The many shapes and sizes of meta-governance. An empirical study of strategies applied by a well-advanced meta-governor: the case of Dutch central government in education

ABSTRACT

This article aims at deepening the understanding of how central governments enact meta-governance. Drawing on meta-governance and policy network theory, a heuristic analytical framework of meta-governance strategies was applied on two contrasting Dutch education policy cases. The results show that the use of meta-governance strategies differs according to the degree of formal responsibility of the central government. Creating nodes in a policy network turns out to be a specific strategy, and the effectiveness of different meta-governance strategies is interrelated. Lastly, the lack of involvement of education practice impinge on the adequacy of meta-governance practice.

KEYWORDS:

• Meta-governance
• governance
• policy networks
• education policy
• policy implementation

Introduction

In increasingly plural, networked societies, governments worldwide experiment with new forms of governance (Christensen and Lægreid 2007; Klijn 2008; Kooiman 1993; Pierre and Peters 2000, 2005; Rhodes 1997, 2007), including indirect remote forms of central coordination and control, which has led to practices of meta-governance (Bell and Park 2006; Fink 2019). Meta-governance entails strategies by which governments seek to control indirectly and steer at a distance, in order to shape and pursue policies (Hammond et al. 2019; Klijn and Koppenjan 2000; Kooiman 1993).

In this study, we aim to deepen our understanding of how governments enact meta-governance. In the substantial amount of conceptual and empirical research on meta-governance, some studies explicitly focus on central governments in their role as meta-governors (see: Damgaard and Torfing 2010; Pahl-Wostl 2019; Ottens and Edelenbos 2019; Qvist 2017; Temmerman, De Rynck, and Voets 2015; Wilson et al. 2017). We build on this line of research by investigating empirically what role central governments take as meta-governor, what strategies they apply and why. Yet, in the growing body of knowledge about strategies that meta-governors deploy, the question of whether and how to exercise their formal responsibilities has not yet found a place. Therefore, in response to the call in the research literature for a better understanding of the conditions under which meta-governance operates and the causal mechanisms behind it (Gjaltema, Biesbroek, and Termeer 2019; Klijn 2008), we seek to understand whether the degree to which central government is formally responsible for policy outcomes is an explanatory factor for different strategies used by central government as a meta-governor.

We consider education policymaking and implementation in the Netherlands a prime example to study central government as a meta-governor and the strategies used. First, the Netherlands has a strong affinity with network governance, due to its long-held consensus culture (Meuleman 2008; Pahl-Wostl 2019). Second, and even more importantly, the Dutch field of education has a longstanding history of, on the one hand, autonomous schools protected by the constitution since 1917, and on the other hand a national government with clear formal responsibilities for education. As the constitution has made straightforward command and control in education unfeasible, meta-governance ‘avant la lettre’ has always offered Dutch government a way out, seeking to steer education at a distance, in an indirect manner (Hooge 2017). Hence, in education, Dutch central government has gained experience with meta-governance for over a century, turning it into a well-advanced meta-governor in dealing with varying degrees of formal responsibility for education.

Our study focuses on the following research question:

What meta-governance strategies does the Dutch central government apply, and how do they relate to the degree of formal responsibility of central government for education?

We focus on two contrasting education policy cases: raising student literacy and numeracy skills and stimulating schools to become learning organizations. These cases represent opposites concerning the degree of formal responsibility held by central government.

This article opens by developing a heuristic analytical framework based on meta-governance theory, compiling meta-governance strategies into three distinct categories. Second, after sketching a general background of education policymaking in the Netherlands and introducing the two cases, we describe our methods of data collection and analyses. Third, the analytical framework is applied to the two cases, providing a thick description of meta-governance strategies enacted by the Dutch government, after which cross-case analyses reveal how these strategies are related to the degree of formal responsibilities. The final part of this article draws conclusions and reflects on the meaning of the results for theory and practice.

Understanding meta-governance: a heuristic analytical framework

Ever since the concept of meta-governance was developed (Kooiman 1993; Jessop 1998, 2002), it has become an important topic in governance and public policy literature and is studied from different angles. We start by indicating how our study fits in this ever-expanding landscape.

The dominant perspective of meta-governance ascribes central governments the role of meta-governor. Because of their special status: central governments provide ‘the ground rule for governance and the regulatory order in and through which governance partners can pursue their aims’ (Jessop 2002, 6). In virtue of special resources at their disposal, such as sizable budgets and personnel, specific powers, access to mass media or democratic legitimacy, central governments can either spark networks to enact steering and policymaking, or disempower, marginalize and even abolish them (Klijn and Koppenjan 2000; Jessop 2002).

On the basis of their recent systematic literature review, Gjaltema, Biesbroek, and Termeer (2019) have identified four ideal types of meta-governance. In the type ‘meta-governance of modes’ the meta-governor balances the three broad governance modes hierarchy, markets and network governance that are traditionally distinguished in the literature (Jessop 1998, 2002; Pierre and Peters 2000; Kooiman 2003). Yet, this ‘overarching’ type of meta-governance of modes is not common because in most cases meta-governance is directed at only one of the three broad governance modes, namely, network governance (Gjaltema, Biesbroek, and Termeer 2019). This is where the other three ideal types focus on: the ‘network meta-governance’ type in which the meta-governor directs a governance network that includes actors from several domains, the ‘multilevel meta-governance’-type where meta-governors from different levels are active in relation to the same network, and the type of ‘meta-governance of multiplicity’ in which the meta-governor orchestrates multiple governance networks.

From its long tradition of activating, strengthening or even specifically creating specific networks for many education policy issues, Dutch central government engages in complex network steering when pursuing its education policy as a meta-governor, orchestrating several networks simultaneously. Hence, the ideal type of ‘meta-governance of multiplicity’, which draws on the concept of network governance and places a strong research focus on networks, seems a fruitful starting point to study how Dutch government engages in meta-governance.

Meta-governance and network governance

Network governance reflects horizontal coordination between mutual interdependent public, private and civil society actors. Analyses of feasible coordination and control mechanisms of these forms of horizontal coordination have led to meta-governance theory (Bell and Park 2006; Fink 2019).

In the literature, the concepts of ‘network governance’ and ‘networks’ are heavily intertwined and rooted in the same strands of theory (Klijn 2008). Network governance focuses on – sometimes ‘self-organizing’ – networks, referred to as ‘governance networks’ (Klijn 2008; Klijn and Koppenjan 2016; Sørensen and Torfing 2009), that function both with and without government to solve societal problems, deliver public services and shape and implement public policy (Hanf and Scharpf 1978; Jessop 1998; Kickert, Klijn, and Koppenjan 1997; Klijn 2008; Kooiman 1993; Marin and Mayntz 1991; Marsh and Rhodes 1992; Pierre and Peters 2000; Rhodes 1997). Klijn (2008) identifies three research traditions looking at governance networks: one stemming from political science focusing on policy networks, another originating from organizational science concerned with service delivery and implementation, and a third rooted in public administration with a focus on governing networks. Over time, the focus of these traditions increasingly overlaps and researchers adopt integrated approaches, particularly with regard to meta-governance, (Klijn 2008; Lecy, Mergel, and Schmitz 2014). Our study must be seen as combining elements from all three traditions: we are interested in the actors that are involved in decision-making and their mutual relations, our focus includes policymaking as well as policy implementation and we ask how horizontal governance relations are connected with central government. We use the term ‘policy networks’, but in a broader sense than Klijn (2008), to accentuate that in all three traditions, networks are related to policy. In deploying meta-governance strategies, government may create, design, change or abolish policy networks as well as lay out their structure and procedures, (Jessop 1998; Klijn and Koppenjan 2006; Marin and Mayntz 1991; Marsh and Rhodes 1992).

Meta-governance strategies

Our aim is to study empirically how central government operates as a meta-governor, to identify, demarcate and describe the repertoire of meta-governance strategies that are used, and examine how these strategies are related to formal responsibilities. In the political science and public administration research literature, various strategies by which meta-governance may be exercised have been identified and classified (Gjaltema, Biesbroek, and Termeer 2019; Kickert, Klijn, and Koppenjan 1997; Klijn and Koppenjan 2000, 2016; March and Olsen 1995; Sørensen and Torfing 2007).

From the angle of institutional design and policy network theory, and relying on the seminal work of March and Olsen (1995) and Kickert, Klijn, and Koppenjan (1997) present a classification of three types of meta-governance strategies. The first category is network design strategies, deliberately attempting to shape and structure networks, such as the in- and exclusion of actors, or facilitating actors to engage in a network. This first category is comparable with what Klijn and Koppenjan (2006) call composition strategies. The second category comprises network framing strategies, focusing on the formulation of goals and objectives to be pursued. And the third category is network participation strategies, meaning that politicians participate directly in networks. We use the classification by Kickert, Klijn, and Koppenjan 1997 with two alterations. First, in the original classification resourcing strategies, such as providing actors with funding, legitimation or knowledge, are considered to be part of network design; we choose to distinguish these strategies as a separate category. Second, we consider network participation strategies as a form of rather direct governance and steering. As our study is on indirect meta-governance strategies, we did not use this category in our framework. We build here on the work of Gjaltema, Biesbroek, and Termeer (2019) who consider process management, and the governor participating and interfering in a network as something quite different from the indirect steering which is something different than the indirect steering meant by meta-governance (see also Wilson, Morrison, and Everingham 2018).

For purposes of our empirical analyses, we break each category of strategies down into components of specific activities. The framework is given in Table 1.

Table 1. Heuristic analytical framework.

Network design strategies	Influence the composition of the policy network Establish new actors Re-arrange and shift positions of actors Influence actors’ roles
Resourcing strategies	Influence the activities undertaken by actors Provide/withhold actors with funds for specific purposes Enable/disable activities by actors through provisions such as regulations and fiscal law Grant/retract actors with knowledge and authority
Framing strategies	Influence the perception and sensemaking of actors regarding The (content of the) policy issue The urgency of the policy issue The purpose of the policy The scope and/or specifics of policy goals by setting indicators, standards and targets

Context, case selection and methods

This section first provides more background on the particularities of policymaking in Dutch education. Next, two specific cases in education policy are introduced. Both cases refer to secondary education, and in line with our research question they represent opposites in the degree to which Dutch central government is formally responsible for the issues addressed. In the first case of the ‘raising standards’ policy, central government has full formal responsibility for the enhancement of student’s literacy and numeracy skills. In the second case of the ‘learning organization’ policy, central government has neither the formal responsibility nor the formal authority to interfere with the way schools organize themselves. By selecting these contrasting cases, we not only expect to discern a great variety of meta-governance strategies, but also to deepen our understanding of the relationship between the strategies employed and the degree of formal responsibility of central government for education. We conclude this section by detailing how various data sources were used and how the data were collected and analysed.

Policymaking in Dutch education

Freedom of education has long constituted the backbone of the Dutch education system. Typically, Article 23 on education in the Dutch Constitution incorporates the inherent tensions between freedom of education and government intervention, as it both protects against and legitimates government intervention in education. On the one hand, the provision of education is free in the Netherlands and education laws and regulations must always meet the freedom to provide education according to religious or other belief. This freedom is in effect a freedom of providers and particularly applies to the choice of learning materials, the hiring of teachers and school’s internal organization (OECD 2018; Author 2017). On the other hand, Article 23 expresses the government’s responsibility for education, formulated as ‘Education shall be the constant concern of the Government’ and allows the government to set standards for schools to become eligible for funding. This legitimates government interventions, as it approves the national government to set statutory requirements to ensure a minimal level of educational quality, and demand that all schools fall under the scrutiny of the Dutch Inspectorate of Education. Central government has the right to translate the statutory requirements into organizational requirements relating to school subjects, class schedules, final attainment levels and examination regulations. However, when designing educational policy and legislation, the desire to adjust the statutory requirements in line with political and social views, may be in conflict with the interpretation of the freedom of education. This repeatedly causes political and social controversy in the Netherlands up until this day. The Dutch Ministry of Education continues to walk a fine line. It is very reluctant to steer education too ostentatiously, to formulate policies and measures that encroach on school autonomy, or to control school districts too directly with law and regulation. If central government oversteps the marks, this draws heavy criticism and triggers controversy from religious and private political/social parties (Author 2017, 38).

Two cases of education policy

Case 1: policy issue ‘Raising standards in literacy and numeracy’

In 2010, Dutch parliament passed a law introducing learning standards for literacy and numeracy, prescribing that students must meet well-defined minimum levels of proficiency at different points in their education careers. To support implementation, school districts received additional funding as part of their block funding. Interestingly, the law works out very differently for literacy and numeracy. Proficiency levels for literacy found their way into examination guidelines for Dutch language and are incorporated in the standardized national exams for Dutch language. Without much discussion, the raising standards policy for literacy was implemented in all secondary schools.

In contrast, from 2013 onwards, proficiency levels for numeracy found their way to a special numeracy test. By including test results in final examinations and in school assessments by the Inspectorate as of the school year 2015–2016, the test was to become high stakes for both students and schools. The latter spurred protests, objections and complaints by the educational field. After tests in pilot schools repeatedly indicated disappointing results and signalled that high numbers of students would fail their exams, the policy around the numeracy test was adapted several times. Ongoing resistance and continued disappointment about test results prompted government in 2017 to abolish the special numeracy test altogether.

Case 2: policy issue ‘Schools becoming learning organizations’

In 2006, government introduced the Education Professions Act (Wet BIO), requiring school districts to establish human resource policies, keep competency files for teachers, ensure that teachers’ competencies are maintained and execute performance interviews with all staff regularly. Due to the constitutionally anchored high degree of school autonomy, the Dutch government has limited possibilities to intervene in the organization of school districts and in their human resources policies, procedures and guidance on school employees and teachers. With this new law, the control of central government in this area has been slightly extended.

In the ‘Teacher’ Action Plan by the Ministry of Education in 2007, the concept of schools as learning organizations emerged for the first time. In 2011, the government presented separate action plans for all educational sectors and an additional action plan for teachers to stimulate schools to become learning organizations. The ‘schools as learning organizations’ policy took further shape when it featured prominently in the Teachers Agenda 2013–2020, launched by central government. The Teacher Agenda provides a long-term policy perspective and is of great significance for the educational field, yet it has no foundation in law. Consequently, central government has little formal control on whether and how schools are to become learning organizations. It is the responsibility, and authority, of school district leaders (superintendents) to either adopt or reject the policy, and decide on measures to promote professionalization and the learning and collaboration of (teams of) teachers and school leaders.

Data collection and analyses

To study the meta-governance strategies applied by the Dutch central government, a three-year timeframe was defined: from 1 January 2013 until 31 December 2015. To discern the strategies empirically and in detail, three different data sources were utilized for qualitative analyses: documents and websites (N = 47 for Case1 and N = 59 for Case 2); face-to-face interviews with key players (N = 4 for Case1 and N = 4 for Case2); and a joint member check in which a total of 19 participants of both policy networks took part.

To determine the actors constituting the policy networks for each of the cases, and to explore the design strategies that led to this formation, we first collected relevant texts issued by the central government. We used Google and the Dutch government search engine ‘https://zoek.officielebekendmakingen.nl’ and (combinations of) the keywords numeracy, literacy, learning organization and secondary education, to collect policy documents. To keep focus, only documents with the policy issue as its main topic were included. Then we analysed these texts to identify actors that were specifically mentioned with regard to the policy issue, based on the trilogy of legal, economic and communicative policy instruments (Vedung 1998). This was followed by a snowball-method to trace additional actors by means of collecting relevant documents and websites using the same keywords as mentioned above. As a check on completeness and relevance, this list of actors was offset against a general overview of actors in the administrative environment of organizations in Dutch education (Author 2017) and policy reconstructions for the two specific cases (Ledoux et al. 2014). A panel of three experts in the field discussed whether actors should be added or removed, resulting in a penultimate list. For the final step, four key players in each of the policy networks were identified and with representatives thereof, semi-structured interviews were conducted, eight in total. During these interviews, we presented the results thus far and asked interviewees to reflect upon, complement and correct, provide additional information, and contradict or confirm the picture painted. Finally, a total of thirteen actors were identified as composing the policy network for ‘raising standards’, and a total of eleven formed the policy network ‘learning organization’. In Table 2, all actors are displayed per policy network (actors involved in the interviews are indicated in bold).

Table 2. Actors involved in the policy networks.

	Policy network ‘raising standards’	Policy network ‘learning organization’
Central government	Ministry of Education, Directorate of Secondary Education	Ministry of Education, Directorate of Teacher Policy
Independent administrative bodies	- Inspectorate of Education (IE) - National Centre for Tests and Examinations (CvTE)	- Inspectorate of Education (IE)
Independent advisory body	- Education Council of the Netherlands (EC)	- Education Council of the Netherlands (EC)
Temporary government-commissioned advisory committees	- Committee on Numeracy Reference Levels (CN) - Committee on Raising Numeracy Standards (CR)
Independent foundations, associations and private companies	- National Institute for Curriculum Development (SLO) - Council for Secondary Education, association of all 334 Dutch school boards in secondary education (VO-raad) - National Institute for Educational Measurement (CITO)	- Association Teacher! (AT!) - Council for Secondary Education, association of all 334 Dutch school boards in secondary education (VO-raad) - Expert and research organizations such as universities, (non-) commercial research institutes and consultancy firms
Actors without legal forms	- Support Office on Raising Standards (SORS) - Program ‘Schools have the Initiative’ (PSI)	- Program ‘Schools have the Initiative’ (PSI) - Political Initiative ‘Learning Together’ (PILT)
Composite actors	- Teacher24, comprising four actors, digital platform of, for and by teachers, aimed at supporting teachers in the practice of their profession - Framework for the Professionalization of Numeracy Teachers, comprising two actors (FPNT)	- Teacher24, comprising four actors, digital platform of, for and by teachers, aimed at supporting teachers in the practice of their profession - Education Cooperative, comprising eleven actors, cooperative society of large teacher unions and smaller professional groups, aiming at empowerment of teachers and organization and improvement of the teaching profession (Coop) - Education Foundation, comprising fourteen actors, platform of employer organizations, employee organizations, student organizations and the Coop, aimed at structural consultation with government about supporting and improving education (EF)

To trace the framing strategies, content analyses were conducted on the documents and websites that were collected during the procedure described above. Starting with four broad coding categories derived from the heuristic framework, a total of 941 text fragments were coded, see Table 3.

Table 3. Documents and coding.

		Case 1 Raising Standards	Case 2 Learning organizations
Documents
Total number		47	59
Coding
Issue-content	Description policy issue, including urgency, purpose	92	115
Issue-other	Other legitimations, e.g. connection with other policies	1	31
Activities	Pursued by the actor re the policy, e.g. conduct or commission research, provide info, support schools	230	165
Info	Information produced or used by the actor re the policy, e.g. monitoring, indicators, targets	137	170
		460	481

As the analyses focused on strategies applied by central government, the text fragments from the policy documents issued by central government were taken as a starting point and cross-checked by text fragments derived from other actors in the policy networks. Data reduction was achieved using MaxQDA in a process of more detailed in-vivo coding, memo-ing, code matrices, data display in tables and networks, summary grids and tables, and drawing and verifying conclusions (Miles, Huberman, and Saldaña 2014). The researchers kept a record of all ideas, interpretations and theories.

During the aforementioned interviews, key actors in the policy networks were also asked about their own and other actors’ role in the network, mutual relations in the network, activities pursued by the actor, and whether targets, indicators or standards were used. The interviews were transcribed in full and analysed by coding and data reduction techniques similar to the ones described above for the text fragments.

For reasons of intersubjective reliability, parts of the coding and data reduction techniques for both the text fragments as well as the interview transcripts, were conducted independently by two researchers, who then discussed their results in order to produce joint coding and a joint interpretation.

Finally, to enhance accuracy and validity of the results, a member check was carried out with a total of nineteen actors who participated in one of the policy networks (Meadows and Morse 2001). No alterations were made to the results and interpretations, but the meeting did spark a number of questions we will get back to in the discussion section of this paper.

Applying the framework for meta-governance strategies

We applied the analytical framework to both cases and present the findings in this section in Table 4, followed by explanatory text.

Table 4. Findings.

	Raising standards	Learning organization
Network design strategies
Establish new actors	– Initiate two new actors SORS and PSI – Stimulate the forming a new composite actor FPNT – Instal two specific temporary committees CN and CR	– Create a new large composite actor Coop
Re-arrange and shift positions of actors	– Position the VO-raad as key player by assigning responsibility to the VO-raad for the obtained results of school districts with respect to ‘raising standards’ policy – Position CvTE, IE and SORS as central actors – Foster the coalition of CN, CR, FPNT, CITO, CvtE, SLO, and SORS within the policy network as a very dense subnetwork which specifically focuses on raising numeracy standards – Bolster the composite actors Teacher24 and FPNT	– Position the VO-raad as key player by assigning responsibility for the obtained results of school districts with respect to ‘learning organization’ policy – Bolster the composite actors Coop and EF and position them as central actors – Integrate Coop into EF – Bolster the composite actor Teacher24
Influence actor’s roles and tasks	– Mandate IE to monitor how schools address literacy and numeracy disadvantage, and whether and how they implement raising standards policy – Formulate questions CN and CR must answer (operationalization and implementation of the policy) and what is beyond the scope of their advice (e.g. aims of the policy as such)	– Mandate IE to include indicators related to school developing as learning organizations in its supervisory framework and to launch an investigation focusing on schools as learning organizations – Mandate IE to support, advise and help schools to become learning organizations
Resourcing strategies
Provide/withhold actors with funds for generic or specific purposes	– Provide project funding to CITO to carry out a pilot project on diagnostic testing and student assessment systems with 300 schools – Provide project funding to CvTE and SLO to develop assignments, tests, exams and teaching methods and material – Financially support VO-raad, PSI, SORS, CITO, SLO and Teacher24 to support, advise and help schools to implement ‘literacy and numeracy education improvement’ – Provide project funding for FPNT to train and professionalize specialized numeracy teachers	– Provide project funding to the VO-raad, Coop, PSI and Teacher24 to support, advise and help schools to become learning organizations – Provide research grants to expert and research organizations to carry out research and consultancy projects on schools as learning organizations – Make yearly progress at the sector level a condition for school districts to obtain additional funds (Functiemix)
Enable/disable actors through provisions such as regulations and fiscal law	– Close an administrative agreement with the VO-raad addressing the ‘raising standards’ policy	– Close three administrative agreements with the VO-raad, Coop and EF addressing the ‘learning organization’ policy
Granting knowledge and authority to actors		– Grant authority to the Coop by referring repeatedly to it in policy documents, presenting it as a bottom-up initiative from teachers – Grant authority to AT! by presenting it repeatedly in policy documents as ‘best practice’
Framing strategies Influence the perception and sensemaking of actors regarding …
… the content of the policy issue	– Communicate the mantra ‘together’ in relation to the ‘raising standards’ policy, using terms as: ‘commitment of all parties involved’, ‘it must be done in the classroom by school leaders, staff, teachers’, ‘make joint efforts’, etc. – Communicate the particular roles for (numeracy) teachers and school leaders with respect to the ‘raising standards’ policy	– Commission research and consultant projects on how to give shape to schools as learning organizations in practice – Activate actors in the ‘learning organization’ policy network to conceptualize, advise and provide examples of good practices of the policy issue
… the urgency of the policy issue	– Communicate in urgent terms about ‘raising standards’ policy (e.g. ‘the alarm clock is ringing’, ‘students are at high risk of getting stuck’, ‘students should not be the victims of poor literacy and numeracy education’, etc.) – Allow schools additional time to implement the policy, but communicate the warning that this cannot be a reason for schools ‘to tone down their ambitions’ – Intensify the ‘raising standards’ policy in the field of numeracy education by communicating and by commissioning SORS to develop activities and instrumentation including setting up counselling interviews with schools.
… the purpose of the policy	– Communicate that all schools are expected to make strong efforts to get as many students as possible to the required attainment levels and pass the test	– Communicate that all schools are expected to become organizations where teachers, teams, leaders and school districts build a learning culture together in which they continuously work to improve the quality of education – Embed the notion of schools as learning organizations in a wider ambition for the education system as a whole, to make the step from ‘good’ to ‘great’
… the scope and/or specifics of policy goals by setting indicators, standards and targets	– Emphasize particularly on raising numeracy standards – Include specific targets for literacy and numeracy in the administrative agreement with the VO-raad: ‘Students succeed with a 0.2 point higher score on the final exam for the subject Dutch, English and mathematics’	– Turn the general policy goals with respect to ‘school as learning organizations’ into firm agreements about concrete targets for which the parties involved can be held accountable, e.g. the proportion of teachers involved in performance and peer review. – Formulate concrete targets at the sector level to be reached within five years, monitor the progress yearly.

Network design strategies

Table 4 shows that central government uses network design strategies widely. With respect to establishing new actors, government establishes notably more new actors in ‘raising standards’ case (five), than in the case of ‘learning organization’ policy (one). A representative of the Ministry recounts on establishing SORS:

The Support Office on Raising Standards is very important to us, it explains our ‘raising standards’ policy and how regulation and measures may be applied in the education practice. SORS is a connecting factor and also acts as a mouthpiece from the schools to the ministry.

The representative of SORS explains:

We have been created by the Ministry to inform the schools as best as possible about the raising standards policy and regulations. We act from a neutral position, however, we are explicitly in favor of the numeracy test. Otherwise, we would not be able to do our job.

Central government also re-arranges the network and (attempts to) shift positions of actors. In both cases, government positions the VO-raad (Council for Secondary Education, association of all 334 Dutch school boards in secondary education) as key player in the policy network, by making the VO-raad responsible for the obtained policy results in school districts and schools by means of an administrative agreement. However, the VO-raad is an association, districts affiliate with voluntarily, to join forces in their relation with central government. That its role is not so evident, is emphasized by a representative of the VO-raad:

The ministry perceives the VO-raad as a kind of ‘super school district’ in secondary education, as if all districts are puppets on our strings. But that’s just not the case.

The standpoint voiced by a representative of the Ministry, illustrates the tension:

In our view, the ‘learning organization’ policy development is not going fast enough, so we call the VO-raad to account, on the basis of our administrative agreement, and they, in turn, must call the school districts to account. Because we cannot call directly on the school districts.

Interestingly, the findings also reveal that central government aims to increase the interconnectedness and interdependency between actors in the network. In the ‘raising standards’ policy network, this is done by facilitating coalition formation of different actors around one specific element of the policy, which is ‘raising numeracy standards’. In the ‘learning organization’ case, central government uses this strategy by integrating the newly established large composite actor Coop (Education Cooperative, a cooperative society of all large teacher unions and smaller professional groups) into the vested large composite actor EF (Education Foundation, a platform of all employer organizations, employee organizations, and student organizations). A representative of the Coop reports:

As Coop we have a common agenda, including stimulating schools as ‘learning organizations’. But beyond that, our members have their own agendas. For example, on issues of employment or school governance. That makes it a rather complex construction for us.

Mandating actors in the policy network for specifically described tasks appears to be a way in which central government applies the strategy of influencing an actors’ role. The findings show how central government asserts its ‘own’ IE (Inspectorate of Education), a double role in both policy networks. On the one hand, government mandates IE to support, advise and help schools to implement the policy, and at the same time it mandates IE to assess schools by including indicators for raising standards and schools as learning organizations in the supervisory framework. Notably, this strategy is similar for both cases.

Resourcing strategies

Table 4 reveals that in both policy networks, central government concludes administrative agreements with the key player, the VO-raad. These agreements not only authorize and legitimize the VO-raad as key player and linking pin within the policy network, as mentioned above, but also commit the VO-raad to carry out activities which are in keeping with the content and goals of the policy. In the case of ‘learning organization’ policy, with no fewer than three administrative agreements, this appears the dominant resourcing strategy. Agreements with key-player the VO-raad, the new ‘government established’ large composite actor Coop, and the vested large composite actor EF (which in turn includes the Coop), ensure central government of policy commitment from the lion’s share of the education sector. The effectiveness of these agreements is reinforced by project funding, combined with specific targets to direct the activities of actors and hold them accountable. Central government also introduced a special funding strategy, by making the additional funds that can be obtained by school districts (Functiemix), conditional on whether targets at the sector level are met.

As shown by Table 4, central government frequently uses funding as a resource strategy, particularly in the form of project funding, to a specific actor for specific activities. Funding is also used to establish a new actor and to commission research. In the case of ‘raising standards’ policy, central government finances vested ‘government related’ actors such as CITO, SLO and CvTE and the new established SORS, to develop and disseminate multiple instruments to promote policy implementation in schools. In addition, a tight network of educational consultancy and support centres is financially supported to develop a large number of instruments and perform activities to ensure that advise, assistance, supervision, information and data are available to support schools to ‘raise standards’. In the case of ‘learning organization’ policy, central government finances key player VO-raad as well as Coop, PSI (Program ‘Schools have the Initiative’) and Teacher24 (a digital platform of, for and by teachers, aimed at supporting teachers in the practice of their profession) to support, advise and help schools to become learning organizations. In this policy network, central government concentrates on knowledge generation and dissemination in its resourcing strategies, by providing grants for research, pilots and experiments to organizations operating independently from government. This is a deliberate choice, as a representative of the Ministry explains:

Our role is to facilitate and create adequate conditions to realize policy ambitions in the educational practice. For instance, by subsidizing Teacher24, SLO or the VO-raad, or granting research and experiments on policy issues. With the aim that they take it further.

The representative of the VO-raad conveys how this actor has taken up the gauntlet:

Our district and school leaders are key actors in shaping the professional development of teachers, the idea of schools as learning organizations and other elements of the administrative agreement. The VO-raad is an important vehicle to set this in motion, for instance, our ‘VO-Academy’ provides training, learning and knowledge dissemination on these topics.

Another notable resourcing strategy is to (not) grant actors knowledge and authority. In the case of ‘learning organization’ policy, policy documents of central government refer repeatedly to Coop as a bottom-up initiative of teachers instead of being established by government, thus bestowing Coop with the authority to play a central role in the policy network. Similarly, the independent Association Teacher! (AT!) is mentioned repeatedly as a best practice, giving it credit. The case of ‘raising standards’ policy may point to the flipside of the resourcing coin, as actors expected to play some role in the policy network, are absent, particularly teacher unions and professional organizations of (mathematics) teachers. There is no evidence from documents or interviews that their absence is the result of a deliberate strategy, but we do note that none of the policy documents grants authority to these teacher organizations.

Framing strategies

Communication appears to enable central government to exert various framing strategies, see Table 4. In the case of the ‘raising standards’ policy, documents issued by central government itself continuously use specific words to underscore that the policy must be a joint effort, requires urgent action, by everybody, for all students. The content analyses of the documents reveal that other actors often repeat these words in their own communication.

In contrast to the direct communication by central government itself in the case of ‘raising standards’ policy, government applies more indirect ways to influence the perception and sense making in the case of ‘learning organization’ policy. For example, by means of commissioning research and consultancy projects on ‘(how) schools (can) develop in learning organizations’, and activating actors in the policy network to conceptualize, advise and provide examples of good practice of the policy issue. As a result, communication on the policy issue emerges in the network itself. With network actors constantly referring to and copying each other, which is very common in this policy network, it is not central government but other actors who ‘spread the word’.

Central government also appears to apply the strategy of influencing the scope and specifics of policy goals in different ways. In the ‘raising standards’ case, government repeatedly prioritizes raising numeracy standards above literacy standards in its communication. To reinforce this narrower scope, it promotes coalition formation within the policy network to specifically focus on raising numeracy standards. In the ‘learning organization’ case, central government leaves the conceptualization to other actors, but does formulate concrete targets and holds actors accountable by making funding conditional on reaching these targets. The representative of the VO-raad indicates that this strategy has been a point of contention:

We have tried to avoid very strict definitions to allow school districts to make their own interpretations of the policy issue. As VO-raad we aim to give direction and start a development, instead of setting strict targets.

Results

Regarding the meta-governance strategies Dutch central government applies, two general results deserve attention.

First, it is not so much that central government applies all the meta-governance strategies distinguished by the heuristic framework that is noticeable, but the interrelatedness between the strategies. By applying one strategy, another strategy becomes available or more effective. For example: by closing an administrative agreement with a key player in the policy network, central government simultaneously re-arranges the position of actors (network design), influences the activities undertaken by the actor by providing funds (resourcing), and influences perception and sensemaking by committing actors to particular wordings, indicators and targets (framing).

Second, promoting the interconnectedness and interdependency between network actors proves to be a salient network design strategy that was not included in the analytical framework a priori. This can be seen as a way to create key nodes in a network (Wilson, Morrison, and Everingham 2018). Dutch government enacts this network design strategy for example by creating composite actors (partnerships or affiliations of network actors) that can be regarded as sub-networks within the policy networks. Other examples are government promoting coalition formation to specifically focus on a ‘raising numeracy standards’ and integrating a newly established large composite actor into a vested large composite actor in the ‘learning organization’ network.

For the relationship between the meta-governance strategies used and the degree of formal responsibility of the central government for education, the results of the comparison between the different cases are relevant.

Cross-case comparison of meta-governance strategies reveals a difference between the direct and indirect use of these strategies. In the ‘raising standards’ case, government mainly applies the network design strategy of establishing new actors, whereas in the other case, only one new, yet very large composite actor Coop is established. By labelling Coop in policy documents as a bottom-up initiative of teachers, it is given an appearance of government independence. A similar pattern is found with resourcing strategies enacted to commission research and diffusion of knowledge. In the ‘raising standards’ case, these activities are geared towards implementation and contracted out to network actors established by or related to government, whereas in the ‘learning organization’ case, such activities are entrusted with ‘government independent’ networks actors who are allowed their own conceptualization.

A difference between direct and indirect use of strategies also characterizes how government applies framing strategies in both cases. In the case of ‘raising standards’ policy, it is central government itself that communicates directly to both the actors within as well as the actors beyond the policy network, such as school leaders, teachers, students and parents. In contrast, in the other case, central government does not so much communicate itself, but rather promotes that actors comprising the policy network communicate on the policy issue. As a result, these actors repeatedly refer to each other and copy each other, also ‘spreading the word’ beyond the policy network.

In other instances, the difference between strategies applied in the contrasting cases is mainly a matter of degree. For example, although central government closes administrative agreements in both cases, it does so no fewer than three times in the case of ‘learning organization’ policy, thus ensuring the commitment of a great many actors.

Lastly, the analyses show that differences between cases may only come into view when observing the combination of strategies. So, while the case of ‘raising standards’ policy exemplifies a direct use of meta-governance strategies across the heuristic framework, direct and indirect strategies are combined in the other case. As pointed out above, central government mainly enacts meta-governance strategies here in an indirect way. Yet, performance-based funding, which is arguably one of the most direct strategies central government can apply to influence an actor, also appeared in this case.

Conclusion and reflection

The main aim of this study was to depict in detail, based on empirical data, what strategies central government applies in its role as meta-governor, and how these strategies differ according to the degree of governments’ formal responsibilities.

Our main conclusion is that the degree to which central government is formally responsible for policy is an explanatory factor for the different meta-governance strategies it deploys. Degree of formal responsibility does not so much help explain which strategies are used, but rather how they are used. The less central government is formally responsible for a policy issue, the more it seems to turn to the indirect use of strategies when acting as a meta-governor. In our case of ‘raising standards’, for which government is formally fully responsible, central government applies network design strategies excessively, particularly to establish new actors and activate existing actors, all operating under the auspices of government itself. Likewise, it is government itself who openly communicates with both the actors within and beyond the policy network to point out the policy’s purpose, urgency and goals. In contrast, in the case of ‘learning organization’, for which government has no formal responsibility, central government takes a more covert approach, by creating nodes in the policy network, stressing the bottom-up character of actors and their independence, and activating actors in the policy network to take care of communication about the policy issue. Consequently, the central government’s role becomes more unobtrusive and less observable, while government not necessarily exerts less control. To avoid overly ostentatious direct control, central government is hiding, so to speak, behind a carefully orchestrated policy network.

Our study reveals a salient meta-governance strategy which was not included in the analytical framework a priori. This points to our second conclusion: creating key nodes in a policy network is a meta-governance strategy in its own right. As a meta-governor, central government can increase the interconnectedness and interdependency between actors in a network by creating composite actors, coalition building and integrating newly created actors into vested actors. These nodes represent (mutually connected) sub-networks in which involved actors take on a role as meta-governor, as, for example, the VO-raad does within the ‘raising standards’ policy network, or the Coop and EF do within the ‘learning organization’ policy network. The latter indicates that the ideal types of meta-governance identified by Gjaltema, Biesbroek, and Termeer (2019), may be combined. For our cases reveal examples of the ideal type of ‘multilevel meta-governance’ that are embedded within the ideal type of ‘meta-governance of multiplicity’ we took as our starting point.

In distinguishing more direct from more indirect use of strategies, and by underscoring that not all actors are created equal, this study builds on the literature describing the concept of meta-governance as controlling at a distance through establishing, strengthening and mobilizing active and responsible actors within policy networks (Jessop 2002; Pierre and Peters 2000; Rhodes 2007). This notion ties into a last conclusion: policy networks are not created in a void. Policy networks are built on existing structures and coalitions of governmental, non-governmental and private organizations, by establishing new actors, and by positioning and bolstering vested actors. It confirms what is asserted in the literature on policy networks: policy networks are not a ‘given’ or emerge accidentally, they are deliberately reinforced, and partly created, by government around specific policy issues (Klijn and Koppenjan 2000, 2006; Marin and Mayntz 1991).

Limitations

Ours is a small-scale study, analysing the strategies applied by only one, albeit well-advanced, meta-governor in just two cases. The findings must therefore be interpreted more as building hypothesis rather than providing conclusive evidence. Future research on strategies applied by other meta-governors is required to determine whether, and if so how, such strategies may be context-specific.

Network design strategies are found to be particularly salient in enacting meta-governance. Although we took great care to identify the actors in the policy networks accurately, our analyses are based on retrospective data and are limited by the drawbacks that come with it. To study the processes resulting in the in- and exclusion of actors in a network in detail, future research might want to focus specifically on design strategies during the actual formation of policy networks.

Reflection

In closing, we reflect on a striking outcome of this study, namely the widely used meta-governance strategy of creating key nodes in a policy network. Our study shows that this strategy enables central government to promote interconnectedness and interdependence between actors in policy networks and to allow actors at different levels to take on a role as meta-governor within the policy networks. What the implications of this intertwining, layering or nesting within policy networks might be for meta-governors and those involved in policy implementation, is of particular interest for future research.

Our second reflection concerns the involvement of practitioners, professionals and administrators in policy networks, which emerged as a key discussion point during the member check of our study. Remarkably, our study reveals that school administrators, school leaders, teachers, students and parents only have a marginal, if any, position in the policy networks we examined. They are primarily treated as policy objects that need to be informed, equipped, enticed and directed in order to achieve the policy objectives. As a result, their practical, professional and experiential knowledge is not used, while it proves to be very relevant for political decision-making and policy development, for example, to detect policy issues and possible solutions, to inform educational policy with expert knowledge and specific expertise from the educational field, and to report on practical implications (Hooge 2013). And conversely, this equates to missing a crucial opportunity increase support for policy implementation among educational practitioners, professionals and administrators. Scholars in the field of policy implementation have long emphasized the importance of implementers’ influence on policy design and their involvement in decision-making, as this enables them to identify with policy and to develop the will and skill for policy-implementation (Mazmanian and Sabatier 1979; O’Toole 2000).

Meta-governance seems a promising new mode of policymaking in increasingly complex, fragmented and multi-layered societies, and offers an enticing lever for ensuring central government influence. Yet, the lack of involvement of practitioners, professionals and school administrators prompts a sobering perspective on meta-governance and policymaking through networks.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Nationaal Regieorgaan Onderwijsonderzoek [405-14-401].