Innovation networks and green restructuring: Which path development can EU Framework Programmes stimulate in Norway?

ABSTRACT

With combined insights from evolutionary economic geography and transition studies, the article examines the engagement of different regions in Norway in the innovation networks created within the European Union’s environmental programmes. The aim is to explore the programmes’ potential for supporting green economy and economic restructuring through branching and new path creation. The authors assess which regions participate in the programmes, which international networks they build, and which organisations participate in different regions. They compare three regions with different restructuring needs and research capacity – the counties of Rogaland, Hordaland, and Sør-Trøndelag (now part of the county of Trøndelag). They find that overall, Norwegian organisations participate relatively frequently in the programmes, but private firms play a marginal role. Their partners are mainly in core EU regions. Regional participation in the programmes is a function of research capacity as well as oil dependence. The authors conclude that in research-oriented regions, research establishments tend to dominate participation, creating potential for restructuring mainly through path creation. In oil-dependent regions, private firms account for a higher share of participants, enhancing the potential for branching. As the former regions participate more, the programme can mainly stimulate path creation.

EDITORS:

• Markus Steen, Kerstin Potthoff, Catriona Turner

KEYWORDS:

• economic restructuring
• evolutionary economic geography
• green economy
• innovation networks
• transition studies

Introduction

Addressing environmental problems is an important motivation for innovation policies. Among other programmes, the European Union’s (EU) Horizon 2020 funds international research collaboration to address environmental issues (European Commission 2018). Additionally, an expert committee appointed by the Norwegian Government recommends directing research and development (R&D) towards addressing environmental issues and promoting international research collaboration (Hedegaard & Kreutzer 2016). The aim of this policy recommendation is to develop green competitiveness and thereby support the transition to a more sustainable economy. Thus, it is interesting to examine the ability of R&D funding to mobilise different types of actors and regions to engage in international research collaboration networks.

Economic restructuring requires new path development, which can take place through branching or new path creation. These normally regional processes require different types of policy stimulus. Branching – moving into new but related industries (Boschma 2017; Grillitsch et al. 2018; Isaksen et al. 2018) – requires mobilising existing industries and bringing in complementary knowledge through international collaboration (Trippl et al. 2017). Branching is more important in industrial regions with a need for restructuring. Path creation implies a central role for research establishments, which can link with leading international institutions and foster breakthrough research (Isaksen 2015). These processes leading to path creation usually take place in regions with strong research capacity.

In this article, we analyse the mobilisation of Norwegian regions and organisations in the EU’s environmental programmes, and ask whether the mobilisation reflects potential for branching or new path creation. We focus on three related questions: Which regions participate in the programmes? What international networks do the participating regions establish? Are there differences in which types of organisations participate? We examine participation in the European Union’s Sixth and Seventh Framework Programmes (FP6 and FP7) and Horizon 2020 (H2020, also informally known as ‘FP8’), and compare three structurally similar counties with different needs for restructuring and research capacity – Rogaland, Hordaland, and Sør-Trøndelag (now part of the county of Trøndelag). We investigate whether the participation in environmental programmes reflects the potential for branching and new path creation towards a greener economy, and whether there are any differences across regions in the impact of the programme.

Green economic restructuring

Transition towards a greener economy

Policymakers increasingly emphasise the need for innovation to solve grand societal challenges, extending the motivation for innovation policy beyond economic competitiveness (Kuhlmann & Rip 2014). Societies around the world need to transition towards a more sustainable economy to reduce global warming and other negative environmental effects of production and consumption. Such a transition requires innovation, for example to improve technologies for renewable energy production, to make transportation and building solutions more efficient, or to develop technologies for carbon capture and storage (European Commission 2011). Accordingly, governments are redirecting R&D funding towards research and innovation in these areas (Johnstone et al. 2010; Wüstenhagen & Menichetti 2012; Mazzucato 2015). Transition also requires more sustainable technologies to replace existing technologies, making user and industry involvement in the R&D projects important.

Transition is not just a function of new technology, but requires transformative changes in production and consumption – in short, a change in the sociotechnical system (Geels & Schot 2007). Literature on sustainability transitions stresses the importance of policy measures that could lead to environmental technological change (Schmidt et al. 2012). Environmental policy strategy is a combination of policy objectives and plans targeting social and economic issues (e.g. economic growth, competitiveness, and new jobs), in addition to environmental objectives.

The participation of different types of organisations in joint R&D projects is a relevant means for developing niche technologies within a broader systemic approach to economic restructuring (Rogge & Reichardt 2016). The term ‘niches’ refer to ‘alternative technologies, practices, structures and actor-constellations deviating from dominant socio-technical systems’ (Kivimaa & Virkamäki 2014, 30). Potential technological transitions at the micro-level (niches) must subsequently be embedded in a broader sociotechnical regime (Markard & Truffer 2008). In the transition, a first phase – generally characterised by new firm entry, knowledge diffused through innovation networks, and uncertainty at the institutional level – is followed by a second phase in which the ongoing transition affects the activity of incumbent firms, new business models emerge, and policymakers outline clearer visions for new development paths (Markard 2018).

The geography of green restructuring

Although economic geographers have long been interested in innovation, its role in solving environmental challenges is relatively unexplored in the field (Truffer 2008; Aoyama 2011). Until recently, the sustainability literature has largely ignored the spatial dimensions of innovations (Hansen & Coenen 2015). Meanwhile, economic geographers have largely not engaged in the discussion on grand challenges (Coenen et al. 2015a). The geography of sustainability transitions literature emerged mainly since 2010, predominantly examining how niches develop in specific places and new technologies emerge. As Essletzbichler (2012) notes, green niches and development paths tend to be geographically localised and often emerge outside the core. Previous research has highlighted multiscalar interactions and non-local relationships (Hodson & Marvin 2010; Essletzbichler 2012; Lawhon & Murphy 2012; Uyarra & Gee 2013), which can break up existing constellations of actors in decision-making networks, allowing for transition. Economic geographers have been called to adopt a multiscalar approach in order to ‘comprehensively understand in which situations and for which purposes relations at different scales matter’ (Hansen & Coenen 2015, 104).

Garud & Karnøe (2003) identify bricolage and breakthrough as two contrasting transition approaches. Rather than aiming for high-tech breakthroughs, the bricolage approach, which is often more successful, emphasises scaling up from simple low-tech solutions and building on widely distributed local knowledge from a range of actors. In the transition towards a green economy, it is not sufficient to develop new technologies. Transition requires regional economies to change their production systems by adopting new technologies and developing new products, services, and business models (Markard & Truffer 2008; Coenen et al. 2015b). The transition to a greener economy is a local problem as well as a global one. Each region must develop competitiveness in areas compatible with the green economy in order to secure future well-being. This challenge is greater for regions where the current sources of competitiveness lie in less green industries. They face a profound transition of their economies, requiring concerted efforts by innovators, researchers, and investors, as well as appropriate innovation policies.

However, academics’ and policymakers’ lack of attention to geography suggests that innovation policy for sustainability transitions might be as spatially blind as traditional innovation policy (Hansen & Coenen 2015). Green innovation policy programmes remain mainly oriented towards R&D and rely on competition between different networks for access to scarce funding. These characteristics make them potentially more attractive to universities and research organisations than to industry (Roediger-Schluga & Barber 2008; Steen & Hansen 2018). Consequently, funding for green innovation projects might predominantly mobilise regions with stronger initial endowments of research capacity, rather than those facing the largest economic restructuring challenges. If such regions were to leverage their research capacity to develop transformative innovations that could be implemented across the economy, it could be an effective strategy. However, for transition to succeed, there is a need to disseminate new technologies into regions and industries that need to adopt them.

How do regions restructure?

Innovation networks, knowledge exchange, and new path development

The need for regional economic restructuring brings sustainability transitions into contact with evolutionary economic geography (EEG), which offers complementary perspectives to those found in transition studies (Essletzbichler 2012; Boschma et al. 2017). The EEG perspective on restructuring sees regional economic development as path dependent. Path dependence may result in regional lock-in, when innovation activities predominantly take place along existing technological paths (Coenen et al. 2017). This makes it harder to move into new directions, resulting in weak regional competitiveness.

For regions locked into unsustainable technological paths, successful transition requires new path development. Regional branching refers to building on knowledge from existing non-green industries in order to develop new competitiveness in green industries (Grillitsch et al. 2018). Path creation refers to the establishment of new firms in new sectors. This is often based on the commercialisation of research or on external investment (Tödtling & Trippl 2013), and may require appropriate policy interventions (Dawley 2014). The EEG literature emphasises the role of innovation in restructuring processes, and the need for knowledge exchange through networks to achieve this (Truffer & Coenen 2012).

Firms that collaborate with other firms, universities, research centres, and other organisations are able to benefit from access to new skills, ideas, and resources. Innovation networks provide firms with a broader knowledge base and hence improve their potential for innovation (Powell & Grodal 2005). Economic geographers have focused on the spatial and relational dimensions of collaborative configurations (Giuliani 2007; Calignano 2014; 2017). Network structures have been examined in-depth with the objective to reveal how the geography of knowledge sources shapes innovation (Balland et al. 2013). Innovation networks often have a set of core regions and such regions act as junctions in the knowledge exchange dynamics, although the identity of those regions differs across technologies (Paci & Batteta 2003). It can be difficult to access the networks, as access presupposes the sharing of knowledge. Such knowledge is hard to acquire and mainly exchanged within the networks, creating self-reinforcing dynamics (Autant-Bernard et al. 2013). Preferential attachment (the tendency of new nodes to connect themselves with nodes that are already well connected; Barabàsi & Albert 1999) and path dependence (Martin & Sunley 2006) play a critical role in innovation networks and shape core–periphery dynamics (Sun & Liu 2016).

Geography of knowledge sources and new path development

While economic geography literature has traditionally been preoccupied with local knowledge exchange, recent research emphasises multiscalar interactions. Actors, networks, and institutions operate simultaneously at various geographical scales. The ability to bridge such scales by means of relationships and stable cooperation patterns is essential for firm innovation. Following Binz & Truffer (2011), innovation should be examined as interdependent processes in a multiscalar perspective. In particular, firms in peripheral regions can potentially benefit from long-distance knowledge flows (Bathelt et al. 2004; Fitjar & Rodríguez-Pose 2011a; 2011b; Grillitsch & Nilsson 2015). These connections are not without challenges, as it is more difficult to transfer tacit knowledge across large distances. Geographical proximity is associated with social proximity or institutional proximity, both of which are important for knowledge exchange (Boschma 2005). Furthermore, long-distance collaboration requires an absorptive capacity, yet this is often lacking, especially in small firms (de Jong & Freel 2010). An important function is therefore held by gatekeepers (e.g. universities or knowledge-intensive firms), which acquire knowledge through international networks and diffuse it to regional firms (Graf 2010; Giuliani 2011).

International innovation networks may be particularly important when economies and, in particular, highly specialised or peripheral regions (Trippl et al. 2017) try to break away from lock-in situations. They may enable countries and regions to recombine competences (Frenken 2000; Fitjar & Rodríguez-Pose 2013), helping them to avoid lock-in to obsolete technological trajectories. Narula (2002) points out that international networks are particularly needed in Norway. He identifies lock-in between dominant industries and leading research institutions as a key challenge, resulting in inertia among firms outside the industrial core, and concludes that ‘relying solely on in-country competences may lead to a sub-optimal strategy’ (Narula 2002, 814).

Different regional innovation systems are associated with different potential development paths and related policy and networking requirements. Fostering path modernisation (upgrading of existing industries based on new technologies) and branching (diversification into related industries) is important in ‘thick and specialised’ systems such as the oil-dependent Norwegian regions (Isaksen et al. 2018). A recent study of the transition from offshore oil and gas to wind suggests how diversifying into related industries (branching) is a critical factor in enabling new path development in Norway (Steen & Hansen 2014).

Innovation networks in European Union’s framework programmes

Findings of previous studies that examined the Framework Programmes

The EU’s Framework programmes (FPs) represent an important policy initiative to promote innovation networks involving distant partners at various geographical scales (Calignano 2017). A number of studies have examined the innovation networks created in these programmes (Breschi & Cusmano 2004; Autant-Bernard et al. 2007; Must 2010; Wanzenböck et al. 2015). Roediger-Schluga & Barber (2008) revealed the existence of network hubs, a stable core of connected actors since the early FPs, and growing integration among the organisations involved in the various programming cycles. Consortia are often based on past collaborations and existing personal or institutional relationships, thus reinforcing the mechanism. Technological and social proximity, more than geographical proximity, drive the formation of new linkages (Scherngell & Barber 2009; Calignano 2014). Universities and research centres play a central role within the networks. Norwegian research organisations have found it difficult to collaborate with user partners and they have struggled to include high-quality industrial partners in their networks (Piro et al. 2016).

Thus, a core–periphery structure tends to characterise the EU innovation networks. Network structures generally coincide with the socio-economic characteristics of regions (i.e. more advanced, innovative, and competitive regions make up network cores) (Calignano 2014; 2017). Innovation activities funded under the FPs are mainly concentrated in more advanced EU regions (Hoekman et al. 2013; Calignano & Hassink 2016; Dotti & Spithoven 2018). Successful applicants typically have a strong scientific reputation and previous experience from participating in EU framework programmes (Enger & Castellacci, 2016). Norway is rarely part of this core (Barber & Scherngell 2013). Norwegian participation tends to follow the same patterns as other peripheral countries, linking frequently to core countries (Piro et al. 2016).

How will environmental programmes mobilise Norwegian regions?

We expect two different influences on the spatial distribution of EU funding for green innovation: First, the need to address grand challenges and the spatially embedded nature of these challenges would mobilise regions where the need for restructuring is more severe. Second, the competition for research funding would imply that regions with strong research capacity are in a better position to attract funding. This would also have implications for path development. Funding directed towards firms in related industries that need to restructure, mainly supports branching. Meanwhile, funding for regions with a strong research capacity, possibly directed towards research establishments in such regions, can mainly support new path creation. As the chances of successfully restructuring are higher with a branching approach (Neffke et al. 2011; Boschma et al. 2013), path creation would be a risky strategy for restructuring. If regional industry is not involved in R&D programmes, research establishments may become ‘cathedrals in the desert’ that struggle to disseminate technologies.

In the Introduction, we raised three key questions: Which regions participate in the framework programmes? What international networks do the participating regions establish? Are there differences in which types of organisations participate?

We expect the most developed core regions to participate in the framework programmes more frequently than the regions that make up the periphery of the network. This expectation follows earlier research, which has shown that such regions attract the majority of the EU funds (Hoekman et al. 2013; Calignano & Quarta 2015). Furthermore, we expect research capacity and the need for restructuring to be important drivers of participation, reflecting path creation and branching, respectively. Second, we expect regions to connect mainly to core EU regions, following the core–periphery structure of the EU FPs (Balland et al. 2013). For path creation, leading European university regions would be important partners. For branching, connecting to regions with related knowledge capacities would be important. Lastly, some authors highlight that research establishments occupy a central position in the European R&D network (Roediger-Schluga & Barber 2008), while others argue that private companies represent more central nodes in the network, especially in core regions (Calignano 2017). The former would be more important for path creation, while the latter are more important for branching.

The geographical context

Norway is facing a pressing need to restructure its economy away from its reliance on oil and gas exports. The long-term challenges of the greening of the economy imply lower demand for the aforementioned exports in the future. However, recent short-term developments have made the issue more acute. The 2014 drop in oil and gas prices had a negative impact on the economy. Norges Bank notes that ‘the Norwegian economy is facing new challenges. Vulnerabilities established during the golden years must be addressed. From being in a unique economic position, Norway is now headed for a period of restructuring’ (Norges Bank 2015). Technological overlap between renewable energy and both oil and gas suggests that there is potential for branching towards greener industries, and firms in the oil and gas industry are to some extent diversifying towards renewable energy (Steen & Weaver 2017; Mäkitie et al. 2018a).

In this article, we focus on Rogaland, Hordaland in south-west Norway and and Sør-Trøndelag in central Norway. While similar in population size and position in the urban hierarchy, they differ in their dependence on oil and gas, as well as in the strength of their R&D systems (Gunnes et al. 2017). Analysis of their presence in the EU’s environmental programmes and their international networks can provide an indication of how international green innovation policy mobilises actors in different types of regions.

In 2014, c.330,000 people – 13% of the workforce – worked in activities directly or indirectly related to petroleum (Blomgren et al. 2015). Together, Rogaland (99,200) and Hordaland (56,700) accounted for nearly half of this percentage, corresponding to 40% and 21% of their workforce, respectively (Blomgren et al. 2015). Sør-Trøndelag (10,300 or 6%) had a much lower level of petroleum-related employment (Blomgren et al. 2015). Rogaland and Hordaland were hit harder by the fall in oil prices. From 2008 to 2016, unemployment grew from 1.1% to 4.5% in Rogaland and from 1.6% to 3.4% in Hordaland, but only from 2.0% to 2.3% in Sør-Trøndelag (Statistics Norway 2017).

Norway is characterised by a strong concentration of R&D activities. In 2015, four counties (Oslo, Sør-Trøndelag, Akershus, and Hordaland) accounted for 70% of R&D expenditures and 78% of university R&D (Gunnes et al. 2017). In 2015, the total R&D investments in Rogaland were less than half of those in Hordaland and a third of those in Sør-Trøndelag (Gunnes et al. 2017). Sør-Trøndelag had the highest R&D investments per capita of any Norwegian county in 2015 (NOK 31,227) compared with NOK 13,267 in Hordaland and NOK 7046 in Rogaland (Gunnes et al. 2017).

The Norwegian Government has established several innovation policy programmes to support the transition towards a green economy. This includes support for environmental technology by Innovation Norway, the establishment of an investment company for renewable energy technology, and the establishment of research centres for environment-friendly energy research (Government of Norway 2017). Nonetheless, the amount spent on petroleum-related R&D was more than three times higher than that spent on R&D for renewable energy in 2015 (Engedal et al. 2017). In the two major environmental programmes funded by the Research Council of Norway over the years 2014–2017 (Klimaforsk and Energix), Sør-Trøndelag was highly successful in attracting funding (NOK 637.4 million or EUR 70.8 million). Despite their restructuring needs, Hordaland and Rogaland performed much worse. Hordaland attracted NOK 67.1 million (EUR 7.1 million) and Rogaland only NOK 12.3 million (EUR 1.3 million).¹ Furthermore, 6 of the 11 research centres for environment-friendly energy research are located in Sør-Trøndelag (4 at the Norwegian University of Science and Technology and 2 at SINTEF (an independent research organisation for industry and technology), while the remaining five are located in the adjacent counties of Oslo and Akershus.

Data and methods

In this article, we focus on FP6 (2002–2006), FP7 (2007–2013), and Horizon 2020 (2014–2020) in order to identify potential changes in participation levels and network formation over time. In each FP, we examine the main green innovation programme, namely the sustainable development programme in FP6, and the environment and energy programmes in FP7 and Horizon 2020.² We include all projects funded by the programmes that involve at least one Norwegian partner organisation. For all partners, we code their location (by Norwegian county or EU NUTS2 region (Eurostat n.d.)) and the type of organisation (higher education establishment, research institute, private company, public organisation, government, or other organisation). Information on the type of organisations is available only for FP7 and H2020.

First, we examine which regions participate most frequently in the programme in a regression analysis. We combine data on FP participation with Statistics Norway data on oil and gas employment and data on regional research full-time equivalents (FTEs) from the Nordic Institute for Studies in Innovation, Research and Education (NIFU). Oil and gas employment is a proxy for the regional need for restructuring. Regions with a high share of employment in oil and gas face stronger pressures to restructure towards green industries. This variable may also capture other influences, such as firms’ resource endowment, but it is the best available measure of restructuring need. Research FTEs is a proxy for the region’s research capacity.

We construct a panel data set of 18 counties over 10 years (N = 180), from 2004 to 2014.³ We control for population size and regional GDP per capita, and fit the following logit regression model, which analyses whether region r starts participating in a new project in year t as a function of four factors:

$Logit\left(P\left(Participatio{n}_{rt}=1|x\right)\right)=ResearchFTE{s}_{rt-1}+Oildependenc{e}_{rt-1}+Siz{e}_{rt-1}+GDPperca{p}_{rt-1}+\tau +\epsilon$

where Research FTEs is the number of FTEs working in research in the county, Oil dependence is the share of employees working in the oil industry, Size is the number of inhabitants, and GDP per cap is the regional GDP per capita. We also include dummy variables for each year, $\tau$, and an error term,$\epsilon$. The dependent variable is binary, all continuous variables are log-transformed, ln(var + 1), and we use robust standard errors. All controls are lagged one period. We subsequently extend the analysis with a multinomial logit regression to consider also participation levels. Due to the heavily skewed distribution of the number of projects, we use a categorical variable distinguishing between regions that do not participate, and regions with low (1–3 projects), medium (4–7 projects), and high participation (8 or more projects), respectively. Year dummies were dropped from the analysis because of the limited number of observations in each group.

Second, we examine the networks that the Norwegian regions (i.e. the case counties) create in the programme, using social network analysis techniques. These analyses build on a complete network of all Norwegian counties’ participation. A link means that the counties collaborated in the same project. Since we did not map the linkages between the other EU regions, this is an ego network analysis for all Norwegian counties. An ego network is a network based on the connections from a single node (‘ego’) (Hanneman & Riddle 2005). We analyse regional, national and international connections to determine the geography of linkages. We focus specifically on whether differences in the research capacity of Rogaland, Hordaland, and Sør-Trøndelag influence the participation of organisations in these regions. Regions with leading research organisations may connect more easily to the EU collaboration network, but may also struggle to develop links to industry and other user partners.

The analysis does not consider other important factors promoting ‘green’ restructuring in Norway (e.g. alternative national funding schemes, increase in internal R&D expenses, new firm births in the ‘green’ sector). However, the volume and geographical distribution of national funding in the field of green technologies, renewable energy, and other green issues are discussed in the preceding section, ‘The geographical context’. We do not have any information on the contents or success of the related projects launched by the Norwegian Government. Hence, we do not know to what extent they actually promote green restructuring, but can only assess their potential.

Furthermore, only partial data is available for the ongoing H2020 programme, specifically for the period 2014–2017. This allows us to assess changes in participation following the oil crisis from 2014. While the lapse of time is limited, several projects started in 2015 and 2016, allowing some time to build research groups, submit applications and start the projects. We count projects in their start year, giving time for projects to emerge. Norwegian oil companies respond quickly to market changes by reorienting towards new markets. For instance, the number of engagements in offshore wind doubled from 2014 to 2015 as the oil price fell (Mäkitie et al. 2018b). Uncertainties within the oil and gas sector influence their interest in new sectors (Steen & Weaver 2017). Nonetheless, the results from H2020 are tentative and participation patterns may shift by the end of the programme.

Norwegian organisations in the environmental programmes

Overall, Norwegian organisations participate frequently in all of the programmes considered in this article (Supplementary Table 1). The share of Norwegian participation is similar in other programmes, such as FP6-SME and FP7-SME, the purpose of which was to enhance the innovation capacity of European small and medium enterprises (SMEs) (Calignano & Hassink 2016). In absolute terms, Norway is ranked 11th in FP6 (330 participations), 9th in FP7 (346 participations) and 15th in H2020 (150 participations). Surprisingly, the share of Norwegian organisations involved dropped from 3.1% in FP6 and FP7 to only 2.0% in H2020. Seemingly, the recent oil price decline did not trigger any immediate increase in Norwegian participation in programmes for green restructuring. This is contrary to our expectation based on the article by Geels & Schot (2007), namely that the external shock of the oil price decline would direct more resources towards green restructuring.

Research establishments (universities, institutions in the higher education sector, and research centres) make up 72% of the Norwegian organisations involved in FP7 and 58% in H2020 (Table 1). Private companies are much less involved in the FPs than are research establishments. The involvement of public bodies increased from FP7 to H2020 (Table 1). The involvement of private companies is considerably higher in core European countries, especially in H2020 (Table 1). The environment and energy programmes have mobilised industrial actors in Norway to a lesser extent than in many other countries. Instead, they have mainly mobilised Norwegian research establishments. Norwegian public sector participation is comparatively high. Hence, the implementation of the programmes in Norway is more suited for path creation than branching, as it relies on new research findings to lead green technology development.

Table 1. Type, number and share of participation in environmental programmes in FP7 and H2020 – Norway and top-five countries

		GOV*	HES	PRC	PUB	REC	OTH	Total
Norway	FP7	-	70 (20.2%)	78 (22.5%)	15 (4.3%)	179 (51.7%)	4 (1.2%)	346 (100%)
	H2020	-	28 (18.7%)	37 (24.7%)	25 (16.7%)	59 (39.3%)	1 (0.7%)	150 (100%)
	FP7	-	322 (22.7%)	433(30.5%)	31 (2.2%)	557 (39.3%)	76 (5.4%)	1419 (100%)
Germany	H2020	-	127 (14.9%)	335 (39.2%)	38 (4.4%)	256 (30.0%)	98 (11.5%)	854 (100%)
	FP7	29 (2.4%)	577 (57.6%)	293 (23.4%)	75 (6.2%)	185 (15.4%)	43 (3.6%)	1202 (100%)
UK	H2020	16 (2.2%)	223 (31.0%)	289 (40.0%)	52 (7.2%)	60 (8.3%)	82 (11.4%)	722 (100%)
	FP7	-	179 (18.9%)	351 (37.0%)	61 (6.4%)	338 (35.7%)	19 (2.0%)	948 (100%)
Spain	H2020	-	90 (10.9%)	349 (42.3%)	107 (13.0%)	228 (27.6%)	52 (6.3%)	826 (100%)
	FP7	8 (0.9%)	220 (24.8%)	270 (30.5%)	57 (6.4%)	316 (35.7%)	15 (1.7%)	886 (100%)
Italy	H2020	4 (0.5%)	129 (17.8%)	320 (44.3%)	76 (10.5%)	144 (19.9%)	50 (6.9%)	723 (100%)
	FP7	-	246 (28.7%)	257 (30%)	51 (6.0%)	208 (24.3%)	94 (11.0%)	856 (100%)
Netherlands	H2020	-	88 (18.5%)	197 (41.4%)	51 (10.7%)	77 (16.2%)	63 (13.2%)	476 (100%)

Notes: *GOV – Government, HES – higher education sector, PRC – private for-profit companies, PUB – public organisations, REC – research centres, OTH – other types of organizations

Which regions participate in the programmes?

Table 2 shows the results of the regression analysis examining which regions participate in the programmes. Model 1 includes the total research FTEs, whereas Model 2 splits research FTEs into FTEs in research establishments and in industry. Participation is significantly and positively associated with region size and oil dependence. Research capacity overall does not significantly affect the likelihood of participation. However, research FTEs in research establishments are strongly positively related to participation (Model 2). Meanwhile, research FTEs in industry has no significant effect on participation.

Table 2. Factors influencing participation of the Norwegian regions: Logit regression model (Standard errors in parentheses; Significance level: * p < 0.10, ** p < 0.05, *** p < 0.01)

Variables	(1)	(2)
Research FTEs	0.583
	(0.451)
Research FTEs in		0.932***
research establishments		(0.252)
Research FTEs in		-0.418
industry		(0.275)
Oil dependence	0.281**	0.451***
	(0.133)	(0.156)
Population size	1.381*	2.591***
	(0.812)	(0.912)
Regional GDP per capita	1.676	-0.800
	(1.845)	(1.891)
Year controls	YES	YES
Constant	-42.81*	-26.96
	(24.17)	(19.25)
pseudo R^2	0.380	0.427
N	180	180

Table 3 shows the results of the multinomial regression analysis. Some differences occur between the factors influencing participation in general and participation at high levels. In this analysis, we only show the results from Model 2, splitting research FTEs into research establishments and industry. Supplementary Appendix 1 shows marginal effects from the analysis, which indicates that research FTEs, in research establishement as well as in industry, are significantly positively associated with high participation. Furthermore, research FTEs in REs are significantly negatively associated with no participation, while research FTEs in industry are negatively associated with low participation. Oil dependence is significantly negatively associated with no participation, and positively (at the 10% level) with medium participation, but not with high participation.

Table 3. Factors affecting participation levels of the Norwegian regions in environmental programmes in FP7 and H2020 (standard errors in parentheses; significance level: * p < 0.10, ** p < 0.05, *** p < 0.01)

	Multinomial logit regression model (baseline: no participation)
	Low		Medium		High
Variables	coeff.	std.err.	coeff.	std.err.	coeff.	std.err.
Research FTEs in research establishments	0.659**	(0.287)	3.211***	(0.872)	5.581***	(1.646)
Research FTEs in industry	-0.342	(0.249)	4.975**	(2.036)	9.031***	(2.288)
Oil dependence	0.388**	(0.153)	2.222***	(0.773)	2.547***	(0.931)
Population size	2.206***	(0.836)	-10.11*	(5.625)	-16.60***	(6.285)
Regional GDP per capita	-2.171*	(1.143)	-5.843**	(2.736)	-10.43***	(3.495)
Constant	-4.164	(12.97)	133.5**	(63.39)	228.4***	(72.88)
pseudo R^2	0.519
N	180

The results support the expectation that regional research capacity and the need for restructuring are associated with participation in the three studied framework programmes. For research capacity, participation is mainly driven by capacity in research establishments, rather than in industry. However, research capacity in industry is associated with high participation. Overall, this supports Enger & Castellacci’s finding that scientific reputation is important for attracting EU funding (Enger & Castellacci 2016). Regions with a higher restructuring need are able to attract funding for restructuring projects, albeit not at the highest levels. Hence, the distribution of projects reflects potential for both branching and path creation, with the latter having the greatest potential in this respect.

Connections established by Norwegian organisations: an ego network analysis

Table 4 presents the social network analysis of Norwegian counties’ participation in the FPs. Their innovation network in FP6 has low density (0.015, i.e. 1.5% of potential connections). Density is the total number of linkages in the network divided by the possible number of linkages (Hanneman & Riddle 2005). The low density is reasonable as the network we have constructed is an ego network, and suggests that Norwegian regions combined were able to connect with many different regions through the programme. Overall, 16 Norwegian counties and Svalbard were active in FP6, establishing linkages with 254 different European NUTS2 regions and 38 non-EU countries. On average, each participating county connected to 4.8 other regions. Organisations in the core regions, such as Oslo, Akershus, Sør-Trøndelag, and Hordaland, participated the most in FP6, compared with the remaining counties in Norway. Their partners were often core regions too, typically capital regions. Hence, participation patterns follow a core-periphery structure. However, the specific regional profiles associated with path creation (leading university regions) or branching (regions with specific complementary knowledge capacities) do not show up prominently in the network. The network structure remains largely the same in the FP7 network.

Table 4. Linkages in FP6, FP7 and H2020; in-degree – collaborations between Norwegian counties and NUTS2 regions on the one hand, and extra-EU countries on the other hand; out-degree – number of projects and collaborations with partners in other NUTS2 regions and extra-EU countries of the participating Norwegian counties; out-degree is the number of linkages from one specific node to other nodes, while in-degree is the number of linkages from other nodes to a specific node

Norway
	FP6		FP7		H2020
Active Norwegian counties	16 + Svalbard		15 + Svalbard		12 + Svalbard
Active EU NUTS2 regions	254		210		176
Active non-EU countries	38		50		24
Density	0.015		0.017		0.008
Average number of links	4.8		4.7		3.0
Degree centrality	Region		Region		Region
Out-degree	Oslo	1862	Oslo	1514	Oslo	747
	Akershus	1588	Sør-Trøndelag	1358	Sør-Trøndelag	489
	Sør-Trøndelag	1265	Akershus	1229	Hordaland	380
	Hordaland	785	Hordaland	642	Akershus	339
	Rogaland	542	Rogaland	561	Rogaland	113
	Troms	430	Troms	281	Troms	104
In-degree
Norway	Sør-Trøndelag	122	Sør-Trøndelag	96	Hordaland	42
	Oslo	92	Oslo	82	Oslo	26
	Akershus	75	Akershus	75	Akershus	23
	Hordaland	74	Hordaland	50	Sør-Trøndelag	20
	Rogaland	72	Rogaland	46	Rogaland	10
EU NUTS2	Île de France	453	Île de France	317	Île de France	100
	South-Holland	211	South-Holland	170	Lazio	80
	Lazio	176	Brussels	153	Madrid	74
	Brussels	163	Lazio	149	Brussels	72
	Attika	159	Madrid	139	Catalonia	61
	Madrid	146	London	137	South-Holland	58
	London	143	Midtjylland	123	Lisbon	46
	Hamburg	143	Oberbayern	116	Helsinki	45
	Copenhagen	137	Swindon/Wiltshire	113	Copenhagen	43
	Helsinki	129	Catalonia	97	London	38
Other countries	Russia	89	Russia	43	Russia	21
	China	60	Turkey	41	USA	19
	Turkey	53	USA	38	Turkey	15
	India	32	Canada	30	Canada	13
	USA	29	Israel	29	China	10
Betweenness	Sør-Trøndelag	850	Sør-Trøndelag	1094	Oslo	626
	Akershus	796	Oslo	899	Sør-Trøndelag	453
	Oslo	649	Akershus	538	Hordaland	316
	Hordaland	370	Hordaland	298	Akershus	277
	Aust-Agder	144	Rogaland	212	Rogaland	68

Norwegian participation is lower in H2020 (see the section ‘Norwegian organisations in the environmental programmes’). Three years after the launch of H2020, the network density and the average number of links had dropped and the number of participating counties had fallen. This may change when H2020 funds more projects. However, fewer Norwegian counties have been able to engage in the green restructuring programmes in H2020 than in previous FPs, and their international networks are less extensive. Hence, the drop in oil prices has not triggered any immediate increase in networking activities through the programme.

Among our case regions, Sør-Trøndelag and, to a lesser extent, Hordaland are among the most active regions. Conversely, Rogaland’s participation is more limited. The economic shock caused by the oil price decline did not provoke major changes in the Norwegian regions’ level of activity in the more recent FP (see Supplementary Tables 2, 3, and 4 for detailed figures related to the participation of organisations located in the three Norwegian counties under analysis).

Types of organisations involved

Besides the differences in participation levels, the types of organisations involved also differ between Sør-Trøndelag, Hordaland, and Rogaland. Private companies are the most active organisations in Rogaland in all of the FPs. In FP6, private companies represent 17 out of 24 cases. The share of private firms is also high in FP7 and H2020 (18 of 24 in FP7 and 7 of 12 in H2020). The national oil company Statoil (now Equinor) was the most active organisation, with 10 participations in FP6 and 17 in FP7, but only 3 in H2020. The participation of the research establishments in Rogaland is limited. The University of Stavanger participated twice and the research institute IRIS six times.

Conversely, the University of Bergen and other research establishments are the most active organisations in Hordaland. They account for 21 of 32 participations in FP6. The share of participation by research establishments located in Hordaland increased further in the transition from FP6 to the two more recent FPs under analysis: 39 of 42 in FP7, and 21 of 24 in H2020. Hordaland shows a very different pattern from Rogaland, relying much more on research establishments and hardly on private firms.

Sør-Trøndelag’s participation has a similar profile to that of Hordaland. In FP6, NTNU and other research establishments in the county participated 68 out of 85 times. This share considerably increased in FP7, when the number of participations of such organisations was 100 of 104. Similarly, research establishments made up 40 out of 43 participations in H2020. As in Hordaland, the number of private firms decreased in later programmes compared with FP6. Rogaland was among the regions with which Sør-Trøndelag collaborated the most in FP6 and FP7. The finding can be explained by collaborations between private firms in Rogaland and research establishments in Sør-Trøndelag, and it reflects the pattern described by Narula (2002), of private firms in dominant industries collaborating with leading national research institutions.

Overall, there are differences in how Norwegian counties engage with the EU’s restructuring programmes. In counties with strong research establishments, participation in the programmes takes place mainly through those organisations. In both Hordaland and Sør-Trøndelag, research establishments have become increasingly dominant over time. In these regions, the programmes can mainly foster new path creation through scientific breakthroughs and commercialisation of research. However, a more industrial county such as Rogaland, where the need for restructuring is pressing, is able to engage with the programmes in other ways. There, industry has a much more central role, including large firms in incumbent industries acting as gatekeepers. These large firms create larger opportunities for bricolage and branching in such regions, as knowledge from international networks can be combined with existing industrial knowledge to create new combinations. This potential can only be realised if the firms that participate in the programme, especially from incumbent industries, transform their activities and develop new business models.

Among the three counties, Sør-Trøndelag participated in the largest number of projects, while Rogaland participated the least. Hence, EU funding more strongly mobilised counties with more mature R&D systems. The greater transition challenges in Rogaland are not sufficient to mobilise participation. Research establishments in Rogaland, being smaller and less mature, have hardly participated in the framework programmes analysed in this article. This has left participation largely up to private companies in Rogaland, who have mainly linked up with research establishments outside the region.

Discussion of the results

In this article, we have examined Norwegian counties’ participation in the EU’s three most recent FPs specifically addressing environmental issues. Participation is mainly associated with research capacity at universities and research institutes. As in other EU FPs, scientific excellence attracts funding (Enger & Castellacci 2016). Additionally, oil and gas employment is associated with participation, potentially reflecting that regions facing large restructuring challenges mobilise for these programmes. Hence, participation patterns reflect the potential for branching as well as new path creation. Norwegian regions mainly link with core EU regions, following the core–periphery structure identified in other EU programmes (Balland et al. 2013). The projects mainly involve research establishments in counties with higher research capacity (e.g. Sør-Trøndelag and Hordaland), but also private firms in more oil-dependent regions (e.g. Rogaland). Hence, the implementation of green restructuring policy differs, depending on the regional context. It follows a more R&D-oriented approach based on path creation in regions with a strong initial endowment of research capacity. In regions facing major restructuring challenges, the result is a more industrial-based approach aiming for branching, involving firms as the main actors (Martin 2010; Tödtling & Trippl 2013; Isaksen 2015). However, the former dominates in terms of the number of projects.

The fact that the projects funded under the three analysed FPs mainly involve regions with comparatively higher research capacity and more mature R&D systems has the potential to develop niches outside the current sociotechnical regime. However, it also risks reinforcing existing patterns and may delay the transition of more oil-dependent regions into the green economy. The net effect might be a slower transition of the Norwegian economy as a whole. Alternatively, the oil-dependent regions might be left behind and further locked-in to path extension. The promotion of path creation through basic research is a high-risk strategy that may fail to produce substantial changes throughout the economy (Neffke et al. 2011; Boschma et al. 2013). When research establishments dominate, the opportunities for bricolage (Garud & Karnøe 2003) are limited and technological change may fail to produce broader changes in the sociotechnical regime (Geels & Schot 2007; Markard & Truffer 2008).

The predominant position of research establishments in the FPs, which consequently limits the potential for changes in the sociotechnical regime, raises the question of how Norwegian innovation policy can be adjusted to complement EU policy and enable Norwegian organisations to benefit more from EU policy instruments. There is a need for instruments that target firms and industrial actors, moving away from the current reliance on universities to drive the transition towards a greener economy. This requires a combination of three approaches. First, policy must support the R&D capacity of Norwegian firms to enable them to participate in the programmes. Norwegian domestic innovation policy prioritises collaboration to the detriment of internal knowledge development in companies, with the effect of reducing companies’ capacity to collaborate meaningfully in R&D projects (Herstad et al. 2010). The lack of capacity makes it hard for them to collaborate at a distance (de Jong & Freel 2010) and to engage in international innovation programmes, as the results of the current study reflect.

Second, Norwegian universities need incentives and assistance to identify suitable industrial partners and include them in their consortia (Piro et al. 2016). The universities are already highly embedded in the programmes and they can provide firms with access to their own networks (Benneworth & Hospers 2007). This would also provide them with a route to impact. The combination of the two factors could help to bring the participation levels of Norwegian firms up towards those of other countries.

Third, Norwegian policymakers must also look beyond R&D to identify mechanisms that make it attractive for non-R&D active firms to invest in green restructuring. Norwegian restructuring policy tends to be R&D focused and often does not fit firms with more experience-based approaches to innovation (Steen & Hansen 2018). Policies to support branching from existing industries are essential for path diversification in Norwegian regions (Brekke 2015).

Overall, the fact that the Norwegian restructuring policies tend to be predominantly R&D oriented and often neglect the importance of experienced-based approaches to innovation involves lessons for policies to support green restructuring more broadly. The analysis highlights the tensions either between branching and path creation (Martin 2010; Isaksen 2015) or between breakthrough and bricolage (Garud & Karnøe 2003) in green restructuring policies. So far, the restructuring debate has not incorporated perspectives from evolutionary economic geography to a great extent (Hansen & Coenen 2015; Boschma et al. 2017) and the roles of branching and new path creation in fostering restructuring are somewhat unclear. However, policy to support restructuring must strike the right balance between the two processes. Restructuring policies that place too much faith in path creation through basic and applied science may struggle to mobilise actors in peripheral regions or regions without a strong science base (Tödtling & Trippl 2005; Isaksen et al. 2018). They may also fail to reach out to industry. Therefore, there is a need to think seriously about how processes of regional branching and path diversification can be unleashed in the pursuit of green restructuring.

Conclusions

We have examined the participation of Norwegian counties in the EU environmental programmes, and have analysed the organisations involved and the characteristics of their international networks with the aim of revealing how regional R&D capacity and the need for restructuring influence participation. Regions with higher R&D capacity, such as Sør-Trøndelag and, to a lesser extent, Hordaland, are more active than more oil-dependent regions, such as Rogaland. The types of organisations involved differ, depending on the characteristics of the regions under analysis. Research establishments dominate in regions with the most developed R&D system, and private companies are more involved in oil-dependent regions. Hence, the implementation of the programmes depends on regional characteristics, resulting in different potentials for stimulating new path development. In research-intensive regions, the programme can mainly foster green restructuring through new path creation. In regions with a greater need for restructuring, the potential for branching is higher.

ORCID

Giuseppe Calignano http://orcid.org/0000-0002-6885-768X

Rune Dahl Fitjar http://orcid.org/0000-0001-5333-2701

Notes

1 See the Research Council of Norway’s website for further details: https://www.forskningsradet.no/prosjektbanken/#/Sprak = en (accessed December 2018).

2 The following three programmes were included in the analysis: Sustainable Development, Global Change and Ecosystems (FP6-SUSTDEV), FP7 (FP7-ENVIRONMENT and FP7-ENERGY) and H2020 Environment and Energy (H2020-ENVIRONMENT and H2020-ENERGY). These programmes address similar topics, such as renewable energy and environmental technology, reflecting continuity in EU environmental policy.

3 Vest-Agder and Aust-Agder are considered as one region in this analysis, because data on regional research capacity is only available for these counties jointly.