Embracing the future: path transformation and system reconfiguration for self-driving cars in West Sweden

ABSTRACT

The past years have witnessed a surge of academic interest into how new industrial paths are developed in regions. Transformation processes of existing regional industries have received less attention. This article focuses on radical innovation-based renewal processes of established paths and investigates how regional innovation systems are tackling challenges related to path transformation. Drawing on insights from the regional and technological innovation systems literatures, we develop an analytical framework that elucidates the relation between path transformation and system reconfiguration. The framework suggests that regional innovation system elements are created or adapted to (i) target the build-up of system functions regionally; (ii) link up to system functions in other locations, and (iii) transplant system functions from elsewhere. The analytical framework is applied to a case study of the transformation of the automotive industry in West Sweden towards self-driving cars. The empirical analysis provides support for the importance of the three types of system reconfiguration and emphasises the relevance of different types of assets. Furthermore, it highlights how actors tend to utilise previous networks and positions in global innovation systems rather than turning to the development of system functions regionally as the ‘default option’ of system reconfiguration.

KEYWORDS:

• Path transformation
• system reconfiguration
• self-driving cars
• automotive industry

Introduction

Regions across the world are to an ever-increasing extent facing the pressure to adapt to a wide range of challenges related the rise of new competitors, accelerated technological development and fundamentally new market conditions. Dealing with these challenges is instrumental to maintain regional competitiveness, jobs and economic growth.

The past years have witnessed an enormous academic interest in regional industrial path development. Drawing on Evolutionary Economic Geography (EEG) models of regional structural change, many studies have sought to explain how new industries emerge in space through path creation, diversification or importation processes (Boschma, 2017; Dawley, 2014; Isaksen & Trippl, 2017; Martin & Sunley, 2006). Renewal processes of traditional industries have received less attention in recent conceptual and empirical work.

This article employs the notion of ‘path transformation’ to describe processes underpinning radical innovation-based renewal of established paths. Mature industries are often supported by a well-established regional innovation system (RIS), implying that system reconfiguration is a crucial part of path transformation. The aim of this article is to contribute to the path development debate in economic geography by investigating both conceptually and empirically how RISs are tackling challenges related to path transformation processes. We combine insights from two strands of research, that is, the RIS and the Technological Innovation Systems (TIS) literatures, which both have elaborated on the link between innovation system changes and transformation processes of industries and technologies. We develop an analytical framework for understanding RIS reconfiguration. This framework builds on the idea that it is a set of system functions (Binz, Truffer, & Coenen, 2016) that all need to be performed for successful innovation-based path creation and transformation. It directs attention to the creation or adaptation of RIS elements for (1) building-up system functions regionally, (2) linking up to system functions in other locations, and (3) transplanting system functions from elsewhere to the region.

The analytical framework is used to investigate path transformation based on the development of Self-Driving Cars (SDCs) in the region of West Sweden. West Sweden is the ‘heart’ of the Swedish automotive industry and hosts firms from all parts of the value chain, supported by a strong RIS with thick organizational and institutional structures. Both its traditional well-established automotive path and the wider RIS structures are currently undergoing major changes due to the advent of SDCs, making this region an interesting case to study the link between radical innovation-based path renewal and system reconfiguration. We address the following research question: What types of system reconfiguration facilitate system functions catering to new activities around SDCs in West Sweden and which local and non-local actors play a key role in shaping such change processes?

The remainder of the article is organized as follows. Section 2 provides a review the current literature on path development and develops an analytical framework for investigating system reconfiguration for path transformation, combining insights from RIS and TIS studies (section 2). This is followed by a methodological discussion and the presentation of the empirical case (section 3). Section 4 contains the empirical analysis. Finally, section 5 discusses the results and concludes.

RIS reconfiguration and path transformation: towards an analytical framework

Economic geographers have long embraced the idea that regional economies are developing in a path dependent manner. Early work focused on forces of continuity and sought to reveal what factors contribute to the persistence of patterns of economic activity and how regions become ‘locked in’ to historically grown trajectories (Grabher, 1993; Hassink, 2005; Martin, 1999). Less attention was initially given to processes through which such trajectories could be ‘de-locked’. However, recent contributions have situated questions about how new industrial paths emerge and why some places are better positioned than others to nurture their future development at the core of contemporary economic geography (Boschma & Frenken, 2006; Cooke, 2012; Martin, 2010; Martin & Sunley, 2006; Storper & Walker, 1989). Inspired by definitions in recent work (see Binz et al., 2016; Hassink, Isaksen, & Trippl, 2019; Steen & Hansen, 2018), in this article an industrial development path is referred to as a critical mass of functionally related firms that are ‘established and legitimized beyond emergence’ (Steen & Hansen, 2018, p. 191). Industrial development paths are embedded in regional innovation systems, consisting of all firms and industries located in the region, organizational support structures, networks and institutional context conditions.

Evolutionary Economic Geography (EEG) portrays the development of new regional industrial paths as an endogenous branching process, in which ‘related’ knowledge assets are combined and provide ‘inputs’ to new paths (Boschma & Frenken, 2011). However, recent work have questioned the narrow view of path development advocated by EEG scholars, pleading for broader conceptualisations to include institutional, social and cultural factors and take account of multi-actor and multi-scalar perspectives of the ‘how and why’ of path development (see Hassink et al., 2019; MacKinnon et al., 2019). The literature has recently also been enriched by offering various more fine-grained typologies to distinguish between different types of path development (see Grillitsch, Asheim, & Trippl, 2018; Isaksen & Trippl, 2016; Martin & Sunley, 2006; Tödtling & Trippl, 2013). These range from the creation of completely new paths to path diversification, path importation, and the renewal of existing paths. Whilst these are ideal types, they point at the wide range of sources, mechanisms and outcomes at play in the evolution of regional industries.

However, current work does not sufficiently explain substantial transformations of existing paths, such as the case of the self-driving cars (SDCs) that is in focus in the empirical part of this article. We use the notion of ‘path transformation’ to capture innovation-based renewal processes of an established path based on radically new technological, organizational or market innovations. Path transformation does not refer to a distinct type of path development, but to the transformation processes underpinning radical forms of path renewal and intra-path changes more broadly. Regions need to embark on new trajectories to prevent negative lock-in and decline. Taking a process perspective comes with advantages over studying the outcomes of path development, in terms of understanding the balance between forces of continuity and change in path dependent evolution (see e.g. Martin’s (2010) ‘path as a process model’).

EEG accounts of path development have been criticized for advocating rather narrow firm- and industry-centred explanations of regional structural change (MacKinnon et al., 2019; Trippl, Grillitsch, & Isaksen, 2018). In response, scholars have forged a link between the RIS approach and EEG models to shed light on wider regional structural factors and conditions that shape new path development (Isaksen & Trippl, 2016). The RIS approach has been used to highlight how existing RIS structures influence the rate and direction of regional structural change and what type of new path development is most likely to occur in regions with different RIS characteristics (Isaksen & Trippl, 2016). The focus in this article is how the transformation of a mature industrial path comes with changes to the broader regional context. In order words: we seek to contribute to a better understanding of the ways in which RISs are reconfigured over the course of path transformation.

Yet, conceptual and empirical work on how the organizational and institutional support structures of RISs need to change in order to facilitate the rise of new paths and the renewal of established ones is scant (Miörner & Trippl, 2017; Tödtling & Trippl, 2013). Whilst the RIS approach has been criticized for its static perspective (Doloreux & Porto Gomez, 2017) and, more recently, for not taking agency into account in explaining system change, actor-based approaches of new path development have neglected the potential inputs originating from and processes taking place at the system level. More recent work has sought to overcome these weaknesses by exploring how agency targeting the system-level may lead to RIS changes (Isaksen, Tödtling, & Trippl, 2018; Isaksen & Jakobsen, 2017; Miörner, 2019).

Building on the contributions outlined above, we investigate how RISs are tackling challenges related to path transformation. Mature industries are often backed by well-established organizational and institutional support structures. If traditional paths face disruptive changes and are substantially transformed, then the wider RIS also needs to be rebuilt. Whilst acknowledging the important role of agency in system reconfiguration, the focus of this article is on how such changes unfold, paying attention to the processes through which a variety of actors facilitate the reconfiguration of RISs.

As noted above, the RIS literature has thus far been fairly vague about what RIS reconfiguration entails. In parallel, other strands of literature, most notably contributions inspired by the TIS concept, have begun to specify system functions that underpin new path development (Binz et al., 2016) and to explicate how system functions might be spatially distributed across various scales (see, for instance, Binz & Truffer, 2017). A key question is thus how a RIS can be reconfigured to develop new functions, or to access or transplant functions performed elsewhere. This resonates with the call to take into account meso-level and macro-level circumstances and the ‘outward’ direction of RIS connections and influences (Martin & Sunley, 2015). In the following sections, we combine insights from the RIS and Technological Innovation Systems (TIS) literatures to unravel the link between path transformation and system reconfiguration.

RIS elements and system functions

Recent work on new path development has challenged the narrow focus on knowledge assets and knowledge dynamics prevailing in EEG. In studies of path creation, scholars have turned to the Technological Innovation Systems (TIS) literature to gain insights into the key system functions that need to be performed in order for new paths to emerge (Binz et al., 2016; Martin & Coenen, 2015; Steen, 2016). The TIS literature outlines six functions (that is, knowledge development and diffusion, entrepreneurial experimentation, market formation, resource mobilization, creation of legitimacy and guidance of the search) that form the basis of a well-functioning innovation system centred around a specific technology (Bergek, Jacobsson, Carlsson, Lindmark, & Rickne, 2008; Hekkert, Suurs, Negro, Kuhlmann, & Smits, 2007). Different variations of these functions can be found in studies of path creation, for example in Binz et al.’s (2016) study of the emergence of water recycling industries in China or Steen and Hansen’s (2018) work on offshore wind in Norway. These contributions suggest that the creation of a new path relies on various assets. These assets are formed through a set of ‘key processes’ that are labelled ‘functions’ in the TIS literature (Bergek et al., 2008, p. 408). In the aforementioned studies, four key assets have been highlighted (knowledge, legitimacy, financial investments and markets), but the particular combination and relative importance of different system functions may vary depending on the industry and its specific needs.

A RIS can be conceptualized as consisting of a set of system elements, referring to actors (both public and private) involved in innovation processes, networks between different actors, and institutions guiding their behaviour (Asheim, Smith, & Oughton, 2011). The RIS approach is useful to understand the regional environment in which industrial development paths are embedded, and thus the spatial context in which path transformation comes about. The TIS framework, on the other hand, allows for analysing the functionality of an innovation system embedded in a socio-technical context, including the emergence and transformation of system functions from a global perspective (Bergek et al., 2008; Binz & Truffer, 2017).

By enriching the regional system perspective found in the RIS literature with the more process-oriented perspective advocated by TIS scholars, it is possible to achieve greater conceptual clarity and to specify in more detail how system elements geographically located within a particular region relate to the formation of assets through system functions taking place in a globally defined context.

Following the TIS literature, system functions can be defined as the key processes through which assets (such as knowledge, skills, legitimacy, directionality, financial assets, and so on) relevant for a particular industry are formed and diffused (Bergek et al., 2008). The framework brought forward here departs from the idea that system elements in a RIS are the localized structures through which assets are provided to regional actors. For example, research and education facilities produce and diffuse knowledge, regulations shape the market conditions, and funding schemes provide financial capital to paths (see Table 1). Importantly, (1) system elements may not only contribute to one but to several functions in the innovation system, and (2) a particular system function could be performed by various system elements. (1) Educational bodies for instance play a key role in the generation (and diffusion) of new knowledge but might also contribute to the legitimacy of an industrial path. (2) The system function ‘knowledge generation’, for example, could be performed by firms, private research facilities, universities, and so on. Existing system functions are often aligned to the established path. Hence, path transformation implies that system functions may need to be changed, adapted, aligned or even created.

Table 1. System functions and system elements.

System function	System elements
Knowledge generation	Education facilities, R&D organizations, vocational training schools, …
Experimentation	Incubators, accelerators, test facilities, …
Market formation	Demand-side policies, platforms, market regulations, action networks, …
Legitimation	Interest organizations, industry associations, consumer groups, standards, norms, …
Direction of search	Visions, strategies, expectations, …
Investment mobilization	Banks, funding schemes, business angels, venture capitalists, …

Source: own elaboration

TIS scholars have argued that system functions are performed across space and scales in networked sets of interdependent subsystems (Binz & Truffer, 2017). Such subsystems are not necessarily spatially defined, which means that regional path transformation processes can rely on system functions that are developed in actor networks and institutional contexts that transcend national or regional borders. RIS elements, that is, actors, networks and institutions, may thus contribute to the formation of assets regionally, or by linking up to system functions elsewhere. In other words, the framework casts light on the relation between RIS elements and what can be referred to as ‘scaled’ system functions.

RIS reconfiguration for path transformation

Regional path transformation requires substantial changes in the institutional and organizational support structures of RISs to assist actors to generate new knowledge, develop markets, create legitimacy and mobilize investments. Previous studies (see Miörner & Trippl, 2017) have demonstrated how RIS reconfiguration may occur through the creation of new system elements alongside old ones. Examples include the establishment of new education facilities and innovation support organizations, formation of new networks or the introduction of new institutional arrangements such as regulations and standards. RIS reconfiguration may also take place by adapting existing system elements, for instance, by reorienting existing educational and research programmes, networks, funding schemes or innovation support instruments to better fit the requirements of paths in transformation.

As noted above, path transformation signifies major changes within an existing regional industry. This implies that the analysis must depart from a regional perspective and adopt an ‘inward-outward’ perspective of structural and agentic circumstances (Martin & Sunley, 2015). We are not analysing the formation of a global innovation system around an emerging technology. We rather take a regional industry perspective and investigate how RISs are reconfigured with the aim to provide ‘old’ transforming paths with new assets that are formed through particular system functions. For that purpose, we outline three types of RIS reconfiguration, which are characterised by different spatial patterns.

Type 1: developing system functions within the region

RIS reconfiguration may take place in order to facilitate the provision of assets within the region. Several previous studies have demonstrated how RIS elements, such as education programmes, research institutes and funding schemes, are created or adapted in order to provide paths with new knowledge and other assets (Dawley, 2014; Miörner & Trippl, 2017; Steen & Karlsen, 2014). This reflects the endogenous dimension of path transformation, where assets are developed regionally and RIS elements are directly associated with one or several system functions linked to new activities in the mature path.

Type 2: accessing system functions elsewhere

In addition to developing assets regionally, system reconfiguration can take place in order to link up to extra-regional assets. This type of system reconfiguration refers to the development of RIS elements that support the mobilization or transfer of assets that emerge from system functions in other regions, and the processes through which such assets are transformed into ‘locally sticky’ ones (Binz et al., 2016). For example, knowledge important for path transformation might be developed through R&D efforts in other regions and accessed by regional actors through the initiation of strategic collaborations and other inter-regional networks, or through the development of formal collaboration platforms focusing on forging links between regional and non-regional actors (Trippl et al., 2018). In other words, this type of system reconfiguration takes into account the fact that system elements may serve the purpose of establishing linkages or ‘pipelines’ (Bathelt, Malmberg, & Maskell, 2004) through which actors access assets formed through system functions elsewhere. Examples of this type of system reconfiguration include the creation of new organizations, such as technology mediating bodies, but also the formation of new network linkages to actors and activities in other regions.

Type 3: transplanting system functions from elsewhere

Finally, system functions required for path transformation may exist in other locations, and system elements that contribute to these functions may be transplanted to the region. An example would be the relocation of R&D units from one place to another, that is, the importation of knowledge production activities. System elements might also be created in order to ‘import’ certain system functions more implicitly, for example by initiating funding schemes aiming at attracting researchers or start-ups that are experimenting with and developing solutions in other regions, but require them to physically relocate to the region in order to access the funding. The spatial characteristics of this type of system reconfiguration are thus different from linking up to extra-regional system functions (type 2), as the focus is not on accessing assets formed elsewhere but on moving system functions to the region.

Path transformation is likely to involve a combination of different types of system reconfiguration, and by studying these we believe that it is possible to gain a deeper understanding of how path transformation unfolds. Thus, our empirical analysis will identify different types of system reconfiguration, but will also pay particular attention to how certain actors coordinate and steer their activities towards pre-defined or emerging goals.

Methodology and case background

We deploy a case study methodology (Flyvbjerg, 2006; Yin, 2013) to conduct an in-depth investigation of system reconfiguration in the context of path transformation. The empirical analysis is based on an extensive document study, including material from a wide variety of sources such as newspaper articles, industry newsletters, reports, PR material, financial information, legal documents, policy documents and video material.

In addition, a total number of 21 semi-structured interviews have been undertaken. Respondents were asked about challenges related to SDC development, what activities actors engaged in, and about the relationships between these activities and elements in the RIS. In that way, the need for RIS reconfiguration could be identified, and it was possible to map the ways through which actors tried to adapt the existing RIS to support their activities as well as how they made use of existing automotive support structures. Customized interview guides were developed prior to each interview, and common questions were complemented with individually chosen sub-questions based on the organizational belonging, role and background of the respondent.

The interviews lasted between 60 and 90 min and were conducted between March 2017 and May 2018. The selection of interviewees is summarized in Table 2. Apart from their organizational belonging, 16 out of the total number of interview partners were chosen based on their long experience with the automotive industry in West Sweden. Informal meetings with three representatives of the regional development organization provided complementary material. Key interview partners were identified during the document study and through a ‘snowballing’ technique. All interviews were transcribed and coded. The coding process for both documents and interviews took place in two steps, first an ‘in-vivo’ coding procedure (King, 2008) identifying interesting themes and patterns, followed by a focused coding based on analytical categories derived from the conceptual framework (Saldaña, 2015). Finally, data from different sources have been triangulated in order to increase the robustness and consistency of the findings.

Table 2. Overview on interviewees.

No.	Role	Organization	Background
1	Expert / Researcher	Research organization	Academia
2	Expert / Researcher	Research organization	Academia / automotive industry
3	Programme director	Innovation support organization	Automotive industry
4	Senior advisor	City of Gothenburg	Spatial planning
5	CEO	Automotive technology firm	Automotive industry
6	Department director	Automotive firm	Automotive industry
7	Project leader	National agency	Spatial planning
8	CEO / Programme director	Innovation support organization	Automotive industry
9	Researcher	Academia	Academia
10	Unit director	Region Västra Götaland	Regional development
11	Project leader	City of Gothenburg	Spatial planning
12	CEO	Public company	Mobility sector
13	CEO	Public company	Energy sector
14	Unit director	IT firm	IT industry / start-up
15	COO	Public company	Automotive industry
16	Senior consultant / Expert	Consultancy firm	Academia
17	Unit director	Innovation support organization	Regional development
18	CEO	Innovation support organization	Shipping industry
19	Unit director	Automotive technology firm	Automotive industry
20	Unit VP	Automotive firm	Automotive industry
21	Board member	Network organization / Innovation support organization	Automotive industry / Public sector

Source: own compilation

The Swedish automotive industry is concentrated in the region of West Sweden (Nutek, 2009). It employs approximately 35,000 people (BRG, 2017), which represents 40% of the total employment of the Swedish automotive sector. Of the total number of workers in the region, 81% are employed in firms manufacturing automotive vehicles, and another 19% are employed by firms manufacturing vehicle parts (VG, 2016). West Sweden is home to large firms such as Volvo Cars, Volvo AB (trucks), HCL Technologies Sweden, CEVT, Autoliv and IAC.

The region of West Sweden hosts firms from all parts of the value chain and this is complemented by strong competencies in the organizational support structures of the RIS. The higher education organizations in the region, most notably Chalmers University of Technology, are important nodes for knowledge generation and experimentation for the automotive industry. In addition, there are several independent research institutes focusing on technical as well as societal issues related to the automotive industry. Furthermore, there are many intermediaries, science parks (such as Lindholmen Science Park), innovation arenas and cluster organizations, contributing to the direction of search as well as other system functions. In other words, system functions are well-developed regionally and strongly aligned to the needs of the automotive industry. The automotive industry thus benefits from ‘thick’ regional support structures (Holmquist & Söderkvist. The region is also home to a strong ICT industry, working closely with some of the automotive firms, pointing to favourable inter-path linkages within the RIS.

The automotive industry in West Sweden is characterised by a high level of competence in the field of active safety technology. Active safety refers to the use of technologies aiming at avoiding accidents, such as the antilock braking system (ABS) and automatic warning systems. Innovations in the field of active safety required creation of networks between previously unconnected actors to ensure the combination of knowledge from a variety of fields, such as mechanics, electronics and IT. The development of active safety in West Sweden started in the 1980s and led to the re-alignment between the support structures of the RIS and the automotive path. For example, SAFER was created in 2006, adopting the role of both a research institute and a network organization for coordinating the work around active safety in the region (James, Vissers, Larsson, & Dahlström, 2016).

In our interviews, Google announcing a self-driving car unit in 2009 was mentioned as a ‘global trigger’ for the automotive industry to start focusing on autonomous vehicles. In West Sweden, the origin of path transformation can be traced back to the year 2013, when the ‘Drive Me’ project was launched. The project is a joint initiative between Volvo Cars, Autoliv, the Swedish Transport Administration, the Swedish Transport Agency, Lindholmen Science Park, Chalmers University of Technology and the City of Gothenburg. It is supported by the Swedish government and sub-projects are partly funded by Sweden’s innovation agency Vinnova. The aim of Drive Me is to study the societal benefits of autonomous driving. The project’s core is the introduction of fully autonomous cars to be driven by ‘real customers’ on public roads in Gothenburg. Based on Volvo’s premium SUV, which include ‘combined function automation’ such as assisted lane-driving, the new autonomous version ‘add hands-off and feet-off capability in special autonomous drive zones around Gothenburg’ (Drive Me, 2016).

Industry and technology experts interviewed within the study directed our attention to a number of challenges surrounding SDC technology that have the potential to give rise to conflicts and tensions between actors with different intentions and interests. For example, there are questions related to the regulation of the ‘ethics’ of SDCs in which the interests of regulators and automotive firms might be incompatible, and broader concerns related to the future role SDCs should play in urban mobility systems. It is possible to observe a divide between public actors, such as city planners and politicians, that perceive the future of SDCs as being ‘shared’ in carpools and the like, and automotive firms who consider SDC capabilities to be a crucial feature in individually owned cars. However, based on the interviews and other material analysed in this study, one striking observation is that the majority of regional actors, whilst acknowledging that discussions take place on the global level, have managed to avoid diverging opinions and tensions materializing into actual conflicts in the Gothenburg area. By drawing on previously built up networks and a history of collaboration between actors in the region, possibly also supported by a ‘culture of consensus’ prevailing in Gothenburg (Thörn, 2011), actors seem to have managed to avoid many of the conflicts that would be expected in processes of path transformation. That said, it is beyond the scope of the study, also limited by the period of observation, to give any assertions of that this will continue to be the case throughout the whole path transformation process.

The transformation sparked by the Drive Me project was initially based to a large extent on the existing technological trajectory centred around the active safety segment of the industry. Nevertheless, for SDCs to become reality, there is a need to not only further develop state-of-the-art technology but also to generate new non-technological knowledge related to other change dimensions, such as consumer attitudes, spatial planning principles, market and business models, infrastructure and so on. It is crucial for the transformation of the regional automotive industry that the RIS supports and facilitates such processes.

Results and analysis

Developing system functions within the region

Over the past few years, an array of activities and initiatives has been undertaken to mould the RIS for progressing with SDCs. In order to support knowledge generation, new organizations have been added to the established system of support organizations. A case in point is ‘Revere – Resource for Vehicle Research’, a vehicle software lab focusing on autonomous driving technology created in 2016 at Chalmers University. Revere was initiated by Chalmers in collaboration with actors from the automotive industry (Volvo Cars and AB Volvo) and is supported by regional public and private funding sources. This new research facility contributes to the development of new technological knowledge around SDCs in distinct ways. It provides an environment in which software developers have access to full vehicles and vehicle systems provided by the automotive industry. This constitutes a break with traditional practices of vehicle research and development taking place at a low modular level.

Furthermore, existing education and research programmes at Chalmers University have been reoriented towards autonomous driving technology and new programmes have also been launched. Evidence from interviews shows that these processes have been facilitated and shaped by long-standing networks between Chalmers and the local automotive companies, providing the latter with a forum to communicate their needs and ‘co-guide’ reconfiguration of the university’s education and research agendas. Several of these networks are formalized in platforms or projects dealing with topics related to technology transfer and competence provision.

Another example for the region’s capacity to build new system elements that serve knowledge generation is the establishment of an ‘artificial intelligence’ (AI) research facility in Gothenburg. In our interviews, Zenuity has pointed to the central role that AI holds for solving some of the remaining technological challenges for SDCs. In order to cope with the shortage of AI competence in the region, Zenuity has joined forces with Volvo Cars, AB Volvo, Autoliv, Ericsson and Chalmers University and initiated the setting-up of an AI centre at Lindholmen Science Park. As stated by an executive at Zenuity and supported by statements in other interviews with representatives of the automotive industry:

We call it a world class AI ecosystem. [..] the idea is that we gather both competence and data at the same place … many in academia are lacking data, and we could give them that, because we have connections to large car fleets, through our customers. (Interview)

Our findings suggest that the RIS building processes analysed above did not only benefit the generation of new technological knowledge needed for SDC development but have also produced side-effects in terms of legitimizing activities around autonomous technology and stimulating cultural shifts within the regional car industry. Several interview partners from the automotive industry and academia alluded to a shift from a traditional ‘engineering culture’ towards a more agile software focused way of working within the automotive industry, and how new system elements such as the AI centre are contributing to a new mindset in the region that is favourable for transforming the automotive path towards SDC development.

Arguably, not all new knowledge required for transitioning towards SDC is developed within the automotive industry. Several interview partners underlined the importance of knowledge assets formed within other regional industries, most importantly the IT sector. Formation of collaborative linkages between Volvo Cars and Ericsson’s local R&D centre can be traced back to the early 2000s, when the companies launched joined initiatives in the field of ‘connectivity’. These initiatives subsequently became a valuable source of technological knowledge for SDC development, but with the ongoing changes in the automotive industry they evolved into also being an important source of non-technological knowledge, for example in terms of business models, ‘mindsets’, and ways of working with software development. This highlights the ways through which actors make adjustments to existing system elements, that is, well established networks, to solve challenges related to new activities and path transformation. Interactions between actors in the automotive and IT industry intensified and took new turns, from being organized as collaborative projects around certain features, to also include labour mobility and facilitate cross-industry recruitment of executives and specialists.

If looking at historically grown RIS elements, such as innovation support organizations and other platforms, the existing system was already strongly supporting the automotive industry. In the early 2000s, the region invested heavily in the creation of test facilities that met the needs of the automotive sector, leading to the establishment of platforms such as ‘Test Site Sweden’, physical test facilities such as ‘Asta Zero’, and road segments with adapted infrastructure. Whilst these were not explicitly targeting autonomous driving at the time of their establishment, through adaptations they have become a key element in the RIS around SDC development. Such facilities play an important role in the formation of technological knowledge related to verification and test methods, but also experimentation, legitimation and investment mobilization. As one representative of the automotive industry expressed it:

There is an explosion of testing, and a lot of work is put into methods development … with all the layers and variants that need to be tested. Asta Zero is world leading in this work. (Interview)

Finally, when it comes to elements related to investment mobilization, existing public R&D programmes for active safety technology have been ‘re-branded’, in some cases by the funding bodies themselves but in most cases by the participating firms, to fit SDC development agendas. For example, Volvo Cars decided to use the remaining funds in a large research programme on vehicle technology for advancing SDCs. Given the close technological relationship between active safety and SDCs, this was mainly a matter of wording and terminology but it had also important signalling effects in terms of demonstrating the importance of publicly funded R&D in relation to SDCs.

Accessing system functions elsewhere

Given the growing importance of software development in the automotive industry, traditional car firms are increasingly linking up with actors in IT regions such as Silicon Valley. New networks are formed by regional automotive firms in order to get access to knowledge and competences that are related to specific issues in SDC development. Volvo Cars and Autoliv have begun to form strategic collaborations with actors such as the process manufacturer Nvidia, the mobility provider Uber and the sensor developer Mobileye. These linkages represent a new form of networks, based on mutual exchange and collaboration around new technologies that have previously not been prominent in the automotive industry. In our interviews firm representatives also highlighted the establishment of ‘probes’ in start-up hotspots such as Silicon Valley in order to come closer to and establish collaborations with start-ups in other regions. This is being done by both Volvo Cars and Zenuity and is argued to serve as an important way to access knowledge available in other regions, as well as an opportunity for Swedish firms to showcase their SDC solutions and possibly attract additional investment. Furthermore, Zenuity has started development units in Munich and Detroit in order to mobilize non-regional competences.

Then there is also evidence that actors in the West Swedish automotive industry have linked up with experts in other regions to further develop non-technological competence, such as knowledge about user behaviour, norms and attitudes towards SDCs, and driver-car interaction. One such example is HEAD, an interdisciplinary inter-regional research project involving researchers from universities in several Swedish regions who are active in disciplines such as ethnography and interaction design. HEAD is hosted by Halmstad University and part of the Drive Me project. This illustrates how Drive Me is used not only to generate new assets within the region but also to coordinate regional activities with knowledge development taking place in other locations.

Furthermore, actors in West Sweden have established new networks with national organizations such as the Swedish Transport Administration (STA) to ensure access to and co-create knowledge about the future of road transport infrastructure and traffic management. These linkages between regional and national actors appear to have a strong influence on the direction of research in this field in West Sweden and are used by the latter in an intentional and strategic way for that purpose. Several interview partners from the automotive industry alluded to this role played by STA, and a representative of STA expressed it clearly as:

[..] in order to solve certain issues, certain challenges need to be dealt with, and we push the companies to become interested in those particular challenges. (Interview)

It is also possible to identify processes of alignment of system functions within and outside the region. Existing initiatives, such as SAFER, Lindholmen Science Park and Test Site Sweden, have become important ‘nodes’ in coordinating system functions in the region and elsewhere. They perform important roles in terms of coordinating activities and aligning the RIS with processes of legitimation and experimentation elsewhere. For example, activities drawing on existing test infrastructure in the region were driven by the major firms but coordinated and in part facilitated by Test Site Sweden.

Transplanting system functions from elsewhere

In terms of investment mobilization, our interview findings indicate that whilst both regional and national funding sources play an important role, the importance of investments by Volvo Cars’ parent Chinese firm, Geely Automotive, should not be neglected. Over the last few years, Geely has invested heavily in the region, locating, for instance, new subsidiaries in West Sweden. There are also plans to build an innovation centre employing approximately 3,500 people. Our findings suggest that adapting the physical environment of the Lindholmen area has played a major role in this regard. Through long-term strategic planning led by city actors, the Lindholmen area has been transformed to attract not only automotive OEMs and suppliers, but also IT firms, a Chalmers University campus and technology consultancies. Several of our interview partners have highlighted the attractiveness of Lindholmen as a crucial determinant for why Geely has located its innovation activities in West Sweden. Another important factor, according to several interviewees, was the establishment of Asta Zero and other test facilitates and test infrastructure, discussed in previous sections.

Moreover, during the past few years, regional actors have taken several initiatives to increase the attractiveness of West Sweden as a ‘the place to be’ (Interview) for engineers interested in autonomous technology. When firms such as Volvo Cars, Zenuity, and Autoliv faced the challenge to attract the ‘right’ competence for solving some of the remaining obstacles related to SDCs, they acknowledged the need for engaging with start-ups in ways not previously done in the industry. Several interview partners from the automotive industry highlighted such challenges, but it was most prominent in the interview with Zenuity, where an executive stated that:

We are searching for start-ups all over the world. It is of course easier if they are close by, so that you can have quicker interactions, build trust quicker and so on. (Interview)

A conventional approach would be to set up incubators or accelerators to support local start-up activities. Automotive firms in West Sweden chose to employ a new strategy to work with start-ups venturing on knowledge development that is of relevance for SDCs. Rather than developing regional functions that support the emergence of new knowledge through start-up activities, major industrial actors in the region initiated Mobility X-Lab, a platform not only for interacting with start-ups from elsewhere but for temporarily relocating small firms to the region. Mobility X-Lab offers a physical environment through which start-ups are co-located with the industry partners, facilitating collaboration and knowledge exchange between the actors. In other words, it is not only about scouting start-ups but to anchor small firms in the RIS around automotive technology at the cross-section between the automotive industry and IT. The creation of Mobility X-Lab does indeed represent implementation of a new system element, but the element is targeting the anchoring of system functions existing elsewhere, in the form of knowledge generation and experimentation. For example, representatives of Zenuity as well as of Mobility X-Lab highlighted the importance of attracting already established start-ups to the region, since the regional entrepreneurial system is somewhat underdeveloped. These start-ups operate in fields such as video recognition and deep-learning microprocessors. They come from entrepreneurial hotspots like Silicon Valley, Finland and Israel, and are thus the outcome of system functions performed elsewhere. Mobility X-Lab represents a way of transplanting them to the region, and one of its representatives expressed that:

We see that it is mainly more mature start-ups that are coming to us, which is an advantage because we are not an incubator with a business development support function. (Interview)

It is also apparent from our interviews that firms engaged in SDC development have tended to ‘externalise’ the coordination of regional system functions to intermediaries such as Lindholmen Science Park. Though they share some goals and interests, their corporate cultures as well as partly competitive relationships prevent the major industrial actors in the region from engaging in collaborative projects without a neutral coordinator. Consequently, regional intermediaries have been found to play an important, if not crucial, role. Interviews with representatives of the automotive industry largely confirmed the perspective brought forward by a representative of Lindholmen Science Park:

it is about making an initiative neutral. Mobility X-Lab was possible to start because Lindholmen could go in as host for it, and make it neutral. Otherwise it would have become a Volvo incubator, a Volvo environment, where Volvo would have tried to invite the others to their thing … And it would not have been the same thing. (Interview)

Our findings clearly indicate that the intermediary role played by Lindholmen Science Park and other organizations has gained in significance over time. This has been connected to processes of adaptation within the organizations, guided by the new needs of actors operating in the automotive industry. Worth mentioning in this context is also the role played by ‘Drive Sweden’, a national strategic innovation programme for the future of mobility, with a special focus on driverless and connected cars. Drive Sweden is not a regional programme but is hosted by Lindholmen Science Park and thus has a strong regional presence in West Sweden. Our research suggests that Drive Sweden serves as an innovation network that helps to transplant system functions from the national level (mainly in the field of new business models and legitimacy for SDCs) to the region. As one interview partner from the automotive industry put it:

In this mission, we are completely national and transparent. It is not that you should try to attract things to Lindholmen and only talk to companies here, but work all over the country and also globally. But then there are situations where the ecosystem can play a role, or the built environment, and then of course you can benefit from being part of Lindholmen as well, of course. (Interview)

Finally, there are many ongoing activities that in different ways are geared towards the formation of a ‘regional SDC agenda’. These are concerned with branding the region as an SDC region, and enhancing international visibility. Whilst one cannot claim that these activities have a direct influence on the formation of new assets, they are important for situating the RIS in the emerging global SDC system. Several interviewees have highlighted that Gothenburg is considered as an attractive environment for firms working with SDC development. However, the formation of an SDC agenda should not necessarily be seen as a case of creation of a new system element, but rather as a process of adapting existing system elements. Of the activities identified in our study, a majority of them seek to ‘implant’ incremental changes to existing plans and strategies. There are several examples of this. Industrial policies have to date not been re-aligned to focus on SDC development. Public actors are trying to find overlapping areas in existing policies related to industrial development, spatial planning and public transport. Experts from the city planning department in Gothenburg are involved in a nationally funded project where they, together with actors from the automotive industry, investigate what SDCs mean for city planning practices and principles as a sub-project of ordinary planning strategy work. Whilst it is formulated as a way for city planners to learn about the implications of SDCs, it has contributed to putting Gothenburg on the global SDC map, as it is actively promoted as a way for the city to prepare for a ‘driverless future’.

Table 3 provides an overview on the key empirical findings and demonstrates that the transformation of West Sweden’s automotive industry towards SDCs is linked to complex processes, involving a combination of different types of system reconfiguration.

Table 3. RIS reconfiguration for SDCs in West Sweden: system elements, functions & actors.

RIS element(s)	System function(s)	Key actors(s)
Developing regional system functions
Adaptation of funding schemes	Investment mobilization	Automotive OEMs; Automotive suppliers
Adaptation of innovation support organizations, platforms (Test Site Sweden, Lindholmen Science Park)	Knowledge generation; legitimation; investment mobilization	Automotive OEMs; Automotive suppliers; Regional policy actors
Adaptation of physical infrastructure	Experimentation; legitimation	Autmotive OEMs; City actors; National policy actors
Adaptation of regional inter-industry networks (Automotive-IT)	Knowledge generation; direction of search	Automotive OEMs; IT firms
Launch of demonstration project (Drive Me)	Legitimation; experimentation	Automotive OEMs
Creation of research centre (AI)	Knowledge generation	Automotive OEMs; SDC firms; IT firms; University
Creation of research lab (Revere)	Knowledge generation; direction of search	Automotive OEMs; University
Establishment of research programme (Chalmers Transport Area of Advance)	Knowledge generation	University
Creation and adaptation of courses and education programmes	Knowledge generation; legitimation	University
Creation/adaptation of test facilities (Asta Zero)	Experimentation; knowledge generation	Regional policy actors; Automotive OEMs; Innovation support organizations
Accessing system functions elsewhere
Adaptation of intermediaries (SAFER; Lindholmen Science Park; Test Site Sweden)	Direction of search; legitimation; experimentation	Automotive OEMs; Automotive suppliers; University
Adaptation of physical environment	Investment mobilization	City actors; Automotive OEMs
Creation of extra-regional network initiatives	Direction of search	National policy actors
Launch of inter-regional research projects	Knowledge generation; direction of search	Automotive OEMs; SDC firms
Creation of inter-regional strategic collaborations	Knowledge generation	Automotive OEMs; SDC firms
Establisment/adaptation of ‘network-probes’ in other regions	Knowledge generation; investment mobilization	Automotive OEMs; SDC firms
Transplanting system functions from elsewhere
Adaptation of agendas and visions (SDC agenda; city planning principles)	Legitimation; Direction of search	City actors; foreign actors
Developing new innovation networks (Drive Sweden)	Knowledge generation; Experimentation; Direction of search; Legitimation	National policy actors; Automotive OEMs, Innovation support organizations
Creation of innovation support organization (Mobility X-Lab)	Experimentation; Knowledge generation	Automotive OEMs; Automotive suppliers; SDC firms; IT firms

Source: own compilation

Discussion and conclusions

This article seeks to contribute to the path development debate in economic geography by investigating how mature regional industries can be substantially transformed and how RIS reconfiguration supports such processes. Combining insights from regional and technological innovation systems literatures we develop a conceptual framework to analyse different types of RIS reconfiguration underpinning the transformation of old industries. By doing so, we establish a connection between RIS elements and the system functions for path transformation.

Our empirical analysis of the automotive industry in Gothenburg (Sweden) and its current transformation towards SDCs confirms the importance of the three types of system reconfiguration identified in our conceptual discussion. System functions are developed and performed regionally, by the creation and adaptation of research labs, education programmes, test infrastructure, innovation support organizations, funding schemes, and intra-regional network linkages. New inter-regional network linkages are also formed, accessing system functions established in other regions, such as Silicon Valley but also linkages to other Swedish regions are found to play a role. Apart from creating new linkages, existing system elements are adapted to take on coordinating roles in managing the external connectedness of regional actors. Finally, we find that the transplantation of system functions, such as investment mobilization and experimentation, through the establishment of new RIS elements, is vital.

The empirical case study also highlights the importance of activities undertaken by incumbents in path transformation. Established automotive firms, such as Volvo Cars and Autoliv, play a crucial role in both initiating new activities and facilitating system reconfiguration processes. To some extent, this finding is contrary to conventional wisdom saying that incumbents are not usually the initial leaders of transition processes (see, for example, Geels, 2011). The study also highlights the role of regional public actors and their contribution to some system functions, such as legitimation and the direction of search. Furthermore, it presents us with interesting examples of how non-local actors can influence RIS reconfiguration. For instance, the national Swedish Transport Administration provides access to system functions, whilst Drive Sweden and the Chinese owners of Volvo Cars contribute to the transplantation of system functions from elsewhere.

By analysing our empirical case using the proposed analytical framework, three important points can be made. First, whilst the relative importance of developing system functions within the region arguably have found been high in previous studies of path development, the relationship with the other types of system reconfiguration is not as clear cut when it comes to path transformation. Actors are guided by visions and strategies that are often developed regionally by industry incumbents, but it is not the case that the default option is to target system reconfiguration to develop other necessary assets regionally. Rather, actors tend to utilise previous networks and their position in global innovation systems to reconfigure the RIS in order to also access and transplant system functions already performed elsewhere. It remains to be investigated whether this is true also for other industries than the automotive industry, and for different types of regions, which calls for this as a key topic for further empirical work.

Second, it is apparent in our study that RIS reconfiguration does not only target the development, access and transplantation of technological knowledge. In line with previous studies, our findings support the idea that it is necessary to account for the formation of a wide range of different assets, such as legitimacy, visions, regional mindsets and culture (see Binz et al., 2016; MacKinnon et al., 2019). The way through which the RIS is reconfigured to facilitate the mobilization of these assets is likely to differ between regions and industries, pointing to another important direction for future studies of path transformation and system reconfiguration.

Third, it is important not to neglect the influence of other regional industrial paths on the transformation of an existing regional industry. The RIS may not only support one but several regional industries. In our case, the RIS was strategically reconfigured to exploit recombination potentials between the automotive industry and the regional IT industry. Elements were created and adapted for ‘tapping into’ existing system functions in the IT sector and using them for transforming the automotive industry. This resonates with findings from work on the importance of inter-path relationships (Cooke, 2012; Hassink et al., 2019; Trippl & Frangenheim, 2018).

Despite the possibility that conflicts might materialize in the future, the empirical analysis illustrates a relatively successful case of system reconfiguration in a particular region. Future studies should be concerned with investigating what determines the capacity of regions to set in motion system reconfiguration processes in different spatial contexts and connected to path development in different industries. Regions are likely to differ in terms of their adaptability and receptivity for innovation system change. Future research should thus cast light on factors shaping the nature of system reconfiguration to better understand the relative importance of the different types of system change and explore what factors enable or constrain system reconfiguration efforts undertaken by actors in variegated regional and industrial contexts.

Acknowledgements

The authors wish to thank David Wolfe for helpful comments on an earlier version of this article. We are also grateful to the editor and two anonymous reviewers for their constructive criticism and suggestions.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Johan Miörner http://orcid.org/0000-0003-0092-5532

Michaela Trippl http://orcid.org/0000-0001-9779-4836