Bouncing forward better – micro-foundations of combinatorial innovation

Abstract

This article theoretically investigates the conditions that empower actors (entrepreneurs and intrapreneurs) to pursue combinatorial innovation processes, and thereby contribute to transformative activities with a high degree of novelty and complexity. We argue that the possibility for combinatorial innovation depends on agents’ time perspective and combinatorial capabilities. We elaborate why institutional incentive structures and knowledge bases condition actors’ time perspective and why network structures and firm routines condition actors’ combinatorial capabilities. We illustrate our arguments with examples and show their practical relevance for policy and innovative actors.

Keywords:

• combinatorial innovation
• sustainable development
• innovative entrepreneurship
• firm adaptability

JEL:

• -Codes: L50, O10, O30, P50

1. Introduction

While the economy bounced back quickly after Covid-19, we need to bounce forward better – this means rapidly and radically changing society and the economy in order to address grand challenges such as global warming, mass extinction of species and staggering inequalities. While not being the silver bullet, innovation plays an important role in enabling such rapid societal and economic changes, and for firms and regions to adapt to these changes. Innovation, however, comes in different forms. Incremental innovations help sustain already existing economic activities or diversify from existing to related economic activities. In contrast, radical innovations based on novel combinations of knowledge and factors of production, which we in this paper coin combinatorial innovations, imply a higher degree of novelty and more disruptive change from existing economic activities, which tackling grand societal challenges requires.

This backdrop leads to the key question addressed in this paper: What empowers actors (entrepreneurs and intrapreneurs) to pursue combinatorial innovation processes? We ask this question from a people’s perspective because it needs people who perceive new opportunities and push for realising them in processes of Schumpeterian innovative entrepreneurship (Schumpeter 1934; Weik 2011). Schumpeter developed the theory that economic change results in essence from novel combinations of production factors, which in the contemporary debate relates to knowledge and resources. To be sure, institutional change is often a necessity to ensure the diffusion of such novel combinations (Granovetter 2005). One key insight in Schumpeter’s work is the distinction in ordinary economic activities where expectations rest on experience (e.g. the baker knows from experience how many rolls to bake), and innovative entrepreneurship where expectations rest on a belief in not yet realised opportunities. Experience – because the combinations are novel – cannot inform the latter, and, therefore, market and technological uncertainty are high (cf. Fleming 2001). Combinatorial innovations are the key driver of change in the economy in Schumpeterian thought (Shane and Venkataraman 2000). This paper follows the Schumpeterian thought and investigates conceptually what enables actors to realise such novel and more complex innovations.

This relates to an important debate on how economies change. Recent evolutionary thinking suggests that Schumpeter might not be right after all because related diversification based on less novel and complex innovations is the most common form of change in the economy (Hidalgo et al. 2018) and over a longer period, many incremental steps may result in a big change (cf. the concept of creeping change in the institutional literature (Streeck and Thelen 2005)). Yet, evolutionary scholars have noted that although less common, combinatorial innovation leading to unrelated diversification and/or new path creation is an outcome that cannot be neglected and is a key mechanism for more disruptive change (Boschma et al. 2017), which may lead to new types of economic activity (Janssen and Frenken 2019). Recent literature has also investigated which macro – and meso-level structures tend to correlate with unrelated diversification as a result of combinatorial innovations. For instance, unrelated diversification is more common in countries that have institutions associated with liberal market economies such as the US and UK while related diversification is more common in countries with institutions typically associated with coordinated market economies (Boschma and Capone 2015). However, as we will discuss later this is dependent on knowledge bases and type of economic activity. Unrelated diversification also tends to be more likely in regions with high innovation capacities (Xiao, Boschma, and Andersson 2018) and high economic complexity (Pinheiro et al. 2022). Yet, also small regions with few related diversification options exhibit unrelated diversification and/or new path creation relatively often (Kuusk 2021).

While previous studies thus provide first insights about the macro– and meso-level structural characteristics important for unrelated diversification, the micro-foundations for combinatorial innovations that may result in unrelated diversification are still largely unexplored notwithstanding some relevant studies. For instance, unrelated diversification is more often associated with new establishments (than with incumbent firms switching an industry) and with the establishment of new subsidiaries of non-local firms, the latter having a higher rate of survival than the former (Neffke et al. 2018). Also, unrelated diversification often requires appropriating external knowledge, access to external markets and financial investments, and building legitimacy through external references (Binz and Anadon 2018). In a recent paper, Altintas, Ambrosini, and Gudergan (2022) argue that unrelated diversification requires different dynamic capabilities than related diversification, the former focussing more on explorative learning and renewal, and requiring a higher-level management involvement. However, we argue that a comprehensive theoretical and conceptual understanding of what empowers actors to pursue combinatorial innovation activities that may lead to unrelated diversification is still lacking. In the current paper, we hope to contribute to such an understanding, by investigating which organisational, network and institutional configurations enable or hinder the development of combinatorial innovations. The current paper theorises about and conceptualises actors in relation to organisational, network and institutional configurations; and how such an embedding affects actors’ possibility to pursue combinatorial innovation processes. Hence, we establish the micro-foundations for combinatorial innovation in relation to the meso-level (e.g. networks and industrial knowledge bases) and macro-level (e.g. institutional conditions). As we will show, such embedding has a clear geographic dimension implying that actors will find varying preconditions across space to push combinatorial innovations.

2. Micro-foundations for combinatorial innovations: a relational view on individual agents’ possibilities

In this paper, we do not view micro-foundations as static properties of individual agents but look out for the necessary relations of individual agents to structural properties that empower them to engage in combinatorial innovation processes (for necessary relations, see Sayer 1992). The individual agents we are interested in are Schumpeterian innovative entrepreneurs (or intrapreneurs) who combine knowledge and resources in new ways to introduce novelty to markets or society under the condition of uncertainty (Schumpeter 1934; Weik 2011). We refer to micro-foundations because we appreciate the role of Schumpeterian innovative entrepreneurs as change agents (Feldman 2001; Shane and Venkataraman 2000). However, our human agency lens also appreciates that the power for exercising innovative entrepreneurship is rooted in knowledge and institutional structures the agent is embedded in. While particular agents come and go, the relative durable nature of relations between actors and institutional and knowledge structures explain what makes it possible – enables and hinders – the engagement of actors in combinatorial innovative processes that may lead to unrelated diversification (cf. Bhaskar 1998).

The Schumpeterian view of innovations as novel combinations of knowledge is reflected in the definition of combinatorial innovations suggested by Grillitsch, Asheim, and Nielsen (2022). Accordingly, combinatorial innovations are viewed as combinations of different types of knowledge bases and sectors, resulting either in unrelated knowledge applied in a related sector (e.g. analytical knowledge (biotech) combined with synthetic knowledge in food production creating new to the market products (functional food)) or related knowledge applied in an unrelated sector (e.g. using knowledge of composite technology in ski production to make aircraft parts as the Austrian firm, Fischer, has done) as shown in Table 1. By combining different knowledge bases the knowledge complexity increases (cf. Balland et al. 2019), as also may be the case when a firm diversifies into new sectors due to additional quality requirements of the products.

Table 1. Combinatorial innovations by knowledge and sector.

	Related knowledge	Unrelated knowledge
Related sector	Related diversification e.g. car – > trucks	Unrelated diversification e.g. food – > functional food based on biotechnology
Unrelated sector	Unrelated diversification e.g. ski – > aviation based on composite material	New path creation e.g. by attracting and anchoring actors/knowledge from outside the region

Source: Modified from Grillitsch, Asheim, and Nielsen (2022).

In Table 1 we use the distinction between the different knowledge bases (analytical, synthetic and symbolic) to determine if the knowledge combinations are related or unrelated. If knowledge combinations comprise more than one knowledge base, the combinations are unrelated and can produce combinatorial innovations. An analytical knowledge base refers to economic activities where scientific knowledge based on formal models and codification is highly important. Examples are biotechnology and nanotechnology. A synthetic knowledge base refers to economic activities, where innovation takes place mainly through the application or novel combinations of existing knowledge. Often this occurs in response to the need to solve specific problems coming up in the interaction with customers and suppliers, and, thus, innovations are user, market and demand-driven. Industry examples are engineering industries such as plant engineering, specialised advanced industrial machinery and shipbuilding. Symbolic knowledge is related to the creation of meaning and desire as well as aesthetic attributes of products, such as designs, images and symbols, and to its economic use. Industry examples include media (filmmaking, publishing and music), advertising, design, brands and fashion (Asheim 2007).

Combinatorial innovation, combining unrelated knowledge and/or unrelated sectors, is thus a necessary mechanism for unrelated diversification or new path creation. Grillitsch, Asheim, and Nielsen (2022), building on March (1991), argue that by integrating unrelated elements, diversification becomes more novel, uncertain and with benefits accruing more distant in time, and that therefore the time perspective of actions play a key role (Fleming 2001; Martin and Sunley 2022).

We use a systemic perspective discussing the causal powers constituted in institutional incentive structures, knowledge bases, network structures and firm-level routines, as illustrated in Figure 1 to theoretically explore the relations that shape the possibilities of individual agents to develop combinatorial innovations. As discussed in sections 3 and 4, institutional incentive structures and knowledge bases affect the time perspective of individual agents, and as combinatorial innovations need a longer time perspective of actions, this constitutes an important micro-foundation for combinatorial innovations. We will further elaborate why and how the network structures (section 5) and firm-level routines (section 6) affect the capability of the innovative entrepreneur to combine knowledge in novel ways, which constitutes another micro-foundation for the possibility of combinatorial innovations. This is summarised in Figure 1. As we will argue in the following sections, the depicted relationships represent causal claims but these causal claims need to be interpreted as tendencies, meaning that a longer-time horizon and enhanced combinatorial capabilities will make it possible for individual agents to engage in combinatorial innovation processes. However, whether individual agents decide to draw on these powers and start engaging in processes of combinatorial innovation, how they decide to do it, and to what intended and unintended outcomes (e.g. whether it leads to unrelated diversification) is contingent, i.e. dependent, on the interplay with other conditions in an open system (for tendencies and causal explanation, see Bhaskar 1997; Jessop 2005).

Figure 1. Theoretical illustration of relations shaping agents’ possibility for combinatorial innovation.

An illustration that links incentive structures and knowledge bases to the time-perspective of action, and network structures and firm routines to combinatorial capabilities. Then it links the time-perspective of action and combinatorial capabilities to the possibility for combinatorial innovations.

3. Institutional incentive structures

The way we are thinking of institutions in this paper is related to the comparative institutionalist approach to varieties of capitalism (Hall and Soskice 2001), which highlights the following five dimensions:

• Financial regulation: Long-term patient capital and debt financing vs. short-term financial markets and equity financing

• Corporate governance: Stakeholder value vs. shareholder value

• Innovation outcomes: Variations in the level and type of innovation across sectors and industries

• Capital-labour relations: Coordinated bargaining, strong trade unions and statutory worker representations vs. decentralised bargaining, weaker trade unions and contentious workplace relations

• Training and employment: Vocational training, apprenticeship, long tenure, low turnover of jobs and low interfirm mobility vs. basic education and firm-specific training, short tenure, high turnover of jobs and high interfirm labour mobility

These five dimensions are highly interrelated, and one important commonality among them is the time dimension, i.e. the distinction between short-term and long-term perspectives. This distinction has a strong bearing on the incentive structures we refer to in this paper. Factors such as corporate governance (stakeholder vs. shareholder interests), ownership (family firms, listed or unlisted firms on stock markets), firms’ place of origin (local, national, or foreign (MNE)) and the type of capitalism in these places have implications for the time requirements of innovation processes in different industries with different knowledge bases. In all these incentive structures the time perspective is an important dimension, i.e. if firms have a long-term planning horizon and have access to patient and risk capital, or if they have a short-term perspective with only access to capital from financial markets, which today are very short term. One could then hypothesise that firms with a long-term planning horizon would be more open to exploiting combinatorial innovations than firms with a short-term horizon, as combinatorial innovations normally will take longer time to accomplish. Firms with a long-term planning horizon would then tend to be either family firms, firms owned by holdings (such as the global healthcare company, Novo Nordic) or unlisted firms, local and national firms from coordinated market economies, and international firms (MNE) also originating from coordinated market economies, all of which in one way or another have a greater degree of leverage to prioritise decisions that require a long-term perspective to be realised and, thus, tend to avoid short-term thinking determined by actors in the stock market, or demanded by a shareholder value focused board (and CEO). This suggests that incentive structures are rooted in formal and informal institutions at the societal (socio-political) level, which influence the behaviour of actors at the micro level, as well as at the firm level.

The increased dominance of short-term thinking, especially in the US and UK, but which also to a lesser extent has influenced coordinated market economies is clearly a result of the revival of the neo-liberalist ideology in the 1980s, pioneered by Reagan and Thatcher, which made it easier to accommodate Milton Friedman’s ideology of maximising shareholders value as the basic governance principle of companies at the expense of broader stakeholder interests, which have resulted in increasing social and regional inequalities nationally and internationally. Corporate governance based on maximising shareholder value is more dominant in the Anglo-American world (liberal market economies) than in continental and Northern Europe (coordinated market economies), but even in the latter group of countries the shareholder value perspective has attained larger influence. Through their prominence, moreover, US transnational companies (TNCs) have also contributed to creating a global economic playing field that TNCs from other countries have had to adapt to be globally competitive (Chang and Andreoni 2020; Wade 2018).

Another outcome of Reagan’s policies in the 1980s, pointed out by Lazonick (2016), was to allow CEOs to be partly paid in share options, which collapsed the division of labour between the CEOs as value creators and the Board as value extractor. This significantly increased the short-termism of the system and is the main reason behind the skyrocketing CEO salaries. Today this has resulted in share buybacks to boost share values in a short-term perspective to satisfy shareholders instead of investing profit in R&D and innovation to secure future growth and job creation. A striking illustration of this is that most of the enormous tax cuts that American companies received from the Trump tax reductions in 2017 were used for share buybacks, and nearly nothing was invested in securing future growth and job creation of the companies, as was presented as the rational justification of the policy (Krugman 2020). Characteristically, the investments in R&D as a percentage of revenue has declined since the 1980s, in part because the share price usually suffers when companies announce this kind of investments, and ‘the bonus culture motivates management to use corporate profits to raise share prices, rather than invest’ (Wolf 2021).

Thus, the liberal market economies have more short-term orientation in financial and employment relations but are, according to the varieties of capitalism dichotomy (Hall and Soskice 2001), associated with more technological breakthroughs. The latter have become less clear cut after 2000, due to the reduced, both private and public, levels of investments in R&D and innovation. A shift has started to occur, according to Soskice (2020, 4), noting that ‘Even if the US held the technological lead, a noted slowdown in US innovation was already developing through the 2000s’, and further

this process which it attributes in part to the progressive retreat of the Federal government from the innovation process, and notably from basic science. This was in part made up by the FAANG [Facebook, Amazon, Apple, Netflix, Google] giants but only in a limited way as far as basic science was concerned

However, this has still been sufficient for the US to be the leading country in the IT/software-based third industrial revolution and provide a hint that knowledge bases play a role for the outcomes as discussed in the next section.

The varieties of capitalism in different countries also influence the behaviour of TNCs subsidiaries in host countries, as they normally bring with them the corporate governance structure and strategy from the location of the headquarter in their respective home countries. This means that a green field investment or an acquisition of a company by a TNC from a liberal market economy will apply the governance system of maximising shareholders value, given no strict regulations in the host country blocking this. The latter is normally not the case. Contrary, a TNC from a coordinated market economy, e.g. a Nordic or continental European TNC, will bring along the corporate governance of their respective home country. This difference has been observed in several empirical studies and will often have a significant impact on the behaviour of the subsidiary with respect to investments in R&D and innovation (Herstad 2005).

Thus, in the coordinated market economies such as the Nordic countries, even if the influence of neo-liberal ideas cannot be ignored, we would still argue that certain firm characteristics such as family firms and stakeholder interests are associated with more long-term interests, which could be more favourable for combinatorial innovations. Do we here see a relationship between the firm and its context? Maybe a family firm in the Nordic context is something else than a family firm in the US context; and a listed firm in the US is something different than a listed firm in the Nordic context? However, even within the Nordic context, we think that listed firms are more exposed to short-term thinking, of course depending on who the shareholders are (their geographical origin), and whether the firm is listed on foreign stock exchanges in liberal market economies. In general, also in a Nordic context, the idea of favouring shareholders value is not absent among investors and businesspeople.

4. Knowledge bases and industries

The effects on innovation outcomes of having a long-term vs. a short-term perspective must, thus, as we already have indicated, be seen in relation to the knowledge base(s) and technology that respective industries are based on, and the products manufactured. In Table 2 we make a distinction between the different knowledge bases (analytical, synthetic and symbolic) and between tangible and intangible types of products. Only the analytical (scientific) knowledge base contains both these two types of products, while synthetic only produces tangible products (engineering) and symbolic only intangible (creative assets such as design, trademarks and copyrights). Along the long-term vs. short-term dimension tangible products tend to be long-term while intangible is short-term. This implies that industries that are based on analytical and synthetic knowledge that produce tangible products need a more long-term perspective for doing R&D and innovation and, thus, need an institutional set-up that can deliver the incentive structures that on the micro level underpin actions leading to combinatorial innovations, while industries based on analytical and symbolic knowledge producing intangible products and services tend to have a more short-term perspective when carrying out R&D and innovation projects.

Table 2. Overview of the different knowledge bases/type of products.

	Analytical knowledge base	Synthetic knowledge base	Symbolic knowledge base
Tangible products	Biotech	Deep tech/hardware engineering industries	–
Intangible products	IT/Software	–	Cultural and creative sectors

Source: Based on Asheim (2007).

Table 2 contributes to a more nuanced understanding of the innovation outcomes in the dichotomised variety of capitalism model. The view that there is more radical innovation in liberal market economies, and specifically in the US, fits very well with software-based industries making intangible output, where the typical institutional set-up in the US of a highly flexible and decentralised system of research and higher education; a flexible and decentralised system of finance; companies capable of scalability (e.g. the FAANGs); and a high-level labour market support radical innovations in a short-term perspective as found in software-based industries (Soskice 2020).

In deep tech/hardware sectors using basically a synthetic knowledge base, where we find engineering industries such as advanced production (e.g. robotics) and energy sectors (i.e. hydrogen), the picture is different. These are sectors that need to have a long-term planning horizon in their R&D and innovation projects, as the innovation process includes engineering, developing of a prototype, as well as manufacturing, producing the prototype to see if it actually works and are reliable before the product can be commercialised in the market. This also spills over when looking at start-ups and scale-up firms in these sectors, which need more long-term patient and risk willing capital than is normally provided by venture capital financing start-ups in software sectors, which represents the majority of funding for start-ups in most countries. With cultural and creative sectors, we refer to activities which centre around cultural values and creative individual or collective expression. A typical example is the fashion industry where the value is less in the tangible material (the fabric) but rather in the cultural value and creative expression based on the brand and design. Hence, innovation in the cultural and creative industries should tend to require a shorter time-perspective than deep-tech/hardware sectors as they are less knowledge, R&D and capital intensive.

Thus, the reduction in both private and especially public funding of R&D and innovation in the last 20 years in the US, explains to a large degree why US has lost its innovation and manufacturing competence and capacity in deep tech/hardware sectors such as e.g. advanced production (manufacturing of robots and other industry 4.0 production technology) and semiconductors. The exceptions are biotech due to strong public funding from the National Institutes of Health (NIH) and military-related industry, where security aspects dominate, and where public agencies such as Defense Advanced Research Projects Agency (DARPA), which managed to ‘sail under the radar’ of the neo-liberal ideology by not indicating that it in reality engaged in industrial policy, has played a similar strategic role as NIH (Wade 2018). This short-termism also caused an almost total stop in the training of technicians (which is typically taken care of by the apprenticeship system in Germany and Austria), and which is a necessary competence to run manufacturing industries in hardware/deeptech sectors. The development of software is on the other hand a rather short-term activity and can, thus, be successful even within the shareholder value paradigm, not the least because it rewards the shareholders very generously because of the network effects in software sectors, where the winners take all.

It is also a question of how important and/or representative breakthrough innovations as is found in biotech are in engineering-based deep tech/hardware sectors, where strategic (combinatorial) innovations are more characterised by the Schumpeterian view of new combinations of existing knowledge. This view of strategic, combinatorial innovations will probably increase in importance in a potential 4th industrial revolution around green technologies (e.g. renewable, and sustainable energy technologies), where an important aspect will be the integration of IT/digital technologies and mechanical and electrical engineering, but taking place in an engineering, manufacturing industry. The breakthrough innovations are typically found in the individual sciences that is integrated in engineering-based industries, but it is the integration of these different knowledge bases, and the subsequent engineering and manufacturing that is necessary to launch a successful new deep tech innovation. And for doing this, the institutional set-up of coordinated market economies is more supportive than the ones found in liberal market economies. Thus, it is not surprising that Germany has been the first driver in industry 4.0 production technology, often pioneered by the so-called hidden champions (Bessant 2019), which are small and medium-sized, family-owned firms located all over Germany, often in small towns in semi-peripheral areas, and not in large metropolitan regions as is the case with the leading US software firms of the third industrial revolution. This also implies that coordinated market economies in the Nordic countries and continental Europe (Germany, Austria, Switzerland), together with Asian countries (China, Japan and South-Korea), which also have a large engineering/manufacturing-based industry sector, might be the first movers of the fourth industrial revolution of green technology replacing the US as the leading country.

5. Network structure

Strambach and Klement (2012) point out that combinatorial innovations typically require accessing and appropriating knowledge residing in different industries and sectors. According to the authors, knowledge does develop in a cumulative manner within industries and sectors. It has also been argued that knowledge becomes increasingly homogeneous between the actors of one industry (Ter Wal and Boschma 2011). Due to the increasing division of labour and specialisation in the modern economy, there will be more possibilities for novel re-combinations but the cognitive and institutional distance between the different fields of knowledge may have become larger, hence, making re-combinations more difficult (Grillitsch 2018).

Social network theory postulates that networks are by definition more dense and homogenous within social groups than between them, and the term ‘structural holes’ identifies the lack of connections between actors belonging to different groups (Burt 1992). For instance, within an industry, suppliers and clients interact frequently within supply chains erected at different geographical scales. These interactions over time lead to a shared knowledge base and institutions made to facilitate interactions. Nelson (1994) referred to this process as co-evolution of industries, technologies and supportive institutions. However, between industries or between sectors (e.g. business, university, government, civil society) the networks are less dense, the knowledge more heterogeneous and the institutions less adapted. This then draws attention to actors who connect structural holes.

Burt (2004, 354) formulates this as follows:

People whose networks bridge the structural holes between groups have an advantage in detecting and developing rewarding opportunities. Information arbitrage is their advantage. They are able to see early, see more broadly, and translate information across groups. Like over-the-horizon radar in an airplane, or an MRI in a medical procedure, brokerage across the structural holes between groups provides a vision of options otherwise unseen.

Burt furthermore connects brokerage across structural holes to the generation of novel ideas: ‘People with connections across structural holes have early access to diverse, often contradictory, information and interpretations, which gives them a competitive advantage in seeing and developing good ideas’ (Burt 2004, 388).¹

As an important sidenote, Burt takes issue with the ‘heroic’ status innovative entrepreneurs have received in the early literature but even still today in public discourse, where it is mainly men receiving the public ‘hero’ status e.g. Bill Gates, Steve Jobs, Elon Musk. The social network theory suggests that a key explanatory factor for good ideas and acting upon them is the position in social networks. ‘People connected to groups beyond their own can expect to find themselves delivering valuable ideas, seeming to be gifted with creativity. This is not creativity born of genius; it is creativity as an import-export business’ (Burt 2004, 388). This is relevant in this paper because – in contrast to the genius of heroes – individual and policy actors can potentially affect network structures and facilitate the combination of knowledge across social groups.

At the micro-level, there is substantial evidence that teams and firms who bridge structural holes show enhanced performance (Zaheer and Bell 2005; Zaheer and Soda 2009). In economic geography and regional studies, a focus has been on the combination of regional and extra-regional knowledge with the result that combining knowledge from different geographical scales tends to be associated with more novel innovation (Fitjar and Rodríguez-Pose 2011; Tödtling and Grillitsch 2015). In terms of social network theory, sourcing knowledge from extra-regional sources provides more opportunities for brokering structural holes, even though this is mediated by regional characteristics. In small and specialised regions, the regional networks are typically dense with few structural holes, while metropolitan regions provide for more opportunities for brokering within the region (Trippl, Grillitsch, and Isaksen 2018). Combinatorial innovation thus also relates to the ability to tap new knowledge globally and anchor it locally, especially when knowledge bases are homogeneous where the innovative agent is located (Binz and Anadon 2018; Crevoisier and Jeannerat 2009; Klement and Strambach 2019).

Concerning bridging structural holes, Hervás-Oliver et al. (2018) finds that combinatorial innovations imply the introduction of new technology-distant knowledge. However, they also find that access to the networks of leading incumbents is a crucial social factor for anchoring such technology-distant knowledge in a local context and thereby promote unrelated diversification or new path creation through combinatorial innovations. The authors argue (Hervás-Oliver et al. 2018, 1395)

In fact, their socially based control of networks act as complementary assets, which regulate the type of knowledge fed into the system, due to the social trust and repetitive interactions with the local firms in their networks. […] The entrance of new firms and knowledge, therefore, requires cooperation between leading incumbents and newcomers. Otherwise, new technology could not penetrate those dense local social ties that prevent any major change.

This resonates with the findings of Leminen et al. (2016) who investigated the novelty of innovations in living lab networks. Accordingly, it mattered whether networks were centralised, distributed, or showed what they called a multiplex structure. Centralised networks are orchestrated and controlled by a single actor. Distributed networks do not have a clear structure nor are controlled by single actors. Multiplex networks are distributed but with some central actors coordinating and facilitating networking. The authors find that innovations with a higher degree of novelty tend to emerge from multiplex networks structure while centralised or distributed networks promoted incremental innovation.

Having established the importance of brokering knowledge between social groups, an important question is how actors build such capability. In this regard, Grillitsch (2018) discusses two key mechanisms, namely multiple positions and positional mobility. The former captures if individuals hold positions in different social groups, e.g. in industry, academy or government, or at the board of firms in different industries. The latter refers to the moves of individuals between positions over time, between e.g. firms, industries, or sectors. With each position (or move), individuals are exposed to new knowledge, routines, institutions and build social networks. For instance, Suvinen (2014) finds that university professors active in innovative ventures often hold other positions in e.g. firms or intermediary organisations simultaneously. Lawton Smith and Waters (2011) highlight the importance of mobility between positions across industries and between universities and industries to promote the circulation of knowledge. Agrawal, Cockburn, and McHale (2006) find that researchers who move to new locations maintain networks with the previous location and mobilise them in research collaborations. Appreciating the importance of positional mobility, Etzkowitz (2012, 768) argues that ‘the absence of a strategy of creating permeable boundaries among the institutional spheres can be a significant retarding factor in regional development’. Grillitsch (2018), however, cautions that because multiple positions and positional mobility are powerful forces for mobilising knowledge and resources across social structures, and even shaping institutional architectures, there is a potential dark side where influential actors may misuse this power. Hence, policies promoting multiple positions and positional mobility must be designed with care, considering potential unwanted and unintended consequences.

6. Firm routines

A point of departure capturing firms` ability for combinatorial innovations lies in the distinction between knowledge exploration and exploitation introduced by March (1991) and further developed in Levinthal and March (1993). ‘Exploitation’ is concerned with refining capabilities that are valued in firms` present markets, whilst the description of ‘exploration’ revolves around ‘the pursuit of knowledge of things that might come to be known’ (Levinthal and March 1993, 105).

Explorative efforts may provide the basis for combinatorial innovations to emerge, however, efforts need to be in place for balancing the explorative and exploitative capacities (captured through the work by O’Reilly and Tushman (2008) on organisational ambidexterity). In other words, firms’ routines and strategies for exploration and exploitation affect their abilities to continue business through known channels or exploit new opportunities.

In this vein, firm routines relevant for combinatorial innovations are best captured through the notion of dynamic capabilities. Dynamic capabilities refer to enterprises’ ‘ability to integrate, build, and reconfigure internal and external competences to address rapidly changing environments’ (Teece, Pisano, and Shuen 1997, 516) and have been highlighted as fostering competitive advantage and enhanced effectiveness (Zahra, Sapienza, and Davidsson 2006).

The dynamic capabilities approach is inclined towards ‘Evolution with design’ (Teece 2013) where processes are not only altered in-house, but also co-evolve with firms` environments. Hereunder, it acknowledges that outside sources are critical to the innovation process as innovation does not take place in isolation (Chesbrough 2003), but through mutual colouring of firms and its surrounding space (Herstad, Solheim, and Engen 2019). The dynamic capability approach highlights the interrelations between agency and structure where ‘today`s complex system of environmental, socio-political, and economic systems, however, is constantly being reconfigured by human behavior and is simultaneously constantly affecting that behavior’ (Hynes et al. 2020, 177). It is, in the literature, however, often underemphasised that dynamic capabilities comprise also alterations to firms` external environment (Schilke, Hu, and Helfat 2018). Dynamic capabilities, however, differ between related and unrelated diversification where the former focusses more on exploitative learning and capability improvement, and the latter more on explorative learning and capability renewal (Altintas, Ambrosini, and Gudergan 2022).

Dynamic capabilities are context-specific and embedded and must be built over time (Helfat and Peteraf 2009) and building these dynamic capabilities will often collide with short-term profitability and what might appear to be the more efficient solution (Hynes et al. 2020). This begs the question of time, and that the contextual surroundings as discussing dynamic capabilities in fast-paced environment, as Arend and Bromiley (2009) argue, should not be confused with repeated, frequent strategic reorientations, which ‘may be so disruptive that firms cannot function effectively. Major strategic reorientations, however, do not occur overnight. Often what looks like a large strategic shift ex post consists of a series of incremental and less disruptive changes’ (Helfat and Peteraf 2009, 95).

Eisenhardt and Martin (2000) uphold a difference for dynamic capabilities in moderately dynamic markets (in which dynamic capabilities share similarities with the more traditional view on routines, such as upgrading), and high-velocity markets (in which they argue processes are fragile, simple and highly experiential comprising unpredictable outcomes, fostering combinatorial innovations/new path creation). Firms` ability to change and adapt is increasingly important in high-velocity markets or lately defined as ‘VUCA’ (Volatile, Uncertain, Complex and Ambiguous) environments (Bennett and Lemoine 2014; Tulder, Jankowska, and Verbeke 2019). The VUCA world has much in common with ‘high-velocity’ markets as defined by Eisenhardt and Martin (2000), as changes are seen as nonlinear and unpredictable, with blurred market boundaries, and market structures being ambiguous and unstable (Li, Easterby-Smith, and Hong 2019).

Relevant herein is to develop an understanding of the microfoundations of dynamic capabilities. Teece (2018) argues that dynamic capabilities can be divided into ‘microfoundations’ and ‘higher-order capabilities’. The microfoundations comprise adjustments and recombinations of firms` existing resources as well as the creation of new ones. In line with this, Eisenhardt and Martin (2000) have identified a panoply of routines that they argue provide certain microfoundations for dynamic capabilities, such as new product development routines, quality control routines, cross functional R&D teams and performance measurement systems. Moreover, the dynamic capabilities to compete in VUCA business environments also rest on entrepreneurs’ ability to tap into external sources of knowledge (Ferraris, Santoro, and Dezi 2017). Combinatorial innovations often require appropriating external knowledge, access to external markets and financial investments, and building legitimacy through external references (Binz and Anadon 2018). Teece (2018, 41) argues that guiding these microfoundations, or what he refers to as ‘second-order dynamic capabilities’ are ‘high-order dynamic capabilities’

by which management, supported by organizational processes, senses likely avenues for the future, devises business models to seize new or changed opportunities, and determines the best configuration for the organization based on its existing form and the new plans for the future.

Teece (2007, 2020) put forward that dynamic capabilities primarily involve activities of sensing, seizing, and transforming. Sensing

include[s] environmental scanning … from internal and external sources … [to] identify new opportunities such as underserved markets or supplemental revenue sources. … [It] requires an internal knowledge network built on decentralized authority, a collaborative organizational culture, and the ability to extract meaning from heterogenous signals (Teece 2020, 11)

Seizing involves ‘the design or updating of business models for new products and services … [and] also encompasses allocating resources, including cash, to high-yield uses, or uses with the potential to become so’. Transformation, or reconfiguration, implies that the enterprise intermittently restructures ‘to maintain evolutionary fitness’ by developing new structures, business models, products or services while abandoning other activities (Teece 2020).

In line with this, Schilke, Hu, and Helfat (2018, 393) highlight that firm capabilities can be divided into two; first, ‘ordinary capabilities’ concerned with maintaining status quo, and second, dynamic capabilities which are targeted towards strategic change, and as such ‘can effect change in the firm`s existing resource base (and the associated support system such as the firm`s organizational and governance structure), its ecosystem and external environment, as well as its strategy’. This becomes increasingly important when it comes to adaptability and resilience, as dynamic capabilities are ‘purposeful’ in terms of intended directionality, and as to distinguish the path chosen form ‘pure luck’ (see Helfat and Peteraf (2009) introducing ‘purposeful’ to the concept of ‘dynamic capabilities’).

Underpinning the microfoundations for combinatorial innovations through dynamic capabilities refers moreover to routines and capabilities to integrate different views and resolve conflicts. Hereunder, lies the importance in ‘breaking down the thought worlds that arise because people do not only know different things, but know those things differently’ (Eisenhardt and Martin 2000, 1109).

7. Conclusions

With this paper, we propose a perspective on the micro-foundations of combinatorial innovations and ask what empowers actors (entrepreneurs and intrapreneurs) to pursue combinatorial innovation processes. This means that we do not see micro-foundations in some static properties of otherwise homogeneous agents (e.g. comparing firm-level R&D expenditures). Rather, we embrace the complex reality with heterogeneous agents who are actively involved in exploring, developing and exploiting opportunities, and thereby act as the agents who often maintain but sometimes transform markets and society through innovative action. Innovative action targeting combinatorial innovations may lead to unrelated diversification and the creation of new development paths, which can play an important role to realise the transformations needed to address the many societal challenges, most importantly climate change and inequalities (Donald and Gray 2019; Frenken 2017).

We find the relations empowering agents in firstly the institutional incentive structures and knowledge bases, which exhibit a tendency to affect the time-perspective of innovative entrepreneurs. The neoliberal project most prominent in the liberal market economies such as the US and UK strongly transformed the time-perspective of action towards short-term goals. This limits the engagement of innovative entrepreneurs in more transformative innovation activities, which are more uncertain and where benefits are more distant in time (Fleming 2001; Grillitsch, Asheim, and Nielsen 2022; March 1991). Also, knowledge bases shape the time-horizon of action where innovation processes in some industries need a longer-time horizon. We attribute this to the knowledge base combined with the degree to which products are tangible. Accordingly, deep-tech industries, which mainly use analytical and synthetic knowledge and have tangible products, require a longer time-horizon as compared to IT/software and cultural and creative industries, which mainly produce more intangible types of products and services. Noteworthy, deep-tech industries related to for instance energy, transport and food will play an essential role in addressing climate change, thus underlining the need for a more long-term perspective.

Secondly, we consider that the nature of network structures and firm-level routines have a tendency to affect the combinatorial capabilities of innovative actors. As regards network structures, a position of actors in structural holes between fields of knowledge in different industries, between sectors, or between academic fields will increase the ability to identify and combine knowledge in novel ways (Burt 1992; Putnam 1995). As regards firm-level routines, the notion of dynamic capabilities best captures firm-internal capabilities to reconfigure knowledge and resources to break with existing paths and respond to changes in quickly changing environments (Teece 2007; 2020). This involves activities of sensing opportunities for novel combinations of firm-internal and external knowledge, seizing these opportunities through developing the opportunities and creating new business models, and transforming the firm but also the environment the firm is operating in. Explorative learning, renewal of capabilities, and involvement if high-level management underpins dynamic capabilities for combinatorial innovations and unrelated diversification processes (Altintas, Ambrosini, and Gudergan 2022).

We argue that an understanding of the relations and thereby conditions that enable innovative entrepreneurs to engage in combinatorial innovation processes is of high academic and practical relevance. It would call for research that interrogates the combinations of conditions that make it possible for individual agents to develop combinatorial innovations. This call is in line with detailed tracing of behaviours, decisions and contexts of involved parties (see Martin and Sunley (2022)). For practitioners, such knowledge will have emancipatory value because these conditions (institutions, knowledge, networks, firm routines) are socially produced and can thus be changed. More specifically, it will create insights into how the short-termism of the neo-liberal project may be addressed and more long-term incentive structures established, and firms supported in taking positions in structural holes, and in building capabilities for sensing, seizing and transforming opportunities. Even though not being a silver bullet, knowledge about the conditions enabling combinatorial innovation is one important mechanism for bouncing forward towards a greener, and more inclusive world.

Acknowledgements

Financial support for this study was provided by the Research Council of Norway, through the project ‘RegReSir – Regional Resilience and Sustainable Industrial Restructuring’ (project number 316539).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Norges Forskningsråd [grant number 316539].

Notes on contributors

Markus Grillitsch

Markus Grillitsch is associate professor of economic geography at the Department of Human Geography and director of CIRCLE - the Centre for Innovation Research at Lund University.

Björn T. Asheim

Bjørn Asheim is professor of economic geography and innovation theory at University of Stavanger Business School. He was co-founder, deputy director and director of CIRCLE at Lund University, 2004-2012. Asheim has previously been professor of economic geography at the University of Oslo and Lund University. He has been editor of Economic Geography and Regional Studies. In 2011 professor Asheim became an Academician at the British Academy of Social Sciences. He is internationally well-known within economic geography and regional innovation studies.

Marte C.W. Solheim

Marte C.W. Solheim is the Pro-Rector for Innovation and Society at the University of Stavanger. She is a Professor in Innovation Studies at the same University and a former head of its centre for Innovation Research.

Notes

1 In a similar vein, Putnam (1995) coined the term ‘bridging social capital’ for networks across social groups, as opposed to ‘bonding social capital’ capturing networks within social groups.