![]()
Part I
Knowledge
![]()
An assessment of industrial clusters' literature
Elisa Giuliani and F. Xavier Molina-Morales
1.1 Introduction
Spatially clustered business activities have existed in economic studies for a long time (Marshall, 1920). Geographic agglomerations of firms specialised in one or more connected industries are here defined as industrial clusters. Introduced by Alfred Marshall at the beginning of the last century, the concept was revamped at the end of the 1970s by Italian scholars interested in understanding the similarities between Marshall at districts and Italian industrial districts (Bagnasco, 1977; Becattini, 1979). Next, the interest in districts or clusters1 spread internationally, becoming a widely studied phenomenon. Cases of agglomerations can be found both in advanced countries (Aydalot and Keeble, 1988; Sabel, 1989; Schmitz and Musyck, 1994; Saxenian, 1994) and developing countries (Schmitz, 1995; Rabellotti, 1995; Cassiolato and Lastres, 1999), as well as in knowledge-intensive manufacturing sectors (Aydalot and Keeble, 1988; Breschi and Lissoni, 2009; Ter Wal, 2009) and more traditional manufacturing sectors (Rabellotti, 1995; Staber, 2011). In spite of the existence of some significant contextual differences between clusters across the world, they are generally considered to share some key characteristics such as local intense formal and informal relationships, which favour inter-organisational trust and external economies.
Since the 1980s, industrial clusters have come to play an increasingly central role in the analysis of the economic growth and competitiveness of countries (Krugman, 1991; Porter, 1998). In particular, a wealth of empirical research has found that clustered firms show a higher innovative capacity than isolated firms (Porter, 1990; Baptista and Swann, 1998; Baptista, 2000). Building upon the original contribution of Marshall (1920), for many years scholars have attributed this success to the presence of three types of external economies: the presence, at the local level, of a pool of skilled human resources; the local availability of inputs; and the presence of technological spillovers. On these grounds, subsequent scholars have developed a wealth of conceptual frameworks and terminologies to explain the power of ‘meso-level’ forces for cluster competitiveness.
Among these meso-level forces, scholars have pointed at the importance of relational aspects. For instance, Italian scholars of industrial clusters have long emphasised the importance of localised social and productive ‘thickening’, an expression that points to the fact that economic actors are embedded within dense local social and productive networks (Becattini, 1989). American economic geographers have used the concept of untraded interdependencies (Storper, 1997) to refer to labour markets, public institutions, and locally or nationally derived rules of action, customs, understanding and values, which were unique within regional and local contexts and which boosted their development potential. Finally, economists have referred to the concept of localised knowledge spillovers, and have interpreted the high innovativeness of clustered firms by conceiving local knowledge as a public good, which spreads pervasively within spatially bounded areas (Audretsch and Feldman, 1996; Baptista and Swann, 1998).
While these concepts were mainstreamed during most of the 1980s and 1990s, the profound transformation of clusters in light of the new global distribution of production and innovation has prompted scholars to call for new interpretative frameworks of cluster innovation and competitiveness. In particular, scholars have started to question ‘meso-level’ interpretations of the success of clusters and have proposed alternative analyses. Among the critical views, we consider the following to be of key relevance for the wealth of studies that have followed thereafter.2 First, scholars questioned the over-emphasis on local knowledge spillovers (LKS) as a concept that explains most of cluster innovativeness, and became dissatisfied with the way that it was measured and understood. In their seminal paper, Breschi and Lissoni (2001) argued that:
The concept of LKS is no more than a ‘black box’, whose contents remain ambiguous. On the one hand, its frequent citation serves well an evocative purpose, i.e. it helps to signal a strong interest in coupling ‘geography’ with ‘innovation’ as research themes. On the other, by contrast, it provides the researcher with an escape route to avoid studying the specific mechanisms through which the two phenomena are linked.
(Breschi and Lissoni, 2001, p. 976)
This critical view sparked the subsequent generation of studies aimed at opening up the localised knowledge spillover black box, as discussed in section 2.
Second, most of the early studies on clusters treated firms as homogeneous actors and generally overlooked the importance of individual agency and characteristics in explaining cluster innovativeness. This means that the extant literature has assumed that there is a high degree of internal homogeneity within clusters. This view still predominates today, to the point that, in a recent contribution, Ter Wal and Boschma (2011, p. 921) argued that research on clusters ‘[does] not pay close attention to the fact that firms are highly heterogeneous in terms of capabilities, strategies, and routines […] in that literature, clusters matter and not so much firms.’
In this chapter we attempt a review of the most salient of the papers which address the limitations mentioned above. In particular, section 2 discusses the advancements in the measurement of localised knowledge spillovers, taking as its reference the recent literature on local knowledge and innovation networks in industrial clusters. Section 3 reviews some of the critical studies that have taken into account the heterogeneity of cluster firms (and other actors) in order to explain their performance. Section 4 concludes by tracing some promising lines of research in this area of investigation.
1.2 Unpacking localised knowledge spillovers
In the concluding section of their article, Breschi and Lissoni (2001) call for a new research agenda in the study of localised knowledge spillovers. The core of their argument is that, within clusters, there are different kinds of knowledge flows, which should not all be placed within a unique bundle. They criticise the fact that the extant literature pays insufficient attention to disentangling the various determinants of spillovers, or to studying whether and how all firms benefit from their presence purely as a result of their co-location. Indeed, the generation of local spillovers – often referred to as the ‘local buzz’ – is only in part a random process of knowledge flowing from one firm to the other. In fact, the overall phenomenon cannot be entirely due to pure chance. Furthermore, knowledge flows through different channels: labour mobility, trade of goods, informal ‘chit chat’, communities of practice, formal meetings, and so on. All of these channels differ in their potential to transfer valuable knowledge and respond to different logics and incentives. More importantly, not all actors in the cluster get exposed to the same amount of local knowledge, as conventional localised knowledge spillover stories would presume. To understand how knowledge is diffused within a cluster, and also the degree to which firms benefit from localised knowledge spillovers, it is necessary to study the networks through which knowledge gets transferred within the cluster. However, the conventional literature is also rather superficial in terms of understanding the characteristics of networks that transport knowledge across organisations. As remarked by Staber:
all economic action in industrial districts is said to be embedded in a dense web of network ties […] But beyond the widespread reference to dense networks as a characteristic of successful industrial districts, many investigators are surprisingly silent about the structure of networks.
(Staber, 2001, p. 537)
The need for a new research agenda oriented toward unpacking the concept of localised knowledge spillovers has sparked, among others, a new wave of cluster studies, which has introduced methodological novelties in the measurement of localised knowledge spillovers and, more generally, in that of local networks. Through methods of Social Network Analysis (SNA) (Wasserman and Faust, 1994),3 scholars have started to look at the way flows of knowledge between cluster firms were distributed within local networks, allowing for a potentially neater understanding of the drivers and consequences of knowledge networks in clusters. In this section we review a selection of influential papers on this subject, a summary of which is presented in Tables 1.1–1.3. Papers in this area of research aim at answering three types of questions.
Table 1.1 Geography for the diffusion of knowledge
The first group of studies (Table 1.1) focuses on whether geography matters for the generation of knowledge linkages, and on the significant drivers of localised knowledge spillovers. While most studies seem to confirm that geography matters (Balland, 2011; Scherngell...
