Forthcoming in British Journal of Industrial Relations

THE IMPORTANCE OF POLITICAL SYSTEMS FOR TRADE UNION MEMBERSHIP,

COVERAGE, AND INFLUENCE: THEORY AND COMPARATIVE EVIDENCE

John W. Budd
Center for Human Resources and Labor Studies
Carlson School of Management
University of Minnesota
Minneapolis, MN 55455 USA
jbudd@umn.edu

J. Ryan Lamare
School of Labor and Employment Relations
University of Illinois at Urbana-Champaign
504 E. Armory Ave.
Champaign, IL 61820 USA
rlamare@illinois.edu

Version Date: October 8, 2020

Abstract

Ideological connections between the state, political systems, and industrial relations have long
been important. But the influence of the structural nature of a country’s political system on trade
union membership, coverage, and influence has been largely overlooked. We uniquely theorize
three channels through which a political system can shape unionization in the workplace
independent of ideology: incentives for inclusionary governance, legislative body composition,
and policy enactment. Empirically, we use multiple European data sets to test the relationship
between political and employee representation using multivariate analyses across more than 25
countries. We find that increased political representativeness, measured by lower
disproportionality and the presence of multiparty coalitions, is a statistically significant predictor
of a greater likelihood of individual trade union membership, coverage, and influence, while
competitive fragmentation, measured by greater numbers of political parties, is associated with
weakened collective voice.

Acknowledgments: The authors thank Weihao Li for assisting with the data collection and
cleaning, and Lorenzo Frangi, Marco Hauptmeier, Miguel Martinez Lucio, Christine Riordan, and
participants at the LERA annual meeting, the ILERA World Congress, the Industrial Relations in
Europe conference, the CIRA annual conference, the Regulating for Decent Work conference, and
the SASE annual meeting for helpful comments.
 Scholars and practitioners have long recognized the important intersections between

industrial relations, politics, and the state. Industrial relations actors seek to influence the state

through supporting candidates for political office, lobbying, leading protests, participating in social

dialogue, and other means. At the same time, the state shapes industrial relations though various

roles—including as a public sector employer, a regulator of labor law, an overseer of

macroeconomic policy and the welfare state, and an arbiter of what constitutes acceptable

economic and social relationships (Traxler 1999; Hyman 2008; Godard 2017; Howell

forthcoming). Much of this is ultimately connected to specific state actions or policies, whether in

the form of laws, rulings, macroeconomic policies, public sector bargaining strategies, or the

acceptance or rejection of social partnership involvement in policy-making, but it can also involve

social norms and values (Lipset 1962; Godard 2002). We add to this by questioning the importance

of a country’s political system for industrial relations. More specifically, we focus on a country’s

electoral system and the presence of multiparty coalitions as key parts of a country’s political

system that may affect trade union membership, coverage, and influence by shaping preferences

for social partnership and creating values that shape “models of behavior” that become

“institutionalized within varying social structures” (Lipset 1962: 97).

Broadly speaking, by making political systems, commonly explored by political scientists,

a key factor in theorizing employment relations institutions and outcomes our ambition is to

expand upon the well-established frameworks for understanding the relationships between

political economies and industrial relations, such as varieties of capitalism (Hall and Soskice

(2001) and other analyses of industrial relations across political economies (Crouch 1993; Traxler

and Kittel 2000; Hyman 2001; Baccaro and Howell 2017). This is consistent with calls for

supplementing varieties of capitalism theorizing with additional ways of understanding key trends

1
and outcomes in comparative industrial relations (McLaughlin and Wright 2018). By focusing on

union membership, coverage, and influence, we also complement work such as Hamann and Kelly

(2010) and Hamann et al. (2015) that explores the influence of electoral systems on the use of

peak-level social pacts.

A country’s electoral system is a significant part of its political system, and political science

has a rich history of studying both the determinants and effects of these systems. Political scientists

commonly identify two key dimensions to compare electoral systems across countries. The first is

the extent to which a national electoral system yields a legislative body that is proportional to the

fraction of votes each party received, and is often measured using a country’s level of

disproportionality, which indicates deviations from the baseline of perfect proportionality or

representativeness (Gallagher 1991). The second is the effective number of parties within the

country’s political system, which adjusts the actual number of parties to account for the relative

strength of each party, usually measured at the electoral or parliamentary level (Laakso and

Taagepera 1979). A third component of political systems that has received significant attention

from political scientists, and frequently operates tangential to comparative electoral systems

scholarship, is the presence of multiparty coalitions. Coalition cabinets are common in Western

Europe and a robust literature has emerged into how these entities form, the degree of bargaining

that occurs within them, and the policies they create (Riker 1962; Laver and Schofield 1998).

Our research engages with this literature to ask two key questions: 1) theoretically, how

might these dimensions of a country’s political system affect trade union membership, coverage,

and influence, and 2) empirically, is there evidence of these relationships. We seek to make unique

theoretical and empirical contributions by presenting a theoretical framework in which a country’s

political system is predicted to affect unionization beyond factors related to a country’s political

2
economy that are already well-recognized (such as legislative ideology, the country’s variety of

capitalism, and Ghent systems of union-run employment insurance). Theoretically, we develop

multiple ways in which a country’s political system might shape unionization, including enacting

public policies, involving trade unions in peak-level corporatist initiatives, enabling direct

relationships between trade unions and legislators, shaping attitudes around political inclusion that

affect workplace agency, and giving social legitimacy to collective voice. Through all of these

channels, a political system that is more inclusive or representative is generally expected to have

stronger trade union membership, coverage, and influence. By contrast, weaker workplace

unionization is likely to emerge under a fragmented political system in which parties behave

competitively rather than cooperatively toward each other.

Empirically, we use multiple well-established European data sets to test the relationship

between political systems and union membership, coverage, and influence using rigorous,

multivariate analyses across more than 25 countries: 1) two waves of the European Company

Survey (containing measures of trade union presence and collective agreement coverage in the

workplace), 2) Round 5 of the European Social Survey (ESS) (containing measures of the presence

and influence of trade unions in the workplace), and 3) all 8 rounds of the ESS pooled together (to

analyze individual trade union membership). Taken as a whole then, we analyze multiple

dimensions of unionization using establishment- and individual-level data.

Disproportionality is a clear-cut (inverse) measure of political representativeness and

across all three data sets, we find supportive results that a lower level of disproportionality (i.e., a

higher level of political representativeness) is a statistically significant predictor of a greater

likelihood of unionization. If this relationship is causal, it likely runs from the electoral to the

workplace level given the limited contemporary channels for unionization to affect

3
disproportionality. Multiparty coalitions typically reflect a political system that accepts and values

bargaining, and we therefore expect that coalition-based systems will correlate with higher levels

of unionization, although the causal direction of this relationship is less clear. Finally, the number

of effective political parties can capture the extent of representative voice to diverse constituencies

or it can indicate competitive fragmentation within a political system that creates extremism,

bargaining complexities, and gridlock (Cox 1997). Across all three data sets, the influence of the

number of political parties on workplace representation is always statistically significant and

negative, consistent with competitive fragmentation. Taken together, we believe that this

significantly adds to the literature on unionization and comparative industrial relations both

theoretically and empirically, and points to broader relationships between the state, political

systems, and industrial relations than are typically recognized.

Three Key Dimensions of a Political System

Political scientists have long puzzled over the design of political systems and their

implications for societies. There exists no single ideal type of political system, so the choice of

any given system must inevitably face tradeoffs (Carey and Hix 2011). Namely, should the system

be inclusive (where political power is broadly shared), aiming to most accurately reflect the

pluralistic array of political interests within a society and attempting to give at least some voice to

those interests? If so, societies should trend toward allocating political power through proportional

representation (Lijphart 1994, Powell 2000). Or, should the system be exclusive (where political

power is deliberately concentrated), aiming to hold politicians accountable to voters while

punishing those parties or individuals whose policies stray too far from what the majority of the

electorate demands? If so, societies should trend toward allocating political power on a

majoritarian basis (Taagepera and Shugart 1989; Cox 1997). These tradeoffs can be thought to

4
distinguish between systems designed to create governing bodies that are a microcosm of the

society in which they inhibit, versus those designed to create agents that make decisions on behalf

of the median voter (Farrell 2001).

In practice, a continuum exists between strictly majoritarian and strictly proportional

representation, with electoral systems differing considerably depending on their specific design

characteristics and construction, which may vary across countries and time (Gallagher 1991).

Consequently, every legislative election results in a unique degree of proportionality when an

election’s votes are allocated to seats by these design characteristics. A standard measure of

(inverse) proportionality is disproportionality, which measures deviations from the baseline of

perfect proportionality in which each party’s vote share equals its share of legislative seats. Larger

values of an electoral system’s disproportionality indicate larger deviations from perfect

proportionality and reflect situations that have less representativeness. The 2019 United Kingdom

general election graphically illustrates the relationship between disproportionality and

representativeness: the Conservative Party received 56 percent of the seats while winning only 44

percent of the vote, while the Liberal Democrats received less than 2 percent of the seats despite

winning 11 percent of the vote. We follow the political science literature by using

disproportionality to measure (inverse) representativeness in the empirical analyses, and therefore

we also use the concept of disproportionality in our theory-building below.

Those who study electoral systems also emphasize the effective number of parties found

within a system as a key second dimension. This dimension is defined as being not simply a count

of the number of parties in a system but rather “the number of hypothetical equal-size parties that

would have the same total effect on fractionalization of the system as have the actual parties of

unequal size” (Laakso and Taagepera 1979: 4, emphases in original). Unlike the proportionality-

5
disproportionality continuum in which greater proportionality is always seen as indicating greater

representativeness, however, the continuum from few to many parties has competing

interpretations. A common argument is that greater numbers of candidates or parties competitively

fragments electoral systems by either creating too many options for voters, leading to wasted votes,

or giving governance opportunities to marginal and extreme small parties (Cox 1997). However,

a greater number of parties can also reflect new parties emerging to represent the diverse points of

view of previously unrepresented constituencies (Clark and Golder 2006). Taken together, these

perspectives result in ambiguous implications for theorizing the role of the number of parties.

A third institutional feature of political systems is whether governance is conducted

through multiparty coalitions or by a single party. When multiparty ruling coalitions are present in

a political system, this indicates at least some degree of bargained compromise between parties in

order to secure governing power (Strom, Muller, and Bergman 2008). Political scientists generally

treat the presence or absence of multiparty coalitions as tangential to electoral systems, rather than

subsuming coalitions into classic electoral systems measures like disproportionality and the

effective number of parties (Riker 1962; Laver and Schofield 1998). But we expect that the

presence of multiparty coalitions could have implications for union membership, coverage, and

influence that operate independent from other electoral systems effects so we add this to

disproportionality and the effective number of parties in our theoretical and empirical analyses. 1

1
We treat each of the three dimensions independently though they might at times interact within
a political system. A full account of the complexity of the interactive relationships between these
dimensions lies beyond the scope of this paper. Moreover, in political science scholarship it is
common to empirically model the dimensions as separate independent variables (Curini and Hino
2012; Matakos, Troumpounis, and Xefteris 2016) and this produces a conservative test of their
individual effects if they are somewhat related because we will find statistically significant
relationships only when there are main, independent effects.
6
 Theorizing the Influence of Political Systems on Unionization

In theorizing how political systems that differ in their degree of disproportionality, the

number of effective parties, and the presence of coalitions might influence union membership,

coverage, and influence, we identify three categories of linkages: those connected to 1) incentives

for inclusionary governance, 2) the composition of the legislative body, and 3) the nature of

policies that are enacted. There has been some theorizing on the influence of aspects of political

systems on peak-level industrial relations. In particular, Hamann and Kelly (2010) and Hamann et

al. (2015) argue that electoral systems shape lawmakers’ preferences for using social pacts as a

shield against not being reelected after the passage of unpopular reforms. But we believe we are

the first to focus on workplace-level representative voice. And although we are unable to directly

examine each of the theorized linkages, we believe that, taken together, they capture linkages with

enough plausibility and diversity to motivate empirical analyses and spur additional research on

variations between electoral systems types and forms of workplace representation.

Before turning to each category, it is important to consider where ideology fits within our

theory. While proportional representation systems generally result in national legislative bodies

with low disproportionality and a large number of parties, they also tend to yield center-left

governing coalitions (Iversen and Soskice 2006). In this way, more representative political systems

would be theorized to be associated with greater union membership, coverage, and influence

because left-leaning governing coalitions will be supportive of the labor movement. This type of

ideological connection between the political and employment relations arenas is well-known in

employment relations scholarship (Johansson and Raunio 2001; Hyman and Gumbrell-

McCormick 2010). We recognize that ideology is an important contributor to any theoretical

framework linking politics and employment relations, but one way in which our contribution is

7
novel is that we theorize ways that a political system can affect the workplace, holding constant

the ideology of the ruling party or coalition and of the legislative body as a whole.

Our proposed linkages are based on a general premise derived from the political science

literature: that spillovers from the political system into wider society are possible. Research

connecting political systems and mass public behaviors shows that these systems commonly grant

political elites the power to formulate their own positions on issues rather than simply mimicking

mass majority attitudes (Sartori 1987). Political elites can then frame policy issues in ways that

shape mass public opinion and attitudes (Druckman and Jacobs 2015). The result is that individual

citizens may form at least some of their beliefs on the basis of elite cues (Converse 1964; Zaller

1992). Within the European context, the assumption that political elites can help to shape wider

public attitudes, opinions, and behaviors been applied to policy issues like EU integration (Gabel

1998) and referendums (Hobolt 2007), as well as immigration (McLaren 2001) and redistribution

policies (Bechtel, Hainmueller, and Margalit 2014). Our theorizing does not focus on elite cues on

specific policies, but this literature is supportive of a generalized assumption that spillovers are

possible from the political system into the wider civic environment, which we use to articulate

several specific possible linkages to the workplace. Generally, then, just as the state can establish

values and norms for industrial relations (Lipset 1962; Godard 2002), our premise is that political

systems also contribute to the values and norms of industrial relations.

This approach is also supported by scholarship on institutional logics in sociology.

Institutional logics are bundles of beliefs and practices based in assumptions, norms, values, and

understandings that “provide the formal and informal rules of action, interaction, and interpretation

that guide and constrain decision makers” (Thornton and Ocasio 1999: 804). Institutional logics

exist at many levels within society, and it is common to theorize that institutional logics around

8
markets and competition shape organizational actions, including workplace representation

strategies (Pernicka and Glassner 2014). But reminiscent of Lipset (1962), state-oriented logics

can also shape and frame how actors in organizations understand their roles and acceptable actions

(Greenwood et al. 2010). We posit that a political system, too, contributes toward and is part of

institutional logics, especially relating to compromise/competition and inclusion/exclusion, that

affect organizations and workplaces.

Incentives for Inclusionary Governance

The first theoretical channel from a political system to workplace voice is rooted in the

distribution and concentration of political power. In more representative and coalitional political

systems, it is difficult for any single party to govern or have outsized influence. In such systems,

then, inclusionary governance arrangements are often required to attain political power, whereas

in less representative and unilateral political systems, parties can be exclusionary in their

governance (Baccaro and Lim 2007; Blais and Bodet 2006). Under more representative and

coalitional systems, unions have greater opportunities to use their extra-parliamentary status to act

as consensus-builders, thus enhancing their value.

Such systems are known to produce policy deadlocks since parties share power but hold

contrasting values, interests, and perspectives (Martin and Vanberg 2004). This sometimes

requires governments to turn to unions to help negotiate consensus-based policies. For instance, in

Austria a right-leaning coalition under Wolfgang Schüssel in the early-2000s was unable to agree

on labor policy reforms. So, the coalition took advantage of a legacy of institutional corporatism

in deferring to social partnerships between unions, employers, and parties in order to enact the

reforms (Traxler 2004; Rathgeb 2018a, 2018b). In contrast, unions and similar entities are less

relied upon when single parties hold most or all political power. For example, in Sweden, the left-

9
leaning Social Democratic Party (SAP) was strongly supported by unions in the mid-1990s but

Sweden’s political system at the time gave it hegemonic policymaking power. It used this power

to unilaterally reform job security and unemployment insurance policies while excluding unions

from the process (Rathgeb 2018a, 2018b). Consequently, we maintain that under more

representative and coalitional systems, unions are likely to have greater legitimacy and influence

in the national political arena, which can spill over into greater legitimacy and influence for unions

in the workplace. The spillover may occur since unions may be likely to be perceived positively

by the public for the role they play in breaking policy deadlocks, and this can happen when the

ideology of the legislature or ruling coalition is either right- or left-leaning. 2

Moreover, because it is difficult for any single party to govern in representative and

coalitional electoral systems, employers have an incentive to seek influence as a social partner

rather than relying on influence within specific political parties (Cusack, Iversen, and Soskice

2007; Martin and Swank 2008). This in turn requires the business community to work with, rather

than against, the labor movement. In certain types of political systems, then, there is a greater

incentive for employers to embrace a culture of inclusion and compromise that recognizes rather

than attacks the legitimacy of labor unions. In this way, we submit that unions and collective

bargaining are less likely to be marginalized when a political system has a higher level of

representativeness, and consequently, unions may have greater workplace influence and workers

2
Theorizing that focuses solely on ideological linkages would predict that the Schüssel coalition
would be hostile to the labor movement whereas Sweden’s SAP would be supportive. This
certainly could have been true in other ways and have been reflected in specific policy initiatives,
but with respect to the influence of the electoral system, there appears to be an influence
unconnected to ideology.
10
are more likely to join. 3 For example, when Denmark introduced a system of proportional

representation in 1920, the country’s employers’ associations viewed cooperation with labor

groups as their best hope to promote industrial peace and avoid radicalism, purely out of tactical

necessity rather than ideological beliefs. Conversely, in the United States, the National Association

of Manufacturers (NAM) initially supported negotiating with labor unions, but the highly

disproportional US electoral system allowed NAM to become entirely aligned with the Republican

Party rather than requiring it to diversify its political activities or cooperate with unions to enact

its preferred policies (Martin and Swank 2008).

Lastly, the distribution of power under different political systems yields varying degrees

of clarity of responsibility (Hamann and Kelly 2010; Hamann et al. 2015). In systems where one

party governs and in those with lower representativeness, there can be a high degree of

accountability because voters can ascribe responsibility to the ruling party, whereas in systems

with lower disproportionality, consensus building can be more important than clarity and

accountability (Powell 2000). Under representative systems, governments looking to enact

unpopular reforms often include unions in policymaking as a way of shielding the government

from punishment by voters, taking advantage of the clouded accountability these systems offer.

This may create a culture of consensus building between governments and unions. Such was the

case in Sweden after a 1998 election took hegemony away from SAP and reduced its governing

status to a minority government. In contrast to prior unilateral reforms, SAP was then required to

emphasize the significance of unions as consensus-based social partners when enacting unpopular

austerity reforms to labor programs, rather than forcing the policy changes through legislatively

3
Moreover, a spirit of inclusion, trust, and compromise among social partners (Martin and Swank
2004; Roche 2007) can set a tone of inclusion and compromise rather than conflict and repression
in the workplace, potentially leading to greater union membership, coverage, and influence.
11
(Hamann and Kelly 2010). Assuming that elite cues supporting consensus-based decision-making

attitudinally spill down into the workplace, systems having lower disproportionality will be

associated with greater workplace support for bargained consensus, and therefore with higher

levels of collective voice in the workplace.

Our theory linking the effective number of parties in a political system to workplace voice

through inclusionary governance is more ambiguous. If parties give voice to previously

unrepresented constituencies in society, a more representative system allows for a wider range of

voices to be included in the decision-making process. This may spill over positively for unions,

since labor-management relations can sometimes behave similar to political governance through

pluralism and can allow for the representation of an array of interests at the bargaining table. But

because the structural nature of a competitively fragmented system may also reward extreme

political groups and allow these smaller parties to compete for relatively outsized power, this may

make creating a spirit of consensus-building and compromise difficult, which could result in

spillovers into organizations and workplaces that have negative implications for union

membership, coverage, and influence.

Composition of the Legislative Body

The second theoretical linkage between a political system and workplace voice that we

develop focuses on the makeup of the legislative body. Specifically, an electoral system with lower

disproportionality and a larger number of parties, as well as one featuring multiparty governance,

will consist of a national legislative body that has a different composition than one in an electoral

system with higher disproportionality, fewer parties, and one-party rule (Martin and Vanberg

2004; Lublin and Bowler 2017). Even holding ideology constant, with a greater diversity of parties

in the former case, there is a higher chance that the legislative body will have at least some

12
members who support trade unions, including legislators who are themselves union members

(Carnes 2012; Sojourner 2013), and a greater likelihood that labor-friendly politicians will occupy

at least some key positions such as ministers, shadow ministers, or members of important

committees. This representation can give unions a source of legitimacy and visibility in the

national consciousness, and it is possible that this gives unions legitimacy in the workplace, too.

In contrast, when there is a lack of worker-friendly legislators, unions can be more easily

marginalized in the public imagination either simply because of their absence (out of sight, out of

mind), or because no one is able or willing to counter or prevent rhetorical attacks. Research in

Europe finds that when political elites express unchallenged exclusionary attitudes toward Muslim

immigrants, mass public attitudes are similarly hostile (Czymara, forthcoming). Similarly, in the

winner-take-all majoritarian US Republican primary campaign of 2015, when then-candidate

Donald Trump labeled Mexican immigrants “rapists” and “criminals,” his views were

unchallenged by most other candidates, and these elite cues led to increases in anti-immigrant

attitudes among many Americans (Flores 2018). When there is high disproportionality and/or a

small number of parties in the legislative body, it is more likely that politicians’ anti-union

statements will go unchecked. This, in turn, can undermine union legitimacy and create negative

public attitudes toward unions, leading to lower levels of union membership and influence.

Compare this to a situation where the low disproportionality and/or a large number of parties in

the legislative body but the same overall ideology. In this case, it is less likely that anti-union

statements would go unchallenged—either because low disproportionality means that politicians

are not as tied to the median voter or because more parties means more voices—and thus there

might be less of an effect on public attitudes. Lastly, when there are multiple parties that might

strive for union support (compared to a legislative body with the same overall ideology but with

13
fewer parties), this could make it less likely that the labor movement is dependent on one party

while also creating a competition among these parties for this support, which could provide greater

rhetorical support for the labor movement in public discourse (Hamann, Johnston, and Kelly

2013).

Policy Enactment

Inclusionary governance and compositional differences can lead to the third category of

linkages from a country’s political system to the workplace: the enactment of policies that lead to

greater trade union influence, coverage, and membership. The most obvious linkage here might be

an ideological connection in which left-leaning governing parties enact policies supportive of

unions and workers (Iversen and Soskice 2006). But we believe it is reasonable to theorize

additional linkages that are separate from ideological explanations.

First, a legislative body with a greater diversity of members might enact policies that reflect

a wider variety of views and interests, because a wider variety of issues would have at least some

members as champions (Blais 1991; Karp and Banducci 2008). So even without a labor-friendly

government in power, there could be a greater chance of labor-friendly legislation in a political

system that has a higher level of representativeness or one that operates as a multiparty coalition.

Relatedly, there could be a greater likelihood of labor-friendly legislation even without a labor-

friendly government because of a greater chance of horse-trading support for different issues when

the legislative body is more diverse. This logic also applies to multiparty coalitions, where

bargained arrangements between different groups are often required to form government and to

create policies (Martin and Vanberg 2004).

Second, when right-leaning governments are in power, a greater diversity of legislative

members or the existence of a multiparty coalition could be more likely to include a voice against

14
repressive policies that would reduce union influence, coverage, and membership. Third, with a

greater diversity of views in a legislative body, more creative and innovative public policies could

result (Orellana 2010), and some of these could be labor-friendly. Yet it is also important to

recognize that the labor-friendly policies that may result from more representative political

systems are unlikely to emerge if these systems engender competition rather than compromise.

Such systems could result in policy stagnation and may be more likely to occur as the number of

parties increases because of uncertainty and mistrust between parties (Timmermans 2006).

Lastly, the importance of all of these channels could be magnified for public sector unions

and workers because the state is the ultimate arbiter of public sector terms and conditions of

employment. So, for example, just as political systems with a greater diversity of members might

be more likely to enact labor-friendly legislation, the same would be true of setting employment

relations policies for public sector workers and unions. Similarly, any incentives for including

private sector unions in social partnerships would apply to public sector unions, too. These could

then directly affect public sector union membership, coverage, and influence, and also indirectly

influence the private sector when these state policies are seen as being a norm-setting exemplar

for the private sector (Godard 2002). These effects may be particularly pronounced under

multiparty coalitions, which have been shown to both produce large public sectors and to spend

considerably on the public sector (Martin and Vanberg 2013).

What about Reverse Causality?

The theoretical channels we have outlined all involve a causal chain from the electoral level

to the workplace level. But what if this relationship is endogenous such that unionization is

affecting disproportionality, the effective number of parties, and the emergence of multiparty

coalitions, rather than the reverse? The presence or strength of trade unions might affect the

15
effective number of parties or the emergence of multiparty coalitions if, for example, unions create

additional political parties or help broker the emergence of a governing coalition. Individual union

membership could be seen as affecting the number of political parties and their presence in

government if union members develop the skills or agency to become more engaged in the political

process (D’Art and Turner 2007; Bryson et al. 2014), leading to support for a greater number of

political parties or a greater likelihood of their being able to bargain to share political power.

In contrast, it is much more difficult to see how unionization can determine

disproportionality. To the extent that disproportionality is determined by electoral rules, unions or

other groups could campaign for changing these rules, but in practice electoral rules that affect

representativeness rarely change (Matakos, Troumpounis, and Xefteris 2016). For instance, in

Europe from 1945 to 2011, only 71 instances of electoral reform led to significant changes in

disproportionality. Just eight of these instances happened from 2002 onward, and six of the eight

involved political elites imposing reforms on the public rather than public-driven reforms

(Renwick 2011). Beyond the exceptionally rare possibility that unions strategically pushed for

electoral reforms during our study timeframe, we are unable to identify contemporary mechanisms

by which workplace voice would affect disproportionality. 4

4
Historically, struggles for political and workplace rights for the working class were closely
intertwined (Lipset 1983; Crouch 1993; Ebbinghaus 1995), so with the emergence of modern
democracies in the 19th and early 20th centuries, the character of the labor movement could have
impacted the nature of the electoral system, or unique factors could have impacted both. If these
founding conditions are enduring enough to manifest themselves in a correlation between union
membership and electoral systems 150 years later, then perhaps our results are remarkable for a
different reason. And our results that include country fixed effects control for unique founding
conditions within a country.
16
 Data

We conduct three sets of analyses using data from two waves of the establishment-based

European Company Survey and from eight rounds of the individual-level European Social Survey

(ESS). As will be described, each data set spans more than 20 countries. To each data set, we

merge political systems measures indicating the country’s level of disproportionality, number of

effective parties, and presence of multiparty coalitions at the date of the interview.

European Company Survey Data

The European Company Surveys (ECS) are conducted by the European Foundation for the

Improvement of Living and Working Conditions (Eurofound) by surveying the manager

responsible for human resources in the establishment, and when possible an employee

representative, across a large number of establishments (European Foundation for the

Improvement of Living and Working Conditions 2010, 2015). We use the data collected from the

management representative in the 2009 and 2013 waves because these provide measures of

workplace union presence. To use the fullest extent of control variables available in each wave,

we analyze the two waves separately. As shown in columns 1 and 2 of Table 1, we are able to

analyze establishments across 29 and 30 countries, respectively, after merging the 2009 and 2013

waves with the electoral system measures.

For the two waves, we are able to construct two dichotomous measures that indicate

whether or not there is trade union representation and whether at least some employees are covered

by a collective wage agreement. The former is not exclusive, meaning that there might also be

forms of nonunion representation present; the latter includes agreements negotiated at the

establishment, sectoral, occupational, and national levels. The sample sizes are in excess of 20,000

establishments for each wave, but vary across these measures because some representation forms

17
are not part of a country’s industrial relations system, so managers in some countries where not

asked about a particular form of representation. 5

For each wave, we also constructed control variables including major industry,

organizational structure, private or public sector, the number of employees, whether or not there

has been a recent organizational change, and demographic characteristics of the establishment’s

workforce. Given how the questions were asked, all are categorical measures. We also create an

indicator for Ghent system countries in which some benefits are administered by trade unions

rather than the state (Ebbinghaus, Göbel, and Koos 2011). There is strong evidence that union

density increases when individuals are incentivized to join unions in order to receive

unemployment insurance, and union decline in some countries can be at least partly explained by

the substitution of union-administered systems with state-administered ones (Böckerman and

Uusitalo 2006). Finally, to account for the well-documented effects of varieties of capitalism on

industrial relations (Hamann and Kelly 2008) we use Schneider and Paunescu (2012) to create

controls for the national context by grouping countries into varieties of capitalism categories:

liberal market economy, liberal market economy-like, coordinated market economy, hybrid, and

state-dominated. Panel B in Table 2 shows the control variables available for the two waves.

European Social Survey Data

The European Social Survey is a cross-sectional survey administered to individuals by

national research councils across numerous European countries using consistent survey

5
In 2009, there is no option for workplace-based union representation in Germany, Austria, Spain,
and Luxembourg, and in 2013 the same is true of Germany, Austria, Netherlands, and
Luxembourg. Because we know there are union members in some workplaces in these countries,
we treat this variable as missing for these countries rather than coding all of them as not having
unions in the workplace. Note that Montenegro and Macedonia are excluded from the data we
analyze because of a lack of electoral systems measures.
18
specifications. It has been administered every other year starting in 2002. The original data are

representative of all persons aged 15 and over living within private households, but we limit our

analyses to workers by excluding those who do not report that their main activity in the previous

seven days was working and by excluding the self-employed. Due to differential question

availability across rounds of the European Social Survey, we conduct two separate analyses of

extracts of workers.

We first focus on Round 5 (ESS5), which was primarily collected in 2010 (European Social

Survey 2010). In this round we can identify whether there is a trade union in the workplace, and

its degree of influence. Specifically, Round 5 uniquely asks “How much influence would you say

that trade unions at your workplace generally have over decisions that affect your working

conditions and practices?” [G44]. From the response option “No trade unions/trade union members

at workplace,” we construct a dichotomous variable indicating if there are union members in the

workplace. From the fuller set of response options, we also construct a 4-point scale measuring

the degree of trade union influence in the workplace, ranging from 1=not much or no influence

(including no unions present) to 4=a great deal of influence. 6 After merging with the political

systems measures, we have a sample of 15,899 workers to analyze, spanning the 25 countries as

shown in the Round 5 column of Table 1.

Second, we analyze a pooled sample of workers across all 8 rounds of the European Social

Survey which were primarily collected between 2002 and 2016 (European Social Survey 2018).

Our primary variable of interest is whether or not an individual is a member of a trade union. This

6
In the first European Social Survey round, there are also questions about trade unions in the
workplace. Creating an indicator for whether or not there is a trade union present, and estimating
multivariate probit models yields similar results to those we present using Round 5. The questions
are worded more directly in Round 5, and there are also a larger number of countries represented,
so we focus on that round.
19
is constructed from the question “Are you or have you ever been a member of a trade union or

similar organisation?” which is asked in each round. We code those who respond “yes currently”

as a trade union member, and all others coded as not a member. Note that this is not a panel data

set; rather, new individuals are interviewed in each round, and we pool them across countries and

years. We exclude a small number of countries that lack electoral measures. As shown in Table 1,

the resulting number of countries in each round ranges from 21 to 27. The overall sample size is

144,507 workers.

In our analyses of European Social Survey data we also include control variables for

demographic, job, and organizational characteristics, including age, gender, years of education,

has children at home, is a citizen of that country, is an ethnic minority in that country, urban

location, is a supervisor, major occupation, major industry, type of employer (the Round 5 analyses

only), and employer size. As with the European Company Survey data, we also create indicators

for Ghent systems as well as Schneider and Paunescu’s (2012) varieties of capitalism categories:

liberal market economy, liberal market economy-like, coordinated market economy, hybrid, and

state-dominated.

Political Systems Measures

To the establishment-level European Company Survey data and the individual-level

European Social Survey data, we merge political system measures using the interview date for

each establishment or worker. As noted earlier, we draw on the political science literature on

comparative electoral systems and multiparty coalitions (Lijphart 1994; Farrell 2001; Strom,

Muller, and Bergman 2008) and use a country’s level of disproportionality and its effective number

of political parties at the level of votes to measure electoral system differences. These measures

are from Gallagher (2019), and we merge them by country based on the most recent national

20
election prior to each observation’s interview date. Consequently, our analysis samples include the

disproportionality score and number of effective parties for the national legislative body at the time

of interview. The disproportionality score ranges between 0.72 (Denmark in 2008 and 2010) and

21.95 (France in 2002). Recall that perfect proportionality equates to a disproportionality score of

0. The number of effective political parties ranges from 2.08 (Malta in 2009 and 2013) to 10.28

(Belgium in 2002). The presence of a multiparty coalition within a political system is dichotomous

and is based on the cabinet situation within a given country when each interview occurred. The

three political measures are not highly collinear with each other. For instance, within the pooled

ESS data, the correlation between disproportionality and both multiparty coalitions and effective

number of parties is -0.21, and the correlation between multiparty coalitions and effective number

of parties is 0.45.

Since we seek to explain the effects of political systems independent of ideology, it is

important to identify and specify how we have incorporated ideology as a key control. We start

with the ideology score for each political party in a country’s legislative body using the ParlGov

database (Döring and Manow 2018). This score uses the 0 (far-left) to 10 (far-right) ideological

scale that is widely used in political science research and has high validity and reliability (Kroh

2007). To control for the ideology of a country’s legislative body that resulted from a given

election, we construct an overall ideology score by weighting each party’s score by its legislative

seat share. Separately, to control for governing ideology, we also include the left-right ideology

score for the ruling cabinet, which is either (a) the ideology score for the ruling party in single-

party cabinets, or (b) an average of the ideology scores for all parties in the ruling cabinet weighted

by seat share if the cabinet is a multiparty coalition. This allows us to account for two discrete

components of political system ideology: that of the specific ruling party (or parties) and the

21
general ideological makeup of the legislative body as whole. And to allow for more flexibility in

the relationship between these ideology measures and our dependent variables, we also include the

squares of these two measures in our models.

European Company Survey Results: Trade Union Presence and Coverage

In analyzing the relationship between unionization, disproportionality, the effective

number of parties, and multiparty ruling coalitions in the 2009 and 2013 European Company

Surveys, we use indicators for whether the following are present in the establishment as reported

by management: (1) trade union representation, and (2) a collective wage agreement that applies

to at least some employees in the establishment. Panel A of Table 2 presents the sample means for

these measures. Slightly less than 35 percent report trade union representation in the workplace

while a collective bargaining agreement is present in nearly 65 percent.

Each of these indicators are used as the dependent variable in a probit model where the key

independent variables are disproportionality, the number of effective parties, and an indicator for

a coalition government. Sample statistics for these three measures are shown in Panel A of Table

3, and full sample statistics are provided in Appendix Table 1. To control for other potential factors

that influence the likelihood of representative workplace voice, the probit models also include the

control variables listed in Panel B of Table 3 (including the two ideology measures described above

and their squared values as well as varieties of capitalism indicators and organizational

characteristics). Returning to Table 2, Panel B reports the marginal effects and standard errors for

disproportionality, the number of effective parties, and coalition government. Throughout the

paper, marginal effects are calculated as the average of each observation’s predicted probability

change from a one-unit increase in the relevant independent variable, based on each observation’s

22
actual values for the variables included in that specification; standard errors are robust to arbitrary

forms of heteroscedasticity. The full results are provided in Appendix Table 2.

First consider the probit results reported in columns 1, 3, 5, and 7. Disproportionality is

statistically significant at the one percent level and is negative in each case. This indicates a

negative relationship between disproportionality and union presence and coverage, holding

legislative and cabinet ideology, varieties of capitalism, and organizational characteristics

constant. Recall that this means that electoral systems that are more representative (that is, a lower

disproportionality score) are associated with a greater likelihood of finding trade unions and

collective agreements in the workplace. The marginal effect of -0.015 in column 1 means that a

one unit increase in disproportionality reduces the probability of a trade union in the workplace by

0.015 percentage points, relative to an overall mean of 34 percent. Switching from the least to the

greatest disproportional country (an increase of 17 in the disproportionality score) implies an

increase in the probability of a union in the workplace by almost 14 percentage points.

In columns 1, 3, 5, and 7, the effective number of parties is also statistically significant and

negative at the one percent level. In contrast to the results for disproportionality, a negative

coefficient here implies that a larger number of parties is associated with a lower likelihood of

finding trade unions and collective agreements in the workplace. This negative relationship is

consistent with the competitive fragmentation aspect of larger numbers of political parties in an

electoral system. The results for the coalition indicator variable in columns 1, 3, 5, and 7 lack a

consistent pattern.

These models in columns 1, 3, 5, and 7 of Table 2 include a large number of control

variables, but do not control for Ghent countries. The models reported in columns 2, 4, 6, and 8

add the Ghent control variable to each specification. The Ghent variable is highly statistically

23
significant in each case, and in most cases is very large in magnitude, too. The Ghent variable also

overrides the importance of disproportionality in three of the four specifications. One interpretation

of this is that the significantly negative disproportionality results in columns 1, 3, and 7 are

spurious and may reflect a failure to control for the complex relationship between Ghent systems

and unionization (Rasmussen and Pontusson 2018). An alternative interpretation is that the Ghent

influence accounts for such a large amount of variance that it is difficult to identify weaker, but

still important, and possibly correlated, relationships. With this in mind, note that the results we

will present for the European Social Survey are robust to the inclusion of the Ghent control, and

the effective number of parties variable remains significantly negative in Table 2 even when

controlling for Ghent countries. It is also the case that our results do not appear driven by countries

with extreme values of disproportionality or effective number of parties. 7

European Social Survey Results: Trade Union Presence and Influence

We can use data from Round 5 of the European Social Survey to conduct a similar analysis

for the presence of trade unions in a workplace. These data span a similar, albeit not identical range

of countries (see Table 1), but are individual-based rather than establishment-based. Column 1 of

Table 4 presents descriptive statistics for the dependent variables and key independent variables.

In the sample we analyze from this round of the European Social Survey, 79.2 percent of workers

indicate that there are trade union members in their workplace. Column 1 of Table 5 reports the

7
When considering extreme values of disproportionality and effective number of parties, it is first
important to remember that even if extreme, the values are valid. Second, these two measures are
right-skewed so we focus our attention on large extreme values. With this in mind, we re-estimated
all of our European Company Survey and European Social Survey regression models in two ways:
(a) by excluding all observations for which disproportionality or effective number of parties exceed
the 95th percentile value for that measure in that model, and (b) excluding all observations for
which disproportionality or effective number of parties are more than two standard deviations
above the mean for that measure in that model. Most of the results are unchanged; in a couple of
instances they are weaker, in a couple they are stronger.
24
key estimated marginal effects from a multivariate probit analysis in which this indicator is the

dependent variable. This probit model also controls for ideology, Ghent systems, and varieties of

capitalism in the same way as in the European Company Survey analyses, and also includes

controls to account for key demographic and job characteristics (for sample statistics and marginal

effects for all of the variables, see Appendix Table 3). Both disproportionality and the effective

number of parties are negative predictors of the likelihood of trade unions in the workplace, and

these estimates are statistically significant at the one percent level. The presence of multiparty

coalitions is positive but small and non-significant for the workplace presence of union members

in the Round 5 data.

A benefit of using the Round 5 data is that we are also able to construct a 4-point scale

measuring the degree of trade union influence in the workplace, ranging from not much or no

influence (including no unions present) to a great deal of influence. As shown in column 1 of Table

4, the average influence is 1.6, which is between no influence and some influence. Returning to

Table 5, column 2 presents the results of a multivariate regression with this as the dependent

variable, and the same controls as in the probit model in column 1. Similar to the results for union

members in the workplace, the level of trade union influence has a negative relationship with both

disproportionality and the effective number of parties, and these are precisely estimated with small

standard errors relative to the coefficient estimates. Additionally, multiparty coalitions are

positively related to higher levels of union influence.

European Social Survey Results: Individual Trade Union Membership

All eight rounds of the European Social Survey ask respondents about their trade union

membership status. We can therefore analyze a pooled sample of 144,507 workers across 31

countries that not only has significant cross-country variability (31 countries), but also variation

25
across time (2002-2017). Column 2 of Table 4 presents the descriptive statistics for the key

variables, including that 30.5 percent of workers across the eight rounds of the European Social

Survey report currently belonging to a trade union. Turning to Table 5, column 3 reports the key

marginal effects from a probit model where individual union membership is the dependent

variable, and includes political system ideology and other control variables similar to the Round 5

models plus dummy variables for the year of the interview (for sample statistics and marginal

effects for all of the variables, see Appendix Table 4). Again, there is a significantly negative

estimate for disproportionality and the effective number of parties. The estimate for the presence

of coalition government is statistically significant and positive.

In column 4, we exploit the fact that most countries are represented in more than one round.

Specifically, this allows the inclusion of fixed country effects while still being able to identify the

independent role of the political systems measures. In the column 4 specification, then, the

influence of the political system variables is coming from within-country variation over time. In

other words, in this specification the results reveal how variation in a country’s political system

over time affects that country’s level of union membership. Even relying on within-country

variation, the estimates for disproportionality and the effective number of political parties remain

significantly negative, though under this most restrictive model, within-country changes in

coalition governance lose significance. We take this as strong evidence that there is an important

relationship between both disproportionality and the effective number of political parties and trade

union membership, and the direction of this relationship is consistent with the other measures of

workplace representation that we have analyzed: greater electoral representativeness is associated

with greater workplace voice, while competitive fragmentation weakens workplace voice.

26
 Implications for Comparative Industrial Relations and Workplace Voice

Our findings have several implications for both unions and scholars interested in the

intersection between industrial relations, the state, and political systems. First, the results offer

support to the notion that characteristics of political systems meaningfully affect union

membership, coverage, and influence in a manner heretofore either missing from the literature or

treated purely as a control. Much research has explored diverse explanations for declining union

density and persistent differences in cross-national union membership and collective

representation (e.g., Ebbinghaus, Göbel, and Koos 2011; Forth, Bryson, and George 2017;

Kranendonk and Beer 2016; Schnabel 2013). While research has analyzed individual attitudes and

demographic characteristics, job, and organizational characteristics, globalization and other

economic trends, trade union activities, the success of right-wing ideological platforms, the decline

of union-administered (Ghent) unemployment insurance, and differences in varieties of capitalism

and industrial relations systems, it appears that the importance of a country’s political system for

influencing trade union membership, coverage, and influence has been overlooked. Our findings

add to this literature by demonstrating that the country’s broader political system is important in

altering the likelihood of individuals belonging to unions, as well as the coverage and influence

unions provide within the workplace. In addition to the extensively-researched varieties of

capitalism framework, attention should be paid to varieties of political systems in comparative

industrial relations. 8

Second, this results imply that although overarching electoral reforms are rare, unions may

stand to benefit from any changes to electoral systems that increase the system’s proportionality.

8
A step-wise approach is a possibility for exploring the incremental explanatory power of
disproportionality, the number of effective political parties, and coalition government. This is a
cautious exercise because most of the dependent variables are dichotomous, but column 1 in
27
When the United Kingdom introduced its Alternative Vote (AV) referendum in 2011, the Labour

Party was split on how much backing it should give to the referendum, and it ultimately failed at

the polls (Qvortrup 2012). Our findings indicate that unions may have benefitted had the reform

effort succeeded (notwithstanding coalitional governance complexities). While labor and other

advocacy groups tend to push for a variety of governmental policies that they expect will further

their goals of higher membership, such as fighting for changes to labor laws that will promote

member mobilization or promoting policies to discourage anti-union employer behaviors, it is rare

for unions and labor advocates to focus on the political system in which they operate as a factor

that might aid or inhibit collective voice.

Third, our findings imply that political systems that encourage social dialogue in the form

of tripartite partnerships and variants of social pacts may provide more indirect benefits to unions

than have been recognized. In countries like Ireland, social partnership has been alternatively

viewed as a tool of economic growth (Roche 2007), a mechanism to constrain worker voice at the

expense of union elites (Allen 2000), or a relatively benign instrument that helped improve

economic productivity before being made irrelevant by financialization during the 2000s (Teague

and Donaghey 2015). While we are unable to test the specific role of social dialogue, our findings

Appendix Table 5 shows the adjusted R2 values for a baseline specification without controlling for
varieties of capitalism and the three key political systems measures but including all other controls
used in Table 2 or 5. Following a step-wise approach, we then add varieties of capitalism indicators
(column 2), and then the three political systems measures (column 3). A partial F-test indicates
that the adjusted R2 value increases by a statistically significant amount in the last step for each
model, meaning that the three political systems measures add explanatory power to every model.
To gauge the magnitude of this increased explanatory power, we use the change in the adjusted R2
value that accompanies the addition of the varieties of capitalism indicators in the previous step
because varieties of capitalism is widely accepted as important. In almost every case, the increase
in the adjusted R2 value is at least as large, if not larger, when adding the political systems variables
compared to the prior step of adding the varieties of capitalism indicators. This is consistent with
our contention that political systems are worthy of additional study.
28
suggest that missing from this analysis is the possibility that political systems that encourage social

partnership and other forms of social pacts may benefit union interests in myriad ways that foster

increased voice in the workplace. This opens up a need for further research investigating these

possible linkages while also suggesting other possible routes for strengthening workplace-based

representation.

Conclusion

We seek to make a unique theoretical contribution by developing a multi-channel

framework in which a country’s political system is predicted to affect trade union membership,

coverage, and influence. By making political systems that are commonly explored by political

scientists a key factor in theorizing employment relations institutions and outcomes, this expands

on the well-established frameworks for understanding the relationships between political

economies and industrial relations. We identify various ways in which a political system can shape

collective voice, and craft a three-channel theoretical framework based on incentives for

inclusionary governance, legislative body composition, and policy enactment. This also

complements and extends the well-known connections between political ideology and industrial

relations by theorizing mechanisms through which a country’s political system can affect the

workplace regardless of the ideology of the governing entity.

This theorizing motivates the empirical analyses that use multiple data sets. As the data do

not allow us to test the specific channels we theorized, the analyses focus on testing the more

general relationship between the political system and unionization. We follow the political science

literature in using disproportionality and the number of effective political parties as the key

dimensions to measure variation in electoral systems, as well as the presence or absence of

multiparty coalitions, which has been identified as another important aspect of a country’s political

29
system. Disproportionality (an inverse measure of electoral representativeness) is hypothesized to

have a negative relationship with workplace collective voice, and multiparty coalitions are

expected to be positively related with voice, while the prediction for the number of effective

political parties is ambiguous owing to the alternative interpretations typically attached to this

measure.

We test the impact of these measures on various measures of workplace representation

based on the availability of different measures across alternative data sets. This includes multiple

waves of establishment-level (the European Company Survey) as well as individual-level (the

European Social Survey) data sets. The European Social Survey results consistently indicate that

disproportionality negatively affects union membership, coverage, and influence, and these

findings are shared by the European Company Survey outcomes except after accounting for Ghent

systems in some cases. The overall strength of these findings are consistent with our theorizing

that unions may specifically benefit from certain types of representative electoral systems. We also

find that the presence of multiparty coalitions often correlates positively with unionization. This

lends at least some support to our theory that political systems that have multiparty coalitions,

which typically require bargaining in order to form governments and enact policies, will spill over

into workplaces that also support systems like collective bargaining. Finally, our results indicate

that as the effective number of parties in an electoral system increases, union membership,

coverage, and influence decrease. This outcome is consistent with the idea that greater numbers of

parties produce competitive fragmentation within a political system. Just as is the case for political

democracy, then, it appears that the optimal political system for collective voice is one that

achieves high levels of representativeness while avoiding excessive competitive fragmentation.

30
 A possible counter to the arguments we articulate above is that our empirical analyses face

issues of causality and endogeneity. Although we theorize that variations in political systems affect

variations in unionization, under some circumstances the reverse may hold true. We do not think

disproportionality likely fits these circumstances. It is difficult to envision the causal arrow running

from unionization to disproportionality since overt electoral reform is uncommon and even minor

changes to proportionality are generally made by political elites, who are unlikely to be heavily

influenced by unions at that stage in their decision-making.

Turning to our other measures of political systems, reverse causality is possible whereby

greater collective voice produces more political parties or multiparty coalitions. Unions may either

form their own political parties or train members to become involved in politics, and these groups

or individuals may bargain for a seat at the table when forming multiparty coalitions. However,

the overarching goal of these organizations is usually to grow membership, service their members’

interests, and advocate for policies that benefit the working class rather than to bring about

electoral reform. Moreover, the literature on comparative trade union density treats union

membership, coverage, and influence variation as the consequence of different micro (individual)

and macro (institutional) effects rather than the cause (Ebbinghaus, Göbel, and Koos 2011). Lastly,

the negative relationship between the effective number of parties and unionization is unlikely to

be subject to reverse causality since it is difficult to imagine unions actively seeking to reduce the

number of parties in a political system. Therefore, although we can discount the likelihood that

reverse causality affects our disproportionality and effective number of parties results, a limitation

is that causal ambiguity remains for the multiparty coalition outcomes.

Relatedly, perhaps there are unobserved norms of inclusivity or other factors that

simultaneously support representative institutions in the workplace and in the political system. We

31
account for an array of potentially confounding determinants of changes in workplace

representation, including country and time effects as well as macro-level political ideology

changes, Ghent systems, and varieties of capitalism, but we may not have exhausted all possible

variables that would explain these variations in collective voice. We therefore encourage additional

empirical analyses that might isolate causal relationships while also investigating the role of

specific channels through which a country’s electoral system may influence the workplace

independent of the political system’s ideology at a particular moment.

While there is more to be done, we believe that we have uncovered an overlooked area

with important implications for comparative industrial relations and collective voice. The

importance of cross-national institutional differences in the nature of trade unions and workplace

voice, supportive legislative policies, and varieties of capitalism needs to be complemented by a

deeper understanding of the role of varieties of political systems, including but not limited to

differences in electoral systems and multiparty coalitions. This also opens up the space to consider

the theoretical and empirical importance of political systems for other employment relations trends

and outcomes.

32
 References

Allen, Kieran (2000) The Celtic Tiger: The Myth of Social Partnership in Ireland (Manchester:
Manchester University Press).

Baccaro, Lucio, and Sang-Hoon Lim (2007) “Social Pacts as Coalitions of the Weak and
Moderate: Ireland, Italy and South Korea in Comparative Perspective.” European Journal
of Industrial Relations 13(1): 27-46.

Baccaro, Lucio, and Chris Howell (2017) Trajectories of Neoliberal Transformation: European
Industrial Relations Since the 1970s (Cambridge: Cambridge University Press).

Bechtel, Michael M., Jens Hainmueller, and Yotam Margalit (2014) “Preferences for International
Redistribution: The Divide over the Eurozone Bailouts.” American Journal of Political
Science 58(4): 835-56.

Blais, Andre (1991) “The Debate over Electoral Systems.” International Political Science Review
12(3): 239-60.

Blais, Andre, and Marc Andre Bodet (2006) “Does Proportional Representation Foster Closer
Congruence between Citizens and Policy Makers?” Comparative Political Studies 39(10):
1243-62.

Böckerman, Petri, and Roope Uusitalo (2006) “Erosion of the Ghent System and Union
Membership Decline: Lessons from Finland.” British Journal of Industrial Relations
44(2): 283-303.

Bryson, Alex, Rafael Gomez, Tobias Kretschmer, and Paul Willman (2014) “What Accounts for
the Union Member Advantage in Voter Turnout? Evidence from the European Union,
2002-2008.” Relations industrielles/Industrial Relations 69(4): 732-65.

Carey, John M., and Simon Hix (2011) “The Electoral Sweet Spot: Low-Magnitude Proportional
Electoral Systems.” American Journal of Political Science 55(2): 383-97.

Carnes, Nicholas (2012) “Does the Numerical Underrepresentation of the Working Class in
Congress Matter?” Legislative Studies Quarterly 37(1): 5-34.

Clark, William Roberts, and Matt Golder (2006) “Rehabilitating Duverger’s Theory: Testing the
Mechanical and Strategic Modifying Effects of Electoral Laws.” Comparative Political
Studies 39(6): 679-708.

Converse, Philip E. (1964) “The Nature of Belief Systems in Mass Publics.” In David E. Apter,
ed., Ideology and Discontent (New York, NY: Free Press): 206-61.

Cox, Gary W. (1997) Making Votes Count: Strategic Coordination in the World’s Electoral
Systems (Cambridge: Cambridge University Press).

33
Crouch, Colin (1993) Industrial Relations and European State Traditions (Oxford: Oxford
University Press).

Curini, Luigi, and Airo Hino (2012) “Missing Links in Party-System Polarization: How
Institutions and Voters Matter.” The Journal of Politics 74(2): 460-73.

Cusack, Thomas R., Torben Iversen, and David Soskice (2007) “Economic Interests and the
Origins of Electoral Systems.” American Political Science Review 101(3): 373-91.

Czymara, Christian S. (forthcoming) “Propagated Preferences? Political Elite Discourses and

Europeans’ Openness toward Muslim Immigrants.” International Migration Review.

D’Art, Daryl, and Tom Turner (2007) “Trade Unions and Political Participation in the European
Union: Still Providing a Democratic Dividend?” British Journal of Industrial Relations
45(1): 103-26.

Döring, Holger, and Philip Manow (2018) Parliaments and Governments Database (ParlGov):
Information on Parties, Elections, and Cabinets in Modern Democracies. Development
version. http://www.parlgov.org/.

Druckman, James N., and Lawrence R. Jacobs (2015) Who Governs? Presidents, Public Opinion,
and Manipulation. Chicago: University of Chicago Press.

Ebbinghaus, Bernhard (1995) “The Siamese Twins: Citizenship Rights, Cleavage Formation, and
Party-Union Relations in Western Europe.” International Review of Social History 40(S3):
51-89.

Ebbinghaus, Bernhard, Claudia Göbel, and Sebastian Koos (2011) “Social Capital, ‘Ghent’ and
Workplace Contexts Matter: Comparing Union Membership in Europe.” European
Journal of Industrial Relations 17(2): 107-24.

European Foundation for the Improvement of Living and Working Conditions, TNS Infratest
Sozialforschung (Munich) (2010) European Company Survey, 2009. [data collection]. UK
Data Service. SN: 6568, doi:10.5255/UKDA-SN-6568-1

European Foundation for the Improvement of Living and Working Conditions (2015) European
Company Survey, 2013. [data collection]. UK Data Service. SN: 7735,
doi:10.5255/UKDA-SN-7735-1

European Social Survey (2010) ESS Round 5: European Social Survey Round 5 Data. Data file
edition 3.4. [data collection]. NSD-Norwegian Centre for Research Data–Data Archive and
distributor of ESS data for ESS ERIC. doi:10.21338/NSD-ESS5-2010.

34
European Social Survey (2018) European Social Survey Cumulative File, ESS 1-8. Data file
edition 1.0. [data collection]. NSD - Norwegian Centre for Research Data-Data Archive
and distributor of ESS data for ESS ERIC. doi:10.21338/NSD-ESS-CUMULATIVE.

Farrell, David M. (2001) Electoral Systems: A Comparative Introduction (Basingstoke: Palgrave).

Flores, Rene D. (2018) “Can Elites Shape Public Attitudes Toward Immigrants? Evidence from
the 2016 US Presidential Election.” Social Forces 96(4): 1649-90.

Forth, John, Alex Bryson, and Anitha George (2017) “Explaining Cross-National Variation in
Workplace Employee Representation.” European Journal of Industrial Relations 23(4):
415-33.

Gabel, Matthew (1998) “Public Support for European Integration: An Empirical Test of Five
Theories.” The Journal of Politics 60(2): 333-54.

Gallagher, Michael (1991) “Proportionality, Disproportionality and Electoral Systems.” Electoral

Studies 10(1): 33-51.

Gallagher, Michael (2019) Election Indices Dataset. [data collection].

http://www.tcd.ie/Political_Science/people/michael_gallagher/ElSystems/index.php.

Godard, John (2002) “Institutional Environments, Employer Practices, and States in Liberal
Market Economies.” Industrial Relations 41(2): 249-86.

Godard, John (2017) Industrial Relations, the Economy, and Society, 5th ed. (Concord, Ontario:
Captus Press).

Greenwood, Royston, et al. (2010) “The Multiplicity of Institutional Logics and the Heterogeneity
of Organizational Responses.” Organization Science 21(2): 521-39.

Hall, Peter A., and David Soskice, eds. (2001) Varieties of Capitalism: The Institutional
Foundations of Comparative Advantage (Oxford: Oxford University Press).

Hamann, Kerstin et al. (2015) “Sharing the Rewards, Dividing the Costs? The Electoral
Consequences of Social Pacts and Legislative Reform in Western Europe.” West European
Politics 38(1): 206-27.

Hamann, Kerstin, Alison Johnston, and John Kelly (2013) “Unions Against Governments:
Explaining General Strikes in Western Europe, 1980-2006.” Comparative Political Studies
46(9): 1030-57.

Hamann, Kerstin, and John Kelly (2008) “Varieties of Capitalism and Industrial Relations.” In
Paul Blyton et al., eds., SAGE Handbook of Industrial Relations (London: Sage): 129-48.

35
Hamann, Kerstin, and John Kelly (2010) Parties, Elections, and Policy Reforms in Western
Europe: Voting for Social Pacts (London: Routledge).

Hobolt, Sara Binzer (2007) “Taking Cues from on? Voter Competence and Party Endorsements in
Referendums on European Integration.” European Journal of Political Research 46(2):
151-82.

Howell, Chris (forthcoming) “Rethinking the Role of the State in Employment Relations for a
Neoliberal Era.” ILR Review.

Hyman, Richard (2001) Understanding European Trade Unionism: Between Market, Class and
Society (London: Sage).

Hyman, Richard (2008) “The State in Industrial Relations.” In Paul Blyton et al., eds., SAGE
Handbook of Industrial Relations (London: Sage): 258-83.

Hyman, Richard, and Rebecca Gumbrell-McCormick (2010) “Trade Unions, Politics, and Parties:
Is A New Configuration Possible?” Transfer 16(3): 315-31.

Iversen, Torben, and David Soskice (2006) “Electoral Institutions, Parties, and the Politics of
Class: Why Some Democracies Distribute More Than Others.” American Political Science
Review 100(2): 165-81.

Johansson, Karl Magnus, and Tapio Raunio (2001) “Partisan Responses to Europe: Comparing
Finnish and Swedish Political Parties.” European Journal of Political Research 39(2): 225-
49.

Karp, Jeffrey A., and Susan A. Banducci (2008) “Political Efficacy and Participation in Twenty-
Seven Democracies: How Electoral Systems Shape Political Behaviour.” British Journal
of Political Science 38(2): 311-34.

Kranendonk, Maria, and Paul Beer (2016) “What Explains the Union Membership Gap Between
Migrants and Natives?” British Journal of Industrial Relations 54(4): 846-69.

Kroh, Martin (2007) “Measuring Left–Right Political Orientation: The Choice of Response
Format.” Public Opinion Quarterly 71(2): 204-20.

Laakso, Markku, and Rein Taagepera (1979) “‘Effective’ Number of Parties: A Measure with
Application to West Europe.” Comparative Political Studies 12(1): 3-27.

Laver, Michael, and Norman Schofield (1998) Multiparty Government: The Politics of Coalition
in Europe. Ann Arbor: University of Michigan Press.

Lijphart, Arend (1994) Electoral Systems and Party Systems: A Study of Twenty-Seven
Democracies, 1945-1990 (Oxford: Oxford University Press).

36
Lipset, Seymour Martin (1962) “Trade Unions and Social Structure: II.” Industrial Relations 1(2):
89-110.

Lipset, Seymour Martin (1983) “Radicalism or Reformism: The Sources of Working-Class

Politics.” American Political Science Review 77(1): 1-18.

Lublin, David, and Shaun Bowler (2017) “Representation of Ethnic Minorities.” In Erik S. Herron,
Robert J. Pekkanen, and Matthew S. Shugart, eds., The Oxford Handbook of Electoral
Systems (Oxford: Oxford University Press).

Martin, Cathie J., and Duane Swank (2004) “Does the Organization of Capital Matter? Employers
and Active Labor Market Policy at the National and Firm Levels.” American Political
Science Review 98(4): 593-611.

Martin, Cathie J., and Duane Swank (2008) “The Political Origins of Coordinated Capitalism:
Business Organizations, Party Systems, and State Structure in the Age of Innocence.”
American Political Science Review 102(2): 181-98.

Martin, Lenny W., and Georg Vanberg (2004) “Policing the Bargain: Coalition Government and
Parliamentary Scrutiny.” American Journal of Political Science 48(1): 13-27.

Martin, Lenny W., and Georg Vanberg (2013) “Multiparty Government, Fiscal Institutions, and
Public Spending.” The Journal of Politics 75(4): 953-67.

Matakos, Konstantinos, Orestis Troumpounis, and Dimitrios Xefteris (2016) “Electoral Rule
Disproportionality and Platform Polarization.” American Journal of Political Science
60(4): 1026-43.

McLaren, Lauren M. (2001) “Immigration and the New Politics of Inclusion and Exclusion in the
European Union: The Effect of Elites and the EU on Individual-Level Opinions Regarding
European and Non-European Immigrants.” European Journal of Political Research 39(1):
81-108.

McLaughlin, Colm, and Chris F. Wright (2018) “The Role of Ideas in Understanding Industrial
Relations Policy Change in Liberal Market Economies.” Industrial Relations 57(4): 568-
610.

Orellana, Salomon (2010) “How Electoral Systems Can Influence Policy Innovation.” Policy
Studies Journal 38(4): 613-28.

Pernicka, Susanne, and Vera Glassner (2014) “Transnational Trade Union Strategies Towards
European Wage Policy: A Neo-Institutional Framework.” European Journal of Industrial
Relations 20(4): 317-34.

Powell, G. Bingham (2000) Elections as Instruments of Democracy: Majoritarian and

Proportional Visions (New Haven: Yale University Press).

37
Qvortrup, Matt (2012) “Voting on Electoral Reform: A Comparative Perspective on the
Alternative Vote Referendum in the United Kingdom.” The Political Quarterly 83(1): 108-
16.

Rasmussen, Magnus Bergli, and Jonas Ponusson (2018) “Working-Class Strength by Institutional
Design? Unionization, Partisan Politics, and Unemployment Insurance Systems, 1870 to
2010.” Comparative Political Studies 51(6): 793-828.

Rathgeb, Philip (2018a) “When Weak Governments Confront Inclusive Trade Unions: The
Politics of Protecting Labour Market Outsiders in the Age of Dualization.” European
Journal of Industrial Relations 24(1): 5-22.

Rathgeb, Philip (2018b) Strong Governments, Precarious Workers: Labor Market Policy in the
Era of Liberalization (Ithaca, NY: Cornell University Press).

Renwick, Alan (2011) “Electoral Reform in Europe since 1945.” West European Politics 34(3):
456-77.

Riker, William (1962) The Theory of Political Coalitions. New Haven: Yale University Press.

Roche, William K. (2007) “Social Partnerships in Ireland and New Social Pacts.” Industrial
Relations 46(3): 395-425.

Sartori, Giovanni (1987) The Theory of Democracy Revisited. Washington, DC: CQ Press.

Schnabel, Claus (2013) “Union Membership and Density: Some (Not So) Stylized Facts and
Challenges.” European Journal of Industrial Relations 19(3): 255-72.

Schneider, Martin R., and Mihai Paunescu (2012) “Changing Varieties of Capitalism and Revealed
Comparative Advantages from 1990 to 2005: A Test of the Hall and Soskice Claims.”
Socio-Economic Review 10(4): 731-53.

Sojourner, Aaron J. (2013) “Do Unions Promote Electoral Office-Holding? Evidence from
Correlates of State Legislatures' Occupational Shares.” ILR Review 66(2): 467-86.

Strom, Kaare, Wolfgang Muller, and Torbjorn Bergman, eds. (2008) Cabinets and Coalition
Building: The Democratic Life Cycle in Western Europe. Oxford: Oxford University Press.

Taagepera, Rein, and Matthew Soberg Shugart (1989) Seats and Votes: The Effects and
Determinants of Electoral Systems (New Haven: Yale University Press).

Teague, Paul, and Jimmy Donaghey (2015) “The Life and Death of Irish Social Partnership:
Lessons for Social Pacts.” Business History 57(3): 418-37.

38
Thornton, Patricia H., and William Ocasio (1999) “Institutional Logics and the Historical
Contingency of Power in Organizations: Executive Succession in the Higher Education
Publishing Industry, 1958–1990,” American Journal of Sociology 105(3): 801-43.

Thornton, Patricia H., and William Ocasio (2008) “Institutional Logics.” In Royston Greenwood
et al., eds., The SAGE Handbook of Organizational Institutionalism (London: SAGE): 99-
128.

Timmermans, Arco (2006) “Standing Apart and Sitting Together: Enforcing Coalition Agreements
in Multiparty Systems.” European Journal of Political Research 45(2): 263-83.

Traxler, Franz (1999) “The State in Industrial Relations: A Cross-National Analysis of

Developments and Socioeconomic Effects.” European Journal of Political Research
36(1): 55-85.

Traxler, Franz (2004) “The Metamorphosis of Corporatism: From Classical to Lean Patterns.”
European Journal of Political Research 43(4): 571-98.

Traxler, Franz, and Bernhard Kittel (2000) “The Bargaining System and Performance: A
Comparison of 18 OECD Countries.” Comparative Political Studies 33(9): 1154-90.

Zaller, John (1992) The Nature and Origins of Mass Opinion. New York, NY: Cambridge
University Press.

39
 Table 1: Countries Present in the Analyses
European Company
European Social Survey
Survey
2009 2013 1 2 3 4 5 6 7 8
Austria (AUT) Yes Yes Y Y Y Y Y Y Y
Belgium (BEL) Yes Yes Y Y Y Y Y Y Y Y
Bulgaria (BGR) Yes Yes Y Y Y Y
Croatia (HRV) Yes Yes Y Y
Cyprus (CYP) Yes Yes Y Y Y Y
Czech Republic (CZE) Yes Yes Y Y Y Y Y Y Y
Denmark (DNK) Yes Yes Y Y Y Y Y Y Y
Estonia (EST) Yes Yes Y Y Y Y Y Y
Finland (FIN) Yes Yes Y Y Y Y Y Y Y Y
France (FRA) Yes Yes Y Y Y Y Y Y Y Y
Germany (DEU) Yes Yes Y Y Y Y Y Y Y Y
Greece (GRC) Yes Yes Y Y Y Y
Hungary (HUN) Yes Yes Y Y Y Y Y Y
Iceland (ISL) Yes Y Y Y
Ireland (IRL) Yes Yes Y Y Y Y Y Y Y Y
Israel (ISR) Y Y Y Y Y Y
Italy (ITA) Yes Yes Y Y Y Y
Latvia (LVA) Yes Yes Y
Lithuania (LTU) Yes Yes Y Y Y Y
Luxembourg (LUX) Yes Yes Y Y
Malta (MLT) Yes Yes
Netherlands (NLD) Yes Yes Y Y Y Y Y Y Y Y
Norway (NOR) Y Y Y Y Y Y Y Y
Poland (POL) Yes Yes Y Y Y Y Y Y Y Y
Portugal (PRT) Yes Yes Y Y Y Y Y Y Y Y
Romania (ROU) Yes Yes Y
Slovakia (SVK) Yes Yes Y Y Y Y Y
Slovenia (SVN) Yes Yes Y Y Y Y Y Y Y
Spain (ESP) Yes Yes Y Y Y Y Y Y Y
Sweden (SWE) Yes Yes Y Y Y Y Y Y Y Y
Switzerland (CHE) Y Y Y Y Y Y Y Y
Turkey (TUR) Yes Yes
United Kingdom (GBR) Yes Yes Y Y Y Y Y Y Y Y
Total Number 29 30 21 22 21 27 25 24 21 22

40
 Table 2: Analyzing Trade Union Presence and Coverage Using the European Company Survey

Is the following is present in the workplace (1=yes, 0=no)?

Trade Union Representation Collective Wage Agreement for at least
(includes Dual Channel) Some Employees
2009 2013 2009 2013
(1) (2) (3) (4) (5) (6) (7) (8)
PANEL A: Descriptive Statistics
Sample Mean 0.348 0.339 0.648 0.634
(standard deviation) (0.476) (0.473) (0.477) (0.482)
PANEL B: Key Marginal Effects from Probit Models
Disproportionality -0.015** 0.010** -0.016** 0.001 -0.032** -0.029** -0.008** -0.001
(0.001) (0.001) (0.001) (0.001) (0.002) (0.002) (0.001) (0.001)
Effective Number of -0.011** -0.022** -0.020** -0.038** -0.078** -0.087** -0.076** -0.099**
Political Parties (0.003) (0.003) (0.003) (0.004) (0.003) (0.004) (0.003) (0.004)
Coalition 0.059** 0.220** -0.129** -0.181** 0.039 0.053* -0.141** -0.144**
Government (0.019) (0.019) (0.010) (0.010) (0.023) (0.024) (0.011) (0.012)
Ghent --- 0.335** --- 0.333** --- 0.087** --- 0.267**
Country (0.018) (0.018) (0.022) (0.020)
Ideology and Other Yes Yes Yes Yes Yes Yes Yes Yes
Controls from Table 3
Sample Size 20,830 20,830 20,224 20,224 24,857 24,857 23,472 23,472
Notes: Each cell in panel A reports the sample mean and standard deviation of the dependent variable in the relevant sample.
Each cell in Panel B reports the marginal effect and robust standard error for the key independent variables from a probit model
for each dependent variable in the relevant sample. The control variables are those listed in Table 3, and the probit models are
estimated using establishment sampling weights.
Statistically significant at the * 0.05 or ** 0.01 level.

41
 Table 3: European Company Survey Independent Variables

PANEL A: Descriptive Statistics of Key Independent Variables

2009 2013
Mean Minimum / Mean Minimum /
(std dev) Maximum (std dev) Maximum
(1) (2) (3) (4)
Disproportionality 5.678 0.72 6.584 0.73
(4.047) 16.73 (4.424) 17.66
Effective Number of 4.466 2.08 4.744 2.08
Political Parties (1.545) 9.04 (1.540) 10.04
Coalition 0.706 0 0.761 0
Government (0.455) 1 (0.426) 1

PANEL B: Other Control Variables

2009 2013
• Ideology of the ruling party or coalition • Ideology of the ruling party or coalition
(weighted by seats) (quadratic (weighted by seats) (quadratic
specification) specification)
• Ideology of the legislative body weighted • Ideology of the legislative body weighted
by seats (quadratic specification) by seats (quadratic specification)
• Country-specific varieties of capitalism (5 • Country-specific varieties of capitalism (5
categories) categories)
• Organizational structure (3 categories) • Organizational structure (3 categories)
• Public or private (4 categories) • Public or private (2 categories)
• Organizational change (3 categories) • Ownership change (2 categories)
• Fraction female (7 categories) • Fraction female (7 categories)
• Fraction part-time (7 categories) • Fraction part-time (7 categories)
• Fraction high skilled (7 categories) • Fraction university degree (7 categories)
• Number of employees (10 categories) • Number of employees (3 categories)
• Industry (11 categories) • Industry (6 categories)
Note: The descriptive statistics are derived from the full analysis samples used for the models in
columns 5-8 in Table 2 (n=24,857 for 2009 and n=23,472 for 2013).

42
 Table 4: European Social Survey—Summary Statistics for Key Variables
Mean (std dev)
Round 5 Rounds 1-8
Analysis Sample Analysis sample
(1) (2)
There are Union Members in the Workplace 0.792 ---
(1=yes, 0=no) (0.406)
Trade Union Influence in the Workplace 1.641 ---
(1=not much or no influence, 4=a great deal) (0.807)
Respondent is a Trade Union Member --- 0.305
(1=yes, 0=no) (0.460)
Disproportionality 5.790 5.884
(4.101) (4.832)
Effective Number of Political 5.138 5.181
Parties (1.670) (1.626)
Coalition 0.849 0.835
Government (0.358) (0.371)
Ghent 0.171 0.186
Country (0.377) (0.389)

Sample size 15,899 144,507

43
 Table 5: Analyzing Trade Union Presence, Influence in the Workplace, and Individual Trade
Union Membership Using the European Social Survey
Round 5 Rounds 1-8
Trade Union Trade Union
Members in the Influence in Respondent is a
Workplace the Workplace Trade Union Member
(1) (2) (3) (4)
Disproportionality -0.014** -0.028** -0.009** -0.002**
(0.001) (0.002) (0.0003) (0.0008)
Effective Number of -0.039** -0.036** -0.022** -0.006**
Political Parties (0.003) (0.005) (0.001) (0.002)
Coalition 0.019 0.268** 0.063** 0.003
Government (0.011) (0.023) (0.004) (0.005)
Ghent 0.164** -0.052* 0.323** ---
Country (0.016) (0.024) (0.003)
Country-specific varieties
Yes Yes Yes No
of capitalism (5 categories)
Ideology and Other Control
Yes Yes Yes Yes
Variables (see note)
Country effects No No No Yes

Sample Size 15,899 15,899 144,507 144,507

Note: Columns 1, 3, and 4 reports the marginal effect and robust standard error from a probit
model; column 2 reports the coefficient and robust standard error from a regression. All models
are estimated using ESS design weights and include the following control variables: ideology of
the ruling party or coalition (weighted by seats) (quadratic specification); ideology of the
legislative body weighted by seats (quadratic specification); whether a citizen; whether belong to
a minority ethnic group; gender; have children at home; urban location (3 categories); years of
education; age (quadratic specification); whether supervises others; number of employees (5
categories); occupation (10 categories); industry (20 categories). Columns 1 and 2 also control
for type of employer (5 categories); columns 3 and 4 also control for year of interview (16
categories).
Statistically significant at the * 0.05 or ** 0.01 level.

44
 Appendix Table 1:
European Company Survey Descriptive Statistics
2009 2013
(1) (2)
Disproportionality 5.678 6.584
(4.047) (4.424)
Effective number of political parties 4.466 4.744
(1.545) (1.540)
Coalition government 0.706 0.761
(0.455) (0.426)
Ghent country (Sweden, Denmark, Finland, Iceland, and 0.155 0.158
Belgium) (0.362) (0.364)
Ideology of the ruling party or cabinet 5.672 5.728
(1.473) (1.513)
Ideology of the ruling party or cabinet (squared) 34.344 35.098
(16.211) (15.811)
Ideology of the legislative body weighted by seats 5.358 5.413
(0.508) (0.559)
Ideology of the legislative body weighted by seats 28.966 29.618
(squared) (5.462) (5.972)
Varieties of capitalism: LME 0.091 0.083
(0.288) (0.276)
Varieties of capitalism: LME-like 0.256 0.257
(0.436) (0.437)
Varieties of capitalism: CME 0.217 0.223
(0.412) (0.416)
Varieties of capitalism: Hybrid 0.283 0.289
(0.451) (0.454)
Domestic ownership (private sector) 0.651
(0.477)
Foreign ownership (private sector) 0.104
(0.305)
Establishment taken over in past three years 0.049
(0.215)
Establishment relocated in past three years 0.057
(0.233)
Significant ownership change in past three years 0.097
(0.295)
Fraction of employees who are female: None at all 0.027 0.020
(0.161) (0.141)
Fraction of employees who are female: <20% 0.287 0.314
(0.452) (0.464)
Fraction of employees who are female: 20-39% 0.179 0.234
(0.383) (0.424)
Fraction of employees who are female: 40-59% 0.186 0.215
(0.389) (0.411)
Fraction of employees who are female: 60-79% 0.159 0.132
(0.365) (0.338)
45
Fraction of employees who are female: 80-99% 0.148 0.077
(0.355) (0.266)
Fraction of employees who are part-time: None at all 0.389 0.333
(0.488) (0.471)
Fraction of employees who are part-time: <20% 0.424 0.492
(0.494) (0.500)
Fraction of employees who are part-time: 20-39% 0.098 0.091
(0.297) (0.287)
Fraction of employees who are part-time: 40-59% 0.039 0.036
(0.194) (0.187)
Fraction of employees who are part-time: 60-79% 0.025 0.025
(0.156) (0.155)
Fraction of employees who are part-time: 80-99% 0.020 0.018
(0.140) (0.133)
Fraction of employees who are high-skill: None at all 0.180
(0.385)
Fraction of employees who are high-skill: <20% 0.402
(0.490)
Fraction of employees who are high-skill: 20-39% 0.176
(0.381)
Fraction of employees who are high-skill: 40-59% 0.078
(0.268)
Fraction of employees who are high-skill: 60-79% 0.065
(0.247)
Fraction of employees who are high-skill: 80-99% 0.075
(0.263)
Fraction of employees with a university degree: None at 0.120
all (0.325)
Fraction of employees with a university degree: <20% 0.467
(0.499)
Fraction of employees with a university degree: 20-39% 0.187
(0.390)
Fraction of employees with a university degree: 40-59% 0.088
(0.283)
Fraction of employees with a university degree: 60-79% 0.060
(0.238)
Fraction of employees with a university degree: 80-99% 0.061
(0.239)
Establishment size: 10 to 19 0.266
(0.442)
Establishment size: 20 to 49 0.273
(0.445)
Establishment size: 50 to 99 0.162
(0.368)
Establishment size: 100 to 149 0.068
(0.251)
Establishment size: 150 to 199 0.038
(0.190)

46
Establishment size: 200 to 249 0.026
(0.159)
Establishment size: 250 to 299 0.028
(0.166)
Establishment size: 300 to 399 0.041
(0.199)
Establishment size: 400 to 499 0.027
(0.162)
Establishment size: 10-49 0.520
(0.500)
Establishment size: 50-249 0.320
(0.467)
Major industry (NACE 1.1): Mining and quarrying; 0.316
Manufacturing; Electricity, gas and water supply (0.465)
Major industry (NACE 1.1): Construction 0.102
(0.302)
Major industry (NACE 1.1): Wholesale and retail trade; 0.148
repair of motor vehicles, motorcycles and personal and (0.355)
household goods
Major industry (NACE 1.1): Hotels and restaurants 0.035
(0.184)
Major industry (NACE 1.1): Transport, storage, and 0.047
communication (0.213)
Major industry (NACE 1.1): Financial intermediation 0.020
(0.139)
Major industry (NACE 1.1): Real estate, renting, and 0.092
business activities (0.290)
Major industry (NACE 1.1): Public administration and 0.062
defence; compulsory social security (0.240)
Major industry (NACE 1.1): Education 0.069
(0.253)
Major industry (NACE 1.1): Health and social work 0.068
(0.251)
Major industry (NACE 2): Industry 0.348
(0.476)
Major industry (NACE 2): Construction 0.089
(0.285)
Major industry (NACE 2): Commerce and hospitality 0.237
(0.425)
Major industry (NACE 2): Transport and 0.067
communication (0.251)
Major industry (NACE 2): Financial services and real 0.043
estate (0.203)

Sample Size 24,857 23,472

47
 Appendix Table 2: European Company Survey Full Marginal Effects Results
Trade Union Representation Collective Wage Agreement for at least
(includes Dual Channel) Some Employees
2009 2013 2009 2013
(1) (2) (3) (4) (5) (6) (7) (8)
Disproportionality -0.015** 0.010** -0.016** 0.001 -0.032** -0.029** -0.008** -0.001
(0.001) (0.002) (0.001) (0.001) (0.002) (0.002) (0.001) (0.001)
Effective number of -0.011** -0.022** -0.020** -0.038** -0.078** -0.087** -0.076** -0.099**
political parties (0.003) (0.003) (0.003) (0.004) (0.003) (0.004) (0.003) (0.004)
Coalition 0.059** 0.220** -0.129** -0.181** 0.039 0.053* -0.141** -0.144**
Government (0.019) (0.019) (0.010) (0.010) (0.023) (0.024) (0.011) (0.012)
Ghent 0.335** 0.333** 0.087** 0.267**
Country (0.018) (0.018) (0.022) (0.020)
** ** **
Ideology of the ruling -0.254 -0.364 -0.059 -0.060** -0.050 -0.038 0.364 **
0.434**
party or cabinet (0.022) (0.024) (0.016) (0.016) (0.027) (0.027) (0.021) (0.024)
** ** *
Ideology of the ruling 0.023 0.031 0.003 0.005** 0.006 *
0.004 -0.039 **
-0.044**
party or cabinet (squared) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002)
** **
Ideology of the legislative 1.410 0.666 -0.193 -0.535** 6.948 **
6.851 **
2.785 **
2.747**
body weighted by seats (0.212) (0.217) (0.131) (0.127) (0.212) (0.213) (0.145) (0.146)
** **
Ideology of the legislative -0.124 -0.057 0.020 0.051** -0.648 **
-0.640 **
-0.264 **
-0.263**
body weighted by seats (0.019) (0.020) (0.012) (0.012) (0.020) (0.020) (0.014) (0.014)
(squared)
Varieties of capitalism: 0.107** -0.096** -0.003 -0.092** -0.069* -0.098** 0.347** 0.296**
LME (0.019) (0.020) (0.016) (0.018) (0.027) (0.030) (0.019) (0.019)
Varieties of capitalism: 0.030 -0.096** -0.030* -0.074** 0.013 0.016 0.504** 0.501**
LME-like (0.019) (0.020) (0.014) (0.015) (0.022) (0.022) (0.014) (0.014)
Varieties of capitalism: 0.058* -0.151** 0.109** 0.050* 0.128** 0.125** 0.441** 0.470**
CME (0.023) (0.024) (0.017) (0.020) (0.026) (0.026) (0.016) (0.015)
Varieties of capitalism: -0.013 -0.045** -0.182** -0.142** -0.072** -0.069** 0.245** 0.260**
Hybrid (0.017) (0.016) (0.014) (0.013) (0.022) (0.022) (0.014) (0.014)
Organizational structure: 0.019 0.019 0.070** 0.066** 0.009 0.010 0.058** 0.053**
Headquarters (0.011) (0.011) (0.010) (0.010) (0.014) (0.014) (0.013) (0.013)
Organizational structure: 0.088** 0.075** 0.156** 0.155** 0.116** 0.113** 0.102** 0.099**

48
Subsidiary (0.011) (0.011) (0.014) (0.013) (0.017) (0.017) (0.017) (0.017)
Public sector 0.154** 0.152** 0.218** 0.224** -0.008 -0.007 0.071** 0.080**
establishment (0.029) (0.029) (0.015) (0.015) (0.042) (0.042) (0.017) (0.018)
Domestic ownership -0.007 -0.009 -0.085* -0.083*
(private sector) (0.028) (0.028) (0.040) (0.040)
Foreign ownership 0.019 0.013 -0.156** -0.157**
(private sector) (0.030) (0.030) (0.043) (0.043)
Establishment taken over 0.025 0.018 0.032 0.030
in past three years (0.016) (0.017) (0.024) (0.025)
Establishment relocated in -0.022 -0.027 -0.001 -0.001
past three years (0.016) (0.016) (0.020) (0.020)
Fraction of employees 0.127** 0.143** 0.079 0.057 0.091* 0.090* 0.021 0.013
who are female: None at (0.033) (0.035) (0.046) (0.045) (0.042) (0.042) (0.051) (0.051)
all
Fraction of employees 0.096** 0.110** 0.091* 0.074 0.072* 0.071* 0.025 0.024
who are female: <20% (0.028) (0.030) (0.041) (0.040) (0.036) (0.036) (0.045) (0.046)
Fraction of employees 0.096** 0.115** 0.080* 0.066 0.018 0.018 0.021 0.020
who are female: 20-39% (0.029) (0.031) (0.041) (0.040) (0.036) (0.037) (0.045) (0.046)
Fraction of employees 0.079** 0.093** 0.071 0.060 0.019 0.019 0.002 0.001
who are female: 40-59% (0.028) (0.030) (0.041) (0.040) (0.036) (0.036) (0.046) (0.046)
Fraction of employees 0.068* 0.082** 0.061 0.052 0.023 0.023 0.030 0.028
who are female: 60-79% (0.027) (0.029) (0.041) (0.040) (0.036) (0.036) (0.046) (0.046)
Fraction of employees 0.067** 0.083** 0.063 0.060 0.001 0.000 0.027 0.026
who are female: 80-99% (0.026) (0.028) (0.042) (0.041) (0.036) (0.036) (0.047) (0.047)
Fraction of employees 0.056 0.037 0.073 0.065 -0.091 -0.091 0.035 0.031
who are part-time: None (0.063) (0.063) (0.044) (0.044) (0.056) (0.056) (0.045) (0.044)
at all
Fraction of employees 0.067 0.055 0.089* 0.082 -0.056 -0.056 0.062 0.059
who are part-time: <20% (0.063) (0.063) (0.044) (0.044) (0.056) (0.056) (0.045) (0.044)
Fraction of employees 0.040 0.026 0.104* 0.097* -0.048 -0.048 0.052 0.056
who are part-time: 20- (0.064) (0.064) (0.046) (0.045) (0.057) (0.057) (0.047) (0.046)
39%
Fraction of employees 0.008 0.006 0.060 0.054 0.002 0.004 0.090 0.094

49
who are part-time: 40- (0.067) (0.067) (0.049) (0.049) (0.059) (0.060) (0.050) (0.050)
59%
Fraction of employees -0.002 0.002 0.078 0.070 0.009 0.009 0.108* 0.114*
who are part-time: 60- (0.067) (0.068) (0.052) (0.052) (0.063) (0.063) (0.053) (0.053)
79%
Fraction of employees -0.016 -0.011 0.077 0.069 -0.032 -0.031 0.006 0.007
who are part-time: 80- (0.067) (0.067) (0.053) (0.053) (0.065) (0.066) (0.055) (0.055)
99%
Fraction of employees 0.059* 0.060* 0.036 0.037
who are high-skill: None (0.024) (0.023) (0.032) (0.032)
at all
Fraction of employees 0.102** 0.096** 0.044 0.043
who are high-skill: <20% (0.023) (0.023) (0.031) (0.031)
Fraction of employees 0.090** 0.082** 0.052 0.051
who are high-skill: 20- (0.024) (0.023) (0.032) (0.032)
39%
Fraction of employees 0.095** 0.086** 0.034 0.033
who are high-skill: 40- (0.025) (0.025) (0.034) (0.034)
59%
Fraction of employees 0.130** 0.121** 0.016 0.014
who are high-skill: 60- (0.027) (0.026) (0.034) (0.034)
79%
Fraction of employees 0.081** 0.072** 0.019 0.018
who are high-skill: 80- (0.024) (0.023) (0.034) (0.034)
99%
Establishment size: 10 to -0.363** -0.360** -0.198** -0.199**
19 (0.013) (0.013) (0.018) (0.018)
Establishment size: 20 to -0.268** -0.266** -0.173** -0.173**
49 (0.013) (0.013) (0.017) (0.017)
Establishment size: 50 to -0.177** -0.178** -0.116** -0.117**
99 (0.014) (0.013) (0.018) (0.018)
Establishment size: 100 to -0.151** -0.147** -0.096** -0.095**
149 (0.017) (0.016) (0.022) (0.022)

50
Establishment size: 150 to -0.097** -0.097** -0.069** -0.069*
199 (0.018) (0.018) (0.027) (0.027)
Establishment size: 200 to -0.083** -0.083** -0.026 -0.026
249 (0.021) (0.020) (0.029) (0.029)
Establishment size: 250 to -0.065** -0.061** -0.048 -0.048
299 (0.022) (0.019) (0.030) (0.030)
Establishment size: 300 to -0.039* -0.040* -0.035 -0.036
399 (0.018) (0.017) (0.025) (0.025)
Establishment size: 400 to -0.024 -0.020 0.025 0.026
499 (0.023) (0.020) (0.028) (0.028)
Major industry: Mining -0.009 -0.001 0.009 0.009
and quarrying; (0.017) (0.017) (0.026) (0.026)
Manufacturing;
Electricity, gas and water
supply
Major industry: -0.068** -0.060** 0.051 0.051
Construction (0.020) (0.020) (0.028) (0.028)
Major industry: Wholesale -0.052** -0.044* -0.033 -0.034
and retail trade; repair of (0.019) (0.019) (0.027) (0.027)
motor vehicles,
motorcycles and personal
and household goods
Major industry: Hotels -0.030 -0.029 0.009 0.008
and restaurants (0.028) (0.028) (0.034) (0.034)
Major industry: Transport, 0.007 0.009 -0.018 -0.019
storage, and (0.022) (0.022) (0.033) (0.033)
communication
Major industry: Financial -0.004 0.012 0.022 0.022
intermediation (0.035) (0.035) (0.042) (0.041)
Major industry: Real -0.045* -0.038 -0.103** -0.103**
estate, renting, and (0.019) (0.019) (0.028) (0.028)
business activities
Major industry: Public 0.066** 0.068** 0.000 -0.000

51
administration and (0.022) (0.021) (0.033) (0.033)
defence; compulsory
social security
Major industry: Education 0.091** 0.102** 0.065* 0.065*
(0.021) (0.021) (0.032) (0.032)
Major industry: Health 0.058** 0.056** 0.058 0.056
and social work (0.021) (0.021) (0.030) (0.030)
Significant ownership 0.041** 0.037** -0.020 -0.022
change in past three years (0.013) (0.013) (0.015) (0.015)
Fraction of employees -0.030 -0.020 -0.003 0.003
with a university degree: (0.030) (0.030) (0.034) (0.035)
None at all
Fraction of employees -0.001 0.007 0.017 0.023
with a university degree: (0.028) (0.028) (0.033) (0.033)
<20%
Fraction of employees -0.006 -0.000 0.002 0.006
with a university degree: (0.029) (0.029) (0.034) (0.034)
20-39%
Fraction of employees 0.009 0.011 -0.023 -0.021
with a university degree: (0.030) (0.030) (0.035) (0.035)
40-59%
Fraction of employees 0.005 0.008 0.000 0.002
with a university degree: (0.031) (0.031) (0.036) (0.037)
60-79%
Fraction of employees -0.038 -0.035 -0.045 -0.042
with a university degree: (0.032) (0.032) (0.036) (0.036)
80-99%
Establishment size: 10-49 -0.273** -0.266** -0.117** -0.117**
(0.011) (0.011) (0.014) (0.014)
Establishment size: 50- -0.122** -0.115** -0.096** -0.093**
249 (0.011) (0.011) (0.014) (0.014)
Major industry: 0.052** 0.055** 0.036** 0.036**
Construction (0.012) (0.012) (0.014) (0.013)

52
Major industry: 0.010 0.014 0.069** 0.069**
Commerce and hospitality (0.017) (0.016) (0.018) (0.017)
Major industry: Transport 0.015 0.018 0.040** 0.041**
and communication (0.012) (0.012) (0.014) (0.013)
Major industry: Financial 0.056** 0.057** 0.052* 0.052*
services and real estate (0.018) (0.018) (0.021) (0.021)
Major industry: Other 0.073** 0.072** 0.077** 0.077**
services (0.021) (0.021) (0.024) (0.024)

Sample size 20,830 20,830 20,224 20,224 24,857 24,857 23,472 23,472
Notes: Each cell reports the marginal effect and robust standard error from a probit model for each dependent variable in the relevant
sample, estimated using establishment sampling weights.
Statistically significant at the * 0.05 or ** 0.01 level.

53
 Appendix Table 3:
European Social Survey Full Means and Marginal Effects Results (Round 5)
There are Union Trade Union
Mean Members in the Influence in the
(Std Dev) Workplace Workplace
(1) (2) (3)
There are union members in the 0.792 Dependent ---
workplace (0.406) Variable
Trade union influence in the 1.641 --- Dependent
workplace (0.807) Variable
Disproportionality 5.790 -0.014** -0.028**
(4.101) (0.001) (0.002)
Effective number of political parties 5.138 -0.039** -0.036**
(1.670) (0.003) (0.005)
Coalition government 0.849 0.019 0.268**
(0.358) (0.011) (0.023)
Ghent country 0.171 0.164** -0.052*
(0.377) (0.016) (0.024)
Ideology of the ruling party or 5.899 0.031 0.098*
cabinet (1.379) (0.021) (0.044)
Ideology of the ruling party or 36.695 -0.004 -0.012**
cabinet (squared) (14.918) (0.002) (0.004)
Ideology of the legislative body 5.394 3.459** 5.231**
weighted by seats (0.457) (0.186) (0.408)
Ideology of the legislative body 29.308 -0.325** -0.481**
weighted by seats (squared) (4.919) (0.017) (0.038)
Varieties of capitalism: LME 0.141 0.116** -0.152**
(0.348) (0.019) (0.038)
Varieties of capitalism: LME-like 0.262 0.138** -0.028
(0.440) (0.018) (0.035)
Varieties of capitalism: CME 0.233 0.119** -0.035
(0.423) (0.017) (0.036)
Varieties of capitalism: Hybrid 0.277 0.018 -0.132**
(0.448) (0.017) (0.036)
Citizen of country 0.956 0.046** 0.068*
(0.206) (0.015) (0.032)
Belong to minority ethnic group in 0.048 0.020 0.073*
country (0.213) (0.014) (0.031)
Female gender 0.501 -0.014* -0.001
(0.500) (0.007) (0.015)
Children living in the home 0.506 -0.000 0.016
(0.500) (0.007) (0.014)
Urban residence 0.224 0.009 0.025
(0.417) (0.009) (0.018)
Suburban residence 0.125 0.014 0.006
(0.331) (0.011) (0.022)

54
Town or small city residence 0.306 0.020** 0.010
(0.461) (0.008) (0.016)
Years of full-time education 13.773 0.002 0.001
completed (3.620) (0.001) (0.002)
Age of respondent 42.195 0.003 -0.004
(11.570) (0.002) (0.004)
Age of respondent (squared) 1914.296 -0.000 0.000
(988.958) (0.000) (0.000)
Responsible for supervising other 0.289 0.027** 0.035*
employees (0.453) (0.008) (0.016)
Central or local government 0.124 0.156** 0.265**
employer (0.330) (0.022) (0.049)
Other public sector employer 0.163 0.185** 0.248**
(0.370) (0.021) (0.047)
State-owned enterprise employer 0.070 0.180** 0.234**
(0.256) (0.023) (0.050)
Private employer 0.617 -0.006 -0.093*
(0.486) (0.019) (0.044)
Establishment size under 10 0.232 -0.203** -0.432**
(0.422) (0.013) (0.025)
Establishment size 10 to 24 0.199 -0.166** -0.337**
(0.399) (0.013) (0.025)
Establishment size 25 to 99 0.263 -0.129** -0.282**
(0.440) (0.013) (0.023)
Establishment size 100 to 499 0.173 -0.057** -0.117**
(0.379) (0.014) (0.026)
Major occupation: Legislators, senior 0.210 0.046** -0.015
officials, managers (0.408) (0.015) (0.030)
Major occupation: Professionals 0.165 0.029 -0.014
(0.371) (0.015) (0.030)
Major occupation: Technicians & 0.108 0.039* 0.014
associated professionals (0.310) (0.016) (0.033)
Major occupation: Clerks 0.166 0.013 -0.004
(0.372) (0.015) (0.032)
Major occupation: Service & shop & 0.007 0.001 -0.168*
market sales (0.085) (0.038) (0.075)
Major occupation: Skilled agriculture 0.114 0.036* 0.034
and fish (0.318) (0.016) (0.035)
Major occupation: Craft and related 0.079 0.036* 0.050
trade (0.270) (0.018) (0.037)
Major occupation: Plant and machine 0.076 0.019 -0.019
operators (0.265) (0.018) (0.037)
Major occupation: Elementary 0.006 0.170** 0.229*
occupations (0.077) (0.057) (0.105)
Major industry: Agriculture, forestry, 0.167 0.064* 0.106*
fishing (0.373) (0.025) (0.053)

55
Major industry: Mining and 0.010 0.140** 0.228**
quarrying (0.102) (0.044) (0.079)
Major industry: Manufacturing 0.008 0.078 0.069
(0.087) (0.040) (0.101)
Major industry: Electricity, gas, 0.062 0.028 0.071
steam, a/c supply (0.242) (0.027) (0.056)
Major industry: Water supply; waste 0.116 0.047 0.061
management (0.320) (0.026) (0.053)
Major industry: Construction 0.061 0.072** 0.171**
(0.240) (0.027) (0.057)
Major industry: Wholesale/retail 0.040 0.061* 0.030
trade (0.195) (0.028) (0.059)
Major industry: Transportation and 0.033 0.045 -0.048
storage (0.179) (0.030) (0.061)
Major industry: Accommodation and 0.036 0.056 0.141*
food service (0.187) (0.029) (0.061)
Major industry: Information and 0.006 0.041 -0.054
communication (0.075) (0.045) (0.093)
Major industry: Financial and 0.053 -0.030 -0.127*
insurance (0.225) (0.028) (0.057)
Major industry: Real estate activities 0.044 0.024 0.112
(0.205) (0.028) (0.060)
Major industry: Professional, 0.070 0.002 -0.040
scientific, technical (0.255) (0.029) (0.059)
Major industry: Administrative and 0.105 0.059* 0.142*
support service (0.307) (0.029) (0.058)
Major industry: Public admin, 0.121 0.000 0.057
defence, social security (0.326) (0.027) (0.056)
Major industry: Education 0.017 -0.009 -0.015
(0.130) (0.034) (0.071)
Major industry: Human health and 0.016 0.045 -0.072
social work (0.125) (0.032) (0.064)
Major industry: Arts, entertainment, 0.005 -0.072 0.015
recreation (0.072) (0.042) (0.090)
Major industry: Activities of 0.007 -0.024 0.122
households (0.085) (0.045) (0.098)

Sample size 15,899 15,899 15,899

Notes: Column 1 reports means and standard deviations. Column 2 reports the marginal
effect and robust standard error from a probit model; column 3 reports the coefficient and
robust standard error from a regression. All models are estimated using ESS design weights
Columns 2 and 3 report the marginal effect and robust standard error from a probit model
estimated using ESS design weights.
Statistically significant at the * 0.05 or ** 0.01 level.

56
 Appendix Table 4:
European Social Survey Full Means and Marginal Effects Results (All Rounds)
Mean There are Union Members
(Std Dev) in the Workplace
(1) (2) (3)
Respondent is a Trade Union 0.305 Dependent Variable
Member (1=yes, 0=no) (0.460)
Disproportionality 5.884 -0.009** -0.002**
(4.832) (0.000) (0.001)
Effective number of political 5.181 -0.022** -0.006**
parties (1.626) (0.001) (0.002)
Coalition government 0.835 0.063** 0.003
(0.371) (0.004) (0.005)
Ghent country 0.186 0.323**
(0.389) (0.003)
Ideology of the ruling party or 5.493 -0.135** -0.027**
cabinet (1.472) (0.006) (0.010)
Ideology of the ruling party or 32.335 0.013** 0.003**
cabinet (squared) (15.919) (0.001) (0.001)
Ideology of the legislative body 5.288 1.220** 0.174**
weighted by seats (0.523) (0.050) (0.052)
Ideology of the legislative body 28.231 -0.115** -0.017**
weighted by seats (squared) (5.585) (0.005) (0.005)
Varieties of capitalism: LME 0.138 0.064**
(0.345) (0.006)
Varieties of capitalism: LME- 0.329 0.040**
like (0.470) (0.006)
Varieties of capitalism: CME 0.236 0.005
(0.425) (0.006)
Varieties of capitalism: Hybrid 0.223 0.060**
(0.416) (0.006)
Citizen of country 0.951 0.049** 0.059**
(0.217) (0.006) (0.006)
Belong to minority ethnic group 0.053 -0.012* -0.004
in country (0.224) (0.005) (0.005)
Female gender 0.483 -0.010** -0.004
(0.500) (0.003) (0.002)
Children living in the home 0.498 0.004 0.002
(0.500) (0.002) (0.002)
Urban residence 0.214 -0.023** -0.012**
(0.410) (0.003) (0.003)
Suburban residence 0.126 0.015** -0.002
(0.331) (0.004) (0.004)
Town or small city residence 0.308 0.004 0.012**
(0.462) (0.003) (0.003)
Years of full-time education 13.587 0.003** 0.001**

57
completed (3.640) (0.000) (0.000)
Age of respondent 42.235 0.016** 0.018**
(11.908) (0.001) (0.001)
Age of respondent (squared) 1925.555 -0.000** -0.000**
(1026.963) (0.000) (0.000)
Responsible for supervising 0.322 0.022** 0.015**
other employees (0.467) (0.003) (0.002)
Central or local government 0.293 -0.201** -0.190**
employer (0.455) (0.004) (0.004)
Other public sector employer 0.192 -0.128** -0.117**
(0.394) (0.004) (0.004)
State-owned enterprise employer 0.242 -0.087** -0.076**
(0.428) (0.004) (0.004)
Private employer 0.159 -0.053** -0.048**
(0.366) (0.004) (0.004)
Establishment size under 10 0.203 0.090** 0.085**
(0.402) (0.005) (0.005)
Establishment size 10 to 24 0.166 0.083** 0.083**
(0.372) (0.005) (0.005)
Establishment size 25 to 99 0.098 0.091** 0.082**
(0.297) (0.006) (0.006)
Establishment size 100 to 499 0.161 0.093** 0.077**
(0.368) (0.005) (0.005)
Major occupation: Legislators, 0.018 0.094** 0.075**
senior officials, managers (0.134) (0.013) (0.013)
Major occupation: Professionals 0.119 0.124** 0.125**
(0.323) (0.006) (0.006)
Major occupation: Technicians 0.076 0.125** 0.124**
& associated professionals (0.265) (0.006) (0.006)
Major occupation: Clerks 0.071 0.074** 0.069**
(0.257) (0.006) (0.006)
Major occupation: Service & 0.006 0.171** 0.129**
shop & market sales (0.076) (0.016) (0.016)
Major occupation: Skilled 0.165 0.029** 0.026**
agriculture and fish (0.371) (0.010) (0.010)
Major occupation: Craft and 0.010 0.152** 0.146**
related trade (0.098) (0.014) (0.014)
Major occupation: Plant and 0.007 0.095** 0.089**
machine operators (0.084) (0.016) (0.015)
Major occupation: Elementary 0.071 0.021 0.006
occupations (0.257) (0.011) (0.011)
Major industry: Agriculture, 0.126 -0.013 -0.025*
forestry, fishing (0.332) (0.010) (0.010)
Major industry: Mining and 0.061 0.099** 0.086**
quarrying (0.239) (0.011) (0.011)
Major industry: Manufacturing 0.040 -0.015 -0.028*

58
 (0.195) (0.012) (0.012)
Major industry: Electricity, gas, 0.025 -0.023 -0.042**
steam, a/c supply (0.157) (0.012) (0.012)
Major industry: Water supply; 0.035 0.046** 0.026*
waste management (0.184) (0.011) (0.011)
Major industry: Construction 0.008 -0.007 -0.014
(0.087) (0.018) (0.018)
Major industry: Wholesale/retail 0.059 0.004 -0.017
trade (0.236) (0.011) (0.011)
Major industry: Transportation 0.022 0.031* 0.014
and storage (0.148) (0.012) (0.012)
Major industry: Accommodation 0.066 0.151** 0.135**
and food service (0.248) (0.010) (0.010)
Major industry: Information and 0.098 0.195** 0.177**
communication (0.297) (0.010) (0.010)
Major industry: Financial and 0.118 0.143** 0.120**
insurance (0.323) (0.010) (0.010)
Major industry: Real estate 0.019 0.052** 0.038**
activities (0.138) (0.013) (0.013)
Major industry: Professional, 0.023 -0.009 -0.008
scientific, technical (0.150) (0.013) (0.013)
Major industry: Administrative 0.006 0.022 0.034*
and support service (0.080) (0.018) (0.017)
Major industry: Public admin, 0.007 0.113** 0.086**
defence, social security (0.086) (0.015) (0.015)
Year: 2002 0.077 0.032** 0.064**
(0.267) (0.007) (0.008)
Year: 2003 0.045 0.030** 0.059**
(0.208) (0.008) (0.009)
Year: 2004 0.088 0.004 0.049**
(0.283) (0.007) (0.008)
Year: 2005 0.035 0.015 0.048**
(0.184) (0.009) (0.009)
Year: 2006 0.091 -0.014* 0.036**
(0.288) (0.007) (0.008)
Year: 2007 0.033 -0.016 0.039**
(0.178) (0.009) (0.009)
Year: 2008 0.079 -0.019* 0.029**
(0.269) (0.007) (0.008)
Year: 2009 0.057 -0.056** 0.012
(0.232) (0.008) (0.009)
Year: 2010 0.063 -0.022** 0.033**
(0.244) (0.007) (0.008)
Year: 2011 0.064 -0.024** 0.022**
(0.244) (0.008) (0.008)
Year: 2012 0.075 -0.024** 0.019*

59
 (0.264) (0.007) (0.008)
Year: 2013 0.053 -0.038** 0.005
(0.224) (0.008) (0.008)
Year: 2014 0.067 -0.006 0.027**
(0.251) (0.007) (0.008)
Year: 2015 0.047 -0.032** 0.022**
(0.213) (0.008) (0.008)
Year: 2016 0.080 -0.016* 0.017*
(0.271) (0.007) (0.008)
Country: Belgium 0.043 0.137**
(0.203) (0.012)
Country: Bulgaria 0.018 -0.132**
(0.134) (0.011)
Country: Switzerland 0.048 -0.112**
(0.214) (0.008)
Country: Cyprus 0.012 0.073**
(0.111) (0.014)
Country: Czech Republic 0.044 -0.187**
(0.204) (0.010)
Country: Germany 0.071 -0.128**
(0.256) (0.007)
Country: Denmark 0.037 0.367**
(0.189) (0.008)
Country: Spain 0.038 -0.116**
(0.191) (0.009)
Country: Estonia 0.037 -0.192**
(0.189) (0.010)
Country: Finland 0.050 0.298**
(0.219) (0.007)
Country: France 0.045 -0.179**
(0.207) (0.015)
Country: United Kingdom 0.053 -0.040**
(0.225) (0.013)
Country: Greece 0.023 -0.056**
(0.151) (0.011)
Country: Croatia 0.006 0.029
(0.075) (0.015)
Country: Hungary 0.026 -0.193**
(0.160) (0.012)
Country: Ireland 0.046 0.044**
(0.210) (0.008)
Country: Iceland 0.005 0.489**
(0.072) (0.018)
Country: Israel 0.035 -0.001
(0.184) (0.010)
Country: Italy 0.014 -0.031*

60
 (0.120) (0.015)
Country: Lithuania 0.023 -0.207**
(0.148) (0.015)
Country: Luxembourg 0.009 0.095**
(0.093) (0.013)
Country: Latvia 0.005 -0.098**
(0.073) (0.019)
Country: Netherlands 0.039 -0.055**
(0.194) (0.008)
Country: Norway 0.051 0.169**
(0.219) (0.008)
Country: Poland 0.037 -0.156**
(0.189) (0.009)
Country: Portugal 0.036 -0.154**
(0.185) (0.010)
Country: Romania 0.004 -0.017
(0.060) (0.018)
Country: Slovakia 0.023 -0.142**
(0.151) (0.011)
Country: Slovenia 0.022 -0.001
(0.145) (0.009)
Country: Sweden 0.050 0.274**
(0.217) (0.007)

Sample size 148,296 148,296 148,296

Notes: Column 1 reports means and standard deviations. Columns 2 and 3 report the
marginal effect and robust standard error from a probit model estimated using ESS
design weights.
Statistically significant at the * 0.05 or ** 0.01 level.

61
 Appendix Table 5: Adjusted R2 Changes When Adding the
Three Key Political Systems Measures Last
Regression Specification
Adding Adding Three
All Controls Varieties of Key Political
except Columns Capitalism Systems
(2) and (3) Controls Measures
Dependent Variable (1) (2) (3)

European Company Survey

Trade Union Representation in the 0.301 0.306 0.314

Workplace, 2009 (0.005) (0.008)
Trade Union Representation in the 0.116 0.139 0.173
Workplace 2013 (0.023) (0.034)
Collective Wage Agreement for at least 0.128 0.223 0.276
Some Employees in the Workplace, 2009 (0.095) (0.053)
Collective Wage Agreement for at least 0.089 0.142 0.199
Some Employees in the Workplace, 2013 (0.053) (0.057)

European Social Survey, Round 5

Trade Union Members in the Workplace 0.167 0.182 0.198

(0.015) (0.016)
Trade Union Influence in the Workplace 0.117 0.121 0.140
(0.004) (0.019)

European Social Survey, Rounds 1-8

Respondent is a Trade Union Member 0.255 0.259 0.271

(0.004) (0.012)
Respondent is a Trade Union Member (with 0.311 n/a 0.311
country fixed effects) (0.000)
Notes: Each entry reports the adjusted R2 value for the relevant regression implied by the
dependent variable (row) and independent variables (column). The values in parentheses indicate
the change in the adjusted R2 value compared to the previous column.

62