The attention that ‘clusters’ have received from policy makers and academics has substantially increased in the last 20 years. Since Porter’s seminal work on The Competitive Advantage of Nations (1990) presented ‘clusters’ as one of the determinants of the international competitiveness of nations and regions, many scholars have adopted and further developed his approach. Porter bases his arguments on what he describes as the globalization paradox, pointing out that despite the logical implications that the globalisation process might have in dismissing the relevance of regional factors, the most competitive firms in world are located in groups geographically concentrated in specific locations. That perspective contributed to attracting attention to the existence of characteristics tied to a local context that could not be accessed by firms positioned elsewhere, and more, to the positive effects that the concentration and the geographic proximity could exert on the firm’s competitiveness.

However influential, Porter’s ideas were not the precursor to discussing the competitive outcomes originating from the geographic concentration of firms (Martin and Sunley, 2003). The roots of cluster theory go back to the industrial districts identified by Marshall (1890), who offered the first detailed description about the economic and social systems created as a result of the spacial concentrations of industrial activities. The Marshallian industrial districts were arrangements of small firms interconnected by commercial operations (buyers and sellers) and other firms engaged in the same or similar activities, that shared productive factors, such as the labour market, infrastructure and tacit knowledge (Becattini, 2004, p. 68). According to Marshal’s descriptions, a group of firms operating in one specific sector within a well-defined, concentrated and relatively small geographic area would experience higher levels of productivity and innovation, indeed the emergence of a fertile environment for technical and organisational developments. Thus the local characteristics would enable the emergence of an ‘industrial atmosphere’ that would increase the firms’ potential to acquire (especially tacit) knowledge, and create positive external economies accessible only to the firms located within the district (Asheim, 2003, p. 416). That perspective tried to evidence that firms geographically concentrated could accesses restricted positive exogenous benefits (exogenous to firms, but endogenous to the district), which would be an alternative to the scale economies achieved by a single (integrated) firm.

Additionally, following some of the seminal ideas proposed by Marshall, it is possible to observe a significant number of economic geographers that also explored regional development using the spatial economic agglomeration to support their ideas. Some examples of concepts emerging from that theoretical trend are ‘regional innovation milieux’ (Crevoisier, 2004), ‘neo-Marshallian nodes’ (Amin and Thrift, 1992) and ‘learning regions’ (Asheim, 1995). More examples can be found in Markusen (1996, p. 297), in which another three different types of industrial districts are described according to the firms configurations, internal versus external orientations and governance structures: a ‘hub-and-spoke industrial district’, which is concentrated around one or more dominant firms; a ‘satellite platform’, formed by a group of unconnected branches embedded in external links; and the ‘state-anchored district’, concentrated on one or more public-sector institutions. Despite the logical and robust assumptions found in many of those concepts, their influence and dissemination were not as successful as the more general cluster framework proposed by Porter. Martin and Sunley (2003) attributes the successful dissemination of the Porter’ concepts to the very general descriptions and delimitations that encompass a wide range of actors and many different structures.

Following much of the concepts proposed by Porter, the description of advantages conferred on clustered firms associated with a general and structured analytical framework stimulated the development and dissemination of academic studies and subsidized the creation of supply-side competitiveness policies directed at structuring and supporting the development of clusters (Pitelis, 2010). That fact resulted in what Martin and Sunley (2003) describe as a ‘policy panacea’ in the use of clusters as a standard (sometimes the unique) target for promoting competitiveness, innovation and economic growth. Moreover, in the last 20 years an increasing number of empirical studies in different countries and sectors have been observed, which aim to identify and discuss the competitive outcomes originating from the concentration of firms and other actors in the same location, for example: Brazil – shoe manufacturing in the Sinos Valley (Schmitz, 2000); Spain – the textile and clothing industries in Catalonia (Porter, 1998); Taiwan – electronic products at the Hsinchu Science Park (Chen, 2008); and the United States – computer and information systems at the Silicon Valley (Saxenian, 1994).

The large significant number of academic studies has resulted in a large number of definitions aiming to describe and establish an accepted cluster ‘template’ (e.g., Enright, 1996; Swann and Prevezer, 1996, Rosenfeld, 1997; Porter, 1998) to support policy makers and academics has led to intense debates and controversial perspectives. Even though the concept of clusters has been increasingly widely disseminated and used by geographers, economists and policy makers, it has suffered from some conceptual confusion. Porter defines a cluster as a ‘geographic concentration of interconnected companies, specialized suppliers, service providers firms in related industries, and associated institutions (for example, universities, standards agencies and trade associations) in particular fields that compete but also co-operate’ (1998, p. 197). However, Martin and Sunley (2003, p. 12) present consistent arguments that indicate the vagueness and superficiality of the concept proposed by Porter. According to their arguments, those characteristics make the concept of cluster means different things to different researchers and policy makers, creating problems for its proper use in the guidance of academics and governments. Those highly controversial aspects of the cluster theory have stimulated the continuous emergence of new concepts and definitions for ‘clusters’.

Proposing a definition aiming to fill some of the gaps and failures found in extant cluster theory, Pitelis (2010, p. 5) defines clusters as ‘geographical agglomerations of firms in particular, related, and/or complementary, activities, with a geographical dimension, that exhibit horizontal and/or vertical intra- and/or inter-sectoral linkages, which operate in the context of a facilitatory socio-institutional setting, and which co-operate and compete (co-opete) in inter-national markets’. That definition tries offer to a more delimited approach that incorporates four major elements: geographical agglomeration, linkages, social-capital [1] and co-opetition (competition and cooperation). The use of those four elements in a single definition offers the possibility to cover the cluster characteristics using delimited criteria to identify and distinguish developed clusters from less complex geographical agglomerations of firms and institutions.

Although it is possible to observe some level of ambiguity encompassing the cluster’s theory, the existence of links interconnecting local actors complemented by geographical dimensions constitute some of the main common points used to guide academics and policy makers with interesting by the competitive outcomes originating from clusters. Those characteristics have frequently been used as the starting point to understand the economic dynamics of clustered firms, putting emphasis on the levels of innovation and productivity emerging from the concentration of different actors in the same area.

Suggesting conditional characteristics to the presence of competitive advantages obtained by firms inside clusters, Ketels (2004) considers that the positive economic effects originated from the geographical concentration will only take place if four critical characteristics are shared among firms and institutions:

Proximity: they must be geographically close to allow the emergence of knowledge spillovers and to share the same common resources;

Linkages: the necessity of similarities in their activities leading to the establishment of connections and synergies;

Interactions: the social interactions developed among firms, clients, suppliers, research institutes, and so on, is what forms the social capital that becomes possible firms to achieve differentiated competitive performances;

Critical mass: it is important to have a significant number of firms and institutions in order to create meaningful impacts on performance of the local actors.

Those characteristics described by Ketels may be used to guide the identification and distinction between ‘developed clusters’ (Pitelis, 2010) from ‘incipient clusters’ (Schmitz, 1999) in order to dismiss some incorrect interpretations associated with the clusters dynamics. Considering that the presence of geographic concentration of firms in the same industry is ‘strikingly common around the world’ (Porter, 1990, p.120), it is necessary the use of specific benchmarks to distinguish and classify different groups of firms geographically concentrated according to their specific characteristics (Gordon and McCann, 2000; Isbasoiu, 2006).

Describing how the existence of local capabilities [2] create differentiated conditions for companies within ‘real’ clusters, Menzel and Fornahl (2010) argue that clusters are essentially formed from path dependencies (Martin and Sunley, 2006), transaction costs economies (McCann and Sheppard, 2003) and small cognitive distances originating from spatial proximity (Maskell, 2001). Thus, that set of factors are expected to create a specific regional dynamics with influence on the firm’s economic performance. Taking into consideration the different stages of a cluster life cycle, and the misunderstandings related to the claims associated with the clusters and competitiveness, Schmitz points out that ‘A group of small producers making the same or similar things in close vicinity to each other constitutes a cluster, but such concentration in itself brings few benefits’ (1999, p. 4), emphasizing that the mere presence of firms in a delimited area does not represent a source of value creation able to improve in a significant way the local economic performance.

Following the arguments above, the differences between regional clusters and simple agglomerations (groups of firms) lie mainly on the interconnected nature and spatial proximity. Thus, clusters are characterized by intense collaborative networks and concentrations of collaboration and competition (co-opetitition) (Pitelis, 2010), conditions which offer significant opportunities and stimulate the emergence of regional competitive advantages (Steinle and Schiele, 2002). Complementarily, another critical characteristic observed within clusters is the diversity of actors. According to Porter (1990, 1998, 2000), an industrial cluster includes suppliers, consumers, related industries, governments, and supporting institutions such as universities. This way, the existence of a regional network formed by a significant group of interconnected local actors is one of the critical factors to understand the differentiated competitive performance of firms within clusters (Steinle and Schiele, 2002). Illustrating that argument, Saxenian (1994) observed that Hewlett Packard and other firms at the Silicon Valley had their performance improved by the development long-term partnerships with suppliers located geographically close. Moreover, based on that observation, Saxenian concluded that, especially in high-tech industries, the physical proximity represents a facilitator to the establishment of efficient collaborative arrangements required to create and manage complex products and services.

1.2 Evolutionary Stages of Industrial Clusters

Despite the vast cluster literature, the number of academic works discussing the evolutionary patterns of clusters overtime is not so extensive. Some examples can be found in Pouder and John (1996), Klepper (2001, 2007), Wolter (2003) and Andersson et al., (2004), and despite the divergent perspectives, it is accepted that clusters follow a kind of life cycle comprised by different phases that significantly differ in their characteristics and influence on firm’s performance. Regarding the cluster dynamics, Pouder and John (1996) argue that comparative analysis between clustered and non-clustered firms during the industry life cycle reveal that firms within clusters outperform those geographically dispersed at the initial stages of development, and have a worse performance at its end. That fact suggests that the cluster life cycle is not just a local representation of the industry trajectory, but is a result from local peculiarities. The comparative analysis developed by Saxenian (1994) between the computer industry in Boston and Silicon Valley illustrates how different clusters belonging to the same industry are very likely to follow different trajectories (Menzel and Fornahl, 2010).

Proposing a different perspective, Klepper (2001, 2007) suggests a model to demonstrate how the cluster’s life cycle is determined by some the industry patterns. Klepper analyzed the automobile, tire and television industries and observed that at the beginning of the industry life cycle it was not possible to observe clear geographic concentrations of firms, with most of the firms spatially dispersed. He observed that in those industries clusters started to emerge and develop according to the industry growth rates. Klepper argues that the local characteristics originating from the spacial proximity (e.g., intensive spin-off process) give the stimulus for the geographic agglomeration of the whole industry, not only for specific groups. At the time the industry growth rate reduces, the attractiveness to remain agglomerated will also decrease and the industry will become dispersed again. That model proposed by Klepper represents a Technology-Product- Industry (TIP) life cycle. The logic behind this model is on the impact that the evolution of products and innovations has on the size, number, and location of firms. Wolter (2003) criticizes the model proposed by Klepper arguing that the growth rate cannot explain the agglomeration process in all industries on equal basis. Moreover, Wolter disagrees with the determinist perspective proposed by the TIP model, once it neglects that mature industries can be reinvented by radical or incremental innovations of new products and process.

Analyzing the economic performance of firms within clusters Pouder and John (1996) attribute to the existence of ‘mental models’ and biased cognitive focus the characteristics responsible for shaping the movement through the cluster’s life cycle. Following that perspective, at initial stages the cluster dynamics creates an innovative environment that exerts positive impacts on the firm’s performance. However, overtime that initial condition is eroded by strong institutional pressures that create a homogeneous macroculture that acts inhibiting the innovative capacity of the firms within the cluster. As in the model presented by Klepper (2001, 2007), that trajectory proposed by Pouder and John may also be criticized by the determinism that ignores the possibility of adaptations or reconfigurations in order to avoid lock-ins and other negative effects.

Considering the arguments proposed by Menzel and Fornahl (2010, p. 8) that ‘very few clusters follow a rigid life cycle from emergence to growth and decline’, it is expected that clusters evolve overtime according to the local dynamics created by economic and social interactions among firms and institutions. That dynamics may be influenced, but not strictly determined by industry patterns (Wolter, 2003). Following a generic and stylized trajectory, within successful clusters the local network formed by inter-firm connections will tend to be intensified overtime, with an increasing number of formal and informal interactions between the long-established companies and new the ones attracted to the cluster. Even though it is more conceivable to assume that the decisions adopted by firms and institutions are shaped by specific circumstances, a generic trajectory can be described following the stages illustrated in Figure 1.

Figure 1: The cluster life cycle

C:UsersLucasAppDataLocalMicrosoftWindowsTemporary Internet FilesContent.WordSem título.jpg Source: Andersson et al. (2004, p. 43)

Agglomeration: It is possible to observe the existence of a number of companies and other actors (e.g. banks, government agencies, universities, accountants, and lawyer’s offices) in a specific region working around the same or interrelated activities.

Emerging cluster: Forming the embryo to the cluster some actors start to cooperation around some core activities, and start to realize the existence of common linkages.

Developing cluster: The linkages are intensified by the emergence and attraction of new actors to the region, resulting in the creation of more interaction. In this context the development of inter-firm-cooperation becomes more evident through the development of joint efforts.

The Mature cluster: This stage is configured by the presence of a certain critical mass of factors that consistently influence the competitive performance of the firms inside the cluster. The internal dynamics is characterized by the presence of an institutional environment, strong linkages, complementarities and the emergence of new firms through startups, joint ventures and spin-offs.

Transformation: Indeed the process of continuous environmental change in markets, technologies, regulations and other process, to be successful a cluster have to innovate and adapt to these new conditions, other way stagnation and decay may affect the cluster dynamics. That process of change/adaptation may happen through the emergence of one or several interconnected clusters with focus in other activities, or by new configuration in terms of networks of firms and institutions.

The presence of economic benefits for clustered firms described by authors like Schmitz and Nadvi (1999), Ketels (2004), Isbasoiu (2006) and Pitelis and Pseiridis (2006) are closely related to the stage of development that a cluster is experiencing. For example, an emerging cluster is not actually a cluster, since the small number of firms is not expected to present a high level of linkages and do not form a critical mass. Moreover, the absence of strong interdependencies such as labour mobility, spin-off, socioeconomic networks and intense exchange of good and services prevent the emergence of local capabilities. Thus, same considering that this stage constitutes the ’embryo’ that determines the future cluster orientation, at this point the firms are not expected to be strongly influenced by a complex local dynamics.

Observing that fact, Menzel and Fornahl (2010) present a skeptical position regarding the effectiveness of any competitiveness policy intended to stimulate the development of clusters at initial stages (agglomeration and emerging), since the existence of horizontal and vertical links among firms concentrated in the same region constitutes a very common fact around the world. Thus, it is almost impossible to distinguish agglomerations with real potential to become a cluster from less complex structures. Consequently, emerging clusters are almost always only described ex-post.

After the initial stages of the clusters life cycle it is expected the development and intensification of interdependencies between firms within cluster boundaries (Press, 2006). Indeed the development of those interdependencies, firms start to resemble more with each other, being observe the emergence of ‘convergent designs’ in terms of technological models (Menzel and Fornahl, 2010), specialized labour market (Cooke et al., 2007), production systems (Pitelis, 2010) and inter-firm relationships (Blien and Maier, 2008). Moreover, developing clusters also attract a high number of start-ups that act stimulating the intensification of intra-cluster relationships. This way, that process of convergence and expansion of the number of firms within the cluster boundaries culminates in the development of self-reinforcing external economies that decrease the heterogeneity among firms at the same time that creates benefits like transactions cost economies and the privileged access to local knowledge.

As clusters reach the stage of maturity, the standards and configurations originating from past decisions become consolidate and it is observed a reduction in the growth rate of firms attracted to the cluster (Klepper, 2007). At this point the cluster trajectory may take two different directions. Keep unchanged, and suffer with a homogenization process that creates bias economic activities and therefore prevent firms to adapt to external shocks (Menzel and Fornahl, 2010). That situation traps firms in previous successful development path and lead to the geographic dispersion of the local actors and to the deterioration of the interdependencies and capabilities. The other possible trajectory is observed in clusters that reach the stage of maturity and successfully sustain the local dynamics by a continuous process of reconfiguration and adaptation to the external shocks (Wolter, 2003).

1.2 Clusters and Economic Performance

The extant theory offers a wide range of explanations to justify the economic and competitive benefits experienced by firms located within clusters. Krugman (1991) stress the existence of increasing returns originating from the concentration of firms in the same area, arguing that the geographic proximity puts together the main parts related to firm’s activities (e.g., labours, firms, suppliers and costumers) resulting in transaction costs economies. Following other perspective, Schmitz and Nadvi (1999) argue that unintentional external economies are not sufficient to explain the competitiveness of firms located within clusters, attributing to the existence of deliberate joint actions (e.g., sharing equipments, associations, strategic alliances and producers improving components) a critical source of the competitive advantages. Pitelis and Pseiridis (2006) explain the levels of competitiveness and productivity associated with clustered firms considering the existence of specialized human resources, infrastructure and befits associated with unit costs economies complemented by the presence of an institutional atmosphere. Stressing a different point of view, Bahlmann and Huysman (2008) adopts the knowledge-based view of clusters to emphasize the relevance of knowledge spillovers among the firms to explain the advantages originated from the agglomeration process. Dupuy and Torre (2006) explains the existence of cluster in terms of the advantages originating from trust relationships that increase confidence and reduce risk and uncertainty about the intra-cluster operations taking place among the firms. Moreover, Zyglidopoulos et al. (2003), describe the positive effects that the reputation of a cluster may exert on the internalization process of small and micro enterprises through the alleviation of strategic constraints associated with factors like qualified work force, financing and reduction of the firm’s legitimation expenses.

Despite that wide range of arguments, the most traditional perspective found in the cluster literature has explained the competitive advantages of clusters in terms of productivity and innovation (Pitelis, 1998; Porter, 1998), suggesting that the special characteristics originated from the economic and geographic proximity have significant impact on those two factors. Supplementary, Enright (1998) considers that the characteristics present inside the cluster’s local environment result in pressures, incentives and capabilities that increase the firms’ competitiveness comparatively to dispersed competitors, explaining the clustering process in terms of geographically restricted characteristics.

Moreover, Solvell et al. (2003) suggests that the competitive advantages emerging from regional clusters may be classified as static and dynamics. According to this perspective, while the agglomeration process triggered and sustained intensively or exclusively by factors like natural resources, low cost labors and government subsidies offers a vulnerable (easy to be copied, substituted or simple eroded by environmental changes) competitive position, clusters based on dynamics characteristics like multi-sectorial externalities, advantages of scale and scope and specific knowledge spillovers are more dynamics and competitive. Extending the arguments presented by Solvell and his colleagues, Andersson et al. (2004) considers that the sustainability of static and dynamic competitive advantages is not strictly determined, arguing that ‘static’ factors are the main responsible for the emergence of clusters, while the dynamics factors are only developed along the different stages of the cluster life cycle.

Complementing the understanding about the influence of the cluster dynamics on the firms’ economic performance some authors like Porter (2001); Garnsey and Heffernan (2007); Karlsson (2008) and Mason (2008) describe the existence of a self-reinforcing process originating from the agglomeration externalities that contribute to create a regional virtuous-circle of increasing productivity, competitiveness and value creation. Following that argument, the economic and geographic proximity will stimulate firms to innovate more indeed benefits originating from local capabilities, which will stimulate even more the agglomeration process through the intensification of inter-firm relationships and the attraction of other firms from outside the cluster, which in turn will strength the local capabilities (Blandy, 2003, p. 101). Thus, the dynamics of clusters is expected to be self-reinforced by agglomeration benefits with significant influence on the firm’s performance.

Putting together the arguments associated with the economic impacts experienced by clustered firms indeed the existence of local factors, it is possible to identify and describe the following positive location-specific externalities:

Cost savings indeed the geographic proximity with specialized suppliers, labours and distributors;

Knowledge-spillovers (intentional and unintentional), since firms inside clusters can benefit from the knowledge dissemination process that may take place especially through inter-firm cooperation, specific linkages and labour mobility;

Deliberate joint actions facilitated by the engagement in alliances and partnerships to achieve strategic objectives;

Trust relationships, that through the geographic and economic proximity minimize the uncertainty associated with commercial operations, resulting in transactions costs economies;

Pressures for higher performance, stimulated by the proximity with competitors;

Specific Infrastructure and public goods that are oriented to attend the cluster demands, like roads, ports, laboratories and telecommunication networks;

Complementarities, associated with firms in different activities but sharing common factors like raw material, clients and technologies that may enhance the cluster efficiency as whole.

Discussing the role of regional clusters in shaping competitive patterns, Tallman et al. (2004) proposes a distinction between the types of competitive advantages emerging from clustered firms: based on traded interdependencies and based on untraded interdependencies. The concept of traded interdependencies is related to the existence of inter-firm transactions inside the cluster, and is observed in formal exchange operations that may take place in form of alliances, commercial operations and acquisitions. On the other hand, untraded interdependencies are related to less tangible effects, and are ‘based on shared knowledge for which no market mechanism exists; with no formal exchange of value for value’ (Tallman et al. 2004, p. 261). To illustrate the mechanisms by which the untraded interdependencies take place, it is possible to mention unintentional external economies associated with tacit knowledge shared through mechanisms like labor mobility.

Those different types of interdependencies, especially untraded, present at the cluster level, represent a source of competitive advantage that is likely to be causal ambiguous (for firms inside and outside the cluster) and high complex in terms of their origins, what consequently constitutes attributes difficult to be replicated by competitors. However, the presence of untraded effects, especially unintentional knowledge spillovers, is viewed Enright (1998) as a constraining factor for firms within clusters, since the establishment of an efficient information flow may limit the firm’s capacity to obtain monopoly profits from the development of innovations.

Complementing the negative effects originating from the clusters dynamics, some authors also describe agglomeration diseconomies that have a negative impact firms located within clusters. For example, congestion effects (Arthur, 1990), institutional sclerosis (Pouder and John, 1996; Pitelis, 2010), rigidities associated with labour mobility and natural resources (Krugman, 1989) and pollution (Fan and Scott, 2003). This way, the dynamics and performance of a cluster is determined by the interplay between positive and negative externalities observed during the different stages of development that a cluster is expected to pass overtime (Wolter, 2003)

Limitations in the Cluster Theory

Notwithstanding the advances in the cluster theory some questions still remain insufficiently explored. One of the main limitations observed in the current state of the cluster literature is the lack of comparative perspectives to explain the advantages and disadvantages of clusters relatively to other alternative models of organization of economic activities. In his very novel approach, Pitelis (2010) suggests that any perspective trying to explain clusters in terms of absolute advantages is at the very best incomplete. In this context, Pitelis proposes the comparison of clusters vis-à-vis to markets and hierarchies in order to understand the reasons and conditions that lead firms to engage in intra-clusters relationships, market operations (outside the cluster) or integrate within the firm’s hierarchy. In fact it is not necessary a great effort to conclude that most of the cluster theory has been developed following a mono-institutional approach (e.g. Porter, 1990, 1998; Saxenian, 1994; Rosenfeld, 1998; Swann and Sennett, 1998; Schmitz and Nadvi, 1999), while some few exceptions concentrated on transactions costs (e.g. Fujita and Thisse, 1996; Iammarino and McCann, 2006; Takeda et al., 2008) and knowledge creation efficiency (e.g. Hendry et al., 2000; Tracey and Clark, 2003; Reinau, 2007; Kongmanila and Takahashi, 2009) have been drawn on a comparative approaches between clusters and open-market operations. Assuming the arguments proposed by many scholars that ‘clusters are engines of innovation’ (Davis, 2006, p. 32), the lack of comparative perspectives do not answer the question ‘why’ clusters are more efficient than markets or the ‘hierarchy’ to improve the firms’ innovative capacity (Pitelis, 2010). Thus, despite the wide number of ramifications observed in the cluster theory such as innovative efficiency, productivity, social capital and social interactions, its explicative power remains almost always restricted