Elsevier

Social Networks

Volume 44, January 2016, Pages 281-294
Social Networks

Multilevel embeddedness: The case of the global fisheries governance complex

https://doi.org/10.1016/j.socnet.2015.03.001Get rights and content

Highlights

  • States’ bilateral and multilateral fisheries agreements interlock as a multilevel network.

  • States’ bilateral clustering is embedded in their shared membership in multilateral fisheries agreements.

  • States’ membership in multilateral fisheries agreements are clustered around secretariats and by similar content.

  • Multilevel alignment is absent, suggesting states use bilateral and multilateral agreements for different purposes.

Abstract

This paper explores how bilateral and multilateral clustering are embedded in a multilevel system of interdependent networks. We argue that in complex systems in which bilateral and multilateral relations are themselves interrelated, such as global fisheries governance, embeddedness cannot be reduced to unipartite or bipartite clustering but implicates multilevel closure. We elaborate expectations for ties’ multilevel embeddedness based on network theory and substantive considerations and explore them using a multilevel ERGM. We find states’ bilateral ties are embedded in their shared membership in multilateral fisheries agreements, which is itself clustered around foci represented by similar content and treaty secretariats.

Introduction

How and why nodes tend to cluster has long been a subject of theoretical and empirical research in social networks. That nodes cluster their relations to form densely connected groups (Moreno and Jennings, 1938, Holland and Leinhardt, 1970, Davis, 1970, Opsahl and Panzarasa, 2009, Opsahl, 2011) has proven a robust finding across both unipartite networks, consisting of ties between a single set of nodes (Simmel, 1955, Holland and Leinhardt, 1970), and bipartite networks, in which ties are defined as occurring between two distinct nodesets (Borgatti and Everett, 1997, Robins and Alexander, 2004).

A number of theoretic motivations for the occurrence of clustering have been proposed ranging from structural balance (Cartwright and Harary, 1956), cohesion and embeddedness (Moody and White, 2003), and propinquity (Festinger et al., 1950). Clustering may also be seen as the outcome of the particular dependencies among tie-variables that we expect to see in social networks (Frank and Strauss, 1986). Different types of dependence assumptions may reflect the extent to which ties are more or less likely to cluster depending on how dyads are embedded in their local social environment (see Pattison and Snijders, 2013). Finding evidence for these dependencies in network data thus serves as a justification for employing methodological tools sensitive to such dependencies.

Clustering also matters substantively. Insofar as social relations provide conduits for transmitting information or influence, clustering can have important ramifications for the diversity or reinforcement of information or influence along such ties (see e.g. Granovetter, 1973). On this basis, clustering has become a particularly pertinent research topic in the study of network governance, where such diversity or reinforcement is theorised to have important implications on the effectiveness of governance arrangements (Provan and Milward, 1995, Provan and Kenis, 2007, Robins et al., 2011, Lubell et al., 2014). Yet, while clustering is an important topic in such literatures and in both unipartite and bipartite contexts, few have investigated how clustering interconnects across interlocking unipartite and bipartite networks. Here we consider how different types of ties are embedded in each other by explicating their positions in a multilevel system.

To explore multilevel embedding, we examine a case from within the field of International Relations and network governance. The global fisheries governance complex provides an excellent example of where unipartite and bipartite networks interlock in a multilevel network governance context. It consists of states and the bilateral and multilateral fisheries agreements they establish between or among themselves to coordinate and cooperate regarding fish stocks that straddle or migrate across international maritime borders.1 This constitutes a multilevel network consisting of two types of nodes – a set of states and a set of multilateral fisheries agreements (MFAs) – and different types of ties between and among them. Here, the ties between states define a micro-level of interaction in the governance network, the affiliations with multilateral fisheries agreements establishes a meso-level link between the states and the macro-level of inter-related fisheries agreements. This multilevel network is a natural superordinate to the socio-ecological network of the fisheries governance complex (Bodin and Tengö, 2012). In our network, the ecological embedding is reduced to the states’ relation to natural resources, such as fishing volumes and national tallies of threatened and endangered species, as well as the geo-spatial embedding of states.

These two types of agreement are structurally different in their negotiation and operation and typically fulfil different functions. The structural differences – bilateral agreements being limited to two parties, whereas multilateral agreements may engage many more – have profound implications for how these instruments are used. International law considers bilateral agreements to operate like contracts, offering parties the opportunity to negotiate terms dyadically. Multilateral agreements though, such as the United Nations Convention on the Law of the Sea (UNCLOS), are regularly employed as normative or “law-making” tools (see Shaw, 2003, 88–89). These structural differences then manifest themselves as functional differences too. Whereas bilateral fisheries agreements tend to concern either allocation of straddling stocks or access to under-exploited stocks, multilateral fisheries agreements tend to concern more collective, normative goals of optimal management and conservation of shared stocks. Because of these structural and functional differences, these different forms of agreement should be treated as constituting analytically distinct networks. On the one hand, states’ dyadic ties through bilateral fisheries agreements (BFAs) can be expressed as a unipartite network (Kinne, 2013). On the other hand, to capture the structural information in states’ membership affiliations in multilateral fisheries agreements, a bipartite representation is preferable.

But while they ought to be treated as analytically distinct, we argue that they should not be treated as independent because each case of a tie of one type existing may depend on a tie of the other type (see also Verdier, 2008, Zawahri and Mitchell, 2011). For example, the existence of a bilateral agreement between two states may affect their propensity to join the same multilateral agreements. Or, similarly, states may find that a shared multilateral agreement may encourage a complementary bilateral agreement or render it unnecessary. We therefore propose that actors’ bilateral and multilateral institutional relationships ought to be treated as two parts of a single, multilevel network comprised of interlocking and thus interdependent unipartite and bipartite networks. Indeed, similarities between multilateral fisheries agreements constitute another unipartite network that may drive bipartite clustering. Since each of these networks interlock, we argue that they should be modelled together as an interdependent multilevel network to make the most robust inference on clustering mechanisms. For example, the fact that multilateral treaties deal with collective issues may mean that states do not sign up to treaties that are similar unless the issues they deal with are of general interest. This gives rise to a particular form of multilevel clustering that suggest that a state cannot force the agenda of a multilateral treaties. This corroborates the evidence for the functional differences offered by associations with attributes of the states and the lack of multilevel alignment between bilateral ties and multilateral treaty content.

To analyse the embedding of ties within and across different types of networks, we draw on the exponential random graph (ERGM) family of models. Frank and Strauss (1986) developed ERGMs for one-mode networks from a Markov dependence assumption, which Snijders et al. (2006) elaborated by proposing parameters derived from the so-called social circuit dependence assumption. Skvoretz and Faust (1999) proposed an ERGM for bipartite networks that was developed further by Agneessens and Roose (2008) and fully extended by Wang et al., 2009, Wang et al., 2013 to include recently proposed dependence models. ERGMS for exploring the joint analysis of ties between different types of nodes were considered by Wasserman and Iacobucci (1991), and then the seminal paper by Lazega et al. (2008), explicating the interrelations between ties of nodes at different levels, motivated the adoption of the new developments for ERGMs to multilevel networks by Wang et al. (2013). We make use of these developments here.

The rest of this paper is structured as follows. The next section introduces the three elemental networks that together constitute the multilevel network this paper concerns. It also reviews and discusses some descriptives for each network, speculating on some exogenous and endogenous mechanisms that may generate such descriptive features. Having introduced the multilevel network, the following section then introduces and discusses several multilevel mechanisms that we expect to play a role in generating the structure of our multilevel network, with special emphasis on cross-level clustering. We then turn to the results produced by modelling the component networks separately and jointly as a multilevel network, and discuss the results and what this tells us about clustering.

Section snippets

A multilevel network

This section describes the three interlocking, elemental networks of the multilevel global fisheries governance complex studied here. These networks “interlock” around the two distinct nodesets. On a micro-level we have ties between states with a potential set of ties that we denote AA; on the macro-level we have ties between the MFAs with a potential edge set BB; and, finally a meso-level connecting the nodes in A with the nodes in B giving a set of potential bipartite edges AB. We use AA, AB,

Multilevel clustering

These three networks could be modelled separately without taking the interdependencies among the different types of ties into account. This would then assume that ties from different networks are completely independent of each other. However, Lazega et al. (2008) demonstrated the analytical advantages of explicitly studying how different levels interact and indicated the added insight that might be obtained from considering ties across different levels jointly as a multilevel network. Aside

Results

This section presents the results from using exponential random graph models to study clustering in and between the three component networks of the multilevel global fisheries governance complex. We fit three models: a bipartite model without the secretariat effect, a model where all networks are modelled independently, and a multilevel ERGM with cross-level effects. All models are estimated using standard procedures in MPNet (Lusher et al., 2013). The maximum likelihood estimates for the

Discussion

This paper has demonstrated how ties’ being embedded in a multilevel structure of interlocking networks affects the propensity of ties to cluster in certain configurations. Chief among our findings is that states’ bilateral ties cluster in a model that treats these ties as independent, but can be better explained by patterns of common multilateral affiliation in a multilevel network model. This is not simply a function of the additional complexity introduced by including multiple networks

References (64)

  • P. Wang et al.

    Exponential random graph (p*) models for affiliation networks

    Soc. Netw.

    (2009)
  • E. Witbooi

    The infusion of sustainability into bilateral fisheries agreements with developing countries: the European Union example

    Mar. Policy

    (2008)
  • F. Agneessens et al.

    Local structural properties and attribute characteristics in 2-mode networks: p models to map choices of theater events

    J. Math. Sociol.

    (2008)
  • J.S. Barkin et al.

    Unilateralism and multilateralism in international fisheries management

    Glob. Gov.

    (2000)
  • R. Berardo et al.

    Self organizing policy networks: risk, partner selection, and cooperation in estuaries

    Am. J. Polit. Sci.

    (2010)
  • Ö. Bodin et al.

    Disentangling intangible social-ecological systems

    Glob. Environ. Change

    (2012, March)
  • R.L. Breiger

    The duality of persons and groups

    Soc. Forces

    (1974)
  • C.T. Butts

    Predictability of large-scale spatially embedded networks

  • C.T. Butts et al.

    Spatial modeling of social networks

  • C.L. Carr et al.

    Multilateral treaties and the environment: a case study in the formation of customary international law

    Denver J. Int. Law Policy

    (1999)
  • D. Cartwright et al.

    Structural balance – a generalization of Heider's theory

    Psychol. Rev.

    (1956)
  • E. Chalamish

    The future of bilateral investment treaties: a de facto multilateral agreement?

    Brook. J. Int. Law

    (2009)
  • J.A. Davis

    Clustering and hierarchy in interpersonal relations: testing two graph theoretical models on 742 sociomatrices

    Am. Sociol. Rev.

    (1970)
  • T. Diez et al.

    The drama of the commons

    The Drama of the Commons

    (2002, March)
  • ECOLEX

    The Gateway to Environmental Law

    (2011)
  • FAO

    FishStatJ – Software for Fishery Statistical Time Series

    (2011)
  • L. Festinger et al.

    The spatial ecology of group formation

  • O. Frank et al.

    Markov Graphs

    J. Am. Stat. Assoc.

    (1986)
  • R. Froese et al.

    Fishbase Database

    (2008)
  • M.S. Granovetter

    The strength of weak ties

    Am. J. Sociol.

    (1973)
  • G. Hardin

    The tragedy of the commons

    Science

    (1968)
  • N. Harrigan et al.

    Differential impact of directors’ social and financial capital on corporate interlock formation

  • Cited by (0)

    View full text