Lessons Learned or Political Persuasion?
Policy-Oriented Learning and Policy Change in the Dutch
Counterterrorism Domain
An Advocacy Coalition Approach
Student: Oscar Laros
Student number: s1491989
Supervisor: Dr. Stef Wittendorp
Second reader: Dr. Joery Matthys
Wordcount: 24867
Assignment: Master thesis
Study: Crisis and Security Management
Institution:
Leiden University
2
Foreword
This master thesis serves as the final assignment to obtain the master’s degree in Crisis and Security Management. For me, writing my thesis on a subject related to terrorism is the logical conclusion of the focus that I have held throughout the master’s and during my internship at the counterterrorism section of the Ministry of Foreign Affairs. The phenomenon of terrorism has always intrigued me, and that clearly helped me a lot in putting all the time and effort in this study. I am very happy and thankful to have had the opportunity to experience and be involved with the Dutch counterterrorism policy during my internship. However, that period also showed me that the policy is not yet perfect, and that measures can in fact have counterproductive effects if they are not coordinated correctly. That notion was the direct incentive to carry out this research. I sincerely hope that this study may contribute to further development of CT-policy – more specifically, the implementation processes of the two studied measures – in order to provide the protection and security, but also the rights and help that everybody has a right on as a Dutch citizen.
I would like to thank mister Wittendorp for the excellent supervision throughout the process. Your advices have been very useful and made me constantly reconsider my own choices and arguments. You have always been open for questions and help, and besides that, you are just a very nice instructor to get lectures from and to work with. In addition, I owe many thanks to the eight respondents that provided very valuable information during the interviews. Without these interviews, this research could not have been done. Lastly, I would sincerely like to thank my parents for their support throughout my study period and my entire life. You two have perfectly facilitated my whole educational development from the start till the end. More importantly, you two enabled me to grow up to the person I am today. I could not have wished for better parents. I hope to make you proud.
3
Content
1. Introduction……….……….……….5
1.1 Radical changes in Dutch counterterrorism policy………...5
1.2 Repressive measures, counterproductive effects……….………..…….…7
1.3 Research question and sub-questions…………..………...…….9
1.4 Academic relevance……….10
1.5 Societal relevance………....12
1.6 Reading guide………..…………12
2. Theoretical framework……….…….………..…13
2.1 The public policy process and learning………..………..13
2.2 Learning in public policy: a definitional debate……….………..13
2.3 Policy-oriented learning and implementation research: towards a new approach....14
2.4 The Advocacy Coalition Framework (ACF)………16
2.5 General functioning of the framework……….18
2.6 Conceptualisation of variables within the framework……….….19
2.7 Four processes that lead to policy change……….…21
2.8 Hypotheses………... 23
3. Methodology……….……….…...…26
3.1 Case study method………...26
3.2 Operationalisation………28
3.3 Data collection……….31
3.4 Method of data analysis: process tracing...33
3.5 Validity and limitations………...………34
4. Case description…..……….….……...36
4.1 Freezing of assets and revoking nationality: an introduction………..………..36
4.2 Freezing of assets……….37
4.3 Revoking nationality………38
4.4 Intended effects………...………..40
4.5 Unintended effects...41
5. Demarcation of the ACF’s actors...43
5.1 Policy subsystem...………43 5.2 Advocacy coalitions...43 5.3 Policy brokers...49 5.4 Exogenous factors...50 5.5 Internal perturbations...52 6. Analysis……….……....53
6.1 Reflection on the hypotheses regarding policy-oriented learning………53
6.2 Conclusion: the causal processes leading to policy-oriented learning ...……..……60
6.3 Reflection on the alternative hypotheses regarding policy change...61
6.4 Political dominance and politicisation in the CT-domain...……64
4
7. Conclusion……….………...69
7.1 Recap of the research...69
7.2 Answer on the RQ...69
7.3 Reflection on the applicability of the ACF...71
7.4 Limitations...72
7.5 Recommendations ...72
References...74
List of abbreviations………...79
List of used policy documents and evaluations………..80
5
1. Introduction
The rise of jihadist terrorism has arguably been the most prompting security issue for West-European countries in this decade. What started as a merely local issue in 2011 with the Arab Spring, evolved into an almost global security issue when jihadist terrorist groups were able to gain a foothold in the chaos and power vacuum that followed the Arab Spring in countries across the Middle-East and North-Africa (NCTV1, 2016: 25). The most infamous jihadist
terrorist group is the Islamic State in Iraq and al-Sham (ISIS), which declared a ‘caliphate’ in Iraq and Syria in 2014. ISIS manifested itself as arguably the most influential terrorist organisation ever by attracting the most Foreign Terrorist Fighters (FTF’s) to a country with a Muslim majority in modern history (NCTV, 2014b: 17). Approximately 40.000 fighters from over one hundred countries, including around 5000-7000 fighters originating from European countries, made their way to Syria and Iraq since 2012 (NCTV, 2018: 13). In addition to the undesired contribution to the armed conflict, (European) jihadist fighters perpetrated multiple terrorist attacks in Western-Europe. Now that the physical caliphate in Syria and Iraq is military destroyed, however, West-European countries are increasingly struggling with the question of how to deal with (returning) FTF’s. All these developments have led to substantial reforms in the counterterrorism policies of West-European states, including the Netherlands in the past five years, in an attempt to address the continuously changing threat of jihadist terrorism to national security. The question is, however, to what extent those policies adequately counter the terrorist threat, what side-effects those policies may have and how we can learn from past policy experiences.
1.1 Radical changes in Dutch counterterrorism policy
This research focuses exclusively on reforms in counterterrorism policy in the Netherlands. The policy domain of counterterrorism has been expanded in the field of security policy in the Netherlands after the murder on Theo van Gogh in 2004 (AIVD, 2014: 59), and the subsequent arrest of members of the ‘Hofstad Group’ in 2004-2005. The Hofstad Group is considered to be the first jihadist terrorist cell in the Netherlands (NCTV, 2014b: 32). New programs were created and new institutions were set up, including the NCTb, the predecessor of the current National Coordinator for Security and Counterterrorism (NCTV). The Dutch signature with these policies was the “comprehensive approach to counterterrorism” (AIVD, 2014: 59), meaning that policy should not focus exclusively on terrorism, but also on preceding processes
6
that can lead to terrorism like radicalisation and polarisation within society (Ibid.). In other words, prevention was seen as the key to success.
In the years thereafter (2006-2013), the problem of homegrown jihadism only incidentally surfaced in the public debate (Ibid.: 5). It was therefore a very unpleasant surprise for the Dutch authorities to see over one hundred citizens leave for Syria during 2013-2014. Moreover, it was a clear indication that the jihadist movement in the Netherlands had strongly increased unnoticed by security and intelligence services (Ibid.: 5-6). As a result, the Dutch General Intelligence and Security Service (AIVD) urged for a new approach with both preventive and repressive measures in June 2014 (Ibid.: 64). This plan of action called “An Integrated Approach to Jihadism”2 (NCTV, 2017c: 1-2) was quickly established by the NTCV in August
2014, including the intention for two (partly) new repressive measures: freezing the financial assets of suspected terrorists, and the adjustment of the law on Dutch citizenship to enable the government to revoke the Dutch nationality of terrorists with a double nationality (NCTV, 2014a: 6-7). However, the actual implementation and execution of the policies took until late 2016 for the freezing measure (Ministerie van Buitenlandse Zaken, 2019a), and March 1, 2017 for the nationality measure (NCTV, 2017a).
The coming into force of three new laws on March 1, 2017 arguably marks the beginning of a new path that the Dutch government has taken in their continuing efforts to counter terrorism. For instance, the option to revoke nationality is but one of three repressive measures that have been implemented (others include contact prohibition, reporting obligation and travel bans). The preventive measures from the earlier comprehensive approach have been supplemented with administrative “special powers” (NCTV, 2017c: 2) in order to ensure that “those who perpetrate terrorist crimes and finance terrorism can be convicted” (Ibid.). The big difference here is that these administrative special competencies that make it possible to impose the two repressive measures are the responsibility of the national government, whereas in almost every other situation whereby a counterterrorism measure is imposed on an individual, the municipality is responsible, as stipulated in the so called ‘local approach’ (Ibid.: 4). As a result, local and national partners need to co-operate intensively to maintain the correct balance of preventive and repressive measures for each specific case, whereby the reduction of security risks and the protection of national security is the overarching goal (Ibid.: 4, 6).
7 1.2 Repressive measures, counterproductive effects
Both measures were criticised from the moment that implementation and the effects became public. Media, academics and lawyers have widely covered the procedures and effects of both the freezing of assets and revoking of nationality measure.
Freezing of assets
If the government freezes your financial assets, you are put on the ‘National sanction list terrorism’ (Rijksoverheid, 2019a). Journalists wondered, for instance, how it is possible to be put on a ‘black list’ without being convicted (Jensma, 2016). Moreover, the utility of the measure is questioned. Sanctioned individuals feel that they are being kicked out of society, which makes the option to disappear in illegality more tempting (Albert & Lensink, 2015). Others give examples of the minimal impact of the measure since the actual financial flows that finance terrorism do not go by conventional banking routes that are cut off by the measure. In addition, the fact that individuals that are in prison – and therefore do not have access to their financial accounts anyway – are also on the sanctions list, is also raising eyebrows (Jensma, 2016). Lawyers have therefore called the measure “disproportionate” (Alberts & Lensink, 2015).
In addition, there are concerns regarding the family of the sanctioned individual. Multiple members of families from individuals who left for Syria have been sued for financing terrorism because they sent their relative money, and therefore faced trial in the Netherlands (Bahara, 2019). Academics also warned the government to remain diligent when imposing the measure, since it is a tough sanction for someone who not (yet) necessarily is a criminal suspect (Wesseling & De Goede, 2018: 229). Further, the measure can have negative effects on individuals that are following a re-integration program. (Ibid.: 163). Also, the efficiency and possible side-effects of the measure remain somewhat unclear, while the legal base to make an individual punishable is very broad (Ibid.: 229-230). Lastly, researchers state that legal protection for suspects and (evaluation of) policy effectiveness are topics that are under discussion in academic literature on terrorism finance, governmental blacklists and sanctions (Ibid.: 164-166).
8
Revoking nationality
Secondly, there is the measure to revoke the Dutch nationality of suspected terrorists. Because international law prohibits countries to make their citizens stateless, this measure can only be imposed on individuals who have a double nationality (NOS, 2019a). Critics claim that this is discriminating since this measure cannot be imposed on a terrorist holding the Dutch nationality only (Kouwenhoven & Eck, 2019). Moreover, actors from the legal domain increasingly question the measure on possible interference with the non-discrimination principle (Klaus, 2017). This topic was also discussed in the Dutch parliament whereby different political parties questioned the legal distinction of civilians that follows from the law. A party supporting the law justified this distinction by stating that it is about “choosing between two evils: revoking Dutch citizenship of nobody, or making this distinction” (Eerste Kamer, 2016b: 1).
Furthermore, the measure is imposed with retroactive effect, meaning that individuals that have been convicted and served their sentence can still be stripped of their nationality. Maher H. is one of those individuals whereby the re-integration and re-socialisation trajectory is disrupted by the intention to revoke his nationality (Kouwenhoven & Eck, 2019). Most importantly, however, may be the fear from both CT-experts and the mayors of the four biggest municipalities in the Netherlands, claiming that revoking nationality can in fact accelerate the radicalisation process (NOS, 2019a). It can take years for an individual who is stripped of the Dutch nationality to get deported, but in the meantime, access to any form of help is restricted since the individual is not a Dutch citizen anymore. This process can – just as with the freezing of assets measure – isolate the individual, which may push him or her into illegality (Kouwenhoven, 2019). At the same time, feelings of injustice may strengthen radicalisation, with all possible negative effects on national security.
Focus of the research
It is clear that both measures have encountered criticism from the media, the legal, academic and the political domain. Nevertheless, a politician responsible for counterterrorism in the dominant party in Dutch parliament recently stated that: “We need to continue on this road by revoking the Dutch nationality from as many terrorists as possible” (Pauw, 2019). It is therefore interesting to investigate to what extent the Dutch government took note of the critics, to what extent those critics may be justified, and, most importantly, to what extent possible problems are solved in further development and implementation of the two measures. In other words, to what extent have the involved CT-actors learned from earlier experiences with the measures?
9 1.3 Research question and sub-questions
This research therefore aims to investigate to what extent there is policy-oriented learning within the Dutch counterterrorism domain, whereby the specific focus is on the implementation of the two repressive measures of freezing of assets and revoking nationality. The measures will not be actively compared, but serve as a handle to investigate the CT-policy domain.3 The research question therefore is:
“To what extent is there policy-oriented learning in the Dutch counterterrorism domain, and
especially among the actors involved with the two new repressive counterterrorism measures between 2017 and 2019?”
The findings regarding policy-oriented learning in this case may help to foster the policy process in the future in addressing the currently perceived bottlenecks in counterterrorism policy. The theory on policy-oriented learning using the Advocacy Coalition Framework (ACF) by Paul Sabatier is used to derive hypotheses addressing policy-oriented learning that guided the data collection and analysis in this research. In order to attain the research goal, sub-questions are formulated below to demarcate different parts of the analysis towards answering the research question in the conclusion. The sub-questions also serve to distinguish the different actors that are found within the ACF, which is necessary to analyse and draw a conclusion based on the specific theory in this research.
1. What changed in the Dutch counterterrorism approach in 2017? How is the change in policy perceived by the CT-actors?
2. What effect do the two nationally imposed repressive measures have on the locally imposed measures from the person-oriented CT-approach?
3. Which CT-actor is responsible for what at each level? How do these actors relate and interact?
4. To what extent are CT-actors involved with the process of policy making and policy evaluation?
Question 1 and 2 aim to get an insight in different actors’ perceptions of policy change and the effects thereof in their daily business. This is important, because different perceptions of an occurring phenomenon can have an influence on the perceived need of policy learning. Question 1 and 2 therefore relate to individuals’belief systems (see 2.6). Question 1, 2 and 3
10
are also used to demarcate different advocacy coalitions within the case. Further, question 3 and 4 aim to uncover the coalitions’ strategies in pursuing their policy objectives. The sub-questions are answered throughout chapter 4, 5 and 6.
1.4 Academic relevance
Counterterrorism research
The field of counterterrorism research has been expanding fast since 9/11. With the rise and fall of ISIS, and the developments around FTF’s, the number of studies focusing on counterterrorism policies of Western states has skyrocketed. It is claimed that there is a disproportionate focus on major Western states such as the USA and the UK in existing CT-literature (Romaniuk, Irrera, Grice & Web, 2017: v). Instead of comparative regional studies, authors suggest comprehensive study of all aspects of CT within a state (Ibid.). This research addresses this gap in knowledge by doing detailed and specific research involving multiple CT-actors in the Netherlands. In addition, this research also serves as a follow-up and more in-depth study of a report on Western states’ CT-policies carried out by Wittendorp, De Bont, Bakker and De Roy van Zuijdewijn (2017). The explorative report mentions and explains the measures of revoking nationality and freezing of assets (Wittendorp et al., 2017: 34-35, 46), but does not incorporate any objectives and effects of the measures. Since this research is executed almost three years later, it provides a great opportunity to investigate those effects of the measures in the Dutch context.
In line with Romaniuk et al., Hellmuth (2016: 5) claims that most studies hold a comparative focus towards the national counterterrorism policies of states. However, according to Hellmuth (2016: 6) literature on counterterrorism policy tends to address state policies using a broad and long-term perspective, thereby lacking an in-depth understanding of the decision-making processes that has led to those state security policies. Hellmuth therefore suggests comprehensive analysis of domestic decision-making processes and responses in order to improve counterterrorism mechanisms (Ibid.). In his comparative study on the UK, France, Germany and the USA, Hellmuth (2016: 282) found that the way power is allocated throughout all levels and agencies of a government is of great influence on counterterrorism policies and the decision-making processes wherein policy reforms are decided upon. Consequently, a change in government structure can alter the decision-making process, restraints, interagency dynamics, institutional designs and thus incentives and objectives for policy reform (Ibid.). He refers to government structure as a stage-setter wherein bureaucratic dynamics may or may not
11
unfold, thereby explicitly stating that counterterrorism policies are therefore not the simple outcome of “predetermined bureaucratic preferences or organizational cultures and procedures” (Hellmuth, 2016: 282).
The findings of Hellmuth’s research are highly relevant, since they to some extent correspond to propositions in the ACF, and serve as an immediate cause to test his assumptions on the case of reforms in Dutch counterterrorism policy in 2017. This research aims to thoroughly analyse the dynamics between actors in the counterterrorism domain involved with the two measures, in order to investigate the allocation of power among these actors using process tracing. By tracing back the developments in the last three years, this research aims to investigate to what extent changes in government structure have fostered policy reform and policy learning. It is therein interesting to see whether government structure serves as a stage-setter for CT-policy reform in this case, as Hellmuth proposed. This research therefore contributes to the literature on counterterrorism policy by testing assumptions following from Hellmuth (2016), while maintaining the scarce domestic, specific, in-depth and short-term focus that will help to enrich the literature on CT-policy and the role of governmental power allocation therein.
Theory development
Secondly, the case provides an opportunity to apply Sabatier’s ACF to a policy domain that has been lacking attention from policy change scholars so far. Compilation studies on the use of the ACF show the strong focus on the domain of environmental and energy policy, and a distinct smaller focus on health and economic policy (Weible, Sabatier & McQueen, 2009: 125-126). Another study overviewing the implementation of the ACF endorses the overrepresentation of environmental and energy policy, as well as the total absence of security policy (Weible et al., 2011: 350). In addition, Pierce, Hicks, Peterson and Giordono (2017: 6, 15) conducted an overview of 161 articles using the ACF, again displaying that environmental and energy policy is far most represented. Particularly absent within the 161 researches is the security policy domain, with just one study conducted (Pierce et al., 2017: 33-45). Considering the almost total lack of attention for the field of security policy, this study aims to fill that gap in the implementation of the ACF. Moreover, applying the ACF to a specific security policy domain as counterterrorism may enrich the literature on policy change and policy learning from a renewing angle.
12 1.5 Societal relevance
The currently perceived bottlenecks in the Dutch counterterrorism approach may have a negative effect on the overarching goal of national security. It is therefore of utmost importance to identify the factors that have led to the current situation, thereby better enabling actors to adjust the policy in a way that measures do not interfere, guaranteeing that individuals receive the help they need and have the right on, and that all actors within the Dutch counterterrorism domain co-operate towards the same overarching interest of national security. Approximately 310 Dutch citizens have left for Syria, of whom 150-155 of them are still in the conflict area (NCTV, 2019b: 2; 2019c: 3). With the (possible) returning of those individuals and their 200 children (NCTV, 2019b: 2), the pressure on the Dutch counterterrorism domain may rise in the future. It is therefore essential that the system is functioning correctly. Moreover, the period between 2006 and 2013 taught the Dutch government a harsh lesson about diminishing attention and incorrect policy regarding counterterrorism (AIVD, 2014: 60-61). Unfortunately, the attack in Utrecht proved that terrorism is still a very imminent threat to national security in the Netherlands, and must therefore be countered by the best suited counterterrorism policy possible.
1.6 Reading guide
The next chapter presents the theoretical framework of the policy learning theory from which hypotheses that guide the research are derived. Afterwards, the methodology is explained. The fourth chapter provides a concise case description. Then, the demarcation of the ACF’s actors in the case is stated in chapter five. Afterwards in chapter six, a reflection on the hypotheses is presented in the analysis. The concluding chapter provides the answer on the research question, also including both limitations of the study, and recommendations for further policy development and research.
13
2. Theoretical framework
This chapter provides an overview of the relevant literature about the policy learning theory. The policy learning theory is positioned within the broader literature on the public policy process and learning. Afterwards, the theory of policy-oriented learning and policy implementation studies constructed by Paul Sabatier is explained. Then, the Advocacy Coalition Framework, that originated from the policy-oriented learning theory is expounded. Hypotheses derived from the framework are used in this research to identify the extent to which policy-oriented learning has occurred, and are therefore presented at the end of this chapter.
2.1 The public policy process and learning
The study of policy learning within governments became a separate field of study in the 1960s, in an era that is associated with comprehensive political, social, technological and economic change across countries that enabled governments within those countries to grow (Freeman, 2008: 2). The study of policy change based on forms of learning challenged the conventional conflict-oriented theories that assumed that policy change is the result of social forces and conflict (Bennet & Howlett, 1992: 275). During the 1970s and 1980s, different theories on policy change and policy learning were established such as Rose’s ‘lesson-drawing’, ‘political-learning’ by Heclo, ‘social ‘political-learning’ by Hall, ‘government ‘political-learning’ by Etheredge and ‘oriented learning’ by Sabatier (Ibid.). This research exclusively uses the latter theory on policy-oriented learning by Paul Sabatier since the aim is to analyse the extent of policy learning after consecutive changes in policies within the Dutch counterterrorism domain. As will become clear with the explanation of Sabatier’s theory in the next sections, his work focuses specifically on policy implementation and is therefore most appropriate in this research.
2.2 Learning in public policy: a definitional debate
Public policy is defined as a statement of government that – either directly or through an agent – contains an intention for an action and a desired outcome to some sort of problem on behalf of the public (Birkland, 2011: 8-9). Since public policy serves the public interest and is always aimed at solving some sort of problem that undermines that public interest, it is of utmost importance that governments adapt and improve their policy when situations change. A way of adapting and improving can happen by policy learning.
However, the definitional debate about policy learning is unavoidable in policy learning literature. For instance, Dolowitz (2004: 26) defines learning as “a process where the learner
14
gains deeper understanding, comprehension, or knowledge of the item under investigation”. The notion of understanding is also a vital part of the definition by Freeman (2008: 2), who states that policy learning is about explaining certain policies: “learning is not only the what and how of public policy but also its why”. Whereas the latter definitions tend to focus on explaining the existing policy that is under review, Busenberg’s (2001: 173) definition on learning in policy has a more forward-looking glance, claiming that it is the process wherein new information and ideas are applied to policy decisions by individuals. Ideally, however, according to Hall, is policy learning based on evaluation of past experience in order to achieve better goal attainment by governments (Bennet & Howlett, 1992: 276). His more instrumental definition thus aims at the adjustment of policy while taking into account the consequences of past policy and new information (Ibid.). The definition of policy learning by Sabatier used in this research also builds on Hall’s notion of experience. Sabatier (1987: 673) proposed policy-oriented learning, which entails: “relatively enduring alterations of thought or behavioural intentions that result from experience and that are concerned with the attainment or revisions of the precepts of one's belief system” 4. Consequently, policy-oriented learning can be a major
determinant of policy innovation and change, and therefore is a highly relevant form of policy learning to analyse in this research.
2.3 Policy-oriented learning and implementation research: towards a new approach
The policy cycle
Policy-oriented learning may occur throughout different stages of the policy cycle as proposed by Bovens, ‘t Hart & Van Twist (2011: 75-76). Bovens et al. (Ibid.: 76) distinguish six separate phases in the policy process: (1) agenda setting; (2) policy preparation; (3) policy provision; (4) policy implementation; (5) policy evaluation; and (6) feedback and reconsidering. In the last phase, policy evaluations and new information are processed aimed at possible continuation, adjustment or termination of the policy (Ibid.). This clearly implies that policy-oriented learning should take place in phase 6 of the cycle. Other scholars, however, have argued that policy learning merely occurs in the implementation phase, which is phase 4 in the policy cycle by Bovens et al.
15
Implementation research
The study of implementation research arose simultaneously with research on policy learning during the 1970s, in order to explain differences in implementation success of governmental programs (Sabatier, 1986: 21). The implementation phase occurs after policy has been formulated, and entails the period that policies are put into effect by governments or representing agencies (Birkland, 2011: 263). The implementation phase is critical for the success of the policy since it relies on the behaviour of the implementers and starts interacting with the policy environment and policy targets (Ibid.: 263-264). When problems occur in the implementation phase, learning can foster adjustments to structure the policy in such a way that the policy can achieve the intended effects by the policy makers (Ibid.: 264). Therefore, the implementation phase is considered a key feature of the entire policy process (Ibid.).
Top-down approach
Until the proposition of the ACF by Sabatier in 1986, two approaches to analyse policy implementation existed: down and bottom-up (Sabatier, 1986: 21). The traditional top-down approach assumed that the process of policy implementation can be understood by looking at the policy goals (usually formulated in a statute) at the highest level of policy design, and examine to what extent the objectives were achieved and why, towards the lowest levels of implementation (Ibid.: 22; Birkland, 2011: 265). The approach is based on a set of assumptions of which the assumption that policy contains clear goals and objectives against which the performance can be measured is considered most problematic (Birkland, 2011: 264-265). In addition, the assumption that there always is a dominant institution that is exclusively involved with a policy is proven wrong, since this perspective neglected the presence and influence of other governmental actors on the policy (Sabatier, 1986: 30; Birkland, 2011: 269). Lastly, the approach underestimated the power of street level bureaucrats – i.e. local level policy implementers – and target groups to commute certain policy for their own purposes (Sabatier, 1986: 31).
Bottom-up approach
Due to the impossibility among scholars to explain certain policy outcomes using the top-down approach, the bottom-up approach was introduced, which is also called “backward mapping” (Birkland, 2011: 268). This approach starts with the lowest level policy implementers by investigating their goals, activities, strategies and networks in order to identify all involved actors at higher scale levels, ending with the policy designers at the top (Ibid.: 268; Sabatier,
16
1986: 32). The bottom-up approach holds some assumptions contrasting the top-down approach. For instance, policy goals can be conflicting and ambiguous, and policy can be more loosely defined than in statutes (Birkland, 2011: 268). Interesting conclusions in early bottom-up studies were that the success of policy programs is far more dependent on the skills of individual local level policy implementers than on high officials in central government (Sabatier, 1986: 32), and that problems with the desired implementation of policy arise when there is “a conflict of interest between implementation agencies and politicians” (Birkland, 2011: 268). However, the bottom-up approach also has some limitations. Sabatier (1986: 34) for example argues that the approach overemphasises street levels bureaucrats’ abilities to frustrate policy designers and thwart the policy goals of politicians. In addition, he states that the power of the central government cannot be underestimated since it eventually structures “the rules of the game” (Ibid.: 35) by distributing the resources and preferences. In an attempt to synthesise the strengths of both approaches into a framework that enables thorough analysis of policy implementation, Sabatier (1986) proposed the Advocacy Coalition Framework.
2.4 The Advocacy Coalition Framework (ACF)
Starting in the late 1980s, Sabatier developed the ACF, and has adjusted and improved the framework throughout the years after hundreds of empirical studies had been conducted that provided valuable insights and critiques (Sabatier, 1986, 1987, 1988, 1998; Sabatier & Jenkins-Smith, 1994; Sabatier & Weible, 2007; Sabatier, Weible & McQueen, 2009; Sabatier et al., 2011). Over time, the theory has grown to an important model of the process of public policy within the academic realm (Birkland, 2011: 298).
The ACF is a model of the policy process that enables to deal with “wicked problems” (Sabatier & Weible, 2007: 189) which involve substantial conflicts in policy goals, extensive technical disputes and the presence of multiple actors from different levels of government involved with the policy. The aim of the conceptual framework is to demarcate what role policy analysis has in policy-oriented learning, and what role policy-oriented learning, in turn, has in the process of policy change (Sabatier, 1987: 650). Policy analysis herein is the accumulation of information about the effects of policy (Ibid.), and is therefore related to policy evaluation. The framework thereby explicitly considers the influence of implementation and feedback on the system (Birkland, 2011: 298). Nevertheless, the ACF’s basic argument is that: “While policy analysis and learning can strongly affect secondary aspects of such belief systems, fundamental changes in subsystem policy are usually the result of alterations in noncognitive, systemic
17
parameters” (Sabatier, 1987: 650). The noncognitive factors playing a significant role in policy change are presented in 2.6 and 2.7.
Some general assumptions underlie the causal logics and hypotheses in the ACF: (1) a central function for technical and scientific information within the policy process; (2) a period of at least a decade to explain policy change5; (3) policy subsystems as the unit of analysis; (4) multiple level actors within the subsystem that can be aggregated into coalitions and (5) the paradigm that programs and policies need to be perceived as translations of beliefs (Sabatier, 1988: 131-133; 1998: 99; Sabatier, Weible & McQueen, 2009: 122). When taking into account these assumptions, the following diagram can be established that graphically presents the ACF. Underneath the figure, the general functioning of the processes in the framework is explained. Thereafter, the concepts are further conceptualised, based on the mentioned articles by Sabatier and co-operating authors.
Figure 1: 2007 Advocacy Coalition Framework Flow Diagram (Sabatier et al., 2009: 123).
5 This research does not cover a timeframe of a decade or more. Elaboration on this limitation is provided in the
18 2.5 General functioning of the framework
The two blocks most left in the figure present two sets of exogenous variables. The exogenous variables affect the constraints, resources and opportunities of the actors within the policy subsystem to pursue their policy objectives (Sabatier, 1988: 132). As indicated by the arrows, for instance, the constitutional structure determines the degree of consensus needed for policy change, which needs to be taken into account by the actors within the subsystem. Since the upper set of variables is more stable over time, it mostly influences opportunities on the long-term. Likewise, a change in public opinion can for instance influence the subsystem actors’ constraints and resources, since increasing public pressure can be used in the strategy to convince the others of your policy objective, or to defend your policy objective. External events are therefore more likely to affect the process on the short-term.
Actors within the subsystem can be divided into advocacy coalitions that consist of a number of people from various organisations and governmental levels who share a set of normative and policy beliefs, and to some extent act in concert to obtain their policy goals (Ibid.: 133). As depicted in the figure, each coalition adopts a strategy to obtain their policy objectives following from their policy beliefs. The strategy is based on their coalition’s resources, and involves innovations regarding existing instruments and policies. The result is two (or more) conflicting strategies that are normally mediated by policy brokers. Their principle task is to reduce conflict and find a reasonable compromise in order to formulate policy (Ibid.). Then, after decision by governmental authorities, the end result is one or more governmental programs that hold agreements on institutional rules, resource allocation and appointments, as indicated in the figure.
The governmental programs produce policy output on the operational level, which is for instance a decision on the tasks and responsibilities of agencies. After being mediated and influenced by other factors (for instance by the policy implementers), the policy output results in a variety of impacts on the problem targeted by the policy as well as possible side-effects (Ibid.). Consequently, based on the perception of the policy’s effects and the adequacy of governmental decisions, advocacy coalitions may revise their policy beliefs and/or adjust their strategy, as depicted by the internal feedback loops in the figure (Ibid.). In addition to these feedback loops, coalitions’ beliefs and strategies may again be influenced by exogenous variables, making the policy process in the ACF an ongoing cycle, which is illustrated by the lowest arrow in the figure. Lastly, events internal to the subsystem and negotiated agreements
19
can influence coalitions’ beliefs and strategies, which is further explained in the next section covering the different variables in the framework more in detail.
2.6 Conceptualisation of variables within the framework
Policy subsystem
The policy subsystem consists of a wide variety of actors who are actively involved with a particular policy problem and seek to influence policy in that particular policy domain. The subsystem is always characterised by a functional and a territorial dimension. The group of policy participants within this subsystem consists not only of the traditional ‘iron triangle’ of legislators, agency officials and interest group leaders, but can also include a variety of public and private organisations like scientists, journalists, consultants, judicial officials, policy analysts and researchers. Normally, the policy subsystem involves all levels of a government. The ACF supposes that policy participants within a subsystem have strong beliefs that they want to translate into policy.
Advocacy coalitions
The ACF argues that policy participants with similar policy core beliefs try to find allies holding those similar beliefs within the subsystem. As a result, any policy subsystem can be divided into (generally two to five) advocacy coalitions. However, the ACF assumes that most policy subsystems are dominated by one single advocacy coalition, and supplemented with one or more minority coalitions. An advocacy coalition is composed of a variety of actors from the subsystem who both “(a) share a set of normative and causal beliefs and (b) engage in a non-trivial degree of co-ordinated activity over time” (Sabatier, 1998: 103). In other words, members of a coalition need to co-operate to some extent towards the goal of achieving shared policy objectives. The ACF thereby explicitly states that coalitions are not exclusive formed by interest group leaders, but normally contain agency officials, legislative and policy officers from multiple levels in the government, researchers and sometimes journalists or other actors involved with the media.
The belief system
Beliefs are regarded as essential criteria to divide individuals within the policy subsystem into advocacy coalitions, because public policies and programs hold implicit theories on how the policy is to reach the desired objective. Since policies involve causal relations, value priorities, perceptions of the problem and assumptions about the solution and execution of the policy
20
instrument, the ACF argues that beliefs and policies can be conceptualised in the same way, thus providing a scale to assess the influence of different actors on policy over time. However, the framework also assumes that the processing of information and one’s perception of the world is affected by cognitive biases and constraints, which results in the notion that “actors always perceive the world through a lens consisting of their pre-existing beliefs” (Sabatier, 1998: 109).
The ACF distinguishes three types of beliefs that are divided in a hierarchical structure going from very broad to more specific beliefs. At the highest and thus broadest level, are the deep
core beliefs that normally span across all policy subsystems within a government. Deep core
beliefs entail very general, ontological and normative assumptions about human nature, and cover fundamental values such as equality, freedom and the balance between government and markets. Deep core beliefs are very hard to change, since they are largely formed during childhood socialisation (Sabatier & Weible, 2007: 194).
The next level covers policy core beliefs, which represents the basic applications of the deep core beliefs across the whole policy domain and thus subsystem. The ACF assumes that policy participants are willing to invest effort in the application of their deep core beliefs to develop policy core beliefs in their specific subsystem. Policy core beliefs are causal perceptions and normative commitments that include fundamental value priorities (such as economic gains versus environmental protection) within a specific subsystem. Causal perceptions address the perception of the urgency of the problem and its principal causes, and therefore relate to fundamental policy choices such as the level of government that needs to deal with the problem, and the choice of broad policy instruments. Since policy core beliefs are subsystem-wide, they can be used to identify different advocacy coalitions within a subsystem. The ACF namely assumes that policy core beliefs are the “fundamental ‘glue’” (Sabatier, 1998: 103) holding an advocacy coalition together. As an effect, policy core beliefs are as well hard to change. The lowest level and thus narrowest in scope are the secondary beliefs, or secondary aspect of the belief system within a coalition. They are less than subsystem-wide in scope and address policy preferences, the design of specific institutions and regulations, budgetary allocation, relative importance of certain causal factors in specials locales and the evaluations of actors’ performances. Because secondary beliefs are narrower in scope, they are more open to change when new evidence, experience or data comes available. In addition, changes in secondary beliefs should require less agreements since less actors are involved, thus making it easier to adjust them.
21
Policy brokers
The conflicting strategies to achieve certain policy goals and decision are normally mediated by a third group of actors within the subsystem that is referred to as policy brokers. Their principle task is to find a reasonable compromise within the conflicting strategies of the coalitions in order to reduce conflict, so that eventually governmental programs can be formulated that produce policy output.
Exogenous factors
The two sets of exogenous variables – the upper relatively stable, the lower more dynamic – affect the constraints and resources of the actors within the policy subsystem. The stable factors include the constitutional structure and fundamental norms and social-cultural values of the country wherein the subsystem is situated, the distribution of resources in the political system and the basic characteristics of the problem. Since these factors are hard to change, they are seldom used as a strategy by advocacy coalitions and therefore rarely are drivers of policy change. They do, however, demarcate the framework wherein the subsystem operates, and therefore influence the opportunities and constraints for subsystem actors. In contrast, the dynamic set of external factors is more likely to change over the years, and is therefore considered as a critical cause for major policy change by the ACF. The factors include changes in the governing coalition, public opinion, socio-economic conditions and movements, and address the possible impact of policy decisions in other policy subsystems.
2.7 Four processes that lead to policy change
Policy-oriented learning
The ACF assumes that each coalition adopts one or more strategies regarding ‘guiding instruments’, which are changes in information, rules, personnel or budgets, in an attempt to influence the behaviour of governmental institutions in a way that corresponds to their desired policy objectives (Sabatier, 1998: 104). The end result – with the help of policy brokers – is one or more governmental programs that produce policy output on the operational level. The ACF assumes then that “on the basis of perceptions of the adequacy of governmental decisions and/or the resultant impacts, as well as new information arising from search processes and external dynamics, each advocacy coalition may revise its beliefs (primarily in the secondary aspects) and/or its strategies.” (Ibid.). This is the moment were policy-oriented learning can play a role in the revision of coalitions’ beliefs and strategies, which is also depicted in the
22
figure with internal feedback loops concerning perceptions on policy effectiveness, external dynamics and probable impacts of alternative policies (Ibid.). The ACF argues that learning is instrumental, which means that policy participants from various coalitions will try to further their policy objectives by seeking a better understanding of the world, i.e. their respected subsystem (Ibid.). However, due to the cognitive bias of individuals and perceptual filtering of information, coalition members will most likely resist to take information into account that opposes their deep core or policy core beliefs. Instead, they will usually exploit formal policy analyses to explain and support their beliefs, and thus attack the opposing coalition’s beliefs (Ibid.: 104-105).
External shocks
However, instrumental, i.e. policy-oriented learning is but one of the four sources of policy change within Sabatier’s ACF. Besides this cognitive activity, external system events or ‘shocks’, internal shocks and a negotiated agreement can have effect on policy change (Sabatier & Weible, 2007: 208). Firstly, external shocks are developments outside the subsystem that can extensively influence the distribution of resources and the composition of coalitions. External shocks are depicted on the bottom left side in figure 1 as changes in the system dynamics. Examples of external shocks are changes in socio-economic conditions, regime change, outputs from other subsystems, and disasters and crises (Ibid.: 198-199). These external shocks can have significant effects on public policy by shifting agendas, drawing attention from key decision-makers and focusing public attention. The external shocks can thereby redistribute resources, and influence the opening or closing of windows of opportunities for policy initiatives. This may lead to the replacement of a dominant advocacy coalition within a subsystem and might alter the policy core beliefs of advocacy coalitions (Ibid.: 199).
Internal shocks
Internal shocks, then, are disasters within the policy subsystem that can lead to major policy change. Examples are oil spill or aviation disasters that often have severe consequences for the policy within that domain. Similar to external shocks, internal shocks can redistribute the critical political resources within the subsystem. This may alter the relation between dominant and minority coalitions. However, an effect exclusively belonging to internal shocks is the influence on the credibility of policy core beliefs, whereby internal shocks mostly confirm the policy core beliefs of the minority advocacy coalition and can increase doubt on the dominant coalition’s beliefs (Sabatier & Weible, 2007: 204-205). In sum, this is also the distinction
23
between internal and external shocks: the former directly questions the dominant coalition’s policy core beliefs, while the relation between external shocks and policy core beliefs is less clear (Ibid.: 205).
Negotiated agreement
The last possibility that can alter public policy is a negotiated agreement after a hurting stalemate, i.e. a policy deadlock. The agreement can occur between two or more coalitions if conditions have occurred that facilitated to negotiate on agreements and the implementation of those agreement in a safe environment, which often is a professional forum (Sabatier & Weible, 2007: 206-207). Nine conditions need to be present in order to establish professional fora that can reach changes in policy after negotiated agreements: incentives to negotiate seriously, i.e. a hurting stalemate, composition, effective leadership, consensus-based decision rule, diverse funding, duration of process and commitment of members, an emphasis on empirical issues, a focus on building trust and an absence of alternative venues (Ibid.)
These four paths towards major policy change lay at the foundation of the fifteen hypotheses that are formulated within the ACF (Sabatier, Weible & McQueen, 2009: 129). Since this research focuses on policy-oriented learning, four of the five hypotheses on learning are selected to be tested, supplemented with two additional hypotheses covering alternative explanations for policy change that serve as a control variable. The hypotheses are presented in the next section.
2.8 Hypotheses
The hypotheses are used to investigate the correctness of causal propositions following from the ACF. This research includes four hypotheses on policy-oriented learning across belief systems (i.e. between coalitions) directly transferred from Sabatier’s framework (1998: 106). The premise herein is that: “coalitions resist changing their policy core beliefs or important secondary aspects, and thus only solid empirical evidence is likely to lead them to do so” (Ibid.: 105).
The first hypothesis claims that a “productive analytical debate” (Sabatier, 1988: 155) can occur between different advocacy coalitions, which fosters policy-oriented learning. The premise for this debate is an intermediate level of informed conflict, which requires two conditions. First, both parties need sufficient information and data in order to be able to criticise the other’s data and causal models, and use their scarce resources to engage in a debate (Ibid.). Secondly,
24
analytical debate is unlikely in the case of a frontal collision of the coalitions’ deep core and policy core beliefs, since those beliefs are hard to change. Therefore, the conflict needs to be between lower levels of the belief system (Ibid.).
Hypothesis 1: “Policy-oriented learning across belief systems is most likely when there
is an intermediate level of informed conflict between the two coalitions. This requires that: (i) each have the technical resources to engage in such a debate; and that (ii) the conflict be between secondary aspects of one belief system and core elements of the other or, alternatively, between important secondary aspects of the two belief systems.”
Since policy core beliefs are hard to change as well, an analytical debate on opposing policy core beliefs is more likely to occur when experts of both coalitions are forced to confront each other in an apolitical setting, which is presented as a forum (Ibid.: 156). A forum urges for professional credibility following from the norms of scientific debate, which leads to a serious analysis of the analytical assumptions underlying the different coalitions’ plans. Eventually, this may converge both parties’ views on the essence of the problem and various policy alternatives to address that problem.
Hypothesis 2: “Policy-oriented learning across belief systems is most likely when there
exists a forum which is: (i) prestigious enough to force professionals from different coalitions to participate; and (ii) dominated by professional norms.”
Besides criteria for engaging in analytical debate, the “analytical tractability” (Ibid.) of a problem plays an important role, reflected in the presence of clear performance indicators and causal relations of a policy problem. Conditions that provide an improved understanding of the problem and proof of possible effects are considered to foster policy-oriented learning.
Hypothesis 3: “Problems for which accepted quantitative data and theory exist are
more conducive to policy-oriented learning across belief systems than those in which data and theory are generally qualitative, quite subjective, or altogether lacking.”
However, since advocacy coalitions may be very steadfast due to their deep core and policy core beliefs, cognitive bias, or a combination of both, an accumulation of technical information does not always alter the view of the opposing coalition. Yet, that information can influence policy change, however not as a result of cross-coalition learning, but by influencing the perceptions of policy brokers (Sabatier & Jenkins-Smith, 1994: 192-193).
25
Hypothesis 4: “Even when the accumulation of technical information does not change
the views of the opposing coalition, it can have important impacts on policy—at least in the short run—by altering the views of policy brokers.”
It is assumed that the presence of policy-oriented learning can to some extent explain certain changes in policy. However, the ACF also distinguishes factors that can influence policy change. The research thus needs a control variable to explain possible policy change in absence of policy-oriented learning. Therefore, an alternative hypothesis is formulated based on the causal processes (see 2.7) and hypotheses addressing policy change by Sabatier (1998: 105-106).
Hypothesis 5: “In the absence of policy-oriented learning, policy change can be
explained by: (i) significant perturbations external to the subsystem; (ii) significant perturbations internal to the subsystem; (iii) a negotiated agreement after a hurting stalemate; or (iv) a hierarchically superior jurisdiction.”
Lastly, the research needs a control variable on the situation whereby policy-oriented learning and/or factors stated in hypothesis 5 are present, but without the result of significant policy change. According Sabatier (1998: 106), this is the result of an ongoing dominance of one coalition within the subsystem.
Hypothesis 6: “The policy core attributes of a governmental program in a specific
jurisdiction will not be significantly revised as long as the subsystem advocacy coalition that instituted the program remains in power within that jurisdiction.”
The next chapter elaborates on the methods used in this research, and provides the operationalisation of the hypotheses in order to enable accurate measurement.
26
3. Methodology
This chapter describes the methods that are used to conduct the research. A base line within this methodology is the starting point that this research is deductive, since it aims to test hypotheses derived from the ACF in a specific case, and that this theory testing is being done in a qualitative way, thus making this research deductive and qualitative in nature. Firstly, the case study method and case selection are justified. Then, the concepts derived from the ACF are operationalised. Thereafter, the methods of data collections and data analysis are explained. Lastly, the validity and limitations of this research are discussed.
3.1 Case study method
This research makes use of the case study method since it aims to thoroughly understand and explain a certain phenomenon (the case) within its context (Yin, 2003: 39). According to Yin (2003: 7-9), the case study method has distinct advantages if the researcher is to conduct explanatory research on “a contemporary set of events over which the investigator has little or no control” (Ibid.: 9). In addition, case studies usually thrive well on prior established theoretical propositions that guide the data collection and analysis, and when there are multiple sources of evidence that need to be converged for the purpose of drawing conclusions (Ibid.: 14). The case of changing Dutch counterterrorism policy between 2017 and 2019 matches the criteria to conduct case study research since it is a developing set of events with no control for the researcher whatsoever. In addition, as will be explained in the section addressing data collection, multiple sources of evidence regarding the case exist, which are guided by theoretical propositions derived from the ACF. Following the goal of testing theoretical propositions on a particular phenomenon, the case study enables to determine what conditions within the case are of influence on the judgement to either verify or falsify the propositions. In other words, since a case study is rich in data, it enables the researcher to gain a better understanding of the event under study, thereby allowing to distinguish conditions critical to the specific case (Resodihardjo, 2009: 30).
Case selection
The selected case in this research is counterterrorism policy in the Netherlands in the period 2017-2019, which corresponds to a single-case design (Yin, 2003: 39). The focus is on two specific measures since they are part of the most recent addition to the arsenal of CT-measures. The two selected measures directly generated controversy, and are therefore most
27
relevant to use as a handle to investigate policy-oriented learning. However, the aim of the research is also to investigate policy-oriented learning in the broader CT-domain. The two measures are therefore not actively and consistently compared in the analysis, but serve as the vehicle to measure the concept of policy-oriented learning among actors in the Dutch CT-domain.
The selected case serves as a critical case, meaning that it enables to critically test the significance of a theory (Ibid.: 40-41). The premise herein is that the case should meet all conditions necessary to test a clear set of propositions, as well as the context in which those propositions are believed to be true according to the theory. The selected case fits these criteria since – as will be showed in the coming sections – different actors from the ACF (advocacy coalitions, external events, etc.) can be distinguished that meet the given conditions that underlie the assumptions and propositions, and enable them to be tested. However, another assumption in the ACF is that the context of the phenomenon is highly relevant. This research therefore uses an embedded single-case study design, which enables to demarcate multiple units of analysis within the selected case (Ibid.: 40, 42-46). Since the domain of counterterrorism policy consists of a wide variety or actors whose possible influence on the phenomenon needs to be taken into account, the embedded design of the single-case study is most appropriate for this research.
In addition, the case provides an opportunity to analyse extensive policy change in a policy subsystem in a timespan less than the decade that is indicated by the ACF. A decade is normally assumed necessary to pass the full policy cycle and encounter the effects of policy change. In this case, however, policy revisions were completed in three years, the effects of the changes were almost directly observable, and steps of the policy cycle like policy evaluation and agenda setting have been present in the shorter timespan as well. This notion, combined with the counterterrorism policy domain being an under researched area using the ACF, makes this case selection appropriate and also relevant to test and possibly further develop the ACF.
Limitations
Importantly, the analysis focuses on the policy domain that implements CT-policy, thereby excluding the role of politics in this policy domain. This research perceives laws underlying CT-policy as given by the legislators (i.e. politicians), and thus excludes the political legislation process that precedes policy formulation and implementation from analysis. As a result, politicians are not incorporated in the advocacy coalitions. In addition, it should be noted that
28
the selected case focuses exclusively on domestic counterterrorism policy in the Netherlands. Despite the fact that certain policy developments are inherently connected to events abroad, this research concentrates on the domestic counterterrorism policies aimed at the overarching goal of national security in the Netherlands.
3.2 Operationalisation
This section operationalises the most important concepts in this study. The concepts are derived from the ACF and the following hypotheses. Concepts are operationalised to provide indicators that indicate the presence of a concept within the collected data. A clear operationalisation is therefore needed beforehand to decide on the methods of data collection, and guides the analysis after data collection afterwards.
Policy-oriented learning
Policy-oriented learning is perceived as present when it is stated that an actor has gained deeper understanding and more knowledge about past policy, and takes the effects of past policy into account while judging about current policy and thinking about possible policy reforms. Following the definition of policy-oriented learning by Sabatier, the concept is present when actors state that past experiences have changed their beliefs regarding the functioning of CT-strategies and the use of specific instruments, and that they are willing to adjust their beliefs in the future for the sake of better goal attainment of the CT-policy or specific CT-measure.
Advocacy coalition
An advocacy coalition can be demarcated within a policy subsystem if actors indicate that they hold (more or less) the same conception on policy objectives and policy functioning regarding CT-policy, and that they seek co-operation with other actors pursuing the same objectives. Indicators of the existence of coalitions among CT-actors therefore are statements about: (i) the role that government should play in CT-policy; (ii) the role of national versus local government in CT-policy implementation; (iii) judgements about the functioning of current CT-policy; (iv) possible ways to improve current CT-policy; (v) perceptions of the criticism of the public on the two new CT-measures; (vi) co-operation with other actors; and (vii) perception of involvement and influence of other actors within the subsystem.
The belief system
The belief system that serves as a vehicle to assess actors’ influence on policy development consists of three levels. Different beliefs can be demarcated by looking for the following
29
indicators in actors’ statements: (i) deep core beliefs: opinions on freedom and equality of persons and perceptions of human nature; (ii) policy core beliefs: causal perceptions about the problem, cause and threat of terrorism in the Netherlands, and the general policy strategy to address that problem. This general strategy involves the level of government that needs to deal with the problem, as well as the general conviction underlying that strategy, for instance a hard, punitive (repressive) or a soft, re-integrative (preventive) CT-strategy; (iii) secondary beliefs: preferences on the design and implementation of specific measures and institutions, and the allocation of budgets and personnel following from specific policies, which in this research are the freezing and revoking measures.
Policy brokers
Policy brokers are the mediating party within the policy process in the subsystem. Indicators for the presence of policy brokers therefore are statements from actors about other actors that do not necessarily have a policy preference but still play a significant role within the CT-domain, and engage in reasonable problem solving rather than harsh persuasion of their own objectives. Actors that serve as the ‘mailbox’ for the complaints of other actors, with the goal to present those complaints to other actors, can also indicate the presence of policy brokers.
Informed conflict
Informed conflict is visible within the subsystem when two opposing coalitions confront each other and engage in a debate on their opposing policy beliefs. These debates can take place in the form of either written correspondence, direct meetings or indirect debate using for instance the media or (academic) research. The two given conditions in hypothesis 1 entail technical resources and the level of beliefs that are at stake in the discussion. Technical resources are data and information that tend to proof certain policy choices and effects of instruments proposed by the coalition, or data and information that can counter the opposing coalition’s plans. The belief levels at stake in the discussion are stated in the hypothesis and are operationalised earlier in this paragraph.
Professionalised forum
A professionalised forum is operationalised by searching for the occurrence of meetings and discussions outside the political realm (apolitical setting). This ‘external’ setting may allow the presence of academics and researchers or the broader public. The participants and topics are important indicators for a forum, since experts from both coalitions need to present, discussing
30
their conflicting policy beliefs following scientific and professional norms. An indicator can therefore also be that the focus in the discussion is on the (side-)effects, effectivity and effectiveness of the policy, instead of political desirability.
Analytical tractability of the policy problem
Analytical tractability of the policy problem relates to the ability of information and data to serve as proof for the effects of certain policy choices. Quantitative performance indicators are herein considered as generally more accepted as proof than qualitative indicators, since the latter generally is more subjective (Sabatier, 1988: 156). The presence of quantitative data on the policies’ effects can therefore foster policy-oriented learning (Ibid.). Quantitative data corresponds to numerical evidence and figures. Qualitative data, in contrast, corresponds to opinions and observations rather than numbers and figures, and is therefore considered more subjective and in general less accepted as substantive proof for policies’ effects (Ibid.). Statements on effect measurement can therefore indicate something about the analytical tractability of the problem.
Significant perturbations external to the subsystem
An event happening external to the policy subsystem that can influence the distribution and composition of resources and coalitions, for instance changes in the governing coalition, public opinion, socio-economic conditions and movements, and addresses possible impact of policy decisions in other policy subsystems. The first three events can be observed in the media, while the last event is more difficult to observe since it takes place in a more closed governmental environment. When statements of actors are present that point at certain events external to the subsystem that had significant influence on the CT-policy domain, an external perturbation may be present.
Significant perturbations internal to the subsystem
Internal shocks are disasters within the policy subsystem that can lead to major policy change. Examples are an oil spill, an epidemic among cattle or an aviation disaster, which can all have significant consequences for the policy within that domain. When statements of actors are present that point at certain events internal to the subsystem – thus related to (counter)terrorism – that had significant influence on the policy domain, an internal perturbation may be present.
