Industrial districts/clusters and smart specialisation policies

The literature on clusters and industrial districts has been growing at an unprecedented pace in the last two decades. While the origin of the notion of industrial district is older and can be attributed to the important work of Marshall (1920), the term “cluster” was introduced by Porter (1990; 1998) in the 1990s, to characterize the emergence in space (clustering) of specific types of specialized agglomerations, where specialized firms and institutions co-evolve and interact (Belussi, 1996). A better general theoretical understanding of the elements representing the constituency of the “model” was developed by numerous contributions at the intersection between economic geography and management studies (Becattini, 1990; Saxenian, 1994; Prouder and St John, 1996; Asheim, 1996; Markusen, 1996; Gordon and McCann, 2000; Belussi, 2006; Maskell and Kebir, 2006; Asheim et al., 2011). Numerous authors also focused their attention on the granularity of the concept, articulating their analysis on various aspects of industrial districts and clusters, studying the growth-factors linked to the elements which form this specific pattern of local development (Becattini et al., 2009): a) the presence of external economies or externalities (Breschi and Lissoni, 2001), b) the process of knowledge creation and diffusion (Belussi, and Gottardi, 2000, Belussi and Pilotti, 2002), c) new firm entry and start-ups (Baptista and Swann, 1998; Stuart and Sorenson 2003; Feldman and Braunerhjelm, 2006), d) learning and capability formation (Amin and Wilkinson, 1999), e) skills transmission and labor market specialization (Sorenson and Audia, 2000), and f) the emergence of indigenous specialized suppliers (Hervas-Oliver, et al., 2017). Another important issue discussed in the literature concerns the evolution of clusters during time. Belussi and Sedita (2009) have adopted the perspective of multiple path dependencies, based on an empirical analysis of Italian cases. The authors highlight that clusters may share some commonalities as regards the factors that underpin their emergence and take-off, but subsequently they give rise to a variety of developments, depending on knowledge variety, innovation intensity, local firm leadership, and external conditions. Other theoretical contributions (Martin and Sunley, 2011; Ter Wal and Boschma, 2011) have suggested the existence of more deterministic cluster trajectories (allowing only a possible adaption) across different stages over time (with time as an irreversible factor) such as emergence, growth, maturity, decline or renewal (for a review see Bergman, 2007). Thus, cluster specialization leads to higher synergies among firms but too much similarity bears the so-called “cluster paradox”: the risks of decreasing returns, uniformity, drop of innovativeness, and at the end, lock-in (Martin and Sunley, 2006; Menzel and Fornahl, 2010; Audretsch and Feldman, 1996). A broad distinction can be made between industry-driven explanations of cluster growth (Ter Wal and Boschma, 2011) and place-based explanations. The former explains the emergence of the clusters as deriving from knowledge discontinuity and the introduction of breakthrough innovations. During the first stage of experimentation and when knowledge is not much codified but grows in a cumulative way, agent proximity and spin-offs create favorable business conditions. Thus, one can observe high levels of industry concentration in clusters. In the maturity phase other firms are created at a global scale in dispersed places, and clusters lose their shape. This picture is clearly significant in the case of high-tech sectors (Menzel and Fornahl, 2010). The latter reflect a cluster-specific view and suggest that clusters can grow or decline independently of the development of the industry, for reasons such homogeneity or heterogeneity of competencies, cluster-specific technological lock-in, institutional or external factors (Belussi and Sedita, 2009; Trippl et al., 2015). Brenner and Schlump (2011) have observed that the transition between stages may be a remarkably slow process. But also the idea of “pre-determined” stages of development has been questioned (Belussi and Caloffi, 2017, ch. ?, this volume); some clusters decline before reaching the stage of maturity or they never follow a high grow-path as observed in some Turkish footwear clusters in Konya and Izmir (Belussi and Caloffi, 2017). Other clusters operating in the same sector but located in other countries like Italy, have reached their maturity in the 1980s, but they have been able to stay at the technological frontier of their sector for a long time (see the case of Montebelluna described in Belussi, 2010). Many regions host clusters (but not necessarily industrial districts, which represent a specific form of clustering), but not all clusters produce high growth. Indeed, if a region has a cluster consisting of industries the demand for whose products is low and/or declining or whose production processes rely on low-skilled labor or a too expensive labor force, the contribution to regional economic growth is likely to be small, no matter what other institutions or specific policies are directed to supporting the cluster. After an initial grow, historically, many clusters and industrial districts relentlessly declined (Belussi and Sedita, 2009; Belussi and Caldari, 2011) Tappi (2005) has shown how complex the process of cluster evolution might be. The author describes the development of a cluster that shifted from musical instruments to the production of ICT components by slowly absorbing microelectronic technologies. Clusters can disappear and then re-emerge, exploiting favourable market conditions, historically accumulated technological capabilities and specialized institutions (see, for example, the case of Swiss watches, Glasmeier, 1991). A paradigmatic case of a cluster that has not undergone decline is of course Silicon Valley, moving its specialization from computers to ICT components, and to social network platforms such as Google and Facebook (Saxenian, 1994; Weil, 2012). Thus, transition does not necessarily imply crisis. Historical accidents are certainly influential to explain the location of some clusters. The Danish mobile-communications cluster emerged recently and has already adjusted numerous times during its relatively young life, a process accelerated by the profoundly rapid pace of innovation in mobile technologies. Dalum, et al (2005, p. 231) highlight the role played by disruptive technologies creating sequential disruptions in the cluster life-cycle. An accelerated scenario of early entrants, enabling institutions and universities, buyouts, mergers, takeovers and exits reflect the highly unstable state of the cluster in North Jutland, which successfully shifted in mobile telephone technologies from NMT to GSM to UMTS, but that entered in a deep crisis recently, when WLAN technologies were developed by large Silicon valley firms (Østergaard and Park, 2015). Important actors for the development of clusters are nowadays leading firms or the investment of MNEs, scientific institutions “feeding” local firms with scientific and technological knowledge, and the creativity of local entrepreneurs. The openness of clusters and industrial districts for FDI inflows and outflows, global supply chains, and the building of external linkages appears as a necessary (Trippl,et al, 2015) but not sufficient condition for successful cluster consolidation and resilience (Becattini and Rullani, 1996; Bair and Gereffi, 2001; Dicken, 2003; Nachum and Keeble, 2003; Guerrieri and Pietrobelli, 2004; Wolfe and Gertler, 2004; Nadvi and Halder, 2005; De Propris and Driffield, 2006; Zucchella, 2006; Belussi, 2015; Boschma, 2015). As discussed by Trippl et al. (2015 p. 2036) there are good reasons to argue that cluster development might by affected by the configuration of RIS (regional innovations systems). For instance, RISs that are home to dynamic high-tech clusters may offer a fertile ground for the rise of new (but related) ones (Boschma and Frenken 2011). Policy-makers may play an important role in supporting cluster development. However, cluster policies have suffered from the creationist myth (Borras and Tsagdis, 2008). Without having essential preconditions in place (e.g. potentials related to high technological dynamics) clusters can hardly be created. As demonstrated by Boschma (2007) and Rodriguez-Rodriguez et al. (2017), institutions typically follow cluster emergence. Often the most effective policies are found in well-developed or mature clusters and less in emerging ones (Tödtling and Trippl, 2013). Policies may be addressed to increase the quality of local resources (supporting vocational training, research, and the provision of collective goods) or to overcome bottlenecks and too high levels of path-dependency. In fact, the development of a broad and comprehensive understanding of cluster evolution still constitutes an emerging topic in evolutionary economic geography and other related disciplines (management, innovation and technological change, etc.). There still does not exist a comprehensive theoretical framework nor ample empirical evidence capable of fully explaining why and how clusters and industrial districts evolve and change over time. One of the reasons for this is that most empirical work on clusters has provided a “static” rather than a more dynamic, longitudinal picture. Also, possibly, another reason lies in the complexity involved in integrating the diverse set of intellectual disciplines required for building a comprehensive theoretical framework capable of addressing all actors and micro-processes involved in the functioning of clusters. The managerial perspective can be useful (e.g. Pouder and St. John, 1996; Wang, Madhok and Li, 2014) to the understanding of the micro/meso unit of analysis to see how cluster firms’ capabilities and strategies can recombine existing and new knowledge from inside and outside the territory. Hence, cluster firms are more prone to cross-fertilize knowledge and technologies between different fields. In this chain of thought, the literature shows that cluster firms are heterogeneous but if they possess different but complementary competences they can sustain cluster evolution. However, the increase of absorptive capacity also plays a role: technologically weak clusters may then absorb new knowledge through building knowledge connections to high-tech or innovative regions. European policies have sustained this process by better connecting weak and strong regions in Europe, integrating them in various R&D cooperative programs. Another important issue relates to globalisation processes of industrial districts and clusters and their relationships with European policies. Cluster dynamics and globalization should be taken into account when designing new policies (Hervas-Oliver and Albors-Garrigos, 2014; Wang, Madhok and Li, 2013; Crespo, 2011). Currently, the openness of territories and their connection to global value chains ask for a novel approach to local development. The impact of globalisation on cluster evolution is occurring not only in terms of flows of exports, but also in relation to a more complex interchange of inward and outward flows of goods, people, knowledge, which often involve MNEs as crucial players in local nodes of global supply chains. FDI by MNEs increasingly takes the form of knowledge-seeking investment, whereby MNEs attempts to augment their knowledge base through obtaining access to foreign pools of knowledge by becoming a participant in various clusters simultaneously. Indeed, being co-localised where new knowledge/technologies/designs are generated is a more effective way to absorb these assets, in comparison with inter-country cross-border transferring. Clusters that have historically developed a high level of capabilities are nowadays among the preferred destinations of MNEs. Specific European policies are required in some cases to counterbalance the excessive power of MNEs in industrial districts and clusters, favouring a more ample access and acquisition to strategic recourses by local firms and SME, in order to contrast the emergence of too “oligopolistic” local systems. “Indigeneous” or “homegrown” MNEs are a novelty in the modern evolution of clusters: they were created during 1990s or 2000s, when small firms invested strategic resources in innovation and expansion, progressively transforming themselves into MNEs (Sedita, Caloffi, and Belussi, 2013; Aznar-Sanchez et al., 2017). Policies can be orchestrated to better embed those firms into their context, providing support for re-shoring processes, and supporting the creation of ancillary service sectors (knowledge-intensive business services, universities, research centres, key enabling actors, and so on). Very importantly, in some circumstances, MNEs are the main actors responsible for giving rise to local clusters, while in others they enter (or emerge in) the local cluster in a subsequent phase of the life cycle (development or maturity). Those MNE-dominated clusters may be particularly fragile, and policies favouring diversification should be promoted.