In Whose Secret Service? A comparative case study on changing relations between the intelligence and policy community in transitioning democracies of Eastern-Europe

(1)

In Whose Secret Service?

A comparative case study on changing relations between the intelligence and policy community in transitioning democracies of Eastern-Europe

Dolinda Molema S1362003

Supervisor: dr. G.G. de Valk Second reader: mr. drs. W.J.M. Aerdts

09-06-2019 The Hague

(2)

2 “What we may be witnessing is not just the end of the Cold War, or the passing of a particular

period of postwar history, but the end [goal] of history as such… That is, the end point of mankind’s ideological evolution and the universalization of Western liberal democracy as the

final form of human government.” – Dr. Francis Fukuyama, 1992 …

Abstract

During the Cold War, invasive intelligence systems in the authoritarian states of Eastern Europe were in direct service of the Communist Party. After the end of the Cold War, these countries started transitioning into democratic governments, whose intelligence services operate in service of the rule of law. It has not yet been researched how the democratic reforms affected the possible change in these relations between the intelligence and policy communities, despite the potentially grave implications for civil liberties and international intelligence cooperation. To fill this knowledge gap, this thesis answers the following research question: How have institutional reforms of intelligence organizations during the democratic transition of former authoritarian communist states in Eastern Europe that are now part of the EU affected the formal relation between policymakers as principal and intelligence agencies as agent? To answer this question, a qualitative comparative case study of eleven Eastern European countries was undertaken. It was found that the creation of a strong legal-institutional accountability mechanism is the most important institutional democratic reform, as it appears to be a sufficient but not necessary condition for a responsive relation between the intelligence and policy communities. The research also suggests that a democratic disposition of the political principals is a necessary condition for a responsive relation. Lastly, there also appeared to be a relation between an active intelligence relation with Russia and increased levels of intelligence politicization.

(3)

3

1 THESIS OUTLINE

1.1. Introduction

In the past few years, an important political change has been taking place. In many countries around the world, the path to more democratic governance appears to be reversing. The growth of (right-wing) populism and the rise of dictatorially inclined rulers are leading to the weakening of power constraints and democratic governance in general (Whitaker 2019: 3). For former communist Eastern-European states such as Hungary under Orbán’s rule or Poland under the Law and Justice party, this means a halt or reset of the democratic reform agenda that was installed after the fall of the USSR. A vital part of this agenda was the democratization of the intelligence apparatus, as these institutions used to be a notoriously powerful tool for the ruling communist elite in Eastern Europe. Despite the intelligence apparatus’ considerable power through secret knowledge, there has been no research as to whether these democratic reforms also entail that these services will operate in service of the rule of law rather than in service of the political parties as was the case during the Cold War. With the exception of East-Germany’s Stasi, which was dissolved altogether, all other Eastern European post-authoritarian states chose to reform the agencies as they could not afford to lose vital intelligence during a rebuilding period (Watts 2004: 11). These reforms consist of institutional elements such as whether oversight committees were designed to oversee or rather overlook, whether old personnel is still in place and whether they train new personnel in the old ways. Depending on the results of these reforms, secret state intelligence could be characterized by continuity or change in their operating procedures. In case of continuity, the agencies could once again be involved in the ruling party’s struggle for growing power consolidation. Especially in light of the current de-democratization movement, the way in which the relation between the intelligence and policy community has been institutionalized during the democratic transition process could have serious consequences for the civil liberties in those states. This is also of great import to the European Union as internal intelligence cooperation would entail severe complications and possibly detrimental consequences if the intelligence apparatus does not operate in service of the rule of law.

1.1.1. Research Question

Consequently, my research question is as follows: How have reforms of intelligence organizations during the democratic transition of former authoritarian communist states in Eastern Europe that are now part of the EU affected the formal relation between policymakers as principal and intelligence agencies as agent? To find an answer to this question, a qualitative case study comparison will be undertaken of eleven countries (Estonia, Latvia, Lithuania, Hungary, Poland, Romania, Croatia and Slovenia (former Yugoslavia), Bulgaria, Czechia and Slovakia (former Czechoslovakia) by means of an interview, process tracing and (historical) content analysis of state documents and legal acts. By reviewing the legal-institutionalized accountability mechanisms, their effect on politicization will be analyzed and explained with the principal-agent theory. For this research to remain feasible, it is limited to the institutional relationship, and does not go in the informal networks.

(7)

7 1.1.2. Societal Relevance

In terms of societal relevance, the underappreciation in the field of democratic transitions for the potential of the instrumentalization of the intelligence sector by political actors has become a pressing deficit given the de-democratization processes in some Eastern European countries (Caparini 2014: 521). To address this deficit, this thesis will assert how the democratic reforms have affected the politicized relations between the intelligence and policy community and what this entails for the civil liberties in those countries. These outcomes of what elements of reform were most or least successful in ensuring an accountable intelligence apparatus in their relation to the policy community could be used for the EU’s or NATO’s membership requirements as well as shed more light on how future transitioning state (for example Arab Spring nations) can best consolidate democratic change. In order to achieve this, the research will also serve a practical policy function as it evaluates the current situation for each country in a comparative setting, thus generating an overview of empirical data on the intelligence organization in all these countries. Such an overview hast not yet been produced in an academic setting. The results should also evaluate to what extent the current intelligence organization of these newer democracies pose risks for intra-EU cooperation in intelligence sharing, which has only become more important as globalized threats such as terrorism, cyberattacks and organized crime affect the borderless European Union. As intelligence consumers use reports for the advancement of their own political interests, the credibility of their intelligence apparatus as well as the trust relation with the international community becomes impaired (Akrap and Tudjman 2016: 76).

The thesis outline will continue with a short overview of what has already been established in the field of intelligence democratization studies in order to ascertain a knowledge gap and the academic relevance of this thesis, which will follow in §1.2.4. After that, the methodology of the research will be outlined.

1.2. Status Quaestionis

In the following section, an introduction into the field of intelligence democratization is provided to understand what the main themes of research are and how this thesis fits into those themes. Then, an overview of what has already been established will be provided to see what has not yet been researched. Finally, by identifying this knowledge gap, the academic relevance and potential contribution of this thesis to the field of intelligence studies can be discussed.

1.2.1. Introduction into the field of intelligence democratization

“Power corrupts, absolute power corrupts absolutely, and everything secret degenerates” (Andregg and Gill 2014: 496). Following these principles, the question arises whether secrecy is bound to be the enemy of democracy, or can the work of intelligence organizations be remedied to fit into democratic governance? While democratic control requires transparency, openness and accountability, secrecy is paramount for the effective protection of national security. In post-authoritarian regimes, the secret nature of intelligence is even more problematic, as the function of intelligence organizations was not embedded in legal frameworks, where these organizations were used against civil liberties rather than ensuring their protection and where security institutions often relied on secrecy for their power. Consequently, the central issue in the field of intelligence democratization is the struggle of the inherent tensions between security and civil liberties, effectiveness and transparency: between intelligence and democracy. Is there a fundamental

(8)

8 incompatibility between these concepts, meaning that the idea of ‘democratic intelligence’ is inherently oxymoronic? (Gill in Whitaker 2019: 1; Holt 1995: 1 in Bruneau and Matei 2010: 757). The tension between accountability and effectiveness in the field of intelligence democratization has consequently led to two main focuses: the first looks into the establishment of accountability and scrutiny mechanisms to remedy these tensions and evaluate democratization processes in post-authoritarian regimes, sometimes compared to those of older democratic intelligence reforms to discover best practices in intelligence democratization. The second focus in intelligence democratization studies is that of the effectiveness of intelligence operations during or after the reforms, which became a top priority after 9/11, Madrid (2004) and London (2005) (Bruneau and Matei 2010: 770). This thesis is only concerned with the first aspect: accountability.

1.2.2. The study of accountability in intelligence democratization

The accountability studies of intelligence democratization consist mainly of empirical case studies and policy evaluations which are mostly comparative and descriptive in nature. The small field of intelligence democratization in post-authoritarian countries consists of empirical (comparative) case studies in three discernible regions respectively: Latin-America, with a focus on the changing civil-military relations through rational choice analysis as military junta regimes transition into democracies, sometimes including Spain in the analyses (Estévez 2014; Weeks 2018; Bruneau and Matei 2010; Boraz 2006), South Africa (Dombroski 2006) and Eastern Europe (Watts 2004; Matei 2014; Bruneau and Dombroski 2014; Caparini 2014; Stan and Zulean 2018). In this literature, there is a lack of theoretical analyses, which leads to an absence of a theoretical debate with mutually exclusive positions. The questions that are most often dealt with are: What measures have been undertaken, how have these affected the democratic level of intelligence organizations and/or what conditions affect these outcomes?

1.2.3. Knowledge gap

The study of accountability in intelligence democratization leaves two important gaps that this thesis aims to address. Firstly, the field of intelligence democratization in transitioning democracies itself has hardly been explored yet, just like research into intelligence organizations outside Anglophone Western states. In the field of security sector reforms in post-authoritarian countries, much attention has been given to the democratization of institutions such as the military and the police, but the governance and reforms of intelligence services is still under-explored both by academics and by policy evaluators (Caparini 2014: 499). Unlike research into older Western democracies, few studies have been conducted into newer democracies. Matei and Bruneau looked at the five most used and from the past ten years and discovered that only 22 of the 161 chapters dealt with the intelligence apparatus in states that are moving from a dictatorship to a democracy (2011b: 657). Another bibliography study on intelligence research by Greta Marlatt showed that a review of over 500 books, articles and reports, only thirty are concerned with newer democracies (Matei and Bruneau 2011b: 657). The most widely used case in intelligence studies is that of the USA, as Andregg and Gill found that of 124 articles, 64% is about the United States, which can be explained by the fact that 70% of all the authors were Americans (2014: 488). This means that there is still a shortage of comparative perspectives of cases outside of the anglophone, established Western democracies (Caparini 2014: 499). The body of literature about intelligence processing and its interaction with politics gives the impressions “that they are all Americans who study this intricate relationship in the American context” (Bar-Joseph 2013: 349). Secondly, in the niche

(9)

9 field of intelligence democratization studies in none-Anglophone countries, there have been no studies specifically concerned with the changing relationship between policy-makers and the intelligence community outside of post-military regimes (Otamendi and Estévez 2016: 5). Consequently, there has been no research into what the effects of democratic reforms are on the relation between the intelligence and policy community.

1.2.4. Academic Relevance

So, how can the following research question add value to the academic field of intelligence studies within the interdisciplinary study of Crisis and Security Management: “How have reforms of intelligence organizations during the democratic transition of former authoritarian communist states in Eastern Europe that are now part of the EU affected the formal relation between policymakers and intelligence agencies?”

The primary academic goal of this thesis is to explore the yet unresearched changes in the formal relation between the policy and intelligence community as a consequence of democratic reform efforts. This should contribute to a further broadening of intelligence democratization studies. The provision of more case study material could also enhance the scope and validity of the literature on best and worst practices of intelligence democratization. Adding more case studies could also provide more validity in the field of intelligence studies in general as it contributes to redressing the current US bias.

A second objective is to test the applicability of principal-agent theory in this field. By treating the intelligence apparatus as a bureaucracy, the principal-agent theory will be used to try and explain the changing relationships. The application of this theory, often used in the field of public administration, tests whether the principal-agent model for bureaucracies and political authority can be extended into the distinct realm of intelligence studies. Considering the special jurisdictions, operations and relations with politics it is interesting to see to what extent these intelligence organizations can be analyzed through theories of ‘regular’ bureaucracies. This could shed more light on the generalizability of the theory in a new environment. Besides proofing the principal-agent theory, four other alternative hypotheses from the intelligence democratization literature will be tested, which should help the falsification process of the new theories in this young field of study. Lastly, four new factors found in the empirical data will be explored in an inductive manner from which one new significant element arises.

1.3. Methodology

To see how this knowledge gap can be addressed with the proposed research question, the following section will go into the applied methodology and a reflection upon its limitations. First the research design will be explained and justified in more detail, then the research methods will be outlined more specifically. As secrecy makes research more necessary, but also more difficult, an adequate methodology is imperative (Otamendi and Estévez 2016: 6).

1.3.1. Research Design

The chosen research design is a qualitative comparative case study with methods of (historic) content analysis, process tracing, and interviewing. For this research in particular, a qualitative comparative case study research design facilitates the comparison of different impacts of varying

(10)

10 accountability mechanisms on differing degrees of politicization, while taking into account the country-specific context as an intervening variable. This should thus provide insight into why some reforms worked out better than others in terms of reducing unwanted political uses of the intelligence system.

There is a general consensus in the field of intelligence democratization studies that quantitative studies are both impractical and objectionable since quantitative databases either do not exist or are not accessible and because there is no agreement on possible numerical measurements of concepts related to democratization and intelligence (such as accountability or political-bureaucratic relations) (Gill 2016; Davies and Gustafson 2013 in Otamendi and Estévez 2016: 6; Yang 2014: 161). Qualitative approaches into the study of intelligence democratization and intelligence accountability are thus preferred as this type of research is more suited for questions whose answers require “an understanding or explanation, occur over time, or are difficult or sensitive to define” (Yang 2014: 159). Since this thesis aspires to generate an understanding and explanation of the type of relation between the changing political and intelligence systems, a qualitative study is necessary to adequately interpret and analyze the differences and similarities. A qualitative research also facilitates comparative research by allowing for understanding, interpreting and analyzing differences and similarities between intelligence organizations within a wider cultural-political context (Whitaker 2019: 1). Since the comparative case study is a tried and tested methodology in the study of intelligence democratization for analyzing and explaining the similarities and differences between national intelligence structures, it will also be used in this thesis to assess the different impacts of variations in democratic reforms and accountability mechanisms on variations in degrees of politicization (Whitaker 2019: 1).

The analysis of the case study comparisons will take place on both a within-case and cross-case level. First the values of both variables are asserted via eleven within-case analyses using the data on the indicators from the case descriptions. With the aggregation of these values, a cross-case analysis is conducted. The main hypothesis that the research wants to test is: “A strong legal-institutional accountability mechanism ensures that intelligence organizations operate both under the rule of law and in service of the rule of law (ergo with low levels of institutionalized intelligence politicization)”. So, by looking at how variations in the independent variable (strength of legal-institutional accountability mechanisms) affect the dependent variable (level of institutionalized intelligence politicization) over multiple cases, the cross-case analysis aims to ascertain a correlation. This constitutes an ‘x-centered’ analysis (or Co-Variational Analysis), a type of cross-case analysis, as this research thus primarily focuses on what difference the strength of the independent variable (X) makes on the outcome of the dependent variable (Y) (Blatter and Haverland 2012a: 81).

1.3.2. Research Methods: Data collection and analysis

As in nearly all research in the field of intelligence, the availability and accessibility of data is a major concern for data collection methods. Most of the studies of intelligence democratization rely on traditional historical and institutionalist approaches for available data, and this one will be no exception (Bruneau and Matei 2010: 771). The first data collection concerning the historical setting of the relation between the intelligence and policy communities before the democratic transitions (Y1) will take place by accessing secondary data in the form of reports and journals for

(11)

11 each of the cases. Then, to outline the democratization processes and institutionalization of intelligence accountability mechanisms for democratic governance (X), mostly secondary data is used from previous research, reports from oversight committees, governments and non-governmental organizations (human rights watches) and law books (concerning the legal accountability framework). Lastly, for data on the current relation between the intelligence and policy communities (Y2), secondary data from investigative journalism and reports will be used

next to primary source data from a semi-structured interview with Mr. Peter Keller on March 19th. Mr. Keller worked for the former Dutch intelligence and security service BVD from 1969-1991, was active in a working group on intelligence for the cabinet and worked for the Directorate of Security for the European Commission from 1991-2007. He is an expert on the relation between the intelligence and policy community and has working experience with the transitioning democracies of Eastern Europe from his work for the European Commission. As an intelligence expert who has visited all the intelligence institutions that this thesis researches, an interview technique with little structure was chosen as to enable Mr. Keller to share as much of his expertise and knowledge as possible. The interview questions and transcript can be found in the appendix §8.3.

To analyze the data, first a content analysis of the reports is conducted by coding the variations in all qualitative indicators (operationalized in §2.4). In the case descriptions (chapter 4), the data on the past relation between the intelligence producers and consumers, on the democratic reforms and on the current relation will be outlined. From this data, the eleven individual within-case analysis are derived that ascertain how the formal relations between the policy and intelligence community changed due to the reforms. In the cross-case analysis (chapter 5), all the differences and similarities in the democratic reforms and the changes in politicization levels will be analyzed across all cases. The emphasis in the cross-case analysis is on the co-variational analysis that assesses the strength of the legal-institutional accountability mechanisms as main democratic reform on the current politicization levels.

1.3.3. Case Selection

For the Co-Variational Analysis (COV), the case selection is crucial in demonstrating that the independent variable acts as a counterfactual in the causal mechanism (Blatter and Haverland 2012b: 10). This means that the cases must have strong differences in the independent and strong, constant similarities in other possible control variables, which is why it is best to use a Most Similar Systems Design (MSSD) (Blatter and Haverland 2012b: 15). Therefore, a subregion was chosen. This sub-region approach is common in comparative case selection for intelligence democratization studies, as important historical and cultural similarities facilitate the explication of similarities and significant differences in processes and outcomes (Andregg and Gill 2014: 490). However, despite the attempt at controlling intervening variables with the MSSD approach, due to the many factors in socio-political settings and contexts, a correlation is often the best possible outcome in the social sciences. So, no claims on causation will be made.

The selected units of analysis are the intelligence and security services within the broader political government in the following eleven states of Central and Eastern Europe (CEE) that moved from an authoritarian, communist state to a democratic member state of the European Union: Estonia, Latvia, Lithuania, Hungary, Poland, Romania, Croatia and Slovenia (former Yugoslavia), Bulgaria, Czechia and Slovakia (former Czechoslovakia). The time frame of the research is to compare the structure in the years before 1989 (the fall of the Berlin Wall and start of fall of the

(12)

12 communist regime in Central and Eastern-Europe) to those of today, so it concerns two points in time. These cases were also selected due to their relevance for wider European intelligence cooperation due to their accession into the EU. With eleven cases, there should be enough variation to extract explanations and patterns. Choosing the same time periods also allows to see what factors work fastest and which cause a lacking behind to neighbor-states.

1.3.4. Limitations

What are the limitations to this research design? Firstly, in terms of reliability, the consistency of the measured values will be enhanced by clear indicators. Despite the aspiration for precision and accuracy to ensure repeated measurements, qualitative studies are subject to inevitable interpretation of the meaning of events. Unfortunately, the reliability of the data could not be enhanced by more interviews as confidentiality in the field of intelligence rendered it impossible to interview other intelligence experts with inside knowledge of the democratization processes in the services of Eastern Europe (P. Keller, personal interview, March 19, 2019).

Secondly, concerning content validity, not all aspects of the concepts are taken into account. The deliberate decision to exclude informal aspects of the intelligence-policy relations and the effectiveness of intelligence to make research more feasible does mean that the total picture of the research is incomplete. However, it is simply not possible to include all elements due to time and scope limitations. Since the democratization process is non-linear, comparing two points in time only gives an indication of the mechanisms at play and does not clearly portray a causal relation. This also affects the internal validity to a certain degree. Alternative explanations of the causal mechanism could include informal structures, but since this is extremely difficult to research due to accessibility issues, the research is forced to exclude these elements for feasibility sake. Lastly, the choice of a comparative case study renders the external validity very limited, as the empirical study into cultural-political, context-sensitive units of analysis does not allow for much generalization. In terms of external validity, the region-specific political and socio-cultural characteristics of the sub-regional case selection do not allow for wider generalization to other cases. This is mainly because other intervening variables might be present in other cases, or because intervening variables might act differently in other cases of the population. For example, in Latin-American countries that transitioned form a military junta into democracies, the (inter)national context is fully different. The main generalization will therefore be back to the principal-agent theory and the causal mechanism.

1.4. Reading guide

The rest of the thesis shall continue as follows. First, there will be a conceptualization of all the used concepts and the operationalization of the most important variables will be outlined. Secondly, the theoretical framework will outline the used principal-agent model followed by the causal mechanism and a number of hypotheses. Thirdly, the eleven case descriptions will follow with eleven individual within-case analyses. Fourthly, a cross-case analyses will be conducted in which the hypotheses will be tested and the outcomes explained. Lastly, the thesis ends with a conclusion that answers the research question and reflects on its societal and academic implications. That final chapter also contains a discussion of the limitations of this research and recommendations for further research.

(13)

13

2 CONCEPTUALIZATION

In this second chapter the main concepts will be explicated and the used variables will be defined in order to create a common ground of what will be discussed. The conceptualization will also facilitate the operationalization later in this chapter, which aims at enhancing the general reliability and validity of the research.

2.1. What are the main concepts in the literature?

To provide some background information and clarity on the terms and concepts that will be the basis of the thesis, the main concepts in the literature on intelligence democratization will be briefly discussed.

2.1.1. Intelligence

At the heart of this thesis lies the concept of intelligence. The purpose of intelligence is to create an information advantage used for a decision-making advantage relative to uncertain and risky states opponents (domestic or foreign), either by advancing one’s own information or by disrupting that of the opponents (Warner 2009). The main duty of intelligence is to give early warnings concerning both dangers and opportunities for the national government. For this to be effective, intelligence operations need to adhere to its central tenets of timeliness, accuracy, completeness, usability, and relevance (Matei and Bruneau 2011b: 660). Whereas there is no one unified definition of intelligence, there is an academic consensus that it can be considered as three things: intelligence as a process, product and as an organization (Lowenthal in Bruneau and Dombroski 2014: 8). This thesis focuses on the third element as it is concerned with the intelligence apparatus and its relation to the policymakers. In this sense, intelligence can refer to a centralized organization part of the national government that is involved in the process and product of intelligence. Within this category, various distinctions between intelligence organizations can be made, for example across the type of intelligence activities they undertake (e.g. SIGINT, HUMINT, IMINT etc.), the type of threat they deal with (domestic or foreign) or whether it is a military or civilian agency. For this research, it is important to clarify the distinction between an intelligence service, that provides an offensive protection for national security from threats abroad, and a security service, that provides a defensive protection for either a political party, a state or the rule of law in general from domestic threats (De Valk 2005). A security service can be granted executive powers of law enforcement akin to those of police forces, such as the ability to criminally investigate, arrest or even imprisonment of suspects. These types of security services are an exception in the Western world, as it is preferred to separate these powers to prevent abuse and enhance the democratic level of the institutions (Vitkauskas 1999: 28). The element of executive powers will return in the cases as a way to distinguish the different agencies in Central and Eastern Europe.

2.1.2. Democratization of Intelligence

In this research, the democratizing reforms of these intelligence organizations form the central process that will be under investigation. The democratization of intelligence is defined in accordance with most literature on the topic, namely as the process of establishing a legal-institutional framework in which intelligence agencies operate under the rule of law and are being

(14)

14 controlled by democratically elected civilians, while maximizing their effectiveness (Matei and Bruneau 2011a: 606; Andregg and Gill 2014: 491). This definition is preferred over the frequently used term ‘democratic transition’ as this implies a destination of democratic intelligence rather than a process (Andregg and Gill 2014: 491). The concept of democratic consolidation is used to define such a final state in which the democratic regime’s legal-institutional structures and processes are becoming stable and that its institutions are supported both by the elites and the masses while no unelected body has authority over popularly elected bodies (Bruneau and Dombroski 2014: 7).

2.1.3. Difference between Intelligence Organizations under Authoritarian and Democratic Rule To clarify the required changes during a democratization process, a further distinction between the aforementioned democratic regime and the authoritarian and totalitarian regimes and their appertaining intelligence organizations needs to be made. Firstly, in authoritarian regimes (characterized by a strong central power, limited political and individual freedoms, lack of constitutional accountability and rule of law), the lack of true legitimacy leads to ruling elite’s use of the security and intelligence agencies to suppress any opposition for the preservation of the regime, without any oversight of their activities (examples include Czechoslovakia's StB, Romania's Securitate, Russia's KGB and East-Germany's Stasi) (Bruneau and Dombroski 2014: 6). As an extreme form of these authoritarian states, in totalitarian regimes (characterized by sitting at the extreme end of the authoritarian regime type with outlawed opposition parties and prohibited individual opposition), the state gives ever more power to the intelligence organizations in order to severely control all aspects of both public and private life (examples include former Brazil’s SNI, Chile’s DINA, Iran’s SAVAK and South Africa’s BOSS) (Bruneau and Dombroski 2014: 7). Here too, oversight or control lacks as the service gains more independence as they only answer to the state without being subjected to oversight. So, in both cases they are in service of the ruling elite (a political party, military junta or single despot) rather than the rule of law.

2.2. Conceptualization of the independent variable:

Democratic reforms (legal-institutional accountability mechanism)

Now that the general concepts are outlined, the most important concepts for this thesis will be discussed: the independent and dependent variables. In the research question (How have institutional reforms of intelligence organizations during the democratic transition of former authoritarian communist states in Eastern Europe that are now part of the EU affected the formal relation between policymakers as principal and intelligence agencies as agent?), the broad independent variable (X) constitutes the institutional democratic reforms and the dependent variable (Y) constitutes the degree of politicization in the relation between the intelligence and policy communities. The most significant democratic reform measure required to transform an intelligence agency under authoritarian rule into one under democratic rule is to create a solid legal framework that establishes clear accountability (Andregg and Gill 2014: 493; Matei and Bruneau 2011b; Bruneau and Matei 2010). The main goal of this research is therefore to discover how (legal-institutional) accountability mechanisms as the most important part of democratic reforms can affect the relation between the intelligence and policy communities. (The main independent variable legal-institutional accountability is a specified element of accountability, which will be extrapolated further in the operationalization in §2.4.1). For now, it is important to first define the broader concept of accountability in order to explain exactly what is meant and to

(15)

15 facilitate the operationalization and measurement of the concept in the empirical research later on. The conceptualization of the dependent variable of institutionalized intelligence politicization will follow in §2.3.

2.2.1. Accountability

Among scholars, there is no consensus on the concept of accountability. It is often interpreted differently by different academics that use their own conceptualizations and frames, making accountability an ever-expanding concept with synonyms such as “good governance, transparency, equity, democracy, efficiency, responsiveness, responsibility, and integrity” (Behn 2001: 3–6; Dubnick 2007a; Mulgan 2000: 555 in Bovens 2010: 947). Generally speaking, the concept of accountability has two dimensions: accountability as a virtue (which constitutes a normative concept) and accountability as a mechanism (which constitutes a narrower and more descriptive concept of being accountable) (Bovens 2010: 946). In the latter dimension, the focus is on the way the institutional accountability arrangements operate in which an actor can be held to account by a forum (Bovens 2010: 946). Since this thesis is merely concerned with formal accountability institutions, the definition will be limited to accountability as a mechanism. Since Bovens created a narrow, functional concept of accountability to facilitate comparative research, his following definition will be used in this research: “a relationship between an actor and a forum, in which the actor has an obligation to explain and to justify his or her conduct, the forum can pose questions and pass judgement, and the actor may face consequences“ (Bovens 2010: 951).

2.2.2. The Forum

To explain this definition of accountability and its link to the later used principal-agent theory, the concept of the forum found needs further discussing. The defining element of accountability is that it entails a relation between a forum and an agent, as seen in the previously established definition. Accounting, therefore, means the requirement of the intelligence agencies to explain and justify their actions vis-a-vis a significant other, who can decide to impose sanctions if the agency exceeded their legitimate authority (also known as accountability error) (Bovens 2010: 951; Peters 2014: 211). The element of sanctions is often under debate by scholars, but since “sanction-based accountability is appropriate in contexts of justified mistrust”, we use the element in this thesis as it is concerned with secret intelligence (Mansbridge 2014: 55). Under these conditions, a forum can balance the power with that of political power of policy community and power through secret knowledge of intelligence agency. Therefore, accountability mechanisms are at the core of democratic reforms, provided they are ‘performing’ material mechanisms rather than inactive, formal mechanisms. To truly assess the strength of the accountability mechanisms, it is therefore important to make a distinction between such accountability mechanisms that were designed to overlook (formal) and those that were designed to truly oversee (material) (Andregg and Gill 2014: 489).

2.2.3. Accountability: Control and Oversight

Effectively, this accountability mechanism for intelligence organizations can be accomplished through control (ex ante facto) and oversight (ex post facto). Control over intelligence agencies means the requirement of asking authorization from a governmental body before undertaking intelligence actions from either elected civilians from the executive branch (e.g. a Minister or Secretary), elected civilians from the legislative branch (committees), appointed members from the judicial branch or from Inspectors General that exercise internal (administrative) control and

(16)

16 management (Otamendi and Estévez 2016: 8; Matei and Bruneau 2011b: 665). Besides authorization requirements, legislatures usually form forward looking control through the creation of legal mechanisms that regulate the use of special intelligence gathering measures (Boraz and Bruneau 2006: 31). Oversight, on the other hand, is the “systematic process of reviewing an intelligence community’s actions, organization, budget, internal policies, and legal constraints with the goal of improving effectiveness as well as ensuring democratic civilian control” (Boraz and Bruneau 2006: 31). Oversight can be conducted by the formal governmental entities (fora) such as the responsible minister, external oversight committees, general parliamentary oversight, judicial oversight, public ministries, ombudsmen or courts of audit. However, it can also refer to oversight performed by informal entities outside of the central government, such as civil society organizations like (international) NGO’s, think tanks, the media, academia, companies and the media (Otamendi and Estévez 2016: 8; Matei and Bruneau 2011b: 665). In this thesis, the oversight bodies (fora) under research are limited to formal, institutionalized governmental entities as the thesis is only concerned with institutional reforms (conform the research question).

2.3. Conceptualization of the dependent variable: Institutionalized Intelligence Politicization

The research question (How have institutional reforms of intelligence organizations during the democratic transition of former authoritarian communist states in Eastern Europe that are now part of the EU affected the formal relation between policymakers as principal and intelligence agencies as agent?) is concerned with the effects of the accountability mechanisms (as discussed in §2.3) as part of the democratic reforms (X) on the politicization of intelligence (Y). Therefore, it is equally important to go into the conceptual discussion concerning the definition of the dependent variable. Moreover, this thesis requires a specific definition of intelligence politicization that is not yet clearly defined elsewhere. Consequently, a new conceptualization of intelligence politicization will be proposed in this section.

2.3.1. Intelligence politicization

Broadly speaking, intelligence politicization concerns the proximity of the relation between the community of policymakers in the executive and the intelligence community. In academic literature, however, there is no consensus about the appropriate interpretation of this relationship (Eisenfeld 2017: 76). As this concerns a relationship between (political) power and knowledge (of secret intelligence) and thus influence, the politicization of intelligence is a contested concept often with negative connotations (Eisenfeld 2017: 78). But the concept of politicization itself in its most neutral interpretation is that of bringing something into the realm of politics, with which intelligence ultimately is concerned (Betts 2002: 1). The negative connotation to the word stems from the idea that politicization can (deliberately) distort or corrupt information to serve the (policy) interests of invested politicians, which leads to a loss of integrity on both sides (Betts 2002: 1). The question whether or not politicization is in fact negative has been ground for a conceptual debate. On the one hand the Kent School of thought warned against the danger of losing integrity and objectivity as too much political involvement would compromise intelligence, and on the other hand the Gates School stresses the useful necessity of political engagement from the intelligence community to ensure the indispensable utility of effective intelligence products (Betts 2002: 4). The executive authorities, being the primary consumers of intelligence products, consequently need to provide the services with direction in the form of guidelines, mission

(17)

17 statements, national security strategies and sometimes daily orders and feedback (Boraz and Bruneau 2006: 31). The input and feedback of policymakers is vital both at the start of the intelligence process with the general guidelines as well as the end where they will have to be able to use the products for national security policies and decisions. As at least some politicization is necessary, the achievement of zero politicization (or pure objectivity) is impossible as it would involve raw data without any connection to political reality, so a bit of political sensitivity is required from intelligence officers as well as the unbiased collection and analysis of intelligence. The relation between intelligence officers and the policy community can thus be seen as a continuum ranging from little involvement to a lot of involvement. On the one end, a shortage of political interest means that intelligence services could either operate without restraints or that their work is ineffective as politicians do not look at their analyses. On the other end of the spectrum, too much political interest leads to a compromised integrity and objectivity (De Graaff 2014: 39). Depending on the use of this continuum, intelligence politicization can have many definitions as it ultimately “exists in the eye of the beholder” (Eisenfeld 2017: 95). Consequently, many scholars have put forward their definitions: Johnson: “putting a spin on or ‘cooking’ intelligence to serve the political needs or beliefs of an intelligence manager or policy official”; Gannon: “the willful distortion of analysis to satisfy the demands of intelligence bosses and policymakers”; Rovner: “the attempt to manipulate intelligence so that it reflects policy preference”; Handel: “interference in the intelligence process by leaders and their close aides”; Treverton: “commitments to perspectives and conclusions, in the process of intelligence analysis or interaction with policy, that suppresses other evidence or views, or blind[s] people to them”; Hastedt: “the overt manipulation of intelligence and the intelligence process” (Eisenfeld 2017: 81-82) and Otamendi and Estévez: “politicization, that is, the temptation to make partisan political use of the state intelligence apparatus” (2016: 5).

2.3.2. Need for a new concept: Institutionalized intelligence politicization

In order to make a fitting concept of intelligence politicization for this thesis, one important adjustment to all these definitions must be made. Whereas the concept will entail a continuum ranging from low to high amounts of political involvement in the intelligence apparatus, it specifically denotes an institutionalized relationship. Thus, informal networks are excluded from the definition. This addition is necessary as it will enable a narrower research and because it enables an accurate description of the relation between the intelligence and policy community under communist rule. As previously discussed, intelligence organizations are often in service of the ruling elite or party in authoritarian regimes, meaning that politicization of intelligence is a defining characteristic of such regimes (Scott and Jackson 2004: 150). However, a defining characteristic of this politicized relation is that it is deeply embedded in the governance of the state. Therefore, this new definition of intelligence politicization necessarily emphasizes the structural use of intelligence organizations for political gains, as part of the governing system rather than emphasizing occasional opportunism as connoted by the previous definitions. The conceptualization also directs the focus on the organizational element of intelligence, as opposed to the aspect of informal networks or the politicization of the actual intelligence product.

On the low end of the spectrum of institutionalized intelligence politicization is the positive politicization as put forward by Gates. To avoid confusion, this is conceptualized as responsiveness since it implies a positive, effective and reciprocal relationship. Responsiveness entails taking into

(18)

18 consideration the needs and views of the principal (policymakers) by engagement and learning to attain a better performance (Blagescu et al. 2005: 11 in Bovens 2010: 958). In this respect it is important to clearly differentiate between this need for responsiveness and the aforementioned accountability (§2.3), since these terms tend to be used interchangeably by different authors. Responsiveness therefore entails the legitimacy and effectiveness of operations by interaction and accountability is about justifying and explaining one’s actions.

The high end of the spectrum of institutionalized intelligence politicization is defined as: an intelligence organization in service of the ruling political elite, who use intelligence activities and products for (partisan) power politics. High institutionalized intelligence politicization fits the use of intelligence for power politics like it originated in the authoritarian state and not for the protection of the current democratic rule of law, compromising both intelligence work and civil liberties by a far-reaching intrusion of the state in society.

2.4. Operationalization

Now that the concepts are clear, the next step is to create concrete indicators to enable measurements and analyses. To conduct a qualitative, comparative case study analysis, clear indicators are important as often very different criteria are used by the evaluators of intelligence democratization in post-communist countries (Watts 2004: 24). Especially with eleven cases, a structured codification is vital.

2.4.1. Legal-Institutional Accountability Mechanisms

Firstly, for the independent variable of democratic reforms, there are a few dimensions that can be looked into. Its definition of “a process of establishing a legal-institutional framework in which intelligence agencies operate under the rule of law and are being controlled by democratically elected civilians, while maximizing their effectiveness” (Matei and Bruneau 2011a: 606; Andregg and Gill 2014: 491), shows three dimensions: an established rule of law; accountability and effectiveness. As stated previously, the element of effectiveness is beyond the scope of this research. More importantly, the most significant democratic reform measure required to transform an intelligence agency under authoritarian rule into one under democratic rule is to create a solid legal framework that establishes clear accountability (Andregg and Gill 2014: 493; Matei and Bruneau 2011b; Bruneau and Matei 2010). Therefore, the concept of democratic reforms will focus on the aspect of legal-institutional accountability as main independent variable. This joint concept combines indicators concerned with intelligence-related legislation and those concerned with the accountability mechanism. The full set of indicators is divided up into four sub-parts. The first part consists of the list of indicators related to the accountability legislation.

Next, accountability was defined as “a relationship between an actor and a forum, in which the actor has an obligation to explain and to justify his or her conduct, the forum can pose questions and pass judgement, and the actor may face consequences“ (Bovens 2010: 951) and is thus concerned with the control and oversight of intelligence agencies (actor) by one or more fora. To measure the nature, degree and effectiveness of institutional accountability, it is necessary to look into how this accountability mechanism is set up in the wider intelligence-legislation, and how the three elements from Bovens’ definition are arranged: the informing, debating and the consequences. Therefore, the second set of indicators are concerned with the informing of the

(19)

19 forum by the actor, the third set is concerned with the debating function of the forum and the last set of indicators is concerned with the consequences of the oversight or control.

The following list of indicators is set up by combining indicators from Bovens (2010: 960) and from Otamendi and Estévez (2016: 11). The conceptual added value of combining the indicators of both actors is that the former is concerned with accountability in the public administrative sense. The latter created a matrix of indicators of accountability specifically relating to comparative case study research into the democratization of intelligence. Combined, an exhaustive and applicable list of indicators can be established. This is necessary as there is much variation in used indicators for the concept of accountability in general as well as in the specific field of intelligence democratization. The selected indicators take into account the difference between formal and material accountability mechanisms and tests to what extent the mechanisms were designed to be performative. This distinction was not yet explicitly made by either Bovens or Otamendi and Estévez. In the appendix §8.1 the origin of each indicator is listed.

Indicators of legal-institutional accountability mechanism for intelligence organizations: 1. Accountability legislation:

- Is there a public law of intelligence services? - Does this law include a precise mandate?

- Does the law define threats and interests of national security?

- Does the law define limits to the special powers/jurisdictions of secret intelligence operations?

- Is authorization required for special procedures or jurisdictions? - Is there a control and oversight mechanism codified in legislation?

- Does the law prioritize and guarantee respect for human right and civil liberties? - What activities can the services undertake under what conditions?

2. Informing of control and/or oversight:

- Are there legislative, judicial and/or expert bodies that monitor the services? - Is there a parliamentary oversight commission?

- Is the forum plural-partisan (how is it composed)?

- Is the principal the same organ as the forum (that conducts oversight and/or control)? - Is the actor obliged to provide information about his or her conduct to the forum? - Does the forum have access to secret information?

- How many times does the forum meet/how many reports do they produce? 3. Debating of control and/or oversight:

- Can the forum engage in questioning and debate? - Does it have the power to require hearings? 4. Consequences of control and/or oversight?

- Can the judgement of the forum have formal or informal sanctions for the actor/principal?

- Can intelligence officers be held legally responsible for their actions?

By assessing the strength of these indicators for all cases, the strength of the accountability mechanism can be established. Firstly, the more elements that are positively present in the intelligence law, the stronger the foundation for the legal-institutional accountability mechanism is. Secondly, the presence of a forum, its formal or material nature and the attributed powers decide

(20)

20 the strength of the accountability mechanism. In the within-case analyses, these two elements make up the strength of the legal-institutional accountability mechanism as a whole, ranging from weak to strong. These values are deliberately qualitative since the measurement levels of the indicators are qualitative as well.

2.4.2. Institutionalized Intelligence Politicization

Secondly is the measurement of the dependent variable of institutionalized intelligence politicization. The concept was defined as: “an intelligence organization in service of the ruling political elite, who use intelligence activities and products for (partisan) power politics”. The absence of existing indicators for this new concept meant that the indicators too had to be invented for this thesis. The indicators were largely derived from the principal-agent theory, which will be explained further in the following chapter of the theoretical framework (§3).

Indicators of institutionalized intelligence politicization:

1. To whom are the intelligence and security services responsive (to whom do they report as the principal)?

2. Organization-wise under which department do the intelligence and security services fall? 3. Who is formally in charge of the services?

4. Who provides general guidance?

5. Who gives is involved in the daily management and gives out specific tasks and orders to the services?

6. Is there an incentive mechanism in place between the principal and the agent?

7. Do political authorities appoint agency heads based on merit, is the process public and open?

These indicators define the formal relation between the intelligence and policy community and measure how strong the political influence is along the politicization continuum, ranging from low (responsiveness as minimal, necessary political involvement) to high political involvement. These values are deliberately qualitative as well since the measurement levels of the indicators are qualitative too. Combined with the strength of the legal-institutional accountability mechanism, it can be approximated whether the agencies are in service of the rule law, the government, a political party or another entity. The measurement levels of the indicators are all qualitative, and therefore so are the values of the independent and dependent variables. In the appendix §8.2, an overview can be found of all the indicators as structured in the case descriptions of chapter 4.

2.4.3. Contextual variables

Lastly, some intervening variables derived from the literature will be taken into account for a possible correlation (see Hypotheses §3.4):

- Are there multiple intelligence organizations or is there a monopoly on state intelligence? - Has the old personnel been vetted?

- What type of political system did the state have and what type did they construct? (presidential/parliamentary system)

- Do the agencies engage in international cooperation within EU/NATO or other important partners?

(21)

21

3 THEORETICAL FRAMEWORK

Now all the concepts are explained, the following section is concerned with how the main concepts are related. As established before, the independent variable is the legal-institutional accountability mechanism and the dependent variable is the institutionalized intelligence politicization. In this section, a principal-agent model is set up to form the link between the two variables in the research question How have institutional reforms of intelligence organizations during the democratic transition of former authoritarian communist states in Eastern Europe that are now part of the EU affected the formal relation between policymakers as principal and intelligence agencies as agent? This link functions as a causal mechanism that should help interpret and explain the possible relation between the variables under different conditions. First, it will be explained why this model is applicable for the research and what its limitations are, then the theory will be explained and the used principal-agent model for this research will be made more concrete. After that, the total causal mechanism will be defined and visualized in more detail, ending with a number of hypotheses resulting from the principal-agent theory. These hypotheses will be supplemented with a few alternative hypotheses, derived from existing literature.

3.1. Why use Principal-Agent Theory?

As the principal-agent model is a non-cooperative game theory, it allows for the explanations of relational behavior between the policy and intelligence communities. Moreover, because principal-agent theory focuses on the responsiveness of the principal-agent’s actions to the principal’s goals within an institutional setting, it is a suitable framework to analyze bureaucratic politicization to higher-level political actors and its outcomes (Gailmard 2014: 90). Principal-agent models are a flexible theoretical framework, which allows for considerable variations in institutional arrangements to compare the varying behavior by principals and agents. Therefore, it has become a commonly used paradigm for analyzing the relation between these two types of institutions (Gailmard 2014: 90). The model is thus both applicable to intelligence-policy relation in authoritarian and democratic regimes, which is a prerequisite for this research.

However, this flexibility also poses an important limitation. Since it enables many explanations of principal-agent relations, that also means that nearly any pattern can be explained, making it hard to falsify the whole theoretical framework in general (Gailmard 2014: 101). Nevertheless, the specific model can be tested in applicability and “the flexibility is not so great that the core idea of political principals exercising direction over policy as best they can is fundamentally compromised” (Gailmard 2014: 101).

3.2. What is Principal-Agent Theory?

Since the core of this research concerns the relation between the intelligence organizations and the policy community, the theoretical framework will explore this relation further. The principal-agent theory is central to this relation, as it is concerned with the responsiveness (and possibly also the accountability) from the agent (the intelligence organization) to the principal (the policy community). To fully grasp the link between accountability mechanisms and possible politicization as proposed in this thesis, first it is essential to differentiate between these concepts

(22)

22 of accountability and responsiveness as these are linked either to the principal (to whom the agent is responsive) or the forum (to whom the agent is accountable). In the following sections, this thesis proposes a new application of the principal-agent theory, resulting from the therefor necessary reconceptualization of accountability and politicization. This finally results in the central premise concerning the relation between the intelligence and policy community: the principal and the forum cannot be the same entity, as the failure of separating the two elements in the umbrella-term of accountability (responsiveness and accountability as conceptualized in this thesis) disables a balance to the executive branch, which would invite the politicization of intelligence.

3.2.1. Responsiveness and Accountability

In principal-agent theory, a broad concept of accountability is used, which is more akin to what this thesis identified as responsiveness. The principal-agent theory is often used to explain the relation between principal and agent in terms of responsiveness and effectiveness. Since responsiveness was defined as ‘taking into consideration the needs and views of the principal (policymakers) by engagement and learning to attain a better performance’ (Blagescu et al. 2005: 11 in Bovens 2010: 958), in this thesis, it can be considered the necessary politicization as put forward by the Gates School in intelligence studies when thinking of the principal as the political authority and the agent as an intelligence organization. In the intelligence cycle, this constitutes the setting of intelligence guidelines and the reception of the final product by policymakers who possibly also provide feedback on this service delivery. It does not concern the accounting for this service or how it was conducted.

3.2.2. The Relation between the Principal and the Agent

Weber identified the principal-agent relationship as an asymmetric one. Whereas the authority is located on the side of the principal, the crucial information advantage of bureaucratic expertise is located on the side of the agent who undertakes action in service of the principal (Weber in Miller 2005: 203). Based on the rational choice perspective that both parties are utility maximizers, and that this utility (consisting of goals and preferences) differs for both, a conflict ensues as the relative power is on the hand of the agent due to the information asymmetry. The agents are believed to be motivated by the public interest and/or their own self-interest, that is said to be concerned with maximizing budgets, bureaucratic ‘empire building’ and the consolidation of a power position (Waterman and Meier 1998: 176; Gailmard 2014: 95; Peters 2012: 13). The principal will try to align the agent’s goals, preferences and actions with their own by means of providing incentives and/or controls (Gailmard 2014: 91; Waterman and Meier 1998: 175). If the principal fails in attaining the alignment of the agent through incentive compatibility or sufficient controls, the result will be agency loss, in which the agent takes on more autonomy to pursue its own goals and interests, resulting in sacrificed responsiveness (Gailmard 2014: 92). The conflict between the principal and agent is not only applicable to democratic regimes, but also to authoritarian regimes as authoritarian leaders too need to both ensure that the entire political-administrative system follows and executes their policies and need to use the expertise of the career bureaucracy to create and implement public policies (Peters 2014: 211).

So, to control their bureaucratic agents, principals can use incentives of hiring and firing such as appointments of top bureaucratic officials, and controls such as monitoring activities setting clear orders for intelligence operations rather than wide guidelines (Miller 2005: 209; Waterman and Wood 1993). The quickest way to align the bureaucracy, with large potential for politicization, is with a ‘spoils system’, which means that the political principal makes the entire top bureaucracy

In Whose Secret Service? A comparative case study on changing relations between the intelligence and policy community in transitioning democracies of Eastern-Europe