4.1 Introduction

In the last few decades a combination of factors, such as the rapid diffusion of reliable high-speed communication networks, reduced trade barriers, freer flows of capital and reduced transportation costs, amongst others, have enlarged the geographic scope of economic activity and market competition, deeply changing the outlook of the world economy ^[1].

While a growing number of developing lower-wage countries became significant players in international trade and investment, stronger growth rates in these countries is also creating vast new consumer markets and therefore opening new possibilities for investment in the opposite direction, i.e. from higher-wage to lower-wage economies.

In addition, the wider use of low cost ICT and web technologies, in both developing and more developed countries, has increased the possibilities for dismantling large vertically-integrated operations into networks of business units located in different countries – forming the so called global value chains (GVCs) – each supplying and producing components, to be combined by yet another business unit, into final products.

Outsourcing and offshoring of activities has been growing, especially in manufacturing industries where product and production processes’ modularity is high ^[2], but also in service sectors such as retailers and branded merchandisers with little or no internal production ^{[3] [4] [5]}.

The development of these so-called GVCs is, however, often associated with the decline of employment in manufacturing sectors located in the developed world. In fact, following a trend to diminish, the weight of manufacturing in most developed OECD countries represents today less than 20% of GDP.

But while these trade and investment linkages between developed and developing countries are not new, their scale and complexity has substantially increased, triggering much debate on the characteristics of this new environment for international economics and on the role of regional innovation policies in promoting local employment, growth and competitiveness.

On the one hand we have those who see the world being ‘flat’ ^[6] – or the world without borders ^[7]. The argument of a flat world means that distance does not matter. You can access production and services capacity anywhere in the world. Also, it is not just production capacity at relatively lower wage costs that you can access worldwide, but also research and development (R&D) and technology. Increasingly we find R&D delocalization and growing investments in global R&D and innovation ^[8].

In line with international product life-cycle theories ^[9] and standard trade theories of comparative advantage, this line of argument sees the decline of manufacturing as a natural and healthy evolution. The sociologist Daniel Bell in 1973 ^[10], has even proposed that the decline in manufacturing in the more advanced economies is a natural and healthy transition to a ‘post-industrial’ society dominated by ‘knowledge workers’.

Delocalization of production to lower wage regions is taken as a healthy symptom of economic development, ‘liberating’ resources so that they can be used in high-value-added sectors, such as services. Moreover, local production is dispensable if you want to move forward towards a knowledge-based economy focused on services and innovation. To enter so-called ‘post-industrial society’ or a globally networked society supported by global high-speed digital networks ^[11], regions should exclude direct manufacturing and rely exclusively on being a hub for high-added-value activities such as R&D, design, testing, or perhaps concentrate on the development of mobile and web applications, or even more recently to focus on promoting the so-called ‘creative industries’.

On the other hand, we have those that suggest that distance matters and therefore you cannot just access any kind of static or dynamic capabilities anywhere in the world. The argument is that distance matters for the creation of localized learning linkages between R&D, design, testing and manufacturing. Proponents of this view argue that building infrastructure, improving educational performance, and strengthening cooperation between public and private institutions is better undertaken at the local level.

Following on from this debate, this paper explores the impacts of these global trends on regional innovation systems and on regional policies for innovation. Our concern is with local responses to globalization through strengthening ‘local capabilities for innovation’. By ‘capabilities for innovation’, we mean the ability to conceive, develop, and/or produce and commercialize new products and services. In particular we focus on the contribution that local learning linkages between manufacturing and industrial design and product development make to the construction of regional capabilities, serving the regional innovation ecosystem.

In Section 4.2 we explore in more detail the characteristics of distributed global value chains. Then, in Section 4.3, we will look at the literature on the influence of the close environment in business innovation and competitiveness. This will lead us to the need to reframe the industrializations debate in Section 4.4. Finally in Section 4.5 we attempt to draw some conclusions for a regional and multilevel innovation policy agenda.

4.2 A world of ‘distributed’ capabilities in global value chains

As hinted above deverticalization of ‘manufacturers’ that have shed internal capacity and have come to rely on an emergent set of global and East Asian contract manufacturers for production appears to be a natural trend of developed economies ^{[5], [6], [7], [8], [9], [10], [11], [12]}.

‘Traditional’ vertically integrated companies may today choose to be ‘networked companies’, i.e. they may choose to focus and specialize on certain stages or functions of the GVC, while outsourcing extensively to suppliers and subcontractors located in the region which does it best at the lower cost.

The well-known case of the global smartphones industry is illustrative of how in many industries the GVC became truly global. However, although many of these subcontractors are now able to supply complex parts and subsystems to anywhere in world, so far they capture a relatively small portion of the value chain. For example, outsourcing the assembly of the iPhone 4 represents less than 2% of total value, and purchase of materials and components (out of the US) represent around 30% ^[13].

Stan Shih, chairman of the Taiwan-based Acer Inc, is perhaps the first to have coined the term ‘smiling curve’ – a U-shaped value distribution across the value chain. In the global smartphones industry, as in personal computers and many other industries, value added is higher in R&D, design and branding, and lower in the midstream stages, which involve labor-intensive processes such as assembly. Because such processes capture only a minor proportion of value added, it is assumed they can and should be outsourced to lower wage regions.

We used to think about local manufacturing industries as important because of the so-called multiplier effects. Bonvillian ^[14] estimated that each manufacturing job creates between 2.5 and 2.9 jobs in other sectors and that manufacturing also operates as an output multiplier. Also, not just with mass standardized products, but with adequate use of new technologies and industrial management strategies, other modes of production, namely mass customization or flexible manufacturing, can also be scalable. Finally, manufacturing contributes directly to productivity gains and because of this it is very unlikely that it would sustain large-scale job growth as in the past.

Although the smiling curve clearly overlooks multiplier effects and local linkages, and highlight the ‘comparative advantage’ of delocalization, reality may however be more complex and difficult to interpret.

Suzanne Berger ^[15] and her team at MIT, studied the actual experiences of 500 companies in North America, Asia, and Europe as they responded to globalization. The study included firms in slow-tech industries like textile and apparel, where technologies and processes evolve incrementally, and fast-tech industries like electronics, where radical disruptive innovations may often cause profound industrial changes. At least two important conclusions emerged from this study.

First, Berger argues that there is no single best model, i.e. there are many different ways to win in the global economy and opportunities are much wider than usually imagined. In some cases successful companies that were once vertically integrated choose to focus on some specialization, while forming different kinds of worldwide partnerships and networks. In other cases, while remaining vertically integrated they find that selective delocalization brings some kind of competitive advantage.

For example while Samsung, a vertically integrated electronics company, chooses to make almost everything in-house, Dell, an American computer company, focuses its own organization on distribution and outsources all the manufacturing and assembly overseas. In the textile and apparel industries, Gap and Liz Claiborne choose to outsource production to foreign countries, but the fastest-growing retailer is the Inditex Group, a Spanish company that makes more than half its clothing in the Galicia region.

Apparently, the technological and organizational changes of the past decades that made modular production possible across a wide range of industries, combined with rising production capabilities in lower wage developing countries contributed to a multiplication of successful models. It is not the simple geographic spread of economic activities across national boundaries, but the functional integration of internationally dispersed activities that is the key to global competitiveness in these different models ^[16].

The picture captured by Berger ^[15] is quite different from that of a flat world where any and every industry can access production, design or development capabilities anywhere. Instead, this new environment created the opportunity for vertically integrated companies relatively concentrated in one region to coexist with network-integrated companies or groups of companies spread worldwide ^[17].

Second, Berger ^[15] argues that the strength of the international companies examined is ‘grounded’ in their local environment. For US companies the key aspects appear to be the establishment of local connections between R&D and end users, strong linkages to venture capital and flexible labor markets. A few years later, in the MIT PIE research study, Berger (^[18], p.14) reinforced the importance of ‘regionally based resources’ such as training, collaborations between firms and universities, suppliers, industrial and technical research centers, and so on. The density, diversity, and abundance of such local resources are key characteristics of successful regional innovation systems.

Another important aspect of the increasing fragmentation of global value chains is that delocalization does not only affect manufacturing. Increasingly it also affects R&D, design, initial tests for proof of concept, and so on. For example, international flows of qualified scientists and researchers and cross-border co-operation in science, technology and innovation, are on the rise, as illustrated by indicators such as expenditures with R&D abroad and co-authorship of scientific publications and patents. There is also higher competition between regions that want to attract the major international scientific centers and often, particularly in smaller regions, the presence of a multinational affiliate may account for a high proportion of local R&D activities.

To summarize, globalization, the rapid spread of new technologies and rising capabilities in developing countries are contributing to an increase in modularization of production and to open innovation processes. There is no single best model of how to compete in this new global environment. However, easier access to global manufacturing capacities and global explicit/codified knowledge does necessarily mean that regionally resident capabilities don’t play an important role. Because implicit/tacit knowledge does not flow so easily across borders and needs to be nurtured at local level and used to search, absorb and complement access to capabilities located elsewhere, the local environment still plays an important role.

4.3 Local environment and regional competitiveness

There have been a number of studies analyzing the significance of the local environment in the competitive strengths of firms. In the 1920s Alfred Marshall was the first to point out the link between geographical agglomeration and the incidence of external economies, in what he called the ‘industrial atmosphere’. Marshall argued that agglomeration of economic activities in a given region created a pool of workers with specialized skills and facilitated the development of specialized inputs and services. Later, geographers such as Perroux ^[19] also argued that higher-growth propulsive industries would affect other industries via backwards and forwards linkages occurring within a particular geographic space.

Following the Marshallian tradition, other authors in the field of regional economics also put forward a similar argument, proposing the notion of ‘industrial districts’. For example Pyke, Becattini and Sengenberger ^[20], point out the importance of local sharing of common values and beliefs, the sense of belonging and the role that institutions such as family, schools, church, local authorities, local union organisations, and so on, play in the diffusion of such localized common values. According to Michael Storper ^[21] the ‘system of social regulation’, including relations of trust, institutional coordination and mechanisms for social consensus, play an important role in coordinating collaboration between enterprises, in fostering the dynamics of local entrepreneurial activity, in the reproduction of labor practices, or more generally in the local dynamics of social reproduction.

Similarly, the concept of ‘innovative milieu’ put forward by the GREMI group (Groupement de Recherche Européen sur les Milieux Innovateurs) ^{[22] [23] [24]} stresses the importance of agglomeration of economic and social activities for ‘collective learning’ and ‘uncertainty reduction’. According to Camagni (^[24], p.3) an innovative milieu may be defined as “the set or the complex network of mainly informal, social relationships on a limited geographical area, which enhances the local innovative capability through synergetic and collective learning processes”. Finally there is also some evidence ^[25] that geographic proximity between industry and universities facilitates knowledge spillovers from university research to private firms and may lead to higher innovation rates in terms of patenting and R&D expenditure.

In short, a number of studies suggest that the environment close to the firms (physical, economic, social) may be an important factor contributing to their ability to compete in a new global world. This is not just because of the reduction of physical distance and associated transport and location costs, but mainly because it facilitates information exchange, lowers uncertainty, increases the frequency of interpersonal contacts, facilitates trust and the diffusion of common values and beliefs, and promotes collective localized learning. Rosalind Williams ^[26] argues that “knowledge is global, but learning is local”. Local environment conditions play an important role because they foster the so-called nontradable socio-institutional input factors (regional capabilities) that are specific, unique and relatively immobile.

Some production input factors are, however, more mobile than others. For example, capital is much more mobile than land, human capital, or labor. Immobile factors are more important because they cannot be easily transferred and reproduced elsewhere and they represent the regions’ unique resource capabilities. This is why companies located in some places have advantages over others by virtue of the dynamics of local learning based upon the appropriate set of local unique resources, such as workers, engineers, managerial talent, suppliers, universities, and so on.

For example, the family-owned firm STIHL based near Stuttgart in Germany, though it could shift production to its lower-wage factories in China and Brazil, often prefers to maintain manufacturing of its most advanced products at home. Similarly, in Portuguese clusters such as moulds, textiles and so on, leading firms often maintain their base manufacturing at home where they can benefit from relatively well trained workers and conditions for achieving better quality.

Another example is the Shan Zhai region in China ^[27]. The term ‘Shan Zhai’ today refers to businesses based on fake or pirated products. However, many businesses with Shan Zhai origins are now becoming innovative disrupters and, in many cases, market leaders. Some of them are fast, flexible, innovative, and willing to take risks. As they achieve a certain scale, they quickly move up the value chain and develop core competencies to differentiate themselves from other imitators. The Shan Zhai phenomenon is not about low-cost fake products; it is about how the local environment enables some Chinese companies to achieve global success.

To further illustrate the importance of the local environment, Pisano and Shih (^[2], p.3) propose to extend the notion of ‘agricultural commons’ – in existence in England up to the first industrial revolution – to the notion of ‘industrial commons’. While ‘agriculture commons’ relate to the land and resources shared by a community of farmers, industrial commons are:

These technical and organizational capabilities are embedded in local networks of people (workers) and organizations i.e. competitors, suppliers, customers, cooperative R&D, universities, and so on, and they flow across organizations through movements of people from one organization to another, through supplier-customer collaborations, formal and informal technology sharing and imitation of competitors ^[2]. According to Berger (^[18], p.14):

In our view these processes of learning and knowledge accumulation may follow different trajectories. Jensen et al. ^[28] propose an interesting dichotomy that contrasts the STI-mode and the DUI-mode of knowledge accumulation.

The Science–Technology Innovation mode (STI-mode) might be referred to as the classic, top-down, internal, research and innovation (R&I) model first practiced in large corporate laboratories, transformed into an externalized model of university laboratory research adapted to technological innovation through ‘academic entrepreneurship’. It is the source of start-up and spin-out small and medium enterprises (SMEs) in high-tech clusters, sometimes characterized in terms of a ‘patenting –seed/angel/venture fund – incubator’ model of new business growth.

This contrasts markedly with the Doing–Using Interacting (DUI-mode) approach to innovation. This is not immediate exploitation of laboratory bench knowledge, although some such knowledge may lie behind the existent state-of-the-art technology or even contribute to its furtherance. DUI involves knowledge recombination among diverse knowledge and practice sets. Accordingly it is fundamentally interactive among firms and/or intermediaries characterized by ‘related variety’ in the first instance. However, research shows that such is the potential of Schumpeterian knowledge recombination that many innovations integrate very different firms/sectors or institutional knowledge-sets.

Accordingly, DUI is diversified in that it thrives on cross-fertilisation or cross-pollination of ideas and practices from different fields, for example the intelligent textiles for stay-clean car seats that inspired the innovation of bacteria-free medical uniforms. This means that DUI is inclusive for firms that have the needed information about a shared innovation possibility, provided demonstration effort is made. The entailed knowledge for DUI is thus implicit rather than codified, and regional/Local rather than globally available.

4.4 Reframing the industrialization debate

While in the previous section we argued that the creation and use of unique immobile nontradable socio-institutional local factors associated with different kinds of localized knowledge trajectories and learning is very important for regional competitiveness, in this section we frame the role that local manufacturing capability, particularly in the DUI-mode (but also in STI mode with new possibilities for micro manufacturing), may play in the construction and sustainability of these relatively unique local knowledge resources.

It is not, as often argued ^[14] that manufacturing is important because of employment creation, or because of multiplier and scalability effects which impact on productivity. The presence of local manufacturing enables the development of specific localized learning linkages between production, R&D and design, hence contributing to the success of innovation. Both Pisano and Shih ^[2] and Berger ^[18] argue that manufacturing is one key local capability of any innovation ecosystem, and as such cannot be delocalized without creating ‘capability holes’.

Many other studies undertaken in the US such as ^{[29] [30] [31]}, amongst others, also point out the importance of retaining and counteracting the delocalization of manufacturing activities. Apparently, according do Pisano and Shih ^[2], if a region loses its capacity to manufacture it may also eventually lose its capacity to innovate. Local manufacturing works like an ‘anchor’ for local learning, dissemination of technology and good management practices.

However, although manufacturing and innovation share the same ‘commons’, the crucial linkages between product design, prototyping, testing and production that fuel regional innovation, do not work in the same way across all industries. Amongst other things, it depends on the product design choice of modularity. Because of the widespread use of new design and production technologies enabling integration of product and processes, the choice of modular product architectures is today much wider.

The innovation management literature sees modularization as being related to product architecture i.e. components and subsystems as modules and how they connect to each other ^[32]. However, we also have to look to the rapidly increasing industrial capabilities in developing countries if we want to find out why and how companies combine the choice of product architecture, with strategies to delocalize manufacturing and access remote capabilities ^[33].

In addition, we need to look beneath the surface of physical components and sub-products. Behind each product or component, we find core competencies grounded in technical and organizational capabilities that enabled their creation, production, and delivery. In addition, these underlying networks of capabilities are dynamic ^[34] i.e. new capabilities emerge to change the possibilities of both products and processes.

One interesting framework to analyze the extent to which manufacturing is tightly linked to the local innovation system, i.e. whether it does or does not play a key role in learning and innovation is to look at various degrees of modularity and process maturity. Pisano and Shih ^[2] propose to use the matrix in Figure 4.1

Fig. 4.1. Product modularity vs process maturity.

If a product falls into the upper-left quadrant (process-embedded innovation), design tends to be less modular, i.e. product architecture is integrated, and therefore the commons between product design and process manufacturing are much higher. In this situation one cannot easily disentangle R&D and design from production, and therefore the presence of local production capabilities is vital for the regional innovation system. Craft industries, or high-end wine brands (such as Porto and others), specialty apparels (e.g. fire resistant, swimming, etc.) are examples of industries where product modularity is relatively low, manufacturing processes are quite developed and mature and therefore design cannot be separated from manufacturing.

If a product falls into the lower-left quadrant (process-driven innovation) the situation is more or less similar. In sectors such as biotechnology drugs and nanomaterials, product architecture is essentially integrated but manufacturing processes may not have yet reached maturity in order to be safely and efficiently transferred elsewhere. As in the previous case this innovation requires proximity between product design and manufacturing.

When the product architecture is essentially integrated, i.e. with low modularity, the argument that a region can focus on high value services of design and or R&D and let others do the innovation translational work of testing, preseries and ramp-up manufacturing may lead to loss of both. Possibly the region will first lose manufacturing operations where processes are more mature, therefore easier to transfer, but in time it may also lose less developed or less mature manufacturing operations.

In the upper-right quadrant (pure product innovation) products are highly modular and manufacturing processes fully mature. In industries such as consumer electronics or active pharmaceutical ingredients amongst others, it is very difficult to retain manufacturing operations, since they can be easily transferred to cheap labor locations. For regions with relative comparative advantage in labor costs, it may make sense to capture some share of international outsource investment in these areas, perhaps looking for a near shore strategy at a relatively close location.

Finally, if a product falls into the lower-right quadrant (pure process technology), product architecture is highly modular but process maturity is low, i.e. manufacturing processes are poorly developed though process technology is evolving rapidly. In these sectors such as advanced semiconductors colocation of product R&D and manufacturing may not be essential, and therefore it makes no sense to compete with regions that may offer lower costs and technologically advanced processes.

As we can see from Figure 4.1, higher modularity may offer multiple opportunities, but, on the other hand, it may also have some negative effects. First it encourages fragmentation of production and second it enhances the possibilities for outsourcing and offshoring design and R&D, since it discourages wide-ranging research, i.e. research on every system and component.

4.5 New challenges for a regional innovation policy agenda

Although everyone today agrees that European regions need higher rates of growth and jobs creation, there is little agreement on what regional and multilevel innovation policies should be implemented. In a world of fragmented GVCs, where firms (not just manufacturing firms, but also banks, law firms, accountants, retailers, and others) may have little or no particular interest or commitment to the development of their local community, what can regional policies do to strengthen local capabilities for innovation?

One typical reaction of European regional policy to this new world of distributed industrial and service capabilities presented earlier appears to be entirely based on a post-industrial view of the world ^[10]. Because services as a percentage of total economic activity (GDP) tend to increase as economies become more developed, loss of manufacturing is deemed unproblematic and automatically driven by the ‘forces of globalisation’.

Also, given the extension across national borders of today’s value chains, local linkages and interdependences among domestic firms may be less relevant than extra-regional /national linkages of global scope ^[35] and therefore the relevance of the resources under the control of public authorities in a given territory would tend to diminish. Regional or national industrial innovation policies are therefore often seen as irrelevant, losing adherence and legitimacy not only in the eyes of business owners but also in the eyes of the citizens in general. In this perspective regional policy should only focus on value-added services, design, creative industries, knowledge intensive start-ups and on spin-offs of academic research. It may also consider facilitating the attraction of hubs and decision centers of large multinational enterprise networks.

Second, another typical policy approach is to argue for more of the same. In European policy documents ^[36] we find the belief that industrial policy is needed because it creates jobs and because industry creates and retains upstream and downstream interactions with other sectors. This view leads to reinforcement of more of the same ‘traditional’ policies, aimed at enhancing existing industries or industrial clusters, hence overlooking the degree to which modularity in these industries may eventually lead to delocalization.

We believe all these reactions are partial and often distorted by a misconception of the role of local learning and innovation in international regional competitiveness. As seen in previous sections this new global and more interdependent world is having profound consequences for management of international production and operations in GVCs, for open innovation and for regional innovation policies in a multilevel context. We need to rethink innovation policies in order to better reflect this change and to provide governments with policy tools that effectively help regions to create and protect their capabilities while at the same time being connected to global science networks and to global markets.

Nevertheless, we cannot assume that localized collective learning and construction of regional capabilities is automatically driven just by colocation or agglomeration of complementary or related economic activities. That is, we cannot assume that appropriation of the so-called Marshallian external economies is deemed unproblematic when carried out in physical closeness. Antonelli and Quéré ^[37], amongst others, argue that it is not sufficient that technological externalities are “freely available in the air” for effective technological communication to take place. Regional innovation policies should therefore focus on stimulating and capturing the benefits arising from local collective learning, innovation and entrepreneurship.

Innovation policies should first focus on creating the base conditions for the development of ‘commons’. This is best undertaken at national or European level. Just as the so-called ‘smart specialisation strategies’ ^[38] argue that at the European-level policy should concentrate on multipurpose technologies, we favor the idea that for ‘commons’ to flourish at the European and national levels we need to build a massive portfolio of technological competences. This can only be done by decades of continuous public investment in basic and applied research. The idea is not that this investment leads directly to an increase in product or services design and innovation, but that it leads to the formation of human capital.

In every region, every university and public and semi-public research institutes should seek to participate in large European challenge-driven science projects around general multipurpose science and technology fields in order to create or improve local conditions for ‘commons’. Also at European and national levels policies should enlarge their scope beyond science and technology, to include education, information society, health, transports, agriculture, and so on.

Second, at the regional level innovation policies should focus on creating capabilities for translation and transformation of these multipurpose knowledge competencies into innovative products, services or business models through support to testing, pilot productions, speeding up ramp-up operations, testing of production in international operations, and so on.

As we argued earlier, manufacturing and local production plays a vital role in creating and sustaining capabilities for undertaking such translational work from knowledge into goods, services or business models, since it provides key learning linkages between science and innovation. Local governments should therefore be clear about their targets. They are not targeting specific sectors, industries or clusters. Their target is creation of capabilities to transfer knowledge into prototyping, pilot production, customer development and testing.

However it also follows from our earlier discussion based on Pisano and Shih that local policies should only focus on two types of manufacturing capabilities. Those pertaining to immature, or newly emerging, process technologies and those in contexts in which product architecture is less modular and manufacturing-process innovation is therefore highly integrated with product design and R&D. In both cases, regions will find that their competitive advantage in a global world lies in creating the capabilities grounded on the need to have R&D, design and production processes geographically close.

Bibliography