Are trade unions winners or losers in EU policymaking?

(1)

1

Are trade unions winners or losers in EU

policymaking?

A study of ETUC’s and BusinessEurope’s relative influence in the EU consultation

of social partners under article 154 TFEU

Signe Wolf

Word count: 17 610

Field of study: Political Science Master thesis, 30 credits

(2)

2

Abstract

This thesis investigated the influence imbalance between worker’s and employer’s interest in the EU policymaking. It is assumed that worker’s interests are disadvantaged, and this study is therefore examining the consultation of social partners under article 154 TFEU that considers social policy and can be used as a most-likely case for worker’s influence.

The study is conducted by identifying conflicts between ETUC and BusinessEurope in their consultation replies and compare the different opinions with the Commission’s legislative proposal to find out which organisation that succeeded in influencing the Commission within each conflict.

The results from this shows that each consultation generated between 0 to 6 conflicts and that most of the conflicts were won by BusinessEurope. This strengthens the assumption that there is an influence imbalance in EU legislation that makes the employer’s organisation

(3)

3

1. Introduction

The understanding of influence and which interests win and lose in policymaking is central to the study of politics and democracy. In the study of EU politics, it is of utmost importance since the system of the EU gives a promising opportunity for interest groups to influence policymaking. Studying EU policymaking without the perspective of interest group interests will risk giving an incomplete picture of how policies are produced (Klüver 2013, 1).

A central conflict within EU policymaking is the conflict between workers and employers’ interests embodied in the conflict between the European Trade Union Confederation (ETUC) and BusinessEurope, where the latter represents employers’ organisations. These two groups have fundamentally different interests where the asymmetric integration of EU have favoured the business and employers in Europe more than the workers and trade unions. The asymmetric integration consists of a fundamentally more integration of the economic domain than of the social domain, resulting in a harmonised single market but without a comprehensive social protection for workers (Seeliger & Kiess 2019, 2-3; Alexandris Polomarkakis 2019, 183-184; Bernitz et al 2018, 2).

The different situation for workers around Europe has also made it harder for national trade unions to agree within the ETUC and it is widely assumed that trade unions are structurally disadvantaged compared to employer organizations in EU decision-making (Dølvik & Ødegård 2012, 70; Adamczyk 2018, 179-180). Researcher’s focusing on trade unions in Europe are divided on whether European integration will lead to better or worse situations for workers around Europe (Seeliger & Kiess 2019, 6-7).

This paper will focus on the conflict between employers and workers and investigate the relative influence between ETUC and BusinessEurope in the EU policymaking. Doing this will test the assumption that trade unions are disadvantaged compared to employers in the EU and give a wider understanding of this potential disadvantage.

(5)

5

with a strong upper-class accent” (Schattschneider 1960, 35; Lindgren & Persson 2011, 4-6; Klüver 2013, 1; Kohler-Koch 2009, 55).

If some interests are more influential than others, then the political equality between citizens is deranged and democracy harmed (Dahl 1989, 322-326). Studying the conflicts of interest therefore is necessary to understand democracy within EU policymaking.

This study is measuring the relative influence of ETUC and BusinessEurope in a specific process where the Commission consults the cross-sectoral social partners, European employers’

organisations and trade unions, on issues within the area of social policy relating to workers. This situation is a most-likely case for worker’s influence in policymaking and will therefore give an understanding of ETUC’s overall relative influence. If ETUC is not influential in this process that focus on social policy and where ETUC has a formal entry, then it is unlikely to assume that they will have great influence in any other aspect of the policymaking at EU level.

There are very few studies about the influence of workers and employers in the EU and the consultation of social partners under article 154 Treaty of Functioning of the EU (TFEU) is also insufficiently studied for the understanding of influence. In that way this study is a pioneer study giving some pieces to better understanding workers influence in the EU as well as better

understanding the consultation of social partners under article 154.

Measuring relative influence is conducted by comparing the consultation replies of ETUC and BusinessEurope with the legislative proposal of the Commission to assess the degree of preference attainment. The observations in this study are the conflicts between ETUC and BusinessEurope that are displayed in the comparison of their replies. Their interests in these conflicts are later compared with the legislative proposal to find out which organisation that succeeded in influencing the outcome.

The methodological focus on conflicts gives the study the opportunity to focus on the disagreements between the organisations and measuring influence in these rather than overall influence of the organisation’s opinions.

1.1 Research purpose and question

(6)

6

The guiding research question for this thesis is:

Is the ETUC less successful than BusinessEurope in influencing the legislative proposals presented by the

Commission in the consultation of social partners under article 154 TFEU? If so, in what context does this seem to be the case?

Answering this question will give an important contribution to the study of EU policymaking and contribute to covering the gap of research regarding trade union influence in policymaking at the EU level. The consultation of social partners under article 154 TFEU is an important case to study as it is central for trade unions influence and for the understanding of European

policymaking regarding working conditions.

The conclusions from this study are also important for the understanding of democracy and legitimacy in the EU. Workers in all parts of Europe are affected by the legislation studied in this thesis and to identify what interests that have pushed the legislation in this direction is of

importance to citizens all over the EU. It is also important for the discussion of the trade unions’ relevance at the EU level and as a force in the worker – employer relationship.

1.2 Structure of the thesis

The thesis will first give a brief overview of the role employers and trade unions have in the EU policymaking and the previous research conducted of interest group’s influence within the EU. Then in chapter 4 the hypotheses of the study are presented followed by an account of the methodology and material used in chapter 5. In chapter 6 the results of the study are presented from testing the hypotheses. The results are then concluded and discussed in chapter 7.

2. The role of employers and trade unions in

EU policymaking

2.1 Influencing the European Commission

(7)

7

process have been argued to be most fruitful (Kohler-Koch & Friedrich 2020, 458). Another reason that the Commission is so central in interest group influence is that it is an institution without direct elections with their only citizens input from interest groups (Mahoney 2008, 186). For employer organisations and trade unions, in the EU often referred to as the social partners, the possibilities to affect the legislation through general consultations and participatory channels are complemented by additional channels within the area of social policy. These channels are only for the European social partners and gives them a unique position to influence and in some situations also take part in the formation of legislative proposals (Bir 2019, 77).

The previous studies of interest group influence have focused on the general online consultation from the Commission where the replies are published online for scholars to retrieve. The specific consultation process for social partners have however been sparingly studied and no overall studies on influence within these consultations have been conducted. For the balance of interest between workers and employers the studying of this consultation process is important and the lack of research of it is problematic.

This consultation process is important to study since it is a unique channel for influence regarding legislation affecting the worker – employer relationship. It is also unique in that the social partners themselves are given a possibility to formulate agreements that can become incorporated in the EU legislation. A process that works best if the conditions for negotiations are balanced and could strengthen the importance of European social partners.

2.2 European social dialogue

In 1985 the EU Commission President Jacques Delors invited the leaders of the European social partners ETUC, BusinessEurope (at that time under the name UNICE), and European Centre of Employers and Enterprises Providing Public Services (CEEP) to the famous Val Duchesse in Brussels and started the formalisation of a cross-sectoral European social dialogue and

consultation process (Clauwaert 2011, 170; Bir 2019, 78; Tricart 2020, 71). This was one of the first steps towards the breakthrough on 31st_{of October 1991 when the social partners signed the}

Agreement on Social Policy that made it mandatory for the Commission to consult the social partners about the Commissions initiative on the social domain (Clauwaert 2011, 170;

(8)

8

This marks the start of the European social dialogue and was created at the same time as Delors relaunched the European integration process with the European single market. By creating the European social dialogue, the social partners where involved in the European integration process and was part of the social dimension that many wanted to introduce as a complement to the economic focus of the union (Tricart 2020, 72).

The rules for the Commission to consult the European social partners was then further

institutionalized by its insertion into the Maastricht Treaty in 1993 and has remained through the Amsterdam, Nice and the current Lisbon treaty. There have also been several Commission communications where the rules have further been clarified and complemented (Clauwaert 2011, 170). Today the social dialogue and consultation of social partners is an official part of the EU legislation process and makes it possible for social partners to make significant contributions to the creation of legislation within the social field.

The current legislation of consultation of social partners and European social dialogue is found in the Treaty on the Functioning of the European Union (TFEU) in article 151-154. According to these articles the aim of these processes is to improve the governance in the EU through involving the social partners in both the decision-making stage and the implementation (Bir 2019, 76; Clauwaert 2011, 172; Commission 2016, 6).

At EU-level the European social partners ways to influence can be divided into three parts: tripartite concertation, consultations of social partners, and social dialogue. The tripartite concertation mainly consists of the Tripartite Social Summit where the social partners meet the presidents of the Council and of the European Commission to discuss macroeconomic issues and have a dialogue about the European employment strategy. The consultations of social partners are a way for the commission to take in the opinions of the social partners regarding different policy issues. The social dialogue refers to bipartite (only the social partners) work that stems either from consultations from the commission or from autonomous discussions (Bir 2019, 79).

The consultation and social dialogue are also divided between a cross-sectoral1_{and sectoral level.}

Where the sectoral level is constituted of consultations, committees and dialogues that are specific to workers and employers in a specific sector, and the cross-sectoral level is issues that

(9)

9

affect various or all sectors. In this paper the focus is on the tripartite cross-sectoral consultation of the social partners. This paper will not focus on the social partners own initiatives but is instead limited to the actions that are derived from the consultation process under article 154 TFEU.

2.3 Consultation of social partners under article 154

The consultation process under 154 TFEU implicates that the Commission needs to consult the social partners on the direction and the content of a potential proposal before they can present a legislative proposal within social policy. The consultation is in two stages where the commission in the first stage asks for input on the direction of EU action and if EU action on the issue is advisable. The second stage consultation is mainly on the content of the envisaged proposal (Commission 2016, 6-7).

In both stages the social partners are asked if they want to enter negotiations and come up with their own agreement that would be used instead of a commission’s proposal according to article 155 TFEU. These agreements that are negotiated between the social partners can be

implemented by the social partners themselves or through EU legislation (Commission 2016, 6-7).

There are three prerequisites for launching a formal consultation under article 154: 1. That the aim is a legislative proposal

2. That it is within the policy area of “social policy”

3. That the EU legislative proposal is within the fields relating to workers (defined by 153 TFEU)2

2_{These fields are: workers’ health and safety, working conditions, social security and social protection of workers, protection of}

(10)

10

Graph 2.1: Process under article 154-155 TFEU3

As seen in the graph above, there are three different types of outcomes of the commission consultation: a) that no action is taken, b) that the commission comes up with a legislative proposal, or c) that the social partners reach an agreement. Where the agreement can be either autonomous or incorporated as a directive but that the agreement stage only can be reached given that both employers and workers want to negotiate and can agree on the content.

As agreements needs both parties to agree they have been several failures of negotiations. As the employer organisations did not want to legislate in some areas, they found a compromise in creating autonomous agreements that was not legally binding but would be monitored by the social partners (Tricart 2020, 81).

Reaching an agreement is in these situations in the shadow of hierarchy as the alternative is that the Commission goes forward with their legislative proposal (Smisman 2008, 161). It is

therefore important for the social partners whether they can reach an agreement that is better for themselves than a legislative proposal from the commission. It becomes important here that the Commission is balanced between the two sides different interest as a bias to one side would mean that that side would have ascendence in negotiating another agreement.

(11)

11

Even though consultations between the social partners and the Commission began already in the 1960s, the model of consultation and dialogue according to 151-155 TFEU started in the early 1990s. The first consultation under this system was 1993 about European work councils. Then in 1996 the social partners did after negotiations reach their first cross-industry agreement about parental leave that was incorporated into the “Council Directive 96/34/EC of 3 June 1996 on the framework agreement on parental leave concluded by UNICE, CEEP and the ETUC” (Welz 2008, 386; Bir 2019, 81).

When describing the almost 30 years that these consultations of social partners have taken place the earlier years are often described as the golden years for European social dialogue. Clauwaert has described this in the two periods: 1995 - 2002 as Commission-guided European cross-sectorial social dialogue and 2002 – 2010 as the autonomous European cross-cross-sectorial social dialogue. The main difference between these two periods is that the commission had a more proactive role in the first period where the consultations of the social partners led to agreements incorporated into directives. While the second period is more characterised by fewer

consultations and own initiatives from the social partners. In this second period the agreements where not incorporated into directives but was instead of the autonomous character (Clauwaert 2011, 170).

It is also stated that the tensions between the social partner grew during and in the aftermath of the Eurozone crisis (Tricart 2020, 84). This together with the situation where the Commission decided not to incorporate an agreement between sectorial social partners, the 2012

‘hairdressing’ agreement, challenged the tripartite relations and distrust towards the Commission was exceptional high (Ibid, 73).

(12)

12

The consequence of the trend towards autonomous agreements not transposed to legislation is that they did not result in any significant outcome. This because of the often partial, uneven or limited implementation. Even the existence of an autonomous agreement could have the effect that European institutions and employers that wanted less legislation had an argument to reject legislative initiatives in that area. And with the enlargement and new member states with less connections to social partners the application of these agreements became inadequate (Tricart 2020, 83).

In contrast, the Juncker Commission (2014-2019) did create conditions for closer consultations of the social partners and encouraged the social partners by showing that they were ready to listen to their opinions. It became a restoration of the tripartite social dialogue at the European level (Tricart 2020, 91). Difficulties still existed in the bipartite relations as the employer organisations refused to commit to binding agreements and in that way reduced the European social dialogue to a form of social concertation where the consultation part became the important part. This means that the social partners’ possibility to affect the legislation is restricted to what the Commission allows for. This does not satisfy eager supporters of legislation and threatens to lead to an unbalance within employment legislation between trade unions and employer organisations where the latter have historically been more oppose initiatives leading to legislation (Ibid, 93).

Recently there are many that claim that Europe and the EU is favouring employers and the wealthy in opposition to the less well-to-do employees. This gap, many argue, is widening and pulling the EU apart at the same time as the weakened European Social Dialogue gives fire to the beliefs that there is an imbalance between workers’ and business interest in the EU (Magnusson & Murhem 2018, 144).

The development of the diminishing social dialogue according to 155 TFEU (that social partners own agreements can be incorporated as directives) increases the importance of the consultative part of the process where the partners leave their opinions to the commission before they come up with legislative proposals.

(13)

13

diminishing social dialogue has on the legislative process and for the conflict between workers and employers’ interests.

2.4 Cross-sectorial social partners at the European level

The consultation under 154 TFEU is restricted to representative organisations for social partners at EU level to answer to. There are seven social partner organisations that represent workers and employers at the cross-industry level. The three general organisations are: BusinessEurope, European Centre of Employers and Enterprises Providing Public Services (CEEP) and European Trade Union Confederation (ETUC) (Commission 2016, 4).

Then there are also four organisations that represent certain categories of workers or undertakings that still counts as cross-industry social partners: SMEunited, Eurocadres, European Confederation of Executives and Managerial Staff (CEC), and Eurochambres (Commission 2016, 4).

ETUC represents all workers at the European Union level and has Eurocadres and CEC as affiliated organisation and their representatives participate in the dialogue as part of the ETUC delegation. They have 89 national member organisations in 38 countries (including all EU member states) among these the Swedish organisations SACO, TCO and LO (ETUC 2020). They were established in 1973 and has today over 45 million members (Bir 2019, 77). BusinessEurope is the organisations for European private sector employers. The organisation originates from the aftermath of the second world war and was created by the industrial federations from the six founding member states (Welz 2008, 149). For many years, the organisation was called UNICE (Union des Industries des pays de la Communauté européene) but changed name in 2007 to BusinessEurope, The confederation of European Business. BusinessEurope has 40 member organisations from 35 different countries, and among these the Swedish organisation Svenskt näringsliv (BusinessEurope 2020). They are the biggest, in terms of economic coverage, organisation representing European employers (Bir 2019, 77).

European Centre of Employers and Enterprises providing Public Services (CEEP) is an employer’s association of businesses and organisations within the public sector and local

authorities (Bir 2019, 77). Among their members are the Swedish Association of Local

(14)

14

as the European Union of Craft Industries and Small and Medium-sized Enterprises (UEAPME), has around 67 member organisations and represent employers in small and medium-sized enterprises (SMEs). In 1998 they signed a cooperation agreement with

BusinessEurope that enabled them to take part in the European social dialogue (Bir 2019, 77).

3. Previous studies

3.1. Prior attempts to measure interest group influence in the EU

Studying how the European social partners influence the Commission is part of the broader study on interest groups influence in the EU. This is a field that have been modestly studied and this is probably partly because of the difficulties in measuring influence. It is nevertheless an important field to expand for the understanding of policymaking and as more scholars have investigated influence, the debate about methodological shortcomings and advantages has widened (Dür 2008, 560; Klüver 2013, 13).

Mapping the studies conducted within the frame of interest group influence in the EU shows a diverse field. Besides the more theoretical approaches of categorizing the EU system in

corporatist or pluralist systems, the empirical studies can be divided into four strands according to Eising. These are: 1. Identifying the actors and their access, 2. Analyzing interest group strategies, 3. Analyzing interest groups positions and frames, 4. Measuring interest group success and influence (Eising 2016, 291).

(15)

15

This paper is within the fourth of Eising’s strands which is measuring the success and influence of organisations. As all the three other strands are important to understand the process of lobbying and elements behind influence, they cannot show if influence is reached. Using access as tool to measure influence has also been criticized of several scholars for not being enough to show influence (Klüver 2013, 10; Eising 2016, 298). Also, as this paper examines a consultation process with a fixed population (only specific cross-sectoral social partners can reply), access would not be a suitable proxy to study.

Measuring interest group success and influence is difficult and can be done in different ways. As more scholars' studies influence more methods evolve and Andreas Dür summarizes the most used methods in three broad approaches: process-tracing, the attributed influence method, and assessing the degree of preference attainment (2008, 559).

Process-tracing is a method used broadly within political science and focuses on a specific case and goes more in-depth to find the causal link (Dür 2008, 563). The two main advocates of this approach, George and Bennett (2005), argue that the method is for identifying the intervening causal process between the independent variable (or variables) and the outcome of the

dependent variable. The main strength of this method is that it makes visible the causality of influence while it at the same time is problematic to generalize from the studies and the conclusions become very issue specific (Dür 2008, 563).

This method has been used by Dølvik & Ødegård in their study of the role that ETUC has in the work leading up to the service directive. Their study shows that the ETUC did have an influence in that work and that trade unions can influence EU legislation (2012). However, their study is very specific to the situation of the service directive and to a great extent focus on the interplay with the European parliament.

The second method of attributed influence is often based on a survey where the representative of an interest group can provide self-assessments of their own and other’s influence. This method has been used by Lindgren & Persson, among others, in their study of the REACH regulation (Lindgren & Persson 2011).

(16)

16

different organisations leading to misleading conclusions about influence. For example, organisations that want to show results for their members and strengthen their relevance may exaggerate their influence while organisations that want to avoid the creation of counter-lobbying will try to reduce their perceived influence (Eising 2016, 301).

Assuming that answers could be strategic rather than honest leads to a huge disadvantage for the method and the validity of the results. Even though the respondents answer honestly the method measures perceptions of influence rather than actual influence in that the outcomes of lobbying are not uncovered. In addition to this is the problems that affect all surveys such as interviewer bias and a low response rate (Eising 2016, 302; Klüver 2011, 490).

The third group of studies conducted within interest group influence in the EU is using the method of assessing the degree of preference attainment. Here the researcher compares the outcomes of political processes with the ideal points or positions. The idea behind this is that the distance between the outcome and an organisations position reflects the influence of the

organisation. This method is used in several studies and often over several issues, the most extensive studies are from Christine Mahoney and Heike Klüver and the results from these studies are explained in the next part of the thesis (Dür 2008, 566-567; Klüver 2013, 16; Mahoney 2008).

There are several advantages of this method, it captures all influence through whatever channel it is used and with the organisation's position as yardstick it can sometimes more clearly than process-tracing find influence in the outcomes. It can also be used for a larger number of cases which allows for greater generalizations and studying factors variations. Using a larger number of cases will probably also cancel out errors in the assessment across the cases (Klüver 2013, 15). The drawback that could affect the use of this method is problems in the determination of preferences and controlling for alternative factors that could explain the coincidence of

preferences and outcomes. The first problems occur if the organisation is missing a position or opinion on a specific issue or if the position papers used to assume their position is written strategically. The second problem can be handled in using several cases and in that way

(17)

17

To summarize, there are different studies conducted regarding interest group influence in the EU and the difficulties of measuring influence are seen in the drawbacks of the different

methods used. There is however no existing research that focus on trade unions and employers’ organisations influence moving away from the issue specific setting and giving a broader view. The results from the studies conducted of interest group influence in larger n-cases are presented in the next section.

3.2. Explaining influence with resources and context

There are some scholars that have more directly measured influence either by in-depth studies or by statistically comparing the outcomes of several observations. The advantages of using a study with several cases is that it makes it possible to find patterns and important factors that varies between policies and could help explain why influence is achieved and not.

Going through conducted studies there is a range of contradictory findings. It is partly because of the different methods, definitions and operationalizations made but also a consequence of studying different organisations and policy scopes (Klüver 2013, 2). This opens for more discussions about the methodology and measures of influence but also the need for more research in the area to clarify the contradictions.

From studies that have looked at issues from a broad range of areas some factors that explain influence success has been revealed. These can, according to Christine Mahoney and later also Heike Klüver, be divided into three categories: the institutional context, interest group

properties, and issue-specific factors (Mahoney 2008, 35-36; Klüver 2013, 10).

(18)

18

Looking closer at these three categories of factors and the arguments of their influence show that these three categories are highly intertwined. Given a specific institutional context the possibilities that some interest group properties will lead to higher lobbying success will be higher than in another institutional context. In the same way a specific interest group property will be more useful in policies inheriting some issue-specific characteristic. Theorizing about a factor’s effect on the degree of influence therefore also needs to take regard of the circumstances in the other two categories of factors.

There are relatively few studies conducted on the importance of institutional context and hence information on how this could affect lobbying is low. The factors that have been argued will affect influence is the degree of democratic accountability, the presence of broad-reaching media and the rules of the policymaking process (Mahoney 2008, 36-37). Neither of these have

however been shown to alone affect the lobbying success. When Mahoney compared lobbying in the USA and the EU, she could not find support for that the institutional context alone had affected the outcomes but instead showed that the institutional context affected what strategies the interest organizations used (Mahoney 2008, 193).

Factors about the interest group properties have close connection with the institutional context. Some of the properties discussed are financial and staff resources, membership resources and membership base, and the organizational structure of the organization (Mahoney 2008, 43-44). A word that often is mentioned in this is resources, and that resources are important. But different scholars have drawn different conclusions regarding resources which probably has its foundation in different operationalizations of resources (Klüver 2013, 13).

Different resources have also clearly divergent effects depending on the institutional context. Economical resources are assumed to be more important in situations where policymakers are dependent on financial campaign support, and information (as a resource) is probably more important for institutions where staff and expertise resources are lower.

Regarding interest group properties, a prominent and often discussed hypothesis is that interest groups defending diffuse interest are less influential than groups that represent more

(19)

19

Klüver has developed this in a preference maximizing based theoretical model where influence from the European institutions to interest organizations is exchanged for information, citizen support and economic power. In her model the influence is therefore dependent on both what the specific institution needs and on what the interest organization has to offer. She does also argue that lobbying success in this model varies with the issue context and finds empirically this proven to be correct (Klüver 2011, 483).

Viewing the issue context several issue specific factors has been tested and some are confirmed to have an impact on interest group influence. Mahoney showed that interest groups where less likely to achieve their goals in large-scope issues, issues that affect many people (Mahoney 2008, 204).

Tatyana Todorova also studies influence in EU legislation among several cases. In her study she shows that issue salience is an important factor to explain lobbying success (Todorova 2020, 53). However, she uses attributed influence as method and therefore measures perceived influence. Coming to the conclusion that issue salience is important when measuring perceived influence is problematic since the method in itself is sensitive for salient issues.

Some scholars have showed that more influence is achieved in more technical and complex issues since the decision-makers make use of the expertise and knowledge from interest groups (Klüver 2011, 485-486; Todorova 2020, 49). The idea behind this is that if a policy is difficult to analyze, understand and/or solve then the Commission is in greater need of information and knowledge. A proposal that is highly technical for example requires more resources and creates a wider opportunity for information-influence exchange to compensate for information lack (Klüver 2011, 487-488).

(20)

20

4. Hypotheses and assumptions

To answer the research question four hypotheses are formulated and tested on the data. The first two hypotheses that are presented are directly linked to whether the ETUC has less influence than BusinessEurope, while the second two hypotheses bring knowledge to if this is true for different issues within the consultation.

It is widely assumed that trade unions are structurally disadvantaged compared to employer organizations in the EU decision-making and that there is a fundamental imbalance between economic and social dimensions of the EU. This is based in the asymmetric relationship between how these two dimensions have been handled in the European integration as the economic dimension has been Europeanized to a greater extent than the social dimension (Dølvik & Ødegård 2012, 70; Bernitz et al 2018, 2; Seeliger & Kiess 2019, 1-2).

Using previous research of interest group influence to explain the relative influence when comparing lobbying success of BusinessEurope and ETUC through the consultations of social partners under article 154 TFEU the institutional context is fixed, and the interest group properties only vary over two organizations. To better understand the results this context and relevant properties needs to be analyzed.

Looking at the institutional context and interest group properties we can see that it is not clear who will be most influential. The goal of the Commission is to present proposals for legislation that will be accepted by the European parliament and the Council. From the perspective of Klüver’s model this would mean that the commission want information and legitimacy in exchange for influence.

(21)

21

The conclusions from previous results thereby gives a complex view of the possible influence imbalance between ETUC and BusinessEurope but cannot clearly explain if an imbalance in influence exist and what the reasons for that is. According to the assumption of a structural disadvantage of workers’ interests the first hypothesis for this paper is:

H1: The ETUC has less lobbying success than BusinessEurope in the consultation of social partners under 154

TFEU.

Finding support for this hypothesis would under the assumption of this being a most-likely case imply that BusinessEurope is more influential than ETUC within the EU policymaking.

When studying success in a binary way we also need to check for the degree of success to detect influence bias. If the success for one of the organisations is more compromised in the way that the Commission’s proposal is a combination of both organisations’ interests but more in line with one of them, then the degree of success in that conflict is lower than if the proposal is totally in line with one of the organisations’ expressed preference.

As mentioned earlier in the paper the outcomes in the EU often are of a compromised character and even though a disadvantage or bias cannot be showed with the first hypothesis it can still exist within the conflicts and we therefore test this with a second hypothesis:

H2: The ETUC has more compromised lobbying success than BusinessEurope.

In Binderkrantz, Christiansen & Pedersen (2014) they show results that strengthens the view that business often win but nuance this with showing that business success varies over different policy issues. Their conclusions from these results are that business organisations contributions are characterised by scientific and economic contributions and that they are better equipped to represent their interest in regulatory areas (Ibid, 881). Their findings support the importance of group resources for influence but also show that business influence bias depend on this and therefore will vary when different group resources are needed (Ibid, 893).

(22)

22

Issues regarding occupational health and safety can be rather complex in the sense that they are often based on technical research to be able to set limit values or medical knowledge to

understand how different subjects affect workers health. This requires more resources and expertise from the organisation to influence the details. These issues are also often limited in scope of workers as the sectors that are affected of legislation within electromagnetic fields or carcinogens are narrower than legislation as minimum wage that affect the whole labor market. Issues relating to a larger scope of workers will be important for a larger group of citizens and the citizens support will be more important for the Commission. This together with less technically detailed issues would give ETUC an advantage in social protection issues since the large membership base gives the Commission citizens support. In contrast to technical issues where economic resources to conduct studies and present detailed information about complex issues are an advantage.

Issues within occupational health and safety also differ within the European social dialogue as outcomes of agreements have been very modest and only occurred at the sectoral level mainly on working time or in texts of softer nature. Negotiations have not occurred on the technical

aspects and the social partners themselves say that the technical nature in itself precludes the negotiation part of social dialogue from playing a major role (Smismans 2008b, 889). This does indicate that it is an area that distinguishes itself from other issues within social policy.

This paper will therefore also test the following hypothesis:

H3: The ETUC is relatively more successful on influencing in the area of social protection but not in the area of

occupational health and safety.

Finding support for this hypothesis would strengthen the thoughts about interest group specific resources being important for the understanding of success in different types of issues. It would also give more knowledge about the variation in lobbying success for trade unions and

employers organisations.

Based on the conflict between employers’ economic interests and the workers’ interest of more social policy regulation to protect them in the employment relationship and at their workplace, BusinessEurope should be more sceptical towards introducing further and more comprehensive legislation within the field of social policy than ETUC would be.

(23)

23

BusinessEurope would want to maintain the regulative status quo and in that the uneven integration of the economic and social domains or even decrease social policy legislation. In contrast the ETUC would want to introduce more legislation within social policy to further integrate the social domain in the EU. This paper will therefore also test the following

hypothesis:

H4: The lobbying success of ETUC will lead to increased regulation to a larger extent than the lobbying success

of BusinessEurope.

Finding support for this hypothesis in the results would imply that when ETUC has higher possibility to influence the Commission the results to a higher extent become increased regulation.

5. Method and material

5.1 Case selection

This study tests the relative influence between workers interests and employers’ interests over the social policy of the EU by comparing policy preferences of ETUC and BusinessEurope with the legislative proposal by the Commission.

Studying their relative influence in the area of social policy is interesting as this is a most likely case for trade union influence. ETUC represent the social side of employer – worker relationship and have expertise and an interest in focusing on these issues. These are also some issues that most directly hits the core of worker’s interest. Therefore, we can assume that if the European trade union can affect any policy area it would be within this process.

Studying a most likely case is interesting since the assumption is tested in circumstances that are favourable for finding a specific result (Teorell & Svensson 2007, 154), in this case more

influence for ETUC. This means that if BusinessEurope has more influence than ETUC even in this case where the circumstances are favourable for another result then we can assume that this is the situation for all EU policymaking. Regardless of the results we can assume that there is even less influence for ETUC in the other parts of EU policymaking.

(24)

24

5.2 Ensuring validity: Measuring relative influence on the basis of conflicts

There are several different methods to measure influence and all of them have their advantages and problems. The three most common methods within the measurement of influence are: process-tracing, the attributed influence method, and assessing the degree of preference attainment (Dür 2008, 559). How these methods work and examples of how they have been used in influence studies is described in the chapter of previous research.

There is also a variation within the method of assessing the degree of preference attainment method that is used in this study. Deriving the political preference can be conducted by expert interviews, reading position papers or investigating statements or actions. To compare political preferences with the outcome can also be conducted in different ways either by using a

computer-based coding where the word scores or word distributions are analysed and compared between texts (Klüver 2013, 65-66).

This study has used a variant of assessing the degree of preference attainment where consultation replies have been used as position papers and been compared with the legislative proposals of the Commission by hand-coding the variables.

The concept of influence is in this paper defined in the same way as the influence scholars Klüver and Dür have used it, as: “the ability of an actor to shape a political decision in line with his preferences” (Klüver 2013, intro; Dür 2008, 561). Using this definition, the method of preference attainment approach is most suitable with influence operationalised as success when policy outcomes converges with the expressed policy preferences of an organisation.

To discover this convergence the replies of the social partners in the consultations are compared with the Commission’s legislative proposal that the consultations are preparatory for. This method has been used by Klüver among others when measuring influence. However, as Klüver has explained, the convergence of policy preferences and policy outcomes are only a necessary and not sufficient condition for influence (Klüver 2013, 7-8).

(25)

25

because of other reasons or pure luck. Therefore, the causality between expressed preferences in the consultation replies and the legislative outcomes needs to be addressed.

By looking at the success for several consultations the unsystematic luck and coincidence of expressed preference and policy outcomes are marginalized, and we can assume that the results are of a more systematic character. The results would thus be either because of the organisations’ actions meaning causality or because the Commission has a view that is systematically in favour of specific interests. In this study both results would be interesting.

Another factor that could affect the results are actors not concerned in this study that try to influence the policy outcomes. If there are more influential actors involved, then the results of this study could be because of actions by these actors. However, in the issues studied in this thesis, workers and employers are the main stakeholders and ETUC and BusinessEurope probably holds the most dominant interests within these issues. While national employers’ organisations and trade unions are important actors, it has been showed that the Commission considers opinions of European level organisations to a higher extent than opinions of national organisations (Greenwood 2007, 346-347). So, it is fair to assume that these organisations will not affect the results more than the examined organisations.

A difference between this study and previous studies of influence is that the focus in this study is measuring the relative influence of the two organisations by examining conflicts or

disagreements. Departing from a view where civil society is a sphere of competing interests rather than a view that larger civil society automatically increases democracy, then the dominant or winning interest are more interesting to study than the accumulated influence.

In the area of social policy there is also a direct conflict between the interests of workers and the interests of employers that fortify that the interesting opinions are those that these organisations disagree about. Issues that are not contradicted could still be important to one of the

organisations but are probably not important for the worker – employer relationship.

It is also interesting to look at the conflicts as both organisations have the opportunity to enter own negotiations on the issue. Focusing on the rival interests and conflicts within the

(26)

26

legislative proposal. Therefore, the study of relative rather than substantial influence is interesting.

This is an attempt to study the part of the influence through these consultations that are important for both actors and therefore also what could be hard to agree upon. For this an assumption is that the organisations will in the replies comment on the issues of the proposal that are important for them. As the replies are of that character that the organisations comment both if they are positive and if they are negative toward something then we can assume that if an actor fails to comment an issue it is because that issue is of lower importance to that actor. Studying conflicts of consultations also creates the possibility to study diversity of influence within consultations. Most scholars that have studied influence over several cases has examined the success of the whole proposal or issue and either made this dichotomous or a measure of degree. This is assuming that the proposals are of a single policy dimension, however dividing each consultation into smaller parts could give an understanding of what parts of the

consultations that different interest influence. That Mahoney have concluded that the success within the EU often is a compromised success strengthens the argument that diversity of influence within a consultation is interesting to study (2008, 190).

In this thesis the researcher has examined each proposal/consultation process and distinguished conflicts between ETUC and BusinessEurope within the proposal and by examining the

outcomes in these conflicts studied the success. By doing this it is possible to examine the success at a more detailed level and see if there are patterns between the conflicts in different replies that can explain what part of the issues that the two organisations are stronger within. A hand-coded content analysis has been used and I have analysed the material based on a codebook that is attached to the thesis in appendix A. This is preferable to computer-based quantitative text analysis as it does not assume that word frequencies show policy preferences and can better handle that different terminology is used in the position papers compared with the legislative proposals.

(27)

27

demanding. The method of attributed influence would neither have been suitable to use as it would lead to validity problems with the definition of influence used for this study.

5.3 Data collection

The material used in this study are the replies from BusinessEurope and ETUC to the second stage consultations of social partners and the legislative proposals from the Commission that are connected to these consultations. Since there is no exhaustive list of all consultations conducted at the cross-sectorial level the researcher first had to collect information from different sources to construct a list of the consultations.

The focus on comparing second stage replies are because this phase is where the Commission asks for comments on the content of the legislative proposal. As this study is focused on the content of the proposal the first stage replies are less relevant. However, for some of the

consultations the organisation has referred to comments in the first stage and in those cases this thesis has included the comments in the first stage in the analysis.

The main source for this was the European Commission’s website’s section “Consultations of the Social Partners” as sources refer to this site as an exhaustive list of conducted consultations (Clauwaert 2011, 172). However, to cross-check that all the consultations are listed I also searched through both the ETUC’s website and BusinessEurope’s websites for replies from consultations, and literature about the social dialogue (ETUC 2020; BusinessEurope 2020). From this investigation the list was extended with a handful of additional consultations.

(28)

28

The processing time for handling third part documents from the Commission is longer due to that the third part, in this case ETUC and BusinessEurope, needs to approve that these documents are handed out before the Commission can do so. This meant that some of the replies were collected in the end of November and beginning of December 2020.

The legislative outcomes of the consultations are found by searching through the Commission’s proposals for legislation at the EUR-lex database. In the pretext for every legislative proposal the Commission makes account for the work that precede the proposal and here mention if and when consultation by the social partners have been conducted. In this way every proposal has been linked to the specific consultation. To cross-check that the outcomes are correct I have looked at Commission press-releases and comments on ETUC and BusinessEuropes websites where these exist, and for older consultations in the works of the social dialogue by Welz (2008) and Keller & Bansbach (2000).

Due to teleworking no documents older than 2003 were handed out from the Register of Commission documents since these documents has not been digitalised. This limited the data of this thesis to consultations between 2003– 2020 with a single exception where the replies of one consultation was not found by either the Register of Commission documents or the researcher. During the years, around forty consultations have been conducted at the cross-sectoral level. Among these 23 have ended in a legislative proposal from the Commission, 7 have ended in agreements between the social partners and 10 of the consultations have not led to any form of legislation or agreement. All consultations of social partners at the cross-sectoral level are listed in appendix B in three categories from the outcomes.

From the 23 consultations that ended in a proposal for legislation from the Commission 12 have been analysed in this study. These consultations and the number of conflicts in these

consultations are presented below in Table 5.1. Table 5.1: List of examined consultations

Consultation subject Short

reference

2nd_phase _Legislative proposal

Pages4_Conflicts Certain aspects of the

organisation of working time

Time July 2004 Sept 2004 6+14 6

Simplification of the provisions of the health and safety at work

Simp Dec 2005 July 2006 2+5 0

(29)

29

directives concerning reports on their practical implementation Revision carcinogens and mutagens CMdi 1 May/June 2007 2016 & 2017 4+6 4 Reconciliation of professional, private and family life

PPF July 2007 October

2008

9+26 3

Minimum health and safety requirements regarding the exposure of workers to the risks arising from physical agents (electromagnetic fields)

EMF July 2010 June 2011 7+7 4

Align directives to align regulation on classification, labelling and packaging of substances and mixtures

Align Feb 2011 Feb 2013 3+4 1

Prevention and deterrence of undeclared work

Undec Feb/

March 2014

April 2014 2+2 4

Possible action addressing the challenges of work-life balance faced by working parents and caregivers

Balan Sept 2016 April 2017 6+5 6

Revision of written statement directive

WSD Nov 2017 Dec 2017 9+12 6

Access to social protection for workers and the self-employed

Access 2017 December/ 2018 January 2018 March 9+13 5

Revision carcinogens and mutagens CMdi 2 2017 December April 2018 & September 2020 4+10 3

Minimum wage Wage 2020

September

October 2020

9+17 5

Tot: 47

The table also presents a short reference for each consultation, the year and month of the second phase consultation, the year and month when the legislative proposal was presented, and the number of pages for the consultation replies with pages for BusinessEurope first and for ETUC second. The short reference for each consultation will be used in tables when the full name is too long to refer to.

This collection of proposals are mostly consultations that have ended in a proposal for a

(30)

30

decision for the latter. As showed in an earlier part of this paper this is representative for the whole population of consultations that ended in legislative proposals.

The Table 5.2 compares the examined consultations with all consultations that ended in a legislative proposal. The material is complete with one exception for the time period 2003 – 2020. The single missing consultation replies are from the consultation on a Quality Framework for Traineeships that later ended in a proposal for a recommendation by the Commission. The document register replied that they did not have these replies in their collection and the

contacted persons at either organisation have not been able to find and handle out these. The comparisons show that the examined consultations follow the same division of types of Commission proposals with 10 consultations ended in a proposal for a directive and 2 in other form. The division of policy area for the consultations are however not the same for the

examined consultations as for all consultations. This stresses that if there is a difference between how influence is within these two different types of consultations then that could affect the degree of success in this study.

Table 5.2: Consultations within the time span compared with all consultations

All Examined Consultations 23 12 Ended in directives 19 10 Ended in other form 4 2

Health and safety 8 6

Social protection 15 6

5.4 Creating the data

To measure the influence the researcher created a dataset from the replies and outcomes of the consultations. The creation of the dataset and the variables that are used are presented below and the codebook that I have used is presented in appendix A.

Observations

(31)

31

BusinessEurope disagree with each other about the content of a proposal that the commission is initiating.

First the researcher listed the opinions about the content of the initiated proposal from both organisations separately. Then the researcher compared and paired these opinions to find disagreements. Opinions on an issue from one organisation that the other organisation did not reply to was deleted. This ended in a list of observations (conflicting opinions) for each

consultation.

The units of analysis in this case are the paragraphs where the opinions within the disagreement are expressed. For some observations, the units of analysis are a single paragraph and for others the opinions are expressed in several paragraphs either in direct succession or splintered from different parts of the consultation reply. As several of the replies have a summary in the beginning or end of the reply this increases the number of paragraphs for an observation.

Success

The first variable is to measure which of the organisations opinions that are closest to the Commission’s proposal for legislation. This variable is dichotomous and either measures success for ETUC or success for BusinessEurope.

Here the coder compared the different opinions in the conflict with the corresponding outcome in the legislative proposal and decided which of the opinions the outcome was closest to. In some cases, the outcomes are clearly according to one of the organisations preferences. In other cases, the outcome is more of a compromise where the content is not exactly according to the preferences of one organisation and here the closest preferences is coded as success.

Certainty of coding

A problem of manual hand-coding is reliability (Klüver 2013, 77). As the results should be the same as if someone else replicated the study the manual hand-coding should be systematic in the way that another coder would give the same value to a variable if the study were replicated. As the researcher in this study is a single coder and no cross-check coding has been conducted, an additional variable of certainty has been created. With this variable the coder indicates how certain the coding is and we can control for the certainty in the analysis.

Policy area

(32)

32

variable is not disaggregated into conflict level but are coded after the issue type of the whole consultation making all conflicts within the same consultation having the same value.

Degree of interest agreement

Two variables were constructed to compare the outcome, the legislative proposal, with the preferred position of the organisation within that conflict. These two variables that reflected the interest agreement of the two organisations was then compared to create a new variable that explains if the outcome was a compromise or a clear victory for one of the organisations. Comparing the agreements of the two organisations the coder created a three graded scale of clear success, some compromise, and compromise. Where clear success is a distance of 4 meaning the situation where one organisation has total agreement and the other has total

disagreement. Some compromise is a distance of 3 or 2, and compromise being a distance of 1 or 0.

Increased regulation

This variable examines the proposal from the Commission and looks at if the result of the conflict have increased regulation. The situation of not increased regulation is divided into two categories, decreased regulation or status quo for the situation where the regulations remains the same as before the Commission’s proposal. Using this together with the variable of success shows if success for ETUC means increased regulation and success for BusinessEurope means decreased regulation.

This variable is coded from the documents that are connected to the consultation and no vast investigation has been conducted. However, a background to the legislation is often presented there and when it is a revision of a directive the previous directive is referred to in these documents. In that way the legislative proposal has been compared with previous regulation to create this coding.

5.5 Statistical analysis

(33)

33

Using statistical analysis in social sciences has its problems in that many of the variables are categorical and statistical tools often work better for continuous variables. This study is apart from handling categorical variables also limited in sample size, which could be problematic since some tools need a larger sample and more variation to be able to explain relationships.

Estimation of success

To estimate if the number of success differ between ETUC and BusinessEurope the percentages of the number of success is presented. This percentage is explaining the total population of the specific time span.

Estimation of association between variables

To test the variables relationship this paper will use a measure of strength for association to explain the relationship between variables.

Using contingency tables, we can present the results in both absolute counts and percentages. To estimate the association between the variables, in a contingency table the rows and columns, this study uses the chi-square test (or Pearson’s chi-square test). This is a non-parametric tool that is robust with respect to data distribution and does not require equality of variances among examined groups or homoscedasticity in the data. In comparison with other non-parametric statistics this test provides considerable information about the relationship and allows the researcher to better understand the results (Dahmström 2011, 241-243).

The assumptions of this tool are that the contingency table contains frequencies where the variables are mutually exclusive, that not too many categories are used, and that not more than 20% of the cells have expected frequencies below 5 and no cell have less than 1 expected frequency (Mackridge & Rowe, 2018, 59).

This test measures the deviation between observed and expected frequencies to find out if there is a relationship between the variables and how strong that relationship is. The chi-square test can be written as:

𝜒2 _{= ∑}(𝑂𝑖− 𝐸𝑖) 2 𝐸𝑖 𝑘

(34)

34

Where 𝑂𝑖 is the observed frequency, 𝐸𝑖 is the expected frequency, and k is the total number of cells. If the result of the above formula gives a larger value than there is a relationship between the values (Dahmström 2011, 241-243).

From this chi-square value a coefficient explaining the strength of the nominal association can be derived. The two most commonly used coefficients based on the chi-square value are Phi (𝜙) and Cramer’s V. The formulas for these two are presented below:

𝜙 = √𝜒2

𝑛, 𝑉 = √

𝜒2

𝑛 × 𝑘

Where n is the sample size and k are the smallest of (𝑟𝑜𝑤𝑠 − 1) or (𝑐𝑜𝑙𝑢𝑚𝑛𝑠 − 1). In this study the value of k will be 1 since the variable of success is binary and the results of phi and V will be equal. Both these measures vary between 0 and 1 where a stronger relationship gives a value closer to 1 (Kateri, 2014, 262-263).

A general rule of thumb of interpreting the strength of the association for Cramer’s V and phi is: 0 = No relationship,

< 0.2 = Weak relationship, 0.2 - 0.3 = moderate,

> 0.3 = Strong relationship (Kateri, 2014, 262-263).

There is a version of the chi-square test that some statisticians argue would be better when using small sample sizes where the test is added with Yates correction that is supposed to correct the upper bias that comes with using categorical variables. While some argue that this is an

important correction, others recommend to not use this correction as it is too strict and diminishes the results unnecessarily. The version with Yates correction is presented below:

𝜒2 = ∑(|𝑂𝑖 − 𝐸𝑖| − 0.5) 2 𝐸𝑖

𝑘

𝑖=1

Using and comparing both the chi-square test and Yates version can give a better view of where the correct values are (Mackridge & Rowe, 2018, 61).

(35)

35

a proxy for correlation between categorical variables. However, as the distance is only a proxy and not a measurement of association it is not used in this study.

Another tool that could have been used in this type of analysis is logistic regression analysis. However, in this study the sample size is too low to give reliable results from using a logistic regression analysis since the recommendations for this tool is using a sample size of at least 100 observations and at least 5 observations within each category of used variables. To not end up with unsecure results because of the problems for the logistic regression function to converge due to a small sample size that measurement have not been used in this study.

5.6 Reliability

The reliability of a study is high if the results of the study are consistent and systematically conducted in a way that increases the possibility for a replication of the study to give the same results (Teorell & Svensson 2007, 55). As this study has a single coder, the variable “certainty” was added to complement the coding of the key variable “success”. The result of this is analysed here to assess the reliability of the study.

Presented in the Table 5.3 is a contingency table of the coded certainty. The coding of “success” was not experienced as hard with only 1 observation in the “uncertain” category and most observations in the “very certain” category.

Table 5.3: Certainty of coding and success

ETUC BusinessEurope Total

Uncertain 0 1 1 Rather certain 8 8 16 Very certain 11 19 30 Total 19 28 47

(36)

36

Table 5.4: Chi-square test of success and certainty of coding 𝝌𝟐 Phi/Cramer’s V

Chi-square test 0.487 0.102

Chi-square with

Yates correction 0.151 0.057

The conclusions drawn from this is that the coder was slightly more certain when coding success of BusinessEurope than coding success of ETUC. At the same time, the association is weak, and this indicates that the certainty in coding does not bias the results.

5.7 Limitations

There are some limitations to this study. The main limitation is that this study measures the association of variables but does not investigate the causality of these associations. To examine the causality a different study must be conducted that more deeply studies the connections and process of the consultation and how it affects the decisions taken by the Commission. That the process of causality is not more deeply studied means that the relationship found in this study could even though it is systematic rather show the agreements of the Commission and social partners rather than showing influence.

This study measures the preferences expressed in the consultation replies, this means that even though they could be strategic they are treated as the opinions of the organisations. An

assumption for this study is that the organisations’ expressed opinions in the replies are what their opinions are.

This study does not take into account that the opinions of an organisations could change between the time of the reply and the time that the Commission’s proposal is presented. For most of the consultations the time is so short between these two occasions that it is unlikely that the organisation would change its opinion. The only consultation that has a larger time lapse is the first revision of the carcinogen’s directive (CMdi 1). Here the causality between the

consultation reply and the legislative outcome can be problematized.

(37)

37

seen as low, however these 25 % could be the important parts for an organisation to win and therefore this could still be seen as a positive outcome. Quantifying influence therefore comes with a limitation that affects this study.

This study is also limited to studying consultations that have ended in legislative proposals. This gives a view of the influence on the content of legislation but does not catch the whole view of influence in the consultation process. The earlier stage of influencing whether a legislation will even be up for becoming a legislation is not treated in this thesis.

The focus of this study is on the conflicts in the consultation replies. These conflicts do not represent all conflicts between worker’s and employer’s interests and are also limited by the content of the consultation documents of the Commission.

Despite these limitations the results of this study will still be an important contribution to the understanding of relative influence between workers’ and employers’ interests in the EU. This is a relatively unstudied area and the limitations of this study rather emphasis the need for

additional studies within the area.

6. Results

The results of this study will be presented by examining each of the hypotheses and what the results tells us about these. For each of the hypotheses one or several examples of conflicts will be used to better understand the results and the conflicts presented are mainly from the most recent consultations. Before examining the results of the hypotheses some general remarks are presented.

The number of conflicts between the different consultations vary between 0-6 conflicts for each consultation. The most conflictual consultations were about the organisation of working time (Time), the revision of Written Statement Directive (WSD) and about work-life balance (Balan). All these three consultations are revisions of already existing EU legislation where the two parties have strong opinions.

Are trade unions winners or losers in EU policymaking?