Broadening Our Knowledge on Cluster Evolution

1. Introduction

The study of regional clusters is a core research topic in economic geography (Markusen, 1996; Martin & Sunley, 2003; Maskell, 2001). Starting from research on industrial districts (Becattini, 2002), the focus has been on the factors that make clusters distinct from spatially dispersed economic activities (Porter, 1998). These studies described the prerequisites of clusters, how particular clusters function at a particular point in time, and how the institutional setting of one place leads to differences in the functioning of clusters at another place (Saxenian, 1994).

Although clusters are regarded as important elements in economic development (Asheim et al., 2006; Martin & Sunley, 2003; Porter, 1998), the strong focus on the way clusters function is contrasted with a disregard for their evolutionary development: how clusters actually become clusters, how and why they decline, and how they shift into new fields and transform over time (Lorenzen, 2005). As a reaction to this gap, and inspired by recent developments in evolutionary economic geography, new life cycle approaches emerged (Staber & Sautter, 2011; Suire & Vicente, 2009; Ter Wal & Boschma, 2011). This strand connects the quantitative development of clusters with underlying qualitative changes and transformations (Bergman, 2008; Menzel & Fornahl, 2010). In doing so, this strand elaborates which dynamics, prerequisites and qualities are connected to the emergence, growth and decline of clusters and in doing so, finds out the patterns of cluster evolution.

The literature mostly differs between the emergence, growth or expansion, decline or transformation of the cluster (Bergman, 2008; Enright, 2003). It has elaborated three different dynamics that drive the cluster through different states of their developments: actors, networks and institutions (Maskell & Malmberg, 2007; Menzel & Fornahl, 2010; Ter Wal & Boschma, 2011). Yet, empirical evidence on these factors from a life cycle perspective is still scarce, and the theoretical contributions describe quite general dynamics, from which single clusters, due to their specific context, most probably deviate.

Furthermore, this strand of literature has a distinct policy perspective. Understanding the dynamics that are connected to the emergence, growth, decline and transformation of clusters allows one to derive targeted policies to govern clusters during different phases of their development (Avnimelech & Teubal, 2008; Brenner & Schlump, 2011). Therefore, this perspective does not solely analyse the path dependencies and contingencies of cluster evolution, but instead the leverages that can be used to contribute to the emergence and growth of clusters as well as to prevent their decline.

This special issue aims to broaden our knowledge on actors, networks and institutions during the life cycle and evolution of clusters in two ways. First, by empirically focusing on hitherto neglected particularities of cluster evolution as well as more nuanced views on actors, networks and institutions in cluster evolution; second, conceptually, by broadening the focus of the prior approaches by integrating different and previously neglected dynamics into this strand of literature. In the following, we first describe new approaches on evolution and life cycles of clusters and how they differ from previous approaches in Section 2. This section also elaborates on actors, networks and institutions as drivers for cluster evolution and through its life cycle. Section 3 describes the contributions of this special issue. We finish this editorial with some thoughts on further research in Section 4.

2. New Approaches on Cyclical Developments

There have always been cyclical approaches to explain the growth and decline of regional industries, for example as a result of the development of new, and the aging of existent, technologies or the change from product to process innovation (Duranton & Puga, 2001; Vernon, 1966). However, these approaches were criticized for being essentialistic, as they consider regional development as a deterministic outcome of often quite general processes (Storper, 1985) and could not explain the different developments of clusters under the same conditions (Saxenian, 1994) or why some industries show another pattern than predicted by the life cycle (Piore & Sabel, 1984).

As a reaction to these critics, and inspired by recent developments in evolutionary economics in general and the evolutionary turn in economic geography in particular (Boschma & Frenken, 2006), new life cycle approaches emerged that focus on micro dynamics instead of the structural dynamics of the old approaches (Menzel & Fornahl, 2010). These new approaches point to three main elements, namely actors, networks and institutions, which differ strongly between different stages of the life cycle and therefore affect the transition from one stage of the life cycle to another. We further explore these three main elements below, but also want to point out here that they are interrelated to each other. Cluster dynamics are in fact derived from the interdependencies of these different elements, something that has been stressed by advocates of the paradigm of relational economic geography (Bathelt & Glückler, 2003; Boggs & Rantisi, 2003). Cluster dynamics are also based on context dependent actions (see also Granovetter, 1985), whereby the context is usually described by institutions or networks. Also Maskell and Malmberg (2007) argue that perceptions of actors that co-evolve with their institutional setting lead to different clustering phases. Ter Wal and Boschma (2011) describe how clusters co-evolve with the respective networks. Menzel and Fornahl (2010) describe knowledge heterogeneity and the way actors utilize it through the life cycle of a cluster.

These new approaches resemble each other in their perspective on clusters not being homogeneous entities. Instead, they derive the dynamics behind cluster evolution from the interplay between heterogeneous agents that change during the cluster life cycle. Due to their focus on micro dynamics, the new approaches overcome the more structuralistic considerations of the older approaches and move from the life cycle analogy towards an ontology of cyclical spatial dynamics (for another view see Martin & Sunley, 2011). In the following we review some literature on the three elements and their interdependencies.

2.1. Actors

Single actors, i.e. individuals, firms and public organizations, are emphasized in the cluster life cycle literature because of their role in generating novelty and variety. Research shows the importance of firm-level learning for cluster evolution. Studies such as Klepper's (2007) analysis of the Detroit automobile cluster or Feldman et al.’s (2005) description of the Washington biotherapeutics cluster exhibit that firms and their strategies and capabilities are strongly intertwined with the evolution of the cluster. The emergence of technology clusters such as Sophia Antipolis in France (Longhi, 1999) and the Research Triangle Park in North Carolina in the USA (Link & Scott, 2003) also show that political actors can affect the emergence of clusters. Yet, studies such as Grabher's (1993) account on the steel and coal complex in the Ruhr area in Germany also make clear that policy-makers unintentionally can be directly involved in the decline of a cluster.

More recent literature on actors analyses the heterogeneity and capabilities of firms and organizations as well as their learning processes during cluster evolution. Mossig and Schieber (2014), for instance, compare the evolution of a declining and a growing packaging machinery cluster in Germany. They found out that their different development depends on the heterogeneity of firms and determined how firms exploit this heterogeneity. Compared to the declining cluster, the growing cluster has a larger heterogeneity of firms and an institutional environment supporting the exploitation of this heterogeneity. Hervas-Oliver and Albors-Garrigos (2014) demonstrate that during phases of cluster transformation, especially incumbent technological gatekeepers face difficulties dealing with disruptive innovations and that mostly newer firms are able to cope with these changes. But during phases of transformation, large incumbents connect to new firms to secure access to new technologies. Potter and Watts (2014) show that in the declining Sheffield metal cluster Marshallian externalities still prevail, yet are especially important for firms that combine ferrous material knowhow with knowledge form technologically related sectors. In doing so, they point to the importance of Jacobs externalities during phases of restructuring. Elola et al. (2012) explain with the example of four clusters in the Basque country in Spain that the initial factor and demand conditions that led to their emergence were no  longer important in the later stages. Instead, firms had to build strategic capabilities. Additionally, cluster transformation relied on firm-specific learning processes. Particularly less specialized, more flexible and outward-oriented firms survived phases of cluster transformation. Together, these studies show that heterogeneity and possibilities and capabilities of exploiting heterogeneity are crucial for cluster growth and renewal.

2.2. Networks

Studies such as those of Owen-Smith and Powell (2004) on the Boston biotech cluster identify that networks evolve with the evolution of clusters. While firms were connected to public research institutions in the early phase of cluster evolution, they built more firm-centred networks in later phases. On a more abstract level, tight networks are, for example, often connected with the decline of clusters while emerging clusters are shaped by unstable networks (Grabher, 1993; Ter Wal & Boschma, 2011).

Shin and Hassink (2011) analyse the life cycle of the South Korean shipbuilding cluster. They found out that the relations of the growing Korean shipbuilders with the maturity of the cluster became more outward orientated, particularly towards more traditional shipbuilding clusters such as Norway. Giuliani (2013) shows with the evolution of networks in the Chilean wine industry that networks in developing countries seem to have a different evolutionary pattern. The strong differences in firm capabilities result in quite hierarchical and stable network structures during the evolution of the cluster. Li et al.’s (2012) study on the evolution of the aluminium extrusion industry in Dali (China) concludes that the cluster transformed due to political-economic changes and that a new generation of entrepreneurs entered the cluster, thereby changing the network structure from a tight and homogeneous into a more dispersed and formalized structure. These studies found out different forms of contingencies that affect the connection between different phases of cluster evolution and structure of relations. This relation is therefore quite complex. While there seem to be rough generalities in this relation (Ter Wal & Boschma, 2011), context-specific dynamics result in idiosyncratic outcomes.

2.3. Institutions

A cluster's institutional setting consists of, for instance, its supportive environments, regional cultures and cognitive frames. Many studies already illustrated how a regional industry culture affects regional development (Piore & Sabel, 1984). Saxenian (1994) and Feldman et al. (2005) describe how institutions evolve with the development of clusters. Other studies, such as Avnimelech and Teubal (2006), show how institutions, especially state-level institutions, precede the evolution of cluster.

Recent research also stresses the co-evolution between actors and their institutional environment. Staber and Sautter (2011) describe how a local identity evolved with the evolution of the cluster, and different constructions of cluster identity became particularly apparent during cluster upheaval and transformation. In the case study by Tomlinson and Branston (2014) purposive adaption and upgrading resulted in a renewal of the declining ceramics cluster in North Staffordshire in the UK. This renewal also led to the transformation of the institutional environment towards new collective supportive organizations, new training and skill courses and collective branding procedures. Skalholt and Thune (2014) compare the reaction of emerging and mature clusters in Norway to the crisis in 2009 and 2010. They discover that especially mature clusters applied strategies of increasing collaboration and competence building and that they have developed more capabilities for collective action than newer clusters. These studies agree with accounts on the co-evolutionary dynamics between cluster firms and their institutional environment (Maskell & Malmberg, 2007). They also point out that their interrelation itself becomes disrupted during phases of cluster transformation, leading to tensions between the established institutional environment and uncertainty about a possible future institutional environment.

All these studies focusing on actors, networks and institutions and their interrelations show that (with the notable exception of cluster decline) a change in quantitative development of the cluster is connected to a qualitative change, marked by changes in firm capabilities, network structures and the institutional environment. They point, in particular, to the disruptions taking place at different levels during cluster transformation.

3. Contribution of the Special Issue

This special issue contributes to this strand of research by adhering to the common framework consisting of the interdependencies between actors, networks and institutions. This framework is used both to connect to the established literature, and to broaden the perspectives on the evolution and life cycle of clusters over different phases. The papers in the special issue are ordered around the dynamics on which they focus. The first two papers focus on actors, in particular on firms and their relations as drivers of cluster transformation.

The theoretical contribution by Gancarczyk (2015) puts the international strategies of lead companies as drivers of cluster evolution into the focus. She argues that contemporary trends of modularization of products and value chains affect the internal organization of clusters as well as the relation between clusters. Additionally, she argues that these trends particularly affect the relation between clusters of developed and developing countries and a more pronounced hierarchical division of labour between them, resulting in R&D in developed, and manufacturing in developing, countries.

Livi and Jeannerat (2015) focus on the relationship between entrepreneurs and multinational companies. Using the case of Swiss medical technologies (Medtech), their paper highlights how the evolution of local Medtech start-ups is shaped by the corporate venture strategies of multinational companies. This strategy allows start-ups to obtain access to capital early. However, this behaviour of multinationals affects start-up processes in a fundamental way. They describe that entrepreneurs starting a firm are less inclined to develop this firm and to produce goods and services in a long-term perspective. Instead, firms themselves are perceived as products that are formed with the intention to be sold. They argue that this change has important implications for the economic development of regions and the life cycle of clusters. They suggest that these finance-led entrepreneurial dynamics create perpetual sequences of the emergence and re-emergence of regional industries.

Both studies have in common that general trends change the evolution clusters. According to Gancarczyk (2015) it is modularization that affects all clusters and might result in a more pronounced hierarchy between clusters. Livi and Jeannerat (2015) demonstrate that the drivers behind cluster evolution change from innovative (i.e. inventing and producing new products that can be sold) towards financial dynamics (i.e. establishing new firms that get access to venture capital funding and can be sold), which also might affect the variety of a cluster. Both studies show that these broad trends result in new patterns of cluster evolution.

While the previous two papers particularly considered inter-firm relations, the next paper by Sinozic and Tödtling (2015) focuses on the heterogeneity of firms and firm capabilities during cluster evolution, using the New Media cluster in Vienna (Austria) as an example. It consists of 480 firms in advertising, video production, information and communication technologies and publishing. Sinozic and Tödtling (2015) indicate that the technological heterogeneity of firms plays a central role in cluster evolution for the expansion of local capacities and opportunities for change, but that it is insufficient as explanation for cluster evolution. Instead, what is required are local technological capabilities, embodied in firms and people, as well as learning conditions for the exploitation of technologies and opportunities.

The next two papers focus on institutions. Berg (2015) describes the emergence and growth of the film and TV cluster in Seoul, South Korea. Around 700 film-making companies are located in Seoul, half of them in the Gangnam Business District. She makes clear how the cluster co-evolved with changes in the broader institutional framework, particularly economic deregulation. She shows that the cluster evolved due to institutional changes on the state level. De-regulation policies allowed other firms to enter the film sector, which led to the growth of the sector and its concentration in Seoul. Currently, the sector is highly affected by location and relocation of important state agencies as part of decentralization strategies of the state as well as by further forms of de-regulations such as the reduction of the yearly screen quota for domestic films. All in all, the case of the film and TV industry in Seoul indicates that its evolution strongly depends on the setting of respective institutions and regulations as well as its institutional surrounding.

In a similar vein, yet with a completely different industry, Martin and Coenen (2015) describe how the institutional context triggered the emergence of the biogas sector in southern Sweden. The cluster consists of firms for feedstock production, collection and transport, pre-treatment and upgrading of biogas, distribution and retail as well as end-use. It consists of about 40 companies that are directly or indirectly involved in the regional biogas value chain. Before the cluster emerged, the prerequisites to form a biogas cluster existed in the region. But it did not emerge before a public programme supported the use of biogas in public transport from 2002 to 2008. After cluster emergence gained momentum, regional institutions also developed to support the cluster.

The case studies by Berg (2015) and Martin and Coenen (2015) therefore describe that clusters did not emerge randomly, but regional prerequisites in connection to state actions, the establishment of new institutions or changes in established institutions affect cluster emergence and transformation by directing resources into or away from a sector. Compared to previous studies, these insights give a more detailed perspective on the co-evolutionary dynamics between clusters and their institutional environment. Institutional changes might precede cluster formation, and institutional changes are the result of strategic and purposeful actions. These outcomes are important for the further development of a policy perspective on cluster evolution, indicating that clusters do not emerge arbitrarily, but that their emergence and evolution can be governed (see also Avnimelech & Teubal, 2006).

The final paper by Trippl et al. (2015) adopts a broader, more theoretical approach. They argue for more attention towards the context in which a cluster, its firms and relations, evolve. In this vein, they propose three areas of further research: first, to integrate place specificity and the regional institutional environment of the cluster, second, to consider that networks and institutions that affect clusters are on different scales, each having distinct effects on cluster evolution and third, to consider how human agency and strategic decisions, for example policy actions, have an influence on the evolution of clusters.

While all the papers in this special issue revolve around a broad framework consisting of actors, networks and institutions, some case studies also point to different elements. The papers on inter-firm relations point towards the changing general dynamics of economic organization. The papers on co-evolution between clusters and their institutional environment point towards a sequence where institutional change precedes cluster emergence.

Moreover, all the papers underline that cluster evolution not only depends on internal dynamics. Instead, external relations are an integral part of cluster dynamics. Trippl et al. (2015) describe the multi-scalarity of relations and institutions, from local to global scales, as an important research issue in the future. In Livi and Jeannerat (2015) it is the global firm that provides regional firms with capital. In Gancarczyk (2015) global firms’ outsourcing strategies result in a stronger hierarchy of clusters. In Sinozic and Tödtling (2015), local firms use the local environment to exploit opportunities and technologies available outside the cluster. In Martin and Coenen (2015), institutional change on the national level provides the framework under which the biogas cluster could emerge. Also Berg (2015) describes how the emergence of the Seoul film and TV cluster started as a reaction towards the change of regulations at the national level.

4. Roads Ahead

This special issue shows the relevance and appropriateness of adopting a life cycle perspective on cluster evolution. Qualitative changes in institutions, networks and heterogeneity of actors are strongly connected to the quantitative change of clusters. More precisely, this special issue confirms the importance of knowledge heterogeneity for the adaptability of clusters. It also concludes that institutional change can precede cluster formation and that the characteristics of relations of cluster firms with firms outside the cluster are not only necessary for cluster renewal, but also affect the actual pattern of cluster evolution. On the basis of these insights three roads for further research are suggested.

First, policy-driven institutional change, such as regulations, affects cluster emergence and growth. In the examples provided in this special issue, investments were directed into the new sector or the state created a new market for a new industry. These institutional effects coupled with fitting regional prerequisites had a strong influence on cluster emergence. While it was common sense that clusters emerged in a kind of accidental and stochastic process (for an early critique see Martin & Sunley, 2006), recent research and also the results of this special issue show otherwise.

Several scholars had already provided insights into how the cluster life cycle concept can inform policies which are better adapted to the stage of the cluster and how these policies differ from static approaches that ignore the life cycle of clusters (Brenner & Schlump, 2011; Hassink & Shin, 2005; Ingstrup & Damgaard, 2013). Examples from Germany allow a comparison of traditional static policies with a life cycle perspective. National cluster programmes, such as BioRegio, InnoRegio or the Spitzencluster-Wettbewerb (Excellence Cluster Competition) as the most recent one (Dohse, 2007; Eickelpasch & Fritsch, 2005), are designed to increase competitiveness and to reinforce existing strengths. This is particularly evident for the Excellence Cluster Competition, which is dedicated to support the most productive and competitive clusters (BMBF, 2009). This kind of strategy bears two kinds of dangers. First, cluster policies might have no additional, or even a negative effect, and this is not recognized, as the future well-being of the cluster is traced back to actual non-effective cluster policies (Fromhold-Eisebith & Eisebith, 2008). Second, this kind of cluster policy reinforces existent cluster structures and developmental patterns, and thus might impede the adjustment of the cluster to changing environments. In doing so it might support its future decline (Hassink, 2010; Menzel & Fornahl, 2010).

In contrast to static approaches, cluster policies that intend to sustain the long-term viability of clusters would focus on supporting the adaptability and changes of a cluster. This not only requires knowledge about the functioning of clusters, but also about the patterns that are connected with their emergence, growth, decline and transformation. Future research would further disentangle the mechanisms and ways of how state actions affect clusters, from regulatory frameworks via supporting institutions to direct involvement.

The second road for further research is the disentangling of what is happening inside and outside a cluster, i.e. the endogenous dynamics of a cluster and when its change depends on external factors. This is apparent both in the papers focusing on institutions and those analysing relations. In both cases, the benefits of adopting a multi-scalar perspectives as proposed by Trippl et al. (2015) become obvious. There is already a large body of literature on how relations at different geographical scales affect clusters (Bathelt et al., 2004) or how external factors, such as field configuring events, change a field (Lampel & Meyer, 2008). It would make sense to combine the strand on cluster evolution and life cycles stronger with that on different spatial scales of relations. Of relevance would be which kind of relations serve which kind of purpose in which phase of cluster evolution. This would allow devising more tailor-made policy measures.

The third road of research revolves around the effects of socio-economic and technological changes on cluster evolution. There is already literature that describes how disruptions in the technologies applied by cluster firms result in decline and transformation of established clusters (Dalum et al., 2005). Yet, of further interest would be how general technological or socio-economic change affects the general pattern of cluster evolution. Leamer and Storper (2001) already hypothesized in 2001 that the upcoming of the internet would have an effect on the geography of production and the spatial fragmentation of economic activity in space. Research on clusters was revitalized by industrial district research, which resulted from a shift from Fordist to post-Fordist flexible forms of production (Boyer, 1997). Such a shift might again occur. Financialization (French et al., 2011) might be such a development that fundamentally affects evolutionary patterns of clusters, i.e. how firms utilize local diversity.

This special issue contributes to further considering the complexity of the interrelations between actors, networks and institutions during the evolution and life cycle of clusters. Yet, there are still many unresolved questions, particularly about the contextualization of cluster evolution. Further connecting the dynamics of qualitative change to the quantitative development of clusters is still required for meaningful, theory and fact-based policies regarding cluster development in particular and regional development in general.

Acknowledgments

This special issue draws heavily from papers presented at special sessions on cluster life cycles at the Regional Studies Association Annual International Conference in Tampere, 5–8 May 2013. All authors, except for Gancarczyk (2015), are project partners in the research project 10-ECRP-007: Cluster life cycles—the role of actors, networks and institutions in emerging, growing, declining and renewing clusters, which is sponsored by the European Science Foundation in the framework of the program “European Collaborative Research Projects in the Social Sciences” (ESF-ECRP).