Network Theory and Security Governance

22
JOBNAME: Sperling PAGE: 1 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014 3 Network theory and security governance Mette Eilstrup-Sangiovanni This chapter explores the potential for combining two theoretical frameworks or research agendas, namely security governance and social network theory. Since the introduction of complex interdependence theory and regime theory in the 1970s, international relations (IR) scholars have paid increasing attention to two facts about international relations: international policy outcomes are shaped by a variety of actors including, but not limited to, states; international cooperation has both formal and informal aspects. As a result, IR scholars have increasingly focused on the informal interactions that occur ‘behind the scenes’ in formal frameworks for rulemaking and implementation and that shape the ways in which formal rules are enacted and applied. Scholars have also studied informal relationships and practices as institutions in their own right – that is, as existing and operating independently of formal structures (see, for example, Brie and Stölting 2011: 20). Both forms of informal institutions have often been found to revolve around actors other than states. By and large, international security has not taken center stage in this scholarship. Owing to the high stakes involved, international security cooperation has traditionally been viewed, on theoretical grounds, as an area in which formal relationships predominate, and in which non-state actors play at most a marginal role (see Abbott and Snidal 2000: 440; Duffield 2006: 633–4; Jönsson 1986: 44–6; Kahler 2000: 555). In recent years, however, two strands of scholarship have emerged to challenge these assumptions. First, the literature on security governance highlights how diverse actors (state and non-state) cooperate at multiple levels and use a variety of instruments and practices (formal as well as informal) to produce security policy outcomes. Second, a growing literature on international security networks uses network theory to analyze the relational structures (formal as well as informal) through which diverse actors make and implement international security policy. Despite their overlapping focus, it is relatively rare for these literatures to speak directly to one another. Although scholars of security governance often adopt the descriptive language of network theory, they rarely go beyond using the term ‘network’ as a heuristic device to depict a generic form of horizontally structured and loosely connected organization. Likewise, scholars using network analysis to examine international security structures often fail to absorb insights from the literature on international security governance. In this chapter, I explore some of the ways in which network theory can advance the analysis of security governance structures, and vice-versa. 1 Unlike other chapters in this volume, my aim is not to evaluate or champion a particular conceptualization of security governance but, rather, to try to stimulate a more direct dialogue between theories of social networks and evolving concepts of security governance. Network analysis, I argue, taps directly into many of the theoretical problems associated with the study of global governance. For example, network theory offers a methodological 41 Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 1 / Date: 17/6

Transcript of Network Theory and Security Governance

JOBNAME: Sperling PAGE: 1 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

3 Network theory and security governanceMette Eilstrup-Sangiovanni

This chapter explores the potential for combining two theoretical frameworks orresearch agendas, namely security governance and social network theory. Since theintroduction of complex interdependence theory and regime theory in the 1970s,international relations (IR) scholars have paid increasing attention to two facts aboutinternational relations: international policy outcomes are shaped by a variety of actorsincluding, but not limited to, states; international cooperation has both formal andinformal aspects. As a result, IR scholars have increasingly focused on the informalinteractions that occur ‘behind the scenes’ in formal frameworks for rulemaking andimplementation and that shape the ways in which formal rules are enacted and applied.Scholars have also studied informal relationships and practices as institutions in theirown right – that is, as existing and operating independently of formal structures (see,for example, Brie and Stölting 2011: 20). Both forms of informal institutions haveoften been found to revolve around actors other than states.

By and large, international security has not taken center stage in this scholarship.Owing to the high stakes involved, international security cooperation has traditionallybeen viewed, on theoretical grounds, as an area in which formal relationshipspredominate, and in which non-state actors play at most a marginal role (see Abbottand Snidal 2000: 440; Duffield 2006: 633–4; Jönsson 1986: 44–6; Kahler 2000: 555).In recent years, however, two strands of scholarship have emerged to challenge theseassumptions. First, the literature on security governance highlights how diverse actors(state and non-state) cooperate at multiple levels and use a variety of instruments andpractices (formal as well as informal) to produce security policy outcomes. Second, agrowing literature on international security networks uses network theory to analyze therelational structures (formal as well as informal) through which diverse actors makeand implement international security policy. Despite their overlapping focus, it isrelatively rare for these literatures to speak directly to one another. Although scholars ofsecurity governance often adopt the descriptive language of network theory, they rarelygo beyond using the term ‘network’ as a heuristic device to depict a generic form ofhorizontally structured and loosely connected organization. Likewise, scholars usingnetwork analysis to examine international security structures often fail to absorbinsights from the literature on international security governance.

In this chapter, I explore some of the ways in which network theory can advance theanalysis of security governance structures, and vice-versa.1 Unlike other chapters in thisvolume, my aim is not to evaluate or champion a particular conceptualization ofsecurity governance but, rather, to try to stimulate a more direct dialogue betweentheories of social networks and evolving concepts of security governance. Networkanalysis, I argue, taps directly into many of the theoretical problems associated with thestudy of global governance. For example, network theory offers a methodological

41

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 1/ Date: 17/6

mer29
Cross-Out
mer29
Inserted Text
,
mer29
Inserted Text
formulated,
mer29
Cross-Out
mer29
Cross-Out
mer29
Cross-Out
mer29
Inserted Text
begun

JOBNAME: Sperling PAGE: 2 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

toolkit that can help us identify, measure and analyze the various relational structuresthat underpin governance frameworks. It also presents a set of theories that explicitlyrelate the structural properties of governance frameworks to international policyoutcomes (see Hafner-Burton et al. 2009: 559). As such, network theory offers afoundation for introducing more precision and rigor into analyses of security govern-ance frameworks. On the other hand, a closer engagement with the concept of securitygovernance can help to unify the fragmented field of political network studies byproviding a common terminology, and by drawing attention to the crucial but widelyoverlooked role of agency in political networks.

The discussion proceeds as follows. In the first section I briefly discuss the conceptof security governance. The second section outlines two forms of network theory –organizational network theory (ONT) and social network analysis (SNA). The thirdsection discusses the applicability of both ONT and SNA to studies of internationalsecurity governance. I highlight five main points of intersection where an activetheoretical and empirical dialogue between the two research agendas holds particularpromise, and illustrate these points with reference to recent scholarship that engages,implicitly or explicitly, in such dialogue. The final section concludes with someobservations about the promise and pitfalls of employing network theory to studysecurity governance.

SECURITY GOVERNANCE

Mark Webber et al.’s (2004: 4) widely cited definition of security governance refers to‘the co-ordinated management and regulation of issues by multiple and separateauthorities, the interventions of both public and private actors, … formal and informalarrangements, in turn structured by discourse and norms, and purposefully directedtoward particular policy outcomes’. This definition highlights two crucial aspects ofgovernance, namely:

+ governance involves multiple different actors, that are separate or autonomousfrom one another;

+ governance is an intentional process of coordination that is enacted deliberatelyby actors with the aim of achieving specific outcomes (see Kirchner 2006: 948).

On these points most scholars appear to agree. There is less agreement, however, onhow to define the structural properties of international governance frameworks. Manyanalysts regard governance as entailing exclusively informal and non-hierarchicalrelationships (see, for example, Daase and Engert 2008; Krahmann 2001; Webber et al.2004). Governance is here sharply contrasted with government, with the formerdenoting explicitly non-authoritative ways of coordinating and regulating policy issues,whereas the latter involves the exercise of formal authority through hierarchicalrelationships among higher and lower tiers of decision-making and implementation.Since they build on informal practices and non-authoritative forms of exchange,governance structures are generally depicted in this literature as more fluid and flexiblethan the formal structures associated with government (Krahmann 2001: 5).

42 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 2/ Date: 17/6

mer29
Inserted Text
research agenda and
mer29
Cross-Out
mer29
Cross-Out
mer29
Cross-Out
mer29
Inserted Text
--
mer29
Cross-Out
mer29
Inserted Text
--
mer29
Inserted Text
primarily

JOBNAME: Sperling PAGE: 3 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

Others take a broader view, describing governance as a form of policy coordinationthat may involve both vertical and horizontal structures. According to Emil Kirchnerand James Sperling, for example, governance ‘can be formal or informal and caninvolve flat, hierarchical or heterarchical patterns of interaction’. Security governance,they insist, neither precludes nor necessitates the privileging of the state or non-stateactors in the security domain (2007: 18; see also Schroeder 2006: 5). It thus leaveswide open the question of how security management is structured and who theprincipal actors are in any given issue area. As we shall see in the next section, thesedifferent conceptions of governance map loosely onto two conceptions of socialnetworks as either 1) a particular form of organization characterized by informality anda predominance of loose, horizontal ties among actors enjoying a high degree ofindividual autonomy, or 2) simply as sets of units or ‘nodes’ that interlink to constitutea discrete structure or system.

NETWORK THEORY

This section introduces network theory as a tool for analyzing governance frameworks.Economists, organizational theorists, sociologists, social-psychologists and anthro-pologists have long used the concept of ‘networks’ to analyze social and economicsystems in which actors are linked through enduring formal and informal relations. Inthe past two decades, IR scholars have begun to experiment with applying networktheory to analyze the interaction of state and non-state actors at both transnational andinternational level (Keck and Sikkink 1998), to study new forms of public administra-tion linking governments and nongovernmental actors (Marin and Mayntz 1991;Rosenau and Czempiel 1992), and to map inter-state institutional and organizationalstructures (Hafner-Burton and Montgomery 2006). In the realm of security studies,scholars have emphasized the role of networks in underpinning transnational terrorismand organized crime (Kenney 2007; Raab and Milward 2003; Sageman 2004). Morerecently, a wide literature has emerged which examines global peace building andpost-conflict reconstruction efforts as well as humanitarian intervention and non-proliferation advocacy in terms of networks (see Carpenter 2011; Lipson 2005;Ohanyan 2010; Paris 2009; Ricigliano 2003).

What is a Network? Structural and Organizational Networks

Despite the growing popularity of the term, there is no clear consensus on how todefine a network. A formal definition describes a network as ‘a specific set of relationsmaking up an interconnected chain or system for a defined set of entities that form astructure’ (Thompson 2003: 54). This is a rather broad definition, indicating nothingmore than a set of ‘nodes’ that are in some way linked. It might refer to a system ofindividuals, groups or organizations, but it could equally describe a system ofcomputers linked to a common server. Much like the definition of governancesuggested by Sperling and Kirchner, it can embrace both non-hierarchical, hierarchicaland heterarchical structures.

Network theory and security governance 43

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 3/ Date: 17/6

mer29
Cross-Out
mer29
Inserted Text
This view
mer29
Inserted Text
social

JOBNAME: Sperling PAGE: 4 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

This open-ended definition is generally associated with a structural approach to thestudy of networks. Structural network analysis (SNA) conceives of networks simply asarrangements of relational ties that define, enable and constrain actors (Hafner-Burtonet.al. 2009: 559). The analytical focus is on how communication, resources, ideas andinfluence flow through particular configurations of relational ties and affect socialoutcomes (Emirbayer 1997: 298; Granovetter 1973). Relational ties, in turn, areunderstood to be (partially) constitutive of actors (Ansell and Weber 1999, Emirbayer,1997: 287; Hafner-Burton et al. 2009: 561). As the name suggests, SNA generatesexplicitly structural (as opposed to actor centered) explanations for both individualbehavior and macro-level outcomes. For example, the resources a node possesses or theability of a node to influence others is seen to depend on its relative position in thenetwork rather than (solely) on its individual attributes (see e.g. Burt 2004). Structuralnetwork analysis has been used by social psychologists to explain, for example, whyindividual job seekers who have weak, rather than strong ties to many other persons areadvantaged on the job market (Granovetter 1973). It has also been used by businesseconomists to investigate how the positions individuals hold within corporate structuresimpact on their salaries and prospects for promotion (see Christopolous 2008). So far,however, relatively few IR scholars have adopted a structural approach to networkanalysis. A notable exception is Hafner-Burton and Montgomery’s (2006) use of SNAto analyze how the position of states in the social network formed by intergovernmentalorganizations (IGOs) impact on international conflict. Measuring the relative positionof states in the organized social setting defined by IGO memberships, they show thatacts of interstate aggression are less determined by individual state attributes or bydirect social relations between any two states, than by the relative positions they occupywithin the wider ‘IGO network’. Another notable exception is Nexon and Wright(2007) who examine the structural (or network) characteristics of imperial systemsusing SNA.

While conceiving of ‘networks as structures’ has been a long-standing popularapproach among sociologists and economists, in a political science context, mostscholars have followed Powell (1990: 301) in conceiving of networks as a distinct formof social organization – different in nature from both hierarchies and markets (seeHafner-Burton et al. 2009). From this perspective a network is not merely a set ofinterlinked nodes, but a particular form of social organization that connects actors inpursuit of common goals. An organizational network can be defined as ‘any collectionof actors (N >2) that pursue repeated, enduring exchange relations with one another andat the same time lack a legitimate organizational authority to arbitrate and resolvedisputes that may arise during the exchange’ (Podolny and Page 1998: 58–9). Incontrast to markets, exchange relations in networks are enduring rather than sporadic;in contrast to hierarchies, networks lack top-down command and authoritative disputesettlement, and decision-making therefore tends to be relatively dispersed and consen-sus based.

This second definition is associated with what we may label ‘organizational networktheory (ONT). In contrast to SNA, ONT conceives of networks not merely as systemsof linked individuals, but as self-conscious collective actors, united by shared goals orvalues. Whereas the analytic focus of SNA is on how network structures enable andconstrain individual agents, the focus of ONT is on how organizational networks make

44 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 4/ Date: 17/6

mer29
Inserted Text
social
mer29
Cross-Out
mer29
Inserted Text
are
mer29
Cross-Out
mer29
Inserted Text
sociologists
mer29
Cross-Out
mer29
Cross-Out
mer29
Inserted Text
find it easier to get a job
mer29
Cross-Out
mer29
Cross-Out
mer29
Inserted Text
discrete

JOBNAME: Sperling PAGE: 5 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

decisions, pool resources and coordinate actions in pursuit of collective goals, and withwhat effects (see Hafner-Burton et al. 2009: 561; Smith-Doerr and Powell 2005: 369).Among business economists and organizational theorists, ONT has been widely used tostudy the structures of firms, where a networked organization has generally been foundto permit flexible, on-demand production, which enables firms to outperform theircompetitors (see especially, Williamson 1975). Frequently, IR scholars have incorpor-ated these insights into theories about the functionality of international networkswithout much revision. For example, the literature on terrorist and drug traffickingassumes that illicit actors benefit from a loose, networked structure, which allows themto adapt quickly to changing circumstances and heightens their resilience in the face ofstate prosecution. Similarly, Chayes et al. (1997) advocate a networked organization ofpeace-keeping and post-war reconstruction efforts on the grounds that organizationalnetworks – with free information flows and authority grounded in knowledge ratherthan formal position – are more responsive to situational demands and therefore betterable to face complex, fast-moving and highly threatening situations in post-conflictzones than more conventional, hierarchical relief organizations.

The two approaches (SNA and ONT) are not mutually exclusive but can supplementone another. In a political science context, analyzing macro-level ‘network effects’often invites us to treat networks as organizational ‘tools’ for solving particular policyproblems. On the other hand, understanding how networks constrain and empowerindividual actors typically requires a more fine-grained examination of the structuralproperties of networks and thus calls for an SNA approach. In the rest of this section Idiscuss the main analytical concepts and methods of organizational and structuralnetwork analysis. The following section lists specific ways in which both variants ofnetwork theory can advance the analysis of security governance and reviews recentscholarship that applies network theory in the relevant ways.

Organizational Network Theory

Organizational typologies contrast networks with hierarchies. As Thompson notes, acore feature of a hierarchy is that it requires some form of overt rule-driven design andadministration (Thompson 2003, 22–4). Whereas hierarchies are based on top-downmanagement and a clear division of roles and responsibilities, networked organizationsare presumed to be flat and decentralized, with decision-making and action dispersedamong multiple actors that exercise a significant degree of local autonomy (Mayntz2004). The presence of some nodes that enjoy a high degree of centrality oftenintroduces aspects of hierarchy into networked organizations. What distinguishesnetworks from organizational hierarchies, however, is that there are no formal,institutionalized lines of command. Lower-level units can form direct relationships bothwith higher-level centers and with nodes at their same organizational level (see Ansell2000: 306; Gulati et al. 2002: 289). Importantly, hierarchical relations in networks tendto be based on the possession of specific resources or structural position rather than onformal institutional rules or mandates.

Several implications follow from this mode of organization. First, unlike hierarchies,which can rely on authoritative rules and legal arbitration to govern relations, networksare self-enforcing governance structures disciplined mainly by reputation and norms of

Network theory and security governance 45

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 5/ Date: 17/6

mer29
Cross-Out
mer29
Inserted Text
for example,
mer29
Cross-Out
mer29
Inserted Text
processes
mer29
Cross-Out
mer29
Cross-Out
mer29
Cross-Out
mer29
Inserted Text
mer29
Cross-Out
mer29
Inserted Text
Along similar lines
mer29
Cross-Out
Mette
Inserted Text
that allocates tasks among participants
Mette
Inserted Text
purely on

JOBNAME: Sperling PAGE: 6 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

reciprocity (Podolny and Page 1998: 60–65; Powell 1990: 301–4; Thompson 2003: 43).Second, the lack of central authority implies that decision-making and coordination innetworks tend to be based on mutual adjustment and consensus seeking, rather thantop-down administrative command. Third, the fact that networks rely on informal,non-hierarchical practices of coordination is generally seen to imply ‘equality ofmembership’ and joint responsibility for decisions and outcomes (Thompson 2003:39-40). In short, networks are fundamentally egalitarian structures. These descriptivecharacteristics are similar to the assumptions of a large swathe of global governanceliterature, which defines governance in terms of informal, decentralized and non-authoritative coordination and management of issues by multiple actors that arenotionally equal partners.

Organizational Network Advantages

Much scholarly effort has gone into illuminating the benefits that flow form anetworked organization. Thus, ONT offers a rich repository of hypotheses about thebenefits of informal, horizontal interactions – many of which carry direct relevance forthe study of security governance. Among the most common advantages claimed fornetworked organizations are:

1. A non-hierarchical organization allows networks to acquire, process and dissemi-nate information faster and more efficiently than hierarchies. In hierarchies,information must typically pass through a centralized processing or decision-making unit, thereby increasing the risk of blockage and delay (Watts 2003).Formal, vertical chains of command can make it difficult for actors to transmitinformation on the local characteristics of problems to central decision-makers(Scharpf 1993: 135). By contrast, the network’s flat, but highly connectedstructure implies that communication can flow unhindered between nodes (Watts2003: 280–81). Information flowing through networks is also thought to be ofhigher quality than information passed through hierarchies, since informationgathered ‘in the field’ from actors working on similar problems or in similarsettings is often more useful than information which is passed down a formalchain of command (Powell 1990: 304).

2. Networks encourage innovation. In hierarchies, top-down command and strongreliance on dedicated routines tend to lock people into particular ways of workingand discourage experimentation (see, for example, Erikson 1981: 188–210;Kenney 2007: 7; Podolny and Page 1998: 63). By contrast, an informal,non-hierarchical form of cooperation based on reciprocity encourages people toshare and collectively interpret information, thereby creating new interpretations(Podolny and Page 1998: 62, 304). Moreover, the free flow of information innetworks is often believed to engender a culture of openness to differentperspectives and new ideas.

3. Networks are flexible and adaptive. The boundaries of networks are generallyeasier to adjust than those of formal hierarchies. This implies that the compositionof networks can be more easily modified to respond to changes in theirenvironment (Podolny and Page 1998: 64–6). For example, due to their informal

46 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 6/ Date: 17/6

Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
a theoretical
Mette
Inserted Text
organizational
Mette
Inserted Text
thought of as
Mette
Cross-Out
Mette
Inserted Text
from
Mette
Inserted Text
also
Mette
Cross-Out
Mette
Inserted Text
physical locations
Mette
Inserted Text
practically
Mette
Inserted Text
and trust
Mette
Cross-Out
Mette
Inserted Text
More generally,
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
Due to their lack of formal, contractual basis, the

JOBNAME: Sperling PAGE: 7 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

organization, networks can ‘scale’ by adding links to new individuals or groups aslong as they are able to communicate within the network and share its values andgoals. Thus, when new problems arise, links can quickly be extended to nodeswith the relevant expertise (Castells 1996: 470–71; Smith-Doerr and Powell 2005:382–4). Similarly, the predominance of ‘weak ties’ between nodes prevents thelocking-in of strong but ineffective relationships. If a particular relationship is notproducing the desired outcomes, actors have the opportunity to search freely forother linkages in the network (Thompson 2003: 144).

4. Networks strengthen local initiative. Thanks to their decentralization, networksencourage diverse solutions to local problems, rather than dictating uniformpolicies. The tendency to design and implement policies ‘closer’ to the people thatare affected by them is in turn held to reduce transaction costs by providingpolicies that are more responsive to people’s needs (Chayes et al. 1997; Paris2009: 19).

Network Disadvantages

While scholars have spent much time highlighting the benefits that networks enjoy overtheir institutional rivals, less attention has been paid to ways in which organizationalnetworks may at times impede collective action. For example, IR scholars oftenhighlight transaction cost savings from a networked organization. When sociologistsand business economists argue that networks reduce transaction costs, however, theytend to emphasize the reliance on trust (as opposed to contractual provisions) as theprimary basis on which costs are reduced (see Podolny and Page 1998: 65). Butwhereas trust may effectively substitute for contractual provisions among individualsthat share a relatively confined social space, it may be more difficult to develop in aninternational context where the ‘social distance’ among nodes (many of whom areheterogeneous groups or organizations rather than individuals) is often wide. In aninternational policy context, networks often bring together actors from diverse socio-political and cultural environments that may pursue conflicting objectives. Thus,trust-based interactions may be difficult to build and sustain. In this context, the lack ofcontractual basis and absence of centralized leadership associated with a networkedorganization may slow down decision-making and increase transaction costs as allplayers try to have their say (Powell 1990: 318).

A loose, distributed structure may also reduce the capacity of organizationalnetworks to make decisions and engage in strategic planning. Often, local autonomymeans that there will be no clear convergence of purpose or allocation of responsibilityto reach common goals. Moreover, in the absence of centralized leadership, collectivedecisions may not be respected as readily due to the lack of an authoritative stamp. Asa result, resources may be allocated poorly, inconsistent tactics may be selected, andactivities carried out that serve narrow, individual interests rather than long-termcollective objectives (Powell 1990: 318; Thompson 2003: 46-7. For a discussion of thelimitations of organizational networks, see Eilstrup-Sangiovanni and Jones 2008).

As I discuss below, by considering both the benefits and drawbacks from anetworked organization, we can build better theoretical foundations for assertions aboutthe general advantages of a ‘governance response’ to complex security problems.

Network theory and security governance 47

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 7/ Date: 17/6

Mette
Cross-Out
Mette
Inserted Text
Thus, when new problems arise, links can quickly be established to actors with relevant expertise.
Mette
Inserted Text
relatively easily
Mette
Cross-Out
Mette
Inserted Text
Organizational
Mette
Inserted Text
(see Powell 1990).
Mette
Inserted Text
some scholars argue,
Mette
Cross-Out
Mette
Inserted Text
undertaken
Mette
Cross-Out
Mette
Inserted Text
of

JOBNAME: Sperling PAGE: 8 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

Structural Network Analysis

As we saw above, from a structural perspective a social network is simply a ‘specificset of relations making up an interconnected chain or system for a defined set ofentities’. Entities (or nodes) can refer to a wide range of actors (individuals, groups,organizations, states) and ties can be conduits for the transfer of both physical assets(money, goods, technology) and non-material resources (information, knowledge,reputation, ideas, norms, and so on). As such, SNA is less a theory than a model foranalyzing how patterns of relational ties affect individual behavior and shape socialoutcomes (see Emirbayer 1997). Simply put, SNA offers a set of methodological toolsthat enables the fine-grained measurement and analysis of network structures – that is,of patterns of ties among interdependent units. It also offers a set of basic theoreticalassumptions (primarily from fields of sociology and mathematics) that relate structuresto social outcomes (Hafner-Burton et al. 2009: 561).

From the point of view of SNA, networks are seen to vary along three maindimensions: number of nodes; structure of the network – that is, the pattern ofconnections between nodes, and structural properties of the network, particularly anetwork’s density, centrality and segmentation (Barabasi 2003: 111–22). Structurally,networks can be divided into three main types: the chain network where goods orinformation move along a line of separated contacts and where end-to-end communi-cation must travel through intermediate nodes; the hub-and-spoke (or ‘scale-free’network) where actors are tied to a central node(s) and must go through those in orderto communicate with each other; and the all-channel network where everybody isconnected to everybody else (Barabasi 2003: 111–22).

In terms of structural properties, the most frequently measured network propertiesare ‘density’, ‘centrality’ and ‘segmentation’. Density refers to the number and strengthof ties among individual nodes (Smith-Doerr and Powell 2005). In formal networktheory, density measures the number of ties observed as a proportion of the totalnumber of possible ties in a bounded group. Thus, dense networks are networks that arehighly connected. Centrality, on the other hand, refers to the tendency of a single pointto be more central than all the others in the network. Segmentation, finally, expressesthe tendency of the network to be fragmented into loosely connected sub-clusters. Thiscan be caused by the presence of ‘cliques’ (groups in which every individual node isdirectly tied to every other individual node) that feature dense in-group ties and sparseout-group connections. Other important network properties are ‘homophily’, whichmeasures the extent to which actors form ties with other nodes that are similar tothemselves, and ‘multiplexity’ which measures the number of content-forms containedin a single tie (like centrality, ‘homophily’ and ‘multiplexity’ can both expressproperties of individual ties as well as be measured as properties of the network as awhole).

What is the applicability of these concepts to the study of security governance?Basically, structural network properties are important for studying the dynamics oflarge groups. At the macro-level, densely interconnected networks are often assumed tobe more efficient at generating and distributing information and ideas than sparselyconnected ones. Dense networks may also be better at aggregating and distributingphysical resources. For example, in their study of imperial orders Nexon and Wright

48 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 8/ Date: 17/6

Mette
Inserted Text
(Thompson 2003:54)
Mette
Inserted Text
Emirbayer notes,
Mette
Cross-Out
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
As Hafner-Burton et al (2009: 561) remind us, it
Mette
Cross-Out
Mette
Inserted Text
in much the same way as IR theories relate institutional design features to institutional outcomes
Mette
Comment on Text
please use italics for 'number'
Mette
Comment on Text
please use italics for 'structure' here
Mette
Comment on Text
please use italics for 'structural properties'
Mette
Comment on Text
italics
Mette
Cross-Out
Mette
Inserted Text
include
Mette
Inserted Text
rather technical
Mette
Inserted Text
social

JOBNAME: Sperling PAGE: 9 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

(2007) show that network density influences the ease with which material resources canbe mobilized. On the other hand, Burt (1992: 17) suggests that dense networks areassociated with higher information costs (per unit of information) because stronglyconnected agents are each likely to discover the same information at the same time.

The level of centralization in a network can also affect its overall efficiency androbustness. Many scholars have argued that high degrees of centralization (such as in ahub-and-spoke network) make networks less robust, since the failure of single nodescan disconnect the entire network. By contrast, decentralized but highly connectednetworks tend to be more robust as many nodes can perish without disconnecting theoverall network. Others have suggested that highly centralized networks are less likelyto absorb ideas from peripheral nodes and therefore tend to be less innovative, whereasa low degree of centralization encourages the inclusion of knowledge and ideas of avariety of actors, including those from the network margins (see Krahmann 2005: 26;Paris 2009: 62–3; Westerwinter 2013).

Segmentation and homophily may likewise influence the functionality of networks.In the case of illicit networks linking terrorists or drug-smugglers, a high degree ofsegmentation can help to build trust among nodes in individual ‘cells’, while at thesame time shielding cells from destruction via infiltration of the overall network. Thus,segmentation heightens resilience vis-à-vis law enforcement efforts (see Kenney 2007).At the same time, however, segmentation is likely to reduce the efficiency ofinformation sharing and may suppress innovation, as information sharing and learningare limited to smaller sub-groups of the network (see Eilstrup-Sangiovanni and Jones2008). Homophily may also influence the kinds of information that pass betweennodes. To illustrate how the degree of homophily in a network can influence networkfunctionality, consider this straightforward example. Studies of the effects of socialnetworks on bank decision-making show that bankers tend to consult closely linkedcolleagues in order to feel more certain about their financial information. However,closer ties do not result in sufficiently candid or fresh assessments of information, butrather reinforce existing opinions (Smith-Doerr and Powell 2005: 7). In this casehomophily reduces effectiveness. It is easy to imagine a similar dynamic at play amongstate actors sharing information on, say, proliferation of weapons of mass destruction ordual-use technologies. If we conceptualize the global system of international non-proliferation and export control regimes as a network, we will see that this network ischaracterized by the existence of a number of small cliques (groups of states that arehighly interconnected) which display a high degree of homophily. These groups willtend to share information and intelligence more readily and are more likely to trust theinformation they receive from other group members. However, by relying primarily oninformation from close, trusted partners they are less likely to learn about some of themost threatening developments on the periphery of the network.

These are examples of how structural network properties influence the functionalityof networks as a whole. However, structural properties are also determinants ofindividual node behavior. Indeed, much social network analysis is devoted to ascertain-ing how the distributions of connections in a network creates opportunities andconstraints on individual behavior and endows agents with different types of resources.A key concept in this regard is node centrality. Simply put, central nodes are thought tobe the most powerful or influential (Freeman 1979: 216). Three main metrics express

Network theory and security governance 49

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 9/ Date: 17/6

Mette
Inserted Text

JOBNAME: Sperling PAGE: 10 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

the centrality of individual nodes: ‘degree centrality’, ‘betweenness centrality’ and‘closeness’. ‘Degree centrality’ indicates the value of the ties that a node has to allother nodes in the network. A node that has direct links to many other nodes is seen asprivileged because it can pass and receive information to/from many other nodes andparticipate in multiple communication processes (Freeman 1979: 219-20). As such, ahigh-degree centrality is thought to endow nodes with ‘social power’, which enablesthem act as agenda-setters or sources of new norms (see Carpenter 2011; Hafner-Burton et al. 2009: 570).

‘Betweenness centrality’ is a measure of how often a node occupies a positionbetween pairs of other nodes. When a node is strategically located on the directcommunication paths linking other nodes, it can gain influence by controlling infor-mation or resources passing between those other nodes (Freeman 1979: 221). Inparticular, the ability to connect other nodes that are otherwise less connected is seen toendow a node with ‘brokerage’ or ‘bargaining’ power (Hafner-Burton et al. 2009: 572).A classic example would be the role of a broker or mediator in a situation of politicalconflict, where the power of the broker increases directly as a function of the barriers tocommunication between two hostile factions (ibid). Other examples would be aninterpreter allowing two people speaking different languages to communicate, or –more commonly in an international governance context – the ability of an actor who isa member of two organizations that otherwise have no overlap in membership to act asan intermediary. It is easy to imagine how such a position could endow an actor withsocial power due to the ability to manipulate information passing between the twoorganizations

‘Closeness’ measures the sum of the paths connecting a node to all other nodes in thenetwork. The ‘closeness’ of a node is defined as its total distance to all other nodes.Closeness is often treated as a measure of how long it takes for a node to distributeinformation to all other nodes (Freeman 1979: 225; Hafner-Burton 2009: 564). Thus itcan be seen as an important resource allowing nodes to control the agenda or define theterms of debate. As I discuss below, these and other network metrics can be very usefulfor understanding patterns of influence and subordination in security governancearrangements.

A NETWORK THEORETICAL APPROACH TO SECURITYGOVERNANCE

In this section I highlight some principal ways in which network theory can advancethe study of security governance, and illustrate by drawing on examples from recentscholarship. As Hafner-Burton et al. (2009: 561-2) emphasize, one of the key benefitsof network analysis to studies in international relations is that it ‘allows structuralinvestigation at multiple levels of analysis, including groups of units of any size as wellas the monadic, dyadic, and systemic levels familiar to international relations scholars’(emphasis added). Moreover, network analysis allows structural investigation of multi-ple forms of interaction (information exchange, exchange of goods or resources,negotiation, deliberation, diffusion of norms and ideas) and various aspects of power

50 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 10/ Date: 17/6

Mette
Cross-Out
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
level of
Mette
Inserted Text
may
Mette
Cross-Out
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
poorly
Mette
Cross-Out
Mette
Inserted Text
opcit.
Mette
Inserted Text
facilitate the establishment of new links between actors or to
Mette
Cross-Out
Mette
Cross-Out
Mette
Cross-Out
Mette
Cross-Out
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
sometimes
Mette
Inserted Text
STUDYING

JOBNAME: Sperling PAGE: 11 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

(material, social, ideational and discourse-based). This makes network analysis particu-larly well suited for analyzing security governance arrangements where manifold actorswork together across different formal settings or issue areas, and where discourses andnorms are thought to structure interactions alongside formal authority and materialpower.

Better Description and Measurement

One way in which network analysis can advance security governance studies is byproviding a set of methodological tools that enable more rigorous description andmeasurement of security governance structures. At the most basic level, network toolsallow scholars to clearly distinguish different forms of governance (non-hierarchical/hierarchical/heterarchical) and to identify and measure important structural propertiessuch as ‘centralization’ or ‘density’, which may affect the functionality of governanceframeworks. This descriptive function of network analysis is particularly importantgiven the manifold definitions of security governance currently in use. As discussedabove, conceptualizations of governance range from a particular form of informal,non-hierarchical organization to a non-specific system of actors that collectively ‘steer’policy outcomes. As Sperling (2008) notes, on the second conceptualization, securitygovernance systems can range from primitive balance of power systems to highlyinstitutionalized security communities. By offering precise definitions and measure-ments of structural properties, the use of network metrics can introduce greaterprecision and clarity into depictions of security governance thereby allowing scholarsto more accurately compare assumptions and share findings.

In particular, network analysis can be a valuable tool for identifying and mappinginformal governance structures. As Brie and Stölting (2011: 19) remind us, informalinstitutions and practices are methodologically more difficult to grasp than formalarrangements. In contrast to their formal counterparts, informal institutions are notofficially codified in constitutions, documents or laws, which makes them difficult tostudy. By offering tools for measuring social interactions and flows of resources suchas information, material goods or ideas, and by providing data collection techniquessuch as ‘snowball sampling’ and respondent-driven sampling, social network analysiscan help scholars to identify the boundaries and content of informal institutions thatmight otherwise remain hidden or obscure (ibid). In particular, sampling techniquesborrowed from SNA can be useful in situations when the boundary of a relevantpopulation or system is not known in advance, which is often the case with informalinstitutions (on data collection for network analysis, see Ward et al. 2011: 255).

Structural network analysis can be used to produce rich, detailed descriptions ofindividual governance arrangements. Security governance often takes the form ofhighly complex systems. Take the example of humanitarian interventions or emergencyrelief missions. As Murdie (2014) notes, emergency relief missions often bring togethermilitary personal and humanitarian NGOs who work side-by-side with internationalorganizations and local partners in new governance arrangements. Similarly, inter-national peace-keeping and peace-building activities typically involve a multiplicity ofactors – from the UN, regional organizations and national governments to internationalnon-governmental organizations (INGOs) and private for-profit contractors (see Lipson

Network theory and security governance 51

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 11/ Date: 17/6

Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
analytic
Mette
Inserted Text
detailed
Mette
Cross-Out
Mette
Inserted Text
theoretical
Mette
Cross-Out
Mette
Inserted Text
empirical
Mette
Inserted Text
,
Mette
Inserted Text
socalled
Mette
Cross-Out
Mette
Inserted Text
often

JOBNAME: Sperling PAGE: 12 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

2005; Paris 2009). When applied to such complex systems, SNA can help analysts todiscover how agents interact and who, among them, leads policy-making andimplementation and controls the flow of information. It can also unveil the simultane-ous existence of vertical and horizontal relations among actors.. Importantly, networkmetrics enable analysts to isolate formal and informal institutional structures and todiscover whether informal patterns of interaction converge or diverge from formalstructures in terms of how they distribute core resources such as knowledge, materialpower, and social influence (Faul 2013).

In addition to providing for more accurate and rich description, network metrics canalso be used to check theoretical assumptions about the structure of interaction withingovernance frameworks. So far, in the literature on global governance, the structuralproperties of governance frameworks have often been theoretically assumed rather thansubjected to rigorous empirically analysis (Hafner-Burton et al. 2009: 682-3). Thus, theliterature tends to describe governance frameworks as characterized by horizontal,informal and fundamentally egalitarian relationships of communication and exchangewithout necessarily verifying these assumptions. Rather than simply assuming a lack ofhierarchy in security governance frameworks, SNA enables the empirical investigationof the distribution of power and influence among diverse policy actors. For example, inher analysis of disarmament advocacy, Carpenter (2011) uses SNA to highlight thegenerally overlooked hierarchical dimension of transnational advocacy networks(TANs). Whereas the literature on TANs has tended to view networks as ‘characterizedby ‘horizontal and reciprocal’ relations of information exchange and common princi-pled understandings’, she finds that TANs in the area of disarmament advocacy tend totake the form of scale-free networks, dominated by a small number of highlyconnected, central hubs. These hubs act as gatekeepers and exercise a predominantinfluence over what issues are selected for campaign purposes, and how these issuesare in turn framed (Carpenter 2011: 72–3).

Although network analysis has yet to be fully embraced by IR-scholars, there arenumerous other examples of studies that employ SNA to describe and analyze securitygovernance structures. Mérand et al. (2011) use SNA to analyse European securitygovernance as embodied in the Common European Security and Defence Policy. Usingmeasures of ‘degree centrality’ and ‘brokerage’ they identify ‘a complex constellationof CSDP actors that feature cross-border and cross-level ties between different nationaland EU policy actors’ (Mérand et al. 2011: 121). Around EU institutions, the networkis quite dense. Ultimately, however, they find that the CSDP network is dominated bya small group of traditional state actors. For example, they find that Brussels-basednational ambassadors (who act as gatekeepers for national governments) are morecentrally positioned in the network than supranational and non-state actors. Thus,network analysis reveals that while CSDP governance is more heterarchical than atraditional intergovernmentalist approach would suggest, it is less flat and egalitarianthan some governance perspectives might suggest.

Another excellent example of the use of SNA for both descriptive and analyticalpurposes is a study by Nexon and Wright (2007) who use variation in the strength anddensity of ties between states to clearly distinguish imperial orders from unipolar andhegemonic systems. Empires, they argue, are characterized by rule through inter-mediaries whereby a central power exercises indirect and informal rule over peripheral

52 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 12/ Date: 17/6

Mette
Cross-Out
Mette
Inserted Text
see
Mette
Cross-Out
Mette
Inserted Text
excellent
Mette
Inserted Text
however
Mette
Cross-Out

JOBNAME: Sperling PAGE: 13 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

polities through representatives, as well as the existence of divergent contractualrelations between a core and its different peripheries. Thus the network structure ofimperial systems is a hub-and-spoke structure in which peripheries remain segmentedfrom one another as their relations are mediated by the core. The absence of collectiveties among peripheral polities limits their ability to mobilize resources for anti-imperialcollective action.

In all these cases, structural network analysis allows scholars to accurately depictnetwork structures and to check theoretical assumptions about the structure ofgovernance by revealing power-inequalities, dependencies and structural relationsamong actors that would not have been easily discoverable by employing otherapproaches (see Maoz 2012).

Evaluation of Structural Effects on Policy Outcomes

A second way in which network analysis can advance studies of security governance isby enabling scholars to test hypotheses drawn from security governance theories. AsHafner-Burton et al. (2009) observe, network theory allows scholars to test hypothesesdrawn from international relations theories at two levels: network effects (macro-level)and structural effects on individual behavior (micro-level).

At the macro-level, organizational network analysis presents a particularly fertileresource for deriving and testing hypotheses about the functionality and boundaryconditions of different forms of governance systems. As we have seen, in the literatureon global governance, an informal, networked structure is generally thought to generatedistinct advantages such as flexibility, responsiveness and a high capacity for innov-ation and learning. On the other hand, both theory and practical experience suggest thateffective strategic planning and commonality of purpose are often lacking in organ-izations with a loose, networked structure. By considering the ‘network effects’ ofdifferent governance structures, we can derive theoretically grounded explanations forwhy informal governance systems may be better adapted to deal with certain complexsecurity issues than formal, top-down government, as well as for why informalgovernance arrangements may prove suboptimal in some other contexts.

Examples of scholarship that use an organizational network approach to formulateand test hypotheses about the functionality of different network structures abound inthe literature on organized criminal and terrorist networks. Within this literature,scholars have largely portrayed clandestine networks as fluid, mobile, highly adaptableand resilient, making them difficult for hierarchically organized law enforcementagencies to combat. A fluid structure has also been found to endow clandestinenetworks with a high capacity for innovation and learning as well as for wide-scale,rapid recruitment (see, inter alia, Cronin, 2002/03; Kenney 2007; Raab and Milward2003; Sageman 2004; Williams 2001). In contrast to this view, Eilstrup-Sangiovanniand Jones (2008) use organizational network theory to highlight vulnerabilities intransnational criminal organizations and to identify potential sources of networkdebilitation. They find, that while a networked structure enhances flexibility, it reducesthe ability of criminal networks to plan and execute complex operations. It also makescriminal networks vulnerable to infighting and infiltration. In line with these findings,Enders and Jindapon (2010: 277) use network analysis to examine the optimal terror

Network theory and security governance 53

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 13/ Date: 17/6

Mette
Inserted Text
(ibid.)
Mette
Cross-Out
Mette
Inserted Text
might
Mette
Inserted Text
or methodologies
Mette
Inserted Text
(ONT)

JOBNAME: Sperling PAGE: 14 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

network structure. Comparing two types of terror networks – centralized and looselyconnected – they conclude that centralized networks are better able to coordinatenetwork activities and better able to respond to counter-terrorism.

Another example of the use of network theory to identify optimal organizationalstructures for addressing particular policy problems is a study of non-proliferationinstitutions by Eilstrup-Sangiovanni (2009). Using an ONT approach, she seeks toexplain why governments have used informal network-based organizations to addresssome proliferation threats, while creating highly centralized, formal intergovernmentalorganizations to deal with others. She finds that organizational networks are adapted tosolving cooperation problems of a different nature than those typically addressed byformal organizational hierarchies such as most treaty-based intergovernmental organ-izations. Whereas formal hierarchical structures reduce the transaction costs of long-term international cooperation among large groups of actors, enhance the credibility ofinterstate agreements and solve problems of incomplete contracting, informal networksreduce short-term contracting costs, and facilitate flexible low-cost cooperation amongsmaller groups. In yet another example, Paris (2009: 75) finds that effective coordin-ation among peace-builders hinges on the existence of a ‘directed network’ whichcombines a decentralized structure with elements of central direction.

As these studies suggest, network analysis can be used to expand and refine existingtheories of institutional design and organizational efficiency. International relationstheorists have long relied on transaction cost theories to generate hypotheses regardingthe conditions under which either hierarchical or dispersed horizontal coordinatingmechanisms are likely to be adopted and to be efficient (see especially Lake 1999,2001; Williamson 1975). By drawing on network theory, many of the assumptions thatunderlie such hypotheses can be tested and refined.

Network theory can also be used to derive and test novel hypotheses regarding theeffects of different structural figurations on individual behavior. For example, Lake andWong (2009) and Carpenter (2011) demonstrate how measures of centrality can help topredict who sets the agenda and controls information flows within specific socialnetworks. At a more general level, structural equivalence (the extent to which nodeshave a common set of linkages to other nodes in a network) may be used to predictbehavior as nodes in similar structural positions vis-à-vis other nodes may be expectedto act in similar ways (on structural equivalence, see Burt 1976; Hafner-Burton andMontgomery 2006: 7).

Network Formation and Evolution

A third way in which network theory can be employed to study security governance isto analyze how governance systems arise and how they evolve over time. In extantliterature, the emergence of governance systems is often explained by reference toexogenous factors such as complexity or environmental uncertainty. For example,Sperling (2008: 10) explains the proliferation of security governance systems by thefact that ‘states by themselves are ill-equipped to respond to the new challenges due totheir complexity. Yet, while an approach stressing generic exogenous factors such astechnical complexity or strategic uncertainty can provide an explanation for the general

54 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 14/ Date: 17/6

Mette
Cross-Out
Mette
Inserted Text
s
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
help to
Mette
Inserted Text
Mette
Inserted Text
organizational
Mette
Inserted Text
of organizational network concepts applied to the study of security governance
Mette
Inserted Text
economic theories of
Mette
Inserted Text
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
drawn
Mette
Inserted Text
(as opposed to general organizational functionality)
Mette
Cross-Out

JOBNAME: Sperling PAGE: 15 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

propensity of governance frameworks to form, it tells us little about their size, shape orwho will form ties with whom (Gulati and Gargiulios 1999: 1440).

To improve our understanding of the evolution of governance frameworks we maylook to studies of the behavior of individual nodes in social networks in order to derivehypotheses about the formation of complex governance systems. As Hafner-Burton etal. explain (2009: 566-7), SNA describes both relational and individual mechanismsthrough which new network ties are likely to be created. Relational mechanismssuggest how the relative location of agents within already existing networks influencesthe likelihood of new tie formation, whereas individual mechanisms indicate qualitiesof individual nodes that make ties among them more or less likely. An individualmechanism that may govern the formation of network ties is homophily – that is, thetendency for nodes to associate with other nodes that share similar attributes (Hafner-Burton et al. 2009: 567). Sociological studies have shown many social networks to becharacterized by homophily (see Ward et al. 2011). Yet, whereas individual character-istics such as age, race, gender, class or occupation may be strongly linked tohomophily in some social contexts, it is not obvious that such traits influence thelikelihood of tie formation in security governance networks or that we should expecthomophily to be a predominant characteristic of such networks at all. Indeed, in apolitical context, we may also expect to see ‘heterophily’ insofar as nodes are likely tolink to other nodes that are dissimilar but complementary in terms of the resources,expertise or world-views that they possess.

Relational mechanisms through which social networks may take form include‘proximity’, which refers to the tendency for units to form links to others that arestructurally close to them (in the sense of having similar sets of connections) or that arephysically or temporally close, as well as the process of ‘preferential attachment’, inwhich nodes form ties to those nodes that already have a large number of pre-existinglinkages (see Hafner-Burton et al. 2009: 568). Proximity computations can be used topredict the likelihood that links between nodes will be established in the future. As Idiscuss below, scholars have found that processes of preferential attachment tend togenerate a skewed degree distribution that follows a power law whereby the ‘rich getricher’ (see Lake and Wong 2009; Goddard 2009; Kahler 2009).

Another network concept of potential relevance to understanding the formation,shape and evolution of governance frameworks is a theory developed in socialpsychology called ‘structural balance’, which describes how relationships in a socialnetwork evolve over time. Focusing on ‘relationship triangles’ involving three indi-viduals, firms, or other entities, the theory suggests that certain combinations ofpositive (friendly) and negative (antagonistic) relationships among nodes are morelikely to be stable than others. Stable combinations include three mutual friendships, ortwo friends who share a common enemy. By contrast two enemies sharing a commonfriend is unlikely to be a stable configuration. If three nodes view each other as rivals,there are incentives for the two nodes in the weaker rivalry to ally against the third (asstandard balance of power theory would predict). This implies that relationshiptriangles containing two friendships are prone to transition to either one or threefriendships (see Marvel et al. 2011).

Structural balance theory has obvious implications for understanding dynamics ofcooperation and competition in social and political networks such as interstate

Network theory and security governance 55

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 15/ Date: 17/6

Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
and
Mette
Inserted Text
of indidivual frameworks, or about
Mette
Cross-Out
Mette
Inserted Text
ibid:
Mette
Cross-Out
Mette
Inserted Text
security

JOBNAME: Sperling PAGE: 16 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

alliances. It can also explain why it is common for some social networks – such as anetwork of competing factions in a civil conflict – to divide into two opposing camps.In some conditions, the theory can be used to predict how small shifts in individualrelationships produce particular outcomes in larger networks. For example, two centralnodes developing an antagonistic relationship can have profound structural implicationsfor the larger network. Such dynamics may be helpful for understanding how and whynetworks of transnational security actors or criminal networks coalesce or splinter overtime.

As an example of how network theory can illuminate the formation of governancesystems, Warren (2010) models interstate alliances as evolving networks. He arguesthat states prefer to form alliance ties with partners who possess a similar pattern ofobligations. Thus, the friend of a friend is likely to become a friend (positive linkage).Yet, in contrast to previous literature on alliance formation, he finds that the enemy ofan enemy is not necessarily a friend, meaning that states will not necessarily formalliances with the enemies of their enemies (Warren 2010: 707). Similarly, havingdeveloped a sophisticated dynamic model of structural balance, a group of researchersat Cornell University used a computer program to successfully ‘predict’ whichcountries joined the Allied Powers during the Second World War, and which lined upwith the Axis (Marvel et al. 2011).

In addition to analyzing the evolution of linkages that develop and shape networks,SNA can also be used to study how certain discourses gain predominance in particularsocial and political communities, or how certain norms and ideas diffuse throughout asocial network. Drawing on Watts (2003), Lake and Wong (2009) suggest a model ofsocial diffusion based on percolation theory. Using Watts’s theory, Lake and Wongshow how ideas or norms that arise in vulnerable clusters of ‘early adopters’ can gain afoothold and then diffuse throughout the network. Percolating clusters are groups ofnodes in which members are tightly connected, so that an initial norm is reinforced by‘internal’ peer acceptance, but loosely connected to the network as a whole so thatnorm adoption is not thwarted by external peer disapproval. Once a norm has takenhold, it can eventually spread beyond the initial cluster through its links to the rest ofthe network (or the initial cluster can ‘grow’ by accepting more members). They applythis model to show how certain norms have become predominant within the globalnetwork of human rights organizations.

Expand and Refine Key Concepts

A fourth way in which network analysis can enrich governance studies is by refiningcentral concepts such as ‘informality’ and ‘power’. In contrast to a substantialistconception which views power as an attribute of individual actors, network theorydefines power in relational terms – that is, as deriving from the structural position ofactors within a network rather than from actors’ individual qualities (Emirbayer 1997:292; Burt 2004, 3-4). ‘Network power’ takes three main forms: ‘social power’;‘bargaining power’, and ‘exit power’ (power arising from a nodes’ ability to de-link anetwork through exit) (see Hafner-Burton et al. 2009: 572; Lake and Wong 2009). Aswe saw above, social power is defined in terms of a node’s ability to reach a largenumber of other nodes. Social power can endow an actor with prestige and enable it to

56 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 16/ Date: 17/6

Mette
Inserted Text
Mette
Inserted Text
generally and structural balance theory in particular
Mette
Inserted Text
(ibid.)
Mette
Cross-Out
Mette
Inserted Text
.

JOBNAME: Sperling PAGE: 17 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

play an agenda-setting role. Central nodes can threaten to sever network links to aparticular node to alter its actions, thereby acting as a gatekeeper (see Carpenter 2011).Social power can also be used to structure actors’ incentives by manipulating the termsof debate (agenda-setting power). Most important, central nodes can set conditions onparticipation in the network as a whole (see Lake and Wong 2009).

Bargaining or brokerage power denotes the ability of nodes to connect differentsegments of the network. ‘Brokerage power’ may attached to central nodes or ‘hubs’that can offer or withhold ties to peripheral players, but can also be possessed byperipheral nodes that are strategically placed in the network. As Burt (1992) famouslydemonstrated, actors that bridge so-called ‘structural holes’ are often advantaged bytheir ability to combine information and resources from diverse groups, and by theircapacity to add value to the network by connecting otherwise isolated networksegments (Burt 1992: 3-4, 18). As conceptualized by Burt, ‘structural holes’ separatenodes or clusters of nodes that possess non-redundant information or resources. Iflinked, these nodes would provide networks benefits that are additive rather thanoverlapping (Burt 1992: 18). Cohesive or structurally equivalent nodes (nodes thatstrongly connected to each other or have similar ties to third parties) are likely to havesimilar information and so provide redundant information (ibid). However, nodes thatare structurally non-equivalent are likely to possess divergent information or resources.An actor that bridges a structural hole is thus able to access and synthesize information,ideas, perspectives and resources from alternative sources and perspectives. This offersinformation benefits and implies that such an actor is a better position to develop newideas and detect novel opportunities (Burt 1992; Burt 2004: 349-350).

By considering forms of relational power alongside other forms of power – such asformal institutional power or material capabilities – we gain a more complete picture ofpower relations. In particular, by highlighting relational aspects of power, SNA capturesforms of social and normative power that are often foregrounded in governancetheories. An important question for governance scholars, however, is how relationalpower (or ‘network power’) relates to other sources of power (such as military oreconomic capabilities). There are several different theories about this. One is thatnetwork power can be used to supplement or substitute for other forms of power. Thisview suggests that poorer and militarily less powerful states might offset their materialdisadvantages through the accumulation of social power or brokerage power (seeHafner-Burton et al. 2009: 573–4). This notion chimes well with studies of theinternational diplomacy of ‘small states’ (such as the Nordics), that are often found topunch above their weight in international diplomacy by virtue of playing the role of‘honest brokers’ or acting as normative entrepreneurs.

Another theory is that network power simply mirrors exogenous power relations. Tieformation, and thus the relative position of nodes in the network, is affected by externalpower differentials. New nodes entering a network will tend to force ties to high-statusnodes (preferential ties). Thus, network members who are considered prominent outsideof the network will also enjoy a central position within the network.

Finally, it is possible that network power can amplify existing power inequalities inthe global system. As we have seen, mechanisms of homophily function by networkmembers forming ties with others who are like them or structurally close to them.Thus, nodes will tend to form clusters or sub-groups with other nodes of similar power

Network theory and security governance 57

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 17/ Date: 17/6

Mette
Inserted Text
in order
Mette
Cross-Out
Mette
Inserted Text
are often able
Mette
Cross-Out
Mette
Inserted Text
a
Mette
Cross-Out
Mette
Inserted Text
to extend or threaten to
Mette
Cross-Out
Mette
Inserted Text
it
Mette
Cross-Out
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
There are few informational benefits to be gained from linking these.
Mette
Inserted Text
in a way that adds value to the network.
Mette
Inserted Text
that can be exploited strategically
Mette
Cross-Out
Mette
Inserted Text
and/or
Mette
Inserted Text
precisely the
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
forge
Mette
Inserted Text
so-called
Mette
Inserted Text
tend to

JOBNAME: Sperling PAGE: 18 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

capability, thereby increasing mutual centrality. Preferential attachment, in turn, impliesthat nodes with a large number of initial links will tend to receive a greater thanaverage number of in-links from other nodes. Thereby, attachments follow a power law,whereby nodes gain new connections in proportion to how many connections theyalready have. If we assume a tendency for powerful actors (by which I mean actors thatpossess superior material or other types of exogenous power resources) to attract agreater number of initial connections, this would trigger an exponential growth in-linksto those actors, which would in turn imply that the network would amplify theirexisting power advantage. Exogenous power resources may be important determinantsof tie formation in the context security governance systems, where preponderanteconomic or military power might make an actor an attractive target for in-links.Moreover, some authors have argued that the informal and non-rule bound interactionswithin organizational networks tend to favor stronger actors (see Eilstrup-Sangiovanni2009).

A ‘Governance’ Approach to Network Analysis?

So far, I have discussed how network analysis can enrich studies of internationalsecurity governance. However, governance studies have a lot to offer network analysis.As many scholars have warned, because they are highly contingent on specificcontexts, the assumptions of social network theory do not always translate well to theworld of international relations (Eilstrup-Sangiovanni and Jones, 2008; Hafner-Burtonet al. 2009: 577; Ward et al. 2010). For example, as Eilstrup-Sangiovanni and Jones(2008) show, while a networked model of organization may be ideally suited to firmsoperating in a competitive global market place it may not work equally well for illicitterrorist groups. Closer engagement with the literature on security governance can helpscholars to ‘translate’ network theoretical concepts into the conceptual vocabulary ofpolitical science and to modify hypotheses drawn from network theories developedprimarily by sociologists, business economists and mathematicians to fit the specificcontext of international security relations.

Another way in which governance theories can inform the application of networkanalytical tools is by bringing ‘agency’ into sharper focus. A frequent criticism offormal network analysis is its tendency to focus only on the structure of relationshipswhile neglecting the diverse attributes of individual actors and organizations that formpart of social networks, as well as the different content of the ties that connect them(see Christopoulos 2008; Kahler 2009: 8). SNA implicitly assumes that all nodes in anetwork possess similar interests and have roughly equal capacity for decision-making,etc.. Yet, in the world of politics, nodes often hold competing ideas, values and goals.Moreover, different actors in a network may have highly differential access to resources(see Christopolous 2008; Ansell and Weber 1999). Thus, in order to present a usefultool for analyzing international security relations, formal network models which tend tosimply assume node preferences, need to be replaced with more agent-sensitive models,which allow for node preferences to differ and conflict, and which consider thepossibility that structurally equivalent nodes might behave very differently according totheir differential access to exogenous resources.

58 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 18/ Date: 17/6

Mette
Inserted Text
by placing fewer formal institutional constraints on their exercise of power
Mette
Inserted Text
2014
Mette
Cross-Out
Mette
Inserted Text
Mette
Inserted Text
social and organizational
Mette
Inserted Text
loose
Mette
Cross-Out
Mette
Inserted Text
,
Mette
Inserted Text
better
Mette
Cross-Out
Mette
Cross-Out
Mette
Comment on Text
Italics!
Mette
Inserted Text
the
Mette
Inserted Text
, tie formation or exit.
Mette
Inserted Text
information-processing,
Mette
Cross-Out
Mette
Inserted Text
material or ideational
Mette
Inserted Text
or the relative availability of 'outside options' (exit power).

JOBNAME: Sperling PAGE: 19 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

A separate problem with the application of assumptions from sociological andmathematical network theories to the world of international politics is that such theoriesrarely conceptualize network actors as strategic (see Lake and Wong 2009). As Kahler(2009) notes, ‘A structural approach to networks typically assumes that agents withinthe networks are not aware of its overall structure and do not act strategically toinfluence or alter that structure’. This, he notes, may be a plausible assumption forlarge-scale social networks. Yet, in the case of smaller, more confined politicalnetworks, it is as plausible to assume that actors are aware of the boundaries andoverall structure of the network and will act deliberately to influence that structure inorder to advance their individual interests. For example, network members may seek tostrategically manage their relationships in order to influence the flow of information ina network, to control overall network membership, or to manipulate the structure of thenetwork itself (Freeman 1979; Scott 2000).

These problems point to a need for better micro-foundations, which can explainbehavior of nodes and provide an element of agency for structural network theories (seeDowding 1995: 140). Such micro-foundations can be developed by drawing on existingIR theories. A governance approach, with its emphasis on complex and overlappingrelationships among diverse actors is particular well suited to providing insights intohow political actors form, maintain and sever relationships, and how (that is, throughwhat strategies or causal pathways) they pursue and gain power within formal andinformal institutional contexts respectively. For example, the wealth of studies ofgovernance within the EU could potentially offer a rich repository of hypotheses abouthow different types of political actors approach the process of network formation andhow they behave within different network structures.

CONCLUSION

In this chapter, I have illustrated that structural and organizational network analysis hasmuch to offer the study of global governance and security. I have argued that many ofthe theoretical and (pre-)theoretical insights of security governance approaches can berefined and subjected to more rigorous empirical testing by using techniques andmetrics of social network theory. Moreover, both structural and organizational networktheories can serve to generate new questions and hypotheses that can expand thesecurity governance research agenda. For example, by challenging and extendingconventional views of power, structural network theory can generate novel conjecturesabout the sources and functions of power in global governance frameworks.

However, there are also reasons to be skeptical of what network analysis can achievein the realm of international relations studies. As Ward et al. (2010: 247) observe,‘because networks reflect structure, network analysis is powerful when the empiricaldata accurately reflect the totality of connections (or accurately reflect the absence ofconnections) between relevant nodes, and when these connections are durable’. Yet,given the difficulties of large-scale data collection, these conditions are rarely met inpractice. Indeed, collecting data for network analysis is often a more complex task thancollecting other types of data typically used in IR research. As Ward et al. (2010: 255)argue, ‘challenges are particularly acute when research questions require complete

Network theory and security governance 59

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 19/ Date: 17/6

Mette
Inserted Text
esp.
Mette
Cross-Out
Mette
Inserted Text
observes
Mette
Inserted Text
(ibid.)
Mette
Cross-Out
Mette
Inserted Text
I suggest,
Mette
Cross-Out
Mette
Inserted Text
might
Mette
Cross-Out
Mette
Inserted Text
more fully on
Mette
Inserted Text
, in particular theories of governance.
Mette
Cross-Out
Mette
Cross-Out
Mette
Inserted Text
sought to illustrate
Mette
Cross-Out
Mette
Inserted Text
governance
Mette
Inserted Text
they note,

JOBNAME: Sperling PAGE: 20 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

network data, since this task requires determining both the boundary of the relevantpopulation and the presence or absence of a tie between each pair of nodes’.Sometimes, as in studies of international organizations, the boundaries of a group ofnetwork members are well defined. In other cases – such as in studies of illicit criminalorganizations or of communities defined primarily by adherence to common norms andprinciples such as for example loose transnational advocacy networks – the boundaryof the relevant population is not known in advance and reliable data on all nodes andlinks may be difficult to gather. This in turn weakens the applicability of SNA asmeasures of network centrality or density may be highly sensitive to missing data(ibid).

Network analysis is no panacea for IR scholars and will not substitute for carefulwork to establish the micro-foundations that provide agency for models of networkedinteractions in a political world. However, the clear focus on the structural determinantsof individual and collective behavior, as well as the wide range of methodological toolsfor identifying and measuring the structural properties of social systems, means thatSNA has a lot to offer scholars studying global governance processes.

NOTE

1. For a discussion of the applicability of network theory to international relations in general, seeHafner-Burton et al. (2009). Siegel (2011) reviews network analysis in comparative politics.

REFERENCES

Abbott, K.W. and D. Snidal (2000), ‘Hard and soft law in international governance’, InternationalOrganization, 54: 421–56.

Ansell, C.K. (2000), ‘The networked polity: regional development in Western Europe’, Governance, 13:279–91.

Ansell, C.K. and St E. Weber (1999), ‘Organizing international politics: sovereignty and open systems,’International Political Science Review, 20 (1): 73–93.

Barabasi, A.L. (2003), Linked: How Everything Is Connected to Everything Else and What It Means forBusiness, Science, and Everyday Life, London: Plume.

Brie, M. and E. Stölting (2011), ‘Formal institutions and informal institutional arrangements’, in T.Christiansen and C. Neuhold (eds), International Handbook on Informal Governance, Cheltenham, UKand Northampton, MA, USA: Edward Elgar.

Burt, R.S. (1976), ‘Positions in Networks’, Social Forces, 55: 93–122.Burt, R.S. (1992), Structural Holes: The Social Structure of Competition, Cambridge: Cambridge University

Press.Burt, R.S. (2004), ‘Structural Holes and Good Ideas’, The American Journal of Sociology, 100: 349–99.Carpenter, C.R. (2011), ‘Vetting the advocacy agenda: network centrality and the paradox of weapons

norms’, International Organization, 65: 69–102.Castells, M. (1996), The Rise of the Network Society, Cambridge, MA: Blackwell Publishers.Chayes, A.H., A. Chayes and G. Raach (1997), ‘Beyond reform: restructuring for more effective conflict

intervention’, Global Governance, 3: 117–46.Christopoulos, D.C. (2008), ‘The governance of networks: heuristic or formal analysis? A reply to Rachel

Parker’, Political Studies, 56: 475–81.Cronin, A.K. (2002/03), ‘Behind the curve: globalization and international terrorism’, International

Security, 27: 30–58.

60 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 20/ Date: 17/6

Mette
Inserted Text
,
Mette
Inserted Text
,
Mette
Cross-Out
Mette
Inserted Text
some
Mette
Cross-Out
Mette
Inserted Text
see opcit.
Mette
Inserted Text
blobal
Mette
Cross-Out
Mette
Inserted Text
context
Mette
Cross-Out
Mette
Inserted Text
a
Mette
Comment on Text
italics
Mette
Cross-Out
Mette
Inserted Text
a
Mette
Inserted Text
sophisticated
Mette
Inserted Text
different
Mette
Inserted Text
provides a
Mette
Cross-Out
Mette
Inserted Text
of
Mette
Inserted Text
the field

JOBNAME: Sperling PAGE: 21 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

Daase, C. and S. Engert (2008), ‘Global security governance. Kritische Anmerkungen zur Effektivität undLegitimität neuer Formen der Sicherheitspolitik’ [Global security governance. Critical comments on theeffectiveness and legitimacy of new forms of security policy], Politische Vierteljahresschrift, 41: 475–98.

Davis, D.R. and A. Murdie (2012), ‘Looking in the mirror: comparing INGO networks across issue areas’,unpublished paper, accessed 21 June 2013 at https://courses.cit.cornell.edu/patel/psac/Davis_PSAC_3_9.pdf.

Dowding, K. (1995), ‘Model or metaphor? A critical review of the policy network approach’, PoliticalStudies, 43: 136–58.

Duffield, J.S. (2006), ‘International security institutions: rules, tools, schools or fools?’, in R.A.W. Rhodes,S.A. Binder and B.A. Rockman (eds), The Oxford Handbook of Political Institutions, Oxford: OxfordUniversity Press.

Eilstrup-Sangiovanni, M. (2009), ‘Varieties of cooperation: government networks in international security’,in M. Kahler, (ed.), Networked Politics: Agency, Power, and Governance, Ithaca, NY: Cornell UniversityPress.

Eilstrup-Sangiovanni, M. and C. Jones (2008), ‘Assessing the dangers of illicit networks. Why Al-Qaedamay be less dangerous than many think’, International Security, 33 (2): 7–44.

Emirbayer, M. (1997), ‘Manifesto for a relational sociology’, American Journal of Sociology, 103: 281–317.Enders, W. and P. Jindapon (2010), ‘Network externalities and the structure of terror networks’, Journal of

Conflict Resolution, 54: 262–80.Erikson, B.H. (1981), ‘Secret societies and social structure’, Social Forces, 60: 188–210.Freeman, L. (1979), ‘Centrality in social networks’, Social Networks, 1: 215–39.Goddard, S. (2009), ‘Brokering change: networks and entrepreneurs in international politics’, International

Theory, 1: 249–8.Granovetter, M. (1973), ‘The strength of weak ties’, American Journal of Sociology, 78: 1360–80.Gulati, R. and M. Gargiulios (1999), ‘Where do interorganizational networks come from?’, American

Journal of Sociology, 104: 1439–93.Gulati, R., D.A. Dialdin and L. Wang (2002), ‘Organizational networks,’ in J.A.C. Baum (ed.), The

Blackwell Companion to Organizations, Oxford: Blackwell.Hafner-Burton, E. and A.H. Montgomery (2006), ‘Power positions: international organizations, social

networks, and conflict’, Journal of Conflict Resolution, 50: 3–27.Hafner-Burton, E., M. Kahler and A.H. Montgomery (2009), ‘Network analysis for international relations’,

International Organization, 63: 559–92.Jönsson, C. (1986), ‘Interorganization theory and international organization’, International Studies Quar-

terly, 30: 39–57.Kahler, M. (2000), ‘Conclusion: the causes and consequences of legalization’, International Organization,

54: 661–83.Kahler, M. (ed.) (2009), ‘Networked politics: agency, power, and governance’, accessed 27 March 2014 at

http://ilar.ucsd.edu/assets/014/6727.pdf.Keck, M.E. and K. Sikkink (1998), Activists beyond Borders: Advocacy Networks in International Politics,

Ithaca, NY: Cornell University Press.Kenney, M. (2007), From Pablo to Osama: Trafficking and Terrorist Networks, Government Bureaucracies,

and Competitive Adaptation, University Park, PA: Penn State University Press.Kirchner, E.J. (2006), ‘The challenge of European Union security governance’, Journal of Common Market

Studies, 44: 945–66.Kirchner, E. and J. Sperling (2007), EU Security Governance, Manchester: Manchester University Press.Krahmann, E. (2001), ‘The emergence of security governance in post-Cold War Europe’, ECPR Working

Paper 36/01, accessed 21 June 2013 at www.one-europe.ac.uk/pdf/w36krahmann.pdf.Krahmann, E. (2003), ‘Conceptualizing security governance’, Cooperation and Conflict, 38: 5–26.Krahmann, E. (2005), ‘Security governance and networks’, Cambridge Review of International Affairs, 18:

19–34.Lake, D.A. (1999), ‘Global governance: a relational contracting approach’, in A. Prakash and J.A. Hart

(eds), Globalization and Governance, London: Routledge.Lake, D.A. (2001), ‘Beyond anarchy: the importance of security institutions’, International Security, 26 (1):

129–60.Lake, D.A. and W. Wong (2009), ‘The politics of networks: interests, power, and human rights norms’, in

M. Kahler (ed.), Networked Politics: Agency, Power, and Governance, Ithaca, NY: Cornell UniversityPress.

Network theory and security governance 61

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 21/ Date: 17/6

JOBNAME: Sperling PAGE: 22 SESS: 3 OUTPUT: Tue Jun 24 13:11:19 2014

Lipson, M. (2005), ‘Interorganizational coordination in complex peacekeeping’, paper presented at theannual meeting of the International Studies Association, Honolulu, 1–5 March.

Maoz, Z. (2012), ‘How network analysis can inform the study of international relations’, ConflictManagement and Peace Science, 29: 247–56.

Marin, B. and R. Mayntz (eds) (1991), Policy Networks: Empirical Evidence and Theoretical Consider-ations, Boulder, CO: Westview.

Marvel, S.A., J. Kleinberg, R.D. Kleinberg and S.H. Strogatz (2011), ‘Continuous-time model of structuralbalance’, Proceedings of the National Academy of Sciences, 108: 1771–6.

Mayntz, R. (2004), ‘Organizational forms of terrorism: hierarchy, network, or a type sui generis?’, MPIfGDiscussion Paper, No. 04/4, Max-Planck-Institut für Gesellschaftsforschung, Cologne.

Mérand F., S. Hofmann and B. Irondelle (2011), ‘Governance and state power: a network analysis ofEuropean security’, Journal of Common Market Studies, 49: 121–47.

Murdie, A. (2014), ‘The ties that bind: a network analysis of human rights INGOs’, British Journal ofPolitical Science, 44: 1–27.

Nexon, D.H. and T. Wright (2007), ‘What’s at stake in the American Empire debate’, American PoliticalScience Review, 101: 253–71.

Ohanyan, A. (2010), ‘Network institutionalism and NGO studies’, paper presented at the InternationalStudies Association Annual Convention 2010, New Orleans, 16–19 February.

Paris, R. (2009), ‘Understanding the “coordination problem” in postwar statebuilding’, in R. Paris and T. D.Sisk (eds), The Dilemmas of Statebuilding, London: Routledge.

Podolny, J.M. and K.L. Page (1998), ‘Network forms of organization’, Annual Review of Sociology, 24:57–76.

Powell, W. (1990), ‘Neither market nor hierarchy: network forms of organization’, Research in Organ-izational Behavior, 12: 295–336.

Raab, J. and H.B. Milward (2003), ‘Dark networks as problems’, Journal of Public Administration Researchand Theory, 13: 413–39.

Ricigliano, R. (2003), ‘Networks of effective action: implementing an integrated approach to peace-building’, Security Dialogue, 34: 445–62.

Rosenau, J.N. and E.-O. Czempiel (eds) (1992), Governance without Government: Order and Change inWorld Politics, Cambridge: Cambridge University Press.

Sageman, M. (2004) Understanding Terror Networks, Philadelphia, PA: University of Pennsylvania Press.Scharpf, F.W. (1993), ‘Coordination in hierarchies and networks’, in F.W. Scharpf, (ed.), Games in

Hierarchies and Networks: Analytical and Empirical Approaches to the Study of Governance Institutions,Boulder, CO: Westview.

Schroeder, U.C. (2006), ‘Coping with complexity. An organizational perspective on European securitygovernance’, EUI Working Papers 2006/09, European University Institute, Florence.

Scott, J. (2000), Social Network Analysis. A Handbook, 2nd edn, London: Sage.Siegel, D.A. (2011), ‘Social networks in comparative perspective’, PS: Political Science, 44: 51–4.Smith-Doerr, L. and W.W. Powell (2005), ‘Networks and economic life’, in N. Smelser and R. Swedberg

(eds), Handbook of Economic Sociology, 2nd edn, Princeton, NJ: Princeton University Press.Sperling, J. (2008), ‘State attributes and system properties: security multilateralism in central Asia, southeast

Asia, the Atlantic and Europe,’ in D. Bourantonis, K. Ifantis and P. Tsakonas (eds), Multilateralism andSecurity Institutions in an Era of Globalization, Abingdon: Routledge.

Thompson, G.F. (2003), Between Hierarchies and Markets: The Logic and Limits of Network Forms ofOrganization, New York: Oxford University Press.

Ward, M.D., K. Stovel and A. Sacks (2010), ‘Network analysis and political science’, Annual Review ofPolitical Science, 14: 245–62.

Warren, C.T. (2010), ‘The geometry of security: modelling interstate alliances as evolving networks’,Journal of Peace Research, 47: 697–709.

Watts, D.J. (2003), Six Degrees: The Science of a Connected Age, London: William Heinemann.Webber, M., S. Croft, J. Howorth, T. Terriff, and E. Krahmann (2004), ‘The governance of European

security’, Review of International Studies, 30: 3–26.Westerwinter, O. (2013), ‘The politics of transnational institutions: power, bargaining, and institutional

choice’, PhD dissertation, European University Institute, Florence.Williams, P. (2001), ‘Transnational criminal networks’ in J. Arquilla and D. Ronfeldt, (eds), Networks and

Netwars: The Future of Terror, Crime, and Militancy, Santa Monica, CA: RAND.Williamson, O.E. (1975), Markets and Hierarchies: Analysis and Antitrust Implications. A study in the

economics of internal organization, New York: Free Press.

62 Handbook of governance and security

Columns Design XML Ltd / Job: Sperling-Handbook_of_Governance_and_Security / Division: 03-Chapter3-Eilstrup-Sangiovannits /Pg. Position: 22/ Date: 17/6