Geographies of Networked Industrial Systems: the Case of Piedmont

6th Conference European Urban and Regional Studies “Boundaries and connections in a changing Europe and Health and well-being in Europe’s cities and regions” Geographies of Networked Industrial Systems: the Case of Piedmont Daniele Ietri and Francesca Silvia Rota+ Dipartimento Interateneo Territorio Politecnico e Università di Torino [email protected] [email protected] Draft, September 2006 Abstract From a quite commonly accepted perspective, competitiveness relies more and more on local actors’ (firms as well as research centres, public bodies and local agencies) ability in creating efficient systems of production and innovation, in order to distinguish themselves from competitors. In particular, specialisation in products and competences is considered a privileged way for firms and territories to face the challenge of Knowledge Based Economies. In the literature, great attention is paid to the concept of cluster and to the development of cluster-based policies as useful tools to steer regional economic development. In an industrial cluster, public agencies, universities, and firms form a complementary network facilitating continuous growth, upgrading, and restructuring. An industrial cluster establishes a regional competitive advantage that allows to attract investments, diffuse knowledge, absorb and create new technologies, and retain skilled workers. Nevertheless, this might not be the only successful way to organise a productive system. As demonstrated by the literature, the networked organisation of a locally embedded productive system may be conceptualised in terms of different models such as: industrial districts, new industrial spaces, supply chains, local productive systems and, of course, clusters (Moulaert and Sekia, 2003). + Despite the paper results from the work of both the authors, the writing of the introduction and paragraphs 1 3 and 4 are to be attributed to Daniele Ietri; while paragraphs 2, 5 and the concluding remarks are due to Francesca Silvia Rota. 1 Assuming the case-study of Piedmont (in Italy), the literature dealing with the description of the regional manufacturing fabric is investigated, in order to point out different models of industrialisation, their trends and reciprocal relationships (continuity, substitution and juxtaposition). As it will be demonstrated, in period 1970-2005, at least eight different models have been produced (comprensorio, industrial district, technological district, concentration/specialisation area, technopole, cluster, filière, localised supply chain) echoing contemporary territorially embedded trends. As a conclusion, the importance of territory in influencing economic processes is firmly stressed and some policy suggestions are pointed out. Keywords industrial systems, competitiveness, cluster 2 Introduction Economic geography has always been concerned, even if with different approaches, about the relationship between industry and territory. In Europe, in particular, the recovery of industry after the crisis of the 70s and 80s has adopted new organizational and territorial forms which need new explanations. In such a context, the interest in spatial concentration of industries has grown in an astonishing way both among scholars and practitioners. Above all, the concept of industrial cluster gained the most widespread diffusion. Anyhow, this concept reveals its weakness as well, in that a real explosion of interpretations and implementations may be found in literature: even if some authors, such as Ketels (2003), believe that current discourses on clusters are signals of a new stage of theoretical development rather than inherent conceptual weakness, there is little doubt that, in current praxis, almost every concentrated productive system may be thought in the form of a cluster. Clusters have become “objects of desire”, resting on the widely accepted assumption that increased local specialisation will lead to increased levels of productivity, growth, employment and innovation (Steiner, 1998). Motivated in large part by Michael Porter’s “Competitive Advantage of Nations” the field has over the last two decades seen a lot of case-driven theory development and the experimental use of clusters in economic development, often pushed by committed individuals (Ketels, 2003, p. 3). In this framework, a critical approach to the concept of cluster appears particularly crucial. Even if the theoretical reflection has sanctioned the effectiveness of the model1 and competition among regions is more and more played on the basis of cluster policies, the outcomes of its application are still to be verified. As stated by Cumbers and MacKinnon (2004), albeit cluster-based policies have been adopted by a range of organisations operating at different geographical scales, “the real test of the clusters approach will be its capacity to provide us with a better understanding of urban and regional growth processes that moves beyond previous approaches” (p. 960). In other words, clusters’ ability to match a rather codified model (namely Porter’s cluster, see below) to the different factors of the competitive advantage of a region is everything but demonstrated. The aim of this paper is thus to discuss whether the cluster represents an universal concept (i.e. the proper one to all the possible contexts) or just an appealing key-word for politicians to collect public favour and to attract investments, quite independently from the characteristics and issues of the territory. In doing that, we will not deal with the theoretical consistency of the concept itself, but with its strength (or weakness) as an operational tool. The main point that will be discussed is if clusters are just an useful instrument in order to promote local competitiveness, while lacking in operability and territorial sensibility as a geographic discourse. In particular, in paragraph 1 we will introduce the concept of cluster both for a theoretical and a practical point of view. As we will state, in current debate on regional development and competitiveness this is in fact a concept that is having an extreme but undifferentiated diffusion. In paragraph 2 1 The Organisation for Economic Co-operation and Development as well as the European Commission, for example, promote clusters in many official documents. See, among the others: OECD, 1999 and 2001; EC, 2002. 3 the cluster as a policy concept will be thus reviewed, pointing out in particular some doubts on its interpretation as an universal theoretical framework. Starting from this, paragraph 3 will deal with an overview on some of the most successful models used by economists and geographers to describe networked industrial systems: industrial districts, technopoles or technological districts, local supply chains or filières. Some of the reflections rising from this part of our research will be then used to formulate a methodological proposal in the analysis of regional productive systems. More specifically ,a “memorandum for analysis” will be proposed, whose strengths and weaknesses will be verified starting from its application to a selected case. As a consequence, paragraph 5 will be dedicated to illustrate some first empirical insights on Piedmont region in Italy, that will be also used in paragraph 6 as a starting point to formulate some concluding remarks. 1 Are them all clusters? The widespread diffusion of the cluster concept is largely due to the theoretical and empirical work of Michael Porter. Although we can trace back its origins in Marshall’s Principles, in Porter’s writings we clearly find the attempt for using clusters in the explanation of the competitive advantage of a given area. Porter’s definition may be considered an evolution from the description of industrial district originally formulated by Marshall (1920): Marshallian districts were in fact identified by a productive area mainly made up of small enterprises, generally locally owned, and taking benefits from the fact that “the secrets of industry float freely in the air”. In a district the concentration of specialised productive processes matches in fact with a flexible labour market and a diffuse industrial atmosphere, i.e. entrepreneurially and technologically oriented. More specifically, Porter (1990) detects at the local scale some key success factors, that can be summarised as follows: - strategic location, in terms of infrastructure endowments, services and high-rank functions in general; local market demand, i.e. the opportunity to rely on a large potential basin of clients/users; the integration among different clusters, i.e. at the regional level; the ability to hire employees from high qualitative human resources. Standing on these factors, Porter defines the components of a “competitive diamond” that includes input factors, such as infrastructures and human resources, the composition of the demand-side, the firms’ structure and competitive capability and – which is most interesting with respect to the argument of this paper – the presence of firms operating in related sectors. According to Porter, a given productive system will be particularly effective in fostering development and competitiveness when firms would be able to develop relationships, that is a typical positive effect (or external economy) of geographical concentration (or clustering). As the literature on clusters has already provided many definitions of clusters, we will avoid to add new formulations. Rather we will assume the probably most famous one and we will analyse it according to the issues which are of the most interest in our point of view. Porter defines clusters as: 4 Geographic concentrations of interconnected companies, specialised suppliers, service providers, firms in related industries, and associated institutions […] in particular fields that compete but also cooperate (Porter, 1998, p. 197) Since it has been formulated, such a definition received a dramatic attention and an uncommon diffusion among scholars and practitioners. Reasons of success are probably the simplicity with which it describes some key features in the new organisation of the production process (inter-firm relationships, the influence of institutions etc.) and the systemic approach that permits to consider many elements (firms, suppliers, clients, institutions, research centres etc.) at the same time and their interactions. In Porter’s definition complexity is not reduced or intentionally ignored at all, but described in a simple sharp way. But, at the same time, the cluster concept and its implementation via policy initiatives have been openly criticised by many scholars (see, for a review: Ketels, 2003). Critics in particular are addressed to the excessive inclusivity of the concept. According to Gordon and McCann (2000), for example, the idea of cluster is deeply multidisciplinary in nature, including theories from regional science, business studies, geography and economic sociology. The main theoretical contributions involved (some of whom will be the object of discussion in the next paragraph) stem from agglomeration theory, supply-chain studies, the implications of embeddedness and institutional thickness in local economic dynamics. Albeit this multidisciplinarity may be a positive element from an analytical perspective, it can turn into a negative one as far as practical issues are to be addressed and is necessary to face critical choices in the definition of the policy agenda. A crucial point is that the very definition of cluster has become characterised by vagueness (Martin and Sunley, 2003; EC, 2002), to the point that nowadays quite every productive agglomeration can be interpreted in terms of clusters, and the concept itself turns into a problematic source of policy advice. In particular, we agree with those scholars such as Newlands (2003) and Benneworth and Henry (2004) who underline that, after its first introduction, the concept has been adopted by too many different theoretical schools, re-interpreting and using it, each from its peculiar point of view. Moreover, many practitioners often adhere quite acrostically to one major definition of cluster apparently (or intentionally) ignoring the theoretical and empirical explanations which lay behind them. For example, in a recent paper, Maskell and Lorenzen (2004) explicitly define clusters as “non-random geographical agglomerations of firms with similar or highly complementary capabilities” (p. 1002, note 1). If practitioners would limit their consideration merely to such a definition, they will probably focus just on firms and their productive activities, failing in considering other important aspects characterizing the cluster (i.e. the importance of relationships among firms and institutions and the dramatic role of the regional geographical scale in the development of a cluster), and that are actually outlined in the following-up of Maskell and Lorenzen’s discourse. Coherently with such a perspective, Benneworth and Henry (2004) state that the potential utility of the cluster concept as an analytical tools relies mainly on its capacity to drive forward diverse theories on industrial agglomeration and bring multiple perspectives. 5 This consideration leads us to another important topic in our discussion: the great success of clusters as an interpretative concept has been widely echoed in an even greater fortune as a policy tool, as confirmed by the diffusion of cluster-oriented policies in every part of the globe. In the next paragraph we will deal with them from a critical perspective. 2 A critical approach to cluster as policy concept Although the cluster concept was originally formulate with an interpretative purpose (i.e. to describe geographically concentrated set of firms and supporting institutions in a limited number of industrial sectors), it has been largely used as a framework for decision- and policy-making, a tool in the design of initiatives oriented towards the promotion of local development. The cluster idea has […] become accepted largely on faith as a valid and a meaningful ‘way of thinking’ about the national economy, as a template or procedure with which to decompose the economy into distinct industrial-geographic groupings for the purposes of understanding and promoting competitiveness and innovation (Martin and Sunley, 2003, p. 9). The main reasons for this success may be referred, first of all, to the presence of some famous success-stories, the most important of whom are Silicon Valley and Route 128. There the presence of place-specific factors, such as the clustering of high technological enterprises, public research centres or universities, and a highly qualified labour forces, are critical factors for the success of the local economy2 (Saxenian, 1994). Moreover, many success stories seem to be the actual evolution of spontaneous development trajectories, to which the cluster “label” has been attributed only after an ex-post interpretation, is spite of an intentional “policy lead” establishment. Secondly, other important reasons are the communicative capacity of the concept (that is highly fashionable) and the simplicity (standardisation) of the policy steps to be elaborated. Thirdly, another key factor is the opportunity to mask, behind the rhetoric of bottom-up local development policies, old-style top-down public financial interventions, especially when supporting old-industrialised areas or relatively underdeveloped regions. Some critics [...] see cluster-based development as a useful approach, but are concerned about its actual use in projects bearing little resemblance to the original framework. These projects are in danger of creating a serious backlash against this approach when the results of so-called cluster efforts fail to live up to their sometimes-lofty promises. Practitioners, however, are under such intense pressure to develop a new, more effective approach to economic development that they can’t afford to wait for these theoretical disputes to be resolved (Ketels, 2003, pp.2-3). In our view critical points are above all related with the contextualisation of initiatives and the selection of the proper geographical scale. It should be noted, for example, that albeit policy discourses enunciate as the leading principle that “one size does not fit all”, they frequently put in practice a quite standardised range of intervention models. 2 But other factors beside geographical concentration intervene as well, guaranteeing the success of those areas, such as military technology and public investments. 6 As to the scale, clusters are more and more pursued not only at the regional scale (that is quite commonly considered by the literature as the proper scale to develop clusters. See, for example: Braczyk, Cooke and Heidenreich, 1998) but at other geographical scales, such as the metropolitan and urban ones, where the competition for attracting firms and investments is getting stronger and stronger (Martin and Sunley, 2003). Moreover, as we stated before, the vagueness of the definition of cluster allows a wide set of possible applications. Here we provide some examples according to the type of industries eligible to assume cluster configurations: - - - firms specialised in traditional sectors. In this case, above all when firms are SME and endogenous dynamics are emphasised compared with a weak role of central government, the concept overlaps with that of industrial districts. Examples are the Italian clusters of Sassuolo (tile district) or Montebelluna (shoes district); technologically-oriented sectors. Examples are the clusters of Cambridge or Sophia Antipolis, based on a strong role of public sectors and the presence of a system of supporting institutions (universities, public/private research centres, etc.); public-owned companies. We refer both to industrial clusters supported by the government, as in the case of the Toulouse aerospace district, and clusters resulting from public funded start-ups, as in the Styrian automotive cluster. Also according to firms’ relationships, different models have been produced. Ann Markusen (1994), for example, claiming that little attention has always been paid in the conceptualisation of clusters and districts’ external cooperation and local/global relationships, distinguishes among four different spatial and organisational cluster structures or “sticky places in slippery space” (1996, p. 293): Marshallian, hub and spoke, satellite platforms, state-anchored. Figure 1 shows a schematic representation of Markusen’s concept on two of these industrial systems. A ) Marshallian industrial district B ) Hub and Spoke district Figure 1 – Spatial organisation of clusters according to Markusen (1996, p. 297) The Marshallian district is made mainly of SMEs among which just few interact with external actors, while in the “hub and spoke” cluster SMEs interact with bigger companies also outside the cluster. In both models, the cluster is supposed to act as a development pole, able to exploit benefits from proximity and networking, reinforcing the competitiveness of single firms, of the whole system as well as of the territory. Those examples show how the original Porterian concept may be easily manipulated and applied in a potentially infinite variety of cases. Starting from these reflections we thus propose a different approach to the analysis of networked industrial systems. We try to show how different geographical models of networked systems do exist, in which 7 local systems and policy interventions play a crucial role. We suggest that, in particular conditions, besides the cluster model, apparently ‘old fashioned’ models could be more suitable to the considered industrial geographies. This approach should avoid the use of cluster as a “universal instrument”, which indifferently applies to many heterogeneous cases; rather it will allow for a rediscovery of the very specificity of the cluster concept (which is still at the centre of contemporary debate). According to this, in the following paragraphs besides introducing the main characteristics of these alternative models, a review of some issues of the cluster concept will be proposed as well. 3. Other models in the theorisation of networked industrial systems Many other models exist describing geographies of networked industrial systems that may be more suitable to local specificities than an approach based merely on clusters From the literature we obtain a selection of some models used by economic geographers to describe the new spatial and organisational forms assumed by industrialised systems. 3.1 Italian industrial district The interpretation of regional economic development in Italy has been deeply influenced by the re-discovery of original Marshall’s work on industrial districts. The theorisation of districts is inspired by research on the phenomenon of “Third Italy” (Bagnasco, 1977) and on the work of Marshall, where he observes that in some industries the productive process can be realised in an interconnected group of small and medium-sized firms. Moreover, the proximity in a district allows the creation of inter-personal relationships which go beyond the pure economic scope, also concerning social and cultural matters. Giacomo Becattini (2000) defines the industrial district as “a socio-territorial entity, characterised by the co-presence, in a territorial area circumscribable in virtue of environmental and historical features, of a community of people and a population of industrial firms” (p. 58-59, our translation). According to Becattini, the community of people incorporates a system of values concerning the organisation of family, work and all the crucial aspects of everyday life: a system which emerges from the historical evolution in the district and which prevails as it is shared and accepted by the most part of the community. The overall image of the district is characterised by an ethic favourable to entrepreneurship and change, which is realised in a complex framework of institutions and rules (explicit and implicit) responsible for the preservation and transmission of the district’s values. Among the firms in the district a process of division of labour is realised, for each enterprise is specialised in a phase of the productive process. Each unit is relatively autonomous from the district’s territory, but is at the same time an integral part of the local system: thus firms and the historical processes which shaped the districts are part of a peculiar and unique reality. As we have mentioned before, the localisation of firms in the industrial district does not only concern with the proximity of their productive units, but also with proximity of actors in the local social network. This gives meaning to personal relationships and family ties among owners of enterprises and between entrepreneurs and employees: a network of interpersonal relations which authors such as Coleman (1988) or Putnam et al. (1993) refer to as ‘social capital’. Each district or local system holds a specific 8 collective identity, which is the emergent result of long-run processes, kept in a quasiequilibrium by institutionalised relationships that the local community inherited by past generations. 3.2 Technological district / Technopole We can define technological districts or technopoles sub-regional geographical areas which demonstrate a high endowment of high-tech activities in a specific industry 3. Actors in a technological district are firms, universities, public and private research centres. Following Piccaluga (2003), we can summarize the main factors of success of technological districts in Italy with: - the approval for (at least formally) bottom-up innovation oriented policy initiatives, more effective and cheaper than top-down interventions; the emphasis on local relationships, especially when activated in order to cooperate for common research and development projects; the perspective of boosting the competitiveness of a region or local system on the national or global scale; the specialisation strategies that concentrate financial and research efforts on few industries with high potentiality; the diffusion of a fashionable concept, also thanks to some success stories and best practices from abroad. Many technological districts are characterised by the presence of strong public research centres or universities, so as the localisation of important enterprises operating in high technological industries; the opportunity to hire qualified workers, an entrepreneurially and technologically oriented culture and the availability of venture capital are considered critical factors for the success of the district. The simultaneous presence of the system of creation of knowledge (university, research centres) and the system of its economical exploitation (high-tech firms, see Autio, 1998) is a fundamental factor for the identification of a technological district. 3.3 Filière / Localised supply chain A filière can be defined for the localisation, in a relatively small area, of the know-how and the productive inputs necessary for the production cycle from raw materials to the final market good. In a filière, actors are geographically concentrated and tied in market and social relationships; moreover, public policies are implemented in order to foster the development of the local productive system (Aimone, Galizzi and Pieri, 1996). Some public agencies perform promotional activities, such as the legal acknowledgement of the protected origin of agriculture and food products (DOCG, DOP, etc. in the Italian 3 In our perspective, technopoles and technological districts are similar, but for the geographical scale: in the case of an area made of the aggregation of a number of municipalities within the provincial level we would use the term ‘technological district’. We would otherwise use ‘technological pole’ when describing a more punctually concentrated area, i.e. the area of a science park or a single municipality or metropolitan area 9 context). In this sense a filière seems to fit the original definition of cluster given by Porter: nevertheless, in the Italian framework the interpretation of this local productive system made by both researchers and practitioners did never adhere to the cluster’s idea, preferring alternative models. In the case of the automotive sector the filière is more frequently defined as a localised supply chain: this model represents a stage in the historical interpretation of the industrial system, which characterised the automotive industry immediately after the crisis of the Fordist era. 4 A geographical reading of networked industrial systems As stated in previous paragraphs, in our view clusters are one particular model used to describe networked industrial system. Beside it, many other models (Italian industrial districts, technological districts or technopoles, filières or localised supply chains) may be used to describe the industrial structure of a region. But they are not interchangeable. Albeit some common features characterise them all (systemness, spatial concentration, inter-firm relations), some others result place and scale specific. These features are: - - - Path-dependence. We refer to the historical evolution which leads to the formation of a cluster. Obviously, it depends on the historical facts that characterise a place. At the same time it exists a sort of inertia in the localisation of firms which, also thanks to the cultural embeddedness of decision makers in a given territory, contributes to hold actors in clusters. In many empirical studies, there is the acknowledgment of the role of small groups of enterprises responsible for the first stages in the formation of a cluster. Instead the role of public bodies is uncertain, as they may equally have influenced positively and negatively the evolution of clusters (in the case of Turin, as an example, see Calafati, 2003); Knowledge and learning processes. In the earliest idea of external economies by Marshall and further applied by Porter, the cognitive advantages offered by agglomeration concern with the opportunity to access collective resources such as specialised good and services suppliers or highly qualified workforce. Moreover, other scholars stress the importance of knowledge formation and diffusion dynamics in the formation on cluster. Here we deal with a theorisation of a tacit and sticky nature of knowledge: in the learning process proximity is essential, as knowledge and information flow almost exclusively locally. This kind of dynamics seems to be more likely to happen when firms are embedded in a thick network of relationships in which knowledge is shared: “supported by close social interactions and by institutions building trust and encouraging informal relations among actors” (Bruschi and Malerba, 2001, p. 819). Learning is thus supported by proximity in space but also by a sort of institutional proximity, which originates in the historical thickness of relationships among actors in a given area; Governance systems. As to the management of knowledge in clusters, Maskell (2001) proposes a knowledge-based theory of clusters, outlining a twofold dynamics of knowledge creation and diffusion. The first, horizontal dimension allows competing firms operating in similar sectors to access information, thanks to the opportunity to keep constantly eye on rival’s activities and to the workforce movements among enterprises in the cluster. This is especially the case of Italian industrial districts, where a particular kind of workforce mobility may be observed 10 when former employees become independent entrepreneurs, rivals or suppliers (and vice versa). Back to Maskell’s discourse, the vertical knowledge dynamics in clusters consists in the relationships which intertwine complementary enterprises in networks of buyers-suppliers of goods and services. In the vertical dimension both operate the effect of specialisation in the supply chain and a centripetal force of the cluster in attracting firms which can benefit from the knowledge creation dynamics which characterises the agglomeration. Clusters are place-specific as far as they are based on interpersonal networks that are locally embedded and supported by the action of local actors. Nevertheless these networks are not per se geographically localised (Boschma, 2005). Albeit most part of the literature on clusters insists on the local (regional and urban) scale, “even when economic activity seems to be spatially clustered, a close inspection will reveal that the clusters rely on a multiplicity of sites, institutions and connections” (Amin and Thrift, 2003, p. 52). So far, we can outline a progressive acknowledgement of the necessity to place clusters in a multiplicity of institutional levels (regional, national, supra-national): thus, an in depth study of relationships between clusters and other geographical scales seems necessary. Bathelt, Malmberg and Maskell (2004) propose a synthetic scheme outlining how knowledge assumes different nature according to geographical scales. At the local scale knowledge is made of “atmosphere, broadcasting, noise, buzz” (p. 38): the sum of local face-to-face relationships provides a continuous flow of information, which is useful in the learning processes. As theorised by Marshall itself, the proximity, the simple colocation in the place in which things happen, is frequently sufficient to acquire, even accidentally, information and know-how essential for the innovative process. At the global level, instead, ‘pipelines’ (long-range world relationships through which codified knowledge flows) allow the diffusion of information at long range and the integration of know-how developed locally with that externally originated. A multiscalar approach to cluster research thus seems to be appropriate, considering scales without introducing (explicitly or implicitly) any hierarchical order (Wolfe and Gertler, 2004). The issues thus summarised provide a theoretical basis for an analytical methodology, proposed in order to detect, from a given regional case of study, different models of industrialisation. In particular, these are to be firstly analysed according to the following criteria: 1. Identification criterion. At least three circumstances can be hypothesized: i) models that are based on precise quantitative criteria; ii) models mainly based on qualitative observations, but eligible to also imply punctual quantitative identification; iii) models that are wholly qualitative as they result from political choices not supported by analytical evidences. In this case the identification is strongly influenced by the geography of power as well as by interests of local lobbies; 2. Geographical scale. It refers to the way the productive systems spatially structure themselves in terms of local units (firms, suppliers, clients, institutional organisations, research centres, etc) and relationships (mercantile or not). In particular, we are interested in the observation of the geographical scales at which the actors of the system interact: models may be very locally constrained or they may be structured in a multiplicity of scales (local, regional, global); 11 3. Actors involved. Here the object of analysis is the identification of the key players of the productive system. In a first tentative taxonomy, they may be private (firms) as well as public (agencies, local administrations, universities, public research centres, development agencies, etc.). Firms can be further classified according to their dimension (SMEs, MNCs, etc.) or sector of activity (high-tech, low-tech); 4. Network structure. In a productive system, relationships among the actors may vary from a condition of total hierarchy/integration (when one or few enterprises impose their power on the system as a whole, shaping its functioning according to their internal rules) to that of total reciprocal autonomy of the various actors (linked just by market transactions)4. According to many scholars, anyway, local networks are more likely to evolve towards more variable, heterogeneous and loose-coupled structures, as those described by Grabher (1993); 5. Innovative activity. As the prevailing industry influences the organisation of local productive systems, we then look for similarities/differences in the structure of systems according to their being characterised by high technology sectors or traditional ones. More specifically, to introduce such a distinction, we adopt the taxonomy proposed by Keith Pavitt (1987), who distinguished among 4 macrosectors: science based, scale intensive, supplier dominated and specialised suppliers; 6. Historical trend. For each detected model we will present some brief considerations on its dynamics, paying attention, in particular, on its fortune as a policy tool (i.e. its adoption or not in policy making processes). Moreover we will consider, when possible, the reciprocal relationships (continuity, substitution and juxtaposition) among the recognised models, in order to point out (if any does exist) a peculiar regional type of industrialisation. 5 Testing with Piedmont The adoption tout court of the cluster model in the reading of a regional economy seems problematic above all in contexts like Italy, where historical events contributed to create a peculiar and varied industrial system: a mosaic of various development patterns corresponding to different territorial contexts (Ietri and Rota, 2004). At the national scale, for example, the well-known model of the “Three Italies” proposed by Arnaldo Bagnasco refers the sub-national divergences in the economic development patterns to the existence of different forms of organization of the industrial production: large-size businesses in the North-West, industrial districts of small and medium sized enterprises (SMEs) in the Centre and North-East, sub-developed productive units in the South. This industrial district was not the only industrial model that characterised the development of the Italian economy; nevertheless (and almost paradoxically), in much of the international debate as well in most of the public opinion, this is traditionally linked with that of its most famous industrial districts Starting from this paradox, we assume as case of study Piedmont region, in the NorthWest of Italy. This is in fact a clear example of the coexistence, in the same region, of different models of industrialisation: from the Fordist-Taylorist model embodied in the 4 Here we explicitly draw inspiration from the well-known dichotomy market/hierarchy proposed by Olivier Williamson (1985). 12 history of the automotive industry, to that of the industrial districts both in traditional and innovative activities. Moreover, Piedmont represents also a dynamic institutional context in that, at least from the 70s, many attempts were spent by the administrative bodies to elaborate new models in order to interpret and (eventually) steer and push the regional economy. In order to apply the analytical model proposed in paragraph 4 to the case of Piedmont, a large amount of documents has been thus considered. These include scientific articles, books, official documents, research reports, planning documents etc. produced between the 70s and the 90s by Italian scholars and practitioners and in which the issue of the analysis of the regional productive system is pursued. Drawing information from this database, we tried to detect some main models and concepts adopted in order to describe the regional industrial geography, especially when a policy aim was recognisable. As a first result, we obtained a preliminary taxonomy 5 of territorial models of industrialisation, that just partially corresponds to those proposed by the literature (see paragraph 3). In other words, besides the reference to “classical” models such as districts, technopoles and filières, some other concepts emerge as well. In the table that synthesises the main results of the analysis (Table 1) we point out three of them: comprensorio, concentration/specialisation and local supply chain. Actually, the comprensorio is not exactly a model describing industrial dynamics. Rather, it is a territorial partition whose limits have been traced down joining together neighbouring Local Labour Systems (defined through the aggregation of municipalities according to daily home-work commuting). On the contrary, the concept of cluster is rarely mentioned. In our view, this condition is to be referred to some peculiar institutional, cultural and historical conditions that we will try to explain further in the concluding remarks of this article. What we want here to underline is a sort of reluctance by policy makers (as well as by local scholars) to use the concept of cluster to define clustered productive activities, that are more preferably described as districts, filières, localised supply chains and concentration/specialisation areas. These describe relatively circumscribed areas (generally at Province – NUTS 3 or metropolitan geographical scale) characterised by a relevant presence of firms specialised in activities (and services) that are similar, related or complementary. These, that correspond to Markusen’s sticky places (see paragraph 2), are here described as networked industrial systems. More precisely, from our analysis it results that over 61 detected territorial industrial systems, 40 can be classified as concentrations or specialisations, 16 as districts, and the remaining 5 as filières or local supply chains. The concepts of concentration and specialisation are thus the most frequently used, but this is probably related to the fact that they are too generic in their meaning 6 . Concentration/specialisation areas identify either local contexts characterised by numerous firms (or employees) in a given industry (in this case they are sometimes referred to as localised industries) or, in alternative, the presence of some local big 5 Results here presented are just provisional. Definitive ones might be provided only when the detected models will be verified with interviews with local experts, as planned as a further phase of our analysis . 6 This results clear above all if we analyse the systems defined by the literature in such a way according to the criteria of geographical scale, actors involved, network structure, innovative activity and historical trend (see paragraph 4). 13 operators in a specific industry (that is then described as industrial division 7 ). Furthermore, examples of concentration/specialisation areas involve niche productions (such as the floriculture industry of Verbania) as well as generic productions instrumental to other industries (examples are the cable industry of Felizzano and Quattordio and the rubber and plastic division of Alessandria and Ovada). Sometimes these areas also include the presence of some major firms pushing a larger system of SMEs. Moreover these concepts are used to describe very different development patterns and inter-firm relationships: industries characterised by processes of diversification of products pushed by a single major firm (as in the publishing industry in Novara) or in which some big plants together with smaller units constitute a relevant system in the national market (i.e. the industry of building materials in the Province of Cuneo) or concentration of medium size firms highly specialised in one technology or product (refrigerating in the industrial pole of Casale Monferrato, or integrated logistic industry in correspondence of regional logistic terminals and commercial platforms). As those examples clearly show, concentration/specialisation areas do not correspond to precise territorial models, rather they are representative of a widespread tendency in the theorisation of the Piedmont industrial fabric, that focuses more on masses rather than on relationships. In these concepts inter-firm relationships are somehow taken for granted, as the result of proximity8. Opposite to this definition is that of industrial district of SMEs. The analysis of the literature shows that districts are quite a diffuse reality in Piedmont, with many municipalities and a relevant share of the industrial fabric involved. Among them the textile district of Biella is probably the most famous. It is a typical Marshallian district as those described by the international literature (Markusen, 1996), made up of a universe of small (quite independent) enterprises specialised in traditional productions and some major units linked together by relationships that are not just mercantile: trust and family ties are important nexus binding together the firms of an industrial district. As to its dynamics, it emerges that the district has been characterised by a strong interaction among firms specialised in the production of fibres and those specialised in their transformation into fabrics and cloths: this process has however been arrested by the crisis that severely hit from the 90s almost all Italian districts and that corresponded with a fall in their competitiveness. Reasons are many (incapacity of SMEs to grow, to enter new markets, to enlarge and internationalise their portfolio of clients, to introduce radical innovations etc.) and this is not the proper seat to address such a complex topic. Other important regional districts that follows quite similar pattern are the jewellery district in the area of Valenza Po as well as that of the metalmechanic district in the Northern part of the region including the taps and fittings industry in Lower Cusio, Borgomanero, Valduggia and Val Sesia and the household goods one in Upper Cusio, Omegna, Varallo Sesia, Stresa. In this case, the original fabric of SMEs favoured the growing of some important firms groups such as Alessi and Lagostina. Quite different is the case of the wine district of Canelli and Nizza. During the last years this district has in fact been more and more evolving towards a filière type organisation of the production with strong relationships among firms producing the grapes, those transforming the agriculture products into wine and derivates, those processing the 7 It corresponds to the Italian word “comprensorio”. The industry of chemical clothing fibres produced in Novara and Trecate, for example, is described as a “blocked isolated sector” missing a local supply chain. 8 14 production rejects as well as those producing wine machinery and plants, or furnishing production services to the such as logistic, packaging or bottling9. Trait d’union in all these cases is the continuation of highly specialised handicraft activities and industrial ones in a territorial context and the qualification of the district and services to firms. As to the cluster concept, very few documents, almost all very recent, deal with clusters. Moreover most of them use the concept as a synonymous of other concepts, sometimes contrasting, eventually confirming our hypothesis on the weakness of the current (too vague) use of this concept. In the presentation of the initiative Torino Wireless, for example, it is used in its meaning of technological district bringing together the most relevant ICT players in the Piedmont area, in order to increase the competitiveness of the territory through the integration of research and development, entrepreneurship and venture capital. The cluster activities are in fact coordinated by the Torino Wireless Foundation, a leading structure that is committed to emphasizing the resources and knowledge assets already present in the area, by developing the contributions of each player and promoting their integration within medium-to long-term strategies. At the same time, clusters are also presented by recent policy initiatives as new productive systems, that are to substitute definitively districts as addressees of industrial funding programmes (Ferrando, 2006). The new territorial systems are thus thought as homogeneous territorial entities, development poles with flexible delimitations, enabling networking between local actors. In this sense – but not every one agrees with this definition –, they are closer to the concept of cluster or filière, rather than to that of district. Despite the crisis of districts, the weight of the perceptions and the cultural references developed around them is still strong and it reflects both on interpretative schemes and on local initiatives and policies. From one side they are still used in every description of the Piedmont economy. From the other side it is often applied in the meaning of the technological district model, i.e. a particular type of district where firms are specialised in high-tech activities such as aerospace (the reference here is to the newly born Turin aerospace district) or ICT (Torino wireless). Characteristics of these districts are the metropolitan location (they all involve Torino or, at least, municipalities of its province). Most important actors of the Piedmont technological districts are university departments or laboratories, research centres and public-private institutions rather than firms. The concepts of local supply chain and filière, on the contrary, are based on the identification of relationships among firms favoured by spatial proximity. In the case of the local supply chain these are mainly localised in a supra-metropolitan area; while in the case of the filière they are likely to embrace a wider area, in any case contained in the regional bounds. In the case of Piedmont, in particular, the concept of local supply chain has been developed to describe the system of multilayered suppliers (mainly small metal mechanic and plastic producers spread throughout the region) depending on Fiat’s production job orders. In a way, the concept of filière as it is now used in the literature describing Piedmont’s economy, can be thus interpreted as the “next stage” of the concept of local supply chain. In particular respect to that the filière is used according to a wider set of productions (not just automotive) and municipalities (not just the Province of Turin). Moreover the type of relationships developing among firms is different: highly hierarchical and rigid in the case of the supply chain, based on an 9 There are also districts that are recently born, as demonstrated for example by the recent new district of wood products and furniture in Saluzzo. 15 rather equal basis in the case of the filière. This hypothesis of an evolution of the term seems to be confirmed by the case of some districts that are more and more described as filières. But it might also be the result of a superficial marketing operation. 16 Table 1 – The conceptualisation of territorial industrial systems in Piedmont ( ) " ( ./ ! ! "# $% &' ! )*+, ' - , 4 ( ( , + 2/ ! 1 )*+, 1 ! -3+ ! " -%& 2% ' ! ! 3( 5 0 5 ,"" + + + # ( " ! ! )*+, 6 ! 3 4 %22& + ! -3 $ 8 3 3 + 7 + : ! 9 : 7 3 1 )*+, ' $ - 0 6 , %222 3 1 17 + 6 )*+, ' 7 ! ( 6 ; # 3 6 5 9 ! 3 6 # 3 5 3 8 5 7 5 ! , ( <9 3 4 * 8 6 ? ! ! * A 9 ! 3 + 7 ! 5 = > 6 @ 6 + %22. 3 %2& A B )*+, : 4 ! 5 3 3 " )*+, $ )*+, - + 6 8 6 C 3 + 0 # 7 18 6. Concluding remarks In this final section some of the most interesting issues raised by the ongoing research are summarised and commented with respect to the present and future agenda of industrial policy in Piedmont. The considerations presented concern in particular: different models of networked industrial systems and their application; the use of the cluster concept by researchers, practitioners and policy makers; potentialities ad limits in the use of the cluster concept in economic geography and policy making. As to the models of networked industrial systems (see paragraph 3), the analysis of the case of Piedmont allows for the identification of a heterogeneous framework. In the interpretation of the regional economy, besides the well-known industrial districts and local supply chains, some other theoretical concepts emerge, such as “comprensori”, Local Labour Systems, concentration/specialisation areas, technological districts. Moreover some concepts, even if assumed by the international debate, are adapted to the specificities of the local system. For example, the local supply chain model in Piedmont is strongly related to the experience of Fiat automotive industry, i.e. to the presence of one big corporation, from which depends a huge system of small productive units. Just in recent years this concept has been applied to other contexts and industries. In this path goes for example the tendency towards the identification of the wine industry in southern Piedmont with the concept of filière, i.e. the idea of a unique system of relations joining together different actors participating to the same production process in a relatively small area. Another interesting result is that clusters find hardly application in Piedmont. Even when geographic industrial concentrations as those described by Porter actually occur, the tendency is to define them according to different models: filières or, in alternative, industrial districts (in the case of concentrated presence of specialised SMEs in traditional, above all agricultural, fields), local supply chain (in the case of automotive industry) and, technological industry (in the case of high tech industries such as Torino Wireless and the aerospace Turin district). In our perspective this is probably due to some structural conditions as well as to cultural and institutional aspects. There is little doubt, for example, that the long term tradition of district development influenced the way industrialisation is interpreted. For many years, especially when the crisis of Fordism begun, industrial districts of SMEs have been seen as the successful Italian way to industrialisation, something to be proud of in the international context. Moreover, in Italy, as well as in Piedmont, they were institutionalised at the national level so as being a district implied an easier access to public funding. We thus perceive a sort of discomfort in using the cluster concept, which first of all depends on the reluctance to define with the term “cluster” systems that have been for a long period called in another way. Moreover, this also deals with accepting a concept that was originally developed to describe development models far away from those of the Piedmont case. The earliest theorisations on the concept and, above all, the first initiatives of cluster building have in fact been quite almost wholly based on USA and Northern Europe experiences, where very different political, economic and institutional conditions existed. In the Piedmont model, for example, there is little room to entrepreneurship, MNCs, network logics, internationalisation. Rather, the industrial geography of the region demonstrates a more intimate relationship with territory. What probably lacked was the capacity of Italian scholars to access the international debate proposing an “Italian way” to clusters, i.e. focusing it more on such concepts of embeddedness, trust, informal ties, long term 19 tradition that are well-known issues on certain Italian literature. As we mentioned in paragraph 3.1, the localisation of firms in the districts does not only concern with the proximity of their productive units, but also with proximity of actors in the local social network and the long-term processes of ‘social capital’ building. Many hints in the story of Piedmont industrialisation, mainly based on participation, bottom-up initiatives and territorial approach, might not be suitably described via clusters, that sustain the hypothesis for a peculiar development model 10. Summarising, the main point we wanted to propose to the debate was if clusters are just a useful instrument in order to promote local competitiveness, while lacking in operability and territorial sensibility as a geographic discourse. Results from the analysis of the Piedmont case lead us to affirm that they are useful, but they have to be contextualised to the local development model. In other words, we hope to have demonstrated the perils of an incautious and unproblematic use of the cluster model. First of all, because it is a very vague concept and it risks in generating confusion especially when applied in policy agendas. Then, because the specialisation from which the competitiveness depends relies not only in the industry filière, but also (and above all) in the institutional and cultural climate a territory offers to firms and investors. The exercise also emphasizes how cluster is basically a theoretical framework, that needs methodological specification to be effective. As argued by Porter himself cluster is thus a tool to foster the performance of economies. From a methodological point of view, it is thus a way to push the position of an industrial system in the competitive scenario, building best practices, and reducing the complexity of regional economy to easy recognisable territorial contexts and industries. In this sense, cluster-based policies can help to steer development and competitiveness. At the some time different models may help to identify areas that are lagging behind, help them to reduce the gap and found their own development model. This means that clusters are useful tools when they are adopted in a complementary way. Finally – but this is just a personal position of the authors – the concept is already going in its declining period. The operational issues of clusters seems nowadays to be part of old-fashioned and probably out-of-date industrial policy in Italy. More appealing seem to be the filières concept even if analytical methodologies are still lacking, especially when a regional scale is considered. With this specification in mind, it is worthwhile to use the cluster model integrating it in scientific analysis, rather than considering it as the only size that fits all. 10 From the praxis, a very recent hint of this model is the recent decision of the Ferrero Spa, one of the most important Italian industrial groups, to introduce in the contract with its employees numerous and important measures of social care. Is that an example of extraordinary importance, comparable with that of Adriano Olivetti, the Italian entrepreneur who tried in the 60s to build its firm on a communitarian model. 20 References Aimone S., Galizzi G., Pieri S. (1996), “La filiera enologica. Il quadro generale e le specificità del Piemonte”, Working Paper, 116, IRES Piemonte, Torino. Amin A., Thrift N. (2003), Cities: Reimagining the Urban, Polity Press, Cambridge. Autio E. (1998), “Evaluation of RTD in Regional Systems of Innovation”, European Planning Studies, vol.6, n. 2, pp. 131-140. Bagnasco A. (1977), Tre Italie. La problematica territoriale dello sviluppo italiano, Il Mulino, Bologna. Bathelt H., Malmberg A., Maskell P. (2004), “Clusters and Knowledge: Local Buzz, Global Pipelines and the Process of Knowledge Creation”, Progress in Human Geography, vol. 28, n. 1, pp. 31-56. Becattini G. (2000), Il distretto industriale, Rosemberg & Sellier, Torino. Benneworth P., Henry N. (2004), “Where is the Value Added in the Cluster Approach? Hermeneutic Theorising, Economic Geography and Clusters as a Multiperspectival Approach” Urban Studies, vol. 41, n. 5/6, pp. 1011-1023. Boschma, R. (2005), “Proximity and Innovation: A Critical Assessment”, Regional Studies, vol. 39, n. 1, pp. 61-74. Braczyk H., Cooke P., Heidenreich M. (eds) (1998), Regional Innovation Systems, UCL Press, London. Calafati A. (2003), “Socialised Development Trajectories: The Case of Turin”, Scienze Regionali, n. 1. Coleman, J. J. (1988), “Social Capital in the Creation of Human Capital”, American Journal of Sociology, vol. 94, pp. 95-120. Cumbers A., MacKinnon D. (2004), “Introduction: Clusters in Urban and Regional Devlopment”, Urban Studies, vol. 41, n. 5/6, pp. 959–969. European Commission (EC) (2002), “Regional clusters in Europe”, Observatory of European SMEs, 2002, n. 3. Ferrando M. (2006) “Il Piemonte riformula i distretti”, Il Sole24Ore - NordOvest, 19/07/2006. Gordon I., McCann P. (2000), “Industrial Clusters: Complexes, Agglomerations and/or Social Networks?”, Urban Studies, vol. 37, n. 3, pp. 513-532. Grabher G. (ed.) (1993), The Embedded Firm, Routledge, London. Ietri, D., Rota F.S. (2005), “The Italian Mosaic facing Competitiveness and Innovation”, in A. Kukli ski and K. Pawłowski (eds.) Europe – The Strategic Choices, RECiFER – Eurofutures Publication Series (REUPUS), National Louis University, Nowy Sacz, pp. 245-260. Istat (1997), I Sistemi Locali del Lavoro 1991, Istituto Nazionale di Statistica, Roma. Ketels C. (2003), “The Development of the Cluster Concept: Present Experiences and Further Developments”, paper presented to the NRW Conference on Clusters, Duisberg, 2003, http://www.isc.hbs.edu/ Markusen A. (1996), “Sticky Places in Slippery Space: A Typology of Industrial Districts,” Economic Geography, vol.72, n.3, pp. 293-313. Markusen A.R. (1994), “Sticky Places in Slippery Space: The Political Economy of Postwar Fast-Growth Regions”, Working Paper No. 79, Center for Urban Policy Research, New Brunswick, New Jersey. Martin R., Sunley P. (2003), “Deconstructing Clusters: Chaotic Concept or Policy Panacea?”, Journal of Economic Geography, vol. 3, n. 1, pp. 5-35. 21 Marshall A. (1920), Principles of Economics, Macmillan and Co., London (first edition: 1890). Maskell P., Lorenzen M. (2004), “The Cluster ad Market Organisation”, Urban Studies, Vol. 41, n. 5/6, pp. 991-1009. Maskell P. (2001) Towards a Knowledge-based Theory of the Geographic Cluster, Industrial and Corporate Change, vol. 10, n. 4, pp. 921–943. Moulaert F., Sekia F. (2003), “Territorial Innovation Models: A Critical Survey”, Regional Studies, vol. 37, n. 3, pp. 289-302. Newlands D. (2003), “Competition and Cooperation in Industrial Clusters: The Implications for Public Policy”, European Planning Studies, vol. 11, n. 5, pp. 521532. OECD (1999), Boosting Innovation: The Cluster Approach, OECD Proceedings OECD (2001), Regulatory reform in network industries: past experiences and current issues, OECD Economic Outlook n. 67, Oxford University Press. Pavitt K. (1987), “Sectoral Patterns of Technical Change: Towards a Taxonomy and a Theory”, Research Policy, vol. 13, pp. 343-373. Piccaluga A. (2003), “Il distretto tecnologico: lo strumento, le potenzialità, le esperienze”, proceedings of the Congress La ricerca nelle strategie di sviluppo del Mezzogiorno, Rome, 13th November 2003. Porter (1990), The Competitive Advantage of Nations, Free Press, New York. Porter M.E. (1998), On Competition, Harvard Business School Press, Harvard. Putnam R., Leonardi R., Ninetti R. (1993), Making Democracy Work: Civic Traditions in Modern Italy, Princeton University Press, Princeton (NJ). Saxenian A. (1994), Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Harvard University Press, Cambridge (MA). Steiner (Ed.) (1998), Clusters and Regional Specialisation. On Geography, Technology and Networks, Pion, London. Williamson O.E. (1985), “The Economic Institutions of Capitalism. Firms, Markets, Relational Contracting”, Macmillan, London. Wolfe D., Gertler M.S. (2004), “Clusters from the Inside and Out: Local Dynamics and Global Linkages”, Urban Studies, vol. 41, n. 5-6, pp. 1071-1093. 22