether this is true of our data, we extracted a random sample of 20% of all union-linked ministers (n = 54). 66% of these individuals ­either have working-class occupational experience or affiliations with working-class-focused unions (e.g. metal, textile, or transport workers).

3 This argument builds on the premise that, during our period of observation, trade unions in Western Europe prefer social democrats to adopt socio-economically left-wing positions, and that SD parties are more likely to deviate from unions’ ideal points by moving to the right than further to the left.

4 The most important reason for this choice is that mere union membership would not facilitate interaction between union and party leaders and can therefore not provide strong linkage between unions and parties. In addition, union membership among social democratic elites may at times ­approach 100%, leaving little room for meaningful variation.

5 For further information on the meta-category and the lists for all countries see ‘Wikipedia Category: Trade unionists by country’.

6 In excluding retirees, students or the main care-takers of the household we follow Gingrich and Häusermann (2015).

7 https://ec.europa.eu/economyfinance/ameco

8 Our results hold when using country-level fixed effects, see Tables 6 and 7 in the online appendix.

9 We present models in Tables 18 and 19 in the online appendix using the presence of statutory ties in 1960 as an indicator. These models return an interaction term with p = 0.107.