Industrial development in thin regions: trapped in path extension?

Transcription

1 Journal of Economic Geography 15 (2015) pp Advance Access Published on 9 July 2014 doi: /jeg/lbu026 Industrial development in thin regions: trapped in path extension? Arne Isaksen y Department of Work Life and Innovation, University of Agder, Jon Lilletunsvei 9, 4879 Grimstad, Norway y Corresponding author: Arne Isaksen, University of Agder, P. O. Box 422, 4604 Kristiansand, Norway. 5arne.isaksen@uia.no4 Abstract Recent theorizing of path dependence supplements the traditional view of regional path-dependent industrial development characterized by lock-in effects with paths dealing with change, that is, path renewal and path creation. Few studies, however, examine why different types of regions experience diverse path-dependent development. This article examines why organizationally thin regions are much less likely to achieve path renewal and path creation than core regions. By use of a case study of industrial development in an organizationally thin and rather peripheral region in Norway the article contends that thin regions often need external investments to avoid being trapped in path extension. Keywords: Evolutionary economic geography, path dependence, regional development, thin regions, peripheral regions JEL classifications: R11 Date submitted: 2 January 2013 Date accepted: 19 June Introduction The question of how regional industries evolve over time is increasingly discussed by scholars. Leading economic geographers argue that long-term regional industrial development demonstrates strong path dependencies. A core argument is that new regional growth paths do not start from scratch but are strongly rooted in the historical economic structure of a region (Neffke et al., 2011, 261). Maskell and Malmberg (2007, 603) phrase in a similar way that a geographical location...can be thought of as having a memory that directs the path of subsequent development. This perspective on regional path-dependent development corresponds with the view that regional industrial development and innovation activity depend much on place-specific factors (of different types). Martin (2010, 20), for example, maintains that innovation is indeed often a highly localized phenomenon, dependent on place-specific factors and conditions. This article focuses on characteristics of path-dependent industrial development in thin regions. The term thin regions corresponds with regional innovation systems characterized by organizational thinness in the conceptualization by To dtling and Trippl (2005). Such regions, that often but not always are peripherally located, have low levels of firms clustering and a weak endowment of knowledge generation and diffusion organizations. The article argues that thin regions experience other typical path-dependent development than core regions. Core regions often have a variety of clusters, firms ß The Author (2014). Published by Oxford University Press. All rights reserved. For Permissions, please journals.permissions@oup.com

2 586. Isaksen R&Dorganizations and a collection of research and higher education institutions (To dtling and Trippl, 2005). Growth in core regions is triggered by a combination of increasing return effects and agglomeration economy, such as the development of a local pool of specialized labour, local inter-firm division of labour, local supporting institutions, local knowledge spillovers and various forms of traded and untraded interdependencies (Storper, 1997). Industrial development in thin regions is based on other pre-conditions and follows most likely different development paths. This fact points to a need for theoretical reflections and empirical investigations of the specifics of industrial development paths in thin regions. Such endeavours are also critical to ensure that industrial policies in thin regions are based on sound knowledge of such regions, and do not routinely build on theories that reflect situations in organizationally thick, core regions. The rest of the article has four main parts. The next part reviews and discusses main theoretical contributions with regard to regional path-dependent development and hypothesizes about the specifics of path dependence in thin regions. The third part describes the case study region (Lister) and the method and empirical data used. Part four interprets the industrial development in Lister in light of the theoretical framework of path-dependent regional development that is constructed in the article. The last section concludes by discussing some general lessons relating to path-dependent development in thin regions. 2. Path- and place-dependent development The idea of path-dependent regional development has been around in economic geography for some decades. Massey (1984, 118), for example, regards local economies as the historical product of the combination of layers of activity. The layers of history consist of economic, cultural, political and ideological strata, and the layers reflect the position of a region s industry within the broader geographical division of labour at different times. This analytical framework interprets the industrial structure, local institutions and specific human resources in a region at one point of time as involving different layers that consist of remnants of former roles that the region has played, and which are critical causes of the creation of new rounds of investment in the region. More recently, Boschma and Frenken (2011a) among others understand the existing industrial structure of a region as determining the extent to which firms are technologically related, which then affects the nature and scope of knowledge spillover among regional firms, and affects which new activities that branch out from existing activities. Regional branching describes how regional industries diversify into new but closely related activities and this approach also maintains that regions follow specific paths of development. The theoretical framework of the article discusses this tradition of focusing on historical and context-specific explanations. It departs from the notion of pathdependent regional industrial development which is characterized by lock-in effects that push a technology, industry or regional economy along one path rather than another. The framework should, however, not be interpreted in a deterministic way. The idea is rather that the past economic development in a region sets the possibilities, while the present controls what possibilities to be explored (Martin and Sunley 2006, 403). The

3 Industrial development in organizationally thin regions. 587 pre-existing industrial and institutional structures constitute the regional environment in which current activities occur and new activities arise. However, in line with Mackinnon et al. (2009), Martin (2010, 2012), Garud et al. (2010) and Martin and Sunley (2011) the traditional path-dependent processes focusing on continuity and stability and restrictive lock-in outcomes are supplemented with paths dealing with change as follows from different forms of reorientation of regional industries. These are, however, still seen as path-dependent processes in the sense that previous events and existing competence and skill affect the probability of future events to occur (Boschma and Frenken, 2011a). In this way regional industrial development is seen as mainly endogenous processes as the local inherited knowledge and skill base of an industry can form the basis of the rise of related new local paths (Martin, 2010, 19). The article does not enter into any comprehensive description and discussion of key concepts and principles in evolutionary economics and evolutionary economic geography [see, e.g. the collection of chapters in Magnusson and Ottosson (2009), and in Boschma and Martin (2010) with regard to the emerging school of evolutionary economic geography). The article focuses on analysing the paths of development that in particular are triggered by specific conditions in organizationally thin regions. A first step in such an endeavour is to distinguish four possible development paths of regional economies in general. The paths may occur simultaneously in the same region, for example, in different sectors, which mean that both continuity and change may take place in a regional economy. The four paths represent however analytical concepts in order to categorize key developments in a region. Discussions of path dependence may start from the notion of path extension, in which increasing returns and positive externalities reinforce local industrial dynamism (Martin and Sunley, 2006, 415). Path extension is characterized by incremental product and process innovations in existing industry and along prevailing technological paths, which in situations of growth can lead to continuity or more of the same in a regional economy. The notion of path extension resembles one of regional resilience, which represents situations in which regional economies continually adjust to changing environmental conditions by the recovering of existing industrial structures (Hassink, 2010a). The basic idea is an extension of established ways of doing things in which informal institutions, that is, socially constructed and collectively shared rules of behaviour, decide much of firms problem solving. A main development path is then the continual adjustment by existing firms in a region. Boschma (2014, 2) extends the concept of regional resilience to the ability of regions to reconfigure their sosio(sic)- economic and institutional structures to develop new growth paths, that is, be able to obtain path renewal and path creation (as is discussed below). Regional resilience depends also in this conceptualization to a large extent on the industrial history of a region. Path extension may in the long run lead to stagnation and gradual decline due to lack of renewal (Hassink, 2010a). Regional industries then face a risk of path exhaustion that refers to situations in which the innovation potential of local firms has been severely reduced or innovations take place along a restricted technological path. A regional industry can be locked into decline if firms have low adaptability with regard to technological and market changes. The literature on regional lock-ins focuses specifically on mature regions with problems such as old, mono-structural industrial regions (Hassink, 2010b). Grabher (1993) conceptualizes the problems as three interrelated types of lock-in: (i) functional lock-in when close and stable linkages

4 588. Isaksen between regional firms reduce boundary-spanning functions; (ii) cognitive lock-in when long-standing personal ties result in groupthink interpretations that determine which phenomena are perceived and which ignored and (iii) political lock-in when symbiotic relations between the politico-administrative system and industry (Grabher, 1993, 254) lead to frequently, reinforced support to already strong regional industries, and indirectly hamper industrial renewal. In such situations firms may be unable or slow to respond to the rise of new competitors and technologies elsewhere (Martin and Sunley, 2006). Firms will then become uncompetitive and decline, so the regional industry shrinks (Martin, 2010, 7). The two remaining paths relate to change rather than continuity. Path renewal takes place when existing local firms and industries switch to different, but related sectors. A main development path is restructuring of existing firms in a region and the establishment of new firms, a process termed regional branching (Boschma and Frenken, 2011b). Path renewal is thus often industry driven as regional industry mutates and widens in terms of industrial specialization and competence. In this case, a number of subsequent incremental innovations may add up to the basis for a new path. A main argument is that industries do not branch into any new industries. Rather, regions are most likely to branch into industries that are technological related to the pre-existing industries in the regions (Neffke et al., 2011, 237). What knowledge and other resources that exist in regional firms shape to some degree the type of renewal that occurs, that is, the type of commodities or services that are produced and the kind of technologies that are used. Technological relatedness then gives rise to pronounced path dependencies in regional diversification processes (Neffke et al., 2011, 238). Path creation denotes the most wide-ranging changes in a regional industry. It includes the establishment of new firms in new sectors for the region or firms that have different variants of products, employ new techniques, organize differently, etc. than what hitherto have dominated in the region. To dtling and Trippl (2013) distinguish two kinds of new industries in a region; first, the rise of established industries that are new for the region (regional path formation in established industries) and second, rise of totally new industries. Path formation may be caused by inward investments and/or sectoral diversification of existing firms through path branching. The second case of new path creation is often research driven focusing on commercialization of research results and grows up through the establishment of new firms and spin-offs. In this case new sectors may not be related to the existing regional industrial base (Henning et al., 2013, 1353). Research driven, new path development is not considered in the regional branching and related variety approaches and the importance of research for the development of new growth paths marks the main difference between path renewal and path creation. Path creation might demand the building of new knowledge organizations and institutional change (To dling and Trippl, 2013). The path-dependent regional industrial development described so far focuses on endogenous factors as regional economies are mainly developed through related diversification of the existing industrial structure or through the combination and reuse of existing competence. However, regional economies develop also as a result of external investments, such as the establishment of new firms by entrepreneurs or corporations from outside the region. Such new firms may come in related as well as unrelated industries. (Neffke et al., 2011, 251) however point to the fact that industries are more likely to enter a region if they are technologically related to the industries that were already present in that region, and Boschma and Iammarino (2009) find that extra-regional knowledge

5 Industrial development in organizationally thin regions. 589 lead to employment growth in Italian regions when the knowledge originates from sectors that are related but not similar to the sectors that are already present in the regions. 3. The specifics of path dependence in thin regions Although much theoretical and empirical studies of path-dependent regional development exist, as yet there have been few systematic accounts of why regions experience different path-dependent development. Why are some regions more capable of path renewal and path creations than others? What regional characteristics may stimulate path renewal and creation? To what extent may organizationally thin regions achieve path renewal and creation? The evolutionary approaches should be well equipped to tackle such questions as it takes history and geography seriously by recognizing the importance of place-specific elements and processes (Hassink, 2010a, 48), and many of the basic mechanisms that give rise to path dependence have a quintessentially local dimension in their form and operation (Martin and Sunley, 2006, 409). Path-dependent processes are thus locally contingent, that is, affected by specific regional capabilities (Fornahl et al., 2012). The still lack of awareness of specific regional characteristics that contribute to different path-dependent processes may reflect a tendency in evolutionary economic geography to emphasize micro-level, firm-based mechanisms and processes (Martin, 2012, 181). This approach results in interests in how firm-specific routines, norms and tacit knowledge lead to path-dependent development. Evolutionary economic geography has hitherto put less focus on collective, systemic, structural and (macro) social and political forces (Martin, 2012, 182). The evolutionary approach that focuses much on the firm level then needs to be extended by an institutional approach that focuses on elements in the regional and wider environments that influence the innovation capability of firms. Martin (2012) also advocates to linking the micro- and macro-level as, for example, macro effects like network and knowledge spillover that are deemed to create regional path dependence arise from the behaviour of micro-level units such as firms and workers. The article discusses characteristics of path-dependent industrial development in thin regions, related to both institutional factors [following the regional innovation system approach, e.g. Asheim and Isaksen (2002)] and firm-specific factors. Thin regional innovation systems have, by definition (in To dtling and Trippl, 2005), few or none higher education institutions and R&D institutes, none or only weakly developed clusters and consequently little local knowledge exchange. The firm structure in thin regions often includes relatively many small and medium sized enterprises (SMEs), and also larger, externally owned firms (To dtling and Trippl, 2005). This structure may in some cases lead to a branchplant culture where locale actors envisage new jobs coming from external investors, which hampers local entrepreneurship and innovativeness (Petrov, 2011). Due to weak innovation systems are SMEs in thin regions often characterized by the DUI (doing, using, interacting) mode of innovation (Jensen et al., 2007; Isaksen and Karlsen, 2012). DUI innovations are typically based on experiences and competences acquired on the job as employees face new problems or new customer demands. This way of innovating leads mainly to incremental changes in products and processes.

6 590. Isaksen Table 1. Typical characteristics of organizationally thin regions Institutional (RIS) characteristics Micro-level (firm) characteristics Typical path-dependent industrial development Few knowledge organizations. Clusters missing or weak. Little local knowledge exchange. Dominated by SMEs. Some large, externally owned firms. DUI mode of innovation. Path extension and exhaustion. Path renewal, if inflow of knowledge and/or investments. The typical innovation deficits in thin regions and the importance of the DUI mode of innovation indicate that such regions often experience path extension and risk falling into negative lock-in and path exhaustion. Thin regions are on the other hand assumed to face problems in renewal of existing and, in particular, in creation of new development paths. Path renewal is triggered by a diverse industrial structure and a variety of firms in a region (Frenken et al., 2007), which may increase local knowledge exchange. Path renewal is thus more likely to occur in well-developed than in thin regions due to the fact that well-developed regions have more diverse industrial structures and larger and more diverse knowledge infrastructures. However, firms also use knowledge sources and find innovation partners outside a region. This demands absorptive capacity of a regional industry (Cohen and Levinthal, 1990; Giuliani and Bell, 2005); that is the ability of at least some firms to identify and acquire external information, interpret and assimilate it, combine it with existing knowledge, share it among members of the firm and other regional actors and then apply it to commercial ends. Following such reasoning, thin regions may achieve path renewal first of all by making use of resources outside the region, which requires some local organizations with boundary spanning functions (Table 1). Path creation is seen to be even more difficult to achieve in thin regions. Path creation includes commercializing new knowledge for a region. Parts of the knowledge are often scientific as it may be difficult to create entirely new use of experience-based knowledge as such knowledge is often embedded in local communities of practice with fixed norms and rules of behaviour (Gertler, 2007). As much scientific knowledge travels fairly easily across geographical distance, the knowledge to be commercialized can, in principle, come from sources outside the region. However, new knowledge will most often be developed and commercialized in the same region because regional actors then will gain early access to the new knowledge and be in contact with persons who have been involved in developing the knowledge. Simmie (2012) emphasizes that new paths are initiated by reflexive, knowledgeable agents but that the agency must be understood in the historical context in which it arises. Path-dependent processes are a result of the interaction between embedded agents and the structure in which they are embedded (Simmie, 2012, 760).This means that agents employ their obtained experience and competence from former jobs and education in their new activity (Garud et al., 2010). Some regions produce more entrepreneurs with the necessary skill and network to initiate entirely new activity that potentially can start a new development path than

7 Industrial development in organizationally thin regions. 591 other regions. This will often be regions with several R&D institutes, higher education institutions and firms or clusters in new and dynamic industries. These are characteristics not found in organizationally thin regions, which leads to the conclusion that path creation seldom occurs in this type of regions. Petrov (2011, 171) draws a similar conclusion in his statement that peripheries normally lack human capital necessary to complete a regional breakthrough. The above arguments point out clearly that thin regions are much less likely to achieve path renewal and in particular path creation than core regions with organizationally thick regional innovation systems. This is explained, as in Table 1, with little local knowledge exchange and a dominance of experience-based learning and innovation activity in firms in thin regions. According to To dtling and Trippl (2005) such situations can be met by policies to strengthen potential clusters in the regions. Such policies should combine endogenous and exogenous elements. Due to the weak or non-existing regional innovation systems the policy recommendations include to attract innovative firms and branches of national research institutions or research centres from outside, and also link firms to partners and knowledge sources outside and inside the region. The last suggestion leads to a need to strengthening the absorptive capacity of regional firms. The strengthening of the innovation capabilities of existing firms and the support of new firm formation is also put forward as policy recommendations in thin regions by To dtling and Trippl (2005). The article now discusses how theoretical assumptions as summarized in Table 1 stand up when subjected to empirical analyses of industrial development processes in the organizationally thin region of Lister in Norway. The subsequent empirical analysis of this region is organized around three main research questions. The descriptive question is what path dependence that characterizes industrial development processes in Lister. This is followed by an explanatory question: what factors, and in particular what institutional and firm-level factors, have led to the specific path-dependent development in Lister? Finally, a theoretical question revolves around what lessons can be drawn with regard to industrial development paths in thin regions more generally from the Lister case. 4. Context and method The Lister region is located in between the cities of Stavanger and Kristiansand in the south-western part of Norway. The distances from the centre of Lister to these cities are 140 and 90 km, respectively. Lister includes two small towns with 510,000 inhabitants in each and otherwise more rural areas. The population increased slowly from 33,000 in 1980 to 36,000 in 2014, that is, with 9% which is much less than Norway in total with population growth of 25% in the same period. Lister has no university, university college or R&D institute and fits well for designation as an organizationally thin region (To dtling and Trippl, 2005). The empirical study focuses on firms in three industries; the process industry, mechanical engineering and technical consulting. The first two of these with about 1700 employees are the relatively most important industries in Lister with location quotients between 2 and 3. These industries were selected for the study exactly due to their relative importance in the region, which made them candidates to possible industrial paths. They amount to nearly 70% of all manufacturing jobs in Lister. The process industry

8 592. Isaksen Table 2. Key information of interviewed firms Type of firms Number of firms Aggregate number of employees 2012 Location of owners Year of establishment Process firms All international 1971, 1974, 1996 Mechanical engineering firms One national, five local 1842, 1860, 1942, 1967, 1969, 2011 Mechanical subcontractors One national, four local 1902, 1969, 1975, 1978 Technical engineering firms 3 20 One national, two local 2006, 2008, 2010 Sum includes mainly two smelters and a supplier of aluminium components to the European car industry. Mechanical engineering contains a ship yard, 5 6 firms with own products and the same number of contract suppliers for (mainly) the Norwegian oil and gas industry. Technical consulting is a small industry in Lister, but with recent job growth. This industry was selected because of its rapid growth and the possibility that the growth may partly stem from outsourcing of activities and recombination of resources from existing firms in Lister. It turned out that many technical consulting firms are in the construction industry and have not emerged from other local manufacturing firms, thus the study focuses mostly on the process industry and the mechanical engineering industry. The main empirical data source is interviews by CEOs, research directors and in some cases other employees in 16 firms in Lister. Some of the interviews were focus group interviews so that, in total, we have 26 interviewees. The focus group interviews were recorded and transcribed; in the other cases minutes were prepared just after the interviews. A compilation and analysis of information from the interviews were sent to the interviewees for a quality check which raised some further comments. The interviews focused mostly on the firms history, their organization of innovation and learning processes, their external innovation partners and knowledge flow, possible spin-off activities and use of input factors. The 16 firms include all firms with 450 employees and a selection of smaller firms. The 16 firms represent most of the jobs in the three industries selected. The three process firms all have international owners from the USA (Alcoa), France (Eramet) and Germany (Benteler Automotive) (Table 2). The other firms are mostly locally owned. Most mechanical engineering firms and mechanical subcontractors have a long history in the region, and the oldest one is actually owned by the fifth generation after the entrepreneur. The only newly established firm here is a spin-off from one of the traditional mechanical engineering firms. The technical engineering firms are, on the other hand new and small. Table 1 includes the central firms in the key industrial sectors in Lister, and it illustrates the possibilities of identifying industrial paths in a thin region. Industrial paths are characterized by the persistence of regional industrial or institutional structures, and that economic agents continue their behaviour under changing external conditions (Henning et al., 2013, 1352). Table 1 demonstrates that it might be

9 Industrial development in organizationally thin regions. 593 manageable to identify industrial paths in thin regions because of few possible paths and a limited number of firms. The study also draws on interviews by the author of the article in 1993 in four of the large mechanical engineering firms and in one firm that have been shut down since then [which is reported in Amdam et al. (1995)]. In addition, the analysis employs various reports and statistical sources to complement and cross-check information obtained through interviews Path dependence in Lister From path creation to path extension. The current process industry and most of the mechanical subcontractors in Lister are the result of industrial development in Lister from the end of the 1960s. The development in Lister is the local expression of what Wicken (2009) regards as the most important part creation process in Norwegian industry during the 20th century, namely large-scale industrialization linked to the exploitation of Norway s natural resources. A huge hydroelectric power development centred on Lister contributed strongly to the establishment of two energy intensive, large-scale process plants by external investors. An aluminium plant was built in 1971 in a joint venture between the Norwegian company Elkem and the US-based Alcoa (now wholly owned by Alcoa) and a producer of mangan alloy was established by the Norwegian Tinfoss Jernverk company in 1974 (now part of the French corporation Eramet). The aluminium company also triggered the establishment of a plant established by international companies (now Farsund Aluminum Casting owned by the German corporation Benteler Automotive), among them Alcoa, in 1996 that produces casted aluminium parts for the European automotive industry. This plant is located just beside the aluminium smelters, and technology and expertise as well as flowing aluminium were (and are) transported from the smelters to the automotive part producer. A dominant development path in Lister centres on these three process firms and their further effects on the development of mechanical subcontractors, as demonstrated below. The path creation was shaped by the development of a natural resource, that is, hydroelectric power, which should be understood in a national institutional context. The development of comparatively inexpensive hydroelectric power and simultaneous development of electrometallurgical processing were an important part of Norwegian industrial policy after 1945 in order to modernize the economy (Sæther et al., 2011). This also includes the building of R&D activity at research instititutes and study programmes in public universities which coevolved with increased research capacity in firms (Moen, 2009). The building of a public research infrastructure contributes to Norwegian metal producers being among the most cost efficient and environmentally friendly producers globally (Moen, 2009). Farsund Aluminium Casting (FAC) was set up at Lister because Alcoa and Volvo would establish a new automotive part factory in Europe. Lister became the preferred location site in competition with other places due to the supply of pure aluminium and skilled workers at the neighbouring aluminium smelter, the possibility of flowing aluminium directly into the foundry line at FAC, but also because of high expertise in aluminium in Norway, particularly at the Norwegian University of Science and Technology (NTNU) and the research institute SINTEF, both found in Trondheim.

10 594. Isaksen The R&D director at FAC right from the start is a PhD in engineering from NTNU and has been a researcher at SINTEF. FAC has continued research cooperation with SINTEF, often in projects partly financed by the Research council of Norway. Favourable local conditions, such as physical infrastructure and overall good knowledge of English at the work force (because the area has a very high proportion of returning Norwegian Americans) also benefited Farsund as a location site. The establishment of FAC points to the fact that the source of path dependence in Lister is less and less the availability of hydroelectric power but rather institutional and knowledge-based capabilities such as high competence in the labour forces and the research capacity in the sectoral innovation system of electrometallurgical processing. The path dependence is, however, also found in sunk costs of local assets and investments, that is, the capital equipment and physical infrastructure. The establishment and growth of the process plants also spurred the growth of some mechanical subcontractors at Lister. One firm was asked to establish near the aluminium plant to produce, repair and maintain production equipment when the plant was established in Another mechanical subcontractor was established as early as 1902, but expanded through contracts to above all the aluminium plant from the 1970s. The CEO and owner of this family business maintains that the aluminium plant built the firm. The process firms were demanding customers for several mechanical subcontractors. These have the last 1 2 decades mainly turned to the Norwegian oil and gas industry as their main customers. The growth of mechanical subcontractors in the first place reveals to some extent a path interdependent development in Lister as parts of the mechanical engineering industry coevolved with the process industry (Martin and Sunley, 2006). A coherent industrial path consists of technological related firms and industries (Neffke et al., 2011). Firms are often seen as technological related if they belong to the same or nearby code in the standard industrial classification system, if they have input output relations, or have similarities in mixes of occupations (Neffke et al., 2011). In a qualitative study like the one in Lister, relatedness and industrial paths can also be based on information on the exchange of technological knowledge among firms that contributes to cognitive proximity. The Lister study revealed several examples of knowledge flow, such as lean production principles. Ten mechanical engineering firms in Lister have formed a local network (Lister Alliance) for joint marketing and to carry out projects that require cooperation of several firms. This type of locally generated cooperation on production can also be seen as an indicator of relatedness and a coherent industrial path. After the start-up of the two process firms in the 1970s and their ripple effects, the industry in Lister, also the traditional mechanical engineering industry predating the 1970s, has mainly continued to further develop their current specialization by building on existing products, services and expertise. Thus, path extension has been the dominant development path within the three industrial sectors studied in Lister. The firms carry out both product and process innovations, but the larger firms focus on efficiency improvements to lower production costs. Some attempts to take forward more radical innovations are found. This can be illustrated by a carbon thermal pilot plant at Alcoa Lista (the aluminium smelter) that aims to develop a completely new method of aluminium production, also supported by public funds. If this long-term development succeeds path renewal would occur, but path extension is currently dominating.

11 Industrial development in organizationally thin regions. 595 The path extension development in Lister is seen in few spin-offs from existing firms, which is illustrated by the many relatively old firms in Table 2. At all, Lister has relatively few new establishments. From 2001 to 2012, Lister got 3.8 new establishments per 100 existing firms in average each year compared with 9.6 new establishments nationwide. (Source: Processing of statistics from Statistics Norway). The figures include all types of start-ups and are thus rough estimates. However, the pattern is evident and consistent with observations from the case study: Lister creates relatively few new establishments, which consequently indicates less potential for path renewal and path creation. A few examples of new firms that develop from existing activities are still found in Lister. As already noticed FAC started in 1996 partly because the aluminium plant was located at the same place. A more recent example is a small firm (12 employees) which produces biofuel plants and is a merger in 2011 of the skill and products from two other firms. Another example is the small technical engineering firm Enerquip which spun-off from a mechanical subcontractor in The subcontractor used its long experience as a supplier to the oil and gas industry to develop its own product that won a competitive bidding for a public customer. The product and further product development were transferred to Enerquip, whereas the parent firm continues as a contract supplier. The two other technical engineering firms in Table 1 are also fairly young. Both have developed completely new products and services with growth potentials. The firms and the entrepreneurs are, however, not linked to existing industry and competence in Lister. The two firms could have been established anywhere and would most likely had better opportunities to recruit skilled labour elsewhere in Norway. The entrepreneurs are from, or have lived, in Lister for a long time, which explain the establishments here Institutional and firm-level explanation of path extension The few spin-offs and start-ups, and the path extension in general, can be explained by specific institutional (RIS) and firm-level characteristics, in line with Table 1. Particular important reasons for path extension are little local knowledge exchange and dominance of the DUI mode of innovation. Little local knowledge exchange reflects the fact that Lister firms are largely self-contained with competence; the firms have most of what they need for production and innovation activities internally. A limited market then exists for new suppliers and service firms to the existing industry. The smelter Eramet, for example, keeps a vehicle repair shop at the firm because an independent repair shop very likely would find few, if any, other customers locally. Relatively little extent of local innovation networks is also demonstrated in the R&D and innovation survey carried out by Statistics Norway (and which is the parallel to Eurostat s Community Innovation Survey). The survey demonstrates that firms in Lister with external innovation cooperation in the period had significantly less cooperation with regional partners than the average for Norway: 31% of firms in Lister had regional innovation cooperation compared with 67% nationwide. Contributing to relatively little local networking and high self-sufficiency is also the fact that mechanical engineering firms and subcontractors in Lister often compete through flexibility and fast deliveries, part of which is achieved by having key skills available internally in a thin region. In regions with far larger industrial activity, firms can find much more relevant partners and more competence nearby. As it is now Lister

12 596. Isaksen firms do not contribute much, if anything, to development of related competence in the rest of the regional industry. The rather self-contained firms reflect the fact that Lister is relatively peripheral, has thin industrial and knowledge milieus and a small labour market. Lister firms report of a fairly stable workforce; a fact that is confirmed by data on labour mobility by Statistics Norway. The turnover of employees is lower in Lister than in Norway in general, both in manufacturing firms and in industry in total. The CEO of the mechanical engineering firm AMV (with 150 employees) claims that the firm could hardly survive in Kristiansand due to faster turnover in the work force there, as a result of more competition from better paid jobs in the oil sector, which would mean that accumulated expertise more readily disappears from the firm. Low turnover means that workers may achieve high experience-based competence to be employed in innovation processes. This leads to another main explanation of the tendency to path extension in Lister, which is the importance of the DUI innovation mode. Innovation activity then builds mainly on employees experience and tacit knowledge where the employees typically know little or nothing of the origin of the knowledge or how they have come to know it, but its here and it works (Maskell and Malmberg 2007, 605). The DUI innovation mode then mostly contribute to incremental innovations, and it also leads entrepreneurs to set up firms that tend to reinforce the local patterns of specialization (Maskell and Malmberg, 2007). The DUI innovation mode is seen in the fact that many Lister firms employ experience and skills in all parts of their organization in innovation processes, among other things through dense cooperation between engineers, skilled craftsmen and service people. An example is the firm AMV that develops and produces tunnel and mining equipment for a global market. The skills of both engineers and production workers are used in innovation processes, as seen in the following statement from the CEO: If a drawing includes errors, an engineer can be sure that within two minutes a worker will come and tell that this does not work. A corresponding example is found in Alcoa Lista. The USbased Alcoa corporation typically develops innovations in separate R&D units which are then spread to and adopted by the plants, in accordance with the linear innovation model. Alcoa Lista employs to a greater extent the competence in all parts of the plant. Operators are, for example, named in patents that have been developed by Alcoa Lista, which is almost unheard of in more hierarchical oriented plants elsewhere in Alcoa. Both examples demonstrate how experience-based skills of the workforce contribute to frequent improvements which underpin path extension. The examples reflect organizational traits such as decentralization of responsibilities and decision making and initiatives on the shop floor, which incorporate learning and improvements in daily work. This way of working resembles the so-called Norwegian or Nordic model because it reflects characteristics of large parts of Norwegian and Nordic work life (Gustavsen, 2011). A characteristic of Lister firms that can enhance tendencies of path extension, but also contribute to path renewal, is considerable extent of extra-regional networks. Many firms are part of, benefit from and contribute in strong Norwegian clusters and innovation system within the offshore oil and gas, maritime and metal industry. Aeron, for example, that produces heating, ventilation and air conditioning (HVAC) systems for ships cooperates closely with Norwegian marine consultants and designers when developing prototypes of new ships (especially offshore vessels), and uses also the research Institute SINTEF for testing and Det Norske Veritas (DNV) for approval of products. The

13 Industrial development in organizationally thin regions. 597 Eramet plant in Lister has set up an R&D unit with three persons in Trondheim together with two other Eramet mangan alloy plants in Norway. The unit is linked to the R&D activity on ferroalloys at NTNU and SINTEF, which is among the foremost in the world in this field. Such links embed firms in Lister by linking them to specific competence in industrial clusters and knowledge organizations in Norway. This may on the one hand contribute to further specialization and path extension, but may on the other hand bring in supplementary competence that gives potentials for path renewal. 5. Conclusion This article raised three research questions. The first two of these involve what path dependence that characterizes industrial development processes in Lister and what factors that have led to the specific development in this region. The current dominant manufacturing industry in Lister departs to a large extent from a path creation process, more precisely regional path formation in established industries, initiated by the establishment of two large-scale process plants in the region in the 1970s. After these investments and their ripple effects, path extension dominates within the sectors in question in Lister. The path creation process was initiated by external investments, inflow of external knowledge and technology and in line with Government industrial policy to exploit a local natural resource, that is, hydroelectrical power. Path extension is based on specific regional and national factors. The regional factors are, in particular, little local knowledge exchange and dominance of the DUI mode of innovation. Both factors inhibit creative linking of different types of local expertise which is important for the sake of path renewal. The national level is crucial because firms are incorporated in sectorial innovation systems centred on Norway, which supports firms existing specialization. Our final, theoretical issue treats what more general lessons that can be drawn with regard to industrial development paths in organizationally thin regions. Three such general lessons are drawn based on the empirical investigations in Lister. The first is that thin regions have some drawbacks when it comes to path renewal and path creation based on regional resources, which is in line with the analytical framework in Table 1. Thin regions have, by definition few resources to redeploy existing industries into new paths of development and to create entrepreneurs with the competence to start off new paths. Lister firms have dealt with the drawbacks of being located in a thin region by building up deep and varied expertise inside firms and through links to national and international customers, experts and R&D organizations, which have contributed to some world class firms. Developing crucial competence internally for production purpose and incremental improvements is probably a typical strategy by firms to cope with locational conditions in thin regions. A second and related lesson is that thin regions like Lister need external investments to avoid being trapped in path extension. In Lister national state institutions played a vital role in developing the hydroelectrical power potential and in establishing the three process firms. Path renewal and creation in thin industrial milieus can hardly build entirely on scarce regional resources but demand inflow of technologies and knowledge. Martin and Sunley (2006) term this mechanism as transplantation, which refers to the importation and diffusion of a new industry or technology from elsewhere. Isaksen and Trippl (2014) demonstrate through theoretical reasoning and empirical illustrations that transplantation and further regional industrial development can be of two basically

14 598. Isaksen different types. It can be based on the establishment of research and higher education institutions or on inflow of companies. Both types can spark the formation of new industrial path under certain conditions, for example, when the institutions and the companies operate within relatively new technologies and when supportive institutional structures are built. A transplantation strategy is in line with the innovation policy approach for organizationally thin regional innovation systems put forward by To dtling and Trippl (2005). They advocate attracting innovative firms and knowledge organizations from outside, and to link firms to external knowledge sources. These proposals are meant to help strengthening potential regional clusters, which in particular will contribute to path extension. That is important; however, the cluster approach is less relevant as a policy guide in thin regions. As shown in the Lister case, thin regions may have few players and resources to build clusters with extensive local networks. This is consistent with the conclusion by Lagendijk and Lorentzen (2007) about regional innovation policy in non-core areas. Such policy should not attempting to nurture local and regional knowledge networks (Lagendijk and Lorentzen, 2007, 465). The policy should rather foster local entrepreneurship and learning in the context of global knowledge networks (Lagendijk and Lorentzen, 2007, 465). New knowledge coming from the inward transplantation of firms is particularly critical for initiating new local paths of development in thin regions with less endogenous dynamism. A third lesson may be that much existing theorizing on industrial development processes is scarcely adapted to thin regions like Lister. It departs mainly from situations in typical core regions with large potentials for endogenous development. Martin and Sunley (2006, 424) thus claim that the attention of geographers has been almost exclusively...on those regions that seem to be the loci of indigenous innovation and technological development. The empirical basis in core regions is seen in how research agendas are put, for example, it would shed light on...the way regions create new variety (like new sectors) and how they transform and restructure their economies over time (Boschma and Frenken, 2011b, 69; emphasis added), or the view that one of the most intriguing questions...is why some regional economies manage to renew themselves (Hassink, 2010a, 45; emphasis added). Theorizing on industrial development in thin regions demands more attention on exogenous processes for such regions to achieve more than path extension. As pointed out above, this is not a new insight as it has been demonstrated that analyses of adaptability in thin regions (and in old industrial areas as well) in particular must consider the institutional context at the regional as well as the national and supranational levels. Nevertheless, such insight is more or less absent from theorizing of the regional path-dependent processes in evolutionary economic geography which highlights endogenous growth processes. Funding The Research Council of Norway (Project numbers , ). References Amdam, R., Isaksen, A., Olsen, G. M. (1995) Regionalpolitikk og bygdeutvikling. Drøfting av lokale tiltaksstrategiar (Regional politics and rural development. Discussion of local development strategies). Oslo: Samlaget.