ساختار و تحول خوشه های صنعتی: معاملات، تکنولوژی و سرریزهای دانش

In this paper, we investigate the relationship between location patterns, innovation processes and industrial clusters. In order to do this we extend a transactions costs-based classification into a knowledge-based taxonomy of clusters, along the lines suggested by a critical revision of the main assumptions underlying most of the existing literature on spatial clusters. Our arguments show that the transactions costs approach and the innovation and technological regimes framework are broadly consistent, and that real insights into the microfoundations, nature, and evolution of clusters can be provided by these classification systems.

Over recent years, the interrelationships between technology, innovation and industrial location behaviour have come to be seen as essential features of regional development. Much research and policy-thinking has been devoted to understanding the factors explaining why particular types of technologies appear to blossom in particular localities, and how this affects local economic growth. Lessons are often drawn from observations of particularly successful ‘innovative’ regions as a means of re-modelling both industrial and regional policy. It will be argued in this paper that insufficient consideration is still devoted to both the nature of innovation processes and the structural conditions under which technical change occurs across space. In order to explain the observed variety of geographical models, it is necessary to take into account the nature of new knowledge in different production sectors. In particular, technological regimes, industrial structures and organisational practices, as well as their dynamics, are often overlooked in favour of simplified and stylised constructs, which appeal to consultants or government policy-makers wishing for easy answers to complex problems. An example of this is the literature promoting industrial clusters. This paper attempts to classify industrial clusters on the basis of the existing literature, trying in particular to integrate transactions costs views and innovation and technology perspectives to give account of both the diversity of cluster structures and the multiplicity of their evolution paths. In doing so, the following questions are here indirectly addressed. How can we explain the variety and distinctiveness of geographically bounded industrial clusters? Why particular types of technologies tend to thrive in particular localities? How do different types of clusters evolve over time? The paper is structured into 10 sections. In the following section we discuss the various hypotheses which exist concerning innovation and geography. In Section 3 we outline the generally-held arguments regarding the relationship between geography and knowledge spillovers, and in Section 4 we present a transactions costs classification of different types of industrial clustering previously developed elsewhere, which is explicitly based on the implicit assumptions underlying most of the existing literature on agglomeration and clustering phenomena. Such a classification is very informative regarding identifying the nature and organisational logic of clusters, and on this basis Section 5 of the paper addresses the limits of the hypothesised advantages of clustering by considering the effects of unintended knowledge flows. Section 6 then explains the limitations of the transactions costs view in analysing the processes of cluster evolution, whilst Section 7 briefly introduces evolutionary perspectives on technical and structural change. Such perspectives are adopted in Section 8 to extend the transactions costs classification proposed in Section 4, in order to give an account of the diversity and multiplicity of possible evolutionary paths of industrial clusters. Section 9 uses selected empirical examples to show the importance of both transactions costs and knowledge regimes in explaining patterns of cluster development. Section 10 outlines some brief conclusions.

In the light of the arguments presented in this paper, it becomes clear that all industrial clusters can be characterised in terms of both transactions costs and relations characteristics as described in Table 1, and also in terms of technological regimes and knowledge characteristics along the lines depicted in Table 2. Our aim, as in all attempts to classify units of analysis by reducing the complexity of the whole population, was to maximise differences among the categories. However, as Pavitt himself said about his own taxonomy, the main weakness of our attempt “is the high degree of variance still found in each category” (Pavitt, 2000, p. xi). This is all the more true here as, while Pavitt approach was inductive and based on detailed empirical observation of individual units of analysis such as firms (Archibugi, 2001), ours is deductive, based on different streams of the literature on the geography of innovation, and it attempts to classify composite units of analysis such as clusters. From theories of innovation and technical change we know that innovators will tend to emerge in locations where technological opportunities are the highest. When there are conditions of high opportunity, high appropriability and high cumulativeness, innovators will tend to be geographically concentrated, giving rise to emergent clusters. Nonetheless, whether these types of situations will arise depend on the nature of knowledge in both the industry and the firms. Whereas technical knowledge tends to be prevalently tacit, complex and systemic, the transaction costs- and knowledge-based arguments here suggest that, in some circumstances, the transfer of such knowledge will be facilitated via informal personal contacts and exchanges in situations where firms are geographically clustered. Conversely, geographical concentration will be far less important when the industry knowledge base is simple, well codified and conditions of low opportunity, low appropriability and low cumulativeness prevail. However, the possible alternative characteristics of clusters presented here indicates that technological and knowledge features alone are not a sufficient guide to the types of clusters that are likely to emerge, nor are industry characteristics. Rather, as we have seen, knowledge and innovation processes, organisational, firm and industry-specific characteristics, and institutional and governance settings, all play a role in explaining the diversity of industrial clusters and also their evolutionary trajectories. Indeed, as any single firm (particularly when large and multinational) can follow more than one technological trajectory (Pavitt et al., 1989), clusters may well be engaged in a prevalent but not exclusive trajectory at any given point of time. Process-based classificatory attempts, such as that presented in this paper, help thus explain multiple trajectories and patterns of evolution. Once we account for innovation and knowledge creation processes, it becomes very difficult to apply simple stylised cluster constructs, because there is neither a representative Marshallian firm nor an illustrative ‘innovative’ cluster. Co-location therefore may or may not offer structures, organisations and institutions which improve the likelihood of local innovation. Understanding this diversity, and in particular both the transactions costs features and also the knowledge features of any particular cluster, should be the underlining base for any policy actions geared at finding actual solutions to particular regional or industrial problems. On this basis, our future research will follow a two-fold path: (1) extend dynamic comparisons among empirical cases, to have feedbacks on the scope and limitations of our classificatory attempt; (2) achieve a workable definition of the appropriate unit of analysis for assessing knowledge spillovers, and ultimately drawing policy implications.