Entrepreneurial ecosystems and clusters: How can economic geographers advance debates for regional development?

The entrepreneurial ecosystem concept is now one of the most popular policy tools for regional development following a surge of interest in entrepreneurship-oriented academic circles, yet has experienced little critical engagement within economic geography discourse. We argue that economic geographers should engage with the entrepreneurial ecosystem concept because (1) it describes a shift in spatial socio-economic organisation that has thus far been underexplored by economic geographers and (2) it is an inherently chaotic concept that requires significant conceptual development, not least in relation to the cluster concept. The entrepreneurial ecosystem concept is considered a close relative and potential successor of the cluster concept, which itself rapidly achieved policy stardom despite academic concerns over its conceptual clarity. We argue that there are significant similarities and intersections between the two concepts with implications for broader regional development literatures, enabling economic geographers to enrich academic debates and consequent policy decisions.


I Introduction
The entrepreneurial ecosystem (EE) concept has achieved rapid expansion in academic and policy fields in the last 15 years.The EE approach is seen as bringing together different research traditions across entrepreneurship and regional development, enabling scholars to 'open up new research questions and avenues of inquiry into both policy-related issues regarding how to support economic growth and prosperity as well as more fundamental social science questions such as the relationship between structure and agency in modern capitalism' (Wurth et al., 2022: 2).Such is the growth of this literature, the term has surpassed 'buzzword' status and moved firmly into international policy discussions (Mason and Brown, 2014).
However, this entry into international policy domains has transcended its conceptual certainty, and there is a risk of bad policy prescriptions (Wurth et al., 2022).Fundamental questions still concern exactly what EEs and their inner workings are, and for key proponents of the concept such as Stam (2015Stam ( : 1764)), 'the entrepreneurial ecosystem approach has run ahead of answering many fundamental conceptual, theoretical, and empirical questions'.Accordingly, definitions are wide-ranging and broad, often serving to muddy rather than clarify the concept, with no clear definition emerging (Cao and Shi, 2021;Malecki, 2018).Furthermore, Wurth et al. (2022) highlight that there are different ontological and epistemological perspectives that underpin uncertainty in the empirical basis for EEs: for some they are perceived as densely populated concentrations of entrepreneurial actors in particular areas of cities and regions that are actively created, while for others they are pre-existing frameworks of variables that can be identified across all regions and countries, that have different levels of intensity and development depending on their success.Relatedly, this raises further questions about whether EEs are a new phenomenon that correlates in age with the literature and requires new theorisation, or that can be retrospectively identified (Stam and van de Ven, 2021;Wurth et al., 2022)?Ultimately, what is meant by an 'entrepreneurial ecosystem' is unclear and prone to different interpretation.
There is an interesting parallel here between the development of the EE literature and the cluster literature in economic geography, from which the EE concept owes significant conceptual lineage (Acs et al, 2017;Rocha and Audretsch, 2022;Spigel and Harrison, 2017).Borne out of Porter's (1990) accessible business language and successful marketing as a policy prescription, the cluster concept quickly rose to prominence as an international policy tool for regional development, outstripping its own conceptual development.Martin and Sunley's (2003) seminal critique highlighted how the cluster concept, while not without merit, was built on such shaky conceptual foundations that it should be considered a chaotic concept.They urged that the cluster concept needed significant conceptual tightening for effective policy usage.Similarly, we contend that the EE concept is chaotic, in need of significant conceptual development.
We argue not to rubbish the notion of the EE concept in favour of clusters, quite the opposite, as we believe that EEs capture a significant shift in the global capitalist economy that economic geographers may have underexplored.However, we believe that the literature on EEs is in danger of reinventing the wheel, while economic geographers have over three decades of progressive research on clusters as resources and lessons on how to develop the EE concept further.Furthermore, we argue that the distinction between EEs and clusters is oversimplified in the entrepreneurship literature, which has attempted to distinguish EEs from clusters rather than understand the relationships between them.Consequently, this represents fertile land for economic geographers to develop, applying contemporary debates such as on agency, evolution and multi-scalarity, to understand the relationships between the concepts and for understanding broader regional development.Moreover, the inherent spatiality of the EE concept means that this is a topic that economic geographers should not be ceding ground on to academics in other disciplines (James et al., 2018).
In Section 2, we explore the popularity of the EE concept, identifying four reasons for its rapid upsurge, which echoes the trajectory of the cluster concept.Following this, Section 3 highlights the conceptual confusion of the EE concept, investigating five areas in which the chaos of the concept shines through.Section 4 highlights the two major ways in which EEs are distinguished from clusters in the EE literature, arguing that these distinctions are not as clear as suggested, which is further inhibiting the development of the EE concept and restricting the mutual development of the two concepts.We believe that this sympathetic critique offers considerable upside for economic geographers, who can expand on understandings of regional development through engaging the EE and cluster concepts, and Section 5 explores three possible avenues for this engagement.

II Why entrepreneurial ecosystems:
The parallels with the cluster concept The EE concept is the most recent approach to understanding the uneven geography of entrepreneurship and the rise of the entrepreneurial economy.The perception of entrepreneurship as a crucial driver of economic growth in modern economies started with Birch's (1979) study on 'The Job Generation Process', which contrasted the prevailing opinion at the time by finding that small and young enterprises were responsible for significant employment growth between 1969 and 1976.Broadly speaking, an EE consists of 'interdependent actors and factors that enable and constrain entrepreneurship within a particular territory' (Stam and Van de Ven, 2021: 1).It extends older economic geographical work on entrepreneurship that focused mainly on the effects of regional economic structures on entrepreneurship (Sternberg and Wagner, 2004), and extends older entrepreneurship research focused on individual motivations or characteristics (Lumpkin and Dess, 1996), by integrating external factors in the decision to form a firm.However, as EE scholars themselves point out, the idea of interdependent actors colocating in places to produce mutual benefit is not new (Stam, 2015).Acs et al. (2017) document the lineage of the EE approach to the strategy and regional development literatures.The former literature proposed 'business ecosystems as a form of economic coordination in which a firm's ability to create and appropriate value critically depends on different groups of actors that produce complementary products or services' (Acs et al., 2017: 2, original emphasis).The latter is a collection of literatures such as the cluster, technopole, industrial district or regional innovation system literatures, that share a focus on how predominantly regional systems are developed that increase the innovativeness or productivity of firms located there.While following these literatures in taking a focus on how predominantly regional systems of interconnected actors can provide a beneficial environment for firms, Acs et al., (2017: 3) distinguish the EE concept by suggesting 'the novelty of the entrepreneurial ecosystem approach lays in the focus on (productive) entrepreneurship as an output of the ecosystem'.
However, as other EE scholars have pointed out, the focus on entrepreneurial actors co-locating in places that provide mutual benefit is also not new, not even within the strategy and regional development literatures (Stam and van de Ven, 2021).Within entrepreneurship studies, the 1980s and 90s saw a movement away from individualist notions of entrepreneurship towards broader community perspectives (see, for example, Aldrich, 1990;Dubini, 1989;Pennings, 1982), with the concepts of entrepreneurial infrastructures (van de Ven, 1993) or environments (Gnyawali and Fogel, 1994) explored.
Similarly, entrepreneurship has often been considered a core component in a long lineage of geographical concepts in the regional development literature.For example, technopoles have been suggested as places where 'entrepreneurial activity thrive[s]' (Scott, 1990(Scott, : 1602)), industrial districts have been seen as encouraging a 'particular type of small firm, in which the entrepreneurial character and dynamics are extremely high' (Fumagalli and Mussati, 1993: 29;also, Staber, 1997), while the cluster literature itself has a wealth of studies on entrepreneurship (see, for example, Feldman and Francis, 2006;Feldman et al., 2005;Mason, 2008).Why then, given this history of attention to productive regional systems and to entrepreneurship within them, is the EE concept attracting such interest now?
While the relationship of the cluster concept to the EE concept is described as one of conceptual ancestry, there are also strong parallels in terms of their trajectories in becoming popular concepts.Indeed, just as the idea of interdependent actors co-locating in places to produce mutual benefits is not new to EEs (Stam, 2015), so was the central cluster idea of specialised co-location not new to Porter (1990).Marshall (1890) had written on the idea a century before, and more recently economic geographers had identified various similar concepts in the 1980s, such as new industrial spaces (Scott, 1988) or industrial districts (Amin, 1989).Indeed, by looking at the rise in popularity of the cluster concept, we can see a mirrored pattern based on four factors that explain the surging popularity of the EE concept.
Firstly, the emergence of both cluster and EE concepts can be seen in relation to the changing global capitalist economy.The cluster concept gained popularity at a time when the global economy was shifting.With the transformation from Fordist mass production to post-Fordism, the role of spatial proximity in the organisation of economic activity was re-discovered (Amin, 1994).A network-based production system of vertically disintegrated firms that was able to quickly adjust to changing demand conditions evolved (Piore and Sabel, 1984).Firms grew that were able to quickly adjust to these changes.Spatial proximity reduced transactions costs to deal with this volatility, which resulted in the emergence of new concepts like regional production systems (Becattini, 2002), newly industrialised spaces (Scott, 1988) and regional clusters (Porter, 1998).
Later, in the development towards a knowledgebased learning economy (Lundvall and Johnson, 1994), the focus shifted away from the flexible organisation of production towards the prioritisation of innovation.In the knowledge-based theory of the regional cluster, firm competitive advantage lies in their access to regionally embedded knowledge that is difficult to transfer and in innovation-oriented knowledge exchanges facilitated by geographical proximity (Malmberg and Maskell, 2002).The knowledge-based cluster (and the resulting theory) was thus a response to global changes in capitalism, where the determinants of firm competitiveness moved from cost to innovation (Johnson and Lundvall, 1994).
The EE represents another shift in socio-economic organisation.A quick perusal of the world's largest companies by market cap sees increasing domination by technology companies which have grown rapidly and penetrated our daily lives.Increasingly, the success of economies is being measured by the number of 'unicorns' or similar buzzwords such as 'gazelles' that they host, along with the size of venture capital flows (Aldrich and Ruef, 2018;Bos and Stam, 2014;Florida, 2016).While the cluster literature describes embeddedness in networks and institutions as crucial elements for firms (Amin, 1994), digitalisation has meant that digital platforms providing marketplaces, storage, communication tools and other services are becoming increasingly important (Kenney and Zysman, 2016).
Relatedly, a second similarity between the concepts is therefore that they offer explanations for shifts in spatial inequalities.Indeed, while some researchers predicted digitalisation and globalisation would see the death of geography, this unevenness has increased in recent decades, manifesting a 'spiky world' (Florida, 2006).These developments follow the argument by Maskell and Malmberg (1999) that the more everything is available everywhere, competitive advantages increasingly rely on factors that cannot be transferred easily, like regionally embedded competencies, networks or institutions.
The cluster literature explains the different development of regions through differences in institutions, specific regional capabilities and networks.It emphasies the importance of geographically close connections to established firms, production complexes, industrial parks and R&D centres.Successful new firms were often spinoffs closely located to their parent firm (Klepper, 2007) and, accordingly, typically located in industrial or science parks on the periphery of cities or around universities (Castells and Hall, 1994).As a result, the region was the dominant scale of relations within clusters.For EEs, proximity to investors is becoming more valued than proximity to engineers (Feld, 2012), and the importance of localised start-up communities more prevalent (Rossi and Di Bella, 2017).As a result, inner-city areas and, in particular, creative or bohemian areas (Nathan and Vandore, 2019) have become the nexus for these new start-up communities (Florida, 2016;Rossi and Di Bella, 2017).
This shift from a regional to an urban context is visible in Silicon Valley.The number of new startups in Silicon Valley annually has decreased since 2018, while the numbers of start-ups in San Francisco have strongly increased since 2010, and now outnumber Silicon Valley start-ups in sum (Siliconvalleyindicators.org, 2022).This demographic shift is accompanied by a spatial shift from Mountain View and Palo Alto to San Francisco and its South of Market Street area (Economist, 2014).Relatedly, Silicon Valley's ability to reinvent itself has meant it remains the pre-eminent focal point of the global technology industry and an example of a successful cluster or EE that many other governments attempt to replicate (Harris and Menzel, 2023).We now have ecosystem rankings (e.g., Startup-Blink; StartupGenome) and ecosystem 'playbooks' (Kauffman, 2019) for getting your EE higher up on these global rankings.'Entrepreneurial ecosystem' has itself become a buzzword in the tech industry and every city should aim for a globally competitive EE.EE proponents therefore argue that the concept is best placed to understand the economic fortunes of economies around the world (Acs et al, 2017).
While the changing global capitalist economy helps to explain why an entrepreneurship-focused concept has become so popular, it does not necessarily explain why the EE concept in particular, has become so popular.A third similarity with the growth of the cluster concept helps to explain this, namely, that early proponents of the EE concept echoed the doctrine of Porter and sold it on the basis of the competitiveness it can offer an economy.Isenberg's (2010) seminal essay The Big Idea: How to Start an Entrepreneurial Revolution started with the story of Rwanda's rise from 143rd to 67th on the ease of doing business ranking offered by the World Bank, citing Rwanda's president, Paul Kagame, to emphasise that 'entrepreneurship is the most sure way of development'.The essay then outlines nine prescriptions for governments to achieve the 'holy grail' of EEs, such as '5: Get a Big Win on the Board' or '6: Tackle Cultural Change Head-On'.In a follow-up essay, Isenberg (2014) offers a further set of clarifications for policy makers and emphasises the importance of getting EE policies right after the failure of 'dictated industrial policy, [and] barren "cluster" strategies'.Other key proponents offered similar language.Brad Feld's (2012: 1) equally notable book Startup Communities: Building an Entrepreneurial Ecosystem in Your City stated that 'entrepreneurial ecosystems are driving innovation, new business creation, and job growth', and the book 'documents the strategy, dynamics, tactics, and long-term perspective required for building communities of entrepreneurs'.Following this, the EE concept was readily endorsed by the OECD local employment and economic development programme, being thought of as a way to move from failed transactional policy-making towards interventions that encouraged relational forms of support (Mason and Brown, 2014).Evidently, EEs have often been presented as a policy panacea that can be achieved in any place, simply by following the sets of recommendations and prescriptions found in these texts.
A fourth parallel with the cluster concept is that the EE concept lends itself to wide, varied and ambiguous application (Malecki, 2018).The concept bypasses sectoral considerations with proclamations of its sector agnosticism, and the ecosystem metaphor navigates any attempts to put scalar limitations on it, so that it can be applied universally to any place no matter the size (Stam, 2015).While a bias exists towards the regional scale, EEs are now a concept that can be applied ubiquitously, that all policy makers can turn to.And turn to it they have: the sheer growth in popularity and excitement in the concept at academic and policy levels despite continued confessions from its strongest advocates that simple definitions have been eluded (Stam and van de Ven, 2021), echoes Steiner's (1998: 1) claim of clusters having 'the discreet charm of obscure objects of desire'.The allure of the concept for driving economic growth through entrepreneurship ultimately clashes with the concept's potential as a policy panacea offering development for all, because of the 'spiky world' that such entrepreneurship-inspired economic growth creates (Florida, 2006).The literature has struggled to address this potential contradiction because the concept has been used ambiguously and applied ubiquitously to places of all shapes, successes and scales.
Following the cluster concept, then, the EE concept has achieved its surge in popularity because of the changes occurring in capitalism that it purports to explain, changing geographies of economic activities, its acclaim as a policy panacea, and its ambiguity which lends to wide-ranging applications.There is no denying that entrepreneurship is an important driver of growth that warrants significant attention.However, concepts that emerge and grow quickly, securing rapid policy support, often lack conceptual rigour and are likely to be riddled with inaccuracies and complicationsjust as the cluster concept was (Martin and Sunley, 2003).Consequently, and as the next section develops, we believe that EEs warrant the 'chaotic concept' brand, and urgent clarification is needed to stop policy failure.

III Following the chaotic footsteps of clusters?
Despite its rapid rise in popularity, proponents are critical of the EE concept, noting it to be 'highly undertheorized, and not yet adequately measured' (Stam, 2018: 1), and 'mainly used metaphorically with unclear relationships to other theories of innovation and (regional) economic development' (Wurth et al., 2022: 1), with which there is significant 'confusion and ambiguities shrouded in these similar-sounding […] concepts' (Rocha and Audretsch, 2022: 18).While this ambiguity might be a result of its wide diffusion, conceptual clarity is needed to provide effective policy (Markusen, 1999;Martin and Sunley, 2003).In their example of the regional cluster, Martin and Sunley (2003) deconstructed the ambiguities of the concept.Their starting point for explaining the chaos of the cluster concept was in the ambiguity surrounding Porter's (1990) definition, which created a situation where 'we know what they're called, but defining precisely what they are is much more difficult' (Martin and Sunley, 2003: 10).They used the term chaotic concept to define those concepts that are 'conflating and equating quite different types, processes, and spatial scales of economic localization under a single, all-embracing universalistic notion' (Martin and Sunley, 2003: 10).The argument being that when a concept and its constituent components can have different meanings, interrelations and processes, which are all interchangeable, statements based on these chaotic concepts become arbitrary at best.Consequently, this section assesses to what extent different types, processes and scales are conflated under the EE concept.
Malecki's (2018: 6) review of the concept highlighted no less than 14 'selected definitions' of the concept to demonstrate the variations in definitions offered.This ambiguity is an ongoing problem, and it has become almost a necessary caveat of recent papers to declare that 'there is not yet a shared definition' (Rocha and Audretsch, 2022: 10), to outline how the 'entrepreneurship literature generally defines the entrepreneurial ecosystem' (Theodoraki and Catanzaro, 2022: 384, emphasis added), before offering their own particular definition.
While there is no shortage of available definitions for EEs (Malecki, 2018), for simplicity we provide three here as popular examples.The first is one of the most cited definitions of an EE, while the latter two are recent and widely accepted definitions offered in the literature: "a set of interconnected entrepreneurial actors, institutions, entrepreneurial organisations and entrepreneurial processes which formally and informally coalesce to connect, mediate and govern the performance within the local entrepreneurial environment" (Mason and Brown, 2014: 5).
"as a set of interdependent actors and factors coordinated in such a way that they enable productive entrepreneurship within a particular territory" (Stam, 2018: 2).
"as the regional collection of actors (such as entrepreneurs, advisors, workers, mentors, and workers) and factors (cultural outlooks, policies, R&D systems, and networks) that all contribute to the creation and survival of high-growth ventures" (Spigel et al, 2020: 484).
These definitions, and the others that can be found throughout the literature, all sound generally the same.However, there are important distinctions between them that create ambiguities.We identify five areas for analysing to what extent the conflation of different things under the same term takes place, namely, (1) the component parts, (2) the interconnections and processes, (3) the scales at which they exist, (4) their governance, (5) and their proposed outcomes.
A first area of uncertainty is understanding exactly which components, i.e., 'combinations of social, political, economic and cultural elements' (Spigel, 2017: 2), make an EE.Here, it is commonplace for authors to suggest EEs consist of different factors, each presenting different variables, and there is no definite agreement.At first glance, for Mason and Brown (2014) EEs consist of actors, institutions and processes, while Stam (2018) and Spigel et al. (2020) mention sets or collections of actors and factors.Within the broader literature there is significant variation in the number of these factors, their specificities and the language used to describe them.To provide a non-exhaustive and hasty overview, Isenberg (2010) offered nine prescriptions before suggesting six domains (Isenberg, 2014); Feld's (2012) seminal analysis of the Boulder ecosystem identified nine attributes; the World Economic Forum has eight pillars (WEF, 2013); Spigel (2017) has ten attributes or elements; and Stam (2018) suggests a total of ten framework and systemic conditions.
The literature appears to be rallying around Stam's (2018) ten framework and systemic conditions.However, uncertainty remains about exactly which factors or conditions should be included and their degrees of importance.For example, Spigel (2017) places a significant emphasis on 'mentors and role models' as one of his ten attributes while Stam (2018; Stam and van de Ven, 2021) make no mention of mentors but includes role models loosely under 'leadership'.Spigel ( 2017) also emphasises open markets, which are not exactly correlated to Stam's (2018) 'demand' element, while 'universities' again are thought to warrant their own attribute for Spigel (2017), but go unmentioned in Stam (2018), and receive negligible mentions in Stam and van de Ven (2021).Evidently, while there is significant overlap across many of these variables and a general consensus can be said to be solidifying, considerable ambiguity remains about what factors are important in EEs, which exemplifies the chaos of the concept as EEs mean different things to different authors.
A second area of uncertainty is the interconnections, processes and causal mechanics between these component parts.As with early cluster research, EEs are lacking knowledge about cause and effect, leading to claims of tautology where 'entrepreneurial ecosystems are systems that produce successful entrepreneurship, and where there is a lot of successful entrepreneurship, there is apparently a good entrepreneurial ecosystem' (Stam and van de Ven, 2021: 810).Within the literature it is recognised that early research typically provided 'only long laundry lists of relevant factors without a clear reasoning of cause and effect' (Stam, 2015(Stam, : 1764)), which meant the literature was 'still in search for a clear analytical framework that makes explicit what is cause and what is effect' (Alvedalen and Boschma, 2017: 893).In response, authors have pushed for more relational (Spigel, 2017) or process-oriented (Spigel and Harrison, 2017) perspectives, towards the utilisation of complex adaptive systems theory and its inclusion of upward and downward causal dynamics (Roundy et al, 2018), to better incorporate evolution and path dependence (Harris, 2021a), or to include interecosystem links (see, for example, Velt et al, 2020).
Stam and van de Ven (2021: 814-815) have sought to infuse Stam's (2018) list of ten framework and systemic conditions with some causality through three propositions that (1) 'The entrepreneurial ecosystem elements are mutually interdependent and co-evolve in a territory'; that (2) 'The ten observable entrepreneurial ecosystem elements explain the levels of entrepreneurial activity in a territory' through upward causation; and that (3) 'Prior entrepreneurial activities feedback into entrepreneurial ecosystem elements in a territory' through downward causation.While there is undoubtedly co-evolution along with upward and downward causation occurring in EEs, this circular conceptualisation does little to tackle the tautological argument of what comes first.Similarly, Spigel (2017) conceptualises the supportive and reinforcing processes between material, social and cultural attributes, again speaking to the ongoing co-evolution of EE attributes but offers little to explain their causal emergence.Roundy et al., (2018) suggest that it is the interplay between micro, meso and macro forces that influence the emergence of EEs as complex systems, but do not go further into identifying specific relationships and processes between specific components parts of EEs that are directly responsible.While this recent drive to label EEs as complex adaptive systems, fuelled by co-evolutionary dynamics and upward and downward causation at various scales, can be beneficial, it has thus far struggled to disentangle a discrete sense of causality.With all of the various and yet to be consensually defined component parts embroiled in constant co-evolution and complex interactions, it makes the question of 'How do I know it when I see it'? (Markusen, 1999: 870), it being causal dynamics and processes key to EEs, particularly difficult to answer.
A third area of uncertainty is the insufficient geographical precision for what is an inherently geographical concept; the above three quotes separately use the terms 'regional', 'territory' and 'local'.Taking a broader selection of EE definitions, it has been proposed that EEs exist on the scales of the 'region' (Spigel, 2017) or 'a specific regional context' (Mack and Mayer: 2120); as 'communities of agents, social structures, institutions, and cultural values' (Roundy et al., 2017: 99), or a 'group of actors in a local geographic community' (Cohen, 2006: 3); as an 'entrepreneurship hotspot' (Isenberg, 2014), or 'entrepreneurship within a given territory' (Theodoraki and Messeghem, 2017: 56).At one end of the spectrum, Andrews et al. (2022) map out a vast number of potential EEs in the US, so that cities as small as Muncie, Indiana, are represented.At the other, Isenberg (2010) discusses national scale EEs, as do others in the literature (for example, Rannikko and Autio, 2015;Stam, 2014).In between, some authors prioritise EEs as existing at the regional scale, for example, Spigel (2022: 484) definition of 'entrepreneurial ecosystems here as the regional collection of actors […]'.This has led to some authors suggesting that every region is a separate EE, for example, Leendertse et al. (2022) study of European EEs divides the continent up into adjacent EEs.Other definitions have tried to demarcate the boundary to 'a 60-mile (100-km) radius around a centre point' (StartupGenome, 2017: 24) or 'a limited region within 30 miles (or 1-h travel) range' (Cukier et al., 2016: 1).Furthermore, other authors point to the potential 'nestedness' of EEs, existing at different scales within each other (Brown and Mason, 2017;Spigel, 2022).
When the EE concept can refer to any spatial scale, it follows that whatever factors and processes make EEs beneficial are scale-independent and difficult to identify, losing analytical precision (Martin and Sunley, 2003).Yet, the term 'ecosystem' hints at a spatiality described by the interactions between its elements.These interactions differ between scales, as different factors are responsible for these interactions such as spatial proximity (Owen- Smith and Powell, 2004), technological trajectories (Dosi, 1982) or institutional settings (Lundvall et al, 1988), which amplifies the problem of insufficient precision on the components and processes highlighted above.Indeed, it is worth noting that Stam and van de Ven's (2021) ten elements constitute a mix of resource endowments and institutional arrangements, the latter of which are notably multi-scalar (Harris, 2020).The development of proper relational (Spigel, 2017) or process-oriented perspectives (Spigel and Harrison, 2017) will rely on tackling questions surrounding the multi-scalarity of EEs as these relations will not take place solely in isolated arenas.
A fourth area of uncertainty is the governance of EEs.Two main approaches to EE governance reside in the literature, a 'bottom-up' approach that embraces the ecosystem metaphor and assumes EEs are self-regulating and naturally evolving ecosystems (Isenberg, 2010(Isenberg, , 2014)), and a 'top-down' approach which assumes that they can be managed by leaders (Colombo et al, 2019).These approaches are represented in the above definitions, contrasting Mason and Brown's (2014) explicit recognition of 'govern' and Stam's (2018) inclusion of 'actors and factors coordinated in such a way', with Spigel et al's (2020) more vague notion that actors and factors 'all contribute'.Typically, the literature assumes these leaders to be entrepreneurs for their direct knowledge of the entrepreneurial process, with government policy makers playing a supportive role (Feld, 2012).However, this poses a problem for policy makers lured in by the dominant discourse purveyed by this literature that EEs are vital contributors to economic growth that should be pursued and encouraged.How should policy makers pursue EEs without interfering with the self-regulating and naturally evolving ecosystem that the bottom-up approach suggests?Are they only to listen to the thoughts of leading entrepreneurs?Or, returning to the issue of causality, what should policy makers do in lieu of strong entrepreneurial leaders to create new EEs?The policy question is pivotal to avoid EE policy mimicking what Isenberg (2014) calls the failure of "dictated industrial policy, [and] barren "cluster" strategies" that went before.
Finally, a fifth area of uncertainty is in the proposed different outputs of the ecosystem.The three definitions provided at the start of this section talk about the 'performance within a local entrepreneurial environment', mention 'productive entrepreneurship' or 'contribute to the creation of high-growth ventures'.The first definition considers entrepreneurship broadly and without restriction, while the latter two more recent definitions link EEs to the idea of 'productive entrepreneurship' (Wurth et al., 2022), which is seen as a proxy for high-growth firms or scale-ups (Stam and Bosma, 2015), that are ran by 'ambitious entrepreneurs… who attach importance to performing (more than) well with their business' (Stam, 2015(Stam, : 1760)).This approach excludes certain types of entrepreneurial activity such as selfemployment, small businesses and even failures (Stam, 2015).However, when a definition of productive entrepreneurship is supplied, it typically refers to Baumol's (1993: 30) definition of 'any entrepreneurial activity that contributes directly or indirectly to net output of the economy or to the capacity to produce additional output', or more recently 'as entrepreneurial activity that creates aggregate welfare' (Stam, 2018: 6).These definitions of productive entrepreneurship are potentially very broad in the economic activities that they subsume as most entrepreneurial activity would create aggregate welfare or (in)directly contribute to the net output of an economy, and appear incongruent with the particular focus on high-growth firms or scale-ups.Moreover, some authors have chosen to ignore the productive entrepreneurship label in favour of 'highgrowth ventures' or 'high-growth entrepreneurship' (see, for example, Abootorabi et al., 2021;Spigel, 2022), Regardless of the definition used, the intent of the productive entrepreneurship focus is to exclude certain types of entrepreneurial activity, which seems somewhat at odds with the concepts name, but more pertinently infers that the concept again means different things to different people.Indeed, as Stam and van de Ven (2021: 811) suggest regarding productive entrepreneurship: 'this claim seems too exclusive', raising the question about what entrepreneurial activities are included or excluded within entrepreneurial ecosystems.
This section has highlighted some of the key areas in the definitions of the EE concept and how they lead to ambiguity, and ultimately the label of a chaotic concept.There is uncertainty around the component parts and processes.The lack of any inherent geographical scale has created a situation where this concept can be applied to almost any place or scale, from the city, through the region, to the national level, enabling academics to create their own EEs.The actors involved in governance remain unclear, as does exactly what is meant by the term's entrepreneurship or productive entrepreneurship, and we have questioned whether it truly reflects the nature of entrepreneurial activity.The concern with this conceptual chaos is that, combined with the popularity identified in the previous section, the EE is following early usage of the cluster concept in becoming a concept with a brand identity that policymakers buy into as a policy panacea, and that academics can repurpose and reimagine as they see fit due to the loose definitions (Martin and Sunley, 2003).Importantly, there remains fundamental ontological-epistemological debates still surrounding the concept.An important aspect of defining concepts can be how they align with other concepts, providing its identity.Consequently, the following section will explore the supposed conceptual and empirical distinctions between EEs and clusters to see if the proposed distinctions are appropriate.

IV The conceptual ambiguity between entrepreneurial ecosystems and clusters
The conceptual ambiguity of the EE extends past its definition and into the supposed differences with the cluster concept that forms a significant part of its conceptual lineage.The EE literature acknowledges that there are significant similarities between the concepts, particularly the shared 'focus on the external business environment' (Stam, 2015(Stam, : 1761)).Indeed, Rocha and Audretsch (2022) argue for three common dimensions, as geographical agglomerations, that contain inter-organizational networks and socio-economic processes, which in turn lead to them both being classified as regional systems of production and exchange.Given these fundamental similarities, the EE literature has made numerous attempts to distinguish EEs from clusters conceptually, with some offering short appraisals (see, for example, Auerswald and Dani, 2017;Stam, 2015), while others debate the differences more comprehensively (see, for example, Rocha and Audretsch, 2022;Spigel and Harrison, 2017).However, we use this section to raise caution that these distinctions could be construed as rather simplistic and unreflective of both the existing literature and the empirical reality, with the distinctions between EEs and clusters remaining unclear.We focus here on two main differences shared across these appraisals, as (1) the role of key actors involved and (2) the importance of industry sectors and types of knowledge.For the EE concept to become a useful policy tool, it must be clearly distinguished from other similar policy tools, like clusters.

The role of key actors
The first theme is the difference in key actors driving these concepts, with the state accused of playing a prominent role and entrepreneurs side-lined in clusters, while entrepreneurs take the leading role and government policymakers a supportive role in EEs (Stam, 2015).Contradictory to these claims, and as mentioned in Section 2, entrepreneurs have been present in the cluster literature for a long time and are seen as 'a critical element in the formation of clusters… [who in adapting] to both constructive crises and new opportunities create the factors and conditions that facilitate their business interests and, in turn, contribute to the development of external resources' (Feldman et al., 2005: 129).They are seen as vital to the ongoing vibrancy of clusters as without new firm creation and firm spinoffs clusters would decline and eventually fail (Menzel and Fornahl, 2010).Stam (2015) argues that cluster research does not pay enough attention to the individual entrepreneur, focusing instead on the enterprise.This has some generalised truth to it; however, studies have highlighted the importance of entrepreneurs as individual agents driving the emergence (Feldman and Francis, 2006;Feldman et al., 2005;Harris, 2021b;Mason, 2008) and renewal (Hervas-Oliver and Albors-Garrigos, 2014;Sedita and Ozeki, 2022) of clusters.Thus, entrepreneurs and their activities are known to play a central role in not just the day-today activities but the growth, revitalisation and survival of clusters.
The other aspect of this claim is that government policy makers play a background role in EEs compared to a driving role in clusters: 'the importance of entrepreneurs as central players (leaders) in the creation of the system and in keeping the system healthy… decreases the role of government compared to previous policy approaches… as a "feeder" of the ecosystem than as a "leader"' (Stam, 2015(Stam, : 1761(Stam, -1762)).Firstly, while government policymakers often do play pivotal roles in clusters (see, for example, Smith et al., 2017), and the formal institutional environments that clusters are embedded within are always important (Maskell and Malmberg, 2007), there are also significant studies that focus on firm dynamics as drivers of clusters (for example, Arthur, 1994;Ter Wal and Boschma, 2011).Government policymakers are by no means even a requirement for cluster emergence and functioning; indeed, the literature significantly highlights that policymakers are often the downfall of clusters (Lovering, 1990).Ultimately, there is no reason that cluster 'leaders' should be government policymakers any more than they should be entrepreneurs.
Furthermore, recent articles in the EE literature note that 'At this point in time, research interest in ecosystems is driven by its intense popularity in policy circles rather than more fundamental research questions' (Spigel et al., 2020: 486).Indeed, one of the pivotal founders of the concept in Isenberg (2014) has previously debunked the 'myth' that entrepreneurs drive the entrepreneurship ecosystem, stating instead that 'There is no one driver of an entrepreneurship ecosystem because by definition an ecosystem is a dynamic, self-regulating network of many different types of actors'.We agree and suggest that the difference between entrepreneur or government leadership is not so much a conceptual distinction between EEs and clusters, but rather a contextual distinction based on the places that are studied.Furthermore, greater nuance is necessary when considering government intervention.In some instances, government policymakers can be quite hands on, directly creating formal policies and even forming new state departments to reach their goals.In others, government policymakers can be more hands off, liaising heavily with local entrepreneurs and allocating funding and other resources to independent or public organisations who distribute them to entrepreneurs.Thus, the forms of government intervention and delivery of resources may differ significantly.However, consideration of government policymakers is necessary.If Silicon Valley and Silicon Wadi are to be considered two of the most successful EEs, then the well-documented and vital role of government policymakers in encouraging their growth must also be considered as important lessons for EE development (Engel, 2015;Wonglimpiyarat, 2016).Indeed, the emergence story of Singapore's EE is that government policy makers saw what was happening in Silicon's Valley and Wadi, decided that Singapore's future economic development was dependent on having their own EE, and so set about creating one through generous new policies (Harris, 2021a).Thus, while EEs clearly have a stronger focus on entrepreneurship, this first distinction made to separate them from clusters due to differences in emphasis paid to entrepreneurs and government policy makers is not as significant as suggested, and the opposite can often be true.

The importance of industry (sub)sectors
The second distinction made to distinguish EEs from clusters is through the importance of industry sectors and types of knowledge.Auerswald and Dani (2017: 98) see 'clusters as bounded by an industry, whereas entrepreneurial ecosystems cut across industries'.Similarly, Rocha and Audretsch (2022) argue that the only distinction between clusters and EEs is the type of socio-economic processes, with clusters focusing on sectoral value-chain activity and EEs on entrepreneurship.Spigel and Harrison (2017: 156) have a slightly broader definition that extends to clusters having a focus on 'a particular industrial sector or between sectors that spur innovation', but maintain the industry agnosticism of EEs.
EE authors have taken a rather simplified identification of the cluster literature here.The cluster literature has been aware of the benefits of crosssector pollination and diversity dating back to Jacobsian externalities (Jacobs, 1969), and discuss the importance of (un)related variety and diversification for cluster evolution (see, for example, Trippl and Otto, 2009;Sedita and Ozeki, 2022).Furthermore, the cluster literature has talked about the spillover of entrepreneurial knowledge in depth (see, for example, Audretsch and Aldridge, 2008;Hervas-Oliver et al., 2017), and so is capable of understanding the spread of entrepreneurial knowledge across sectors.
The supposed sector agnosticism of EEs also reflects a rather simplified empirical reality.While it is true that digitalisation has diminished many sectoral distinctions and the focus of the EE concept is on entrepreneurial knowledge, high-growth start-ups cannot rely solely on entrepreneurial knowledge to get them to the exit stage, they will need at least some sector-specific knowledge to continue their growth.Indeed, looking at any major EE around the world you are likely to see subsector-specific actors and resources: the first FinTech accelerator launched in New York in 2011 (TechCrunch, 2010), then in London in 2012 (TechCrunch, 2012).StartupBoot-Camp, one of the world's leading accelerators, has specialist accelerators in the finance, fashion, cybersecurity, energy, sports, decarbonisation and sustainability sectors (StartupBootCamp, 2022).Spigel (2022) has argued that FinTech firms in the UK have crossover between the 'tech' and financial sectors, which while not present across all FinTech companies, amounts to a significant amount of activity and produces 'nested' EEs.Furthermore, the proliferation of EE rankings by organisations such as Startup Genome or Blink are increasingly diversifying from general EE rankings to sector-specific rankings.For example, Startup Genome has now produced EE rankings for the FinTech (2022a), Blue Economy (2022b), AgTech and New Food (2022c), CleanTech (2022d) and Life Sciences (2021) subsectors.These rankings show that there are EEs around the world that specialise in particular sectors, such as Zurich, Hong Kong and Singapore for FinTech.It is not, therefore, just sector-agnostic entrepreneurial knowledge that EE actors contribute, but also sector-specific.Furthermore, it is not simply about producing knowledge, but how the different EEs value sectoral knowledge, that will differ.Again, as Spigel (2022) demonstrates with different rates of cross-sector activity in different UK EEs, this will be context-dependent.Thus, while digitalisation and the rise of entrepreneurship have softened many sectoral boundaries, there remains important sector-specific knowledge that will be important for start-ups throughout the entrepreneurial process and which EEs will place different priorities on.Furthermore, this may differ on a sector-by-sector basis, with Spigel (2022) suggesting that there will be some 'very specific' industries such as Biotech or Aerospace that may have a greater demand for sector-specific knowledge.It is therefore likely that the relationships between EEs and clusters are going to be variegated, in large part along sectoral lines.
Current understandings of the EE concept, then, amount to a one-size-fits-all general theory of EEs, driven by broad guidelines, that does not properly consider context and variation along the lines of actor roles or sectoral activity.More research is needed to identify ways in which EEs may differ (see, for example, Ferreira et al., 2023) and exploring their connections with clusters.By questioning these distinctions between EEs and clusters we have shown that the EE concept has considerable conceptual ambiguity within its internal definitions but also in its demarcations with other concepts, reifying its position as a chaotic concept.Given the similarities and conceptual lineage of these concepts, a conceptual endeavour needs to occur that explores the dynamic interactions of clusters and EEs as distinct but interrelated systems (Autio et al., 2018;Martin and Sunley, 2015), particularly if EEs are to be considered to exist primarily at the regional scale and be integral to the region's stability and changing socio-economic structure (O'Connor and Audretsch, 2022).This means seeing them in relation to each other, not as isolated phenomena (see, for example, Auerswald and Dani, 2017), and with a more precise understanding of the distinctions between the concepts.This is, of course, difficult when those distinctions are fuzzy and unclear.
V Where entrepreneurial ecosystems can engage with the cluster concept: A research agenda Following the critique of clusters as conceptually chaotic (Martin and Sunley, 2003), cluster scholars sought to redefine the concept in relation to similar popular concepts of the time, such as industrial districts and regional innovation systems.Similarly, the EE concept would do well to engage with the wide-ranging economic geography literature on clusters and regional development that extends over three decades.Not only would this enhance the EE concept itself, but also develop the conceptual relationships between the EE and cluster concepts required to advance understandings of regional development.In this section we focus in on a limited selection of three areas that the two literatures could benefit from particularly closer engagement.

Evolution
One of the main criticisms of clusters at the turn of century was the tautological argument suggesting that the presence of externalities makes good clusters and that good clusters produce externalities (Martin and Sunley, 2003); or as Krugman (1991) put it, you cannot describe the emergence of agglomerations with agglomeration economies.The very same tautological critique is found in the EE literature (Stam, 2015), whereby a good EE results in many new entrepreneurs, and when there are many entrepreneurs, there must be a good EE.Similarly, there is little in the way of detailed explanations of EE emergence (see, for exception: Roundy et al., 2018).
In response to this critique, cluster theorists had an evolutionary turn towards understanding how clusters and their externalities are produced and change over time (see, for example, Martin and Sunley, 2006, 2007, 2012, 2015).While it is generally accepted that there can be a high degree of randomness or serendipity in the emergence of clusters (Martin and Sunley, 2003), scholars investigated the preconditions and processes that affect the probability of cluster emergence (see, for example, Isaksen, 2016).For example, clusters can form based on diversification from related industries (Boschma and Wenting, 2007), due to the emergence of new technologies (Martin and Sunley, 2006), or the transplantation like formation of a new firm or branch from somewhere else (MacKinnon et al., 2019).Others turned towards complexity science to explain evolutionary dynamics, focusing on the complex systemic interactions within clusters and regions and exploring their adaptability (Martin and Sunley, 2011).This literature argues that a cluster does not exist simply as an aggregate of its parts, for example, a certain amount of firms, organisation or employment.Instead, a cluster emerges when the interplay between its parts creates synergies or distinct dynamics.In this argumentation, a cluster is not there because all the elements of the cluster are there, but because the interplay between the elements is there.Others have expanded further, arguing for the importance of social legitimation and narratives in cluster emergence (Harris, 2021b).Furthermore, economic geographers interrogated path dependency and produced cluster life-cycle models to explain the stages that clusters evolve through (Bergman, 2007;Menzel and Fornahl, 2010), as well as the path dynamics of regional development more broadly.Indeed, the literature has produced many factors affecting cluster evolution, such as institutional frameworks, firm dynamics, networks, sectoral dynamics and varied actor agency (Harris, 2020).Elements of this literature have been utilised in EE discussions but could be developed further.For example, the life-cycle model has been applied but in limited ways (see, for example, Cantner et al., 2021;Mack and Mayer, 2016), focusing on largely quantitative measures of firm counts and success to describe different stages, as well as missing some of the important additions of the non-linear dynamics that Menzel and Fornahl (2010) added.
Interestingly, EEs describe an interesting situation where both the EE and the start-ups themselves have a life cycle of sorts.While the cluster literature typically sees firms as generally the same, and so focuses on the life cycle of the cluster as a whole, EEs would have to consider two life cycles: one for the entrepreneurial process from the early-stage start-up to a global firm, and one for the EE itself, including the manifold complex interactions between the two.Current discussions on the co-evolution of multiple systems or paths (Frangenheim et al., 2020;Gong and Hassink, 2019;Hassink et al., 2019) might give some indications for theorising the evolution of EEs in connection with its start-ups.Indeed, the co-evolution literature should be further addressed when thinking about how clusters and EEs co-evolve.Start-ups will benefit from knowledge sources in both EEs and clusters at various points and so how start-ups experience interactions with these two systems is important.Furthermore, understanding how these systems can interact with one another for mutual benefit is important for policy recommendations for regional development.

Agency
Related to discussions about evolution, agency has become a focal point of economic geography discussions.Early cluster theorists suggested that clusters emerged out of random chance or serendipity, with little understanding of how actors could create and change them, inhibiting policy knowledge (Arthur, 1994;Krugman, 1991).Critics of the early approaches to understanding cluster evolution that focused on firm dynamics, institutional frameworks and industrial dynamics, argued for more agency-oriented perspectives that drew on wider debates in economic geography (Giddens, 1984;Yeung, 2005) and aimed to understand specifically how actors could drive cluster evolution (Harris, 2021b;Hassink et al., 2019;Trippl et al., 2015).This literature can be applied to EEs, particularly as it gains policy-making attention, to understand how diverse actors can change EEs and drive their evolution.
Agency debates have also distinguished several types of agency with different effects, for example, the literature distinguishes between transformative agency and reproductive agency as those actions that are capable of significant changes in economic systems or smaller, incremental, moments of change (Benner, 2023;Coe and Jordhus-Lier, 2011).Transformative agency is particularly important given its ability to have systemic impacts.Grillitsch and Sotarauta's (2020) contribution identifies three types of transformative agency driving regional development: innovative entrepreneurship, institutional entrepreneurship and place-based leadership, which can be transposed directly to changes occurring within EEs.Especially important might be the distinction between firm-level agency and systemlevel agency (Benner, 2023;Isaksen et al., 2019).Furthermore, how system-level agency differs between clusters and EEs is important to understand for regional development.These approaches all offer the potential to understand the role of different actors, entrepreneurs, firms, policymakers or others, in driving changes in EE evolution within context.Agency is essential to not just understand the interactions and processes within different EEs at different stages, but to make accurate policy recommendations based on what is understood.

Multi-scalarity
After early critics of the cluster concept argued the focus on the local or regional scale offered a territorial trap, the cluster literature began to emphasise how relations and knowledge processes within and outside clusters differ (Boschma and Ter Wal, 2007).Bathelt et al. (2004), for example, distinguish between the learning processes within regional networks, the 'local buzz', and the structured knowledge flows of the 'global pipelines' in strategic partnerships such as alliances, projects or research collaborations.While 'local buzz' helps to develop shared values and interpretation schemes through diverse interactions and to organise the complex information about markets and technologies, 'global pipelines' integrate new knowledge into the cluster.
The cluster literature also emphasises the role of gatekeepers, which are actors who are located between two groups of actors and organise the exchange of knowledge between them as intermediaries (Graf, 2011).From a geographical perspective, these are groups respectively inside and outside a region.Gatekeepers are thus 'actors' with 'strong connections outside the cluster which also contribute to the diffusion and recombination of external knowledge within the local context' (Giuliani, 2011(Giuliani, : 1330)).For EE research this means that analyses which are confined to processes within defined geographical boundaries are likely to fall short in analysing the relational dynamics at different geographical scales influencing the internal working of EEs.
It can be suggested that this multi-scalar and relational perspective might be even more important for EEs than clusters, given the presence of localised communities meeting with global accelerator networks and global venture capital firms (Florida, 2016).Further research that would be helpful here is to explore transnational entrepreneurs (Brown et al., 2019;Schäfer and Henn, 2018), returnee entrepreneurs (Kenney et al., 2013) or how accelerator and coworking networks expand across EEs (Harris, 2021a;Harris and Menzel, 2023).Furthermore, there is a possibility for new kinds of extra-local connections between EEs and clusters in different places that would broaden our understanding of the various types of connections between different types of systems.

VI Conclusion
This paper has demonstrated that the EE concept is a chaotic one, meaning different things to different people, and embodying significant fuzziness in meanings.We identified five areas of the definition in which there is significant confusion between authors offering different perspectives on the same aspects of the definition: the elements of an EE, their interactions, the scales at which EEs exist, the actors involved in their governance, and the proposed outputs.Despite this, the potential of the EE concept matches its popularity, and it can be used to investigate a significant recent shift in global capitalist dynamics and regional development which economic geographers have thus far underexplored.However, as we have demonstrated in this paper, the literature could benefit from discussions with the cluster literature and economic geography more broadly so as not to reinvent the wheel.
While recognising that the concept does have a discernible lineage from the cluster and other related concepts, we believe that this ancestry has been simplified to the detriment of engagement between the concepts.Better understandings of how the EE and cluster concepts relate and co-exist would greatly develop both literatures, open up a wealth of literature from economic geography that can be used for mutual benefit, and improve knowledge about regional development.In that sense, entrepreneurial ecosystems can be seen as 'boundary objects', arrangements that 'allow different groups to work together without consensus' (Star, 2010: 601).For economic geographers, particularly, engagement with this literature offers the potential for a new research synergy and opportunity for engaged pluralism that has been the focus of much recent debate in our discipline (see, for example, Hassink et al., 2014;Martin, 2021;Rosenman et al., 2020).The EE discourse represents an inherently spatial discussion that we are currently ceding ground to business schools on, and offers a prime opportunity for economic geographers to show how an economic geography approach can add value to other disciplines (James et al., 2018).
The collaborative avenues offered in Section 5 are a limited array of options for economic geographers and EE scholars to consider.However, they offer promising routes to tackle some of the main issues facing the EE concept and relations between the two, as well as advancing broader debates about evolution, agency and multi-scalarity within economic geography.Many of the pressing conceptual issues that the EE concept faces have been debated and explored in economic geography research over many years.Perhaps most pertinently, however, there is a need for better conceptual and empirical investigation that combines EE and cluster approaches, to help delimit the differences and identify the intersections and relationships.The EE concept has great interdisciplinary potential, and we hope that this paper can help stimulate such discussions across the literatures identified in this paper.