Cross-National Policy Networks and the State Paper prepared for presentation at the Ninth Biennial Conference of the European Union Studies Association, March 31 – April 2, 2005, Austin, Texas, Panel 1D: Interest Groups and Lobbying in Comparative Perspective Beate Sissenich Asst. Professor Department of Political Science Woodburn Hall 210 Indiana University Bloomington, IN 47405 812-855-4198 bsisseni@indiana.edu Work in progress. Please do not cite without author's permission. Acknowledgments: I thank James Russell for technological consultation and Antje Schwennicke for research assistance. Indiana University's Russian and East European Institute provided funds for ongoing survey research, while Cornell University and the Social Science Research Council assisted with earlier data collection efforts. Helpful comments on previous versions came from Tim Bartley, Matthew Evangelista, Bernard Grofman, Jeffrey Isaac, John Krinsky, Peter Katzenstein, Mike McGinnis, David S. Meyer, Sid Tarrow, Charles Tilly, and Barry Wellman. Abstract This paper challenges overly optimistic claims about the power of nonstate actors in cross-border networks that link them with states and intergovernmental organizations. In a most-likely case design, the paper examines the social policy network between the European Union (EU) and Poland and Hungary prior to accession. The analysis focuses on two dimensions: whether states act as gatekeepers and whether national borders restrain communications. The paper demonstrates that while nonstate actors easily communicate with intergovernmental organizations such as the EU, contacts fail to cut across national borders. Network data show that the EU, as an intergovernmental actor, significantly controls the flow of communication, a fact that runs counter to the notion of networks as fluid entities that enable all actors to link freely with others. Both state and intergovernmental actors have an interest in network construction and control. Only by empirically tracking network contacts, the paper argues, will we be able to estimate the capacities and limits of nonstate actors in transnational politics. Introduction To what extent is interest representation in EU policy formation and implementation transnational? This paper examines two dimensions of transnationalism: First, to what extent do governments act as gatekeepers in EU policy formation? If interest representation in the EU were indeed transnational, then we would expect governments not to play any significant role as gatekeepers in policy negotiations. Instead, interest groups should be able to access decision- making forums at the European level without the mediation of national governments. Second, to what extent do national boundaries shape interest representation in EU policy? In a transnational policy network, we would expect policy-related communications to cross national boundaries rather than be segmented by them. I focus on EU social policy as a most likely case of a transnational policy network. This case presents two factors that are likely to produce dense cross-border networks among diverse sets of actors. First, the EU is the world's most highly institutionalized system of regional integration. If transnational interest representation exists anywhere, we should certainly expect to find it in the EU. Second, EU social policy grants certain nonstate actors (i.e. employer organizations and trade unions) a specific role in policy-making and implementation. Concretely, European social dialogue is a procedure for peak-level consultations between European-level employer and labor organizations with the option of concluding agreements that will lead to Community legislation. Such legislation may then be implemented in the member states by labor and employers themselves or through legal transposition. Social dialogue is also an element in the European Employment Strategy, at both the European and national levels. The social partners regularly consult with the Council and the Commission in the Standing Committee on Employment and with the heads of state or government of the past, current, and future EU presidencies. Furthermore, in order to discuss macroeconomic policy, the social partners meet regularly with the Commission, member state economic and finance ministers, and the European Central Bank. In this paper, I present evidence on transnational policy networks drawn specifically from accession preparations of Poland and Hungary. I also describe an ongoing survey on communications in the field of EU social policy among the 25 member states and Bulgaria, Romania, and Turkey. Accession preparations constitute a special scenario for the development of a cross-border policy network. On one hand, given that such a network takes time to evolve and accession negotiations were primarily intergovernmental, we would expect both border effects and states' gatekeeping function to be stronger than in a network among member states. On the other hand, however, accession conditionality required the adoption of EU rules across all policy fields, including social dialogue, giving candidates an incentive to include the social partners in policy adoption and implementation. My argument proceeds as follows: I begin by relating my research to two bodies of literature: policy networks and transnational politics. Next, I set up my model of transnational networks along two dimensions: the extent to which territorial states shape network interactions and the ability of governments to mediate between intergovernmental organizations and nonstate actors. The bulk of the paper is devoted to a case study of network interactions around Poland and Hungary's adoption of EU social policy prior to accession. The case illustrates that at least in this instance, border effects constrain transnational communications. I follow up with some remarks on ongoing survey research into the much larger case of EU social policy among 25 member states and three candidate countries. Relevance My research seeks to bridge two lines of investigation that have remained strangely separate from each other: transnational politics and policy networks. Transnational networks that link nonstate actors with intergovernmental organizations and governments have recently been credited with many accomplishments: achieving policy change at the international and domestic level, improving the status of nonstate actors in their domestic settings, giving nonstate actors leverage over international relations, and creating a transnational political public (Boli and Thomas 1999; Fox and Brown 1998; Gordon and Turner 2000; Guidry et al. 2000; Keck and Sikkink 1998; O'Brien et al. 2000; Risse et al. 1999; Smith et al. 1997; Young 1997). Intergovernmental organizations may provide societal actors with new access points for articulating their interests. Much has been made, for instance, of nonstate actors' ability to render the operating procedures of international financial institutions more transparent and accountable (Edwards and Gaventa 2001; Fox and Brown 1998; Hulme and Edwards 1997). Sectors that pose inherently transnational problems, such as environmental policy, would appear to lend themselves to the influence of nonstate actors and their cross-border networks (Keck and Sikkink 1998; Wapner 1995; Young 1997). Some analysts have hailed the increasing density and visibility of cross-border advocacy coalitions as proof of an emergent transnational or even global civil society (Edwards and Gaventa 2001; Fiorini 2000, pp. 7-8). The sources of influence of these coalitions include mass membership, knowledge, money, job-creating capacity, access to international organizations and national governments, and moral authority (Josselin and Wallace 2001, p. 253). Like the literature on transnational advocacy coalitions, theories of network governance emphasize cooperation between state and nonstate actors. The concept of policy networks is based on the idea that where policy-making and implementation are widely dispersed, state and nonstate actors occupy overlapping spheres and organized interests permeate state institutions. They are seen as relatively open, combining formal and informal exchange and conflict as well as cooperation (Bogason and Toonen 1998; Kenis and Schneider 1991; Pappi and Henning 1999; Rhodes 1996, 1997). But whereas transnational advocacy networks are ostensibly delineated by substantive agreement, policy networks are demarcated by stakes in a given policy area. And while transnational advocacy networks are often seen as substitutes for democratic accountability, policy networks tend to be depicted in more sober terms, as technocratic and depoliticized forms of decision-making that circumvent the separation of powers in the interest of technical knowledge and political expedience. Policy networks, it is argued, are likely to arise in settings of multilevel governance, high issue complexity, and dispersed decision-making. Just how states and intergovernmental organizations interact with nonstate actors is a question we need to investigate more closely. If a network's boundaries are assumed to be perpetually in flux and outsiders can become insiders at any moment, empirically assessing transnational interactions may seem daunting. Yet unless we actually trace cross-border contacts, we cannot know whether it is possible to have a vibrant transnational network that nevertheless fails to influence domestic or international policy. Similarly, mapping network interactions is necessary if we want to assess whether networks emerge spontaneously or are constructed deliberately by certain actors. Is the opportunity for cross-border contacts sufficient for generating network contacts? Who constructs networks, and to what end? The failure to chart network interactions may lead us to misunderstand power relations in transnational space. Whether cross-border networks are indeed devoid of hierarchies, as the imagery often implies, requires empirical investigation. Similarly, we are likely to overlook the politics of cross-border interactions if we define networks on the basis of shared values. Durable trends of international NGO proliferation (Boli et al. 1999) have spurred considerable research on states and intergovernmental organizations as targets or allies of nonstate actors' pursuits. Scholars see nonstate actors as capable of setting agendas, changing the normative frames of reference among state actors, and altering organizational modes of operation, state policies, and behavior. Keck and Sikkink's (1998, pp. 12-13) famous "boomerang pattern" of influence describes states and intergovernmental organizations as potential allies of societal actors in repressive systems. In some cases, the mere existence of supranational organizations and rules may enhance the impact of certain issue networks (Gelb 2002). However, the account of nonstate actors enlisting states and intergovernmental organizations for their own purposes misses the extent to which states act as filters for the cross- border activities of nonstate actors. States control access to their territory, though not always to their citizens (Huntington 1973; Vallier 1971; Waldinger and Fitzgerald 2004). Their political institutions affect modes of organization and mobilization by transnational actors (Josselin 2001a; Krasner 1995). States continue to provide the main opportunity structures for political mobilization and focal points for collective identities (Riker 2002; Tarrow 2001a and 2001b). International markets depend on well-defined property rights, whose formulation and enforcement remain primarily the responsibility of nation-states (Halliday 2001; Keohane and Nye 1971; Strange 1988; Wapner 1997). States remain the constitutive elements of intergovernmental organizations, many of which have become targets as well as hubs of transnational activism (Clark, Friedman, and Hochstetler 1998; Keohane and Nye 2000). The similarity of states' domestic institutions and social values may facilitate common interests and thus improve the odds of international cooperation (Keohane 2002, p. 34). Levels of nonstate- actor activity appear to follow levels of government activity in a given area (Skjelsbaek 1971). Whether a country's NGOs develop transnational ties may partly depend on whether the country is a member of intergovernmental organizations and a signatory to human rights treaties (Sikkink and Smith 2002, p. 41). Furthermore, states act as guarantors of internationally recognized principles and international regime obligations (Young 1999, p. 21). As a result, state bureaucracies may expand rather than shrink when countries engage in international cooperation (Boli and Thomas 1999, p. 48). Countries' participation in international NGOs tends to increase following national independence (Boli et al. 1999, p. 61). Few nonstate actors, revolutionary movements included, seek to destabilize the state system altogether (Bell 1971; Josselin and Wallace 2001, p. 258). In fact, when activists pursue transnational strategies, they often invoke citizenship rights, which continue to be grounded in territorial states (Gentile 2003). Finally, even at the domestic level, the notion of civil society's bottom-up flourishing is increasingly countered by skeptics who emphasize the primacy of state structures in shaping popular mobilization (Tarrow 1998; Tilly 1978) and associational life (Skocpol, Ganz and Munson 2000; Crowley and Skocpol 2001; Skocpol 2003). The state-as-filter argument acknowledges the influence of state structures (as well as intergovernmental organizations) on the activities of nonstate actors. Yet it falls short of recognizing the possibility of deliberate entrepreneurship by states and intergovernmental organizations in dealing with nonstate actors. If NGOs can employ government support for their cross-border activities, then the reverse might be true as well (Gilpin 1971, p. 54; Nye and Keohane 1971, p. xx). There are multiple reasons why both states and intergovernmental organizations might act as network engineers. States and intergovernmental organizations may initiate, shape, or fund transnational networks in order to access nonstate actors' technical expertise, moral authority, or constituents (Donnelly 2002). Intergovernmental agencies keen on extending their mandate are likely to cultivate NGO networks as possible clients. Both states and intergovernmental organizations also routinely delegate policy implementation to nonstate actors (Korzeniewicz and Smith 2001; Sassen 2002). Contacts with nonstate actors may function as substitute for democratic legitimation, especially for intergovernmental organizations that lack electoral accountability (Keohane 2002, p. 36; Keohane and Nye 2002). By empirically mapping cross- border ties, my case study seeks to shed a skeptical light on the ability of transnational networks to democratize governance outside the territorial state. Cross-Border Networks: Border Effects and States as Gatekeepers Two dimensions guide my network analysis of EU social policy. The first dimension asks whether governments perform a gatekeeping role in dealing with intergovernmental organizations (IGOs). In a transnational policy network that grants nonstate actors access to decision-making and implementation, we would expect to see direct links between state, nonstate, and EU actors. The second dimension captures border effects in networks. Where transnationalism is at work, we would expect national boundaries to be insignificant. Thus, actors from one country communicate with their peers from another country; organizations such as the EU may serve as the hub of multiparty interactions. By contrast, a network falls short of true transnationalism if contacts tend to be between EU and national actors, with the latter failing to reach across national boundaries to their peers elsewhere. As Table 1 illustrates, the two dimensions generate four possible combinations: (1) The first cell refers to a network in which a government interacts with the intergovernmental organization, but not with other governments directly. Societal actors remain excluded from the interaction. (2) The second scenario is typical of how Neorealists conceive of multilateral organizations: Here, governments negotiate with other governments as well as with the organization as a whole. Nonstate actors may be connected with their own governments and their peers across borders, but are considered irrelevant for policy negotiations. (3) In the third scenario, governments do not control the access of nonstate actors to the intergovernmental organization. However, border effects are relevant in that neither governments nor nonstate actors communicate directly with their peers in other countries. (4) In the fourth scenario, all actors are, at least potentially, connected to one another, regardless of type of organization and country of origin. This is the ideal type of a "global civil society." Figures 1-4 illustrate these four scenarios for the EU. Table 1: Types of Cross-Border Networks Border Effects Yes No Governments as Gatekeepers for Nonstate Actors Yes 1 Bilateral negotiations between a state and an intergovernmental organization 2 Traditional multilateral organizations, such as NATO No 3 Relations between a metropole and its colonies 4 Transnationalism or "global civil society" Fig. 1: Border Effects and Governments as Gatekeepers Notes: Graph produced by NetDraw (Borgatti et al. 2002) EU = EU Institution SA1 etc. = State Actor in Country 1 etc. NSA1 etc. = Nonstate Actor in Country 1 etc. Fig. 2: Governments as gatekeepers, but without border effects Fig. 3: Border effects exist, but states do not act as gatekeepers Fig. 4: Transnational policy network: no border effects, governments do not act as gatekeepers. Case Study: EU Enlargement and Social Policy Transfer to Poland and Hungary The study of EU enlargement and social policy allows us to examine network emergence in a most-likely scenario. At the same time, the controlled paired comparison of Poland and Hungary offers insight into domestic determinants of organizational contacts. European integration presents the highest degree of institutional density to be found anywhere in contemporary international politics. If transnational networks emerge, they are most likely to do so in an environment in which policy-making is shared among states or delegated to joint institutions. Preparations for EU enlargement in 2004 raised the likelihood of cross-border network contacts because all entrants faced the same set of accession requirements and EU preparatory instruments. Most importantly for our purposes, accession states had to demonstrate full compliance with EU law and policies, which meant transposing them into national law and implementing them nationally and sometimes subnationally. The accession preparations raised the likelihood of cross-border contacts among national actors. First, national actors had an incentive to communicate regularly with EU institutions. This was obviously the case for government actors negotiating the accession conditions and demonstrating their country's readiness. But nonstate actors were also likely to benefit from EU contacts, whether through material support or by exchanging information. The EU, in turn, was likely to communicate not only with candidate governments but also with nonstate actors in order to build constituencies for its own work. Similarly, the EU could expect to enhance its monitoring capacity by interacting directly with groups targeted by EU policies. Second, cross- border contacts among peers from different accession countries were likely for two reasons: On one hand, candidate countries faced the same EU demands and could learn from one another's strategies and practices. Governments struggling to adopt a set of EU laws could shop for problem-solving approaches among their peers elsewhere. On the other hand, EU institutions specifically created an infrastructure for cross-border dialogue among actors from the candidate countries. It did so through international conferences and consultative forums, financial assistance for cross-border cooperation, and support for international affiliations of nonstate actors. Both cross-border networks and involvement of nonstate actors were to be expected for another reason as well. Unique among EU policies, social policy grants nonstate actors a specific role in law-making and implementation. Specifically, European social dialogue is a procedure for peak-level consultations between European-level employer and labor organizations with the option of concluding agreements that will lead to EU legislation. This legislation may then be implemented by national labor and employer organizations or by national governments. The social dialogue implicitly requires functioning structures of economic interest representation at the national level, though EU law does not specify what particular structures would satisfy this requirement. In this paper, I focus on Poland and Hungary's accession to the EU. Both countries can be expected to be well-connected to European and international institutions: They were consistently among the most advanced EU accession candidates. Both embarked on the democratic transformation with extensive experience in market socialism and levels of liberalization not seen elsewhere. Both countries were highly integrated with the EU economy even prior to 2004. Thus, in 2000, 83 percent of foreign stock in Hungary's economy came from EU countries, compared to 77 percent of foreign stock in Poland (UNCTAD 2003). Despite these similarities, we can expect Poland and Hungary to vary in terms of cross-border contacts and government gatekeeping due to the fact that the countries differ in size , exposure to the international economy, government stability, and legacies of collective action under state socialism. For a long time, Hungary's economy was more open to foreign direct investment than Poland's. In 2000, FDI stock accounted for 43 percent of Hungary's GDP, compared to 22 percent of Poland's (UNCTAD 2003). Politically, Poland has experienced frequent changes in government since 1989 and faces highly unstable coalitions (Blazyca and Kolkiewicz 1999), whereas Hungary has had relatively stable governments. Poland, with a history of labor-based mass mobilization against state socialism, now has a highly bifurcated landscape of interest representation in which previous allegiance with the regime still constitutes a major cleavage. Hungary experienced a more gradual reform of state socialism from within the regime; present socioeconomic interest representation is highly fragmented on both the labor and employer sides, but less politicized than in Poland (Ekiert and Kubik 1999; Seleny 1999). Hence, while we should expect both countries to have developed extensive cross-border contacts, Hungarian actors are likely to be more connected with other European actors than is the case for their Polish peers. In order to investigate how nonstate actors communicate with governments and EU institutions, I map out patterns of regular contact among organizations with stakes in EU social policy and enlargement. Network analysis allows us to trace where actors turn in order to advance their goals (Scott 2002; Wasserman and Faust 1994). The next section therefore specifies tests for the extent to which governments act as gatekeepers and national borders shape communication patterns. Propositions I use three different network concepts to analyze the extent to which states act as gatekeepers and borders affect interactions in the EU enlargement and social policy network: cliques, structural equivalence, and brokerage. First, tracing cohesive subgroups allows us to compare within- and between-group interactions. The concept of cliques denotes a subgroup of actors that fulfills the following two conditions: (1) All subgroup members have direct ties with one another. (2) There is no other actor in the network with direct ties to all of the subgroup's members (Wasserman and Faust 1994, p. 254). Cliques in a network may overlap, i.e. have members in common. Co-membership of different types of actors in the same cliques indicates opportunities for easy flow of communication among EU, state, and nonstate actors. Co- membership of actors from Poland and Hungary indicates that borders do not inhibit communications between both countries; by contrast, if cliques are limited to actors from one country and EU institutions, border effects are operating. In other words, cohesive subgroups should not cut off nonstate actors from EU institutions, nor should they isolate national actors from their peers in other countries. Proposition 1a: If governments act as gatekeepers, cliques will not contain EU and nonstate actors at once. If governments do not act as gatekeepers, some cliques will contain both EU and nonstate actors. Proposition 1b: If borders inhibit communications, clique membership will be limited to actors from the EU and one country. If communications cross national borders, by contrast, we can expect actors from more than one country to share cliques. Second, at a more general level, we can argue that if governments no longer act as gatekeepers, the communication patterns of nonstate actors should not differ radically from those of EU and state actors. Structural equivalence captures similarities in the network positions of different actors. Actors are structurally equivalent if their ties to and from all other network actors are identical (Wasserman and Faust 1994: 356). Because actors are rarely fully structurally equivalent, existing routines measure the extent to which the network positions of actors converge. I compare profile similarities based on the correlations among actors' ties. Proposition 2a: If governments act as gatekeepers, equivalence clusters are likely to separate EU, state, and nonstate actors. By contrast, if governments do not interfere with nonstate actors' access to the EU, then equivalence clusters will contain actors of several types (nonstate, state, and EU). Proposition 2b: If borders inhibit communications, we can expect clusters to contain national actors from only one country. If communications cross national borders, by contrast, we can expect equivalence clusters to contain national actors from more than one country. Finally, some organizations are likely to stand out as vital to the network, connecting actors that may not have any direct way of communicating. Intermediaries have particular control over communications among other actors. A measure that captures inter-group mediation derives from the idea of brokerage. According to Gould and Fernandez (1989, pp. 92-94), a broker is any actor who assists communications in a network whose participants differ in their interests and activities. This differentiation implies that exchanges vary in meaning depending on the actors involved. Brokerage, specifically a measure termed liaison, allows us to examine who facilitates between-group communication among EU, state, and nonstate actors. Given three actors and a network partition into non-overlapping groups, Gould and Fernandez define a liaison as an actor who facilitates communications between two separate groups. All three actors belong to different groups. We can formulate the following propositions about nonstate actor participation and border effects: Proposition 3a: If governments control the access of nonstate actors to the EU, we can expect them to have higher liaison scores than either EU or nonstate actors. If governments do not mediate between the EU and societal actors, EU and government actors will not differ systematically in their liaison scores. Proposition 3b: If borders inhibit communications between actors from Poland and Hungary, then we might expect the EU to function as liaison to mediate between the two countries. If border effects are nonexistent, we would not expect the EU to score systematically higher than other actors. Having specified expectations for the three measures of cliques, structural equivalence, and brokerage, I will now discuss the results of the network analysis. Data Analysis Despite the inherently fluid character of the network around EU enlargement and social policy, the use of quantitative measures requires the imposition of some type of boundary. The actors in this network were selected for their presumptive theoretical relevance. Thus, the purpose of the investigation is whether actors who should theoretically participate in the transfer of EU social policy to Poland and Hungary are in fact doing so. Of the thirty-two actors included here, four are EU institutions; two belong to the Polish government and four to the Hungarian government; four are Polish and twelve are Hungarian nonstate actors (employer and labor organizations); three represent Europe-wide nonstate actors; and three belong to other categories (see Appendix 1 for a complete list of organizations and their abbreviations). Ties were measured via a survey that asked respondents (one representative per organization) to indicate with which of the thirty-two actors they were in regular contact (see Appendix 2 for a discussion of the survey). The analysis was conducted with UCINET 6 (Borgatti, Everett and Freeman 2002). Cliques There are thirty-two cliques with a minimum of four members. Tables 2 and 3 show the frequency of clique co-memberships for four types of organizations (EU, state, nonstate, other) and four types of origin (EU, Poland, Hungary, other). Examining the cliques yields three insights: First, clique patterns differ significantly between Poland and Hungary. Hungary has nine cliques limited to actors from Hungary, while Poland has none that are limited to Polish actors. This is largely due to the differences in the organizational make-up of interest representation in both countries. Poland has two main employer and labor organizations, respectively. Ideological competition divides both labor and employer federations between organizations founded in opposition to state socialism and those that conformed to the system prior to 1989. Hungary, by contrast, has nine employer and six labor federations. Thus, Hungary's extreme organizational fragmentation increases the likelihood of intra-country cliques. In Poland, by contrast, the smaller number of organizations and the fierce competition between the two labor and employers' organizations, respectively, reduce the likelihood of intra- country cliques. Second, Hungarian nonstate actors share seven clique memberships with EU actors, compared to two for their Polish equivalents. Hungarian state actors share nine clique memberships with EU actors, compared to three for Polish state actors. The larger number of Hungarian government actors probably accounts for part of this difference, but it also suggests that Hungarian actors are better connected at the international level. Third, the EU is party to 18 cliques in terms of organization type and 17 cliques in terms of organizational origin. This makes the EU the focal point in the network. Table 2: Frequency of Clique Co-Memberships by Organization Type Clique combinations, by organization type Frequency SA, NSA 7 EU, SA 6 EU, NSA, O 5 EU, SA, NSA 5 NSA, O 5 EU, O 1 EU, SA, NSA, O 1 SA, NSA, O 1 EU 1 Note: EU = EU institution; SA = state actor; NSA = nonstate actor; O = other. Table 3: Frequency of Clique Co-Memberships by Origin of Organization Clique combinations, by origin Frequency HU 9 EU, HU 6 HU, EU, O 4 HU, O 4 EU, PL, O 2 EU, PL 2 EU 1 HU, PL, O 1 PL, O 1 EU, O 1 EU, PL, HU 1 Note: EU = EU institution; HU = Hungary; PL = Poland; O = other. Following Proposition 1a, if governments act as gatekeepers, we would expect EU and nonstate actors not to share clique memberships. As Table 2 shows, there are a total of eleven cliques in which both EU and nonstate actors are members. Thus, it appears that nonstate actors can easily access EU institutions without the mediation of their governments. Following Proposition 1b, if borders inhibit communications, we would expect clique membership to be limited to actors from one country and the EU. Table 3 shows that there are only two cliques in which both Polish and Hungarian actors participate. This is a rather small number and suggests that border effects might in fact be operating. Table 4 shows the results for associations between clique co-membership and organization types. Three variables describe actor attributes: organization type, state versus nonstate actor, and country of origin. I use two measures of association: Pearson's correlation and multiple regression quadratic assignment procedure (QAP regression). QAP regression regresses a dependent matrix on one or more independent matrices. In Table 3, the dependent matrix is that of clique co-memberships, the independent matrix for each bivariate regression is based on actor attributes ("organization type", "state/nonstate", and "origin", respectively). Clique co-membership is not significantly associated with organization type or the state/nonstate actor distinction. In other words, EU, state, and nonstate actors frequently overlap in their clique memberships, as expected in a network that allows nonstate actors direct access to the EU. However, there turns out to be a statistically significant relationship between clique co- membership and organizations' origin. That is, cliques link actors at the national and the EU level, but rarely connect Polish and Hungarian actors directly. There are two exceptions: One clique includes the Polish and Hungarian delegations to Brussels. Here we can suspect geographic proximity to aid in cross-national communication. The other exception is a clique that links the Polish and Hungarian post-Communist successor unions (OPZZ and MSZOSZ). Their tie goes back to cross-national networking among trade unions under state socialism (author's interview, Budapest, 11 February 2000). Both the substantive analysis of clique patterns and the statistical analysis of clique co- memberships yield the same results: Governments do not act as gatekeepers, but border effects impede communications between actors from Poland and Hungary. Thus, the EU is the focal point of networking by state and nonstate actors, but neither government nor interest groups tend to seek out their peers in other candidate countries. This finding is important. It implies that even under conditions of extensive networking among intergovernmental, state, and nonstate actors, relations may remain purely bilateral. Intergovernmental organizations may set up opportunities for networking, but incentive structures do not necessarily favor cross-border interactions. The EU, it seems, encourages activism directed toward Brussels, but has not provided any incentives for actors from several countries to interact directly with one another. Table 4: Measures of Association between Clique Co-Membership and Organizational Characteristics (Thirty-Two Actor Network, Clique Size: 4, Number of Cliques: 32) Organization Type State/Nonstate Origin Pearson's Correlation (Significance Level) .021 (.387) .046 (.183) .259 (.000)** R-square1 (Probability)2 .000 (.818) .002 (.362) .067 (.001)** Note: N = 992. Statistical significance: * p = .005, ** p = .001, two-tailed test. R-square based on QAP regression. Probability based on the proportion of random trials yielding an R-square as large as or larger than observed. Structural Equivalence If governments intercede between nonstate actors and EU institutions, we can expect the relational profiles of actors to differ systematically according to their organization type (Proposition 2a). Similarly, if borders inhibit communications, then we can expect equivalence clusters to separate actors from different countries (Proposition 2b). As a measure of structural equivalence in the patterns of ties, I use the Pearson product correlation coefficient of every pair of profiles (Wasserman and Faust, pp. 368-370). Table 5 shows the partition resulting from splitting actors based on convergence of iterated correlations. Clearly, border effects separate the profiles of Hungarian and Polish actors. Furthermore, there is some evidence of states acting as gatekeepers. Although each cluster contains actors of more than one type, there are no clusters containing both EU and nonstate actors. In other words, while the profiles of nonstate actors resemble some state actors, they do not resemble EU institutions. Note that there is some similarity in the profiles of EU institutions and certain two Polish government actors (cluster 1), as well as in the profiles of certain nonstate actors and two international groups (cluster 3). Table 5: Structural Equivalence Clusters (CONCOR) Cluster Organization Name Organization Type Origin 1 K-DGEnl-HU K-DGEnl-Pl K-DGESA PHARE PL-Mlab PL Del Bx EU EU EU EU State State international international international international Poland Poland 2 PL-NSZZ PL-OPZZ ETUC PL-PKPP PL-KPP UNICE CEEP ILO-CEET FES Nonstate Nonstate Nonstate Nonstate Nonstate Nonstate Nonstate Other Other Poland Poland international Poland Poland international international international international Cont'd next page. Cluster Organization Name Organization Type Origin 3 HU-PM HU-MFSA HU Del BX HU-Liga HU-MOSZ HU-ASZSZ WB-HU State State State Nonstate Nonstate Nonstate Other Hungary Hungary Hungary Hungary Hungary Hungary international 4 HU-MGYOSZ HU-MFA HU-AMSZ HU-VOSZ HU-AFEOSZ HU-MSZOSZ HU-OKISZ HU-IPOSZ HU-CEHIC HU-MOSZ(E) Nonstate State Nonstate Nonstate Nonstate Nonstate Nonstate Nonstate Nonstate Nonstate Hungary Hungary Hungary Hungary Hungary Hungary Hungary Hungary Hungary Hungary Note: Self-ties ignored. Depth of splits: 2. Convergence criteria: 0.2. R2 = 0.207. Liaison The liaison measure allows us to estimate the extent to which any actor controls the flow of communication among groups in the network. Table 6 shows two sets of weighted liaison scores, i.e. by organization type and origin. Nonstate actors score low on both measures. This makes sense: Even in a highly participatory network, we would not expect them to aid communication between governments and the EU. Four of the five highest-scoring liaisons between organization types are EU or international actors who help nonstate actors communicate with their governments. Two EU institutions serve as principal mediators between nonstate actors and governments: the Commission DG on Employment and Social Affairs and PHARE, a funding agency assisting countries in preparing for accession. Two other international actors, the International Labor Organization's Central and Eastern Europe Team and the German Friedrich Ebert Foundation, also act as important mediators between governments and nonstate actors. Following Proposition 3A, we can conclude that governments do not act as gatekeepers that intercede between the EU and nonstate actors. Among state actors, only the Hungarian Ministry of Foreign Affairs stands out as a significant liaison. Given the usual image of foreign ministries as operating at the intergovernmental level, this is somewhat surprising. It suggests that Hungarian nonstate actors can easily approach a ministry normally concerned with diplomatic contacts, but they do not depend on the ministry's mediation in order to be heard by EU institutions. As for cross-border communication, international actors are clearly important as mediators between Poland and Hungary. Four of them stand out: The Commission DG on Employment and Social Affairs leads once again, followed by the ILO's Central and Eastern Europe Team, the Friedrich Ebert Foundation, and PHARE. European umbrella organizations of labor and employers are relatively less important as mediators among actors from Poland and Hungary. Following Proposition 3b, the evidence points in the direction of significant border effects. The analysis of variance (Table 7) shows organization type to be significantly associated with liaisons between types of organization. Consistent with Proposition 3a, this provides evidence of a network in which EU actors broker relations between state and nonstate actors. Rather than states acting as gatekeepers for contacts between the EU and nonstate actors, it is the EU that occupies this influential position as a mediator. Liaisons among countries differ systematically by organizational origin, making international actors the only relevant liaisons in the network. This means that for Polish and Hungarian actors to communicate, they usually require an international mediator. Consistent with Proposition 3b, this means that border effects are operating. Table 6: Weighted Liaison Scores (Matrix "A reports any tie with B") Organization (Type) (Origin) Liaison by Organization Type Liaison by Origin K-DGESA (EU) (international) 47.85 45.38 PHARE (EU) (international) 20.56 15.08 K-DGEnl-HU (EU) (international) 1.36 0 K-DGEnl-PL (EU) (international) .73 .25 HU MFA (state) (Hungary) 13.58 .25 HU Del BX (state) (Hungary) 7.80 .33 HU MFSA (state) (Hungary) 6.66 0 HU-PM (state) (Hungary) 4.62 0 PL Del BX (state) (Poland) 4.08 .5 PL Mlab (state) (Poland) 3.57 .06 PL-OPZZ (nonstate) (Poland) 1.80 .61 HU-Liga (nonstate) (Hungary) 1.75 .34 HU-MOSZ (nonstate) (Hungary) 1.75 .34 PL-PKPP (nonstate) (Poland) 1.36 .70 HU-MSZOSZ (nonstate) (Hungary) 1.28 .34 PL-NSZZ (nonstate) (Poland) 1.12 .64 UNICE (nonstate) (international) 1.04 3.42 Table 6 Cont'd Organization (Type) (Origin) Liaison by Organization Type Liaison by Origin HU-CEHIC (nonstate) (Hungary) .91 .2 HU-AMSZ (nonstate) (Hungary) .54 0 ETUC (nonstate) (international) .52 3.73 HU-ASZSZ (nonstate) (Hungary) .50 0 HU-VOSZ (nonstate) (Hungary) .43 0 PL-KPP (nonstate) (Poland) .41 .14 HU-IPOSZ (nonstate) (Hungary) .32 0 HU-MGYOSZ (nonstate) (Hungary) .29 0 HU-OKISZ (nonstate) (Hungary) .16 0 HU-AFEOSZ (nonstate) (Hungary) .06 0 HU-MOSZ(E) (nonstate) (Hungary) .00 0 CEEP (nonstate) (international) .00 0 ILO-CEET (other) (international) 13.57 26.11 FES (other) (international) 13.08 20.21 WB-HU (other) (international) 1.16 0 Mean (N = 32) 4.78 3.71 Standard Deviation 9.33 9.78 Note: Individual scores are based on N = 32 x 31 nodes = 992 observations. Table 7: Analysis of Variance for Liaison Scores Grouping Variable Liaison by Organization Type Liaison by Origin Organization Type: Chi-Square (Significance Level) 17.04 (.001)*** 3.27 (.352) Origin: Chi-Square (Significance Level) 2.04 (.361) 9.40 (.009)** Note: N = 32. ANOVA based on Kruskal Wallis test for nonparametric data. Statistical significance: * p = .05; ** p = .01; *** p = .001. We can now sum up the data analysis (see Table 8). First, there is considerable evidence that nonstate actors participate in the network and can access both governments and EU institutions to exchange information and voice preferences. States, in other words, do not act as gatekeepers. Second, border effects separate Polish and Hungarian actors. Candidate countries apparently did not seek each other out for consultation on negotiations with the EU concerning their adoption of EU social policy. Instead, contacts in the network are directed toward Brussels. Regardless of countless opportunities for labor and employers to interact with their peers from other candidate countries (such as in the context of European peak organizations or EU- sponsored conferences), actors appear to have had few incentives to build links among candidate countries. Third, one EU agency (the Commission DG-Employment and Social Affairs) controls much of the communication flow among groups in the network, suggesting that the initiative of network construction may lie with intergovernmental and/or state actors, an aspect often neglected in scholarship on transnational advocacy. The data also show considerable variation in the network positions of actors within any given group. Both EU and state agencies have a division of labor that puts some at the forefront of network communications, while others engage solely in intergovernmental negotiations. Whether nonstate actors can have any substantive impact on policy formation and implementation may depend on political incentive structures that are exogenous to the specific policy network. In the case of EU enlargement and social policy, the intergovernmental framework of accession negotiations appears to have trumped the structural network access that nonstate actors had. The strategic context outside of the network may have favored competition among candidates for a speedy accession rather than cooperation in the face of similar challenges. Table 8: Summary of Network Evidence Indicator States as Gatekeepers Border Effects Cliques No Yes Structural Equivalence Limited Yes Brokerage No Yes Extending the Network Research: Ongoing Network Survey of 28 Countries The analysis of pre-accession communications in social policy among the EU, Poland and Hungary reveals important border effects. But to what extent is this pattern representative of the EU social policy network more generally? My ongoing research on social policy communications among approximately 180 organizations from the EU, member states, and the three candidate countries of Bulgaria, Romania, and Turkey investigates whether border effects operate primarily in candidate countries or also among member states. The selection criteria for organizations to be included are the same as in the earlier project: Do the players whom we should theoretically expect to be most active in EU social policy indeed participate in the network? While this nominalist approach risks omitting potentially important actors (e.g. nongovernmental organizations representing the rights of the unemployed), it tests whether the EU's own rules for social dialogue (whereby certain nonstate actors are privileged over others) do indeed apply to communications in social policy. The logic is again one of a most-likely test case: If the designated social partners, i.e. employer and labor organizations, do not form a transnational social policy network with governments and EU institutions, then we have even more reason to be skeptical about the ability of less privileged nonstate actors to do so. Unlike the earlier network data on Poland and Hungary, which were gathered primarily in face-to-face interviews, the current survey is being conducted online. It is too early to tell how this affects the response rate and the quality of responses, but collecting the same data among 180 organizations in 28 countries is understandably difficult without the aid of either mail or electronic questionnaires. Conclusion: Implications and Directions for Further Research Analyzing the apparently vibrant network around EU social policy and enlargement highlights the limits of cross-border communication. This case combines multiple factors that make it unlikely that states act as gatekeepers and national borders restrain communications. European integration constitutes the world's most highly institutionalized environment of cross- border exchanges. EU social policy grants nonstate actors an explicit role in policy formulation and implementation. Candidate countries for the 2004 EU enlargement faced the same set of demands and were likely to benefit from sharing information about best practices and rule adoption strategies. All these factors promote a network in which nonstate actors participate alongside state and EU agencies and where actors of several countries reach out to one another. Yet the evidence shows that, although nonstate actors can access state and EU actors, cross- border communications among national actors remain scarce. In this network, national actors did not seek out their peers from other countries, but instead directed their efforts toward Brussels. In other words, the mere availability of structural opportunities for cross-border networking was insufficient for generating such ties. Furthermore, nonstate actors did not take the lead in network construction. As the brokerage measures indicate, certain EU and other international actors controlled communication flows in the network, whereas nonstate actors tended to be at the receiving end. This contrasts sharply with the bottom-up view of transnational relations that prevails in much of the research on activist networks. This study illustrates that we need to differentiate between network structures and exogenous political opportunities if we are to understand transnational politics. Political incentives ultimately set by states and intergovernmental organizations may reward bilateralism (such as between the EU and national actors) regardless of ample opportunities for cross-border networking. Just as states influence the conditions that guide how nonstate actors mobilize, intergovernmental organizations do the same at the international level. Furthermore, by offering actual measurement of interorganizational contacts in transnational space, this study advances the debate on how best to trace cross-border networks systematically in a variety of contexts. Beyond the ongoing survey research among 28 countries, there are several ways of extending this research on transnational political mobilization: 1) by studying networks of varying sizes; 2) by comparing networks in different sectors. Extending the sample will add variation in terms of economic and political integration, past experience with mass mobilization and socioeconomic interest organizations, and progress toward democracy and market capitalism. All these factors would seem, prima facie, to increase the likelihood that states do not act as gatekeepers and communication flows freely across borders. Research on cross-border networks in other policy areas would add variation in terms of institutionalization of interest representation and modes of political mobilization. Whereas labor and employers tend to mobilize via elections and close ties with political parties, NGOs often choose a range of unconventional tactics (Josselin 2001b, p. 180). Cross-sectoral comparisons would also help us understand the influence of substantive policies. For instance, the cost of implementing EU social policy in the new member states may be negligible, therefore triggering little activism, whereas environmental policy requires significant investments for decades to come and is likely to divert resources from other acute areas of need. In sum, the present study offers three important insights about cross-border politics: First, NGO interactions with one another as well as with states and intergovernmental organizations may remain nationally distinct despite formal opportunities for cross-border exchange. Second, networks may be the result of careful construction by intergovernmental agencies that place themselves at the center of interactions. By supporting NGO activities, intergovernmental organizations may create constituencies that will in turn provide political backing for institutional expansion. Third, political opportunity structures are seldom fully contained within the network of cross-border communications. Diverse intergovernmental agencies relate differently to NGOs. One agency's close relations with NGOs do not automatically produce a network in which nonstate actors participate freely and national actors communicate with their peers from other countries. Intergovernmentalism may prevail even when individual agencies successfully build networks in which nonstate actors participate. Appendix 1: Table 9: Actors Included in the Network Organization Name Abbreviation EU Institutions Commission Directorate General for Enlargement, Hungary Team K-DGEnl-HU Commission Directorate General for Enlargement, Poland Team K-DGEnl-PL Commission Directorate General for Employment and Social Affairs K-DGESA Commission Directorate General for Enlargement, PHARE PHARE Polish Government Institutions Polish Ministry of Labor and Social Affairs PL-Mlab Poland's Delegation to the EU PL-Del-BX Hungarian Government Institutions Hungary – Office of the Prime Minister HU-PM Hungarian Ministry of Foreign Affairs HU-MFA Hungarian Ministry of Family and Social Affairs HU-MFSA Hungary's Delegation to the EU (HU-Del-BX) Hungarian Trade Unions Democratic League of Independent Trade Unions HU-Liga National Association of Hungarian Trade Unions HU-MSZOSZ Autonomous Trade Unions' Confederation ASZSZ HU-ASZSZ National Alliance of Workers' Councils HU-MOSZ Polish Trade Unions NSZZ Solidarity PL-NSZZ All-Poland Alliance of Trade Unions PL-OPZZ Hungarian Employers' Organizations National Association of Entrepreneurs HU-VOSZ National Federation of Consumer Cooperatives HU-AFEOSZ Federation of Hungarian Manufacturers HU-MGYOSZ Hungarian Industrial Association HU-OKISZ Hungarian Association of Craftsmen's Corporations HU-IPOSZ Confederation of Hungarian Employers' Organizations for International Cooperation HU-CEHIC National Federation of Agricultural Cooperatives and Producers HU-MOSZ/E Union of Agrarian Employers HU-AMSZ Polish Employers' Organizations Polish Confederation of Private Employers PL-PKPP Confederation of Polish Employers PL-KPP Table 9 Cont'd Organization Name Abbreviation European Peak-Level Organizations European Trade Union Confederation ETUC Union of Industrial and Employers Confederations of Europe UNICE European Center of Enterprises with Public Participation and of Enterprises of General Public Interest CEEP Others ILO Central and Eastern Europe Team ILO-CEET World Bank – Hungary Office WB-HU Friedrich Ebert Foundation FES Appendix 2: Constructing the Database The network survey was administered between February 2000 and May 2001. It documents ties as reported by one representative of each organization. Information for each node was gathered separately in order to cross-check mutual actor perceptions of a given directional tie (Pappi and Henning 1999). Thus, each actor was asked whether it was the source as well as the target of regular contacts with every other actor, as well as whether it was a member of any of the other organizations in the network. The information from the three directional relations ("we initiate regular contact," "they initiate regular contact," and "we are members in" ) was compacted into a matrix with 32 x 31 (= 992) observations (ignoring self-ties because these are meaningless here). This matrix ("A reports any tie with B") registers any tie an organization reported with another actor. It is thus based on directed ties, but leaves open whether the organization in question is the source or the recipient of contacts or membership affiliations. In order to increase reliability, I symmetrized this matrix by counting only those ties on whose existence source and target agreed. Thus, only reciprocal ties are registered. The resulting matrix ("confirmed ties") is a conservative estimate of regular contact and affiliations within the network and serves as the basis of the analysis presented here. The matrix of confirmed ties has a density of .27, which means that twenty-seven percent of all possible ties are actually present in the network. Except for brokerage measures, which require directional ties, the analysis was performed on the matrix of confirmed ties. Brokerage scores were measured on the matrix "A reports any tie with B." References Ansell, Christopher T. 2000. "The Networked Polity: Regional Development in Western Europe. Governance 13 (3): 303-333. Ansell, Christopher T., Craig A. Parsons, and Keith A. Darden. 1997. "Dual Networks in European Regional Development Policy." Journal of Common Market Studies 35 (3):347-375. Batliwala, Srilatha. 2002. "Grassroots Movements as Transnational Actors: Implications for Global Civil Society." Voluntas 13 (4): 393-409. Bell, J. Bowyer. 1971. "Contemporary Revolutionary Organizations." Pp. 153-168 in Transnational Relations and World Politics, edited by Robert O. Keohane and Joseph S. Nye, Jr. Cambridge: Harvard University Press. Blazyca, George, and Marek Kolkiewicz. 1999. "Poland and the EU: Internal Disputes, Domestic Politics and Accession." Journal of Communist Studies and Transition Politics 15 (4): 131-143. Bogason, Peter, and Theo A.J. Toonen. 1998. "Introduction: Networks in Public Administration." Public Administration 76 (Summer): 205-227. Boli, John, and George M. Thomas, eds. 1999. Constructing World Culture: International Nongovernmental Organizations since 1870. Stanford: Stanford University Press. Boli, John, Thomas E. Loya, and Teresa Loftin. 1999. "National Participation in World- Polity Organization." Pp. 50-77 in Constructing World Culture: International Nongovernmental Organizations Since 1870, edited by John Boli and George M. Thomas. Stanford: Stanford University Press. Borgatti, Steven P., Martin G. Everett, and Linton C. Freeman, L.C. 2002. Ucinet 6 for Windows: Software for Social Network Analysis. Harvard, Mass.: Analytic Technologies. Breiger, Ronald, S. Boorman, and Paul Arabie. 1975. "An Algorithm for Clustering Relational Data, with Applications to Social Network Analysis and Comparison with Multi-Dimensional Scaling." Journal of Mathematical Psychology 12: 328- 383. Clark, Ann-Marie, Elisabeth J. Friedman, and Kathryn Hochstetler. 1998. "The Sovereign Limits of Global Civil Society: A Comparison of NGO Participation in UN World Conferences on the Environment, Human Rights, and Women." World Politics 51 (1): 1-35. Coleman, William D. 2001. "Policy Networks, Non-State Actors and Internationalized Policy-Making: A Case Study of Agricultural Trade." Pp. 93-112 in Non-State Actors in World Politics, edited by Daphné Josselin and William Wallace. New York: Palgrave. Crowley, Jocelyn Elise, and Theda Skocpol. 2001. "The Rush to Organize: Explaining Associational Formation in the United States, 1860s-1920s." American Journal of Political Science 45 (4): 813-829. Donnelly, Elizabeth. 2002. "Proclaiming Jubilee: The Debt and Structural Adjustment Network." Pp. 154-180 in Restructuring World Politics: Transnational Social Movements, Networks, and Norms, edited by Sanjeev Khagram, James V. Riker, and Kathryn Sikkink. Minneapolis: University of Minnesota Press. Edwards, Michael, and John Gaventa, eds. 2001. Global Citizen Action. Boulder: Lynne Rienner. Eising, Rainer, and Beate Kohler-Koch. 1999. "Introduction: Network Governance in the European Union." Pp. 3-13 in The Transformation of Governance in the European Union, edited by Beate Kohler-Koch and Rainer Eising. London: Routledge. Ekiert, Grzegorz, and Jan Kubik. 1999. Rebellious Civil Society: Popular Protest and Democratic Consolidation in Poland, 1989-1993. Ann Arbor: University of Michigan Press. Fernandez, Roberto M., and Roger V. Gould. 1994. "A Dilemma of State Power: Brokerage and Influence in the National Health Policy Domain." American Journal of Sociology 99 (6): 1455-1491. Fiorini, Ann M., ed. 2000. The Third Force: The Rise of Transnational Civil Society. Washington, D.C.: Carnegie Endowment for International Peace. Fox, Jonathan A., and L. David Brown, eds. 1998. The Struggle for Accountability: The World Bank, NGOs, and Grassroots Movements. Cambridge: MIT Press. Gelb, Joyce. 2002. "Feminism, NGOs, and the Impact of the New Transnationalisms." University of California International and Area Studies Digital Collection 1 (9), available at (accessed 11/18/03). Gentile, Antonina. 2003. "The Neo-Liberalising State: Restructuring Domestic and Transnational Labor Contention—The Case of Dockers." Paper presented at the Social Science Historical Association Convention, Baltimore, 13-16 November. Gilpin, Robert. 1971. "The Politics of Transnational Economic Relations." Pp. 48-69 in Transnational Relations and World Politics, edited by Robert O. Keohane and Joseph S. Nye. Cambridge: Harvard University Press. Gordon, Michael E., and Lowell Turner, eds. 2000. Transnational Cooperation among Trade Unions. Ithaca: Cornell University Press. Gould, Roger V., and Roberto M. Fernandez. 1989. "Structures of Mediation: A Formal Approach to Brokerage in Transaction Networks." Sociological Methodology 19: 89-129. Guidry, John A., Michael D. Kennedy, and Mayer N. Zald, eds. 2000. Globalizations and Social Movements: Culture, Power and the Transnational Public Sphere. Ann Arbor: Michigan University Press. Halliday, Fred. 2001. "The Romance of Non-State Actors." Pp. 21-37 in Non-State Actors in World Politics, edited by Daphné Josselin and William Wallace. New York: Palgrave. Hulme, David, and Michael Edwards, eds. 1997. NGOs, States and Donors: Too Close for Comfort? New York: St. Martin's Press. Huntington, Samuel. 1973. "Transnational Organizations in World Politics." World Politics 25 (3): 333-368. Josselin, Daphné. 2001a. "Non-State Actors in World Politics: A Framework." Pp. 1-20 in Non-State Actors in World Politics, edited by Daphné Josselin and William Wallace. New York: Palgrave. Josselin, Daphné. 2001b. "Back to the Front Line? Trade Unions in a Global Age." Pp. 169-186 in Non-State Actors in World Politics, edited by Daphné Josselin and William Wallace. New York: Palgrave. Josselin, Daphné, and William Wallace. 2001. "Non-State Actors in World Politics: The Lessons." Pp. 251-260 in Non-State Actors in World Politics, edited by Daphné Josselin and William Wallace. New York: Palgrave. Keck, Margaret E., and Kathryn Sikkink. 1998. Activists Beyond Borders: Advocacy Networks in International Politics. Ithaca: Cornell University Press. Kenis, Patrick, and Volker Schneider. 1991. "Policy Networks and Policy Analysis: Scrutinizing a New Analytical Toolbox." Pp. 25-59 in Policy Networks: Empirical Evidence and Theoretical Considerations, edited by B. Martin and Renate Mayntz, 25-59. Frankfurt/Main: Campus. Keohane, Robert O., and Joseph S. Nye, eds. 1971. Transnational Relations and World Politics. Cambridge: Harvard University Press. Keohane, Robert O. 2002. Power and Governance in a Partially Globalized World. London/New York: Routledge. Keohane, Robert O., and Joseph S. Nye. 2002. "The Club Model of Multilateral Cooperation and Problems of Democratic Legitimacy." Pp. 219-244 in Power and Governance in a Partially Globalized World, edited by Robert O. Keohane. London/New York: Routledge. Knoke, David, Franz Urban Pappi, Jeffrey Broadbent, and Yutaka Tsujinaka. 1996. Comparing Policy Networks: Labor Politics in the U.S., Germany, and Japan. New York: Cambridge University Press. Korzeniewicz, Roberto Patricio, and William C. Smith. 2001. "Protest and Collaboration: Transnational Civil Society Networks and the Politics of Summitry and Free Trade in the Americas." The North-South Agenda Papers No. 51 (September). Coral Gables: University of Miami. Krasner, Stephen. 1995. "Power Politics and Transnational Relations." Pp. 257-279 in Bringing Transnational Relations Back In, edited by Thomas Risse-Kappen. New York: Cambridge University Press. Laumann, Edward O., and David Knoke. 1987. The Organizational State. Madison: University of Wisconsin Press. Laumann, Edward O., P.W. Dayton, and P. Greenwald. 1983. "The Boundary Specification Problem in Network Analysis." Pp. 18-34 in Applied Network Analysis, edited by Ron Burt and M.J. Minor. Beverly Hills: Sage. Nye, Joseph S., and Robert O. Keohane. 1971. "Transnational Relations and World Politics: An Introduction." Pp. ix-xxix in Transnational Relations and World Politics, edited by Robert O. Keohane and Joseph S. Nye. Cambridge: Harvard University Press. O'Brien, Robert, Anne Marie Goetz, Jan Aart Scholte, and Marc Williams. 2000. Contesting Global Governance: Multilateral Economic Institutions and Global Social Movements. Cambridge: Cambridge University Press. Pappi, Franz Urban, and Christina H.C.A. Henning. 1999. "The Organization of Influence on the EC's Common Agricultural Policy: A Network Approach." European Journal of Political Research 36: 257-281. Rhodes, R.A.W. 1996. "The New Governance: Governing without Government." Political Studies 44: 652-667. Rhodes, R.A.W. 1997. Understanding Governance: Policy Networks, Governance, Reflexivity, and Accountability. Philadelphia: Open University Press. Riker, James, V. 2002. "NGOs, Transnational Networks, International Donor Agencies, and Prospects for Democratic Governance in Indonesia." Pp. 181-205 in Restructuring World Politics: Transnational Social Movements, Networks, and Norms, edited by Sanjeev Khagram, James V. Riker, and Kathryn Sikkink. Minneapolis: University of Minnesota Press. Riles, Annelise. 2000. The Network Inside Out. Ann Arbor: University of Michigan Press. Risse, Thomas, Stephen C. Ropp, and Kathryn Sikkink, eds. 1999. The Power of Human Rights: International Norms and Domestic Change. Cambridge: Cambridge University Press. Sassen, Saskia. 2002. "The State and Globalization." Pp. 91-112 in The Emergence of Private Authority in Global Governance, edited by Rodney Bruce Hall and Thomas J. Biersteker. New York: Cambridge University Press. Sassen, Saskia. 2003. "Globalization or Denationalization?" Review of International Political Economy 10 (1): 1-22. Scott, John. 2000. Social Network Analysis: A Handbook, 2nd ed. London: Sage. Seleny, Anna. 1999. "Old Political Rationalities and New Democracies." World Politics 51 (4): 484-519. Sikkink, Kathryn, and Jackie Smith. 2002. "Infrastructures for Change: Transnational Organizations, 1953-93." Pp. 24-44 in Restructuring World Politics: Transnational Social Movements, Networks, and Norms, edited by Sanjeev Khagram, James V. Riker, and Kathryn Sikkink. Minneapolis: University of Minnesota Press. Skjelsbaek, Kjell. 1971. "The Growth of International Nongovernmental Organization in the Twentieth Century." Pp. 70-92 in Transnational Relations and World Politics, edited by Robert O. Keohane and Joseph S. Nye. Cambridge: Harvard University Press. Skocpol, Theda, Marshall Ganz, and Ziad Munson. 2000. "A Nation of Organizers: The Institutional Origins of Civic Voluntarism in the United States." American Political Science Review 94 (3):527-546. Skocpol, Theda. 2003. Diminished Democracy: From Membership to Management in American Civic Life. Norman: University of Oklahoma Press. Smith, Jackie, Charles Chatfield, and Ron Pagnucco, eds. 1997. Transnational Social Movements: Solidarity Beyond the State. Syracuse: Syracuse University Press. Smith, Jackie, and Hank Johnston. 2002. "Globalization and Resistance: An Introduction." Pp. 1-10 in Globalization and Resistance: Transnational Dimensions of Social Movements, edited by Jackie Smith and Hank Johnston, 1- 10. Lanham: Rowman and Littlefield. Strange, Susan. 1988. States and Markets: An Introduction to International Political Economy. London: Pinter. Tarrow, Sidney. 1998. Power in Movement: Social Movements and Contentious Politics. Cambridge: Cambridge University Press. Tarrow, Sidney. 2001a. "Contentious Politics in a Composite Polity." Pp. 233-259 in Contentious Europeans: Protest and Politics in an Emerging Polity, edited by Douglas Imig and Sidney Tarrow. Lanham, Md.: Rowman and Littlefield. Tarrow, Sidney. 2001b. "Transnational Politics: Contention and Institutions in International Politics." Annual Review of Political Science 4: 1-20. Tilly, Charles. 1978. From Mobilization to Revolution. Reading, Mass.: Addison-Wesley. True, Jacqui, and Michael Mintrom. 2001. "Transnational Networks and Policy Diffusion: The Case of Gender Mainstreaming." International Studies Quarterly 45 (1): 27-57. UNCTAD. 2003. World Investment Directory. Geneva: UNCTAD. Vallier, Ivan. 1971. "The Roman Catholic Church: A Transnational Actor." Pp. 129-152 in Transnational Relations and World Politics, edited by Robert O. Keohane and Joseph S. Nye. Cambridge: Harvard University Press. Waldinger, Roger, and David Fitzgerald. 2004. "Transnationalism in Question." American Journal of Sociology 109 (5):1177-1195. Wapner, Paul. 1997. "Governance in Global Civil Society." Pp. 65-84 in Global Governance: Drawing Insights from the Environmental Experience, edited by Oran B. Young. Cambridge: MIT Press. Wasserman, Stanley, and Katherine Faust. 1994. Social Network Analysis: Methods and Applications. New York: Cambridge University Press. Young, Oran B., ed. 1997. Global Governance: Drawing Insights from the Environmental Experience. Cambridge: MIT Press. Young, Oran B. 1999. Governance in World Affairs. Ithaca: Cornell University Press. The legal basis of EU social policy consists of the Treaty on European Union (OJ C340 [10.11.1997]: 145-172, Art. 2 [consolidated version]) and the Treaty establishing the European Community (in force since 1/1/1958, OJ C 340 [10.11.1997]: 173-308 [consolidated version], esp. articles 39-42, 125-130, 136-150, 152, and 158-162. The social dialogue was established through the Agreement on Social Policy of 1991, annexed to the Maastricht Treaty via the Protocol on Social Policy and incorporated into the Treaty on European Union via the Amsterdam Treaty (Art. 138-139 TEC). For pragmatic reasons, I adopt Josselin's inclusive definition of nonstate actors as organizations that are "largely or entirely autonomous from central government funding and control: emanating from civil society, or from the market economy, or from political impulses beyond state control and direction" (Josselin 2001a, p. 3). This definition covers groups conventionally termed "nongovernmental organizations" (NGOs) as well as for-profit actors, religious entities, think tanks, terrorists, and criminals (see also Halliday 2001, pp. 21 and 25). The definition is problematic, of course, in that it takes as given what should be the subject of empirical investigation: the autonomy of nonstate actors. The literature varies in the extent to which it presumes (Eising and Kohler-Koch 1999) or shows (Knoke et al. 1996, Laumann and Knoke 1987, Pappi and Henning 1999) policy networks to encompass a wide array of state and nonstate organizations. Among students of transnational advocacy coalitions, a few have attempted to map network interactions. Riles (2000) offers a discourse analysis of transnational interactions, while True and Mintrom (2001) measure attendance at UN Conferences on Women (focusing on actor attributes rather than network characteristics). Defining networks by common values presumes an outcome that may or may not result from transnational communication. Neglecting material or professional incentives for network participation in favor of idealistic motivations discounts the possibility of advancing democracy through material claims-making, for instance by trade unions. But see Ansell, Parsons and Darden (1997) and Ansell (2000) for a use of the concept that emphasizes strategic shifts in power among network actors. Keck and Sikkink argue that the methodological difficulties and costs of investigating network relations empirically cancel out any theoretical gains (Keck and Sikkink 1998, p. 31). But see Smith and Johnston (2002) for an argument about increasingly similar opportunity structures for actors from multiple national contexts. By contrast, Sassen (2003, p. 7) acknowledges the role of the state in producing cross-border networks, but argues that the result is indeed a denationalization in favor of "a hierarchy of nested scales" that includes both local and global sites of exchange. Earlier definitions of transnationalism focused on societal links across national boundaries, in contrast to transgovernmentalism (cross-border links among lower levels of government) and intergovernmentalism or interstate relations (cross-border links among state leaders) (Keohane and Nye 1971). Outside of the EU, an example would be negotiations between the International Monetary Fund and a government over macroeconomic stabilization. Outside of the EU, NATO would be an example. Outside of the EU, relations between a metropole and its colonies might illustrate this scenario, where the colonies are economically integrated in the economy of the mother country, but not of fellow colonies. Accession criteria and procedures are the same for all candidates of the 2004 and later waves: (1) stability of institutions guaranteeing democracy, rule of law, human rights, and respect for and protection of minorities; (2) existence of a functioning market economy; (3) capacity to cope with competitive pressures and market forces within the Union; and (4) ability to take on the obligations of membership, including adherence to the aims of political, economic and monetary union (Copenhagen European Council, Conclusions of the Presidency, June 1993). The last criterion calls for the adoption of roughly 100,000 pages of EU legislation. The conditionality of accession means that whereas enforcement mechanisms vis-à-vis member states are differentiated and specific, accession candidates face one categorical threat: indefinite delay or outright denial of entry. Hence, the pressures to conform to EU demands were exceptionally high. Hungary's population size is roughly one fourth of Poland's. The measure produces scores between 0 and 1, 1 representing perfect structural equivalence. Given flow 1 ? 2 ? 3, where 2 is the broker, the role of a liaison can be described as follows: A? B ? C (all three actors belong to different groups) (Borgatti et al. 2002). Note that no claim is made here that liaison scores should not differ systematically among all three groups. We would not expect nonstate actors to act as liaison between governments and the EU. The nature of the EU and the accession process implies that the EU and governments will interact directly. This is a nominalist approach, which sets network boundaries based on theoretical considerations. Whether the actors involved experience the network as such is subject to empirical investigation (Laumann et al. 1983). Only organizations that responded to the survey could be included in the network. The imbalance in the number of Hungarian versus Polish state actors is due to the lower response rate among the latter. The imbalance in the number of Hungarian and Polish nonstate actors is due to the more fragmented organizational landscape in Hungary. Whereas Poland has two main employer federations and two main labor federations at the national level, Hungary has nine employer and six labor federations. Also termed CONCOR. For details, see Wasserman and Faust 1994, pp. 376-381; Breiger et al. 1975. Weighted brokerage scores take into account the existence of alternative brokers that connect any two actors. When B is one of two brokers connecting A and C, weighting gives B only half of the credit for the link between A and C. See http://www.polsci.indiana.edu/Surveys/TakeSurvey.asp?SurveyID=9MJ5n3KM9541G for the online questionnaire. Twenty-three face-to-face interviews and nine electronic responses. Among employers' organizations and trade unions, this was usually the international affairs secretary or the person assigned to European integration issues. In the case of ministries, the informant tended to be a representative of the department handling EU accession. Among EU institutions and Europe-wide nongovernmental organizations, the informant was generally the person in charge of enlargement affairs. In the case of perfect reciprocity of perceptions of contacts, the Pearson correlation between matrix "we initiate" and the transpose of matrix "they initiate" would be 1. In fact, the Pearson correlation is .44, indicating that only in 44 percent of all observations do both actors agree on the tie. The matrix "we initiate regular contact" has a density of .333, which means that 33.3 percent of all possible ties are actually present. Software used for analyzing the network data: Borgatti et al. 2002. The density of matrix "they initiate regular contact" is slightly lower, with 30.9 percent of possible ties actually present. The Pearson correlation between matrix "we initiate" and matrix "they initiate" is .77, suggesting a high level of reciprocity among actors in initiating regular contact (statistically significant at the .0005 level, SD=.051). With only 1.6 percent of possible ties present, the matrix "we are members in" has by far the lowest density. This cautious estimate screens out contacts that are interpreted differently by source and target. Such disagreement is itself worth studying, but does not constitute the focus of this article. The Pearson correlation between "A reports any tie with B" and "confirmed ties" is .732 (statistically significant at the .0005 level, SD=.048). By contrast, the directional matrix "A reports any tie with B" has a density of .41. 43