Looking inside the spiky bits: a critical review and conceptualisation of entrepreneurial ecosystems

The spatial concentration of economic activity is one of the most enduring traits of contemporary capitalism (Marshall, 1890; Myrdal, 1957; Krugman, 1991; Fujita et al., 2001; Scott, 2006). Rather than a world of equal opportunity where globalisation and telecommunications eradicate the importance of geographical distance—the so-called ‘flat world’ thesis (Friedman, 2007)—the contours of the world economy appear inherently and, enduringly, ‘spiky’ (Florida, 2005). Arguably, technological advancement has accentuated this process making the world ‘more curved’ than ever before (McCann, 2008, p. 368; Rodríguez-Pose and Crescenzi, 2008).

As a consequence of these powerful centripetal forces, entrepreneurs are drawn to and, inextricably bound together, with other core entrepreneurial actors in close geographic, institutional and relational proximity. In recent years, entrepreneurial ecosystems (EEs) has become the latest conceptual ‘fad’ (Martin, 2015) seeking to help explain the dynamics of these entrepreneurial ‘spiky bits’ (Neck et al., 2004; Isenberg, 2011; Mason and Brown, 2014; WEF, 2014; Stam, 2015). While this systemic concept is intuitively appealing, its rapid adoption has tended to overlook the heterogeneous nature of specific ecosystems. Given the increasing attention being paid to the concept by scholars (Mason and Brown, 2014; Audretsch and Belitski, 2016) together with its increasing prominence in public policy circles (Isenberg and Onyemah, 2016), it appears timely to scrutinise the concept in greater depth to help develop this important area of research (Borissenko and Boschma, 2016).

Accordingly, this paper offers a critical review and conceptualisation of the ecosystems concept by unpacking its theoretical limitations, core dynamics, measurement approaches and use in policy-making. In doing so, it aims to provide a way forward for scholars and, potentially, help policy makers apply the concept more fruitfully. To highlight their heterogeneous and path dependent nature, a preliminary taxonomy of archetypal ecosystems is proposed. The paper concludes that entrepreneurial ecosystems are highly variegated, multi-actor and multi-scalar phenomenon which therefore requires bespoke policy interventions. The remainder of the paper is as follows. First, it provides a review of the literature then assesses the definitional ambiguities surrounding the concept. It then delineates the main components of EEs. It then critiques how scholars have assessed, measured and conceptualised the concept as well as proposing a classification of EEs. In the penultimate section, policy implications are explored before conclusions are offered.

Analysing the process of spatial agglomeration has been a source of great fascination for scholars during the last century and more. This began with Marshall’s (1890) pioneering analysis of the industrial concentrations in Victorian England which led to the identification of ‘agglomeration economies’. Firms accrue multiple pecuniary and non-pecuniary advantages firms from spatial co-location with firms in the same sector, notably the development of specialised pools of human capital, the creation of specialist suppliers and the creation of specialist infrastructure benefiting firms in the same sector (Marshall, 1890). These Marshallian externalities contrast with so-called Jacobian externalities (Jacobs, 1969). Jacobs argued that knowledge may spillover between unrelated industries within urbanised agglomerations as ‘ideas developed by one industry can be applied in other industries’ (van der Panne, 2004, p. 595). However, the literature is largely inconclusive over whether Marshallian specialisation or Jacobian diversification externalities most favours regional innovation (van der Panne and Van Beers, 2006).

Marshall’s work attracted little attention for much of the twentieth century when neo-classical economics dominated. However, interest in geographical foundations of industrial success re-awakened during the 1980s and 1990s when the ‘new economic geography’ took centre-stage (Feldman and Tavassoli, 2015). While some scholars question the true novelty of these perspectives (Martin, 1999), instrumental to this trend was the large number of investigations on the evolutionary dynamics of the so-called ‘Third Italy’—the dense networks of inter-related SMEs based around traditional industrial sectors such as ceramics, machine tools and textiles based in northern Italy (Piore and Sabel, 1984; Harrison, 1992). In the 1990s, research shifted to geographical ‘clusters’ (Porter, 2000; Martin and Sunley, 2003). In these locations, firms were thought to benefit from the local sectoral specialisation and knowledge spillovers (i.e. both Marshallian and Jacobian externalities). Socially embedded ‘untraded interdependencies’ (external to the firm but internal to a region) provided firms with access to human capital, knowledge exchange and specialist suppliers (Scott, 1988; Storper, 1995). However, this phenomenon is highly variegated (Markusen, 1996). Adopting a similar neo-Marshallian perspective, Saxenian’s (1996) groundbreaking study contrasted the nature of Route 128 and Silicon Valley to show that different clusters operate in fundamentally different ways. Although both are highly successful regions, their internal dynamics and workings were deemed fundamentally different and distinctive (Saxenian, 1996).

Despite this long lineage of work, it has been Michael Porter whose name has become inextricably linked to the cluster concept. As well as stimulating a vast literature examining clusters his work had the important role of translating the concept for policy makers who then produced a panoply of cluster policies that were adopted around the world (Martin and Sunley, 2003). This, in turn, propelled the cluster concept even higher up the agenda for policy makers (Hospers, 2006) and the quest to generate the next Silicon Valley become something of the ‘holy grail’ for regional policy makers intent on ‘replication’ (Neck et al., 2004; Isenberg, 2011; Feldman, 2014). Indeed, it is hard to overstate the prevalence of ‘cluster policy’ which became the ‘principal tool deployed by places to generate a strong and sustained economic performance’ (Audretsch, 2015, p. 7).

In parallel with the growth of the cluster literature, innovation scholars devised the concept of innovation systems to understand the systemic processes underpinning localised knowledge generation and transfer (Freeman, 1995; Lundvall, 2010). A key focus of this concept is its emphasis on the relational aspects between different institutional actors and how this facilitates the innovation process. Economic geographers were quick to see the attractiveness of this approach to examine the regional construction of the knowledge architecture in many economies by initiating the concept of regional innovation systems (or RIS) (Cooke et al., 1997; Asheim et al., 2011). Key actors within a RIS are universities, research organisations, technical training colleges, regulatory bodies and venture capitalists. These actors are thought to play a key role in orchestrating the innovation process within regional economies (Cooke et al., 1997). The novel feature of this work was the emphasis it placed on the wide array of institutional actors which play a role in the construction and dissemination of knowledge.

On a micro-level, the related concept of technological ‘competence blocs’ was introduced by scholars to denote the role different institutional actors play in aiding the transformation of knowledge into commercial products in science based contexts in and around university campuses and science parks (Eliasson, 2000). Competence blocs are the defined as the infrastructure necessary to create, select and diffuse new ideas throughout clusters of firms (Eliasson and Eliasson, 1996). Eliasson viewed these institutional factors as crucial for aiding Marshallian externalities. These systemic approaches to innovation have had a powerful impact in shaping industrial policy across many OECD economies (Carlsson et al., 2002; Warwick, 2013).

Following in the wake of these concepts, EEs is the latest conceptual tool designed to shed light on these centripetal agglomerative forces. In contrast to the innovation systems literature where institutions play an overarching role, within the EE construct entrepreneurs are the central actors. The EE literature ‘aims to explain (ambitious) entrepreneurship’ from a systemic perspective (Borrissenko and Boschma, 2016, p. 14). While entrepreneurial agency is at the core of this concept, it cannot be viewed in isolation given that ‘entrepreneurship takes place in a community of interdependent actors’ (Stam, 2015, p. 1761). This implies that the construction of successful entrepreneurial regions is not simply a function of firm-specific attributes, but is mediated by the wider context within which ventures operate (Mason and Brown, 2014). Emphasising the relational and institutional foundations (and synergies) of economic success within localised contexts suggests that the dynamic interactions between actors produces more than the ‘sum of their parts’. Unsurprisingly, much of the work on EEs strongly corresponds with the systemic literature on innovation systems outlined above (Borissenko and Boschma, 2016), especially the focus on the relational elements between multi-actor networks within regions which govern entrepreneurship and knowledge creation.

The theoretical concept of EEs has been informed by a variety of different literatures. Various scholars have attempted to explore and interpret ecosystems with respect to their social, cultural, behavioural, institutional and biological determinants. Indeed, the fusion of these diverse perspectives is perhaps its strongest asset but at the same time makes definitional and measurement issues extremely problematic (Audretsch and Belitski, 2016). While scholars have traditionally viewed the process of new firm formation through a resource-based lens (Garnsey, 1998), more recently greater attention has been attributed to the behavioural, social and cultural underpinnings of entrepreneurship (Sarasvathy, 2001; Baker and Nelson, 2005; Welter, 2011). The primary outcome from this wider research focus has been a much stronger focus on the ‘localised’ determinants of entrepreneurship (Acs et al., 2014; Autio et al., 2014). Arguably, this stronger focus on the multiple relational and spatially embedded processes which coalesce to shape entrepreneurial ecosystems amounts to something of a ‘spatial turn’ within entrepreneurship. Indeed, it is combination of distinctive localised characteristics—be they social, institutional and relational—within ecosystems which lend the concept its explanatory power.

The EE concept therefore offers a potentially new and insightful theoretical framework for analysing the underlying dynamics of how new venture formation occurs and is more plentiful and growth-oriented in certain geographical locations than others. But while conceptually and intuitively appealing, the literature lacks a common understanding of what EEs are and has not been subjected to sufficient rigorous theoretical and empirical scrutiny (Stam, 2015). Furthermore, some early work adopted a normative approach speculating how to ‘create’ ecosystems (Cohen, 2006). While attractive to policy makers, this lack of specificity increases the likelihood of misapplication. Like clusters, how ecosystems become established and evolve is varied (Martin and Sunley, 2011). Specifically, there is ‘little understanding’ of how successful ones come into existence (Feldman and Braunerhjelm, 2006, p. 1). EEs are heavily path dependent and rooted in their historical and institutional trajectory (Nelson and Winter, 1982; Feldman and Braunerhjelm, 2006). Consequently, every ecosystem is unique with its own distinctive idiosyncrasies and characteristics which are spatially, relationally and socially embedded.

EEs have been subject to a large degree of interpretative flexibility, which is largely a function of their diverse intellectual antecedents. The biological metaphor was first coined by James Moore (1993) when he stated ‘business ecosystems condense out of the original swirl of capital, customer interest, and talent generated by a new innovation, just as successful species spring from the natural resources of sunlight, water, and soil nutrients’ (Moore, 1993 p. 76). This link to biological ecosystems denotes the complex interactions and interdependencies which shape EEs. Drawing on work undertaken on biological ecosystems these evolving organisms can be viewed as ‘prototypical examples of complex adaptive systems’ where ecosystem properties constantly change and evolve, often in non-linear ways (Levin, 1998, p. 431). As Moore states ‘innovative businesses can’t evolve in a vacuum’ (p. 75). While originally introduced by Moore (1993), it has been the US entrepreneurship academic Daniel Isenberg who popularised the concept, particularly amongst non-academic audiences (Isenberg, 2010, 2011). Indeed, while the concept can claim multiple ancestries (Mason and Brown, 2014), its origins are in the ‘grey’ business literature and practitioner communities (Isenberg, 2010; Napier and Hansen, 2011; Feld, 2012). Accordingly, it offers a practical perspective rather than a purely theoretical one (Kantis and Frederico, 2012; Isenberg, 2011; Napier and Hansen, 2011).

There is no uniform or commonly accepted definition of EEs however various attempts have been made recently to define the concept. Typically, definitions convey the importance of localised and interdependent relationships between different entrepreneurial actors as key ingredients driving the performance of EEs. Stam (2015) offers a broad definition of EEs as a ‘set of interdependent actors and factors coordinated in such a way that they enable productive entrepreneurship’ (p. 1765). Mason and Brown (2014, p. 5) set out a more comprehensive definition of EEs as a ‘set of interconnected entrepreneurial actors, entrepreneurial organizations, institutions and entrepreneurial processes which formally and informally coalesce to connect, mediate and govern the performance within the local entrepreneurial environment’. Clearly, the dynamic and systemic nature of the concept encompasses multiple actors, institutions and processes.

The concept has been used differently within the literature making it a rather ‘chaotic’ (Martin and Sunley, 2003) or ‘fuzzy’ concept (Markusen, 1999). According to some, they represent ‘a conceptual umbrella encompassing a variety of different perspectives on the geography of entrepreneurship rather than a coherent theory’ (Spigel, 2015, p. 1). Part of this confusion stems from the fact that the term has been applied in both geographical and non-geographical contexts. As Moore (1993) did himself, some view ecosystems as a facilitator of innovation where different actors interact and work to help cumulatively co-produce new knowledge (Malecki, 2011; Zahra and Nambisan, 2011). The example of Google’s Android innovation ecosystem is often held up as an exemplar of this phenomenon (Nambisan and Baron, 2013). This approach has considerable value and has been shown to reveal powerful insights how a population of technologically differentiated firms ‘mutually and iteratively discover, create and enact innovation opportunities’ (Best, 2015, p. 12).

However, the majority of observers view ecosystems primarily as a spatial concept (Feldman and Braunerhjelm, 2006; Mason and Brown, 2014; Audretsch and Belitski, 2016) to explain why certain places have high levels of entrepreneurial activity (Spigel, 2015; Stam, 2015). So rather than having innovation at its core, entrepreneurship is the fundamental driver behind the concept. In this paper we view EEs through this lens.

One aspect of EEs often overlooked in the emerging literature is the fundamental role played by social and cultural factors in shaping entrepreneurship (Venkataraman, 2004). In many respects, entrepreneurship takes place within the framework of ‘sociocultural structures’ (Spilling, 1996, p. 92) which are fundamentally determined locally and are heavily path-dependent (Gertler, 2010; Welter, 2011). A good example of the deeply embedded nature of entrepreneurship is the world-renowned German Mittelstand companies with their close connections to regionalised banks, schools, local government and research centres (De Massis et al., 2017). Other scholars have shown how the experience of Israeli entrepreneurs has been heavily shaped by military experience where non-hierarchical structures encourage problem-solving and innovative thinking (Senor and Singer, 2009). In other words, social and organisational ties are deeply ‘intertwined’ (Ferrary and Granovetter, 2009).

Another issue that has been overlooked concerns the temporally unfolding and evolutionary nature of EEs (Borissenko and Boschma, 2016). A key feature of ecosystems is nonlinearity. Significant changes can occur within ecosystems over time, not least due to changes in government policy, resulting in multiple possible outcomes in their developmental trajectories. This kind of evolutionary thinking has shaped the way in which scholars conceptualise the evolution of clusters (Martin and Sunley, 2011). Given their similarities, such an approach may be an equally fruitful mechanism for understanding EEs.

In summary, the initial conceptualisations of EEs appear to be somewhat under-socialised, lacking a time dimension and fail to incorporate the full complexities of the socio-spatial context mediating entrepreneurship. These omissions are evident in the mechanistic forms of measurement which have been utilised to assess ecosystems. This myopic focus is common within the entrepreneurship literature as a whole which is preoccupied with characteristics and behaviours of individuals and firms (Borissenko and Boschma, 2016), whilst often ignoring how the context of firms regulates their ‘behaviour, choices, and performance’ (Autio et al., 2014, p. 1099).

The concept of ecosystems is an inherently dynamic one which acknowledges the importance of entrepreneurial processes and cognitive belief systems which underpin interactions within an economy. Just as there is an evolutionary logic to cluster formation (Feldman and Braunerhjelm, 2006), some observers note that ecosystems are ‘a naturally evolving system’ (Isenberg, 2010). Figure 1 presents the diverse types of actors within EEs.

In contrast to some practitioner viewpoints (Feld, 2012), a myth we wish to refute is that ecosystems are predominantly about start-ups. According to some, this focus is symptomatic of the misplaced obsession with SMEs and start-ups in terms of their ability to generate innovation and growth (Isenberg, 2012; Fritsch and Storey, 2014; Mazzucato 2014), despite significant evidence to the contrary (Nightingale and Coad, 2014; Shane, 2009). As Vivarelli (2004, p. 48) notes ‘the conventional wisdom is that start-ups are good per se and that all the potential entrepreneurs have to be helped’. However, research in both advanced and developing countries (Vivarelli, 2004; Vivarelli, 2013; Quatraro and Vivarelli, 2015) shows that this tends to encourage low-quality entrants and associated deadweight (Shane, 2009). This owes to the underlying causal drivers of entrepreneurship which often fosters so-called ‘necessity entrepreneurs’ (Baumol, 1996) pushed into entrepreneurship by ‘regressive’ drivers such as fear of unemployment (Vivarelli, 2013). Growth-oriented opportunity entrepreneurs on the other hand have a very different entrepreneurial and behavioural make up (Baumol, 1996; Amit et al., 2001).

The crucial aspect of ecosystems are the actors, processes and institutions which are not directly related to start-ups, such as large firms, universities, public sector bodies, health care systems, banks and stock markets (Isenberg 2010; Mason and Brown, 2014). Entrepreneurial universities, in particular, are widely seen as a vital entrepreneurial actors (Guerrero et al., 2016) while others take the view that their role has been over-stated (Brown, 2016). The role of large existing firms, in contrast, is often downplayed with the EE literature. However, there is considerable evidence which shows that large incumbent firms often play a central role in configuring some ecosystems (Mayer, 2013), as attractors of skilled labour (Harrison et al., 2004); the incubation of entrepreneurs, the spill-over of knowledge and as important initial customers (Eliasson, 2000). In some areas, large exogenous defence companies play a central role in shaping EEs (Adams, 2011). These roles are pivotal in configuring the nature of a local entrepreneurial context but tend to be overlooked in some of the EE literature (WEF, 2014; Spigel, 2015).

One thing ecosystems do have in common is a spatial boundedness. Close geographic proximity fosters network formation and knowledge exchange and most ecosystems follow a strong spatial logic (Glückler, 2007). While some are focused around large urban conurbations, others operate at a regional level or in some rare case at the national level (e.g. Estonia). In some cases, EEs are characterised by more complex ‘nested geographies’ (i.e. EEs located within larger EEs) which involves multi-scaler interactions with other entrepreneurial actors on a number of different spatial levels, both domestically and internationally. So while the role of local institutions matters, so too does the ‘connectivity between those elements’ (Motoyama and Knowlton, 2016, p. 20). This is particularly important for enabling learning processes which are ‘intrinsically social and collective phenomena….involving joint contributions to the understanding of complex problems’ (Teece, et al., 1994, p. 15). This echoes other work which emphasises how networking and relational factors are crucial for early stage ventures more generally (Sullivan & Ford, 2014; Witt, 2004).

In order to delineate the main actors, interactions and cognitive mind-sets within these complex systems, we propose a taxonomy featuring four main coordinative aspects of EEs which we will consider in turn: entrepreneurial actors; entrepreneurial resource providers; entrepreneurial connectors and entrepreneurial culture (see Fig. 2 below).

A growing number of supranational bodies like the OECD, national governments, think tanks and economic agencies around the world are actively embracing the ecosystems concept as a tool for policy making in the sphere of entrepreneurship (Coutu, 2014; Mazzarol, 2014; WEF, 2014; Stam, 2015). However, policy formulation runs the danger of running ahead of its theoretical and empirical underpinnings. Given the fundamental peculiarities and specificities within the internal workings and dynamics of ecosystems, there is no standardised strategy for effectively developing EEs (Audretsch, 2015). Yet, from our reading of most ecosystem policy frameworks, the adoption of bespoke policy frameworks does not seem immediately evident.⁷

Three main beliefs seem strongly embedded in current policy approaches. First, when policy makers refer to the ecosystems concept they invariably pre-fix the term with the term ‘start-up’ (see, Schreiber and Pinelli, 2013; WEF 2014). This could be labelled the ‘start-up monoculture’ (Stangler, 2015). This can be damaging for several reasons. It ignores the fact that the needs of firms change as they evolve. For example, if new ventures grow and upscale they will need to migrate away from the actors and interactions which facilitated their initial development. By just focusing on start-ups, some ecosystems may not have the full range of actors and interactions necessary to upscale businesses. In many economies, there are sufficient new entrants (Acs et al., 2016) but insufficient quality entrants (Nightingale and Coad, 2014). Consequently, the view that a focus on start-ups is ‘bad public policy’ (Shane, 2009) has gained currency, arguing that policy should focus on growth-oriented firms instead (Mason and Brown, 2013). Other scholars claim much greater focus should be granted to promoting innovative start-ups rather than the majority of start-ups who engage in ‘copycat’ entrepreneurship (Colombelli et al., 2016).

Equating ecosystems purely with start-ups is therefore misconceived and potentially misleading (Isenberg, 2011; Mason and Brown, 2014). Ecosystems need more innovative scale-ups and blockbusters (Napier and Hansen, 2011; Coutu, 2014), indicative of a shift in policy on quality rather than quantity of entrepreneurship (Stam, 2015). How this translates into concrete policy proposals is a challenging task given the considerable difficulties of identifying firms with ‘blockbuster potential’. While not wishing to offer specific policy prescriptions to help firms scale-up, policy makers will most likely require a different set of policy instruments to those designed to promote entrepreneurial entry. In this respect, peer-based relational support mechanisms which foster ‘adaptability, enhance the dynamic capabilities and increase the outward orientation of firms’ may be more effective than traditional transactional forms of support such as grants and assistance with funding typically targeted towards HGFs (Brown and Mawson, 2016, p. 830). Given start-ups programmes have proved highly ineffective at counteracting the powerful path-dependent nature of entrepreneurship within economies (Van Stel and Storey, 2004; Fritsch and Storey, 2014), designing appropriate policies to foster scale-ups may prove equally illusive.⁸

The second core element embodied in current policy frameworks is the overwhelming focus on technology-based form (TBFs), and university spin-off companies (USCs) in particular. This is common in cluster policies, innovation policies and most entrepreneurship policies (Brown and Mason, 2014). This often entails high-tech firms being accorded a central place as dynamic drivers of knowledge-based economies. However, this emphasis on high-tech firms is incompatible with the composition and industrial structure found in most ecosystems. It is virtually impossible in regions lacking high-tech industries to build a strong technology base (Qian et al., 2013). By heavily promoting these forms of activities in economies with little prior core competencies or advantages, policy makers are ignoring the evolutionary logic and path-dependencies which shape their home environments. Given that most successful ecosystems typically build upon existing agglomerative forces based on particular industries or groups of industries, ensuring that these existing competencies are developed is vitally important (Isenberg, 2011; Best, 2015). If scaling-up firms is the hallmark of successful ecosystems, it would appear that the promotion of firms irrespective of their sectoral background should be a feature of systemic entrepreneurship policies.

A final feature that is evident in many of the ‘new’ policy frameworks is their reliance on traditional narrowly-based policy interventions. The vast majority of interventions promoted under the auspices of ecosystems strongly resemble single-actor interventions—utilised under other conceptual frameworks like clusters, innovation systems and learning regions (e.g. promoting start-ups, fostering networks, incubators, public sector VC etc.)—which focus on assisting the creation and growth of high-tech firms. Most interventions target particular network actors, be they entrepreneurs, universities or firms. While this kind of support can benefit some—the Arch Grants programme in St Louis being one such example (Motoyama and Knowlton, 2016)—this does not in itself constitute systemic entrepreneurship policy. Unlike the field of innovation policy which reveals considerable experimentation with systemic interventions designed to assist and connect innovation systems as a whole (Smits and Kuhlman, 2004; Wieczorek and Hekkert, 2012), entrepreneurship policy rarely attempts to help foster vertical connections across ecosystems. Yet arguably, in order to help foster ‘blockbuster entrepreneurship’ more system-wide and outward-oriented approaches are likely to be more effective.⁹

Like others, we see considerable merit in adopting a systemic approach to aid our understanding of the world’s entrepreneurial ‘spiky bits’. The strength of this concept is, undoubtedly, the emphasis it places on the role of entrepreneurial ‘agency’ in shaping EEs. However, when examining EEs it is important both structure and agency are examined together to appreciate the full complexity of the dynamics of entrepreneurial activity in any given context. At present, scholars and policy makers have failed to comprehend the full complexity of these complex organisms. Crude forms of measurement approaches towards EEs could potentially amplify this problem. Our proposed taxonomy is therefore an attempt to delineate this complex, variegated and temporally discontinuous phenomena.

Importantly, misconceptions about the nature and functioning of entrepreneurial ecosystems create the potential for misconceived policy interventions, signified by the continued emphasis on start-ups and the lack of genuinely systemic policy instruments.¹⁰ Given their pervasive heterogeneity, there is unlikely to be a ‘one-size fits all’ policy prognosis for developing different types of ecosystems. Scholars therefore need to further dissect, conceptualise, theorise and empirically examine this complex phenomenon much more closely to move our understanding forward. This paper aims to provide a start in this direction.