Delegation of authority to the EU agencies authority by the EU legislators

Delegation of authority to the EU agencies

A study to the relation between the autonomy of an agency and the delegation of

Master’s Thesis

Author: Wies Loeffen (S1576364)

Study: Public Administration

Track: International and European Governance

University: Leiden University, Campus The Hague

Supervisor: Kutsal Yesilkagit

Second reader: Sanneke Kuipers

Word Count: 19.089

I. Abstract

In the 1970’s the EU started the creation of independent agencies. The idea behind the creation of these agencies was that expert knowledge could be brought together within such an agency. Currently, more than 40 agencies operate on a range of different topics. The tasks of these agencies differ, from information gathering to authoritative decision-making. This means that there are all kinds of authorities that the official EU institutions, such as the Commission and the Council, delegated to these agencies when they were established through EU legislation. However, some of the agencies have had new delegation that expanded the task they were already granted with when they were founded. It is this form of delegation to independent agencies of the EU that I tried to explain in this thesis. One of the factors that have been previously connected to delegation is the level of autonomy of an organization. The central focus of this study looked to how the autonomy of an agency is of influence on the amount of delegation to that agency. To answer the question, the principal-agent theory was examined to find reasons for delegations. This theory describes that principals are hesitant to delegate authority and that they will build constraints to be able to control their agent when they do choose to delegate. This led to the expectation that when principals delegate, they are more likely to delegate tasks to agencies with a low level of autonomy. To test this expectation, this study used a quantitative method to measure the effect of autonomy on the delegation of tasks to the EU agencies. Data on the delegation of new authority was gathered and added to an existing dataset that measured the autonomy of EU agencies. Next to that, a division into two types of delegation was made. First order delegation points to the addition of completely new tasks, whereas second order delegation is delegation that enhances the instruments to perform already existing tasks. As a control variable the age of an agency and the types of authority (agency tasks) were included. All 25 agencies established and operating in 2009 are included in this dataset (N=25). The results of this study show that a relatively low amount of new delegation was granted to the EU agencies (fifteen in total, spread over thirteen agencies). Furthermore, the formal autonomy of an agency does not portray a significant correlation with the amount of new delegations granted to the EU agencies. This was contrary to the expectation that a lower level of autonomy would be connected to a higher amount of delegation. The assumptions formulated in the theoretical chapter that autonomy would be of influence, are not confirmed.

II. Table of Contents

I. ABSTRACT ... 3

III. LIST OF ABBREVIATIONS ... 6

IV. LIST OF FIGURES AND TABLES ... 7

FIGURES ... 7

1.4OUTLINE OF THIS STUDY ... 11

2. THEORETICAL FRAMEWORK ... 12

2.1LITERATURE REVIEW ... 12

2.1.1 Delegation ... 12

2.1.2 The Theory of Institutional design ... 14

2.2GAPS IN LITERATURE ... 17

2.3FOCUS OF THIS STUDY ... 18

2.3.1 Hypothesis & conceptual model ... 19

3. METHODS ... 21

3.1RESEARCH QUESTION –OPERATIONALIZATION ... 21

3.1.1 Dependent variable: Delegation of Authority ... 21

3.1.2 Independent variable: Autonomy ... 22

3.1.3 Control variables ... 25

3.2DATA COLLECTION ... 26

3.2.1 Measurement of Delegation ... 27

3.2.2 Measurement of autonomy ... 28

3.3DATA ANALYSIS STRATEGY ... 28

4. THE EU AGENCIES – BACKGROUND ... 29

4.1AGE OF THE AGENCIES ... 30

4.2THE AUTHORITY OF THE AGENCIES ... 31

4.2POLICY FIELDS OF THE AGENCIES ... 32

5.1.1 First order delegation ... 35

5.1.2 Second order delegation ... 36

5.1.3 Delegation & age of the agency ... 38

5.2ORGANIZATIONAL AUTONOMY ... 39

5.3DELEGATION & AUTONOMY ... 42

6. ANALYSIS PART 2: INFERENTIAL STATISTICS ... 43

6.1CORRELATIONS ... 43

6.1.1 Correlation dependent and independent variables ... 46

6.1.2 Correlation dependent and control variables ... 47

6.2LINEAR REGRESSION ... 48

6.2.1 Simple regression ... 48

6.2.2 Stepwise regression ... 49

6.3INTERPRETATION AND DISCUSSION OF RESULTS ... 51

6.3.1 Delegation to the agencies ... 51

6.3.2 Relationship between autonomy and delegation ... 52

7. CONCLUSION AND DISCUSSION ... 54

7.1CONCLUSION ... 54

7.1DISCUSSION ... 56

8. LITERATURE ... 58

APPENDIX A. EXAMPLES OF CODING EU AGENCY REGULATION ... 61

APPENDIX B. RECODING OF AUTONOMY VARIABLES ... 63

APPENDIX C. ASSUMPTIONS FOR REGRESSION ... 65

III. List of Abbreviations

AMOCEB European Agency for the Management of Operational Coordination at the External Borders of the Member States of the EU

CECA Community Fisheries Control Agency

CEDEFOP European Centre for the Development of Vocational Training

CEPOL European Policy College

CPVO Community Plant Variety Office

EAR European Agency of Reconstruction

EASA European Aviation Safety Agency

ECDC European Centre for Disease Prevention and Control

EDA European Defence Agency

EEA European Environment Agency

EFSA European Food Safety Authority

EMA European Medicines Agency

EMCDDA European Monitoring Centre for Drugs and Drug Addiction

EMSA European Maritime Safety Agency

ENISA European Network and Information Security Agency

ERA European Railway Agency

ETF European Training Foundation

EU European Union

EUISS European Institute for Security Studies

EUROFOUND European Foundation for the Improvement of Living and Working Conditions

EUROJUST The European Union’s Judicial Cooperation Unit

EUROPOL The European Police Office

EUMC European Monitoring Centre on Racism and Xenophobia

EUSC European Union Satellite Centre

OHIM Office for Harmonization in the Internal Market (Trade Marks and Design)

IV. List of figures and Tables

Figures

2.1: Conceptual model 20

4.1: The years of establishment of the EU agencies 30

5.1: Years of establishment & delegation to the EU agencies 38

5.2: Autonomy & delegation to the EU agencies 42

Tables

3.1: Operationalization of delegation 22

3.2: Operationalization of formal autonomy 23

3.3: Operationalization of control variables 26

4.1: Authority of the agencies 31

4.2: Agencies categorized 33

5.1: Delegation to the EU agencies 34

5.2: Autonomy of the agencies 40

6.1: Gamma-correlations 44

6.2: Simple regression: second order delegation & executive appointment procedure 49 6.3: Simple regression: second order delegation & authoritative decision-making 49

1. Introduction

The European Union (EU) does not solely exist of one organization, but can be seen as a collection of institutions operating within one system (Shanks et al., 1996). The European Commission, the Council of Ministers, and the European Parliament together possess the legislative power of the European Union (European Union, 2015a). The executive power of the EU lies in the hands of the European Commission (European Parliament, 2015). However, the EU legislators have created other institutions, the EU agencies, to which they delegate some of this executive power. Due to the creation of these agencies, part of the executive function has moved from the Commission to these independent agencies (European Union, 2015b). In the 1970s, the European Union started with the creation of these agencies. They were designed to operate relatively independently from the Commission. The agencies were mostly invented as a means to expand the knowledge and expertise of the EU and to provide information regarding a variety of topics (Christensen and Nielsen, 2010). The creation of these agencies introduced a new set of actors that further increased the decision-making actors in the EU (Busuioc, Curtin and Groenleer, 2011). However, it is not yet fully known what factors cause the delegation of authority to these agencies by the EU legislators. The agencies are not fully capable of shaping their own policy; they are dependent on the EU legislators to delegate authority to the agencies. It is this topic of delegation of authority and the explanatory factors behind this delegation that is central to this study.

Currently, there are over 40 EU agencies operating on a ranch of different topics, ranching from food safety (EFSA) to the European Police Office (EUROPOL) (European Union, 2015b). All the agencies are established by regulation. This founding regulation can be seen as the first form of delegation of authority from the EU legislators to these agencies. However, some agencies have had new tasks delegated to them, after they were already established. Recent events or changes in the environment of both the EU and of the individual agencies can create a need for the EU to change some of the tasks of the agencies and to extend the authority delegated to these agencies. For example, the growing threat of terrorism caused the Commission to expand the task of EUROPOL in order to enhance their capabilities in combating this terrorist threat (Council Decision of 3 December 1998). This addition of tasks can vary from the creation of totally new tasks that were not part of the scope of the first regulation of the agency, or they can expand the authority of agencies on tasks that were already a task since the start of this agency. What factors can explain the additional delegation of tasks to EU agencies, after they have been formally founded?

Different scholars have tried to explain the delegation of authority from a legislator to an executive (Epstein and O’Halloran, 1999; Bendor, Glazor and Hammond, 2001). There are different elements that influence the likelihood of a legislator delegating authority to an (executive) agency (Epstein and O’Halloran, 1999). In this study, the institutional design (autonomy) of an agency is linked to the delegation of authority to that specific agency. All agencies are established as officially independent from the EU (Christensen and Nielsen, 2010). However, they are created and supervised by the official EU institutions, such as the European Commission and the Council of Ministers. Although they are all officially independent from the EU, the agencies differ in their levels of autonomy, with some having a relatively high amount of formal autonomy and some having low amounts of formal autonomy (Christensen and Nielsen, 2010; Wonka and Rittberger, 2010). The autonomy of an organization is part of the institutional design of an organization. Autonomy of an agency can be seen as “the capacity to manage your own affairs, acting and deciding unbound by the preferences and interests of their principals.” (Busuioc, Curtin and Groenleer, 2011: 850). Depending on their level of autonomy, an organization is either more or less independent from their principal. Various studies have looked at the autonomy of EU agencies (Busuioc, Curtin and Groenleer 2011, Christensen and Nielsen, 2010, Wonka and Rittberger, 2010). These studies will be further discussed in the theoretical chapter of this study. It is this factor of autonomy that will be used in this study to address the issue of institutional design. While other studies that focused on the EU agencies have looked at the relation between autonomy and authority, or at the relation of autonomy and accountability, this study uses the concept of autonomy to look at the delegation of tasks to the EU agencies, after the agencies have already been established.

1.1 Central question

The goal of this study is to provide insights into the way the authority of the EU agencies has developed after they have been founded and to identify whether there is a relationship between the autonomy of an agency and the tasks/authority delegated to an agency after its establishment. To accomplish this goal, I will gather data regarding the delegation of authority to EU agencies and analyse this data together with data regarding the autonomy of the agencies. The autonomy of an agency is linked to the delegation of legislation to these agencies, in order to gain knowledge regarding possible explanatory factors for legislators to delegate authority to agencies. This research goal leads to the following central question:

‘To what extent does the autonomy of EU agencies influence the authority

delegated to the EU agencies after they have been established?’

1.2 Theoretical Relevance

Although quite a lot of scholars have focused on the topic of delegation and on the topic of EU and EU agencies, there are still interesting theoretical questions to be examined. This study therefore contributes to the body of knowledge related to delegation and the body of knowledge related to EU agencies.

First of all, agentification is something that is happening both on a national as well as on an international level. Agentification refers to the phenomenon of agencies that are being placed at a distance of the formal government institutions (Wonka and Rittberger, 2010). This phenomenon forms an interesting study object. What reasons make legislators create these institutions and what makes them delegate power to these institutions? Adding knowledge to this topic enhances our knowledge in both the field of delegation of powers as in the field of EU studies. There are multiple scholars that have studied the question of delegations from legislator to executive institutions/independent agencies (Epstein and O’Halloran, 1999; Bendor, Glazor and Hammond, 2001). Next to that, these scholars used multiple theoretical approaches trying to explain the reasons behind this delegation. To enhance the knowledge regarding the practical fitness of some of this theoretical knowledge can enhance existing theoretical knowledge. Furthermore, most of these studies on delegation focused on the American governmental system (Epstein and O’Halloran, 1999). This study expands knowledge, by applying and combining existing insights on the European governmental system. The specific way the European system is shaped might lead to new insights into factors explaining delegation of authority.

Furthermore, existing studies have focused on investigating the autonomy of EU agencies (Busuioc, Curtin and Groenleer 2011, Christensen and Nielsen, 2010, Wonka and Rittberger, 2010). Using their findings and data as a starting point, this study combines this concepts of autonomy to the delegation of authority. Where Christensen and Nielsen (2010) tested the relation between autonomy and authority of the agencies, this study slightly changes the approach by not focusing on the level of authority received at the founding of the agency, but at the question whether authority changes throughout the existence of the agencies. This contributes to existing literature, by adding insights into factors that explain delegation in this special governmental system of the EU and by testing the relation between the concepts of autonomy and delegation.

1.3 Societal Relevance

Although this study is mostly aimed at contributing to scientific knowledge related to autonomy, delegation and authority, this study also has some practical/social implications. The amount of EU agencies has been growing over the years, leading to a total of over 40 agencies (European Union,

also on a national level the number of agencies is rapidly growing (Geradin, Muñoz and Petit, 2005). This phenomenon of agentification seems to be happening in multiple governmental systems around the world (Wonka and Rittberger, 2010). Therefore, not only for policy makers in the European Union, but also for policy makers on a national level, knowledge regarding the working of agencies and their autonomy can be valuable for future decisions. This study can function as an overview of how the authority of the EU agencies has developed and has changed over the years, which can be valuable information for future policy decisions related to agencies and their authority.

1.4 Outline of this study

This chapter has formed the introduction to this study. The following chapter will focus on the theoretical framework. I will present a literature review, point to gaps in literature and make the focus of this research more concrete by formulating hypotheses and a conceptual model. Chapter three presents the methodological framework of this study. This chapter consists of a description of the design of this study and the methods and techniques used. Next to that, attention is paid to the reliability and validity. The first three chapters can be seen as the first part of this study, providing the body and knowledge in order to perform the analysis. The analysis forms the second part of this study. I have divided the analysis into three parts. The first part of this analysis, chapter four, focuses on providing background knowledge on the population of EU agencies. After presenting this information regarding the characteristics of the agencies, chapter five consists of the first part of the analysis, in which the descriptive statistics of all relevant variables are discussed. The expectations formulated in the theoretical chapter are further tested in chapter six. This chapter consists of the inferential statistics, presenting a quantitative analysis of the relation between delegation of authority and autonomy of the EU agencies. In this chapter, I will also discuss the findings and elaborate on possible explanations for these findings. Finally, in chapter seven, the conclusion will follow and I formulate an answer to the research question. This chapter also focuses on the discussion, in which I elaborate on both limitations to this study as well as on suggestions for further research, based on the findings of this study.

2. Theoretical Framework

In this second chapter I discuss existing literature and theories that have addressed (parts of) the question that is central in this study. The goal of this chapter is to specify the relationship of this study with existing literature and theory. This provides a provisional answer to the thesis question, resulting in a conceptual model and hypothesis in order to further test these assumptions in the rest of this thesis. I will start with presenting a literature review in section 2.1. This literature review focuses on describing existing literature within the topic of delegation to agencies. Next, in section 2.2, the gaps in literature will be mentioned. This section focuses on the knowledge that is still missing in the literature in order to answer the question central in this study. The third part of this theoretical chapter addresses the expectations and hypothesis central to this study and results in a conceptual model.

2.1 Literature Review

The EU agencies cannot determine their own policy and tasks. They need their principals, the Commission or the Council, to delegate authority to them. The definition of delegation used comes from Curtin (2005: 91): “an authoritative decision, formalised as a matter of public law, that (a) transfers policy making authority away from the established, representative organs (those that are directly or indirectly elected, or are appointed by elected politicians) to (b) a public non-majoritarian institution.” A non-majoritarian institution refers to an organization whose budget comes from public money, with a public function, but with a certain level of independence from elected politicians (Ibid.). According to this definition, EU independent agencies would qualify as a non-majoritarian institution. To determine what is already known in literature regarding delegation of authority to agencies and the relation with the autonomy of an agency, it is important to deliberate on previous scholars who paid attention to this topic. Recent literature has showed an increased interest in the problem of delegation (Bendor et al., 2005). Different theoretical approaches try to explain the logic behind delegation (Ibid.). In the next two sections I explain two of the common approaches used to explain delegation to public actors (Epstein and O’Halloran, 1999).

Transaction costs politics theory

One of the theories that help to explain the concept of delegation and reasons for legislators to delegate authority is the transaction costs politics theory (Epstein and O’Halloran, 1999). The transaction costs theory is a theory used to explain economical transaction between different parties, but Epstein and O’Halloran (1999) argue that it can also be used to explain the decision for a

Congress can either decide to delegate some authorities or that policy can be made by Congress itself (Ibid., 1999). The legislator will weigh the cost of either making policy itself or delegating authority to an agency. They argue that for ‘informationally intense policy areas’ the Congress will decide to delegate, since these areas are often dependent on expert knowledge, which is present within agencies. Furthermore, most often the public and the media will not pick up these kinds of topics, unless something goes wrong. This makes it attractive for Congress to delegate this authority. In contrast, Epstein and O’Halloran argue that distributive policy issues will be most likely be made by Congress itself. So, the first part of the argument why legislators choose to delegate authority has to do with the topic of the issue area, on which the legislator will base its decision to delegate. The more informationally intense a policy area, the higher the chance the legislator will choose to delegate authority to an agency, where expert knowledge is available.

Principal-agent theory

Related to the arguments formulated by Epstein and O’Halloran (1999) for delegation and the transaction costs politics theory, is the principal-agent theory. A principal (in a political study, most often a legislator) can choose to delegate authority to an agent (the executive, or in this study, an agency). A principal is defined as: “those institutions that use their authority to establish non-majoritarian institutions through a public act of delegation”, whereas an agent is defined as: “those who govern by exercising delegated powers.” (Curtin, 2005: 92). The principal-agent theory includes different theoretical aspects. First of all, it is based on the rationalist assumption that actors involved act in a way that maximises their own benefits (Magnette, 2005). As long as the benefits outweigh the costs, a principal will therefore not delegate authority. Next to that, the functionalist assumption included is that “the nature of an agent is defined by the function it is supposed to perform” (Ibid., 2005: 5). According to Majone (2001) the two most important reasons for delegation are the enhancement of the credibility of policy commitment and the reduced costs of decision-making, for example due to the use of expertise of the agent. Thatcher and Stone Sweet (2002: 4) formulated four common rationales that are related to the reasons described by Majone (2001). Principals will delegate authority to an agent in order to:

• Help them “resolve commitment problems.” This means that agents can perform work that will improve the credibility of the principal;

• Solve information asymmetries, mostly in technical areas of governance. An agent can gather expertise and information necessary to make public policy, thereby enhancing the efficiency of decision-making and reducing the decision-making costs.

• Improve the rule making efficiency, since agents are able to address specific problems and issues, whereas principals can deal with more general aspects of policy;

• Delegate unpopular policy areas and policy goals, to avoid having to take blame for unpopular decisions.

However, it is often assumed that principals (governments) generally prefer to stay in control over policy themselves, instead of delegating it to an independent agency (Magnette, 2005). When governments choose to delegate, they are sooner inclined to delegate to a ‘weakly independent agency than to a truly regulatory body’ (Magnette, 2005). Delegation of authority to an agency often involves designing control mechanisms to make sure the principal has some level of control over the agent. If this control fails and if agencies actions deviate from the preferences of the principal, this results in agency losses. Kiewiet and McCubbins (as cited in Pollack, 1997) describe the two problems that principals might face when the preferences and actions of agents are different from the wishes of the principal, which can result in agency losses. First, bureaucratic drift or ‘shirking’ happens when agents act opportunistically and take action according to their own interest (Ibid.). Second, the problem of ‘slippage’ refers to the emergence of perverse incentives for the agent to behave differently from the preferences of its principal, due to the way the delegation is structured. The principal can use certain ‘administrative and oversight procedures’ to limit the chance for agency losses. Administrative procedures are used ex ante and they deal with formulating the scope of the agency and the creation of rules and procedures for the agency (Ibid.). These procedures are often aimed at defining “the agent’s mandate as narrowly as possible (Magnette, 2005: 11). Oversight procedures, on the other hand, are ex

post procedures, enabling the principal to monitor and to influence the behaviour of the agency

through the use of positive and negative sanctions (Pollack, 1997). These oversight procedures allow the principal to restore some of the information asymmetry existing between principal and agency. However, Kiewiet and McCubbins (as cited in Pollack, 1997) argue that these procedures can be costly themselves and that a consideration needs to be made to create a balance between the risks of agency losses and the costs of these administrative and oversight procedures. The creation of oversight and administrative procedures are related to the institutional design of an agency. In the next section, I will elaborate further on this concept of institutional design.

2.1.2 The Theory of Institutional design

The institutional design of an organization is often seen as being an influencing factor for delegation (Pollack, 1997: 103). Broadly speaking, the theory of institutional design focuses on the principles that are desirable when designing institutions and organizations (Goodin, 1996.). These choices for the design of an institution have consequences for the functioning of such an institution later on. According to Goodin (1996), there are some general principles that are related to institutional design. First, revisability, a concept that relates to the flexibility of institutions. Institutions need, to a certain extent, be able to adapt itself to the ever-changing society. Second, robustness, which relates to the fact that an institution needs to “adapt to new situations only in ways that are appropriate to the relevant respects in which the situations are new” (Goodin, 1996, p. 41). Institutions should be robust,

motivational. People have all kind of motives, both self-interest as altruistic, that make them do the

things they do. Fourth, publicity, relates to the question how one should ‘frame’ the institution. An institution needs to be: “at least publicly defensible” (Ibid. p. 42). Lastly, he names variability, which focuses on the fact that different kinds of institutional design should be ‘tested’. Experimenting, altering and developing institutional design throughout different situations can help create insights in how institutions work and how they can be improved. Galbraith (1977, p. 5) defined organization design as “the search for a fit between the strategy/goals of the organization, the organizing mode of the organisation and the people who do the work.”

This general explanation of the theory of institutional design shows what general principles can be used to (effectively) design institutions. These principles can be kept in mind when designing or changing institutions. But how then, is an effective institution practically designed? The theory of institutional design by Goodin (1996) describes the ideological principles behind institutional design, but the practical filling in of this design is not yet discussed. Depending on the type of institution, the design can very greatly. In the next section I therefore focus on an aspect of institutional design that is relevant for most types of institutions, namely the organizational autonomy of an institution/organization.

Organizational Autonomy

Looking at the principles for effective institutional design by Goodin (1996), described in the previous paragraph, it seems that these principles are often related to flexibility and the possibility to change and adapt. One of the elements that in practice can shape this flexibility is the level of autonomy of an organization. With autonomy I refer to: “the capacity to manage your own affairs, acting and deciding unbound by the preferences and interests of their principals.” (Busuioc, Curtin and Groenleer, 2011: 850). The autonomy of an organization is a design question. If you look at agencies from a principal-agent perspective, an agency can be defined in terms of autonomy. The autonomy of the principal-agent increases when the degree of independence from its principal increases. However, organizational autonomy is a broad concept that entails many aspects. Within the existing literature on organisational autonomy scholars have used measured different parts of this concept. For example, one can look at whether the organization is financially independent, legally independent or has independence on the policy they make (Verhoest, Peters, Bouckaert and Verschuere, 2004) One of the types of autonomy that are often studied in relation to the EU agencies, is the formal autonomy of an organization. This is the type of autonomy that also forms the central point of attention in this study.

Various studies have looked at the formal autonomy of EU agencies (Busuioc, Curtin and Groenleer 2011, Christensen and Nielsen, 2010, Wonka and Rittberger, 2010). Christensen and Nielsen (2010) have provided a systematic insight in the autonomy of the EU agencies. Instead of using the method of

a qualitative case study, as most studies have done, they performed a quantitative analysis on the total population of EU agencies up until 2009. These authors divided formal autonomy into two dimensions: structural autonomy and legal autonomy (Christensen and Nielsen, 2010). Structural autonomy can be seen as: “The insulation of the agency from respectively Commission and member government control.” (Christensen and Nielsen, 2010: 177). Legal autonomy relates to the “authorization of the agency to act in its own capacity, and ranges from purely informational tasks to binding decisions.” (Ibid., p. 177). I discuss the exact way to measure formal autonomy in the operationalization in the method chapter.

2.1.3 Autonomy & Delegation

In the previous paragraphs the concepts of delegation and autonomy were explained, through an examination of existing literature. However, scholars have also combined these two concepts in previous research. To describe the relationship between autonomy and delegation, the principal-agent theory can again be used. The idea behind this theory is that agents are granted with a certain level of authority/discretion and that in exchange for this discretion they provide certain tasks for the principal (Epstein and O’Halloran, 1999). The level of independence, the level of autonomy, then depends on the authorities/tasks that have been transferred to the agent and the extent to which the agents themselves can decide how to handle certain tasks. The more autonomous the agent, the less it is bound by restrictions imposed by the principal on how to fulfil its tasks. However, agents are still dependent on the principal to determine what kind of tasks they may deal with and what kind of tasks are outside the scope of the agent. Principals can choose between drawing up long and detailed legislation, which leaves very little discretionary space for the agent and therefore makes the principal capable of “micromanaging the policy process” or to leave the legislation vague and short, which leaves a lot of space for the agent to determine how it will fulfil its task (Huber and Shipan, 2002: 2). Autonomy and delegation are therefore inevitably connected with each other. Without at least a minimal level of autonomy, delegation is not possible. And the autonomy of an agency implies that there is some sort of delegation to this agency, because without delegation there would be no autonomy. What then makes up the relation between these two concepts?

There seem to be different interpretations of the relationship between autonomy and delegation by previous scholars. First of all, when you relate autonomy with discretion, there seems to be a positive relation between delegation and discretion. For example, Epstein and O’Halloran (1999) positively associate discretion with delegation. Discretion can be seen as the moving space agencies have to determine the exact content of the policy. Thatcher and Stone Sweet (2002: 4) describe discretion as: “(a) the sum of delegated powers granted by the principal to the agent, minus (b) the sum of control instruments available for use by the principals.” If legislators don’t fill in a lot of details of the

discretion is, according to Epstein and O’Halloran (1999), related to a higher level of delegation. Put differently, delegation is only possible through transferring discretion from a principal to an agent (Thatcher and Stone-Sweet, 2002). The concept of autonomy can be related to the concept of discretion, since autonomy deals with the “the capacity to manage your own affairs, acting and deciding unbound by the preferences and interests of their principals.” (Busuioc, Curtin and Groenleer, 2011, p. 850), which has similarities with discretion. This leads to the assumption that autonomy, just like discretion, is positively related to delegation.

Secondly, a different approach can be found in literature. As was mentioned in paragraph 2.1.1, legislators are expected to be hesitant in delegating authority. Autonomy can therefore be constrained by placing both ex ante and ex post place control mechanisms on the agency (Pollack, 1997). Tallberg (2002: 28) argues that the decision to delegate in essence comes down to a consideration between two choices: the choice what powers to delegate and the choice for the control mechanisms that need to be designed. Since, as has been mentioned, legislators are hesitant to delegate due to the loss of control, a principal is likely to set up certain mechanisms to keep its agents under its control, when it does decides to delegate (Magnette, 2005). These mechanisms can be seen as limiting the autonomy of an agency, since they increase the ties an agency has with its principal. Christensen and Nielsen (2010) found that the level of constraints placed on an agency increases when the authority of an agency was seen as higher. The formal agency design, measured through formal autonomy, showed an inverse relationship with the delegation of decision-making authority to EU agencies (Ibid.).

2.2 Gaps in literature

In the previous paragraph, I discussed the parts of the puzzle that we already know. In this paragraph, I look at what we don’t yet know about the research question. Describing these gaps in existing literature functions to create a specific focus of this study and it shows how this research can contribute to existing literature. A large part of the literature on which theory on delegation in the public sector is based focused on ‘classical’ relations of delegation between Congress and agencies in the American governmental system. As Epstein and O’Halloran (1999) describe, there exists a large spectrum of different governmental systems. Testing the predictions of Epstein and O’Halloran in the particular constellation of the European governmental system is necessary to improve the theoretical and empirical knowledge regarding delegation to agencies. Next to that, the studies by Pollack (1997), Kiewiet and McCubbins (1991) and Epstein and O’Halloran (1999), on which much of the studies on delegation by public actors are based, were all written a relatively long time ago. Applying these insights to a more recent study is useful to test whether some of the assumptions have changed in a practical environment.

Moreover, there seems to be a kind of tension involved in the relation between autonomy/discretion and delegation, with scholars reporting both possibilities for a positive and a negative relationship. However, the specific relationship between autonomy and the amount of new delegation to already existing agencies and the exact direction of this relationship between these two concepts is of interest and has not yet been examined before, creating the need for more specific testing of this relation in empirical research.

Lastly, the most important gap in existing literature is that there are not yet insights into the way the authority of agencies changes after their establishment. Although the EU agencies have been researched quite extensively, on all kinds of topics, existing studies have not focused on changes in delegation from a principal to an agent. Mostly, studies have taken into account only the founding regulation of an agency to look at the delegation of tasks to agencies. But what happens after the agencies have been established? Do the authorities of these agencies increase while they are operating? And if so, what can explain why some agencies receive more authority than others? Therefore, in this study to EU agency delegation, I look into already existing agencies, which means that the overall, broad, topics of delegation are already known, since each of these agencies have their own specific goal and topic. With the establishment of each of these agencies, they get some authority delegated to them. The changes in their authority after their establishment are not yet researched. In their study, Epstein and O’Halloran (1999) analysed 257 laws to analyse delegation of agencies. They found out that in some cases this delegation was done by the creation of new agencies, but around half of the time authority was delegated to agencies that already existed. This means that after agencies get established, their authorities can change and grow. Since the EU governance system is different from the system in the USA, one can question whether this will be the same for the European Union. This study tries to address this gap in existing literature.

2.3 Focus of this study

After describing existing literature on delegation and of autonomy of agencies and pointing towards gaps in this literature, the next part of this theory chapter is dedicated to the specific focus of this study. I expect that, based on the literature available, the design of an agency (in this case of the EU agencies) will be of influence on the amount of delegation that is delegated to them. The EU agencies get tasks delegated to them by the EU legislators (the Commission and the Council) and depend on these legislators regarding their institutional design. Will the level of autonomy have an effect on the levels of delegation? If this is the case, then the analysis will show that there is indeed a significant relationship between the level of autonomy and the amount of delegation.

Instead of focusing only on the founding regulation and authority delegated to the agency, in this study I focus on delegation granted to the agency while they were already operating. This makes it possible to make an overview of how the tasks and authorities of the agencies have changed over the years and this can be seen as a contribution of this study to existing literature.

2.3.1 Hypothesis & conceptual model

Combining the transaction costs politics approach and the principal-agent approach, one can see that a principal will only choose to delegate when there is a clear benefit for the principal. As described in the previous paragraphs, different theoretical assumptions can be made regarding the relationship between autonomy and delegation. When looking at autonomy as in discretion, a positive relation can be expected, since delegation implies handing over some form of discretion regarding the policy choices and content made by agencies (Epstein and O’Halloran, 1999). However, when looking at formal autonomy by looking to the constraints placed on an agency, one can also expect a negative relationship, due to the likelihood that the more a legislator delegates, the more it will try to remain oversight on that agent it delegates to (Ibid.). Following the relationship Christensen and Nielsen (2010) found when focusing on autonomy and authority of EU agencies, I expect that the higher the possibility to control and influence the agent, the more likely a principal will be to delegate authority to that principal, since this leaves the principal with a higher level of control on this agent. This leads to the assumption that the level of autonomy of an agent and the amount of delegation delegated to that agent are negatively correlated. Since I focus on the choice for legislators to delegate authority after the founding regulation and delegation, I expect the autonomy of an agency to be already mostly shaped at its establishment. The principal (the EU legislators) therefore knows what constraints are already in place when it decides to delegate. Considering the fact that principals are hesitant to lose control (Magnette, 2005, Pollack, 1997), I expect that legislators will sooner delegate more authority to agencies who are under stricter control, hence, to agencies with lower amounts of autonomy.

I therefore hypothesize that the institutional design of an agency (in this case, of the EU agencies) influences its policy. The dependent variable in this study is delegation of tasks from the EU legislators to the EU agencies. I expect that the formal autonomy of an organization to be negatively related to the amount of delegation. If an agency has lower levels of autonomy and is therefore more formally constraint by their principal, the principal will be more likely to delegate authority to that agency. Combining the independent variable of autonomy with the dependent variable of delegation, this leads to the following hypothesis:

Agencies with low degrees of formal autonomy will more often receive delegation of authority than agencies with higher amounts of formal autonomy.

This hypothesis can be translated into a visual conceptual model, with the level of formal autonomy of an agency forming the independent variable of this study and the delegation of authority to an agency by the EU legislators, after the agency has been established as the dependent variable. I expect that there will be a negative relationship between the independent and the dependent variable.

3. Methods

In this chapter I give an overview of the choices I have made to be able to answer the research question. In the first paragraph I discuss the operationalization of the dependent and independent variables. Paragraph two continues with the data collection strategy in which I explain the methods that have been used to collect the necessary data. The last paragraph in this chapter focuses on the data analysis strategy and looks at the methods that have been used to analyse the data. Throughout this chapter, I pay attention to the validity and reliability of the choices that have been made and the measures I took to enhance the validity and reliability.

3.1 Research Question – Operationalization

A first step that is necessary to further perform this study is the operationalization of the main variables. Both the delegation of authority to the agencies (the dependent variable) and the formal autonomy of an agency (the independent variable) are concepts that can be explained and interpreted in multiple ways.

3.1.1 Dependent variable: Delegation of Authority

The way the agencies can have their tasks extended is through delegation from either the Commission or the Council. They need to be granted with new authority/tasks through delegation by the Commission or the Council. The European Commission, the European Parliament and the Council of Ministers together form the legislators that can delegate new jurisdiction to the EU agencies. The table below shows the definition used to measure delegation and the values this definition can take. These values have been formulated for the purpose of this study, in order to create a more detailed overview of the type of delegation to EU agencies. I have divided delegation into two orders. I regard first order delegations as delegations that delegate a whole new field of policy to an agency. This means that the agency gets authority over a whole new domain that was not part of the agency before. Second order delegations are the delegations that expand the tasks of the agency, but within an already delegated field of policy. This can be seen as more of a widening of already existing tasks, or as the creation of new mechanisms to deal with tasks the agencies were already delegated with. This subdivision into different sorts of delegation can help to gain more insights into the types of delegation to EU agencies.

Table 3.1 Operationalization of delegation

Variable Operationalization Values Descriptive statistics

Delegation “An authoritative decision, formalised as a matter of public law, that (a) transfers policy making authority away from the established, representative organs (those that are directly or indirectly elected, or are appointed by elected politicians) to (b) a public non-majoritarian institution” (Curtin, 2005: 92). The tasks or instruments need to be related to an expansion of the jurisdiction/authority of the agency. Changes in regulation that are purely administrative or related to the status of employees are not seen as delegation of new tasks.

First-order delegation: delegation related to new policy fields/tasks for the agency. Second-order

delegation: delegation of new instruments to an agency to perform already existing tasks

First-order delegation = 5 (spread over 4 agencies) Second-order delegation = 10 (spread over 9 agencies)

3.1.2 Independent variable: Autonomy

As has become clear from the literature review, the amount of organizational autonomy is an important part of institutional design. The concept of autonomy can be broken down into different parts and definitions, but as a starting point it relates to the ability of organizations (the agents) to act without being bound to the preferences of their principals (Busuioc, Curtin and Groenleer, 2011). In order to make the concept more suitable and measurable for this study, it is important to break it further down. When studying organizations and agencies, oftentimes the concept of autonomy is divided into two parts. First of all autonomy can point to the formal autonomy of an agent. This relates to the autonomy on paper, the autonomy that is granted to the agent through formal rules and decisions. In the case of the EU agencies, this formal autonomy has been granted to the agency through the founding regulation (Busuioc, Curtin and Groenleer, 2011). Next to that, there is the de facto autonomy, which relates to the level of autonomy the agent has in practice (Verhoest et al. 2004; Yesilkagit 2004; Busuioc, Curtin and Groenleer, 2011). This de facto autonomy can differ in relation to the formal level of autonomy. In this study, I focus on the formal level of autonomy of the EU agencies. This formal autonomy looks at the way the hierarchical lines of the organization are structured. The use of formal autonomy as independent variable knows some limitations and can influence the validity of this study. The amount of autonomy on paper might not be the same as the de facto autonomy of the organization (Yesilkagit and Van Thiel, 2008). Yesilkagit and Van Thiel (2008)

(Ibid.). In some instances, depending on the level of politicization of the topic/of the agency, the agency might have more or less de facto autonomy than formal autonomy (Yesilkagit, 2004). Since I will only measure formal autonomy aspects, this might lead to a decrease of the validity in this study. Conclusions drawn from solely formal autonomy aspects might differ than when I would incorporate other types of autonomy as well. However, due to availability constraints on data on the de facto autonomy, the formal autonomy will be taken as the main variable used to measure autonomy. I believe testing the formal autonomy of EU agencies and their relation to delegation of authority can form a first starting point in this study field.

In table two I operationalize the independent variable of formal autonomy. The descriptive statistics of the eight indicators that make up the formal autonomy are derived from the data by Christensen and Nielsen (2010). It is important to keep in mind that the level of autonomy of agencies can change over time and that the data only portrays the level of autonomy at a certain point in time. Some agencies might develop a more autonomous operating climate over time, while others might become more restricted in their functioning. This part is not included in this study, since the data used was limited to a specific point in time (2009). The eight indicators/values that are part of formal autonomy are discussed more deliberately on the next page.

Table 3.2 Operationalization of formal autonomy

Variable Definition Values/Survey Question Descriptive statistics

Formal autonomy*

“The capacity to manage your own affairs, acting and deciding unbound by the preferences and interests of their principals.” (Busuioc, Curtin and Groenleer, 2011: 850),

1. Advisory board added to agency (0 = no, 1 = yes)

2. Appeals board added to agency (0 = no, 1 = yes)

3. Regulated by comitology procedure (0 = no, 1 = yes)

4. Overall degree of embeddedness (mean of first three, scale of 0-1)

5. Procedure of decision-making at management board (0 = unanimity, 1 = majority)

6. Obligation to secure information and transparency of activities (Mean of ‘has to make an annual work programmes’, ‘has to give annual account’, ‘has to make an

No = 12 Yes = 13 No = 21 Yes = 4 No = 20 Yes = 5 Min = 0 Max = 1 Mean = 0,29 Std. Dev. = 0,28 Unanimity = 3 Majority = 22 Min = 0.33 Max = 1 Mean = 0.93 Std. Dev. = 0.17

annual report of last years activities’ (for all individual variables: 0 = no, 1 = yes). Scale going from 0–1

7. Executive appointment procedure (0 = internal, management board, 1 = external, commission and/or council)

8. Number of external stakeholders represented on governing board (scale from 1-9)

9. Overall amount of autonomy

Internal = 17 External = 8 Min = 1 Max = 5 Mean = 2.72 Std. Dev. = 1.31 Low = 11 High = 14

* Indicators of formal autonomy based on the indicators established by Christensen & Nielsen (2010). These indicators of autonomy are specifically established for the collection of EU agencies, making them relevant to use for this study as well. The data in this table is based on Christensen & Nielsen, since I use their dataset for the analysis as well (2010: 186).

Looking at the eight indicators of formal autonomy, it is important to describe the direction of the expected relationship between the individual indicators and the level of autonomy. The first three variables, ‘appeals board, ‘advisory board’ and ‘regulated by comitology procedure’ are either present (1) or not present (0) within an agency. If an agency scores positive on one of these variables, this can be seen as a decrease in autonomy, since this means the agencies are under supervision of the EU legislators through either an appeals or advisory board. Following Christensen (2010) I have combined these three variables into a new variable that measures the overall degree of embeddedness. The lower the overall degree of embeddedness, the more the agency can be seen as formally autonomous, after all, they are then less ‘embedded’ by EU institutions and regulations. The fifth variable, the procedure of decision-making at the management board, is divided into two values: majority voting or unanimity. Majority voting forms a sign of high autonomy, since it is easier for the agency to make decisions under majority voting than under unanimity voting. The sixth variable, the

obligation to secure information, is measured through three other variables. The questions whether the

agencies have an annual work program, an annual account and an annual report that they have to complete every year are combined in this variable. The higher the agency scores on the obligation to secure information, the lower the amount of autonomy for this agency, since a high obligation to provide information to the EU legislators means that they are more strongly connected to the EU and therefore less autonomous. The seventh variable, the executive appointment procedure, can either be

autonomy since this means that the agency itself can decide over its executive. Lastly, variable eight is the number of external stakeholders that are represented in the governing board. When this number is high, this is a sign of high autonomy. The more external stakeholders, the lower the influence of official EU agencies on the agencies, therefore increasing the level of autonomy of the agency.

3.1.3 Control variables

The expected effect of autonomy on the amount of delegation to the agencies is probably influenced by different factors. It is therefore of importance to include certain control variables, to make sure that the effect that is found is not in fact caused or influenced by other variables. The first variable that is expected to have an effect on the relationship between autonomy and delegation is the age of an agency, since the time an agency exists might have an effect on both the autonomy and especially on the amount of delegation to these agencies. One can expect that the longer an agency has existed, the higher the chances that more authorities/regulation has been delegated to that agency, hence, the more chances the agency has to have faced some sort of delegation. Therefore, it is important to measure the effect of the age of an agency on the amount of delegation and to include this variable in the analysis and in the regression analyses. The age of an agency is measured through the date when the agency was officially established. Although it sometimes took quite some time for an agency to become operable after the official establishment, this date is chosen since it indicates the starting point for the existence of the agency.

A second variable that might have an influence on the relationship between autonomy and delegation to EU agencies is the type of authority of the agency granted at the agencies when they were founded. Authority can be defined as the tasks the agency is granted with and it ranges from informational to decision-making (Christensen and Nielsen, 2010). It might be of influence on the dependent variable, delegation, what kind of authority the agencies were granted with in the beginning. With authority I refer to the sorts of tasks that are delegated to the agency. Christensen and Nielsen (2010) differentiate between four types of authority an agency can be granted with. These four agency tasks are displayed in table two in such a way that they portray an increasing level of agency authority. First of all, and of most frequent occurrence, agencies are granted with information-gathering tasks. This can mean that the agency is granted with tasks that can vary from collecting information, collecting surveys, evaluations, etc. The next type of authority is the coordination of networks, which means that their tasks include keeping in touch with national authorities and coordinating relations and networks between these national authorities (Ibid.). Third, there is the task of policy advice to the Commission. Lastly, and the highest level of authority, is the task of making “authoritative decisions on specific rule applications” (Ibid.: 185). When granted this authority, agencies have ‘higher’ authority, meaning that they for example can issue certificates to companies or that they may hold inspections (Ibid.). It is possible to possess multiple types of authority; in fact most agencies in the EU have two or more types

of authority. The type of authority might influence the delegation of new authority to EU agencies. It might be that agencies with some forms of authority receive more new delegation than agencies with other types of authority. Therefore, it is important to take these four types of authority into account when performing an analysis on the relation between autonomy and the delegation of new authority to agencies.

Table 3.3 Control variables*

Variable Operationalization Survey Question/Values Descriptive statistics 1. Year of

establishment

The date of the regulation that officially established the agency

The year the agency was officially established 1975 = 2 1990 = 2 1993 = 2 1994 = 1 1995 = 2 1996 = 1 1998 = 1 2000 = 2 2001 = 1 2002 = 5 2004 = 5 2006 = 1 2. Agency task (authority of the agency)

The tasks an agency is granted with through its establishment

1. Information gathering (No = 0, Yes = 1)

2. Coordination of network (No = 0, Yes = 1)

3. Policy advice to the Commission (No = 0, Yes = 1) 4. Authoritative decision-making (No = 0, Yes = 1) No = 5 Yes = 20 No = 4 Yes = 21 No = 14 Yes = 11 No = 18 Yes = 7

* This table is based on data gathered by Christensen and Nielsen (2010).

3.2 Data collection

After providing an operationalization of the dependent and independent variables, it is important to discuss the method of measuring these variables. As the basis of my data collection, I used the dataset by Christensen et al (2010) on EU agencies. However, this data is not completely up to date, since they only incorporated agencies up until 2009. Therefore, this study only includes data up until 2009. Furthermore, data about delegation is not included in this dataset. Therefore, data regarding delegation have been manually collected and included to this dataset.

3.2.1 Measurement of Delegation

There are 25 agencies in the dataset; these are all agencies that were included in the dataset that I use in this study and that were operating in 2009 (Christensen and Nielsen, 2010). For all these agencies I look at the laws that were delegated to them at their initiation and at the laws that were delegated to them in a later stadium of their existence. This makes it possible to compare how much legislation has been added to the individual agencies and to compare this data with the level of autonomy these agencies possess. To collect the information on the legislation, the EUR-Lex database has been used. This is a database that publishes all EU law and EU documents. However, not all the new regulation regarding the EU agencies signed by the legislators can be seen as the creation of new authorities/as new jurisdiction for the EU agencies. Some regulations mainly deal with procedural changes, or are focused on financial regulation. This made it necessary to make a clear distinction between what regulation can be regarded as a delegation of authority and what regulation merely counts as procedural regulation. For example, the largest part of regulation on EUR-Lex regarding the European Medicines Agency is regulation related to the adjustment of the fees of the European Medicines Agency to inflation rates. Although this is official EU regulation, this information does not provide any useful information regarding delegation to this agency. Therefore, while coding the regulation I have focused on the question whether the regulation included a change in the authority/the tasks of the agency. Different types of legislation can be used to delegate authority. Council regulation, commission regulation and commission decisions are the types of legislation most often used to delegate authority to agencies.

Sometimes, the decision whether or not a piece of regulation could qualify as delegation depended on the interpretation of a text. By comparing the founding regulation text with the amendments, I was able to determine whether or not tasks had changed. Carefully reading, breaking down the text in different parts and making tables was necessary in order to make the correct judgment whether or not it counted as delegation. However, it needs to be taken into account that, due to the large level of legislation connected to each of the agencies, that interpretation mistakes might have been made. This can potentially have a negative effect on the reliability of this study. However, all legislation on EUR-Lex has been checked twice, to make sure that the possibility of mistakes was as low as possible. After collecting all relevant legislation through EUR-Lex, I have used the websites of the individual agencies to check whether the information was complete. For example, the closure of the European Agency for Reconstruction was not mentioned in one of the EUR-Lex documents and added to the table after consulting the website of the agency. The fact that not all the relevant legislation can be found on EUR-Lex makes it important to triangulate data sources and check whether all legislation and regulations are present in the table, in order to enhance the reliability of this study.

Lastly, I have made a ranking in the types of delegation, to differentiate between first and second order delegation. This again came down to interpreting legislation texts and judging whether a delegation fell within a first order delegation or within second order delegation. An example of this ranking process can be found in appendix A. In this appendix I show an example of the process of determining whether or not regulation could be judged as delegation of authority. This helps to enhance the reliability of this study, due to the fact that I create insights into the way certain choices were made.

3.2.2 Measurement of autonomy

To measure the formal autonomy of the agencies I have used the dataset by Christensen and Nielsen (2010). As has been displayed in table 3.2 (page 24) there are eight variables that together measure the level of formal autonomy of each agency. The measurement of eight indicators of autonomy was therefore not necessary, since data was available through this existing dataset. However, it was also useful to combine the eight individual independent variables into a new variable that measures the overall degree of formal autonomy per agency. This total level of formal autonomy is based on the sum of the eight indicators, with each indicator having an equal weight. Appendix B shows the process of recoding and restructuring the variables in order to create this new variable.

3.3 Data analysis strategy

There are multiple methods that I have used to analyse the data. First of all, in chapter four, I will provide a background on the EU agencies. Part of the data from this background has been collected through primary and secondary sources and consists of qualitative data regarding the EU agencies. Since this background is only meant to be a way to gain some extra insights in the dataset, there is no real judgment/interpretation attached to the content of this part of the analysis.

The second part of the analysis, the quantitative analysis, will be analysed through the means of a statistical analysis. Both the descriptive and the inferential statistics have been analysed in SPSS. Following Christensen and Nielsen (2010) I use the Gamma-correlation instead of Pearson-correlations, since not all variables are interval variables and Gamma-correlations are therefore more appropriate. I will perform the inferential statistics analysis by a linear regression analysis, with the independent variable being the overall degree of autonomy (and also measured with the variables of autonomy one by one) and the dependent variable being the delegation of tasks to the agency. In order to use a linear regression analysis, certain assumptions need to be fulfilled (De Vocht, 2010). Although the N of this study is rather small, most of the assumptions are fulfilled. However, regarding the assumption of multicollinearity, certain independent variables are quite strongly correlated. Therefore, some measures have been taken, which are presented in appendix C, together with the other assumptions for regression. Both simple and stepwise regression analyses were used to analyse the

4. The EU agencies – Background

This chapter focuses on providing a short background on the EU agencies. I concentrate on the types of agencies currently present in the European Union, the specific type of agency subject to this study, the reasons for their establishment and on the different types of fields they operate in. Furthermore, I pay attention to a description of the two control variables in this study, the age of the agencies and the authority of the agencies.

There are multiple sorts of agencies operating in the European Union. The general definition formulated by the European Union of EU agencies is: “EU agencies are distinct bodies from the EU institutions – separate legal entities set up to perform specific tasks under EU law.” (European Union, 2015b). Kelemen (2005: 175) further describes EU agencies as “public authorities with a legal personality and a certain degree of organizational and financial autonomy that are created by acts of secondary legislation in order to perform clearly specified tasks.” Both these definitions focus on the facts that these agencies perform specific tasks, separately/independently from the other EU institutions. The biggest group of agencies are the decentralised agencies. Next to the decentralised agencies the EU currently has multiple agencies under common security and defence policy, executive agencies, EURATOM agencies and lastly the European Institute for Innovation and Technology, which forms a distinct category (Ibid.). These different types of agencies all perform different tasks for the EU, but are all set up differently. The decentralised agencies together with the ‘Agencies under Common Security and Defence Policy’ form the focus of this study. The Agencies under Common Security and Defence Policy fall within the same category as the decentralised agencies, but are bundled together in a group due to the fact that they all perform tasks within the framework of the EU’s Common Security and Defence Policy (Ibid.). Although the policy fields in which the agencies operate vary greatly, the agencies were all established with the same goal in mind. The administrative capacity of the Commission was perceived as relatively low for the demands the EU faced (Christensen and Nielsen, 2010). Since the pressure on the EU regulatory capacity was increasing, the EU needed a way to solve this challenge. In order to expand specific and expert knowledge of the Commission and to bring together expertise from both EU institutions and national authorities, the idea of creating independent agencies was invented (Christensen and Nielsen, 2010; European Union, 2015b). These decentralised agencies are set up to perform different types of tasks, such as technical, scientific or managerial. These tasks are aimed at helping the EU institutions to make and to implement policies. The locations of the agencies are spread throughout the EU member countries. The agencies have their own staff and for each agency there are different and individual provisions for financial, budgetary and staff regulations.