1 Introduction
It is commonly accepted that new knowledge is a primary source of economic growth, as suggested in the endogenous growth literature (Lucas
1988; Romer
1986,
1990). However, endogenous growth theory does not specify the mechanisms through which new knowledge is converted into economic activity and growth. Braunerhjelm et al. (
2010) have suggested that diffusion of new knowledge via entrepreneurship provides the missing link. Acs et al. (
2009) propose that spillovers from R&D activity in universities and incumbent firms are a key mechanism; R&D activity endogenously creates entrepreneurial opportunities that are materialized by new firms “outside” universities and incumbents.
Although the creation, diffusion, and utilization of new knowledge is a key focus in innovation systems (Carlsson and Stankiewicz
1991; Cooke
2001; Freeman
1987; Lundvall
1992; Nelson
1993), there is a dearth of research on the links between innovation systems and economic growth. This is largely due to the fact that most of the literature on innovation systems is static and descriptive rather than dynamic and analytical, focusing on the supply side (the creation of technology) rather than on how innovations are converted into economic activity and growth via the market (the demand side, and the interface between supply and demand, see Carlsson
2016) and on how innovation systems evolve over time.
The purpose of this paper is to address this gap in the literature. Entrepreneurial activity is inherently experimental in nature (Kerr et al.
2014). We articulate the function of entrepreneurial experimentation as an essential mechanism for translating new knowledge into economic activity and growth created in innovation systems. To this end, we argue that spinoffs and acquisitions—in addition to de novo firms and intrapreneurship—constitute micro-level mechanisms and processes of industrial dynamics that give rise to system-wide entrepreneurial experimentation that creates, selects, and scales up new technology and innovations. This means that entrepreneurial experimentation is central in driving both the supply- and the demand-side dynamics of innovation systems and is thereby crucial in understanding the link between innovation systems and economic growth. Entrepreneurial activity is a key function in technological innovation systems as originally formulated by Carlsson and Stankiewicz (
1991), but its role has been largely ignored in the literature.
The paper is organized as follows. In the next section, we discuss the links between knowledge creation, entrepreneurship, and economic growth. We then review extant perspectives on Systems of Innovation and identify important implications of the differences between them. In the fourth section, we describe the role of entrepreneurial experimentation in systems of innovation. This includes proposing two key mechanisms for entrepreneurial experimentation (in addition to de novo startups).
1 We propose that spinoffs (both corporate and academic) and acquisitions are key for a successful entrepreneurial experimentation and the dynamics of innovation systems. In the final section, we discuss the implications of these suggestions and propositions for innovation and entrepreneurship research and practice.
2 Knowledge creation, entrepreneurship, and economic growth
Most new knowledge that generates economic growth is created in universities and existing (often large) firms. According to Carlsson et al. (
2009), 60% of R&D in the USA is carried out by businesses to develop better production processes and products (the development or D part of R&D). Twenty percent of U.S. R&D is applied research, mostly in business firms, while 20% is basic research (of which 60% is done in universities). A similar pattern can be found in Sweden where more than 2/3 of R&D is carried out in businesses (mainly a few large firms), and around 27% in universities (SCB
2014).
Most new knowledge generated in universities benefits the economy via a better educated labor force, but some new knowledge (probably less than 10%) is commercialized via licensing to existing firms (the main avenue) or via new start-ups (academic spinoffs). This creates new business opportunities and greater variety. Most of the new knowledge created in large firms results in expansion of existing lines of business (via cost reduction and product improvement), but some of it results in new business lines in existing firms or in the formation of new businesses via corporate spinoffs or via acquisitions or corporate reorganization/recombination. Acquisitions represent a major avenue for diversification and for scaling up newly started businesses. Some new firms (de novo firms) are based on previously existing or traditional (exogenous) knowledge (de novo startups represent a much larger percentage of the number of startups than of the contribution to economic growth).
Thus, most R&D is carried out in existing business firms and results in incremental innovation (improved products and processes) that expands existing businesses via investment and higher productivity. This is the primary source of economic growth. Our concern in this paper is with new knowledge and technology—radical innovation—that is commercialized through experimentation rather than through expansion of existing lines of business. Some of this new knowledge is created in incumbent firms and some in academic institutions. In each time period, this is only a small portion of the knowledge created, but over time, it represents a major source of economic growth. Radical innovation gives rise to new products and eventually new industries that compete with and sometimes replace old ones (creative destruction). Radical innovation pushes out the knowledge frontier and creates new opportunities that transform the economy over time. As innovation systems evolve, incremental innovation in incumbent firms tends to become dominant.
This process works out differently in different domains. To illustrate, it is useful to distinguish between two types of technological regimes (Carlsson
2013). Design-driven regimes are typical in well-developed engineering fields. Mechanical engineering, electrical engineering, design and fabrication of semiconductor devices, and software are examples of technology areas operating predominantly under the design regime. In these industries, the innovation system is mature, and innovation is typically of the incremental type. Technical problems are usually attacked through “analytical design”—presupposing a well-articulated design space (set of relevant technologies). The search processes taking place in that space are sequential and iterative rather than parallel. The relatively high efficiency of the development processes reflects the fact that the design space utilized is strongly bounded and the performance requirements well defined and easy to operationalize. Design-oriented innovation processes are demand rather than opportunity-driven. Incumbent firms are the primary drivers. Commercialization takes place when incumbent firms expand existing lines of business and diversify by creating new business lines (intrapreneurship or corporate venturing), when employees leave their employer (corporate spinoff) to found a new technology-based firm based on knowledge and ideas acquired in the existing firm, or when new businesses are formed through re-combination or acquisition of existing businesses.
In contrast, discovery-driven regimes are characteristic of fields with poorly articulated or structured design spaces and loosely defined innovation systems. The limited extent to which functions are clearly identified and mapped on the known structures and processes means that the solutions to problems have to be discovered rather than designed (Stankiewicz
2002). Typical for discovery regimes is that innovation is driven by opportunity rather than demand. Technological advances, particularly radical ones, tend to be triggered by serendipitous discoveries, often through application of existing technologies to new areas. Product performance requirements are hard to fully specify and operationalize early in the process. There is often strong dependence on some form of field trials (experimentation). Examples are biotechnology and applications of semiconductor devices. In the life science industry, innovation is increasingly carried out in new entities, dedicated biotechnology firms, for the purpose of commercializing intellectual property (IP) developed in universities. Once commercialization is attained, such spinoffs are typically acquired by existing pharmaceutical firms for scaling up (production, marketing, and distribution). Commercialization of academic research takes place via start-ups (academic spinoffs) or licensing to existing firms. New (exogenous) ideas may also emerge outside the knowledge creation system and result in the creation of de novo startups. Once new knowledge results in a new business or line of business, selection occurs at many levels in the market, by consumers, venture capitalists and so on. Thus, entrepreneurial experimentation is a vehicle for both variety creation and selection.
The development of driverless cars is an example of radical innovation. New technologies are being developed outside the automobile industry (Tesla) or are introduced into the industry via partnerships between automobile manufacturers and IT firms. Volkswagen’s partnership with Nvidia and Aurora (a new Silicon Valley startup), Volvo’s partnership with Uber, and Fiat Chrysler’s partnership with BMW and Intel are examples here. It is also interesting that Google has set up a new entity, Waymo (a new company within the Alphabet group), to partner with existing car makers (Mercer
2018).
It is also noteworthy that the scaling up of automobile production in Detroit (itself the result of a combination of technologies in previously existing mechanical industries such as the manufacture of carriages and wagons, bicycles, and engines; the initial innovations were new combinations through the assembly of standard parts) took place via the Ford Motor Company, essentially a spinoff from Oldsmobile. Ford imported the idea of the moving assembly line from the meat processing industry (Klepper
2004; Carlsson
2013). After this initial radical innovation, innovation in the automobile industry has been much more incremental.
Our analysis focuses on entrepreneurship as a key process in which new technologies and new knowledge are converted into innovations that drive growth. The starting point is Carlsson and Eliasson’s (
2003) theory of the experimentally organized economy (EOE), which outlines a framework for micro-based endogenous growth (see also Eliasson
1991). A key idea in this framework is that economic growth is a result of experimentation that results in new technologies, followed by selection in dynamic markets and hierarchies, and of the capacity of the economic system to capture winning businesses and innovations, while letting go of losing ones. It thus has a two-pronged focus: (i) creation of variety of new technologies and ideas and (ii) selection and retention of “winning” innovations in the form of commercialization of new technologies. The selection process may be thought of as a kind of sorting or filtering process, in which viable and innovative high-impact businesses and innovations are selected by market forces—for example via acquisition of startups by incumbent firms—and scaled up. Based on the EOE framework, we argue that the central function of entrepreneurial experimentation in innovation systems involves creation as well as selection and scaling-up of innovations. Entrepreneurial experimentation relates both to the “supply-side,” in terms of the system’s capacity to develop variety of new technologies and business ideas that become subject to selection, as well as to the “demand-side,” in terms of effective selection and scaling up of innovations and businesses on the market.
2
The need for technical experimentation on the supply-side is required because of genuine Knightian uncertainties (Knight
1921) regarding which technologies may be useful and feasible.
3 A large variety of projects raises the odds of developing and selecting “good” technologies, i.e., it is not known a priori which technologies will turn out to be important innovations (commercialized technologies). Likewise, experimentation is crucial in the selection stage. For a given new technology, there are often no established business models or markets, no well-defined areas of implementation, and there is uncertainty about synergies with existing technologies and products (Kemp et al.
1998). No single Schumpeterian entrepreneur (new or established) knows beforehand “what works;” nor can knowledge of business models, market niches, and areas of technology implementation easily be deduced from some set of first principles. Entrepreneurial experimentation is crucial in the innovation process itself (“technical experimentation”) as well as in translating innovations into economic activity.
We hold that entrepreneurship at the systems level is fundamentally about experimentation (Kerr et al.
2014; Klepper
2015), and that entrepreneurial experimentation therefore is a key element of the domains of innovation systems and entrepreneurship research. Dynamic innovation systems, which feed innovation and growth, must promote or at least accommodate entrepreneurial experimentation.
Entrepreneurial experimentation may be conceived of as part of the absorptive capacity within the system, consisting of receptivity to new technologies and ideas, as well as the ability to act and experiment on them. Similar to Shane’s (
2000) argument about the discovery of entrepreneurial opportunities, it is possible that limited technological change generates large economic output because entrepreneurs experiment with different ways to exploit new technology. Conversely, significant technological change might generate limited economic output because of lack of entrepreneurial experimentation. Not only the discovery of opportunity, but also the decision to exploit opportunity, is crucial for entrepreneurial experimentation to take place (Schumpeter
1934). Policy-wise, this perspective naturally entails a focus on institutions and incentive structures that stimulate actors (firms, individuals, organizations) to undertake entrepreneurial experimentation (Carlsson and Eliasson
2003).
This line of argument implies that there is a need to explicate the systems features that lead to and stimulate entrepreneurial experimentation. Contributions by Freytag and Thurik (
2007) and Fritsch and Wyrwich (
2014) note the relative stability of differences in entrepreneurial activity across countries and regions and suggest that non-economic factors such as entrepreneurial culture and attitudes are important determinants. Marx et al. (
2009) focus on enforceability of non-compete clauses as an institutional feature that influences labor mobility and hence the diffusion of innovation. Yet, a common critique of the traditional innovation systems literature is that it stresses feedback loops and interdependencies, but is rather vague on where those effects come from, how they are materialized, as well as how they are linked to behaviors and incentives of actors in the system (Braunerhjelm and Henrekson
2016).
Given our definition of the function of entrepreneurial experimentation, our task is then to articulate and specify actors and respective actions involved in the three central processes, i.e., creation, selection, and scaling-up of innovations, within a given institutional setting. We also need to specify at the micro level—in terms of actors and actions—how the three processes link up to and feed each other in interdependent ways; it is such interaction that motivates the systems perspectives.
To this end, we put entrepreneurship and experimentation at center stage and develop our framework by stressing the interaction and symbiosis between new technology-based firms and established businesses and universities.
4 We identify two examples of critical mechanisms—spinoff and acquisition—that drive entrepreneurial experimentation in the system. Both are examples of mechanisms that are relevant in high-technology and knowledge-intensive contexts (see, e.g., Norbäck and Persson
2014; Andersson and Xiao
2016; Gans and Stern
2003). They also bear directly on a systems’ capacity to (i) create new technology and products (creation), (ii) experiment with regard to their applicability in various market and business contexts (selection), and (iii) scale up the commercial potential by embedding new technology and innovations into global sales networks and value chains (scaling-up).
While the role of spinoffs has been acknowledged in the literature on the evolution of industries and industry dynamics (see, e.g., Klepper
2001,
2002), acquisitions are often neglected in discussions of entrepreneurial activity. Established firms purposefully select pertinent acquisition targets. When targets are technology-based new firms, the incentive for acquisition for the acquiring firm is often to embed the target’s technologies, products, or services in existing systems and sales networks, or to strengthen the acquirer’s technological or knowledge assets. From the point of view of the founders of new firms, they and their organizations typically lack the human capital and financial resources needed to scale up their innovations and to fully exploit their commercial potential. Such resources therefore have to be acquired from outside, for example by being acquired by an established business. This implies that many entrepreneurs have an incentive to be acquired. For example, Baumol (
2002) argues that the different roles of new and established firms may be described as a “David-Goliath symbiosis.” The life science industry is a perfect example. New and small firms are more likely to develop radical innovations and technologies but lack resources to scale up. Established firms with global sales networks and resources, such as multinational corporations, can enhance or embed the novelties developed by small entrants into existing products and production processes and bring them to a global scale. Acquisition of innovative entrants is one way in which the complementary roles may be realized, and thus exemplifies one form of interaction between new technology-based firms and established firms that could contribute to system-wide innovativeness and economic growth.
To motivate our case in point, we draw on four main sets of literatures: the innovation systems literature (Carlsson
2006,
2007; Lundvall
1992; Nelson
1993), the literature on the origins of new technology-based firms and university and corporate spinoffs (Andersson and Klepper
2013; Klepper
2015; Lindholm Dahlstrand
1997), the literatures on commercialization strategies of innovative start-ups (Norbäck and Persson
2014; Gans and Stern
2003), and the symbiosis between established and large firms (Andersson and Xiao
2016; Baumol
2002; Lindholm Dahlstrand
1997). Insights from these literatures allow us to articulate what makes the system in the sense that we introduce spinoffs and acquisitions as distinct functions that are directly linked to behaviors and incentives of well-defined actors. We also elucidate how these functions are related to system-wide entrepreneurial experimentation and how they induce feedback effects and interdependencies in system-wide entrepreneurial experimentation capable of generating innovations and economic growth.
5 Summary and concluding remarks
An institutional environment that facilitates experimentation is central to maintaining a vibrant system of innovation. We argue in this paper that the central function of entrepreneurial experimentation in innovation systems is the creation, selection, and scaling-up of innovations. Entrepreneurial experimentation relates both to the “supply-side,” in terms of the system’s capacity to develop a variety of new technologies and business ideas that become subject to selection (technical experimentation), as well as to the “demand-side,” in terms of the efficiency of selection and scaling up of innovations and businesses on the market (market experimentation). We also argue that both technical and entrepreneurial experimentation are critical for the development of well-functioning systems of innovation. Both forms of experimentation, combined with a willingness to let losing incumbents fail, may in fact be argued to constitute the underlying notion behind Schumpeter’s (
1942) process of “creative destruction.” The rate of entrepreneurial experimentation has implications for what types of innovations will occur, who will pursue them and when. As Stern (
2005) argued, “a favorable environment for entrepreneurship and a high level of economic experimentation go hand in hand”. Although experiments can be conducted in large companies or in the public sector, new technologies and innovative products are often commercialized by entrepreneurs and often cluster at particular times.
Reviewing the extant literature on innovation systems, it is our conclusion that it does not yet articulate the role of entrepreneurs and entrepreneurship in innovation systems. In particular, what is lacking is an analytical framework that, with reference to explicit micro-level mechanisms and processes of industrial dynamics, articulates how the behavior of entrepreneurs and entrepreneurship give rise to system-wide entrepreneurial experimentation that creates, selects, and scales up new technology and innovations.
This paper tries to fill this gap in the literature by focusing on entrepreneurial activity as the mechanism through which innovations are converted into economic activity and growth. In emerging innovation systems, radical innovation is typically carried out by new entities—university or corporate spinoffs or de novo enterprises—whereas in mature innovation systems, innovation is more incremental and carried out by incumbent firms through diversification and acquisition. Entrepreneurial experimentation comprises both “technical” and “market” experimentation and is the active ingredient in creating variety within the system. Entrepreneurial activity in the form of new entrants may intensify competition and lead to the weeding out of inefficient incumbents and thus contribute to the selection of viable products. It is also the vehicle for scaling up production; diversifiers from related industries often account for the early successful entrants into new industries. As a result, the nature of entry may change over the life cycle of the innovation system.
In this paper, we conceptualize entrepreneurship analytically in terms of its function in innovation systems rather than as an outcome of an innovation system. That is, from a systems perspective, we claim that entrepreneurship should be looked upon as a function, similar to the way the traditional innovation systems literature treats organizations and institutions when considering functions that determine a system’s ability to produce and exploit scientific discoveries and technological innovations that generate economic growth.
To take a step towards the development of a framework that accommodates these requirements, we identified two mechanisms for the entrepreneurial experimentation in the system: spinoffs and acquisitions. Both are micro-based and relate directly to processes of industrial dynamics. We make a strong case that the spinoff mechanism (both corporate and academic) is critical for the creation of high-quality new firms, and that the acquisition mechanism is important for the scaling up of exploitation activities in such firms. Building in particular on Baumol’s (
2002) conjecture of a symbiosis between new entrants and established firms, we argue that new firms and established incumbents have different advantages and disadvantages in entrepreneurial experimentation, and that they, in interaction with each other through spinoffs and acquisitions, fulfill different roles. Our argument is that a system of innovation where large and new firms interact through spinoffs and technology-related ownership changes, under certain conditions, can be highly conducive to innovativeness and growth. Using the concept of collaborative innovation blocs discussed by Elert and Henrekson (
2018), one could argue that large established firms, universities, and new entrepreneurial firms have complementary skills in the innovation system and that their interaction through processes of industrial dynamics are ways in which complementarities are exploited at a system-wide level.
Although our main focus is not on policy, the framework suggests that policy must consider both technical and entrepreneurial experimentation, which means that a careful consideration of the broader institutional framework is necessary, not just the domain of traditional “innovation policy.” The experimentation view also suggests that there may be systematic market failures when the costs associated with experimentation are too high or the returns are too uncertain and far into the future. This means that the institutional framework needs to incorporate a long-term perspective. On a general level, an infrastructure of organizations that invest in new knowledge and technology combined with an institutional governance system that promotes interaction between incumbents and new firms and provides incentives of individual action is critical for the functioning of the innovation system. For example, for spinoffs to happen there needs to be an institutional framework that provides incentives for individual action and fosters labor market mobility. In fact, one may argue that a crucial feature of an innovation system is to provide incentives for individuals to engage in entrepreneurship and exploit opportunities (Braunerhjelm and Henrekson
2016).
Our framework reinforces the argument of Kerr et al. (
2014), i.e., that constraints on the ability of entrepreneurs and investors to experiment effectively can shape which industries, organizations, and time periods see the most radical innovations. It also sets the framework for understanding where barriers to experimentation may lead to market failures. When the time horizon for commercialization is extremely uncertain and distant, such as in the case of basic research, institutional regimes may be critical to enable experimentation in areas that are of importance to society but where a process of serial entrepreneurial experimentation by profit-seeking investors is unlikely to provide a set of stepping-stones to the technologies behind disruptive innovation.
Being mainly conceptual, the framework in this paper would need to be complemented and supported by further empirical research, particularly on spinoff and acquisition mechanisms. Important research questions to be considered are the following:
-
What are the different roles of entrepreneurial experimentation in transformative change?
-
Are the roles different in different sectors and at different points in time?
-
If so, why are they different? And to what effect?
Analyzing these questions includes studying the roles that ownership changes (acquisitions and different kinds of spinoffs) in different technological sectors over long time periods. It should also be considered that systems change over time, and that formative phases might be substantially different from phases of expansion or stagnation. This in turn is likely to have important policy implications.