European Governance in Adult Education: On the comparative advantage of joining working groups and networks

This article examines the working of complex intergovernmental policies that have brought about new opportunities and structures in European adult education since the 2008 global financial crisis. Drawing on political sociology, it restricts attention on the Renewed European Agenda for Adult Learning (2011), to examine its historical development, and how it bundles together various governance mechanisms, policy instruments, and social actors to govern the adult education policy domain through policy coordination. This points at regulatory politics as a distinctive quality of European governance in adult education. Then, through Social Network Analysis, it explores in depth one of its policy instrument (i.e., coordinated working groups/networks) and the form of network governance it creates. This analysis pinpoints at the comparative advantage of some organizations (i.e., the ministries of Latvia, Finland and Belgium), which partake in this form of network governance. This produces unpredictable contingency in EU policy coordination.


Introduction
In the European Union (EU), in the wake of the 2008 global financial crisis, complex intergovernmental policies have brought about new opportunities and structures in adult education. For instance, all member states are now committed to help adults with a low level of skills to increase their literacy, numeracy and digital skills and/or to progress towards an upper secondary qualification or equivalent (i.e., Upskilling Pathways initiative). At the same time, they are also committed to have their national budgets and reforms' priorities closely monitored by the institutions of the EU (i.e., European Semester). These new opportunities and structures also increased the challenges of European governance in the adult education policy domain, and specifically that of effective policy coordination.
A glance at the literature on governance increasingly points at different mechanisms, instruments and tools as key for the act of governing education among other policy domains (Dill, 2000;Dill & Beerkens, 2010;Jordan, Rüdiger, Wurzel, & Zito, 2005;Kassim & Le Galès, 2010;Erkkilä, 2016). All this comes against the backdrop of numerous actors intervening in policy-making, and a variety of disciplinary perspectives and approaches to identify and study the policy devices through which multi-actor governance occurs also through policy coordination, at both European and global levels.
Some point at networking, seminars, reviews, expert groups as 'soft governance' devices that lay at the core of the Europeanization of education (Lawn & Grek, 2012). But the apparatuses for multi-level governance and effective policy coordination within the EU hold some peculiarity that is worth attention. In fact, Europeanization points not only at the process of creating a European policy space (Ibid.) but most importantly to an all-encompassing process of 'domestic adaptation to European regional integration' (Graziano & Vink, 2006, p. 7), which occurs through regulatory politics and a 'joint decision mode' (Kohler-Koch & Rittberger, 2006). Therefore, researching European governance in adult education, and particularly policy coordination, requires attention to the EU's specific 'actorness'. In fact, the EU multi-level nature involves interdependence of governments representing different territorial levels, as well as interdependence between governments and non-governmental actors (Hooghe & Marks, 2001;Bache & Flinders, 2004;Zito, 2015). Moreover, EU governance tends by its very nature towards Europeanization, and this largely implies the transformative effect of the EU governance system on the political institutions, policies, and political processes of EU member states, and beyond European countries (Sabel & Zeitlin, 2010). Thus, we contribute to a growing area of interest in researching European politics and Europeanization through scrutinizing specific mechanisms or instruments, and by assessing the struggle between their legitimacy and effectiveness (Tholoniat, 2010;Walters & Haahr, 2005;Chatzopoulou, 2015).
In education, particularly the Open Method of Coordination, and policy 'instrumentation' more broadly (Lascoumes & Le Galès, 2007), have attracted a growing interest among scholars when analysing decision-making within the EU and its member states, and its contribution to Europeanization. Ravinet (2008), for instance, addressed the effects of a governance mechanism in the higher education policy domain (i.e., the 'follow-up' mechanism, or the process of re-enforcing continuity of cooperation through various working groups), and its diverse tools (i.e., the devices linked to the actions utilised by working groups and other actors) to show how its evolving structure played a significant and influential role for national adaptation.
While this furthers our understanding of European governance in education, including policy coordination, and its domestic adaptive effects, most studies concentrate on the analysis of single policy instruments (e.g. Open Method of Coordination), and seldom consider the adult education policy domain.
Against this backdrop, our main concern lays with the working of complex intergovernmental policies that have brought about new opportunities and structures in European adult education since the 2008 global financial crisis. Specifically, we are interested in the development, process and practice of policy coordination that these policies entail, and their possible effects for domestic adaptation.
In this paper, we restrict attention on the Renewed European Agenda for Adult Learning (Renewed Agenda hereof), endorsed by the European Council in 2011 (Council of the European Union [Council], 2011). We do so with a double aim. One is to illustrate how this intergovernmental policy bundles together various governance mechanisms, policy instruments, tools, and social actors to govern the adult education policy domain through policy coordination. The other is to examine policy coordination, and explore the forms of comparative advantage it produces for some organizations, and the countries they represent.
The paper is structured in three main sections. First, we introduce our conceptual framework, and explain how we conceptualise complex intergovernmental policies as policy mixes (Del Rio & Howlett, 2013). Then, we present the Renewed Agenda as a policy mix that is performing three authoritative functions (i.e., legal, epistemic and procedural) and which has put in motion a new process of instrumentation in the adult education policy domain. Such process is illustrated through a brief account of the historical development of the Renewed Agenda, and the governance mechanisms and policy instruments through which it works. Then we focus on one such policy instrument (i.e., coordinated working groups/networks) and, through Social Network Analysis, explore the form of network governance it creates. In so doing, we pinpoint at the comparative advantage of some organizations that partake in this form of network governance.

Conceptual clarifications
The emergence of policy coordination as a governance technique (Armstrong, 2010) brought into light the variety of mechanisms, instruments and tools as central conceptions enacting European governance, but also as the analytical instruments that allow examinations of the coordinating function of EU institutions. Based on a review of these concepts, how they have been signified, and to which end, by those engaged with public administration, political science, and particularly public policy and education governance, this section provides a conceptual background to the way we employ an instruments approach to examine the coordinating function of EU institutions in the adult education policy domain.
Several approaches in the literature deal with the way policy instruments and tools are understood and used for policy analysis. Some focus on the effect of instruments by analysing practical problems and practical knowledge in government organizations, which may help improving the quality of policy processes (Bruijn & Hufen, 1998). Others are interested in policy instruments and the role they play at policy formulation and implementation stages, particularly, the choice of instruments. Specifically, instrument choice has been increasingly analysed through a constitutivist lens (Linder & Peters, 1998), which calls for attention to the subjective meanings (symbolic, ethical, and so on) of policy instruments, and how these are interpreted and mediated through different values and perceptions of the actors involved in the policy process. Policy design studies (Del Rio & Howlett, 2013;Barton, Ring & Rusch, 2017) also point at existing conflicts or synergies between different policy instruments employed in the same bundle of more complex policy mixes, which involve multiple governments as well as multiple domains or policy goals, and are at the heart of European governance. In recent years, policy design studies have concentrated attention on the formulation of intelligent design of policy mixes, policy 'portfolios ', or 'bundles' (Del Rio & Howlett, 2013;Barton et al., 2017), with the scope of establishing optimality of complex policy mixes (Mandell 2008, Howlett and Rayner, 2007, Lanzalaco, 2011, institutions and networks (John 2011), and policy layering or layering of tools (Thelen, 2004;van der Heijden, 2011;Daugbjerg & Swinbank, 2016;Considine 2012). Methodologically, these studies take into account the types of tools, the policy objectives, but also institutional and behavioural contexts (Del Rio & Howlett, 2013).
Against this backdrop, and in light of the multilevel governance that characterises the EU, we borrow from Del Rio and Howlett (2013) the concept of 'policy mix'. A policy mix embeds 'horizontal' complexity -as each mix relates to different policy instruments and actors within a level of policy-making, as well as 'vertical' complexity -as each mix addresses a number of policy goals, domains and/or governments. In other words, the 'horizontal' dimension of a policy mix relates to a number of instruments (e.g. funding schemes) and relationships existing between them within a single level of policy-making (e.g. European). At the same time its 'vertical' dimension refers to the involvement of multiple goals (like economic growth, adult's up-skilling, etc.), policy domains (such as economy, labour, education, etc.) and governments (e.g. national, regional and local governments in EU's member states, as well as candidate and associate countries). By utilising such a perspective on the Renewed Agenda we reveal the extraordinary policy coordination challenges facing policy-makers and researchers.
The next step enabling a better understanding of this policy mix is based on Lascoumes and Le Galès' (2007) take on policy 'instrumentation' and their distinction at the 'levels of observation' between 'instruments', 'techniques' and 'tools'. Their contribution has two analytical merits. At epistemological level, it includes space for analysing the values, history and nature of instruments. Most importantly, however, at conceptual level, it distinguishes, and clarifies the relations, between policy 'instruments' and 'tools', as tools are the micro devises through which meanings (symbolic, ethical, and so on) are construed. It is on this ground that instruments can shape social practices.
Accordingly, we define the concept of 'governance mechanism' as a policy process aimed at reaching specific policy objective(s) that naturalizes these objectives and the effects it produces. Empirically, it focuses on power and interests, or the debates surrounding the creation and introduction of policy objectives, the ways they were modified, and their controversies. Unlike in our other studies (Milana & Klatt, 2019a, Milana, Klatt & Vatrella, forthcoming), this paper is not focused on investigating the details of such naturalization process, but the 'governance mechanism' perspective enabled us to map and analyse the complex soft forms of governance being used in EU policy development. Moreover, we define the concept of 'policy instrument' as the means used to reach policy outcome(s), in the sense of more or less stable frameworks that structure collective action.
In short, we believe productive to put different standpoints into dialogue. A functionalist standpoint (cf. Del Rio & Howlett, 2013, among others) helps in identifying complex policy mixes, and the relationships between instruments or tools and possible existing conflicts and synergies. In our case, it helped identifying and describing policy mechanisms and instruments utilised in the Renewed Agenda following our historical analysis of the conflicts and synergies in the three phases of its development. At the same time a social constructivist standpoint (cf. Risse, 2004, among others) enables identifying possible influence of network governance on individual member states. By analysing various characteristics of coordinated working groups under the Renewed Agenda, and drawing inferences between the network nodes through Social Network Analysis, we attempted to identify the 'two-way constitutivness' of social environments and individual member states, and the impact of network governance on domestic adaptation and innetwork influence.
Drawing on the above heuristic model that conceptualises and clarifies the relations between policy mixes, governance mechanisms, policy instruments and tools we identified a few policy mixes that, agreed among EU institutions and member states, may connect to and influence national and local governments' decisions on adult education within the EU. One such policy mixes is the Renewed Agenda that, in the next section, we scrutinize from an historical perspective, before examining the mechanisms and instruments through which it works.

Exposing the Renewed Agenda
We differentiate between three periods in the history of the Renewed Agenda to appreciate its formation as a policy mix on its own rights, existing conflicts or synergies and how these changed over time, and its mode of working. 1 The first period, signposted by the establishment of 1996 as the European year of lifelong learning and the Resolution on lifelong learning by the Council of the EU (June 2002), is a 'pre-foundation stage' in which the EU sets the ground for adult education to emerge as a policy domain distinct from education and training. At this stage, the adult education dimension of lifelong learning is teased out in dialogue between the institutions of the EU, which bolsters the ties between European education and training and employment policies. This creates the ground for adult and further education to be seen as an intergovernmental and multi-sectorial policy domain with multiple goals. Therefrom, EU institutions concentrated on existing statistical data gaps at the microlevel (learner-centred), and stronger knowledge exchanges and collaboration across member states, and with other international organizations with an interest in adult and further education.  (Council, 2008), is the foundation stage in which adult education became a clearly defined policy domain. At this stage, a policy mix governing adult education within the EU starts to take its current shape, upon initiative of the European Commision.
On these precedents, the outbreak of the global financial crisis also impinged on the tuning of the Renewed Agenda. Specifically, two elements of Europe 2020, though indirectly, bear higher significance for the adult education policy domain: a European benchmark on tertiary education for young adults (i.e., at least 40% of the younger generation should have a tertiary degree), and a flagship initiative linking skills to better job prospects (i.e., An Agenda for new skills and jobs).
In 2011 the Council of the EU approved a Resolution on the Renewed Agenda. Council Resolutions have no legal effect on EU member states, as they are non-binding documents, but express political positions on a specific topic, and set out future work within a particular policy domain that is not (or not entirely) of EU exclusive competency. Accordingly, they may have different scopes that span from inviting member states or other EU institutions to take action in a particular area to coordinating member states' actions by setting objectives, and proposing assessments and monitoring procedures.
Previous to the Renewed Agenda, the Directorate General for Education and Culture of the European Commission had put forward an Agenda for adult learning in one of its 2006 Communications, and a corresponding Action plan was proposed in 2007 to the Council of the EU, the European Parliament, the European Economic and Social Committee, and the Committee of the Regions (EC, 2007). This led in January 2008 to its adoption by the European Parliament through to a Resolution on adult learning (European Parliament [EP], 2008). The Renewed Agenda built on these previous normative steps, yet tailed the global financial crisis that had made its effects felt in Europe too, when member states from the Eurozone became unable to repay or refinance their government debt. It was to contrast this and related social consequences that in 2010 the EC reconsidered the union's growth strategy in Europe 2020. Within this scenario the Renewed Agenda recognises that to face both the short and long-term consequences of the economic crisis, there is a need for adults regularly to enhance their personal and professional skills and competences… [but] adult learning is currently the weakest link in developing national lifelong-learning systems… [and] Implementing the Action Plan [for adult learning] has also highlighted the difficulty of adequately monitoring the adult-learning sector, due to a lack of sufficient statistical data and evaluation of policy measures. (Council, 2011 p. C372/2) Accordingly, it sets new priorities in this policy domain that are 'to be seen in the context of a longer-term vision for adult learning which -in the period up to 2020 -will endeavour to raise the sector's profile' (Ibid. p. C372/3). This vision stresses: enhancing the possibilities for adults to engage in learning activities; developing new approaches based on learning outcomes and lifelong learning guidance systems; increasing awareness among employers of adult learning's benefits for productivity; encouraging higher education institutions to embrace non-traditional students; promoting learning opportunities in support of seniors' active, autonomous and healthy aging; enhancing the involvement of civil society, social partners and local authorities on the basis of shared responsibility; and promoting adult learning as a means to increase solidarity between age generations and cultures.
Short-term priorities for 2012-2014 invited member states to better liaise with ministries and other stakeholders; use lifelong-learning tools agreed at EU level; use Grundtvig, Leonardo da Vinci and the Structural Fund to co-finance activities; use the Open Method of Coordination to promote mutual learning; and designate a national coordinator to facilitate cooperation with other member states and the EC in implementing the Renewed Agenda.
Moreover, the EC was invited to ensure complementarity and coherence between policy initiatives; establish close liaison with the national coordinators designated by the member states; enable the sharing of information through peer-learning activities and reviews, conferences, workshop etc.; commission studies and reinforce the capacity of existing research structures; pursue and intensify collaboration with other international organizations, and particularly the Organization for Economic Cooperation and Development (OECD) to exploit the results of the Programme for the International Assessment of Adult Competences (PIAAC), but also the United Nations and its Educational, Scientific and Cultural Organization (UNESCO), and the Council of Europe; harnessing available EU funds to support the Renewed Agenda; and report on its implementation as part of the joint progress report of the strategic framework for European cooperation in education and training (ET2020).
Although not binding, we argue that the Renewed Agenda constitutes a policy mix that performs substantive authoritative functions at legal, epistemic and procedural level, which ease European governance in the adult education policy domain.
Legally, although Resolutions are non-binding documents like Communications, according to EU Law the latter set out the EC's own thinking on a particular matter, whereas the former are legal instruments that encourage all those addressed to act in particular ways, hence enable EU institutions to establish non-binding rules for member states. So, the Renewed Agenda elevated political authority on adult education from the EC (accountable to appointed impartial and independent commissioners) to the Council of the EU (accountable to national governments) (Klatt, 2014). A precedent had been established in 2008, when the previous Action plan on adult learning had gained legitimacy through the EP's Resolution on adult learning (EP, 2008).
Epistemically, the Renewed Agenda's short-term priorities and longer-term vision legitimate an 'instrumental epistemology' in the adult education policy domain that, as Bagnall and Hodge (2018) argue, has come to be favoured in contrast to alternative, competing ones (i.e., disciplinary, constructivist, emancipatory) in the contemporary cultural context.
Procedurally, the Renewed Agenda sets the objectives of member states' action (e.g. liaise ministries and other stakeholders, co-finance adult learning activities, promote mutual learning) and of EC's action (e.g. ensure complementarity and coherence between policy initiatives, establish close liaison with member states, enable knowledge sharing, reinforce research capacity of existing structures, pursue and/or intensify collaboration with other international organizations). But it also prescribes the policy instruments through which these shall be achieved. Finally, it interlocks the short-term priorities in adult education, and related policy instruments, to ET2020, a different policy mix.
In short, the Renewed Agenda, through its legal, epistemic and procedural functions, has put in motion a complex process of instrumentation in the adult education policy domain, which frames 'adult learning' as the process leading to the acquisition of skills by adult citizens, and which, in turn, increases the pool of skills available in a country, and, by extension, within the European region as a whole, an undivided territory, in its racing for global competition.

Governance mechanisms and policy instruments
Our examination of the mode of work of the Renewed Agenda (Milana & Klatt, 2019a, Milana, Klatt & Vatrella, forthcoming) points at the following as its core governance mechanisms, 2 operating under the principles of the Open Method of Coordination: • Standard-setting: It involves normative actions and setting common goals (including benchmarking and standardization) that concur towards the establishment of a single, European model in the area of adult education and learning, to which all member states should conform. • Capacity-building: It promotes 'good' or 'best' practices that help orienting the practical implementation of policy solution in the area of adult education learning to what framed as common European problems, by EU institutions, national governments and other stakeholders. • Elite learning: It instigates changes in the value system of national actors through peer learning, peer counselling etc.
• Financial redistribution: It implies that EU's wealth is shared out between member states as a deliberate effect of joint decisions that include conditionality, and are used in support of reforms and activities in the area of adult education and learning.
But several policy instruments contribute to the working of these mechanisms; those surfacing in the analysis include: • Coordinated working groups/networks: Groups established and coordinated by the EC, whose members, appointed by member states' governments or the EC, represents different elite positions (i.e., governmental agencies, other stakeholders, experts), and are assembled, over a period of time, to work on important policy issues in the area of adult learning. • Mutual-and peer-learning arrangements: Occasions for representatives of member states, and EC's staff that support this activity, to identify and learn about initiatives and practices in place in different member states (and beyond) in the area of adult learning. • Data generation: The gathering of quantitative and/or qualitative data, the method used to generate data from different sources, and the procedure through which data reaches a database or otherwise organized collection of data. • Benchmarks: Accepted standards at European level, at times negotiated and agreed among Heads of states and governments, by which member states' performances in the area of adult education and learning can be measured, compared, and thus their level of quality judged. • Funding schemes: Plans or arrangements designed by EU institutions to encourage governments, organizations or people to attain a particular objective or to put an idea into effect by providing money to finance an activity, a program, or a project entirely or in part.
In the next section, we focus attention on one among the instruments depicted above, working groups and networks coordinated by the EC, as these represent also the preferable 'working method' identified by the EC within the Open Method of Coordination. We will then examine them through Social Network Analysis so as to reveal the complexity of interest representation, and how it contributes to Europeanization processes. individuals representing a common interest / policy orientation (B); organisations (C); local, regional or national member states' authorities (D); or other public entities (E).

Zooming on coordinated working groups/networks
Unless there are overriding priorities or emergency conditions, all appointed members are selected through public calls for applications, with the exception of public authorities (i.e., D and E). Selected features of the working groups/networks under consideration here are presented in Table 1. All working groups/networks were tasked to assist the EC with the implementation of existing EU legislation, programmes and policies and to coordinate with member states, through views' exchange. Only the Working Group on the Implementation of the Action Plan on Adult Learning, active at foundation stage, was tasked also to assist in the preparation of legislative proposals and policy initiatives.
At consolidation stage, however, changes in EU education governance impinged on the adult education domain. An internal restructuring of the EC moved its responsibility from the Directorate General for Education and Culture to the Directorate General for Employment, Social Affairs and Inclusion since 2013, so the coordination of working groups/networks in this domain shifted accordingly.
Moreover, due to the 2009 agreement ET2020, and its tuning to Europe 2020, the work of these groups/networks slowly altered too, as to better fit the principles of the Open Method of Coordination. Made explicit in the mission statement of Working Group on Adult Learning, such adaptation process is also evidenced in its stress on mutual learning among member states, assistance to member states in coping with country specific 'recommendations' by the EU institutions, and 'concrete and useable outputs' as a result of the group's activity.
Operating under a looser interpretation of the Open Method of Coordination's principles, both the Thematic Working Groups on Quality Assurance in Adult Learning, and on Financing Adult Learning, had a thematic focus (quality vs. finance), and higher interest in research gaps. The Thematic Working Group on Quality Assurance in Adult Learning explored synergies to strengthen the policy links between EU policy development on quality in vocational education and training, higher education and adult learning through three subgroups on indicators, accreditation / governance, and staff competences. In the meantime, the Thematic Working Group on Financing Adult Learning explored existing good practices to produce policy recommendations to assist member states in improving the efficiency and coherence of adult learning financing. Two subgroups focused, respectively, on funding adult learning for re-skilling and up-skilling to support innovation and growth, and funding adult learning for social inclusion and active citizenship. Both working groups appointed also individuals in their personal capacity.
By contrast, the Working Group on Adult Learning, in line with its tighter governance function, did not appoint any individual in his/her personal capacity, and instead increased representation of other public entities, and particularly of candidate countries (now including Albania, Montenegro, Serbia and Turkey). Further, among EU agencies, it replaced Eurydice, a network of institutions that facilitate sharing of information on national education systems, with the European Training Foundation, an agency that supports education, training and labour market reforms in transition and developing countries.
Yet, silent members of all working groups/networks are consultancy firms that, having signed framework contracts with the EC, provide their services as facilitators and rapporteurs for the groups/networks' activities.

2008-2010
Provide the EC with: -Policy advice and assistance in implementing, and following up, the actions set out in the Action plan (2008-2010); -Examples of good practices for dissemination and discussion of proposed actions at EU level, to impact and strengthen adult learning participation at national and regional levels. Having identified the main characteristics of each of the coordinated working groups/networks at work in the adult education policy domain, and considered their contrasting peculiarities, next we focus on the form of network governance this policy instrument creates.

A Social Network Analysis of network governance in European adult education
Coordinated working groups are a significant policy instrument used in the European governance of the adult education domain. Governance refers to an organizational form of political as well as government action that is open to the involvement of private and civil society organizations (Rhodes, 1996(Rhodes, , 1997(Rhodes, , 2000Mayntz, 1999). Accordingly, the EU (2001) itself identifies the general principle of 'participation' as an indicator of 'good' governance. Yet, among the different meanings that governance entails as a peculiar form of organization and coordination is that of a self-governing network: in this sense, governance emerges from self-organization phenomena set up by interdependent actors (Rhodes, 1996(Rhodes, , 1997(Rhodes, , 2000. Therefore, we draw on Social Network Analysis (Wasserman & Faust, 1994) to study the interactions of local, national, European and global policy actors within and across these networks. The horizontality between the actors, or the possibility that they are coordinated between peers during a decision-making process, produces self-organization systems starting from the relational models that the actors produce. In line with this definition Jones, Hesterly and Borgatti (1997, p. 914) speak of 'network governance', which 'involves a select, persistent, and structured set of autonomous firms (as well as nonprofit agencies) engaged in creating products or services based on implicit and open-ended contracts to adapt to environmental contingencies and to coordinate and safeguard exchanges. These contracts are socially -not legallybinding.' The organizations partaking in the coordinated working groups under consideration are heterogeneous so we draw a parallel with the above definition of network governance, as these organizations also engage in producing products and services of some sorts (i.e., new norms and adult education provisions) based on not legally-binding social contracts. Accordingly, to describe the form of network governance in European adult education in which these organizations (as independent actors) engage, we performed a Social Network Analysis starting from a 2-mode matrix. This was generated from the 5 coordinated working groups (or events), and includes a total of 98 actors (or nodes) (i.e., organizations representing national ministries, third sector associations, EU agencies, etc.) -see Annex, which includes the number of events in which each actor is involved. For each actor with a national horizon of action, we added a two-letter country code in accordance with the ISO 3166-1 alpha 2 standard. The actor data were collected from the official European Commission's Register of Commission Expert Groups and Other Similar Entities (EC, 2018). The register provides the list of all appointed members in each group, according to their membership type (see Table 1). Appointed members include representatives from member states, candidate countries, European Free Trade Association countries, and relevant EU bodies or agencies (e.g., CEDEFOP, European Training Foundation, Eurydice), education and training associations (e.g. European Association for the Education of Adults), and European social partners (e.g., European Trade Union Confederation) (for a full list see Annex). Starting from this 2-mode matrix, we produced a 1-mode matrix, symmetric and binary, for the nodes, through this matrix we obtained a simple graph of contacts among actors (Figure 1). . We used Ucinet 6 software (Borgatti, Everett & Freeman, 2002) to perform the Social Network Analysis, and the NetDraw software (Borgatti, 2002) to obtain the graph. The 1-mode matrix contains the information about the presence of a contact between any single couple of actors: if these actors were in a same working group at least we can establish that they had an occasion for interacting and cooperating. This strategy presents at least two limitations that we will take into account.
The first limitation concerns the validity of our indicators of tie: we can suppose that two actors interact but we don't know if they really did, moreover we know that two actors do not share any working group membership but we do not know if they interact in other circumstances. At the same time, it must be noted that the occasions of connection prompted by the 5 working groups under consideration here are institutionally directed to support the European governance of adult education and therefore the ties that were activated in these circumstances are semantically well connected with the object of our research.
The second limitation concerns the reliability of the relational data we identified: the activity periods of the working groups are in fact partly different (see Table 1), but the analysis of an inter-organizational network (i.e., of a network of collective subjects), can justify the need of a longer time to be taken into account, and the European governance of adult education here analysed is well considered starting from all the 5 working groups.
This clarified, in the form of network governance under consideration the level of cohesion, which represents the density or 'proportion of possible lines that are actually present in the graph' (Wasserman & Faust, 1994, p. 101) (Figure 1), is equal to 63.20%.

Actors' connectivity
For each of the actors involved in this form of network governance, Table 2 reports its level of centrality in terms of degree (of a node) or 'the number of lines that are incident with it' (Wassermann & Faust, 1994, p. 100), and its normalized measure (with a range of 0-1). This measure is a proxy for an actor's connectivity / level of integration within the form of network governance under consideration. Perhaps expectedly the organizations with the highest level of integration (nDegree 0.94-1) are the administrative divisions of governments that hold responsibility for education at national level, with two country exceptions. One is Belgium, where the Flemish Department of Education and Training is among the organizations with the highest level of integration (nDegree: 1), yet its counterpart, the Ministry of the French Community of Belgium, is among those with the weakest connectivity (nDegree: 0.29) within the network governance under consideration. Another is Portugal, where the National Agency for Qualification and VET showcases a highest level of integration (nDegree: 0.94) when compared with the national Ministry of Education and Science (nDegree: 0.51). But among the ministries of education only three from Northern and Eastern Europe share the highest possible level of integration (nDegree: 1): the Ministry of Education and Science of Latvia, the Ministry of Education and Culture of Finland, and the Flemish Department for Education and Training of Belgium. Interestingly, they represent one among the oldest members of the EU (Belgium), one among those that joined the EU at the time of its 1990s enlargement (Finland), and one among the new members that joined the EU in the biggest enlargement of 2004 (Latvia).
Alongside national ministries, also a few trade and worker unions active at European level (European Association of Craft, Small and Medium-sized Enterprises, UEAPME; European Trade Union Confederation, European Trade Union Confederation) showcase a relatively high level of integration (nDegree: 0.85). Remarkably, however, unions representing workers in the education sector (European Trade Union Committee for Education; European Federation of Education Employers) have a weaker level of integration within the network under consideration (nDegree: 0.57-0.70), when compared to their generalist counterpart (i.e., European Trade Union Confederation) or to trade unions like UEAPME or BUSINESSEUROPE (nDegree: 0.78). At the same time, the social partner representing small and medium enterprises in Europe (UEAPME) is better integrated than the its counterpart representing all-sized enterprises (BUSINNESSEUROPE).
Among EU agencies specialised in education, CEDEFOP (nDegree: 0.85) has the higher level of integration, followed by the European Training Foundation (nDegree: 0.78), whereas Eurydice has a rather weakest connectivity (nDegree: 0.59).
Actors other than ministries yet from the public sector that shows a medium level of integration include organizations like the Adult Education Foundation (Liechtenstein) (nDegree: 0.76) and the Centre for Vocational Education (Montenegro) (nDegree: 0.73) from non-EU member states, but which are either member of the European Economic Area (Liechtenstein) or negotiating access to the EU (Montenegro).
Finally, when we compare the level of integration of the two Directorates General of the EC that are involved in the form of network governance under consideration, the Directorate General for Employment, Social Affairs and Inclusion has a higher level of integration (nDegree: 0.73); this despite the fact that the Directorate General for Education and Culture (nDegree: 0.59) coordinated a higher number of working groups/networks (cf. Table 1).

Actors' brokerage capacity
Furthering our analysis, for each of the actors involved, Table 3 shows its level of centrality in terms of Betweenness, and its normalized measure (nBetweenness, expressed as a percentage) (Freeman, 1979). 'The important idea here is that an actor is central if it lies between other actors on their geodesics, implying that to have a large 'betweenness' centrality, the actor must be between many of the actors via their geodesics' (Wassermann & Faust, 1994, p. 189). Hence, nBetweenness is a proxy for an actor's brokerage capacity within the form of network governance under consideration. The two measures of an actor's level of centrality (cf. Table 2 and Table 3) showcase some similarities. Overall, most actors show a high level of integration (nDegree) as well as a high level of its brokerage capacity (nBetweenness). However, a number of actors point at interesting differences in these measures, which are worth attention. Strikingly the Ministry of National Education of Turkey has a relatively high brokerage capacity (nBetweenness: 1.38) when compared to the ministries of education from a group of countries (Iceland, Netherlands, Lithuania, Bulgaria, Croatia, and Portugal) with a higher level of connectivity (nDegree: 0.94) than Turkey (nDegree: 0.87). It shall be noted that although accession negotiations have stalled, Turkey applied to accede to the EU since 1987, and represents today one of the main partners of the EU in the Middle East. Moreover, since the mid-1990s, the EU-Turkey Customs Union regulates free trade in the area. Turkey connectivity may be driven by its weak educational attainment of the adult population (Eurydice, 2019) and therefore its high aspirations for coordination in this particular field.
Noticeably also the Adult Education Foundation of Liechtenstein has a high brokerage capacity (nBetweenness: 0.75) but a medium level of integration when compared with other actors. In fact, while the level of integration is very close to that of the Ministry of Education and Research of Sweden, a EU member states, the European Training Foundation, a EU specialised agency, and BUSINESSEUROPE, a trade union active a European level (nDegree: 0.78), its brokerage capacity is markedly higher when compared to that of these organizations (nBetweenness: 0.38). Not a EU member state, likewise Turkey, the EU-Liechtenstein relations are shaped by the country's participation to the European Economic Area, and its adhesion to the Schengen Area.
Lastly, there are also a number of organizations that, despite their relative level of integration (nDegree: 0.57), have no brokerage capacity (nBetweenness: 0). These include the ministries of education from a number of candidate countries (Albania, Macedonia, Montenegro and Serbia), the State Secretariat for Education, Research and Innovation of Switzerland, which is not a member of either the EU or the European Economic Area, but also the ministries of two EU founding members (Italy, France). Interestingly, for France, also the Ministry of Employment, Vocational Training and Social Dialogue falls into this group.
Remarkably, the two organizations that represent educational providers at European level (European Federation of Education Employers, European Association of Vocational Education and Training) belong to this group as well. So do a few national institutions other than ministries, but that are under ministerial supervision (State Education Quality Service of Latvia, ex Institute for the Development of Professional Training of Workers, now National Institute for Public Policy Analysis, of Italy, SOLAS of Ireland). Finally, this group includes also the University of Belgrade and the Adult Education Action of Serbia, and the Association for Research and Media in Education of Austria.

Actors' maximal cohesiveness
Advancing our inquiry, Table 4 illustrates the results of the cliques analysis. 'A clique in a graph is a maximal complete subgraph of three or more nodes' (Wasserman & Faust, 1994, p. 254). In other words, a clique here represents a subgroup of organizations where everyone has an unmediated connection to all the others within the same subgroup. A clique is thought to generate consensus among its members else it may fall apart. Our analysis allowed the identification of 8 cliques.
Remarkable the Ministry of Education and Science of Latvia (node n. 4), the Ministry of Education and Culture of Finland (n. 10), and the Flemish Department for Education and Training of Belgium (n. 25) are part, together, of all 8 cliques. The same triad of organizations stick out in previous analysis as having the highest level of integration (cf. Table 2) and of brokerage capacity (Table 3) within the form of network governance under consideration. So, their maximal cohesiveness in all 8 cliques points at a circumstance that can certainly encourage a mutual coordination among these actors. In other words, they are not only involved in all 5 working groups coordinated by the EC under the Renewed Agenda, but they are also involved in all cliques that, together, engage all other 95 actors. This evidence highlights the role of strong coordination played by these 3 actors alone; hence they result as the most central actors in the form of network governance of European adult education.

Discussion
Comprehending the EU, including the role played by the European Commission, in its relations to the member states, and how this affects policy developments in the adult domain is at the heart of a strand of adult education policy research (cf. Milana & Holford, 2014, among others). At the same time, some scholars have paid attention to the working of commissions and task forces instituted at national level within members states (Milana & Rasmussen, 2018). Nonetheless, adult education scholars have paid no attention thus far to the Commission's coordinated expert groups, which are an essential instrument of multilevel governance, and a linchpin in the European policy coordination system.
Our empirical evidence on the coordinated working groups/networks connected to the Renewed Agenda shows how each expert group member is embedded in a relational network. To understand these relationships, social scientists focus on structural location within a network as a source of potential power (Hafner-Burton & Montgomery, 2010; Hanneman & Riddle, 2005 among others). But we are also interested in the mutualconstitutivness of a social network and its individual members, hence in the potential power within the network, and of a network to influence member states' domestic adaptation.
In terms of actors' centrality, in the form of network governance we examined, expectedly the organizations with the highest level of integration are the administrative divisions of governments that hold responsibility for education at national level. This is not surprising as all the member states are encouraged to contribute to the work of coordinated working groups/networks established under the Renewed Agenda. Furthermore, the actors expected to be central in a network are usually those with higher organisational resources, and those who have the incentives to be active (Maggetti, 2014). In this respect it is interesting to observe, however, that countries such as Italy or France, with the largest assumed administrative capacity, due to their size, their positional power within the EU, but also to the high levels of adults without upper secondary education in these countries, are not that well connected. This may illustrate that they have less aspirations to play significant roles within the adult education policy field. It may also be assumed that these countries may be less engaged due to the soft power of this policy instrument, which in turn provides more leadership opportunities (agenda setting, peer learning) to less 'powerful' member states (e.g. Latvia). Moreover, we acknowledge that adult education has general a lower status in several (if not all) member states, when compared to primary, secondary or higher education, and responsibility in this policy domain is often distributed across ministries. This may imply, for instance, that those asked to represent local, regional or national member states' authorities in these working groups/networks may well be qualified in adult education and hold strong ties and relations within this form of network governance, but may have limited access to organisational resources or incentives to be equally active at country level. This is an aspect that have been raised, for instance, by both working groups/networks' coordinators and individual members that we have had the opportunity to interview or hold informal conversations with, while we progress with our study. So, while our analysis showcases the complexity of Europeanization processes within coordinated working groups/networks, we acknowledge that more research is needed to deepen our understanding of such a network centrality positions and their effects for domestic adaptation in the adult education policy domain.
Our analysis also points at noticeable differences between the connectivity of each of the five official groups coordinated by the EC, and the subgroups emerging from the form of network governance these produce (i.e., the cliques). Such distinction produces unpredictable contingency in policy coordination.
When we consider the official groups coordinated by the EC, the two Directorates General for Education and Culture, and for Employment, Social Affairs and Inclusion, are the formal brokers; although belonging to the same organization they are not connected to each other in the form of network governance of European adult education emerging from these groups. Once again, we know from interviews and informal conversations that they interact in a number of other circumstances. Despite this, when we consider the emerging subgroups (i.e., the cliques), three network brokers emerge as each is connected to the other two. These are interesting findings as such divisions may illustrate how actors may be influenced or behave (Hanneman & Riddle, 2005). If the same actors are connected to different subgroups then the possibility of information diffusion grows. The information may spread across different subgroups, and across entire networks. It seems that such ability rests within the three above-mentioned ministries from Latvia, Finland and Belgium. Further qualitative analysis is needed, however, to inquiring the nature of these relations, and particularly the relations each formal broker holds with network brokers, and peerrelations between network brokers.

Concluding remarks
In this paper, we examined the working of complex intergovernmental policies affecting European adult education by concentrating attention on the Renewed Agenda.
First, by depicting its main characteristics, and stages of development, we clarified how this policy mix performs a legal, an epistemic, and a procedural authoritative function to govern European adult education. Then, we identified the governance mechanisms (i.e., standard-setting, capacity-building, elite learning, and financial redistribution) and policy instruments (i.e., coordinated working groups/networks, mutual-and peer-learning arrangements, data generation, benchmarks and funding schemes) at work under this policy mix. This points at regulatory politics as one the distinctive qualities of European governance in the adult education policy domain.
Secondly, we deepen our analysis of a specific policy instrument, namely coordinated working groups/networks, as it can be used in direct linkage to decisionmaking situations or as a space for monitoring the environment, gathering information and socialising. These working groups create a space for Europeanization and national adaptation. But as our analysis shows, it is also important to look at the sources and distribution of power within different forms of network governance, which may lay with actors who are highly connected and have high brokerage connectivity. At the same time, the working groups/networks considered here are linked to several governing mechanisms such as standard-setting, capacity building and elite learning through which they can successfully promote the domestic adoption of soft rules that originate from the network itself. Therefore, it is also possible that high level of centrality in a form of network governance may contribute to influencing other network members; hence actively contribute to the 'joint decision mode' through which multi-level governance works within the EU. Yet, the effects for domestic adaptation are harder to assess, as these are also dependent on organizational backing these members have in their own country. So, this contributes to more efficient European policy coordination where EU policies are not prescribed but 'formed' within specific policy instruments such as working groups/networks. Yet whether such more efficient European policy coordination may in turn result in a higher consistency of implementation and practice within the adult education policy domain across member states (which is the main aim of an effective EU policy coordination) remains questionable.
Finally, we recognise that the study would benefit from data on individual rather than institutional level -understanding how an individual is embedded in the structure of groups within a net may lead to some assumptions about their attributes and behaviour. While such data was not publicly available for all working groups / networks under consideration in this paper at the time of carrying out the analysis herein presented, it is in this direction that we are moving our research forward.