The Rules of the Game: A comparative case study on the conditions for the socialization of permanent representatives in the EU and NATO

The Rules of the Game

A comparative case study on the conditions for the socialization of permanent representatives in the EU and NATO

August Danielson

Master thesis

Fall 2016

Uppsala University

Supervisor: Anna Michalski

Abstract

Cooperation in international organizations is to a large degree driven and sustained by socialization – the process of inducting actors into the norms and rules of a given community.

In the context of international organizations, the most influential state agents are the permanent representatives, the member states’ ambassadors to an international organization. However, systematic studies on the conditions for the socialization of permanent representatives are conspicuously absent in the literature. Instead, most previous research has focused on the conditions for socialization of “high-level officials” in the European Commission. This quite narrow perspective has led to a “N=1” problem and the generalizability of these studies have suffered as a result. In this thesis, I have aimed to broaden this perspective by testing six hypotheses on the conditions for socialization within two committees of permanent representatives in two different international organizations, the PSC (EU) and the NAC (NATO). This has been done by conducting elite interviews with 21 permanent representatives and deputies in Brussels. In contrast to previous research, the results of this thesis suggest that four of the six tested hypotheses should be disregarded, while two hypotheses – the representative’s relation to its MFA and the ambiguity of the international organization’s norms – should be given more theoretical consideration in future research. In addition, the interviews have shown that the degree of “internalization”, which can be understood as the goal and outcome of socialization, is stronger in the NAC than in the PSC. This outcome also contests the conventional wisdom that the EU is a sui generis case of socialization. On the basis of these results, I argue that the socialization of permanent representatives in international organizations is likely to occur if the organization’s norms have unambiguous, existential and materialistic consequences.

Keywords: socialization theory, North Atlantic Treaty Organization, European Union, internalization, international organizations, permanent representatives

Word count: 19,852

Table of Contents

1. Introduction 5

2. Purpose – to fill a theoretical gap in socialization theory 6

2.1 Research Question 8

3. Previous research – the evolution of socialization theory 9

3.1 Constructivism in International Relations 10

3.2 Sociological institutionalism 11

4. Conceptual clarifications – socialization theory 12

4.1 Socialization as an outcome 14

5. Hypotheses on the conditions for socialization 15

5.1 Limitations – what this thesis is not testing 15

5.2 Choice of hypotheses – what this thesis is testing 17

5.2.1 The properties of the international institutions that trigger socialization 17

5.2.2 The properties of the agents who become socialized 18

5.2.3 The properties of the issues or norms regarding which socialization takes place 19 5.2.4 The properties of the interaction between the socializing and the socialized 19

6. Background to the NAC and the PSC 20

6.1 The organizational level 21

6.2 The political level 22

7. Method 23

7.1 Research design 23

7.1.1 Mill’s methods 23

7.2 Measuring the conditions – elite interviews 25

7.3 Generalizability 28

7.4 Operationalization of variables and hypotheses 30

8. Analysis 31

8.1 The dependent variable – the degree of internalization 31

8.2 H1: Organizational boundedness 37

8.3 H2: Relation to MFA 39

8.4 H3: Prior work experiences 41

8.5 H4: Ambiguity of norms 43

8.6 H5: Issue intensity 45

8.7 H6: Insulation/agent autonomy 47

9. Discussion 49

10. Conclusion 54

List of Abbreviations

ASEAN Association of Southeast Asian Nations

CFSP Common Foreign and Security Policy

COREPER Committee of Permanent Representatives

CSDP Common Security and Defense Policy

DG Directorate-General

EEAS European External Action Service

EFTA European Free Trade Agreement

EU European Union

FAC Foreign Affairs Council

IO International organization

IR International relations

IS NATO International Staff

MFA Ministry of Foreign Affairs

NAC North Atlantic Council

NATO North Atlantic Treaty Organization

OSCE Organization for Security and Cooperation in Europe

PSC Political and Security Committee

UN United Nations

WTO World Trade Organization

1. Introduction

The role of international organizations (IOs) is becoming increasingly significant for international policy-making.

Whether it concerns terrorism, humanitarian crises or climate change, IOs are currently involved in every imaginable global issue. IOs have also become distinctly more organized in the post-World War II era, reflected in the upsurge of largely independent international bureaucracies.

These “unelected bodies”, and the civil servants that work for them, are frequently put in positions of considerable political influence as a result of the often vague legal frameworks of IOs and insufficient oversight mechanisms to their member states’ disposal. Therefore, a growing number of IOs have become active supranational policy- making institutions, rather than simply acting as forums for intergovernmental policy coordination.

In the past two decades, the scholarly interest in IOs has also shifted. Compared to the early works on European integration that mainly tried to show that institutions matter, recent research on IOs has rather focused on the mechanisms and conditions under which they matter. Within this trend, constructivist research on IOs’ ability to “socialize” state actors has grown exponentially.

In the context of IOs, socialization can be understood as “a process of inducting actors into the norms and rules of a given community.”

In international socialization research, such actors can include individual policymakers, bureaucrats and even states.

By adopting the norms and rules of a community, socialization implies that an actor switches their “logic of action” from a logic of consequences to a logic of appropriateness – which can be understood as acting on the basis of the type of behavior that is most appropriate in a certain context rather than on the basis of instrumental calculation.

However, even though socialization in some rare cases can lead to state representatives completely adopting their group’s idea of the normatively

“correct” version of a certain policy – instead of working to achieve the instructions given by their respective member states – socialization should not necessarily be understood as

1

I use the term “international organization” to describe intergovernmental organizations. In line with Barnett & Finnemore (2004), an IO should be understood as an organization that has representatives from three or more states supporting a permanent secretariat to perform ongoing tasks related to a common purpose.

2

Barnett & Finnemore 2004.

3

Trondal 2010, p. 6.

4

See Egeberg 1999, 2004; Hooghe 1999, 2005; Kelley 2004; Beyers 2005; Trondal 2010.

5

Alderson 2001; Checkel 2005.

6

However, this thesis will only study the conditions that affect the actions of individual state agents.

7

March & Olsen 1998.

something wrong or harmful. Rather, the socialization of state actors can and should be understood as a fundamental cornerstone of state cooperation in IOs. It can also be argued that it is specifically the pursuit of consensus building within IOs that make states promote norm- driven, global interests rather than their “materialistic” state interests.

Since cooperation within IOs is of fundamental importance to solve global issues such as the ones mentioned above, socialization should be understood as being equally important in combating these issues.

The theoretical argument behind socialization is thus quite sound. For example, it is reasonable to believe that a state actor could “go native” and completely adopt their organization’s norms after living and working in Brussels for a decade or more. However, although socialization processes may favor cooperation and consensus building, socialization does not necessarily entail a complete absence of conflicts.

We cannot simply presume that every bureaucrat or state representative who works within an IO naturally starts sharing their organization’s norms after working in an international setting for a certain number of years. This begs the question:

what actually drives socialization processes? If we are to understand how cooperation in IOs is sustained and, by extension, how global issues can better be dealt with, we must understand the conditions that enable socialization both as both a process and an outcome – the internalization of group norms and role conceptions.

However, even though the literature on socialization in IOs has grown vastly over the last two decades, the specific conditions that can be said to produce socialization have to a large degree remained untouched by scholars within IR theory as well as EU studies.

2. Purpose – to fill a theoretical gap in socialization theory

Some significant exceptions can be made to this statement. Among those who have looked specifically at the personal and organizational conditions that can be said to “produce” or

“enable” socialization are Beyers (2005, 2010), Hooghe (1999, 2005), Lewis (2005) and Egeberg (2004). However, the conditions or factors (these terms will be used interchangeably

8

It should be noted that the pursuit of norm-driven interests does not necessarily entail a switch in a state’s basis of action from instrumental calculation to appropriateness. In other words, to pursue norm-driven interests can also be a “rational” state interest.

9

Beyers 2010, p. 912.

10

I will use both “produce” and “enable” in this context to refer to how socialization is both “created” and

“made possible” by the independent variables, for lack of a better word. This is in line with Checkel (2005).

11

Checkel 2005, p. 7.

in this thesis) that these authors choose to study vary greatly. Some studies include direct controls for pre-selection, domestic recruitment and pre-socialization (for example Beyers 2005 and Hooghe 2005), while other studies do not control for any variables, but instead aim to understand the institutional or organizational context that is needed for socialization to occur (for example Lewis 2005). This methodological diversity is also reflected in the studies’

varying operationalization of socialization as a dependent variable. Some studies have operationalized socialization as the degree of internalization of group norms where a shift in

“role conceptions” or “support for supranational norms” is used to measure the degree of internalization.

Meanwhile, others have conceptualized the outcome of socialization as “pro- social” or “pro-norm” behavior.

The variation in the choice of dependent variable can in part be attributed to the inconsistency of the literature in distinguishing socialization as a process or as an outcome.

For example, socialization can be understood as both the process of moving from one logic to another (acting on the basis of appropriateness instead of consequences) and the end point of working within a socializing group (internalization). These two perspectives result in very different conceptions of what socialization actually is, how we can measure it and what it is caused by.

In addition to this conceptual diversity, most studies on the conditions that enable socialization within IOs have also lacked comparative designs.

For example, studies on socialization in the European Commission have been criticized for the “N=1” problem.

One explanation as to why most studies on socialization in IOs have focused on the European Commission is an unstated premise in the literature that socialization within the EU is something unique and that the process of “European socialization” is, for some implicit reason, not generalizable to other IOs.

For example, using “supranational norms” as a dependent variable for the outcome of socialization is far from being a generalizable operationalization of socialization for an IO that has security in the North Atlantic area as its main goal or norm (e.g. NATO). This premise is also demonstrated in how most previous research has disregarded other international institutions than the EU. One explanation to this could simply be that EU institutions are more accessible and easier to study than other IOs. As stated by Beyers, “it resembles the drunk who

12

Beyers 2005; Hooghe 2005.

13

Lewis 2005; Johnston 2005; Zürn & Checkel 2005.

14

Beyers 2010, p. 910.

15

See Hooghe 2005 and Beyers 2005.

16

Warleigh-Lack & Phinnemore 2009, p. 216; Trondal 2010, p. 23.

17

One rare exception is Johnston (2005) who has called for more comparative research on socialization

processes in IOs.

searches for his keys near the streetlights because this is where he can see”.

However, simply because it is harder to conduct comparative studies on the conditions for socialization in other IOs than the EU is not a sufficient argument for not doing it.

Since I believe that previous research has been built upon an erroneous assumption that the conditions for socialization within the EU are somehow non-generalizable to other IOs, I aim to test this claim. By comparing the existence of the different conditions for socialization within two similar committees of permanent representatives, the Political and Security Committee and the North Atlantic Council, within two different IOs, the EU and NATO, this thesis will show which of the tested conditions that are more relevant and which we should rule out. By comparing these two committees, the aim of the thesis is thus to increase our theoretical understanding of how and why socialization processes develop by “narrowing down” the possible conditions that lie behind these processes. Below, the research question of the thesis will be described and motivated.

2.1 Research Question

As previously mentioned, the conditions that drive socialization constitute an under-researched topic and the generalizability of previous studies have suffered as a result of non-comparative approaches. The purpose of this thesis is to fill the aforementioned theoretical gap in socialization theory by comparing the existence and level of the potential conditions that produce socialization in two comparable cases. Hypotheses that have been proposed in previous studies on socialization in the European Commission will thus be tested on two committees, the NAC and the PSC, with the goal to “rule out” hypotheses that are not applicable to both committees. Comparing socialization processes within two IOs will also allow us to highlight the key mechanisms that lie behind each condition. For example, even if the level of some variables is more or less equal in the two committees, the path to these outcomes may differ between them.

The guiding research question of this thesis is thus the following: under what conditions does the socialization of permanent representatives in international organizations occur?

18

Beyers 2010, p. 910

To be able to draw conclusions as to which of the scope conditions on socialization that have been proposed in previous research should be seen as less significant, we must first establish whether or not the outcome of socialization (the degree of internalization) is equal or different between the committees. Once this is done, we can compare the existence of the possible conditions to explain why the degree of internalization is either similar or different between two comparable cases. However, to be able to accurately measure the degree of internalization in both committees we must also operationalize the concept of socialization so that it is generalizable for the socialization processes within all committees of permanent representatives in IOs. This is of course no easy task, but by conducting a short review of the literature we should be able to get closer to a more generalizable definition.

3. Previous research – the evolution of socialization theory

Socialization theory can be understood as the result of two main social constructivist approaches within the disciplines of international relations, sociology and organizational theory: “IR constructivism” and sociological institutionalism.

Both of these approaches are based on an ideational ontology in the sense that they see ideas as being the main driver of international politics and institutions.

This ontology can be contrasted with rationalist conceptions of human action that understand the world as a result of “individual goal- seeking”.

In comparison, constructivism sees human action as the result of “collectively meaningful structures and processes”.

Constructivism can thus be said to put larger focus on the power of concepts and ideas rather than the power of material self-interest. From a constructivist perspective, material resources “only acquire meaning for human action through the structure of shared knowledge in which they are embedded”.

Since constructivists argue that action has no meaning in itself, the implication of this perspective is that every human action is only the result of a collective understanding of the “appropriate” behavior in a specific context.

However, even though IR constructivism and sociological institutionalism share a common ontology, they differ somewhat in their theoretical focus. Whereas both approaches

19

I use the term “IR constructivism” to distinguish the development of constructivism within two disciplines, IR and organization theory. This is consistent with previous scholars such as Adler (2002).

20

Schimmelfennig 2003, p. 68.

21

Snidal 2002, p. 74.

22

Adler 2002, p. 100.

23

Wendt 1995, p. 73.

24

Adler 2002, p. 100.

assume that the actions of states or state agents is the result of a collective understanding of appropriate behavior, IR constructivists argue that this appropriateness is a result of an international society that is driven by norms and identity. Meanwhile, sociological institutionalists see appropriateness as the result of political institutions, which March and Olsen argue should be understood as “collections of interrelated rules and routines”.

The difference between the two approaches can thus be understood as whether it is the power of the ideas and norms themselves or the power of the institutions in which they are allowed to grow that ultimately creates a collective understanding of appropriate behavior.

3.1 Constructivism in International Relations

Both IR constructivism and sociological institutionalism have old roots in the literature. IR constructivism is essentially the result of the seminal works of Karl Deutsch (1957) and Ernst B. Haas (1958) who suggested that international institutions can create a sense of community and belonging that is distinct from the nation state. While Deutsch emphasized transnational collective identities and social communication as key factors in creating “security communities”, Haas instead focused on the spill-over effects of political integration. Towards the end of the Cold War, some of the key realist assumptions of interstate politics were criticized by neoliberal scholars such as Keohane and Nye (1977) who argued that states were becoming increasingly interdependent of each other as a result of complex transnational connections.

This criticism led to the so-called “third debate” within IR, also known as the “inter-paradigm debate”. From this debate, the works of Deutsch and Haas were once again put in the spotlight of IR scholars. Among those who were inspired by the works of Deutsch and Haas around this period were the “modernist constructivists” Onuf (1989) and Wendt (1992). While Onuf coined the concept of IR constructivism through his ideas on the power of performative language (speech acts), Wendt instead focused on how and why anarchy in the international system is upheld.

Through this perspective, Wendt’s inspiration of Deutsch’s theory of security communities is very clear – they both view the collective identity formation as an essential aspect of the international system. Meanwhile, Ernst B. Haas’s son, Peter M. Haas, developed the neofunctionalist theory of regional integration through his idea of “epistemic communities”, also echoing the idea of collective social learning within Deutsch’s theory. However, while

25

March & Olsen 1989, p. 160.

26

Schmidt 2002, p. 11.

27

Adler 2002, p. 99.

Deutsch’s and Wendt’s theories were more state-centric, Peter M. Haas focused on the

“transnational networks of knowledge-based experts” that facilitate international cooperation by institutionalizing a “shared set of normative and principled beliefs”.

Other notable contributors to IR constructivism have been Kratchowil and Ruggie (1986) who argued that international regimes can create “shared understandings of desirable and acceptable forms of social behavior” as well as the many scholars of the English and Copenhagen schools.

3.2 Sociological institutionalism

The second main theoretical approach that has had a large influence on the development of socialization theory is sociological institutionalism. Sociological institutionalism is a “neo- institutional” theory that has essentially grown out of two different disciplines, organizational theory and sociology.

During the post-World War II era, organizational scholars such as Robert K. Merton (1936), Herbert Simon (1947) and Philip Selznick (1948) argued that the

“Weberian” understanding of organizations as solely rational and efficient had become far too unrealistic. Instead, they argued that institutions can become infused with a distinct set of values, “an identity”, and that they should be understood as more than simply “expandable tools”.

Selznick’s classic study of the Tennessee Valley Authority focused on the transformation process of an organization becoming an institution and, as a result, the process of vested interests (political tradeoffs and alliances) inhibiting the formal, rational goals of an organization.

This “old institutionalist” explanation of how and why organizations are constrained by their social environment was deemed insufficient by neo-institutionalist scholars such as Meyer, Powell & DiMaggio and March & Olsen towards the beginning of the 1980s.

They argued that many of the institutional forms and procedures used by modern organizations were not adopted because they were the most efficient and rational, but rather because they were the most “legitimate” in an organizational culture. DiMaggio and Powell (1991) coined the concept of institutional isomorphism to explain institutional change in cultural terms.

DiMaggio and Powell argued that institutional homogenization is the result of the

“persuasiveness of cultural accounts” from other organizations rather than a common effort to

28

Haas 1992, p. 2.

29

Kratchowil & Ruggie 1986, p. 764

30

DiMaggo & Powell 1991, p. 8. However, organizational theory should not be understood as solely

“sociological” since it is open to both rationalist and sociological considerations (Scott 1995).

31

Scott 1995, p. 18-19.

32

DiMaggo & Powell 1991, p. 12.

increase efficiency.

Isomorphic processes, they argued, can thus make “organizations more similar without necessarily making them more efficient”.

An institutional/organizational culture can also have an affect on the actors within organizations. DiMaggio and Powell argued that institutions “do not simply limit options: they establish the criteria by which people discover their preferences”. In a similar vein, March and Olsen developed neo-institutionalism by viewing institutions as either the “enabler” of collective action bargaining or a set of formal and informal rules that dictate what type of action that is appropriate in a given context.

Our choice of these two perspectives is to a large degree dependent on our view of human nature — whether humans choose to act rationally or appropriately when given the choice to act. March and Olsen described two types of such “bases of action”: the logic of expected consequences (LoC) and the logic of appropriateness (LoA). The logic of appropriateness assumes that humans are “rule followers” that assume identities and rules when put into an institutional context and can essentially be understood as the basis of sociological institutionalism.

As will be shown in the next section, these two bases of action are fundamental in understanding how and why state actors “socialize” in IOs.

As previously mentioned, the main difference between IR constructivism and sociological institutionalism is the theoretical focus on what actually creates the “rules of the game” – the set of informal rules that guide behavior in, for example, an IO. Institutions are thus seen in two distinct ways, as either the promoters or sites of socialization.

By focusing on how and why the social context (i.e. both the actors within the institution and the effect of an institutional framework) can create senses of community and belonging, facilitate attitude change and shift role conceptions, socialization theory thus combines both of these analytical “lenses”. Below, socialization theory is reviewed in greater detail to enable a better understanding of the different hypotheses that will be tested later in the thesis.

4. Conceptual clarifications – socialization theory

In a special issue of International Organization from 2005, socialization theory in the context of international institutions, and the microprocesses behind socialization, was for the first time

33

DiMaggio & Powell 1991, p. 13.

34

DiMaggio & Powell 1983, p. 147.

35

March & Olsen 1998, p. 948-953.

36

Ibid.

37

Checkel 2005, p. 806.

thoroughly fleshed out and defined.

The authors of the issue argued that previous research on socialization theory, such as the theories and ideas proposed by the many scholars presented above, had focused too little on theorizing the mechanisms behind socialization.

While previous IR constructivists and neo-institutionalist scholars had made a convincing case that socialization actually does happen, they had been less clear on exactly how and why it occurs.

Taking a more positivist approach to socialization, the scholars in the issue aimed to map out how states and state agents might become socialized in international institutions.

Socialization can essentially be understood as the process of individuals developing a sense of belonging with a group and adapting their behavior accordingly.

This type of human rationality can be understood as “communicative rationality”, which Habermasian social theory defines as the outcome of successful communication.

Socialization theory thus views state actors’ preferences as “redefinable” through the act of persuasion, which in turn can be understood as arguing for a norm or informal rule.

Attempts at persuasion of a new group member can lead to the adoption of new norms and eventually to the “internalization” of the agent, or in other words “the sustained compliance of an agent to act in accordance to what is socially accepted in a given setting or community”.

Internalization should thus be understood as both the end result of socialization and the conscious or subconscious goal of the group that sustains the socialization process. By adopting these norms and rules of a community, the process of internalization implies that a state agent no longer acts in conformity with the logic of expected consequences, but rather on the basis of what action that is seen as the most appropriate in a certain context or setting, for example in a working group within an IO.

However, the distinction between a “rational agent” and a “socialized agent” is not completely clear cut. Checkel emphasizes that socialization should be understood on the basis of the mechanisms that lie behind an agent’s gradual shift from a logic of consequences to a logic of appropriateness. These mechanisms are defined as “strategic calculation”, “role playing”, and

“normative suasion” and which, in turn, represent three different “modes” of rationality:

instrumental, bounded and communicative. In essence, the process of socialization can thus be

38

I use the terms “international institution” and “international organization” interchangeably in this section, in line with Checkel (2005).

39

Checkel 2005, p. 806.

40

Alderson 2001, p. 416.

41

Beyers 2010, p. 909.

42

Checkel 2005, p. 812.

43

Ibid.

44

Ibid. p. 804.

understood as the incremental shift from instrumentality to communication as the basis of action.

The way in which an actor gradually adopts these mechanisms is also represented by two different types of internalization: Type I and Type II internalization. Type I internalization can be understood as conscious socialization or conscious adoption of roles while Type II internalization implies that an agent fully adopts the interests of his or her social context or community.

These two types of internalization still follow the logic of appropriateness, but the rationality behind them are vastly different. Type I internalization can basically be understood as “playing along” by the rules of the game (role playing) while still acting on the basis of bounded rationality while Type II internalization implies that the state agent acts entirely on the basis of communicative rationality (normative suasion).

The main difference between these two types can thus be understood as whether the agent still has control over what rationality (bounded or communicative) he or she acts on the basis of.

4.1 Socialization as an outcome

As mentioned in the introduction of this thesis, most authors choose to see the dependent variable (DV) and outcome of socialization as the “degree or level of internalization”.

However, depending on which socialization mechanism they choose to focus on, the authors attempt to measure this operationalization in slightly different ways. Hooghe, for example, argues that the level of internalization should be operationalized as “support for the organization’s norms” while Beyers argues that it should be operationalized as “role conceptions” – i.e. whether a state agent adopts the role of an intergovernmental state representative or that of a supranational representative of the organization. A third alternative is presented by Lewis who operationalizes the DV as “the internalization of new role conceptions and conceptions of the self in line with group-community norms”.

As such, Hooghe, Beyers and Lewis all operationalize the outcome of socialization or the degree of internalization as the shift of a state agent’s identity and preferences after he or she has been exposed to a socializing environment. The difference between their operationalizations is

45

Checkel 2005, p. 805.

46

Ibid. p. 804-805.

47

Ibid.

48

Checkel 2005, p. 817.

49

Lewis 2005, p. 940-941.

basically what type of socialization mechanism they are measuring, either the cognitive role playing mechanism, which indicates a Type I internalization outcome, or the normative suasion mechanism, which indicates a Type II internalization outcome.

In other words, to be able to carry out a comprehensive assessment of the degree of internalization within the study’s two case studies, the NAC and the PSC, we should measure the DV as both of these internalization outcomes. I believe that Lewis’s operationalization is the one that comes closest to this. As such, I have chosen to measure the DV as the internalization of new role conceptions and conceptions of the self in line with group-community norms. This operationalization should let us measure the outcomes of both the role playing mechanism and the normative suasion mechanism and thus the outcome of both Type I and Type II internalization. However, since it would not be very fruitful to simply ask a state agent to what extent they have adopted their group’s norms or how they see their role within the IO, we must in some sense also

“operationalize the operationalization” by formulating suitable interview questions.

5. Hypotheses on the conditions for socialization

Since the purpose of this thesis is to test previous hypotheses on scope conditions for when internalization is more likely to occur within the studied committees, a discussion on the study’s choice of hypotheses is in order. First of all, however, a quick note should be made on which hypotheses this thesis is not testing.

5.1 Limitations – what this thesis is not testing

While the articles in the 2005 special issue of International Organization focus on mostly the same hypotheses as this thesis, the difference in methodology between the articles and this thesis prohibits us from testing the exact same hypotheses. While this thesis employs a comparative method to test the explanatory power of the hypotheses, most of the articles in the special issue use regression analysis to test their hypotheses. For example, Beyers and Hooghe test the conditions on each individual respondent and the significance of each condition through a regression analysis based on interview and survey data. However, since a regression analysis requires an adequate sample size in order to accurately measure the relationship between the

50

Checkel 2005, p. 817.

51

A list of these operationalized interview questions can be found in Table A1 in the appendix.

dependent and independent variable, this method was not a viable option to study the effect of the conditions in the two selected case studies – there are simply not enough member states (and thus Permanent representatives) in the two IOs.

Therefore, this thesis is not able to test the effect of conditions that are unique for each individual and thus non-generalizable for the group(s) as a whole. For example, it would be wrong to use gender as a variable since the presupposed “socialization effect” of a person’s gender is extremely hard to estimate. If I were to test the hypothesis that gender does have an effect on socialization, I would also need to estimate the threshold effect of that condition, in other words “how many males or females in a group does it take to affect the socialization process?” Testing such a hypothesis would instead require a regression analysis on the relationship between socialization and gender for each specific state agent, which, as previously mentioned, lies outside the scope of this thesis. Other hypotheses that this thesis is not able to test are country-specific conditions. For example, Hooghe stresses the effect of how the “pre-socialization” of national norms, such as the national administrative tradition, the centralization of national authority or the ideologies of the national political parties, influence the state agent’s adoption of international norms.

Since these variables, and other variables related to country-specific effects, cannot be measured without in-depth studies on each state agent’s country, they simply fall outside the scope of this study.

Another condition that I have not been able to test is “primacy” – the effect of age and experience on a person’s susceptibility to socialization efforts. Since this effect would most likely only be noticeable in the first few months after joining a new committee, this condition would require a time-series research design. Another condition that would require a time-series research design is the effect of “self-selection”, in other words a state agent’s support for an organization’s norms prior to joining the organization. Since we have no control group that we could use to test this effect, the only way to measure it would have been to ask the state agents how they feel about the organizations’ norms prior to joining and comparing those views to how they feel today. As such, these hypotheses also fall outside the scope of the study.

However, I have included all other hypotheses on the scope conditions for the socialization of state agents that I have been able to measure through in-depth interviews. In the next section I will discuss these hypotheses.

52

A common guideline is n >= 30 before you can expect a statistical analysis based on the normal distribution to be valid.

53

Hooghe 2005.

5.2 Choice of hypotheses – what this thesis is testing

The tested scope conditions are based on the hypotheses presented in three articles that aim to identify the main conditions for state agent socialization in IOs. These articles are Beyers (2005), Hooghe (2005) and Lewis (2005). The scope conditions that the hypotheses in these articles are based on can also be organized into four different categories that are presented in Zürn and Checkel (2005): (1) the properties of the international institutions that trigger socialization, (2) the properties of the agents who become socialized, (3) the properties of the issues or norms regarding which socialization takes place and (4) the properties of the interaction between the socializing and the socialized. I will discuss each hypothesis within these four categories and the reasoning behind them in turn.

5.2.1 The properties of the international institutions that trigger socialization

One factor that focuses on the organizational component of the socialization process is Hooghe’s variable organizational boundedness. Based on Allison’s theory of organizational fragmentation from his seminal work on the Cuban Missile Crisis, Hooghe argues that the more

“bounded”, or rather, the less “fragmented” an organization is, the less inclined state agents are to identify with their group’s interest over and above that of the IO as a whole. State representatives are thus more susceptible to norm internalization in a bounded IO since the socialization process within a working group reinforces the norms of the organization if the two share each other’s norms. In other words, the more contact different groups or committees have with the rest of the organization, the greater the chance is that an agent will internalize the norms of the organization.

The first hypothesis is thus the following:

H1 (Organizational boundedness): The likelihood of a state agent’s internalization of norms is tied to how well the international institution can reduce organizational fragmentation. The more bounded the organization, the more effectively it socializes.

54

Hooghe 2005, p. 865.

5.2.2 The properties of the agents who become socialized

Beyers emphasizes that state agents are often constricted by their “multiple embeddedness”, in the sense that they identify with different rules and expectations depending on what context (either domestic or international) they are embedded in.

As such, Beyers argues that state agents are caught in a “representation dilemma” as they are expected to serve many different preferences and consider the views of other state representatives while at the same time stay loyal to their respective member states. However, he argues that socialization is still possible and that the more ambiguous instructions that a state agent receives from their Ministry of Foreign Affairs (MFA) – and the more leeway the state agent has to those instructions – the easier it is for socialization to take place.

The second hypothesis that this thesis will test is thus the following:

H2 (Relation to MFA): The likelihood of a state agent’s internalization of group norms and new roles is conditioned by how they are tied to their domestic environment. The less tied a state agent is to his or her instructions, the easier it is for them to adopt new role conceptions.

The second condition that relates to the properties of state agents is the effect of an agent’s prior work experiences. Beyers argues that state agents who have extensive work experience from IOs are more prone to adopt a supranational role in relation to their member state since they already know the “rules of the game”, the different types of know-how and codes of conduct that exist within most IOs.

Beyers contrasts this with state agents who have greater domestic work experiences. Since such agents have a better understanding of their national sensitivities, he argues, they thus have a harder time to adapt to the prevailing norms of the IO and are more prone to adopt an intergovernmental role conception. The third hypothesis that will be tested is thus the following:

H3 (Prior work experiences): The likelihood of a state agent’s internalization of group norms and role conceptions depends on their previous work experiences. Agents who have greater work experience in the international community have an easier time to adapt to the norms of the organization than agents who have predominantly domestic work experiences.

55

Hooghe 2005, p. 888; Beyers 2005, p. 910.

56

Beyers 2005, p. 933-934.

57

Beyers 2005, p. 912.

5.2.3 The properties of the issues or norms regarding which socialization takes place

A condition that only a handful of scholars have placed any emphasis on is the role of issue or norm properties. One of these scholars, Hooghe, argues that large, abstract values facilitate socialization since they reduce a state agent’s cognitive dissonance and thus preempt rational calculation.

The more diffuse or uncertain a specific issue is, the lower the risk is of inter- group conflict based on the rational calculation over material stakes. In other words, the more ambiguous a concept is, the less you know about its consequences and the easier it is for socialization to occur. In the context of socialization in IOs, one could argue that it is easier to adopt the norms of a certain IO if they are large and diffuse (ambiguous). The fourth hypothesis is thus the following:

H4 (Ambiguity of norms): The likelihood of a state agent’s internalization of group norms and role conceptions depends on the ambiguity of the organization’s norms. The larger and more diffuse that a state agent perceives the norms of the organization, the more likely it is for socialization to take place.

5.2.4 The properties of the interaction between the socializing and the socialized

In his article, Lewis argues that Type II internalization is mainly the result of two scope conditions: issue density and insulation. In other words, he argues that the more intense and secluded the interaction between state agents, the more likely it is that they adopt new roles and support the norms of the organization. Beyers also focuses on the effect of the interaction between state agents in his three first hypotheses. He argues that the more frequent, intense and longer lasting the interaction between state agents, the more likely it is that they internalize new roles and norms. These conditions are summarized in the following two hypotheses:

H5 (Issue intensity): The likelihood of a state agent’s internalization of group norms and role conceptions depends on the intensity of the interaction with other state agents. The greater the intensity, the more likely it is that a state agent’s role conception will shift.

58

Hooghe 2005, p. 870-871.

H6 (Insulation/agent autonomy): The likelihood of a state agent’s internalization of group norms and role conceptions depends on how private and secluded their interaction is with other state agents. The more insulated their meetings are to outside actors, such as their capitals or the media, the more likely it is that socialization will take place.

To be able to test these six hypotheses, we must first understand the organizational and political context that the permanent representatives operate in. The following section thus provides a more “non-theoretical” background to this thesis’s two case studies, the NAC and the PSC.

6. Background to the NAC and the PSC

The North Atlantic Council and the Political and Security Committee are two of the most influential political committees in the world. They are the de facto decision-making body in the area of foreign and security policy in their respective organizations and exist to coordinate the positions of their member states and create common solutions. To gain a better understanding of how and why these committees act and look as they do, this section will describe them on the basis of two analytical levels, the organizational level and the political level. However, I will first provide a short introduction to the more formal and judicial aspects of the two committees.

The NAC was first enshrined in Article 9 of the North Atlantic Treaty and is the principal decision-making body of NATO. The cornerstone of NATO is the principle of collective defense, which is codified in Article 5 of the treaty. It states: “an armed attack against one or more of [the member states] in Europe or North America shall be considered an attack against them all”. The role of the NAC is thus to discuss and decide upon matters related to Article 5 and the other articles of the treaty. The NAC may also convene at the level of ministers or heads of state and government.

The decision to create the PSC was made at the Helsinki European Council meeting in

December 1999 with the intent to make it the “linchpin” of the EU Common Foreign and

Security Policy (CFSP). The main legal responsibilities of the PSC is to “deliver opinions” to

the Foreign Affairs Council and plan, prepare and oversee all EU crisis management operations.

The PSC thus drafts the council conclusions for all FAC meetings based on the agenda set by the High Representative.

6.1 The organizational level

The organizational aspects of the committees are relevant in order to grasp the context of the representatives’ working environments. According to Egeberg, this context has a large effect on the transformation process of state agents adapting to the norms and practices of an IO.

Specifically, Egeberg points to the “organizational locus” (i.e. the physical location of the organization or state agents) and “organizational demography” (i.e. the state agents’ age, sex, education and length of service) as relevant factors to this process.

This section will describe these aspects of both committees.

One of the main differences between the two committees is how the delegations to each IO are spread out. In the case of the PSC, the ambassadors are stationed in their respective Permanent Representations to the EU. These delegations are spread out in Brussels, most of which are located within roughly 20 minutes’ walk of each other. In comparison, all NATO delegations (and thus the permanent representatives of the NAC) are located in the NATO HQ on the outskirts of Brussels.

The demographics of the two committees are more or less similar. Both the PSC and the NAC are made up of exactly 23 men and 5 women, all of which are senior ambassadors with substantial work experience in the international community.

Since there is little personal information about the representatives available to the public, I have not been able to ascertain the average age of the representatives in the two committees. However, based on my personal impressions from my interviews with a relatively large portion of the representatives in each committee, the difference in the average age between the committees should be negligible.

59

The Lisbon Treaty, article 38.

60

Egeberg 2004.

61

As previously mentioned, this thesis is not able to test the effect of conditions that are unique for each individual. However, these factors are still of interest in order to get an understanding of the organizational context that the representatives work in.

62

http://www.nato.int/cps/en/natohq/who_is_who_39074.htm (accessed 15/11/2016);

http://europa.eu/whoiswho/public/index.cfm (accessed 15/11/2016).

6.2 The political level

One interesting similarity between the two committees is that while they have significant influence in the policy-shaping process within their respective IOs, this influence is to some extent dependent on how well the representatives manage to persuade certain agenda-setting actors within their respective IOs. Within NATO, the main agenda-setting actors are the International Staff (IS) and the Secretary General. The IS is composed of about 1,000 civilians who provide administrative support to the NATO delegations and in some cases implement the decisions of the NAC. The head of the IS is the Secretary General who also chairs the NAC meetings. This arrangement has given the IS considerable informal influence on the NAC agenda. One NAC representative emphasized this in the following way:

You’ll find ambassadors after the NAC finishes running over to the head of the table where the secretariat sits and doing their side deals or grabbing people as they walk out of the room, the deputy secretary general, the assistant secretary general and selling them one thing or another, lobbying them for this or that. All the time.

When it comes to the PSC, the main agenda-setting actors which the representatives must influence are the EEAS and the High Representative. Since the Lisbon Treaty entered into force, the PSC has been chaired by a permanent EEAS official appointed by the High Representative and not, as was previously the case, by the ambassador of the rotating presidency. This change has not only led to the High Representative gaining agenda-setting influence in the PSC, but also to a situation where the PSC representatives need to actively influence low-level EEAS officials in order to have any input on the initial text proposals.

One respondent who has work experience from both committees confirmed both the strength and similarity of the agenda-setting influence of the IS and the EEAS:

They both have very strong weight of their own. So you need to be aware of what they are doing, what’s cooking, and how you can influence that. That's similar.

To conclude, this section has shown that both the NAC and the PSC have important roles in the policy-shaping processes within their respective organizations. They are the de facto foreign

63

NAC respondent 9.

64

PSC respondent 4.

65

NAC respondent 1.

and security policy decision makers within the EU and NATO and their influence is not only informal, but permeates the formal organizational and political levels of both IOs. In other words, the NAC and the PSC should be understood as comparable cases.

7. Method

7.1 Research design

As previously mentioned, the purpose of this thesis is to test hypotheses related to the scope conditions for socialization in IOs. I have argued that a fundamental aspect that has been left out in previous research on socialization in IOs is a comparative perspective. By comparing the level of socialization and existence of six different scope conditions within the NAC and the PSC, this thesis is able to test the relevance of the hypotheses presented in previous studies.

However, the logic of a comparative method can not be taken for granted. As such, the research design of this study will be described in greater detail below.

7.1.1 Mill’s methods

When it comes to comparative methods there are a number of different alternatives.

More or less all of these methods are derived from John Stuart Mill’s A System of Logic (1843). In this work, Mill argued that there are two ways to conduct a controlled comparison, the “method of agreement” and the “method of difference”.

The method of agreement is used to identify the independent variables that are similar when two or more cases share a common outcome. The method of difference uses the opposite logic and is thus used to identify the independent variables that are associated with a different outcome. The purpose of these two methods is simply to identify the factor or factors that can explain a certain outcome. The two methods are visualized in the following figure:

66

See for example “Most similar/different systems design” (Przeworski & Teune 1970) and “The comparable-cases strategy” (Lijphart 1975).

67

George & Bennett 2004, p. 153.

Overall similarities

Crucial difference Overall

differences

Crucial similarity

Figure 1: Mill’s methods

The use of Mill’s methods becomes fairly self-evident when our goal is to test hypotheses.

Based on the outcome of the dependent and independent variables, we may use either the method of agreement or the method of difference to explain why the outcome is either similar or different between two comparable cases. Depending on the method that we use, we may thus reject either the hypotheses or conditions that differ between the cases (if we find that the outcomes are the same) or the hypotheses where the conditions have the same level in both cases (if we find that the outcomes differ).

In the case of the NAC and PSC, we are thus not able to choose one of Mill’s two methods until we have established the level of the dependent variable, the internalization of group norms and role conceptions. As shown in the figure above, another requirement of Mill’s methods is that the cases we compare should be either completely different or completely similar in all aspects except the studied relationship (how x effects y).

The logic behind this requirement is that we simply want to isolate the relationship between x and y to be able to show the effect of x, or in this case, the effect of the tested conditions on the degree of internalization. In practice, this is almost impossible to do. It is generally extremely difficult to find two cases that are comparable in all respects. As such, Mill’s methods should be seen more as ideal types that we can use to approximate the relationship between a condition and an outcome. As seen and stated in section 6, the NAC and PSC should thus be understood as comparable cases. Another issue with Mill’s methods is that they can not be used to confirm

68

Adapted from Skocpol & Somers 1980, p. 184.

69

Teorell & Svensson 2007, p. 225-230.

Case 1 Case 2 Case 1 Case 2

a b c x

d e f x

a b c x

a b c not x

y y y not y

Method of agreement Method of difference

Key:

x = casual variable

y = phenomenon to be explained

hypotheses, only make them more plausible.

It is still possible that the conditions may or may not be present (depending on the chosen method) in other cases where the outcome is still either the same or different. In other words, if we were to attribute a causal significance to the conditions/independent variables that seem to be associated with either the correlation or variance of the outcomes, we could be making a type 1 error – a “false positive”.

We can thus only falsify hypotheses, not confirm them.

A third issue that is related to the use of Mill’s methods is that of multiple causation or

“equifinality”. Equifinality is the condition of a social phenomena having “alternative determinants” – that the same outcome can emerge in different cases via a different set of independent variables.

In other words, equifinality makes it a bit harder to falsify hypotheses through a comparative method. However, there is some discrepancy in the literature as to how serious the problem of equifinality is. While King, Keohane and Verba argue that equifinality does not undermine the use of comparative methods to make “probabilistic claims” (such as:

“there is a high probability of poor communication among super powers during crises leading to war”)

, George and Bennett argue that a comparative method must be complemented by a process tracing method to ensure that the hypothesized causal relationship is not spurious.

Unfortunately, to also conduct a process tracing study to confirm the results of the comparative method lies outside the scope of this thesis. Instead, the aim of the thesis is to produce plausible alternative hypotheses that can be tested through a future process tracing study. I believe that laying this groundwork strengthens the probability of the claims that this thesis makes on the basis of the comparative method. In other words, by simply formulating credible and logically consistent alternative hypotheses, the risk of making a type 1 error decreases.

7.2 Measuring the conditions – elite interviews

To test the hypotheses that have been presented above, this thesis will use elite interviews to measure the existence of the conditions that lie behind the hypotheses. Elite interviews are mainly used to target actors that are personally involved in a political process and as such have

70

George & Bennett 2005, p. 156.

71

Ibid.

72

George & Bennett 2005, p. 157.

73

King, Keohane & Verba 1994, p. 87.

74

George & Bennett 2005, p. 156-157.

a unique insight into the chain of events that lead up to a political outcome.

Since a common trait of high-level committees in IOs is that their work to a large degree goes on “behind the scenes”, elite interviews are necessary in order to gain a sufficient amount of insight as to which conditions exist in our two cases. As with every method, elite interviewing has its strengths and weaknesses. When it comes to the validity of the method, elite interviews are specifically useful to minimize systematic measurement errors, as would be the result of interviewing “non-elites”, or lower level bureaucrats, who have an inferior understanding of the “actual” way that socialization works in our two cases. One precaution that must be taken into account when interviewing elites is to not pose too sensitive questions that the respondents might not be able to answer because of confidentiality.

Since the members of the NAC and the PSC have access to extremely sensitive information, some level of “censorship” should be expected. However, to minimize the risk of the respondents not answering the questions, the conditions have been operationalized as “non-sensitively” as possible without being too abstract or unclear.

While the level of validity of the elite interview method is largely dependent on minimizing the conditions that could affect the truthfulness or correctness of the responses given, a high level of reliability is the result of minimizing non-systematic measurement errors. This has been achieved mainly by making sure that the operationalizations of the tested hypotheses and conditions, in other words the interview questions, are as concrete and precise as possible. King, Keohane and Verba argue that the best questions to ask in interviews are those that do not let the respondent “do the work for us”.

For example, if we want to know if a certain independent variable affects our dependent variable, we cannot ask our respondent if there is a causal effect between the two, but rather ask for measures of the two variables and estimate the causal effect ourselves. The risk of letting the respondent “do the work for us” is that they could give very different answers depending on how they interpret the question.

The way in which a researcher poses a question about a causal effect (for example implying that it is something positive or negative) can also influence the response. Avoiding these pitfalls increases the reliability of the interview data and the interview questions have thus been formulated with this in mind. Another factor that may affect the reliability of the interview data is how structured the interviews are. A fully structured interview runs the risk of not providing a completely

75

Beamer 2002, p. 86.

76

Ibid., p. 93.

77

King, Keohane & Verba 1994, p. 112.

78

King, Keohane & Verba 1994, p. 112.

accurate depiction of reality since the respondent may not freely speak his or her mind (lower validity) while an unstructured interview is more prone to unsystematic measurement errors as the respondents may interpret the questions differently (lower reliability).

A good “middle way” is thus to conduct semi-structured interviews so that one can not only follow a prepared list of questions, but also pose follow-up questions. I have thus chosen to use this interview strategy to ensure high levels of both validity and reliability.

When it comes to the sampling of respondents, I have simply chosen to interview as many representatives within the two committees as possible. One could argue that a more structured sampling method would have been preferable to maximize the generalizability of the sample (for example by ensuring that the sample consists of an equal amount of permanent representatives from Southern, Northern and Eastern European member states), but since the total possible sample size is in itself quite small (28 for each committee), it would not be beneficial to limit the sample size any further. I have thus contacted every delegation in NATO and the EU via a template e-mail in which I have specified some background information about my thesis and requested an in-person interview in Brussels. Of the 56 delegations, I received a reply from 28, 7 of which either replied that they were unavailable or did not respond to follow- up e-mails. As such, 21 delegations invited me to meet with their permanent representative or deputy permanent representative. Since the deputies within these committees are one rank below the member state’s ambassador, one could argue that they do not possess the exact same insight as the representatives themselves. However, as the deputies are always present during the committee meetings, take part in most (if not all) preparatory discussions before committee meetings and will fill the representative’s position if they are unable to join a meeting, their insight into the workings of the committee and the existence of the tested hypotheses and conditions should be understood as more or less the same as that of the permanent representatives. In total, I have interviewed 14 permanent representatives (7 from the NAC and 7 from the PSC) and 7 deputy permanent representatives (3 from the NAC and 4 from the PSC).

This interview data is, to my knowledge, the largest of its kind on the permanent representatives of the NAC.

Larger studies on the socialization of PSC-representatives have previously been

79

Teorell & Svensson 2007, p. 89.

80

The only previous study on the socialization of NAC representatives is, to my knowledge, Gheciu

(2005). However, of her 80 respondents, only 4 were current permanent representatives or deputies.

done by, for example, Jolyon Howorth.

The complete list of respondents can be found in figure A1 in the appendix.

A few short comments shall also be made on the more practical aspects of the interviewing process. During the first week of October 2016 I travelled to Brussels to conduct most of the interviews. This was very helpful since most of the respondents preferred to be interviewed in person rather than via telephone. 18 out of 21 interviews have thus taken place in Brussels, either at NATO Headquarters or at the member state’s permanent representation to the EU. The rest (3) of the interviews have been conducted via telephone or Skype. The respondents were all informed that the interviews would be treated with anonymity and the respondents are therefore referred to as “NAC respondent 1”, “NAC respondent 2” and so on in the following sections. Of the 21 interviews, 15 have been recorded and transcribed. The combined transcriptions of these 15 interviews constitute a total of 55,342 words from 7 hours and 34 minutes of recordings.

The rest of the respondents preferred to not be recorded and the material from these interviews is therefore limited to my own notes. The interviews have lasted in average around 30-40 minutes, with some lasting longer (2 hours) and some shorter (20 minutes).

7.3 Generalizability

The generalizability of this study is limited to committees of permanent representatives in IOs.

As such, this study is not interested in the same actors as the previously mentioned studies on the conditions for socialization in the European Commission. These studies focused on “high- level officials”, i.e. mostly civil servants employed by the Commission to work in the many Directorate-Generals (DGs). A significant distinction should be made between these actors and the permanent representatives studied in this thesis. Jeffrey Lewis argues in his article The Janus Face of Brussels that there are “qualitative differences” between the expert-level working groups in the European Commission and the main preparatory body of the Council, COREPER.

Since COREPER consists of national permanent representatives, he argues, there are substantial organizational characteristics that differentiate it from the Commission expert- level working groups, for example in terms of the frequency and intensity of socialization

81

Howorth 2010.

82

A redacted version of this document can be made available upon request.

83

Lewis 2005, p. 968.