TEORIJA IN PRAKSA let. 56, 3/2019 853 Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA* POLICY NETWORKS: LOOKING FOR THE DECISIONAL CENTRE IN EU POLICYMAKING** Abstract. Responding to the call for cooperation between the policy network and governance literature, we apply social network analysis (SNA) to a study of European multi-level policymaking using empirical data gathered as part of the INTEREURO project. We focus on the stud- ied network’s characteristics and, building on hypoth- eses developed in the policy network and governance literature, judge the potential capacity to coordinate EU networks. Based on our analysis, we redefine the deci- sional centre of EU networks and argue that coordina- tion capacity varies among different policy fields. Keywords: interest groups, social network analysis, European Union Introduction During the last major globalisation wave of policy coordination, politics ever more strongly became a matter of interlinked actors and activities tak- ing place beyond nation-states. Indeed, it is networks rather than formal institutions that have been increasingly taking over governance. Global public policy networks are on the rise (Reinicke, 1999–2000). Like at the global level (Reinicke, 1997), also in framework of the regional EU political system (Scott, 2009) questions were raised concerning whether one can talk about supranational policy as governing without government. The policy network literature shows that policy networks can explain the policy process and results of policymaking at the national level (see e.g. Howlett, 2002) and EU level (see e.g. Daugbjerg, 1999; Peterson, 1995; Peters, 1998). In particular, authors from governance theory have encour- aged policy network analysts to contribute to network governance in future ARTICLES * Danica Fink-Hafner, PhD, Professor, Faculty of Social Sciences, University of Ljubljana, Slovenia; Meta Novak, PhD, Assistant Professor, Faculty of Social Sciences, University of Ljubljana, Slovenia; Luka Kronegger, PhD, Assistant Professor, Faculty of Social Sciences, University of Ljubljana, Slovenia; Patrycja Rozbicka, PhD, Lecturer in Politics and International Relations, Aston University, Birmingham, United Kingdom. ** The authors acknowledge the financial support from the Slovenian Research Agency (research core funding No. P5-0136), German Science Foundation (grant no. EI461/6-1) and the European Science Foundation (10-ECRP-008). Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 854 research (Torfing, 2016; Sørensen and Torfing, 2005a). Such research under- takings are also crucial for studying EU policy networks beyond the simple distinction of national, sub- and supra-national levels of government. Based on that literature, we start from the thesis that the EU policy networks are NOT built based on national borders nor are constituted by the same level actors. More precisely, we view the EU policy process as a ‘marble cake’ of multi-level and multi-national networks. We respond to a call to integrate governance theory and policy net- work studies and move beyond identifying the EU as merely complex policy networks. Our study’s goal is to judge the potential of EU networks’ coordination capacity. Specifically, we refer to two competing hypotheses concerning the centre of decision-making within EU policy networks. The interest intermediation theory proposes it is impossible to locate the centre of decision-making (e.g. Heclo, 1978; Hanf and Scharpf, 1977). In contrast, the literature on lobbying in the EU political system contends the central role is held by the European Commission (e.g. Richardson, 1994). However, and that is when policy network literature is able to help, when considering Olsen’s (2002: 593–4) research on institutional change, it is more reasonable to expect there is not just one principal actor in the centre of polycentric, multilevel policy networks. The notion of a policy network centre has yet to be sufficiently clarified and we seek to help close this gap. Further, based on Peters (1998) and van Waarden (1992), both with a background in the study of policy networks, we suggest the coordination capacity of EU networks depends on the stability of relations within EU networks/network sections, the variety of types of actors active on the same kinds of EU policies, the inclusion of actors from different countries and which level they stem from. In line with Rhodes (1997), who highlighted both the EU’s extraordinary differentiation and the varying relationships found between interest groups and government in certain policy areas, we expect that the potential of EU networks’ coordination capacity varies among EU policy processes. We reveal EU policy networks as constructed during the preparation of national positions on a selection of the 20 most salient directive propos- als tabled between 2008 and 2010 (Table A in the Appendix). Data were obtained from national policy officials as part of the INTEREURO Multi-level Governance Module (Beyers et al., 2014; INTEREURO, 2014). Our analy- sis focuses on the actors’ positioning with respect to particular issues in those proposals recognised at the European and national level while the Commission which prepares EU policy proposals is regarded as the centre (it is excluded from the data analysis to allow the relations among stakehold- ers to be revealed). We conduct the study using Social Network Analysis (SNA), which helps us better understand the nature of actors’ relations in multi-level policy circumstances. Using concepts and terms from the SNA Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 855 like closeness and betweenness centrality, we operationalise the policy actors’ complex multi-level relations, and treat SNA as an analytical tool. In the next section, we outline the theoretical framework for the analysis, which is followed by the methodology section. Based on our empirical anal- ysis, the EU policy networks under study are described and hypotheses are commented upon. In the concluding section, we summarise the strengths and limits of the analysis while proposing further research avenues. Theoretical approach In this article, we link several approaches to the study coordination capacity of networks and policy networks in general (Table 1). While pol- icy analysis primarily focuses on the meso level of politics (that is, public policymaking), the governance literature is increasingly looking at ways to include the meso level into the macro one. Table 1: APPROACHES TO STUDYING POLICY NETWORKS Approach Authors Key characteristic Level of analysis Aspects studied Broader framework Policy analysis Van Waarden (1992) Policy network (PN) concept; analytical tool based on the actor approach Micro, meso PN described with the number and characteristics of actors, PN functions, centrality… Description of policy networks Studying the impact of (changes in) PN characteris- tics on policy outcomes Scharpf (1993) Policy network as a specific social structure Meso PN as a form of organisation Description of policy networks Governance literature Börzel (1997) Sørensen and Torfing (2005) Torfing (2016) Networks as a form of gover- nance Macro Policy networks as theory of governance Networks of networks Studying governance and the governance of gover- nance (meta-governance) Social Network Analysis For an overview, see Wasserman and Faust (1994) and Carrington et al., eds. (2005) An ana- lytical approach to studying social networks Adaptive (micro, meso, macro) Relationships among social entities; characteristics of social networks based on mathematically developed analytical tools and software An ever-growing range of SNA applications spilling over to various academic disciplines and subjects of research Source: Authors’ own analysis. Coordination problems Contemporary governments’ efforts that seem to exacerbate their inher- ent coordination problems have led to various authors hypothesising dif- ferent solutions. Unlike other authors who focused on hierarchical coordi- nation (Kochen and Deutsch, 1980; Davis, 1995), the market as a steering mechanism (Marin, 1990), and “new institutionalism” (March and Olsen, Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 856 1989), Peters (1998) proposed concentrating on networks of loosely linked actors and the achieved degree of coordination between them along with assessing the causal factors supporting that. He argued that fully analysing and comprehending the decisional capacity of government requires think- ing about the interactions among not just individual organisations, but also about how ‘networks’ of organisations interact. The proposed network per- spective on the coordination problem had the advantage of allowing the inclusion of non-state actors and involved more negotiation and mediation patterns than in more traditional models. After Scharpf (1997), Peters (1998) argued the continuous interaction of a network’s members and their shar- ing of at least some values may “generate sufficient trust to permit more effective problem solving and positive-sum solutions” (1998: 299). Policy networks can help explain not only the ‘second-order’ decisions on ‘how do we do it’? (Peterson, 1995), but also decision-making in new policy sectors, and to predict conflicts in EU policymaking and implementation (Howlett, 2002). From policy networks to governance Policy networks have thus far been related to two main theoretical schools – the interest intermediation and governance schools. While the interest intermediation school regards policy networks as types of interest group–state relations at the meso level (policymaking and implementation), the governance school views policy networks as a theory of a specific form of governance (Börzel, 1997). The governance school (initially the ‘Max Planck’ School composed of Renate Mayntz, Fritz Scharp, Patrck Kenis, Volker Schenieder and Edgar Grande) regards policy networks as a particular form of governance, whereby “policy networks present themselves as a solution to coordina- tion problems typical for modern societies” (Börzel, 1997: 5). They were also known as “governance without government” (Rosenau, 1992). On the global level, policy networks were seen as being potential coordinators of activities at times when the word “government” is neither desired nor feasi- ble (Kenis and Raab, 2003). Although attempts to develop a network theory per se have been unsuc- cessful, the search for theories compatible with particular policy network analysis has continued (e.g. linking policy network analysis with resource dependency theory and diffusion theory; Kenis and Raab, 2007). Previous labels (such as neo-corporatism, policy communities, private interest gov- ernment, advocacy coalitions) used when talking about governance net- works have today been replaced by governance networks being interpreted as informal governance arrangements, partnerships, joined-up government, Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 857 co-governance mechanisms, strategic alliances, deliberative forums, advi- sory boards or policy task forces (Sørensen and Torfing, 2009: 237). Governance literature has shown particular interest in the question of the democratic potential of governance networks and – in this framework – in networks as tools of government. The stress remains on both the effective- ness and democratic character of networks (Sørensen and Torfing, 2005b). This orientation demands a stronger focus on the governance–governing relationship, including the issue of decision-making centres. Interestingly, various schools and authors have disagreed on whether policy networks actually have centres. Van Waarden (1992: 31) concluded that networks do not necessarily have a power centre or are coordinated by hierarchical authority, but instead thrive based on horizontal bargaining. Heclo (1978) and Hanf and Scharpf (1977) even maintained it is impossi- ble to locate a decision-making centre in policy networks. On the contrary, researchers looking at EU policy networks have stated that there is indeed a centre of policy networks to consider. Richardson (1994: 140) believes this is the European Commission acting as a policy broker supported by none static key actors and groups. This agrees with Majone’s emphasis on the Commission’s role in the EU as a “regulatory state” (Majone, 1994). Nevertheless, the lack of policy coordination continues to be stressed while studying EU policymaking. Hypotheses on the functionality of policy networks Peters (1998) developed several assumptions (A) regarding the function- ality of networks. He first argued that (1) less integrated networks are less likely to coordinate effectively (A1). This is largely to do to with the build- ing of trust and experience among members of networks. He added that (2) coordination is less likely when organisations are in similar policy areas but lack common ideas about solutions (A2). Put differently, networks that have a more unified ‘epistemic community’ are capable of generating coor- dination more easily than networks consisting of organisations that hold conflicting views. Third, he pointed out the importance of the international character of current policymaking, saying that (3) coordination should ini- tially occur at the national level in order to provide for successful interac- tion at higher levels (A3). An extension of those assumptions came in Peters’ call to primarily concentrate on the inclusion of those organisations most affected by the results of policymaking: national, regional and local groups, with their presence guaranteeing a focus on practical and realistic aspects of implementation. While Peters’ assumptions can guide our analysis in the search for EU net- works’ coordination capacity, we still need to operationalise the variables. Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 858 Here, we deploy indicators proposed by Jordan and Schubert (1992) within policy analysis literature and operationalise them with the use of SNA con- cepts (see Table 2). We consider three main criteria for evaluating the EU policy network. We look at: (1) the level of a network’s institutionalisation, especially its stability across policy issues; (2) the scope of the policymaking arrangements (whether the network is sectoral or trans-sectoral); and (3) the network’s participants (how diverse is the network’s membership). A simi- lar categorisation was used by van Waarden (1992). He looked at the num- ber and type of actors involved (referred to as the number of participants in Jordan and Schubert 1992: 12), the structure, institutionalisation, and rules of conduct – which Jordan and Schubert combined under the label of “level of institutionalisation” – along with power relations and actors’ strategies (within the scope of the policymaking arrangements). By combining all of these, we hope a more comprehensive picture of the EU policy network will emerge. Table 2: BREAKDOWN OF HYPOTHESES AND OPERATIONALISATION As su m pt ion s Peters (1998) Cr ite ria of ne tw or k e va lua tio n Jordan and Schubert (1992) Op era tio na lis at ion Operationalisation and SNA concepts less integrated networks are less likely to co-ordinate effectively (A1) (1) the level of a network’s institutionalisation, and in particular its stability across policy issues Non-stable relations (activity of pairs or more of actors on single directives) and stable relations (activity of two or more identical actors within more than one directive) SNA: value of ties (links) between nodes (actors) H1. If we identify more non-stable relations within EU network/network sections, we can conclude that the EU networks are less likely to co-ordinate effectively. co-ordination is less likely when organisations are in similar policy areas but lack common ideas about solutions (A2) (2) the scope of the policy- making arrangements (whether a network is sectoral or trans-sectoral) Are actors within the network diversified according to type of organisation? Are groups active on the same type of directives? Can we identify actors or a group of actors responsible for integrat- ing networks? SNA: diversity and density of a network, betweenness and closeness centrality H2. If we identify that different types of actors are active on the same types of directives within network/network subsections, we can expect less coordinated networks. (3) network’s participants (diversification of a network’s membership) the co-ordination should initially take place at the national level in order to provide for successful interaction at higher levels (A3) (3) network’s participants (diversification of a network’s membership) Are actors within the network diversified according to country and level of origin? SNA: diversity and density of a network, relations among network participants, number and types of actors, cross-policy relations, centrality measures, structural holes, network stability H3. If we identify that actors within network/ network sections are coming from different countries and levels of origin, we could expect more effectively coordinated networks. Source: Authors’ own analysis. Here is the list of questions guiding our empirical research: What is/are the social structure/s of the revealed networks (their diversity)? What can we say about the density of the policy networks and thus about the effects on their coordination capacity (density of network and ties/links between Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 859 actors)? Is there one or several centre/s and what is its/their structure? But, most of all, how do particular aspects of the network matter? Based on the above reading of Peters, what can be expected to occur and is it possible to develop any hypothesis on casual relations focusing on the policy net- works’ impact on a future policy process? Methodology - Social Network Analysis It is only social network analysis that allows flexibility while analysing networks at various levels – micro, meso and macro. While we agree with Varone, Ingold and Jourdain (2016: 323) that SNA offers a valuable method for empirically describing and assessing the key variables that determine a policy network’s characteristics (which may thereby also determine the policy process), we also believe it holds potential while analysing networks’ capacities. This potential lies not only in its mathematical, conceptual and analytical foundations facilitating the description of rational configurations, but also in its precision and visualisation, offering new relevant insights into complex configurations and relationships (Schneider, 2017). Indeed, the visualisation of policy networks can go beyond mere ‘illustration’ and “help to improve communication about the data to third parties and help the researcher to explore specific properties of certain networks better or facilitate the exploration of differences across several networks; or it could even help to discover explanations for policies” (Brandes et al., 1999: 76). Data We reconstruct EU policy networks as they were created in the process of establishing national positions on the 20 most salient EU directive pro- posals between 2008 and 2010. The data were collected within the frame- work of the INTEREURO Multi-level Governance Module (Beyers et al., 2014; INTEREURO, 2014). The proposals concern three policy fields: the environment and energy, finance and the economy, and rights. For exam- ple, the Directive Proposal on Deposit Guarantee Schemes falls within the finance and economy policy field, the Directive Proposal on Patients’ Rights in Cross-border Health Services is a representative of a policy field focused on rights, and the environmental field is best denoted by the Directive Proposal on Promotion of Renewable Energy Sources. In a much finer pol- icy field categorisation, we distinguished between policy fields which are distributive, redistributive and regulatory in their nature (Lowi, 1972). Interviews were conducted with national and European policy officials who were asked to indicate all non-state and para-state actors involved in the debates on several of the most conflictive issues related to the studied Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 860 directives’ proposals. We asked policy officials to place different stakehold- ers active within the debates on those issues on a scale from 0 (a position that favours less integration or less regulation) to 100 (encouraging full harmo- nisation on the European level). Following Helbling and Tresch (2011), we assumed the most controversial issues would have entailed larger debates, thus allowing us to better control for the size of a network. We considered actors from five EU member states (Germany, Netherlands, Slovenia, Sweden and United Kingdom) and further expanded by including European-level organisations and institutions. The country selection controls for several national characteristics, ensuring variation with respect to population size, national wealth, state-interest groups’ tradi- tional models, and political system. Four countries have income per capita higher than the EU average, with Slovenia falling below it. Lijphart’s index of interest group pluralism (2012) for the selected countries shows quite a spread, with the most corporatist being Sweden (0.5) and the most pluralist being the UK (3.38), with Slovenia and Germany being placed closer to a weak version of corporatism (0.88). In total, we identified 56 issues. The full list of units included 254 actors. For the analysis, the European Commission was excluded from the data due to its special role as a central actor in all EU policymaking networks (it is not exclusive in our 20 directives). Thus, the network’s border (i.e. the list of eligible actors) consisted of 253 units (254 minus the Commission). The majority of actors have an economic background (142), followed by non- governmental organisations (65), para-state actors (30) and expert organisa- tions (16). Network description and revisiting the hypotheses The analysed data are conceptualised as a two-mode (bivariate) network where one set of units represents the network’s stakeholders, while the sec- ond set of units represents directives in a policy proposal. The relationship between these two sets is defined as an organisation’s involvement in con- flicting issues that form part of a certain directive proposal. The focus of the analysis is the networks of organisations involved in more than one directive. Therefore, organisations connected to individual directives were removed. The network was further reduced by eliminating disconnected directives. The reduced data consisting of 17 directives and 27 organisations were reordered into the analysed one-mode network of 27 organisations (Figure 1). Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 861 Figure 1: ONE-MODE REDUCED NETWORK WHERE TIES AMONG ACTORS INDICATE ACTIVITY ON THE SAME DIRECTIVE Source: Authors’ own analysis. Relations among actors in our analysis Two units (actors) in this network are linked when they worked on the same directive, meaning there were indicated in an interview as being active on the same particular directive. For example, the World Wildlife Fund and Greenpeace Europe were mentioned as being active on the Directive on Renewable Energy Sources, thus in our social network they will be linked by a tie. Ties (links) between two actors are valued according to the number of different directives they were both involved in together (e.g. the more directives they were involved in, the larger the value of the tie). Given the characteristics of the available data, which do not include information on actual relations between actors but actors’ activity on issues concerning the same directive, we are talking about a proxy for network relations. Number and types of actors Table 3 includes all our units/actors in the network after the reduc- tion (namely, after removing actors that are only active with respect to one directive). We can already observe several things about the network. With regard to the country of the origin of the actors, we see the dispro- portional representation of EU-level and Slovenian organisations, respec- tively 10 and 8 (66.5% of the network). Those actors’ overwhelming pres- ence is explained by the fact most interviews were conducted with EU-level Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 862 officials and Slovenian policy officers (20 and 34, respectively, compared with 8 in Germany, 5 in the Netherlands, 10 in the UK, and 5 in Sweden). A comparison of the ratio of interviews with the number of actors mentioned would not apply here because we reduced the network to only actors that were active in relation to more than one directive (253->27). The net- work includes 13 economic groups (e.g. banking associations, investment manager associations), 1 expert organisation (the Slovenian Jozef Stefan Institute), and 12 NGOs (mostly environmental organisations, e.g. European Environmental Bureau), and 1 para-state organisation (VNG – Association of Dutch Municipalities). Table 3: ACTORS IN THE NETWORK AFTER THE REDUCTION Name Country of origin Type of organisation No. Direc- tives Aggregated constraint (structural holes) Degree Closeness Betweenness centrality Alstom INT Economic 2 0.277545 13 0.577778 0.006044 Association of Management of Investment Funds - GIZ SL Economic 2 0.542496 6 0.382353 0.000000 Bosch DE Economic 3 0.277545 13 0.577778 0.006044 BP GB Economic 2 0.277545 13 0.577778 0.006044 Business Europe EU Economic 4 0.236552 16 0.619048 0.030623 EURELECTRIC EU Economic 2 0.256342 15 0.604651 0.012527 European Association of Cooperative Banks (EACB) EU Economic 2 0.542496 6 0.382353 0.000000 European Association of Public Banks (EAPB) EU Economic 2 0.542496 6 0.382353 0.000000 European Banking Federation (EBF) EU Economic 2 0.542496 6 0.382353 0.000000 European Consumers Association (BEUC) EU NGO 3 0.328511 9 0.541667 0.369231 European Environmental Bureau (EEB) EU NGO 2 0.277565 7 0.553191 0.059780 European Fund and Asset Manage- ment Association (EFAMA) EU Economic 2 0.542496 6 0.382353 0.000000 Focus- Society for Sustainable Development SL NGO 2 0.256342 15 0.604651 0.012527 Friends of the Earth Europe EU NGO 4 0.268598 13 0.577778 0.006044 Greenpeace Europe INT NGO 5 0.216079 17 0.702703 0.370733 Jožef Stefan Institute SL Experts 2 0.285642 10 0.520000 0.003077 Investment Management Association (IMA) GB Economic 2 0.542496 6 0.382353 0.000000 Legal Information Centre for NGOs SL NGO 2 0.227220 14 0.500000 0.147692 Natuur NL NGO 3 0.241540 16 0.619048 0.030623 Save the Children EU NGO 2 0.849490 2 0.342105 0.000000 Shell NL Economic 2 0.277545 13 0.577778 0.006044 Slovenian Chamber of Commerce SL Economic 2 0.256342 15 0.604651 0.012527 Slovenian E-Forum, Association for Energy Economics and Ecology SL NGO 2 0.285642 10 0.520000 0.003077 Slovenian Philanthropy, Association for the Promotion of Volunteering SL NGO 2 0.849490 2 0.342105 0.000000 Umanotera Slovenian Foundation for sustainable development SL NGO 2 0.256342 15 0.604651 0.012527 World Wildlife Fund (WWF) INT NGO 3 0.244639 15 0.604651 0.012527 VNG NL Para-state 2 0.482614 3 0.481481 0.000000 Source: Authors’ own analysis. Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 863 Cross-policy relations In order to gain an insight into a network’s fabric, we look at cross-direc- tive collaborations where actors are jointly active on more than one direc- tive. The process of reducing the network to those actors left us with 27 units. The reduction process also led to lowering the number of directives from 20 to 17. While we see more economic actors in the network, it is worth point- ing out that the majority of them were only active in relation to two direc- tives (the exception BusinessEurope for 4 and Bosch for 33). Cross-directive activity better characterises NGOs (2 to 5 directives), with Greenpeace being active with respect to 5 directives. All directives in which Greenpeace was active are examples of environmental directives (on waste manage- ment, ship source pollution, renewable energy resources, CO2 storage, radioactive waste disposal) (Table A in the Appendix). A similar conclusion emerges with regard to all the other NGOs’ activities (expanding the list for Greenpeace with directives on the energy performance of buildings and greenhouse gas emissions). When focusing on the actors’ country of origin, the only conclusion we can draw concerning the cross-policy relations is that organisations from Slovenia (without distinguishing the type of actor) and those at the EU level are, more often than others, working on more than one directive. Given the cross-policy character of EU-level organisations, this is a reasonable conclusion. The argu- ment for Slovenia, unlike for other countries studied in the article, could stem from the fact that the country is much smaller population-wise while its inter- est groups system is less professionalised and thus we can expect that certain national organisations become more cross-directive in orientation due to the limited number of groups that are active on European issues. Centrality measures and structural holes Closeness centrality. Our analysis expands when we look at variables that are more characteristic in social network studies. Closeness measures ties across all networks, not just in the immediate vicinity of an actor. Thus, para- phrasing, we can see who is the most central in the network, the closest to all other actors. The actor with the highest value for closeness centrality is Greenpeace (0.70), followed by Natuur (0.62) and BusinessEurope (0.62). These three actors are the most embedded in the network. The potentially have the broadest access to all actors in the analysed part of the network. Betweenness centrality and bridging actors. Betweenness centrality measures the role of an actor as a hub. Actors with the highest between- ness-centrality value hold the greatest potential to link different sections Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 864 of the network. A ‘bridging’ actor in this sense is the most important actor for linking parts of the network. In our analysed section of the network, Greenpeace has the highest value (0.370), followed by BEUC (0.369). This is seen especially in Figure 1. Greenpeace and BEUC play the role of artic- ulation points where removing them from the network would cause it to fall apart in two separate components. Interestingly, while BEUC has a very broad role as a bridge, it is connected to the network’s section only through ties with a value of 1. Considering our definition of stable/non-stable net- works (see below), BEUC is part of an unstable relationship. Bridging structural holes. The kind of relationship (as described above) is important only in addition to the sheer number of relations. Having many relations within a group exposes an actor to the same information over and over again, whereas relations outside of one’s group yield more diverse information which is worth passing on or retaining to make a profit (de Nooy et al., 2012). The higher the presented aggregate constraint of an actor, the less ‘freedom’ that actor has to withdraw from existing relations or to exploit structural holes. People or organisations with a low aggregate constraint are hypothesised to perform better (Burt, 2001). In the analysed network, those organisations with the lowest constraints, and therefore the greatest opportunity to create new relevant ties to fill the structural holes, are once again Greenpeace Europe, BusinessEurope, but also the Legal Information Centre for NGOs. This again underlines the quite privileged role and function of Greenpeace and BusinessEurope. Network stability We are interested in cross-directive collaborations. Actors working on more than one directive have a wider interest and a more elaborate agenda. Instead of being just part of an issue network (involved only in debates on individual issues), they are part of multiple-issue policy networks. If two actors engage on two common directives or more, we may talk about a more stable relationship (in social network terms) compared to two actors only engaged on one common directive (see Figure 2). In mathematical- SNA terms, we calculate a network’s stability by the weight/value of the ties between actors. A more stable relationship (stable policy network) has a value of at least 2 and a less stable one (defined as only a single-issue net- work) has a value of 1. In the analysed network, actors/units have at least 1–3 directives in common. Most ties have a weight of 1 (e.g. two actors were active on the same directive). The strongest connection is between Friends of Earth and Greenpeace, which is the sole tie with a value of 3. Both organisations were involved in three directives, namely the Directive Proposals on: Renewable Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 865 Energy, Carbon Capture and Storage, and Emissions Trading Scheme, with all three being categorised as environmental directives. If we focus on more stable relations (ties with a value of 2 or more), the policy network that is revealed is clearly divided into three independ- ent components (e.g. sections; Figure 1): two smaller ones and one larger. The larger component in Figure 1 represents the network of interest groups active in environmental policy fields (regulatory policy), while the two smaller ones are in financial policy fields (distributive policy). The first smaller component consists of banking associations (European Banking Federation – EBF, European Association of Public Banks – EAPB, and European Association of Cooperative Banks – FACB). All three actors are categorised as exactly the same type (‘economic actors’) and level (European). They were active on exactly the same two directives, catego- rised as a ‘finance’ directive. The second small component consists of the European Fund and Asset Management Association (EFAMA), Association of Management of Investment Funds (GIZ), and the Investments Management Association (IMA). Three groups worked on the same two Directive Proposals: Investments Schemes and Deposit Schemes. All three actors are categorised as ‘finance’ actors. While EFAMA is a European-level group, GIZ is Slovenian, and IMA is British. Finally, the largest component consists of a mix of organisations. We have here representatives of environmen- tal European (e.g. WWF, EEB) but also national NGOs (e.g. the Slovenian E-Forum). However, we also find actors classified as finance and business- related (e.g. Dutch Shell, German Bosch, international Alstrom, or European- level – BusinessEurope). There is a clear indication of a cross-country and cross-level structure as well as the cross-organisation type. It is a complex Figure 2: NETWORK STABILITY – SIMPLIFIED MODEL FOR EXPLANATION* Source: Authors’ illustration. *A1, A2, A3, A4 – different hypothetical units/actors within a network Directives A, B, C – three hypothetical directives r1 and r2 – ties in a two-mode network, indicating unit/actor activity on a particular directive Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 866 and dense network with multiple connections (ties) between the actors. In some cases, a tie’s value is larger than 2, sustaining the network’s stability. When looking at the details of the biggest component, another conclu- sion emerges. If we focus just on ties with a value of 2 or more (namely, actors engaging jointly on 2 or more of the same directives), we can identify three central actors that are bridging three sub-sections of that component. The most central (talking about betweenness centrality) is the WWF, the sec- ond is BusinessEurope, and then there is Natuur. The sub-sections they are bridging are actually made up of different types of actors and different lev- els. That adds to our conclusion that the network’s dense sub-section is actu- ally constructed from cross-level, different interest group types and interest groups with different countries of origin. Empirical findings Our findings support the thesis that it is possible to reveal a policy net- work centre, even where this centre is not a single actor. Instead, it is the core of a network composed of clusters of actors that appear in policy pro- cesses related to various draft directives. Further, it is not one network but a set of components. They differ with regard to policy type. Distributive and redistributive draft directives are represented by two smaller components (Figure 1), with a small number of active actors and ties of the same value. The much more complex component (due to the diverse types and levels of actors as well as the mix of tie values) encompasses actors that are active in relation to regulatory policies (primarily environmental). Returning to our hypotheses, we can provide a few conclusions. First, we argued that the stability of a network (ties between nodes with a value equal to 2 or higher) improves the network’s coordination capacity (reversed H1). While we observed it within the identified components, a worrying sign is the lower values between them. Two smaller components are clearly separated from the rest of the network and even remain strangers between themselves. In conclusion, coordination between the diverse policies in the finance sector (in our case, also identified as distributive and redistributive policies) is less likely to be effective. In contrast, the larger component including a number of regulatory policies (in our sample chiefly represented by environmental policies, but also protection of rights) does not have that problem, suggesting bet- ter coordination takes place between the development processes of dif- ferent directives within those policy fields. However, another issue arises here. Paraphrasing hypothesis 2, the diversification of actors’ types within a component actually works against effective coordination (H2). When it comes to environmental regulations and the protection of rights, namely Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 867 the network’s biggest component, we noticed several diverse actors across the types. It could consequently be argued that the networks built around those issues are less likely to be coordinated effectively. Our findings sug- gest that, when it comes to financial regulations, the established compo- nents are quite stable, including the same types of organisations, albeit from different levels (however, given H3 on the more effective coordination in networks comprising cross-level and country actors, that is not a problem). The hope for improving the coordination capacity within the largest component of the network lies in its increased cross-country and cross-level structure (H3). Summing up, while a mix of cross-type actors negatively influences a component’s coordination capacity, a cross-level and cross- country mix acts as a mitigating factor. Here, however, a caveat must be noted. We observed a number of EU-level organisations and those coming from Slovenia in that set. Potentially, this therefore allows us to argue that those two sets of actors are in particular responsible for the component’s coordination capacity. But, as mentioned, our data are imperfect and there is an issue of the overrepresentation of the Slovenian actors, rendering the argument inconclusive. Conclusion Our aim in this article was to identify the potential of EU networks’ coordination capacity. The two main points of focus were the search for a decisional centre of the network and the information on network stability. Using inputs from policy analysis, governance literature and network stud- ies, we operationalised our variables and deployed SNA as an analytical tool to help us theorise about coordination capacity. Our argument is that such approach brings added value, especially when applied to the peculiarities of EU policy networks emerging in complex decision-making, which links the national and supranational (EU) levels. Our approach holds consider- able potential to further develop the concepts and theories of governance as well as theories of governing in the postmodern era. Our research findings speak in favour of the thesis that EU governance may be described as a network of networks. While our findings do not directly indicate country as a significant variable, they advocate the need for cross-country and cross-level engagement in order to improve the coordina- tion of a network. Countering the notion that individual policy networks are more or less isolated, our findings show links both among policy networks and within them (especially looking at the full network and ‘individual’ pol- icy networks as only its components). Yet we need to include the noticeable difference between regulatory policy areas on one hand and distributive and redistributive policy fields on the other. Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 868 Based on our analysis, greater attention must also be given to particular interest organisations through which policy networks and their components become connected. Although network analysis is very descriptive, we can make comments about the roles played by the groups (e.g. bridging roles) and their importance for the networks. Additional research is needed in relation to interest organisations. Indeed, in line with expectations, while a huge cloud of actors is involved in the network, not all actors matter equally for it. Given the network’s revealed characteristics, the potential to make a difference varies considerably among the actors and a relatively small share of the actors divide it into individual policy components. As shown in the empirical section, network components may differ con- siderably in terms of the size and variety of the actors involved. Among the ‘all EU-level policy network’, ‘all-one-industry EU policy network’ and ‘over- all heterogeneous EU policy network’, only the latter (the environmental EU policy network) appears to be in line with the expectations that networks in EU policy processes will generally be internally heterogeneous. Further research is required to examine whether including all 28 EU member states and broadening the analysis to a larger sample and including more direc- tive proposals in the analysis would bring different results or confirm our findings. Future studies should also consider defining the network’s stabil- ity through the use of temporal data (e.g. interactions measured across time, not only across directives) and by focusing more on the substance of an actor’s activity (implementing not only the actor’s activity, but also its char- acter, e.g. if groups were pro or against a particular Directive Proposal). All in all, our contribution supports the position taken by Kenis and Schneider (1989) that it is unrealistic to search for a theory of policy net- works but more viable to contribute to the development of network theo- ries in certain academic fields, including governance. However, governance also seems to be more a concept than a theory and the need for theorising governing in the world of today has already been expressed in the need for the ‘governance of governance’ or ‘meta governance’ (Sørensen and Torfing, 2005). While there is quite obvious potential in linking the study of policy net- works in EU policy processes and governance studies, it should not be for- gotten that policy actors and policy networks operate in an institutional context. This not only relates to the institutional characteristics of the EU political system and its dynamics, but also the national and international intergovernmental institutional contexts which directly link national and supranational policymaking. Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 869 Appendix Figure 1: CAPTION: REDUCED TWO-MODE NETWORK INCLUDING DIRECTIVES AND INTEREST ORGANISATIONS AS NODES, TIES BETWEEN UNITS INDICATE ACTORS’ ACTIVITY ON A PARTICULAR DIRECTIVE Source: Authors’ own analysis. Table A: ANALYSED DIRECTIVES Directive No. of issues Description 001 3 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL laying down the framework for the deployment of Intelligent Transport Systems in the field of road transport and for interfaces with other transport modes 003 4 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on waste electrical and electronic equipment (WEEE) (Recast) 004 4 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL laying down minimum standards for the reception of asylum seekers 011 5 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the energy performance of buildings (recast) 035 2 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the protection of animals used for scientific pur- poses 039 3 Proposal for a COUNCIL DIRECTIVE amending Directives 92/79/EEC, 92/80/EEC and 95/59/EC on the structure and rates of excise duty ap- plied on manufactured tobacco Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 870 Directive No. of issues Description 046 4 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the application of patients’ rights in cross-border healthcare 063 2 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL amending Directive 2005/35/EC on ship source pollu- tion and on the introduction of penalties for infringements 077 2 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the promotion of the use of energy from renewable sources 078 4 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the geological storage of carbon dioxide and amending Council Directives 85/337/EEC, 96/61/EC, Directives 2000/60/EC, 2001/80/EC, 2004/35/EC, 2006/12/EC and Regulation (EC) No 1013/2006 079 1 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL amending Directive 2003/87/EC so as to improve and extend the greenhouse gas emission allowance trading system of the Community 080 1 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the safety of toys 104 2 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on Alternative Investment Fund Managers and amend- ing Directives 2004/39/EC and 2009/…/EC 108 1 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL amending Council Directive 78/660/EEC on the annual accounts of certain types of companies as regards micro-entities (Text with EEA relevance) 120 1 Proposal for a COUNCIL DIRECTIVE on the management of spent fuel and radioactive waste 136 4 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL amending Directive 97/9/EC of the European Parlia- ment and of the Council on investor-compensation schemes 137 3 Proposal for a DIRECTIVE …/…/EU OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on Deposit Guarantee Schemes [recast] 142 4 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on preventing and combating trafficking in human be- ings, and protecting victims, repealing Framework Decision 2002/629/ JHA 143 3 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on combating the sexual abuse, sexual exploitation of children and child pornography, repealing Framework Decision 2004/68/JHA 144 3 Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the right to interpretation and translation in crimi- nal proceedings Source: Authors’ own analysis. Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 871 BIBLIOGRAPHY Beyers, Jan, Laura Chaqués Bonafont, Andreas Dür, Rainer Eising, Danica Fink- Hafner, David Lynn Lowery, Christine Mahoney, William Maloney, Daniel Naurin (2014): The INTEREURO Project: Logic and Structure. Interest Groups and Advocacy 3 (2): 126–140. Börzel, Tanja. A. (1997): What’s So Special About Policy Networks? – An Exploration of the Concept and Its Usefulness in Studying European Governance. European Integration online Papers (EIoP), 1–16. Brandes, Ulrik, Patrick Kenis, Jőrg Raab, Volker Schneider and Dorothea Wagner (1999): Explorations into the Visualization of Policy Networks. Journal of Theoretical Politics 11 (1): 75–106. Burt, Ronald S. (2001): Structural Holes versus Network Closure as Social Capital. In: Lin, Nan, Karen Cook and Ronald Burt (eds.), Social Capital: Theory and Research, 31–56. New York: Aldine de Gruyter. Carrington, Peter, John Scott and Stanley Wasserman (eds.) (2005): Models and Methods in Social Network Analysis (Structural Analysis in the Social Sciences). Cambridge: Cambridge University Press. Davis, Glyn (1995): A Government of Routines: Executive Coordination in an Australian State. Melbourne: Macmillan Education. Daugbjerg, Carsten (1999): Reforming the CAP: Policy Networks and Broader Institutional Structures. Journal of Common Market Studies 37 (3): 407–428. Hanf, Kenneth and Fritz W. Scharpf (1977): Interorganizational Policy Making. Limits to Coordination and Central Control. London: Sage. Heclo, Hugh (1978): Issue Networks and the Executive Establishment. In: King, A. (ed.) The New American Political System, 87–124. Washington D.C.: American Enterprise Institute. Helbling, Mark and Anke Tresch (2011): Measuring Party Positions and Issue Salience from Media Coverage: Discussing and Cross-validating New Indicators. Electoral Studies (30): 174–183. Howlett, Michael (2002): Do Networks Matter? Linking Policy Network Structure to Policy Outcomes. Evidence from Four Canadian Policy Sectors 1990–2000. Canadian Journal of Political Science 35 (2): 235–267. Jordan, Grant and Klaus Schubert (1992): A Preliminary Ordering of Policy Network Labels. European Journal of Political Research 21: 7–27. Kenis, Patrik and Jörg Raab (2003): What Do Policy Networks Do? Paper at the Conference on Democratic Network Governance. Copenhagen, Denmark, 22–23. Kenis, Patrik and Jörg Raab (2007): Taking Stock of Policy Networks: Do They Matter? In: Fisher, F., Miller, G. J. and Sidney, M. S. Handbook of Public Policy Analysis, 187–200. CRC Press: Boca Raton, London, New York. Kenis, Patrik and Volker Schneider (1989): Policy Networks as an Analytical Tool for Policy Analysis. Paper for conference at the Max Planck-Institute, Cologne, 4–5 December 1989. Kochen, Manfred and Karl W. Deutsch (1980): Decentralisation: Sketches towards a Rational Theory. Cambridge. Mass.: Oelgeschlager, Gunn and Hain. Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 872 Lijphart, Arenth (2012): Patterns of Democracy. Government Forms and Performance in Thirty-Six Countries, 2nd ed. New Haven, CT: Yale University Press. Lowi, Theodore J. (1972): Four Systems of Policy, Politics and Choice. Public Administration Review (333): 298–310. Majone, Giandomenico (1994): The Rise of the Regulatory State in Europe. West European Politics 17 (3): 77–101. March, James G. and Johan P. Olsen (1989): Rediscovering Institutions: Organi- zational Factors in Political Life. New York: Free Press. Marin, Bernd (1990): Generalized Political Exchange: Preliminary Considerations. In: B. Marin (ed.), Generalized Political Exchange: Antagonistic Cooperation and Integrated Policy Circuits. Frankfurt: Campus Verlag. de Nooy, Wouter, Andrej Mrvar and Vladimir Batagelj (2012): Exploratory Social Network Analysis with Pajek. Cambridge: Cambridge University Press. Olsen, Johan P. (2002): Reforming European Institutions of Governance. Journal of Common Market Studies 40 (4): 581–602. Peters, Guy. B. (1998): Managing Horizontal Government: The Politics of Co-ordination. Public Administration (76: Summer): 295–311. Peterson, John (1995): EU Research Policy: The Politics of Expertise. In: Carolyn Rhodes and Sonia Mazey (eds), The State of the European Union, Vol. 3: Building a European Polity? 391–411. Boulder and Essex: Lynne Riennerand Longman. Reinicke, Wolfgang H. (1997): Global Public Policy. Foreign Affairs (November/ December): 127–38. Reinicke, Wolfgang H. (1999–2000): World Wide Web: Global Public Policy Networks, Foreign Policy (117): 64–157. Rhodes, R. A. W. (1997): Understanding Governance. Buckingham: Open University Press. Richardson, J. (1994): EU Water Policy: Uncertain Agendas, Shifting Networks and Complex Coalitions. Environmental Politics 3 (4): 139–167. Rosenau, Jeremy (1992): Governance, Order and Change in World Politics. In: Rosenau, Jeremy and Ernst-Otto Czempiel (eds.), Governance without Government: Order and Change in World Politics, 1–29. Cambridge: Cambridge University Press. Scharpf, Fritz W. (1997): Games Real Actors Play: Actor-centred Institutionalism in Policy Research. Boulder, CO: Westview. Schneider, Volker (2017): Comment on: Studying Policy Advocacy through Social Network Analysis. European Political Science 16: 335–336. Scott, Collin (2009): Governing Without Law or Governing Without Government? New-ish Governance and the Legitimacy of the EU. European Law Journal 15 (2): 160–173. Sørensen, Eva and Jacob Torfing (2005a): Network Governance and Post-Liberal Democracy. Administrative Theory & Praxis 27 (2): 197–237. Sørensen, Eva and Jacob Torfing (2005b): The Democratic Anchorage of Governance Networks. Scandinavian Political Studies 28 (3): 195–218. Danica FINK-HAFNER, Meta NOVAK, Luka KRONEGGER, Patrycja ROZBICKA TEORIJA IN PRAKSA let. 56, 3/2019 873 Sørensen, Eva and Jacob Torfing (2009): Making Governance Networks Effective and Democratic through Metagovernance. Public Administration 87 (2): 234– 258. Torfing, Jacob (2016): Metagovernance. In: Christopher K. Ansell and Jacob Torfing (eds.), Handbook on Theories of Governance, 538–550. Cheltenham, UK and Northampton, MA, USA: Edward Elgar Publishing. Van Waarden, Frans (1992): Dimensions and Types of Policy Networks. European Journal of Political Research 21: 29–52. Varone, Frédéric, Karin Miryam Ingold and Charlotte Jourdain (2016): Studying Policy Advocacy through Social Network Analysis. European Political Science 16 (3): 322–336. SOURCES INTEREURO (2014): Networks, Strategies and Influence in the EU – Project. Accessible at http://www.intereuro.eu/public/, 20. 1. 2019.