The geography of innovation literature has recently shifted from a dominant focus on the rate and determinants of innovation to tackle issues concerning its direction and societal relevance (Uyarra et al., 2019; Mazzucato, 2020). The emerging focus is not primarily on analysis of innovations and their processes per se, but rather on how these can help to solve ‘wicked' problems or “grand challenges” (Mazzucato, 2020) and/or reduce environmental risks (e.g., global warming, pollution, threat of nuclear war), which, despite being caused by economic growth in modern societies (i.e., capitalism), are considered manageable by continuous scientific and technological development and innovation, or “technological fixes” (i.e., applying technology to solve problems). Moreover, while this literature has traditionally emphasized the roles of firms, academia, and government, it has recently begun attending to how new innovations are created by intentional actors with agency—defined as “the ability of people to act and have an effect” (Gregory et al., 2009, p. 347)—aiming to stimulate new economic activity and influencing and changing formal and informal institutions¹ (Carvalho and Vale, 2018, Grillitsch and Asheim, 2018). Isaksen et al. (2019) approach this dualism by arguing that regional actors have the capacity to enact both “firm agency” and “system agency,” where the latter is how both nonindustrial and industrial actors aim to change system conditions (e.g., institutions) to stimulate regional development.

In line with this thinking, the main focus of this article is not to primarily explain institutions or gradual institutional change per se (see e.g., Mahoney and Thelen, 2009 for a discussion here), but rather agency processes where actors could influence technological and institutional change, either intentionally or unintentionally. Following up this ambition, we adhere to the assertion made by Grillitsch and Sotarauta (2020) that institutional entrepreneurship, Schumpeterian innovative entrepreneurship, and place-based leadership—the trinity of change agency—are important in regional innovation processes. However, as emphasized by Bækkelund (2021), this perspective has been less able to account for the contribution of agency to path dependence, suggesting that reproductive agency (i.e., how agents and their actions contribute to incremental change that maintain existing structures, industry practices and institutional logics) is also important. Thus, change agency need not be defined at the individual level, but can also be a form of collective agency enacted by formal or informal networks or actor groups (Geels, 2020).

Institutional entrepreneurship is conceptualized by Grillitsch and Sotarauta (2020, p. 708) as “actions that are directed toward transforming existing or creating new institutions”, and describes how actors work to leverage “particular institutional arrangements and mobilize resources, competences, and power to create new institutions or to transform existing ones” (Sotarauta and Pulkkinen, 2011, p. 98). Moreover, the literature has increasingly focused on how actors work with ‘missions' in industry restructuring (Simmie, 2012; Steen, 2016; Miörner and Trippl, 2017; Hassink et al., 2019), particularly in relation to environmental sustainability (Fløysand et al., 2022). Institutional entrepreneurs can be found both within the public and private sector and generally involves the development of broader socio-political engagement and alignment around both informal and formal institutional change (Pacheco et al., 2010). This can for example occur through “framing” or “lobbying” processes (Woolthuis et al., 2013), which can help legitimize and empower emergent regional paths or technologies (Grillitsch and Sotarauta, 2020). However, reproductive agency is also present both insofar as it is in many actors' interests to resist institutional change due to hegemonic positions in the regional economic system (MacKinnon et al., 2019) and because reproductive agency related to institutional entrepreneurship for green industrial restructuring can unintentionally carry practices that (de)legitimize greening (Fløysand et al., 2022).

The trinity of change agency also includes Schumpeterian innovative entrepreneurship, defined as “willful attempts to realize novel combinations of knowledge and resources coupled with the search for a not-yet-realized potential” (Grillitsch and Sotarauta, 2020, p. 708). This agency is thus initiated by firms, either alone or cooperatively, that engage with “innovation” and “technological development” to introduce new products and services to the market (Shane and Venkataraman, 2000). This can occur when new products or processes come about, or not, through upscaling successful demonstration projects toward more sustainable solutions for reducing environmental risk. Reproductive agency in this case would be defined as incremental changes that follow a path-dependent innovation trajectory in which firms replicate existing innovation. When creating new paths to reduce environmental risk, replicative innovations will not drive new paths. However, an industry with the ability to quickly absorb and reproduce innovation when new paths are created is considered positive for path evolution. Thus, change and reproductive agencies caused by Schumpeterian innovative entrepreneurship have different roles, depending on the path evolution stage.

Finally, the trinity of change agency also includes place-based leadership, defined as agency that aims “at transforming particular places by pooling competencies, powers and resources to benefit both agents' individual objectives and a region more broadly” (Grillitsch and Sotarauta, 2020, p. 708). According to Sotarauta et al. (2017, p. 212), place-based “leaders as individuals, and groups of individuals, tend to possess a greater range and depth of assets–including commitment to advancing the region–than other actors”. This individual- or collective-based leadership form coordinates regional development efforts among stakeholders. Place-based leaders can be individuals or actor collectives working to inspire and direct heterogenous actors toward a shared vision or direction. This means that place-based leadership can be connected to both formal authorities and “informal leaders” (e.g., philanthropic business leaders who lead through charisma and inspiration rather than through “formal” means). It can also be connected to reproductive agency if place-based leadership processes can maintain regional agency patterns that uphold existing production or innovation practices, or, at best, very little change from the status quo. Thus, societal pressure to advance regional sustainability does not necessarily lead to radical changes in leadership orientation and industrial practices. However, path-dependent (i.e., reproductive) place-based leadership can also decrease environmental risks in those instances where such processes lead to e.g., sustainable governance (by regulatory and industrial actors) which can prevent a “tragedy of the commons” (Ostrom, 1990).

In sum, we define transformative capacity as an outcome of the interplay between change agency and reproductive agency among all three layers of the trinity of change, which play out on different scalar levels; Institutional entrepreneurship on the national level, Schumpeterian innovative entrepreneurship on the firm level and place-based leadership on the regional level. We therefore structure our analysis around both a vertical dimension (how the change agencies interact with each other over multiple spatial scales) and a horizontal dimension (how change agency and reproductive agency interact with each other at each scale). While such a dynamic or transformative capacity can also be found in regional industry clusters (see e.g., Njøs et al., 2017), our focus on the interplay on multi-scalar agency still merits an application of our theoretical framework.

Our interviews quickly revealed an awareness of environmental risk and related challenges. However, while the risks assessments among our informants from the public sector and NGOs focused on risks such as infectious diseases and parasites that affect both farmed and wild salmon, water quality degradation from farm discharges, and carbon footprint issues related to soil degradation and deforestation in soya feed production regions abroad (see also Norwegian Ministry of Fisheries Coastal Affairs., 2009; Rainforest Foundation Norway Future in Our Hands., 2018; Føre et al., 2022), the informants from the industry sector focused on innovations challenging the risk such as those linked to electrification of production processes and supply boats, more sustainable feeding systems, etc. (see section Schumpeterian innovative entrepreneurship).

Nevertheless, the interviews revealed how the environmental risk situation had triggered institutional entrepreneurship in the form of policy experimentation on the national level, as also emphasized by others (e.g., Hersoug et al., 2019; Jakobsen et al., 2021), in terms of new national regulations including the “Traffic light system” (2017) (Tveterås et al., 2020) and the DL program (2015) aimed at inducing new industrial technology development (Hersoug et al., 2019; Jakobsen et al., 2021). The traffic light system assigned production zones a red, yellow, or green light based on the lice levels of farmed salmon³ (Figure 1).

In 2015, the Norwegian government issued Development Licenses (DLs) (Føre et al., 2022), i.e., temporary development concessions that promote technology. The purpose of introducing these licenses was to incentivize industrial actors to innovate new sustainable technological solutions; as described by the Directorate of Fisheries (n.d.a):

Although lice reduction is not a specific DL program priority, nearly all approved DL projects aimed to reduce lice levels. A total of 104 applications for 892 licenses⁴ were submitted to the Directorate of Fisheries (Hersoug et al., 2021), of which 24 from all over Norway were approved (Directorate of Fisheries, n.d.b). The fact that a total of 892 licenses “equals 85% of the total number of ordinary licenses” (Hersoug et al., 2021, p. 7) demonstrates the program's popularity and degree to which competition was stimulated among industry actors (Hersoug et al., 2019; Vormedal et al., 2019; Aarset et al., 2020; Føre et al., 2022; Osmundsen et al., 2022). All awarded licenses included measures against sea lice (Føre et al., 2022). As of September 2022, 24 applications for 119.5 licenses, covering a total of 88.919 tons of maximum allowed biomass, have been granted (Directorate of Fisheries, n.d.b).

In relation to the trinity of change agency, the emergence of DLs came about as a process of institutional entrepreneurship on the national level; the effect on reducing environmental risk is clear, as DLs have changed the rules of the game for industry actors toward developing more sustainable production technology:

Most of DL projects in the program are thus technological expressions of environmental risk reduction. Several of them have emerged within offshore, semi-closed, and closed salmon farming technology (Figure 2). However, while DLs have the potential to stimulate green technological development, they do not necessarily lead to industry greening. Nevertheless, if companies attaining DLs accomplish their projects, their licenses can be converted to “normal” licenses even if the project fails at commercialization. This creates additional conditions for growth in total produced biomass within existing technologies (Hersoug et al., 2021) and is thus an unforeseen consequence of institutional entrepreneurship, because it can lead to actors “reproducing” environmental risk. It is important to note that producers have always sought “special purpose licenses” (Hersoug et al., 2021) and that this incentive represents a major firm subsidy, despite high investment costs (Vormedal et al., 2019), given that the cost of a new license today is almost NOK 200 million (NOK 171.4 million in 2020) (Osmundsen et al., 2022). A recent proposition to create “eco-technology licenses” is also worth mentioning, as these would specifically stimulate developing closed or semi-closed technologies (Osmundsen et al., 2022). This is a potential criticism of the DL program, which so far has failed to successfully divert salmon producers' practices away from the open-net pen technologies that play a major role in upholding the very environmental and animal welfare risks the program originally sought to eliminate.

While the emergence of DL projects has contributed to significant technological development in the industry, this does not paint the whole picture of environmental risk reducing efforts that are taking place. Evidence of place-based leadership reducing environmental risk on the regional level was also reported as important for environmental risk reduction by several informants. Exemplifying this, several informants turned to the region of Northern Trøndelag. This region hosts several salmon farming firms, most of which were historically classified as small-and-medium sized businesses (SMBs). These are family firms with a strong community commitment and engagement in regional industry cooperation, who historically have tended to follow collective rather than individual strategies (Jakobsen, 1999). Regional cooperation between firms is common, as emphasized by this informant:

Another informant argued that these patterns of cooperation have created firm economic benefits:

Currently, local industry stakeholders and those from the regional public sector have created new forms of self-organizing, bottom-up collaborations around area cooperation, defined as “a cooperation between two or more private or public actors connected to aquaculture within a defined geographical area” (Karlsen et al., 2019, p. 1). This is because local natural conditions and regional farm agglomerations have created, quite literally, negative “spillover” effects of salmon lice between farms. This effect was revealed by one of our informants:

As a solution, the firms in the region have taken the initiative to swap locations to better control lice spread, creating a self-organized, bottom-up initiative that sought better governance of the commons by applying scientific models and measurements of factors such as how lice spread in currents and in proximity to other fish pens:

Thus, this area cooperation among regional firms occurred along with the regulatory authorities. Another interesting aspect is that this self-organization does not seem to have been forced on the industry. While Karlsen et al. (2019) point out in their nationwide study that some area cooperation is required by law, most cooperation is nonetheless voluntary. They stress that Northern Trøndelag is particularly engaged in area cooperation, underline the presence of a clustering milieu consisting of smaller companies with common challenges as a reason behind this (Karlsen et al., 2019). The long-term collaborative strategies in this region are thus path-dependent and serve as an example of reproductive agency insofar as regional actors' agency maintains existing collaborative structures and explicitly and implicitly upholds incremental changes toward seeking better common governance and, consequently, environmental risk reduction. In this region, reproductive agency can be seen as positive with regard to environmental risk reduction because the collaborative milieu in the region has a history of interdependence and trust. Yet this form of reproductive agency-as-place-based leadership can be challenged on the basis that several merger and acquisition processes have occurred since the 1970s, ultimately leading to the merger of Midt-Norsk Havbruk and Salmonor in 2021, and, most recently, a further merger with Norwegian Royal Salmon and Salmar—one of the largest salmon producers in Norway (Intrafish, 2022). Accordingly, the regional conditions seem to be changing toward an ownership structure in which a concentration of larger companies gains power at the expense of smaller companies. This indicates that the place-based leadership observed to date may be challenged by change agency that does not necessarily seek to improve common governance.

The industry's adaptation to the United Nation's Sustainable Development Goals to reduce undesirable salmon production outcomes has become a common strategy among industrial players. Most firms point to several innovation initiatives that address what the industry perceives to be environmental risks. These include innovations linked to electrification of production processes, supply boats, and more sustainable feeding systems:

Nonetheless, reproductive agency is observed insofar as path-dependent innovation trajectories produce new environmental risks as by-products of an otherwise rather successful development trajectory. Reproductive agency is present in those innovations that carry observable and thoroughgoing environmental risks, and in related fish welfare impacts within the geographical areas hosting salmon farming production:

In the next quote, the dynamics between change agency to eliminate environmental risks and reproductive agency to increase industrial growth and environmental risk (i.e., an awareness of an underlying risk ambiguity) is evident:

The lice problem is a particular focal environmental risk in the industry and is considered by some as a persistent and “wicked” problem (Osmundsen et al., 2017). Many have argued that the greatest environmental risk of lice is not directly related to increased mortality rates among farmed salmon, but rather the risk it poses to the wild salmon population in Norway if farmed lice-infested salmon escape. For farmers, this environmental risk has necessitated increased treatment of the farmed salmon, which then leads to higher mortality rates among these penned salmon (Norwegian Ministry of Trade Industry Fisheries., 2021 p. 15):

To treat salmon lice, the industry applies mechanical delousing, which causes substantial salmon death in the fish farms. As shown in Figure 1, a high mortality rate is a general problem for the industry. In addition, there are large variations in mortality levels across production zones. In Zones 3 and 4, we observe particularly high mortality. In contrast, Zones 1 and 7–13 stand out for their low salmon mortality levels, among which the lowest is Northern Trøndelag (Zone 7; Figure 1). Over time, this zone has displayed low salmon death numbers⁵.

In the following, we argue that these low salmon mortality levels in the region can be partly linked to firm-level Schumpeterian innovation entrepreneurship, exemplified by the regionally embedded DL project Aquatraz. This project was granted four concessions and the right to farm 3,120 tons of farmed salmon per year (Directorate of Fisheries, n.d.b). The new technology in this project has been developed by Seafarming Systems, who in 2016 agreed to build several new semi-closed containment pens for the salmon farming company Midt-Norsk Havbruk (Sandstad et al., 2022). The main aim of this project is to significantly reduce the environmental risk of salmon lice growth by “shielding” the salmon through increased pen water circulation and decreasing salmon escape through sturdier construction or architecture (Figure 3). In so doing, the project has also applied many solutions from the maritime and offshore industry and has had to develop new verification standards for the technology in cooperation with DNV (Sandstad et al., 2022).

Aquatraz and its stakeholders received praise from the media (Hernes, 2021) and interviewed informants, and has delivered on all areas of the project's initial goals, namely to reduce salmon lice, increase growth, improve salmon coloration, eliminate escaped salmon, eliminate personal injuries, and maintain a good work environment (Salmonor, n.d.). It has therefore generated much knowledge on how to reduce environmental risk among involved actors over several project generations (Kyst.no, 2020). The third Aquatraz generation to operate in Eiterfjorden (Northern Trøndelag) also showed significant positive results in 2021 compared with the control pen (Table 1) (with respect to, e.g., costs, sales price, delousing numbers, mortality rates) (Aunsmo and Aunsnmo, 2021)⁶. Now in the fourth generation, involved actors are more focused on commercialization. This long-standing emphasis—in addition to aspiring to future convert of DLs to normal licenses—distinguishes the Aquatraz project from most other DL projects. According to one industry informant connected to the project:

This industry informant also argued that “we have come so far in the run that we are in the last phase and that this actually is a commercial product”. Another unique aspect of the project dynamic is the strong ties between the technology developer (Seafarming Systems) and the salmon producer (Midt-Norsk Havbruk). In contrast to most of our industry informants, this case highlights a particularly strong identification and joint commitment among the project participants—expressed here as “we”—in their efforts to upscale the technology to the market:

Moreover, many extra-regional industry actors hail Aquatraz as a promising solution to persistently difficult industry problems, as emphasized by a key representative for the Norwegian salmon farming industry:

Aquatraz is thus a good example of a technological fix, a solution envisioned by industry actors which can mediate environmental risks. Not only does it exemplify how industrial actors have developed a technological demonstration project that significantly reduces the environmental risk of salmon lice and escaped salmon, it is also an example of how such a fix has the potential to move beyond testing and into a commercial market application stage. Nevertheless, as the interviews revealed, stakeholders connected to Aquatraz are also driven to convert the project DLs to normal licenses. While it remains to be seen whether Aquatraz will represent an emerging change in the dominant industry standard (farming salmon in open pens), it shows significant promise of becoming a viable commercial product. It also shows a high degree of legitimacy in that several industrial (wharf and suppliers who needed to develop, test and coordinate the solutions) and R&D stakeholders (e.g., NIVA measuring water quality, NOFIMA measuring product quality, and SINTEF Ocean providing test and verification data and analyses) have been quite involved, and have had healthy discussions with Midt-Norsk Havbruk. Finally, the main stakeholders connected to the project have in a recent report (Sandstad et al., 2022) suggested that future regulatory specifications should emphasize “functional demands”, for example qualitative assessments with regard to the risk of escaped salmon from sea farming installations (e.g., semi-closed pens), rather than “absolute restrictions for salmon density in pens” (Sandstad et al., 2022). This, it is argued, can (p. 8, our translation): “stimulate to development of new technology which contributes to good animal welfare…and reduces the risk for escaped salmon”, whilst simultaneously allowing the industrial actors to increase (or at least not restrict) their production volumes. As such, this illustrates an emerging attempt by industry stakeholders to influence regulatory stakeholders—“bottom-up”—based on a demonstrated technology.

While the change agency processes described herein represent a departure from conventional salmon farming practices, reproductive agency elements are echoed in many firm-level innovation activities. Although, these have created both Schumpeterian innovations to reduce environmental risk, and incremental innovations to reduce environmental risk, their main target seems to be to increase growth. Accordingly, reproductive agency is an inherent ambition in Norwegian firms' business strategies:

This “inherent tension” between change agency, which attempts to reduce environmental risk, and reproductive agency, which maintains and stimulates economic growth through increased biomass rather than e.g. value-adding through processing or product diversification, has led to several protests and NGO activism. It has also led to a court case in which several salmon farming companies sued the Norwegian state for, in their opinion, wrongly depicting the cause of high pen lice levels that halted salmon production. Most stakeholders (e.g., researchers, NGOs, environmentally conscious politicians) claim that the lice problem is an inherent consequence of a volume-oriented production system that deprioritizes environmental risks. On the one hand, salmon farming is considered important for economic growth, through increased volumes; on the other hand, the volume-oriented growth strategy leads to stakeholder environmental risk concerns:

In sum, it appears that both institutional entrepreneurship and Schumpeterian innovative entrepreneurship, involves both implicit change and reproductive agency. Reproductive agency can also be seen in actions that reduce environmental risk, such as sustainable commons governance through place-based leadership. However, this may be challenged by foreign ownership and leadership constellations that might steer in a direction where the consequences for particular places is at core. This can be explained by an overarching industrial growth logic (i.e., reproductive agency) in an industry that does not “accept” biomass reduction that would otherwise go a long way toward reducing the lice problem.