Networks of Clusters: A Governance Perspective

Elke Schüßler
Freie Universität Berlin
School of Business and Economics
Department of Management
Boltzmannstr. 20
D-14195 Berlin, Germany
Phone: +49 30 838 56807
Fax: +49 30 838 56808
e-Mail: Elke.Schuessler@fu-berlin.de

Carolin Decker
WHU - Otto Beisheim School of Management
Burgplatz 2
D-56179 Vallendar, Germany
Phone: +49 261 6509 333
Fax: +49 261 6509 339
e-Mail: Carolin.Decker@whu.edu

Frank Lerch
Helmut-Schmidt-University Hamburg
Department of Economics and Social Science
Holstenhofweg 85
D-22043 Hamburg, Germany
Phone: +49 40 6541 2080
Fax: +49 40 6541 2087
e-Mail: frank.lerch@hsu-hh.de
 Networks of Clusters: A Governance Perspective

Abstract

This paper aims to further our emerging knowledge on the external linkages of clusters. We
adopt a network governance perspective and study connected clusters as goal-directed, multi-
level whole networks that we denote as ‘cluster networks’. Based on an analysis of four
empirical cases varying in regional scope, age, and industry context, we identify two
governance forms: internally governed cluster networks are formed to establish ties among
cluster representative organizations to share knowledge and pool resources on selected
activities; externally governed cluster networks are formed to systematically develop cross-
cluster ties and competences on and across levels and are brokered by a central inter-cluster
administrative organization and several decentralized lead organizations. Our findings show
that cluster connectivity can go beyond organization-based pipelines and personal
relationships to include clusters as governed entities, albeit with different intensities
regarding the brokerage of ties on lower levels.

Keywords: regional clusters, cluster collaboration, cluster network, cluster connectivity,

whole networks, governance, global pipelines, territorial innovation systems

Acknowledgements: We gratefully acknowledge the insightful comments received from

participants at research seminars at Freie Universität Berlin and Tilburg University and from
conference participants at the European Group for Organizational Studies Colloquium (2011)
and the Workshop on Organization Studies of the German Association for Business Research
(2012). We are particularly indebted for their detailed feedback on earlier drafts to Stephan
Kaiser, Leon Oerlemans, Christina Günther, Vincent Mangematin, and Jörg Sydow. Joachim
Thiel helped us in refining the title. The editor Mark Lorenzen and three anonymous
reviewers have contributed significantly towards improving the paper through their valuable
developmental advice. All remaining errors are ours.

JEL: O18, D85, L14

1. Introduction

Regional clusters, defined as “geographic concentrations of interconnected companies,

specialized suppliers, service providers, firms in related industries, and associated industries

in a particular field that compete but also co-operate” (Porter, 1998: 197), have attracted

researchers’ interest for decades (e.g. Bresnahan et al., 2001; Hoover, 1937; Marshall, 1890;

Scott, 2002; Staber, 1997). In efforts to specify the functioning of this highly contested

concept (Martin and Sunley, 2003), research has paid attention to the actual networks

forming within (Visser, 2009; Visser and Boschma, 2004), but also between different clusters

(Asheim and Isaksen, 2002; Boschma, 2005; Humphrey and Schmitz, 2002; Lorenzen and

Mudambi, 2012). Such often non-local, external linkages of clusters are established by

gatekeepers (Rychen and Zimmermann, 2008), technical, entrepreneurial or transnational

communities (Saxenian, 2002, 2006; Saxenian and Hsu, 2001), global pipelines (Bathelt et

al., 2004), or temporary encounters (Maskell et al., 2006).

Although external linkages are recognized as crucial for developing and maintaining

the innovativeness of clusters (Ter Wal and Boschma, 2011), only few studies examine the

idiosyncratic, multi-level ties established between clusters to achieve certain goals (e.g. Engel

and del-Palacio, 2011). Such an analysis is necessary to understand how cluster-

administrative bodies or other organizations within clusters can support the development of

shared goals and institutionalized mechanisms of problem-solving, the accessing of markets

and funding sources, or the transfer of knowledge in the complex web of centralized and

decentralized, strong and weak, organization- and person-based ties characterizing cross-

cluster linkages (Bathelt et al. 2004; Gilding, 2008; Lorenzen and Mudambi, 2012; Visser

and Atzema, 2008). We therefore propose to study connected clusters as multi-level whole

networks (Provan et al., 2007) that require governance to achieve common goals (Provan and

Kenis, 2008). A governance perspective allows us to shed light on a variety of goals and

1
forms of cluster connectivity by considering the cluster and the inter-cluster levels as

additional levels of analysis on which linkages may be formed. We analyze four empirical

cases of cluster collaboration varying in age, size, regional scope, and industry to understand

what governance approaches are used to ensure goal alignment on and across cluster and

inter-cluster levels.

Our contributions are twofold. First, we identify internally and externally governed

cluster networks as different forms of governing cluster collaboration. Whereas organization-

based pipelines (Gupta and Govindarajan, 2000; Mudambi and Swift, 2012) and person-

based relationships across clusters (Saxenian, 2002, 2006) are more or less emergent from a

cluster perspective, cluster networks are designed by and systematically involve cluster-level

actors. These can be private organizations acting on behalf of clusters, cluster-administrative

organizations (CAOs), or inter-CAOs. Second, we discuss the impact of structural and

relational factors on the formation of cluster networks to elaborate on the rationales behind

different governance forms. Especially the absence of strong global pipelines can be a

motivation for forming cluster networks, and externally governed cluster networks are more

likely to occur in neighboring rather than globally distant regions.

We proceed as follows. First, we outline levels and factors pertaining to the

governance of cluster collaboration. Second, we elaborate on our cases, data sources and

analysis. Third, based on our findings, we specify and compare two governance forms and

outline the implications of our results.

2. Governing Cluster Collaboration

2.1 Connected Clusters as Goal-Directed, Multi-Level Whole Networks

Existing research on cluster connectivity distinguishes person-based and organization-based

linkages (Lorenzen and Mudambi, 2012). Person-based linkages result from transregional or

2
transnational entrepreneurs promoting the transfer of knowledge from one cluster to another

(Saxenian, 2002, 2006). Organization-based linkages are typically formed by multi-national

enterprises (MNEs) that establish relationships with subsidiaries or alliance partners and

gatekeepers in other clusters to nurture the transfer of proprietary knowledge (Cheung et al.,

2011; Giblin, 2011; Vapola, 2011). Although a limited number of actors within a cluster

benefits from such ties, their utility for entire clusters is unclear (Henn, 2012).

We conceive of cross-cluster ties as goal-directed rather than serendipitous (Kilduff

and Tsai, 2003), based on “conscious efforts to build coordination” (Provan and Kenis, 2008:

231) across clusters. Although serendipitous interactions may also occur, clusters cooperate

for a specific cluster-related purpose such as enhancing participants’ innovative abilities,

attracting public funding, or strengthening regional economic development. They are

frequently supported by national or regional policies (Asheim and Coenen, 2005; Dohse,

2007). This goal-directedness requires governance “to ensure that participants engage in

collective and mutually supportive action, that conflict is addressed, and that network

resources are acquired and utilized efficiently and effectively” (Provan and Kenis, 2008:

231).

Building on and extending Provan and Kenis’ (2008) network governance approach,

we define cluster networks as multi-level whole networks that, compared to networks,

additionally comprise the levels of clusters and inter-cluster linkages. More precisely, actors

participating in cluster networks are network partners residing in different clusters that, in

turn, are embedded in networks among clusters. A shared institutional context enables

learning within and across clusters. It guides the transfer of knowledge, project guidelines,

and joint goals from organizational to network, cluster, and inter-cluster levels or directly

from the organizational to the inter-cluster level – or vice versa (Bathelt et al., 2004; Bunker

Whittington et al., 2009; Gilding, 2008; Menzel and Fornahl, 2010).

3
 Governance can be categorized along the dimensions of brokerage and leadership

(Provan and Kenis, 2008). On the cluster level, hub firms (Jarillo, 1988) can have

orchestrating functions. Alternatively, similar to network administrative organizations

(Human and Provan, 2000), CAOs are established as external cluster leaders (Sydow et al.,

2011). They monitor goal-directed activities on the inter-cluster level and endow an inter-

CAO with resources, power, and legitimacy to coordinate decisions and collective actions so

as to establish a centralized, external lead organization. Organizations or individuals residing

within clusters can act as brokers of ties to other clusters on and across different levels to

bridge “structural holes” (Burt, 1992).

The degree of centralization depends on the number of specialized investments in

cluster collaboration. Generally, the leaders of cluster networks are responsible on behalf of

their clusters for “making things happen” on the inter-cluster level (Huxham and Vangen,

2005). Such leadership involves intense brokering of ties on and to ‘lower’ levels if only a

limited number of actors have sufficient power and resources to take a leading role. If power

and resources are symmetrically distributed, network actors refrain from a centralized

brokerage (Provan and Kenis, 2008).

The formalization and institutionalization of inter-cluster leadership depends on the

extent of public funding and on financing models. Formal institutions need to be established

if national or supra-national innovation policies support cross-cluster linkages (Asheim and

Coenen, 2006; Dohse, 2000). Further contingencies reflect the goals and multi-level character

of cluster networks and influence their governance (Provan and Kenis, 2008).

2.2 Relational and Structural Contingencies of Cluster Networks

Goal consensus requires agreement on network-level content and process-related

goals, but the goals of network members need not be similar (Provan and Kenis, 2008).

Brokers and leaders in cluster networks define common goals and mobilize for goal

4
alignment (Huxham and Vangen, 2000). Goal consensus is also influenced by the regional

scope of a cluster network. For instance, supra-national initiatives such as those pursued by

the European Union nurture cluster collaboration within clearly defined geographic and

political boundaries and align goals to some extent through funding incentives (Kuhlmann

and Edler, 2003). Cluster networks can also comprise inter-continental relationships (Bathelt

et al., 2004; Engel and del-Palacio, 2011; Gilding, 2008) to manufacture cognitive, cultural

and temporary proximity (Boschma, 2005; Bunker Whittington et al., 2009). Goal consensus

may be moderate or high on the inter-cluster level, but much lower on the level of

organizations residing within clusters. If long-term cluster network sustainability is to be

achieved, the governance of cluster networks should involve at least a subset of member

organizations on different levels (Provan and Kenis, 2008).

Trust is related to governance (Provan and Kenis, 2008) and depends on the

intensities and durations of the linkages between clusters. Ties may be recurring or sporadic

(Granovetter, 1973). While trust density may remain low in cluster networks overall, it can be

high in selected inter-organizational networks across clusters. Trust-based ties among

governing actors on the inter-cluster level suffice to make the cooperation effective.

Generally, the nature of cross-cluster ties is expected to correlate with the nature of local ties

(Lorenzen and Mudambi, 2010). If clusters lack prominent lead organizations and instead

have trust-based personal relations, shared governance on the inter-cluster level is likely.

Cluster networks comprise ties between various numbers of participants on different

levels (He and Fallah, 2011). On the inter-cluster level, collaboration is unlikely to exceed a

small number of clusters, but each cluster typically comprises many organizations residing

within them whose networking needs to be facilitated by adequate governance structures. The

higher the number of participants on different levels, the higher the need is for a central entity

that coordinates their needs (Provan and Kenis, 2008).

5
 Cluster networks require cluster and inter-cluster level competencies that depend on

the nature of the tasks that are performed and the demands and needs that are faced by the

participants (Provan and Kenis, 2008). If innovation is the main goal, a competence must be

developed that ensures participants’ access to complementary knowledge located in the

collaborating clusters and its transfer to the place where it is needed (Herrmann et al., 2012).

The need for cluster and inter-cluster level competences also varies depending on the industry

in which cluster networks are established. Industries can be supplier-dominated and

production-intensive, where existing knowledge is exploited to nurture incremental

innovation that benefits from interactive learning with customers and suppliers, such as the

automotive industry. Science-based industries such as the biotechnology industry aim at

radical innovation often pursued collaboratively by companies and research organizations

(Asheim and Coenen, 2005; Pavitt, 1984). Governance depends on the kind of knowledge

that is exchanged and the type of innovation to be pursued in an industry (Lorenzen and

Mudambi, 2012). While formal governance is beneficial in transferring proprietary

knowledge among selected actors, radical innovation requires a governance approach that

allows broader spillovers across clusters (Bunker Whittington et al., 2009).

3. Case Analysis

Since our knowledge about cluster networks is limited, we have chosen a multiple-case

design that allows us to uncover different facets of this phenomenon (Yin, 2009). We have

selected four cases of collaborating clusters differing in age, size, regional scope, and

industry so as to identify and explain commonalities and differences in their governance. We

have identified these factors as relevant for the governance of cluster networks on the basis of

theoretical considerations. The knowledge base of an industry, for instance, influences the

6
formality of governance forms; size affects the degree of centralized brokerage; regional

scope has an effect on goal consensus and the extent to which governance is shared.

The first case is the Automotive Cluster East Germany (ACOD), a national

collaboration among five state-based automotive networks and cluster initiatives initiated in

2004 and supported by national public funding. The second case is Germany goes 3D that

emerged in 2010, a national collaboration between the three leading German pop music

industry clusters in Berlin, Cologne, and Hamburg. The third case is the transcontinental Tri-

Cluster Berlin-Tucson-Ottawa Alliance between the photonics clusters in Berlin-

Brandenburg, Tucson (Arizona), and Ottawa (Ontario), initiated in 2004. The fourth case is

the ScanBalt BioRegion, a transnational network of biotechnology clusters and organizations

from eleven countries in the Baltic Sea Region initiated in 2001.

We conducted five to six interviews per case between 2007-2012, focusing on

informants from different organizations who had been involved – continuously or at key

points in time – in governing cross-cluster linkages either as cluster leaders or as

representatives of the inter-CAOs. Typically, we have addressed our interviewees in two

roles: as representatives of a cluster with respective interests, and as experts in the goals and

governance of the respective cluster network. We also interviewed one European

Commission official involved in fostering cluster collaboration in the photonics industry to

gain insights into policy issues. Altogether, we conducted 29 semi-structured interviews (see

Table 1), which took 30 minutes on average.

--- Insert Table 1 about here ---

The interviews covered five broad areas: (1) general information about each initiative,

such as starting-point and evolution, initiators and members, expectations and goals, (2)

interaction, such as opportunities for communication and cooperation on and across levels,

rules, guidelines, and practices for cluster collaboration, (3) governance, such as formal and

7
informal control mechanisms, conflicts, and mechanisms for achieving goal consensus and

conflict resolution, (4) background information and interviewees’ personal assessment, such

as specific purpose of the collaboration, key communication and cooperation partners, type of

knowledge to be transferred, time, efforts and money to be invested by the partners, and (5)

context, such as competition, territorial innovation policies (including public funding), and

regulative and legislative pressures. If the interviewees agreed, the interviews were digitally

recorded, transcribed and coded to categorize the collected data. For triangulation purposes,

we analyzed a broad range of documents (e.g. strategy papers, newsletters, roadmaps, and

reports) as background information, and, where possible, studied the initiatives’ websites for

additional material. Attending meetings and workshops gave us further cross-validation

opportunities.

--- Insert Table 2 about here ---

As a first step of analysis, we wrote up brief histories of the development of each case before

coding for our core theoretical constructs in more depth, which are outlined in Table 2. The

constructs are useful because they reflect the goal-directedness and multi-level character of

cluster networks, as well as the different bodies involved in governance and their interactions.

We have thus engaged in a largely deductive coding process based on theoretically derived

dimensions. At least two authors analyzed each of our cases in depth, so that we could

compare our impressions and gain shared insights into all of our cases. Once we had roughly

coded and discussed the governance structures and contingencies for each case, we compared

and contrasted our findings across cases. It quickly became obvious that ACOD and ScanBalt

were similar regarding their goals, complexity and formality, whereas the Tri-Cluster

Alliance and Germany 3D were comparable regarding their informal and hollow nature.

Comparing the similarities and differences within each of these pairs helped us to develop

and refine our typology. For instance, even though Tri-Cluster and 3D differ substantially

8
regarding their regional scope and industry, they nonetheless pursue similar goals such as the

sharing of limited resources and gaining political visibility, which has affected the

governance form chosen.

4. Findings

4.1 Case Histories

4.1.1 ACOD

ACOD, founded in 2004 on the initiative of the original equipment manufacturers (OEMs), is

engaged in the automobile industry in East Germany. It comprises five regional state

initiatives (AutoMobilZulieferer Sachsen, automotive Berlin-Brandenburg e.V., automotive

Mecklenburg-Vorpommern e.V., automotive Thüringen e.V., and MAHREG

Automotive/Sachsen-Anhalt Automotive e.V.) and about 370 organizations (OEMs, suppliers

and service providers, research institutes, associations, and other institutions that are related

to the automotive industry). ACOD’s goal is the development of the automobile industry in

East Germany, which mainly comprises small and medium-sized enterprises (SMEs) that aim

to cooperate with the OEMs but do not fully match the requirements of these globally

operating companies:

“We need to organize for a higher added value and innovation performance in the supply
chain, and we need to increase the R&D performance of the local supplier companies. In
particular, we are supporting the small and medium sized companies to professionalize and to
climb the tier-hierarchy.” (ACOD interview)

ACOD fosters the national and international interest in the East German automotive industry

by providing platforms (e.g. International Motor Show Cars, the Hanover International Motor

Show Commercial Vehicles, or Subcontracting Fair “Z” in Leipzig). To promote East

Germany as a European center for high-tech products for the automotive sector, it provides

contacts to influential political institutions on a national and a state level.

9
4.1.2 Germany goes 3D

Founded in 2010, Germany goes 3D connects the three main popular music regions in

Germany: Berlin, Hamburg, and Cologne. It is represented by Cologne-based c/o pop GmbH,

the Berlin Music Commission, and the Hamburg-based Reeperbahn Festival GbR and IHM

(Interessengemeinschaft Hamburger Musikwirtschaft). The collaboration has been

established as a working group in which members share selected interests and topics that

permit the pursuit of joint aims such as resource sharing and gaining visibility among national

politicians despite otherwise fierce regional competition. Salient activities are, for instance,

joint presentations at trade fairs abroad and the 3D Receptions during the Berlin Music Week,

the Reeperbahn Festival in Hamburg, and the c/o pop in Cologne. These receptions have the

advantage that international partners can meet representatives of each partnering region

during each event rather than having to choose between one of the locations.

4.1.3 The Tri-Cluster Berlin-Tucson-Ottawa Alliance

The Tri-Cluster Alliance, formed in 2004, is an international collaboration of three clusters in

the photonics industry. It aims to facilitate market access, intensify communication, and

increase knowledge about the competences of the different firms residing in the respective

clusters in order to, for example, support the search for suppliers for specific goods and pool

resources where possible. A further aim is to increase the visibility of the value of the

photonics industry vis-à-vis policy makers and customers. The collaboration started with the

organization of a joint Tri-Cluster breakfast at the Photonics West trade fair in San Jose and

has been followed up by several mutual visits. The Tri-Cluster breakfast has taken place

regularly, alternating between the Photonics West and the “LASER World of Photonics” in

Munich until 2011. It is currently being debated whether another breakfast will take place in

2013, so that the future of this cluster network is unclear.

10
4.1.4 ScanBalt BioRegion

ScanBalt is a meta-region in life sciences around the Baltic Sea that includes Denmark,

Estonia, Finland, Iceland, Latvia, Lithuania, Norway, Poland, Sweden, the northern part of

Germany, and the north-western part of Russia. It currently comprises about 4,000

organizations (2,543 life science or life science related companies, 1,012 academic institutes

and institutions, 238 hospitals and clinics, and 112 investors). The central goal of ScanBalt is

to generate synergies and ensure the optimal exploitation of joint resources in the Baltic Sea

region so as to achieve scale in global competition:

“We know that the universities in the Baltic States are not top rated, but in every university
there are several groups that lead the world in their fields. So, why not go together and
network so that the ScanBalt BioRegion can become critical and visible.” (ScanBalt
interview)

After initial collaborative efforts in 2001, in June 2004 a not-for-profit membership

association was established. In 2006, ScanBalt suggested the foundation of a Baltic Sea Fund

for Innovation, Research and Education that became a flagship project in the EU Baltic Sea

Region Strategy.

4.2 Comparing Cluster Networks

4.2.1 Governance Structures

The governance of ACOD and ScanBalt is formal, complex, and brokered to a high extent.

Both initiatives have founded legal entities that act as external lead organizations. At ACOD,

the central executive body is the Board of Directors, composed of members of the inter-

regional association (ACOD e.V.) that defines the basic cluster development processes. Its

members – among them at least one representative of a state initiative – maintain key

contacts on national, international, and regional levels and have to consider the special

characteristics of each state initiative. Leadership is rotating, i.e. the chairman’s role changes

regularly among member organizations. The association ACOD e.V. has statutes that

11
delineate the basic lines of collaboration. It is the shareholder of ACOD GmbH, a limited

liability company that executes operational activities and constitutes the inter-cluster

administrative office. ACOD GmbH acts as a broker of information between the competence

networks, the state initiatives and the organizations residing within them, and representatives

from the local political realm:

“(…) we have the association, and there is the member assembly, which is the formal way.
Once a month we have a meeting with the state initiatives about activities and outcomes that
are pertinent. The other contacts focus on events and activities and are bilateral. But in
general, we have one meeting per month with the state initiatives and two member assemblies
and six board meetings per year.” (ACOD interview)

ACOD GmbH initiates and manages activities on the inter-cluster level such as workshops,

the ACOD conference, the presence of ACOD at trade fairs, and various competence

networks that cover a wide range of expertise:

“These competence networks are governed by one or two persons who do not come from the
central administrative office but are representatives of research institutes or companies and
pursue their own interests with a topic.” (ACOD interview)

The five state initiatives are directly connected to the association ACOD e.V. via the Cluster

Hub, a coordination platform. Its task is to align the goals and activities of the state initiatives

that are responsible for supplier development in their respective regions and to pursue “a

constructive dialogue without segregating conflict” (ACOD interview). The OEMs and the

state initiatives have strong voices at the inter-cluster level. When it comes to specific

projects such as the development of new technologies, individual research organizations or

companies take the lead in organizing the project activities. There is a clear separation of

competences between the cluster and the inter-cluster level, because the state initiatives adopt

different approaches and opportunities to support their members:

“Activities across states are the ACOD’s task, we are more likely to focus on the states.”
(ACOD interview, representative of a state initiative)

12
However, competences that are developed on the state level, such as the ability to cooperate

with regional clusters abroad, can be transferred into ACOD so that other participants may

benefit.

ACOD e.V. draws its funding from membership fees. Paying members are the OEMs

with plants in East Germany, a handful of systems suppliers, and influential research

institutes. It is also co-financed by public funds, but these are due to decrease over the next

years. Funding differs among the state initiatives that benefit from ACOD as a unified

network, in that it helps them to apply for national funding:

“As a platform we have the best conditions to raise funds from the Federal Ministry of
Economics and Technology.” (ACOD interview)

ScanBalt has founded a non-profit organization with a defined membership structure, an

Executive Committee with a Chair, General Secretary and a Secretariat, while formal Statutes

and Standing Orders are approved and adopted by a General Assembly. The central network

broker, the one-person Secretariat based in Copenhagen supported by four liaison offices

Copenhagen (Denmark), Tartu (Estonia), Gdansk (Poland), and Groningen (Netherlands) has

mainly coordinative functions such as initiating EU funding applications or distributing

information both within the macro region and between it and the EU. On an operational level,

leadership within ScanBalt is more dispersed and based on projects. Despite the

establishment of a central governing body, which became necessary due to an EU-funded

project, the collaboration has retained a highly decentralized structure:

“It was very important for us to stress the bottom-up nature, since you need to have the
regions take ownership of such a thing as ScanBalt Bioregion because you cannot sit at the
top and expect people to participate in anything and be responsible to a centralized
organization; that would be hopeless. The second point is that money cannot cross borders.
This means you need to have activities placed out regionally and to have regions themselves
do things but referring to an overall strategy, vision and mission.” (ScanBalt interview)

Both ScanBalt and ACOD combine a centralized external broker responsible for distributing

information, developing a shared strategy and vision, providing an interface for policy-

makers, and coordinating activities with decentralized internal lead organizations

13
representing the different member clusters or interorganizational networks that have formed

or are forming within and across the collaborating clusters.

The ScanBalt central leadership promotes a variety of activities. For instance, it has

initiated thematic networks to nurture collaboration within, for example, clinical research,

marine biotechnology, or stem cell research. It has fostered shared SME services between

clusters and public/private collaboration in the field of biosystems technologies. The

ScanBalt Academy has been implemented to act as an external advisory and ambassador’s

body. ScanBalt exerts socio-political influence by publishing opinion papers, for example,

focusing on stem cell research, and helps to create public awareness for the biotechnology

sector in the partner regions and social acceptance for sometimes contested research projects.

Since 2004, there have been approx. 1,700 mentions in the press. ScanBalt members also

initiated the ScanBalt Health Region. It became a flagship project in the EU Baltic Sea

Region Strategy and modeled its governance structure on ScanBalt’s bottom-up approach,

which insists that most of the work needs to be done in concrete projects in the regions:

“There is only a very small baseline funding, and it should stay like this. It is the only way
you can have a sustainable organization, because making money across borders is not easy.”
(ScanBalt interview)

On the decentralized level, different actors act as brokers at different points in time. One such

actor is the biotechnology park at the University of Tartu in Estonia, which has been

conducting the Estonian Genome Project since 2001 and has taken a leading role within

ScanBalt. Generally, for each project idea a lead organization needs to be identified that is

also willing to invest resources up-front and to oversee the project application and

coordination.

In a similar way to ACOD, whose “state initiatives have emerged for different reasons

and act under different conditions” (ACOD interview), ScanBalt stresses that the macro-

regional cooperation should not try to remove competition between regions because each has

14
specific technological or research-based competences. The idea is rather to connect central

regions, such as the Medicon Valley in the medical technology sector, and smaller, more

peripheral regions in an “added-value macro-regional chain” (ScanBalt interview):

”(…) any region has some kind of competence to drive collaboration. This is very important
because otherwise it would just be a matter of the capital regions taking their lead.” (ScanBalt
interview)

Accordingly, ScanBalt rarely holds its general assemblies in the capital cities, but in smaller

cities and peripheral regions.

ScanBalt, like ACOD, has a highly formalized governance structure. It has taken

several years of negotiation to agree on a set of formal statutes, but now these statutes are not

questioned:

“It was not easy to establish the statutes. It took years of discussion about what should be the
role of ScanBalt itself, how much direct power should it have, how much will the regions
have, who are the most important universities, or companies, or clusters, etc. (…) it was very
good that we spent all this time because today there are no serious discussions about it.”
(ScanBalt interview)

Over time, ScanBalt has developed a differentiated membership structure and now includes

individual members as well as regions. Different rights and fees are attached to each

membership status. It has adopted this structure so that ScanBalt can avoid “stealing”

members from its member clusters, where most organizations already pay membership fees:

“The backbone of the ScanBalt BioRegion is the regional cluster organizations. They are
themselves member-based associations and most do not have strong financing. ScanBalt
should never ever compete against its own members. So therefore we say, those who are
members of regional clusters are per definition also members of ScanBalt. We should not
compete against a regional cluster.” (ScanBalt interview)

It is hereby necessary to distinguish administrative funding for organizing cluster networks

from operational level funding, which is tied to projects and comes, for example, from

specific EU or national programs or from firms interested in their outcomes.

The Tri-Cluster Alliance and Germany goes 3D have informal governance structures.

In 3D, the only formal document is a one-page joint vision paper, but the collaboration has

formal selection criteria due to several membership requests. These criteria – each

15
participating region must have an internationally renowned cluster in the music industry and

host an internationally relevant industry event – are communicated through press interviews

or interpersonally. 3D has considered becoming an independent legal entity, but has decided

to delay this step until it becomes necessary, for example, because of joint funding

applications. The participants fear that formal guidelines and requirements would put their

personal relationships at risk:

“It is still guerilla-like. (…) Everything works on a voluntary basis. (…) Formal statutes could
put our good personal relationships at risk, because once we have formal statutes, we would
have to be stricter with each other.” (3D interview)

The Tri-Cluster Alliance is also based on verbal agreements between representatives of the

Berlin, Tucson and Ottawa photonics clusters. There are no formal structures or dedicated

resources for coordination and control. Relations within this collaboration are rather loose

and sporadic:

“It is more an agreement by handshake. We have no contract. We (representatives of the

regional network of optical technologies and the WISTA technology park in Berlin) sent an
official letter of intention to Tucson, and that was it.” (Tri-Cluster interview)

Germany goes 3D is currently paid for directly by its three cluster representative

organizations, as its current goals do not require intense project funding. The Tri-Cluster

Alliance also lacks dedicated inter-cluster financing, with activities being financed out of the

internationalization budgets of the clusters’ administrative offices or the technology parks.

Not even the name Tri-Cluster Alliance has been established officially, since our

interview partners also sometimes speak of the Tri-City Breakfast or the Tri-Region Alliance.

It is often equated with the Global Advantage Network, an international networking program

of the University of Arizona that was meant to grow out of the three clusters’ collaboration

and reach out to additional partners, but for which funding had been cut.

Both cases have not specified a central lead organization. While the Cologne cluster

views itself as the main initiator of 3D, all interviewees stressed how informal and friendly

relations as well as the current scope of activities have made it unnecessary to specify a

16
formal lead organization and governance is currently shared. In the Tri-Cluster Alliance,

informal leadership moved from the regional development office in Tucson to the

international office of WISTA management GmbH, manager of the Berlin Adlershof science

and technology park, when budgets were cut in the U.S. The Tri-Cluster Alliance thus has

changing and informal internal lead organizations.

4.2.2 Contingency Factors

In each case there are both shared and partly diverging goals. Germany goes 3D and the Tri-

Cluster Alliance aim to gain visibility and increase political leverage, but their cluster

representatives stress that they are not directly mandated by their member organizations but

rather pursue these goals in their role as cluster managers. Despite being in very different

industries, the main activities of these two cluster networks revolve around organizing joint

trade fair representations and pooling resources for communicating the special competences

of the participating clusters in their respective industry context to policy makers and potential

customers. As a Tri-Cluster representative stated with reference to the often overlooked

strategic importance of the photonics industry for other industries:

“There are too few of us and (…) we don't really compete against each other, we complement
each other. (…) The value of what we invent and create needs to be propagated better.” (Tri-
Cluster interview)

Similarly, Germany goes 3D focuses on developing a common voice and sharing financial

burdens:

“We are competitors, but we also maintain an open dialogue with each other. Our primary
target is to develop greater bargaining power towards the outside, to become foreign ministers
for music, so to speak.” (Germany goes 3D interview)

Trust is based on the intensity and duration of collaboration. The youngest cluster network is

Germany goes 3D, which started only in 2010. Nonetheless, the collaborating partners refer

to the high level of trust on the inter-cluster level, based on having known each other

personally for several years from meeting at different music events. This perception of trust

allows for shared participant governance, but is highly person-dependent. Germany goes 3D

17
is furthermore embedded in numerous inter-organizational and inter-personal relations

between the three music regions that are not directly brokered by 3D.

The Tri-Cluster Alliance has existed since 2004, but due to the geographic distance

between the clusters there are only sporadic inter-cluster contacts, so that trust needs to be

built and institutionalized. The network activity so far has consisted of mutual visits of cluster

delegations and the regular cluster breakfast at the annual Photonics West industry exhibition

and fair. These activities have increased the knowledge about the connected clusters’

members and competences and have contributed to establishing loose ties, but a more central

brokerage by leading business organizations and research institutes is considered necessary to

extend the collaboration towards joint investments and projects:

“The cluster breakfast is a tradition by now. In the beginning it was a small circle. There you
had the opportunity to really meet and chat and to become visible. It grew larger and larger.
Now that we have seen the profiles, we need to reorganize to provide more focused content.
Noncommittal meetings do not work anymore.” (Tri-Cluster interview)

In ACOD and ScanBalt, trust among inter-cluster actors but also increasingly among the

organizations residing in the different clusters has developed over several years of intense

collaboration. While ACOD is very active in establishing competence networks that comprise

members from different state initiatives and work on themes such as aluminum or energy-

efficient production, collaboration targets at ScanBalt are auctioned via a number of EU-

funded projects, indicating that ScanBalt has developed skills and routines in raising public

funds for specific projects:

“One thing that has been very important over the years is that we are now a group of people
who really know each other very well and have been with each other in various projects, and
who can distribute their projects to the right partners. So we have established a kind of trust
over these years and I think this is really a very valuable capital, which is extremely
important.” (ScanBalt interview)

In terms of size, ScanBalt is the largest collaboration in our sample with 13 connected

clusters and over 50 formal members, followed by ACOD with 5 participating state initiatives

and about 370 organizations. Since these two initiatives decidedly reach out to project-based

18
collaboration among the clusters’ member organizations, intense brokerage is needed both on

an inter-cluster level and below.

Regarding the nature of the task, Germany goes 3D has not tried so far to extend its

current activities towards, for example, a joint funding application. The Tri-Cluster Alliance

has attempted to broker exchanges among the different organizations within the connected

clusters as well. Aiming to attract international subsidiaries from partner clusters, for

instance, an exclusive business welcome package provided by the local business and

technology parks for photonics companies from other clusters has been developed. Twice, a

rotating international photonics summer school has been held in Berlin. It is currently unclear

whether one of the Tri-Cluster partners will take a lead in pursuing these activities further, as

goals and needs diverge to some extent. Attracting international subsidiaries, for instance, has

mainly been a goal for the Berlin cluster historically lacking large photonics firms, but this

goal has not been reached. The main aim for Tuscon was to gain knowledge about and

establish contacts to suppliers, so that special orders could be handled quickly and this goal

has been reached by the efforts so far. While Germany goes 3D continues its joint meetings

and public relations, representatives of the Tri-Cluster Alliance thus describe the cooperation

as increasingly dormant:

“The thread may be thin but the communication line is still there.” (Tri-Cluster interview)

Goals also diverge within ScanBalt, limiting the extent of activities pursued on the inter-

cluster level. For instance, ScanBalt tried to set up a Nordic Baltic Expat Platform on which

members would share information about expatriates so that they could be attracted back to

the ScanBalt regions. This project, like some other pilot projects, had to be stopped due to the

members’ unwillingness to share crucial information. Nonetheless, in both ACOD and

ScanBalt goals are aligned by a joint interest in gaining national or EU funds for projects –

not a debated option for the intercontinental Tri-Cluster Alliance. The need for cluster and

19
inter-cluster level competences is highest in ACOD and ScanBalt, because both initiatives

require significant degrees of interdependence among partners on the inter-cluster level and

among the organizations residing with the clusters in order to steer and conduct joint projects

and coordinate activities on and across different levels in line with their goals.

5. Discussion

We identify two forms of cluster networks based on the differential emphases put on goals,

governance approaches, and contingencies (Figure 1).

--- Insert Figure 1 about here ---

Germany goes 3D and the Tri-Cluster Alliance are examples of internally governed cluster

networks. Cluster managers from CAOs or hub firms acting as cluster leaders deliberately

establish relations with the partnering clusters’ representatives. In both cases, the connected

clusters aim to increase their visibility towards policy makers and to distribute and find

information or potential partners in the other clusters. Collaborative ties between

organizations residing in the different clusters may exist or emerge, but are not intensely

brokered from the inter-cluster level, although jointly organized institutionalized events

provide opportunities for actors on lower levels to detect new business opportunities and

extend, for example, their knowledge about technologies or funding opportunities.

Governance is informal in both cases, although both refer to an official letter of intent that

indicates the existence of some formal rules.

ScanBalt and ACOD are examples of externally governed cluster networks,

characterized by interactions both on inter-cluster and inter-organizational levels, while

involving vertical linkages across these two levels. Ties in and between the connected

clusters are brokered by institutionalized, cross-cluster governance structures on different

levels in the form of an inter-cluster lead organization external to the clusters and several,

20
sometimes changing, decentralized internal lead organizations. Whereas the central body is

responsible for administering funds, creating, for instance, external visibility, and brokering

lower-level ties to some extent, decentralized lead organizations are responsible for exploring

project ideas, developing funding proposals, and implementing permanent or temporary

collaborative cross-cluster structures.

Our findings show that internally governed cluster networks result when the cross-

cluster activities remain sporadic and do not require intense ties on lower levels. Although the

Tri-Cluster Alliance and Germany goes 3D differ substantially regarding their regional scope

and industry, they are comparable in their goals and the nature of the tasks pursued, such as

the sharing of limited resources and gaining political visibility, and in the resulting low need

for cluster and inter-cluster level competences. Leadership in these constellations can be

shared or centralized, depending on the intensity of interaction between participants on and

across levels, and on goal consensus. Internally governed cluster networks differ from

organization-based linkages across clusters in the form of global pipelines (Lorenzen and

Mudambi, 2012) in that lead organizations need not be flagship firms or MNEs, but can

embrace small and medium enterprises or CAOs acting on behalf of clusters precisely

because global pipelines are absent. The Tri-Cluster case also shows that cluster networks can

potentially be terminated, although cluster connectivity may continue in the form of inter-

personal or inter-organizational ties.

Externally governed cluster networks are established when the goal of the

collaboration is to create intense “interlocking” among clusters in a wider geographical

context (Asheim and Coenen, 2006). Their governance structures are implemented to develop

co-specialized supply chains and increase the innovative capacity of often small and medium-

sized suppliers (Smith et al., 2002) within a bounded environment. In our sample, externally

governed cluster networks were established in neighboring regions, although these were not

21
necessarily nationally bound. Externally governed cluster networks can thus be studied as a

form of territorial innovation system (Moulaert and Sekia, 2003; Simmie, 2005) and enrich

this stream of research, which is sometimes critiqued for neglecting actors and institutions

outside of national boundaries (Cooke, 2005). They can also be understood as an intermediate

structure between localized industrial clusters and global production networks (Rutherford

and Holmes, 2008).

Our study applies the network governance framework (Provan and Kenis, 2008) to the

study of clusters and inter-cluster linkages, but adds levels of analysis on which linkages may

be formed. Lead organizations on the cluster level are either CAOs or hub firms acting on

behalf of clusters. They may share governance on the inter-cluster level, or one cluster’s

representative organization may act as an internal lead organization. Centralized external

governance is provided by inter-CAOs and often complemented by several decentralized

brokers. Whether governance is shared depends in part on the nature of the clusters

themselves. The music clusters, for instance, are all characterized by rather informal network

structures, which characterize the cluster network as well (cf. Lorenzen and Mudambi, 2012).

Furthermore, our findings indicate that strong and more powerful clusters tend to play a more

central role in cluster networks. When budgets were cut in the Arizona photonics cluster, for

instance, the Berlin cluster took on the leadership role in the Tri-Cluster Alliance.

Governance tends to be less formalized when fostering regional development and raising

public funds for the cluster network and for specific projects are of minor importance.

The findings reveal that different combinations of contingencies can lead to the same

governance form. The comparison of ACOD and ScanBalt shows that both cases of

externally governed cluster networks rely on similar approaches to leadership, brokerage, and

formalization, and the extent of trust. They both coordinate relatively high numbers of

participants. However, the cases differ regarding their regional scopes, goals, and cluster- and

22
inter-cluster level competencies. They are engaged in different industries and pursue different

tasks. The comparison of Germany goes 3D and the Tri-Cluster Alliance shows that these

cases of internally governed cluster networks are also similar concerning leadership,

brokerage, formalization, and the number of participants. They further resemble each other

with respect to the tasks that they pursue and the intensity of collaboration. However, they

differ referring to their goals, regional scopes, and industries. Thus, the decisive factors for

the choice of a particular governance approach are similarities in leadership, brokerage, and

formalization, as well as trust and the number of participants.

The network governance lens illustrates that structures for brokerage and leadership

exist on the cluster and inter-cluster levels to ensure that resource exchanges are goal-

oriented and help all participants to reap joint and individual benefits (Provan and Kenis,

2008). Our findings add to our limited knowledge regarding the effectiveness of cluster

networks (Mudambi and Swift, 2012). ScanBalt has been successful in gaining visibility and

fostering knowledge transfer between the connected clusters and their member organizations.

The amount of project-related funding and the increasingly tight connections to Brussels

corroborate this point. Single actors (e.g. research institutes) benefit from the collaboration by

getting access to knowledge held in other clusters, which they could not have obtained

otherwise. ACOD increases its members’ technological added value by jointly developing

knowledge in special areas of expertise and opening up business opportunities that they

would not get otherwise. Germany goes 3D has raised public interest in politically motivated

endeavors to restructure the German music industry, fostering the external legitimacy of the

existence of several decentralized clusters instead of a spatially concentrated allocation of

organizations in the German capital. The Tri-Cluster Alliance has established knowledge

about the connected clusters’ competences but has failed to expand its activities. Based on

our data, we cannot measure the effect of cluster networks on organization-level

23
performance. Due to the existence of cross-cluster ties in many different industries, their

multi-level character and complexity, and the lack of established measures (Mudambi and

Swift, 2012; Provan and Kenis, 2008), a promising area for future research would be the

development of new, possibly industry-specific and multi-dimensional measures of the

outcomes of cross-cluster linkages.

A limitation of our study is its cross-sectional character. Changing contingency

factors may lead from loosely coupled clusters to institutionalized, integrated cluster

networks or – as indicated by the Tri-Cluster case – vice versa, while participants’ increasing

collaboration experience may foster a dynamic evolution in their governance on various

levels (Powell et al., 2005; Ter Wal and Boschma, 2011). Further research could include, for

example, identity (Romanelli and Khessina, 2005) or power relations (Christopherson and

Clark, 2007) as additional contingency factors. We do not claim that our forms of cluster

networks are exhaustive, but a governance lens provides an important addition to our current

knowledge on cluster connectivity, which should stimulate further research on this

phenomenon.

24
References

Asheim, B. T. and Coenen, L. (2005) Knowledge bases and regional innovation systems:
Comparing Nordic clusters, Research Policy, 34, pp. 1173-1190.
Asheim, B. T. and Coenen, L. (2006) Contextualising regional innovation systems in a
globalising learning economy: On knowledge bases and institutional frameworks,
Journal of Technology Transfer, 31, pp. 163-173.
Asheim, B.T. and Isaksen, A. (2002) Regional innovation systems: the integration of local
‘sticky’ and global ‘ubiquitous’ knowledge, Journal of Technology Transfer, 27, pp.
77–86.
Bathelt, H., Malmberg, A. and Maskell, P. (2004) Clusters and knowledge: local buzz, global
pipelines and the process of knowledge creation, Progress in Human Geography,
28(1), pp. 31-56.
Boschma, R. (2005) Proximity and innovation: a critical assessment. Regional Studies, 39, pp.
61-74
Bresnahan, T., Gambardella, A. and Saxenian, A. (2001) 'Old economy' inputs for 'new
economy' outcomes: cluster formation in the new Silicon Valleys, Industrial and
Corporate Change, 10(4), pp. 835-860.
Bunker Whittington, K., Owen-Smith, J. and Powell, W. W. (2009) Networks, propinquity,
and innovation in knowledge-intensive industries, Administrative Science Quarterly,
54, pp. 90-122.
Burt, R. S. (1992) Structural Holes: The Social Structure of Competition (Cambridge, MA,
London: Harvard University Press).
Cheung, M.-S., Myers, M. B. and Mentzer, J. T. (2011) The value of relational learning in
global buyer-supplier exchanges: A dyadic perspective and test of the pie-sharing
premise, Strategic Management Journal, 32, pp. 1061-1082.
Cooke, P. (2005) Regionally asymmetric knowledge capabilities and open innovation:
Exploring 'Globalisation 2' - a new model of industry organization, Research Policy,
34(8), pp. 1128-1149.
Christopherson, S. and Clark, J. (2007). Power in firm networks: what it means for dynamic
regional innovation systems. Regional Studies, 41, 1223-1236.
Dohse, D. (2000) Technology policy and the regions - the case of the BioRegio contest,
Research Policy, 29 (9), pp. 1111-1133.
Dohse, D. (2007) Cluster-based technology policy - the German experience, Industry and
Innovation, 14(1), pp. 69-94.
Engel, J. S. and del-Palacio, I. (2011) Global clusters of innovation: The case of Israel and
Silicon Valley, California Management Review, 53(2), pp. 27-49.
Giblin, M. (2011) Managing the global-local dimensions of clusters and the role of “lead”
organizations: The contrasting cases of the software and medical technology clusters
in the west of Ireland, European Planning Studies, 19(1), pp. 23-42.
Gilding, M. (2008) ‘The tyranny of distance’: Biotechnology networks and clusters in the
antipodes, Research Policy, 37, pp. 1132-1144.
Granovetter, M. (1973) The strength of weak ties, American Journal of Sociology, 78, pp.

25
 1360-1380.
Gupta, A. K. and Govindarajan, V. (2000) Managing global expansion: A conceptual
framework, Business Horizons, 43(2), pp. 45-54.
He, J. and Fallah, H. (2011) The typology of technology clusters and its evolution – evidence
from the hi-tech industries, Technological Forecasting and Social Change, 78 (6), pp.
945-952.
Henn, S. (2012) Transnational entrepreneurs, global pipelines and shifting production
patterns. The example of the Palanpuris in the diamond sector, Geoforum, 43, pp. 497-
506.
Herrmann, A. M., Taks, J. L. and Moors, E. (2012) Beyond regional clusters: On the
importance of geographical proximity for R&D collaborations in a global economy –
the case of the Flemish Biotech sector, Industry and Innovation, 19(6), pp. 499-516.
Hoover, E. (1937) Location Theory and the Shoe and Leather Industries (Cambridge, MA:
Harvard University Press).
Human, S. E. and Provan, K. G. (2000) Legitimacy building in the evolution of small-firm
multilateral networks: A comparative study of success and demise, Administrative
Science Quarterly, 45(2), pp. 327-365.
Humphrey, J. and Schmitz, H. (2002) How does insertion in global value chains affect
upgrading in industrial clusters? Regional Studies, 36, pp. 1017–27.
Huxham, C. and Vangen, S. (2000) Leadership in the shaping and implementation of
collaboration agendas: How things happen in a (not quite) joined up world, Academy
of Management Journal, 43, pp. 1159-1175.
Huxham, C. and Vangen, S. (2005) Managing to Collaborate (London: Sage).
Jarillo, J. C. (1988) On strategic networks, Strategic Management Journal, 9(1), pp. 31-41.
Kenis, P. and Provan, K. G. (2006) The control of public networks, International Public
Management Journal, 9(3), pp. 227-247.
Kilduff, M. and Tsai, W. (2003) Social networks and organization (London: Sage).
Kuhlmann, S. and Edler, J. (2003) Scenarios of technology and innovation policies in Europe:
Investigating future governance, Technological Forecasting and Social Change, 70,
pp. 619-637.
Lorenzen, M. and Mudambi, R. (2012) Clusters, connectivity and catch-up: Bollywood and
Bangalore in the global economy, Journal of Economic Geography, DOI:
10.1093/jeg/lbs017.
Marshall, A. (1890) Principles of Economics: An Introductory Volume (London: Macmillan).
Martin, R. and Sunley, P. (2003) Deconstructing clusters: chaotic concept or policy panacea?
Economic Geography, 3, pp. 5-35.
Maskell, P., Bathelt, H. and Malmberg, A. (2006) Building global knowledge pipelines: The
role of temporary clusters, European Planning Studies, 14(8), pp. 997-1013.
Menzel, M.-P. and Fornahl, D. (2010) Cluster life cycles – dimensions and rationales of
cluster evolution, Industrial and Corporate Change, 19(1), pp. 205-238.
Moulaert, F. and Sekia, F. (2003) Territorial innovation systems: a critical survey, Regional
Studies, 37(3), pp. 289-302.

26
Mudambi, R. and Swift, T. (2012) Multinational enterprises and the geographical clustering
of innovation, Industry and Innovation, 19(1), pp. 1-21.
Pavitt, K. (1984) Sectoral patterns of technical change: Towards a taxonomy and a theory,
Research Policy, 13, pp. 343-373.
Porter, M. E. (1998) On Competition (Boston, MA: Harvard Business School Publishing).
Powell, W. W., White, D. R., Koput, K. W. and Owen-Smith, J. (2005) Network dynamics
and field evolution: The growth of interorganizational collaboration in the life
sciences, American Journal of Sociology, 110(4), pp. 1132-1205.
Provan, K. G. and Kenis, P. (2008) Modes of network governance: Structure, management,
and effectiveness, Journal of Public Administration Research and Theory, 18, pp. 229-
252.
Provan, K. G., Fish, A., and Sydow, J. (2007) Interorganizational networks at the network
level: A review of the empirical literature on whole networks, Journal of Management,
33, pp. 479-516.
Romanelli, E. and Khessina, O. M. (2005) Regional industrial identity: Cluster configurations
and economic development, Organization Science, 16(4), pp. 344-358.
Rutherford, T. D. and Holmes, J. (2008). ‘The flea on the tail of the dog’: power in global
production networks and the restructuring of Canadian automotive clusters. Journal of
Economic Geography, 8, 519-544.
Rychen, F. and Zimmermann, J.-B. (2008) Clusters in the Global Knowledge-Based
Economy: Knowledge Gatekeepers and Temporary Proximity, Regional Studies,
42(6), pp. 767-776.
Saxenian, A. (2002) Transnational communities and the evolution of global production
networks: The cases of Taiwan, China and India, Industry and Innovation, 9(3), pp.
183-202.
Saxenian, A. (2006) The New Argonauts. Regional Advantage in a Global Economy
(Cambridge: Harvard University Press).
Saxenian, A. and Hsu, J.-Y. (2001) The Silicon Valley-Hsinchu connection: Technical
communities and industrial upgrading, Industrial and Corporate Change, 10 (4), pp.
893-920.
Scott, A. J. (2002) A new map of Hollywood: The production and distribution of American
motion pictures, Regional Studies, 36, pp. 957–975.
Simmie, J. (2005) Innovation and space: a critical review of the literature, Regional Studies,
39(6), pp. 789-804.
Smith, A., Rainnie, A., Dunford, M., Hardy, J., Hudson, R., and Sadler, D. (2002). Networks
of value, commodities and regions: reworking divisions of labour in macro-regional
economies, Progress in Human Geography, 26, 41-63.
Staber, U. (1997) Specialization in a declining industrial district: The case of Baden-
Württemberg textiles and clothing, Growth and Change, 28(49), pp. 475-495.
Sydow, J., Lerch, F., Huxham, C. and Hibbert, P. (2011) A silent cry for leadership:
Organizing for leading (in) clusters, Leadership Quarterly, 22(2), pp. 328-343.
Ter Wal, A. L. J. and Boschma, R. (2011) Co-evolution of firms, industries and networks in
space, Regional Studies, 45 (7), pp. 919-933.

27
Vapola, T. J. (2011) The laws of attraction: What attracts innovative start-up firms to
partnerships with global MNCs? Journal of International Entrepreneurship, 9, pp. 39-
61.
Visser, E.-J. (2009) The complementary dynamic effects of clusters and networks, Industry
and Innovation, 16(2), pp. 167-195.
Visser, E.-J. and Atzema, O. (2008) With or without clusters: Facilitating innovation through
a differentiated and combined network approach, European Planning Studies, 16(9),
pp. 1169-1188.
Visser, E.-J. and Boschma, R. (2004) Learning in districts: Novelty and lock-in in a regional
context, European Planning Studies, 12(6), pp. 793-808.
Yin, R. K. (2009) Case Study Research: Design and Methods (Thousand Oaks: Sage).

28
"This is an Accepted Manuscript of an article published by Taylor & Francis in Industry and Innovation
on 26 Jun 2013, available at http://www.tandfonline.com/10.1080/13662716.2013.805929"