Examining the Connections within the Startup Ecosystem: A Case Study of St. Louis

2.1 Background

In the past century, the study of entrepreneurship has evolved dramatically, in both level and complexity of analysis. Unchanging, however, is the acknowledgement that entrepreneurship is vital to a market economy, driving scholars and practitioners alike to understand successful entrepreneurship. For many years, entrepreneurship was conceptualized at the microeconomic level as relating to individuals and firms. Van de Ven (1993) summarized that the focal questions of these past studies were who the entrepreneurs were and what kind of individual traits led them to success. Janssen (2009) classified those studies as “internal factor” analyses.

The past few decades, however, have seen a shift in the literature toward a more interdependent, interactive view of entrepreneurship. From a macro perspective, this took hold with Van de Ven’s pioneering work on the “social system framework” (Van de Ven 1993), which identified actors broadly engaged in innovations and entrepreneurship at the local scale, such as the university and its scientific research, financing and insurance arrangements, and a human competence pool. This has led to the current, ubiquitous belief that “the unit of entrepreneurship is not the isolated individual but networks of actors” (Grabher and Stark 1997).

Reflecting this environmental perspective, several studies emerged in the 2000s that contributed to a wealth of knowledge on different elements that may contribute to a successful entrepreneurial system. This has come to be known as the “ecosystem perspective.” For example, Neck and his colleagues (Neck et al. 2004; Cohen 2006) cultivated their entrepreneurial framework from observations in Boulder, Colorado, and identified six elements that contribute to a system of entrepreneurship: incubators, spin-off firms, formal and informal networks, physical infrastructure, and culture. Feldman (2001) uncovered that pioneering entrepreneurs, supportive social capital, venture capital, entrepreneurial support services, and engaged research universities constituted the successful establishment of an entrepreneurial culture in the Washington D.C. area. Similarly, Mack and Mayer (2015) investigated the regional entrepreneurial ecosystem in Phoenix, Arizona, from an evolutionary perspective. Like Feldman’s approach, they identified elements based on a literature review and asked how each interviewee perceived the state of each element in the region.

Additional recent studies have pointed out the importance of spin-offs from local anchor firms and “entrepreneurial recycling,” (originally termed “flexible recycling” by Bahrami and Evans 1995, and later “entrepreneurial recycling” by Mason and Harrison 2006) in which successful entrepreneurs cash-out to start new and even more successful companies (Mayer 2011; Mason and Brown 2014; Stam 2015; Spigel 2015). While these findings could have powerful implications, the roots of those spin-off firms, or recycled entrepreneurs, could be identified only retroactively, and did not help in understanding the current conditions of the ecosystem or the methods needed to generate more spin-offs or recycles.

Thus, the major limitation of these past studies was their focus on identifying elements without an analysis of the relationships between those elements. Without understanding the way the elements of an entrepreneurial ecosystem interact, we cannot understand the ecosystem dynamics. The limitation becomes particularly apparent when the interactions between elements, such as networks and social capital, are identified as elements of the ecosystem (Feldman 2001; Neck et al. 2004). Our primary objective is to shed light on how the various elements of an ecosystem may interact, as well as to disentangle those interactions from the elements themselves.

Incidentally, the major elements identified by the past studies of entrepreneurship ecosystems resembled those of cluster theory. This began with Porter (Porter 1994, 2000), who’s aim was to holistically summarize industrial agglomerations. He listed risk capital (such as venture capitalists), specialized support services, research universities and corporate research labs, core customers, and a labor force (Porter 1998b). His stylized diagram of the wine industry in California (Porter 1998b) displayed a loose configuration of related sub-sectors, yet it fell short in explaining how and how much those actors were connected (Motoyama 2008). So far, the entrepreneurship ecosystem literature has followed the same trajectory.

Here, we hold a close view with Breznitz and Taylor (Breznitz and Taylor 2014), who criticized the past “element-focused framework.” They conducted a case study of Atlanta, Georgia, which scored high in all elements of the ecosystem, such as highly-ranked research universities, anchor companies, skilled labor, venture capital funding, and favorable policies, but which performed poorly in entrepreneurial outcome. Instead, they called for the “structure-focused approach” in which they analyzed “the structure of the relationship among agents, be they firms, individuals, associations or governments within the cluster” (2014, 376). Their case highlights the importance of the idiosyncratic value of elements in a given system as well as the importance of how organizations interact with each other and local entrepreneurs.

In our effort to investigate actors and connections between elements, we adopt the unit of analysis and methodology used by social network studies. There is a rich literature on this subject as various scholars have investigated the influence of individuals’ network patterns on entrepreneurship from a micro perspective. Notably, building on Granovetter’s (1985) work, Aldrich and his colleagues (Aldrich and Martinez 2005; Aldrich, Rosen, and Woodward 1987; Aldrich and Zimmer 1986) emphasized the critical importance of the embedded nature of entrepreneurial activity in a surrounding social context. Jack and Anderson (2002) and McKeever et al. (2015) have built on this line of research by linking social embeddedness with the type of opportunities an entrepreneur sees and thus the type of venture they start, concerning both social and economic spheres. Research has also investigated the role of social ties for outcomes of entrepreneurial ventures in greater detail, such as questioning if volume of networking activities by entrepreneurs is related to individual firm success (Witt 2004) and how an entrepreneur’s network may systematically vary according to the stage of his or her venture (Greve and Salaff 2003).

While past social network analyses have examined individual connections at the micro level, we extend our analysis at the meso level, seeking to understand how connections among individuals, between individuals and organizations, and among organizations, come together to form the ecosystem as a whole. Our premise is that our micro- and meso-level approach will allow us to identify and measure the clear connections between actors, agents, and organizations within an ecosystem, rather than examining aggregated elements, such as universities, research labs, industries, labor, and policy. We discuss the operationalization of this micro- and meso-level analysis in the next section.

2.2 Scope and Framework

First, we focus on a specific segment of the entrepreneurship ecosystem by applying the approach developed by Lyons, Lichtenstein, and their colleagues (Lichtenstein and Lyons 2006; 2010; Lichtenstein, Lyons, and Kutzhanova 2004; Lyons et al. 2012). They emphasize segmenting the pipeline of entrepreneurs and firms according to stage of development, and cultivating specific skills in entrepreneurs as a method of economic development. Moreover, they point out that such skill development can be facilitated intentionally and structurally by forming a community of entrepreneurs. For instance, this could occur at an incubator or by bringing together otherwise geographically dispersed entrepreneurs, as in Saskatchewan (Lichtenstein and Lyons 2012). Within this framework, we focus on the beginning of the pipeline, or earliest stage of a venture: the startup. In doing so, our intent is to begin our integrative analysis from the simplest angle. Further, by taking this approach, we narrow our focus to the skill attainment and learning of entrepreneurs, as opposed to capital attainment.

Second, as we seek overall to understand entrepreneurial ecosystems as networked structures, we conduct a highly exploratory analysis of elements and connections between elements within an ecosystem. Here, we integrate the analysis of micro and meso levels and propose a multi-level framework. Our first level of observation starts with the fundamental question: at minimum, what constitutes an entrepreneurial system? Our answer is entrepreneurs, which is the level that the bulk of social network analysis has focused on. Until entrepreneurs are present, incubators have no one to incubate and VC firms have no one to fund. These things may be present, but by definition there is no entrepreneurial activity without entrepreneurs. Then, those entrepreneurs use or need resources – funding, training, and other things that otherwise enable them to start their businesses and obtain the skills they need to move through the pipeline. Thus, our next level of observation is the entrepreneurship support organizations and their connections to other similar organizations, which is the level that the past entrepreneurship ecosystem literature has primarily focused on. ^[1]

Lastly, we seek to understand the connections beyond what the two bodies of literature, entrepreneurship ecosystems and social networks, have discussed independently and extend our analysis to the interaction between individual entrepreneurs and support organizations. Support may come from individuals as well as organizations, however – such as angel investors or more experienced entrepreneurs. As such, we also investigate interactions with individual supporters within the ecosystem.

Thus, we seek to discover the content and use of connections within a system and consider entrepreneurial networks not as an element in the system, but as the overarching structure of the system itself. We investigate the interactions among entrepreneurs themselves, between entrepreneurs and support organizations, among support organizations, and with other supporting individuals. Through this highly exploratory, descriptive, and bottom-up approach, we attempt to minimize the normative debate about what the system should look like, and to maximize conceptualization of the ecosystem through entrepreneurs’ perspective of the connections in the system.

3.1 Methods

After learning about Arch Grants, we chose the 2013 class of recipients as our sample of startup entrepreneurs. In selecting a sample, we sought a collection of startups that had shown some measure of success by generating revenue, that had heterogeneity across industries and backgrounds, and that we believed would give us a window into the startup community of St. Louis. We chose the 2013 cohort because they would be more likely to be working in St. Louis, under the rules of the Arch Grant, and thus more likely to be available for an in-person interview, than other cohorts. In selecting this sample, we did not assume that receiving an Arch Grant necessarily made an entrepreneur successful, but that it was a signal that a firm was able to launch its business with a modest amount of cash awarded by the semi-public sector. Furthermore, any collective efforts between the public, semi-public, and private sectors that are captured by our research can demonstrate a case with direct policy implications.

The 20 recipients were from various sectors, with products and services ranging from women’s triathlon apparel to mobile advertising solutions. For purposes of this report, we anonymized startups according their sector. We classified eight companies as “Biotech,” another eight companies as “IT/Information Services,” and the remaining four as “Other” – which were either manufacturing or education focused.

We interviewed the Arch Grant recipients in late 2013 and were able to reach 16 out of 20 of them (80 %). Three companies had multiple founders participate in the interview, leading to 21 total interviewees, of which two were female (11 %). To uncover network interactions, the firms were asked where and from whom they had received support for their startup, as well as what form that support took. Then, we expanded our interviews to key support organizations that those startup companies mentioned in their interactions. This included seven support organizations. While these entrepreneurs and support organizations are the core of our analysis, we conducted additional interviews with people who were identified by entrepreneurs, such as mentors and other entrepreneurs outside the Arch Grants recipients, and also supplemented with multiple field observations of networking and speaker events within the ecosystem.

5.1 Connections as Representative of Culture

Many past economic geography studies have used broad terms such as culture (Neck et al. 2004; Cohen 2006) or social capital (Feldman 2001) in describing an entrepreneurial ecosystem. Scholars who have studied cultures of organizations and populations have put forth definitions varying from culture as “shared systems of meaning” (Hofstede 1980), to culture as “ (a) a pattern of basic assumptions, (b) invented, discovered, or developed by a given group, (c) as it learns to cope with its problems of external adaptation and internal integration, (d) that has worked well enough to be considered valid and, therefore (e) is to be taught to new members as the (f) correct way to perceive, think, and feel in relation to those problems” (Shein 1990). While these concepts are helpful in expressing vague phenomenon, alone they lack the power to decompose the structure or process of connections between people and organizations. For example, embracing a “risk-taking culture” is indeed important, but where does perception of this culture come from? It will emerge when specific role models, processes, and networks are present. We believe our research has deepened understanding of cultural evolution in this way. That is, mentorship between experienced and nascent entrepreneurs provided role models, and the way that support organizations structured their programs, events, and resources shaped the development of connections among those entrepreneurs, forming the processes and networks that became the framework for the local entrepreneurial culture.

Recent empirical work by Huggins and Thompson has delineated indices of a regional community’s culture that impact rates of entrepreneurship, including a community’s (1) engagement with education and work, (2) social cohesion, (3) femininity and caring attitudes, (4) adherence to social rules, and (5) collective action (Huggins and Thompson 2015). This work provides a much-needed connection between ideas and opaque experiences of culture and measurable behaviors. We go even further to show a primary tool by which these indices are enacted and spread – through intermediary support organizations and the interactions they create. Our qualitative and interaction-based analysis shows how a region’s culture may influence not just rates of entrepreneurship but the nature of an entrepreneurial ecosystem that develops. Attitudes toward risk-taking circulated as entrepreneurs shared what they had been going through, leading to a broad perception of how safe it was to take risks in this environment. Further, as entrepreneurs commented on each other’s plans and received feedback on their own, they could help each other improve. This mutual process of reciprocity was perceived as a supportive culture. While it is difficult to change the general culture of a people or region from a top-down approach (Griswold 2013), we find that meso-level organizations may be able to adjust these micro-level, specific ways in which people interact, thereby changing the higher-level culture in a bottom-up fashion.

We note that these connections, used for feedback and learning between entrepreneurs, were not present in St. Louis several years ago. A director of a support organization expressed the common sentiment in this analogy:

The typical problem I saw with entrepreneurs five years ago was like this: “I do this business alone, and I don’t know other startups in town. I don’t know investors here, and there is only old money by big corporations in St. Louis, so I go to Silicon Valley to find an investor.” Then, if you talk to investors, they would say: “I don’t find any prospective startups in St. Louis, and, in fact, there may not be any startups here, so I go to Silicon Valley to find companies to invest.” So somehow, they might find each other in Silicon Valley, but not in St. Louis.

(Brasunas, interview, 10 December 2012).

These remarks present a sharp contrast to how our interviewees currently perceived the supportive and friendly environment of the area. Generally, culture is believed to change over long periods of time, but we found the specific ways people interacted changed in a matter of five years, and our findings suggest that this was due in large part to the role that the local entrepreneurial support organizations played.

5.2 Geography within St. Louis

Geographic density surfaced as an important catalyst contributing to the interactions among entrepreneurs and between entrepreneurs and support organizations. The locations of entrepreneurs and support organizations were extremely concentrated, as shown in Figure 5.

Figure 5:

Map of Arch Grants recipients and support organizations

Here, we see scattered activity except for two areas of high concentration. The first is in downtown St. Louis, more specifically at 611 Olive Street, ^[3] where eight Arch Grants recipients and three support organizations resided (from those interviewed). Close to the Mississippi River, this location was a former department store building, built in 1914. At the time of the study, it housed T-Rex, a space provider for entrepreneurship, which occupied five floors. In addition to the eight Arch Grant recipients of 2013, there were 72 other startup firms operating in the space. The second area of increased concentration is between downtown and Forest Park, more specifically at 4041 Forest Park Avenue. At the time of our study, this area was being developed into a technology park known as CORTEX. The 4041 building housed two support organizations, the Center for Emerging Technologies and BioGenerator, as well as three of our interviewed biotech firms. Arch Grants did not require, but encouraged its recipients to be located in one of these two facilities.

We cannot make a pure causal argument, but we do observe that these physical colocations substantially correlated with the interactions among firms. Arch Grants required its recipients to relocate to the St. Louis area for at least one year as a condition of the grant. Three out of twenty recipients had not completely moved to St. Louis yet, at the time of the interviews, and we identified the least interaction between these startups and other recipients or support organizations. In contrast, the most interactive companies were located either at T-Rex or on Forest Park Avenue.

Proximity alone is not sufficient, however, to create an interacting community of entrepreneurs. As Lichtenstein and Lyons (2006, 2010) rightly stressed, even entrepreneurs within the same incubator will not necessarily interact. But, catalytic activities to connect them are crucial, often led by a person who can diagnose and assess different stages of development of entrepreneurs. Within St. Louis, these two conditions were met: entrepreneurs who were connected through their Arch Grants cohort and colocation, as well as the support organizations, also located in the same facility and who provided trainings and organized catalytic activities for the entrepreneurs.

5.3 Types of Organizations and Support

The interviews revealed that different support organizations in the region provided somewhat overlapping yet still different types of support, based on startup stage and function. By analyzing the open-ended interview question “What kinds of support do you provide?”, we identified the primary types of support offered by each organization and categorized these in Table 3. This table is not intended to be comprehensive, but it still captures what each organization considers its primary support offerings, and displays a good amount of overlap among the organizations.

We classify the support offerings into two large categories. The broad support types are mentoring, finding the talent, connecting, and financial. The functional support types are more specific, such as refining the business model, practice pitching to investors and customers, due diligence, and space. Note that the orthodox kinds of support for economic development (financial and incubation space) were only two of the many functions that these support organizations provided. Moreover, there was no organization that provided money or space alone. The support organizations provided these traditional types of support in conjunction with other types of support.

Previous studies on entrepreneurship ecosystems have provided little discussion of the content of support services, but rather treated them as a black box. At best, Feldman (2001) identified entrepreneurial support services as one of the elements of the ecosystem and observed somewhat different types of support. We find a number of different types of support, separated into two groups and provided heterogeneously across organizations. Most of these support types do not have the clear-cut categories appearing in business school textbooks, such as finance, accounting, marketing, and so on. Instead, the categories of support provided by support organizations and identified by entrepreneurs were much more diverse and specific. Thus, we point out the important need for further research on what types of support startups seek, use, and find valuable, and how that compares to what is offered.

The most heterogeneous support was mentoring, which many support organizations identified as their primary service, though the form and content of this mentoring could have varied greatly. We had only limited knowledge about the exact content of mentoring. It could have been about some of the other functions listed in Table 3, such as refining the business model, finding the best-fit talent, managing cash-flow, or balancing work and life. In all cases, individuals mentored nascent entrepreneurs on a purely voluntary basis and did not have monetary incentives for consultation. Rather, the mentors had gone through similar processes in their past and they were willing to give back to the new generation. In that sense, the mentor-based support found here is contextually different from the presence of professional and support services that cluster theory (Porter 1998a) or global city literature (Sassen 2001) identified; it is much more locally based and non-commercially oriented. Accordingly, local “entrepreneurial recycling” can benefit a region not only directly, as an entrepreneur re-invests their own energies and capital to start new firms and fund others, but indirectly as well, as they serve as role models and mentors to the next generation (Dubini 1988; Malecki 1997; Mason and Harrison 2006).

Further research involving mentors directly is needed, as recent research has started to indicate the presence and importance of mentorship (Motoyama et al. 2013). As mentorship is studied, it will be important to collect data at the source (the mentors) because looking to either support organizations or mentored entrepreneurs will fail to give the whole picture. Support organizations do not necessarily perform the mentoring themselves, but introduce mentors to entrepreneurs. Further, entrepreneurs may give biased responses because they may not disclose their most problematic issues to researchers.

Another implication we can draw from our findings is that the presence of multiple organizations providing somewhat similar, overlapping support functions at the regional level helped entrepreneurs to progress through different development stages. Earlier interview quotes revealed that entrepreneurs got financial support from multiple organizations at different times, which helped them to sustain their operation. As such, even within the same function, different support organizations had different niches and training methods and catered to different stages of the entrepreneurial process. For instance, ITEN provided the Mock Angels Program, in which they trained entrepreneurs to present their business ideas to angel investors. That was contextually different from the pitch practice provided by Capital Innovators, whose primary target was venture capitalists and other institutional investors.

We observed that each organization had different assets, even within the same function of training, and intended to use these assets to provide complementary support to companies. At the same time, the emergence of several support organizations in the past few years, while providing many new opportunities in the region, has substantially changed the landscape of entrepreneurship: four out of seven interviewed support organizations initiated since 2011, in addition to the inception of Arch Grants in 2011. Some organizations were thinking that there were unnecessary redundancies across support systems at the time of the interviews. Indeed, there was some debate about merging some organizations to create efficiencies and to simplify the landscape for entrepreneurs. During our four-month interview period in 2013, two support organizations named new executive directors, with strategic intentions to reorganize their missions. This could suggest constant reorganization of support organizations within a region, at least at the nascent stages of an ecosystem. This further indicates that identifying missing types of support may not be sufficient for creating a healthy ecosystem. Injecting a new element will likely create adjustments by other elements and in the way different elements interact. Indeed, the entrepreneurship ecosystem seen in this study appeared to be a self-regulating system, where filling in missing elements seemed to be taking place through a mechanism similar to market evolution at the organizational level, rather than a top-down method imposed by the public sector.

Through interviews, we further observed that entrepreneurs tried different support groups without formally joining their programs. This shows a selective behavior by entrepreneurs, as well as compatibility concerns between a support organization and an entrepreneur. Support organizations may be tempted to recruit entrepreneurs formally to claim an accomplishment and effectiveness of their programs, but being proprietary in this sense may be harmful. The collaborative nature of the support organizations in this study, and its effect on entrepreneurs, suggests that a delicate balance is needed. In order to maintain collaboration, certain conditions should be avoided, such as organizations free riding on one another, or getting help without committing diligence. Constant communications among support organizations seemed to help this aspect. Support organizations were aware of how and how much other organizations had supported their entrepreneurs. Again, the interconnection among support organizations, the second layer of connections, was highly beneficial in this aspect.

5.4 University Support

We would like to highlight the specific roles that several universities in St. Louis played, stemming from past research on entrepreneurship ecosystems which has discussed the role of universities, particularly Feldman’s (2001) emphasis on “engaged research universities” as an element of the ecosystem. There has been much debate about the role of the university as the engine of regional economic development (Fritsch 2002; Kitagawa 2004; Lawton Smith 2007). Past studies argued that the influence of the university was larger for high-tech or knowledge-intensive industries. Generally, we found this to be true – for our case, only firms in the pharmaceutical or biotech sectors were actively involved with universities, not firms from other sectors, such as information technology or manufacturing. Moreover, past research has treated “high-tech” sectors as relatively homogeneous, usually including biotech and pharmaceutical, information technology, and other specialized manufacturing such as aerospace and precision machinery (See commonly used high-tech definitions (Saxenian 1999; Milken Institute 2011)). Expanding this debate, we find substantial heterogeneity within the high-tech sectors in this regard, suggesting that it may be beneficial for future research to delineate more specific technology industries.

Second, as discussed in the analysis section, a common pattern we found was that for startups with university connections, the form of this connection was that the founders were students or post-doctoral researchers at universities and applied the technologies they studied to commercial settings. Similar findings have been found for students at MIT (Roberts and Eesley 2011). In this sense, it was not professors or the technology transfer office that made possible the commercialization of university-based technologies. Transfer or commercialization took place through individuals, and not in a way suggested by a standard linear model, such as in the Triple Helix model (Etzkowitz 2008; Etzkowitz and Leydesdorff 2000). The creation of patents and licensing were not the observed method for the university to engage with local entrepreneurship, and we did not see scientific knowledge trickle down by itself. Compared to faculty and university staff, students and post-doctoral researchers are more mobile, and they carried and applied their scientific knowledge outside of the university setting.

The Skandalaris Center at Washington University in St. Louis was one of the most active support organizations, but was not a typical college entrepreneurship center. It was a special independent unit that directly reported to the university chancellor, as opposed to being under the business school or related to a technology transfer office. Its endowment came from an entrepreneur, who specifically instructed its executive director to be non-academic. The key function that The Skandalaris Center played was to motivate and prepare students for early-stage entrepreneurship and to connect students with local startup entrepreneurs. It did this by establishing a platform to bounce early business ideas between students, providing student resumes to startups, and allowing local entrepreneurs to serve on the judging panels for competitions. Thus, the role that this university played in the local system of entrepreneurship was substantially different than only providing a technology transfer office or courses through the business school. In this sense, we saw the university as one of the elements in this ecosystem, because it provided a specific mechanism to create flow between the student population and local startup firms. This type of university involvement is not completely unique to The Skandalaris Center, as a similar case was found at the University of Waterloo (Bramwell and Wolfe 2008). We still find room for more research on this type of university involvement in entrepreneurship, however, as the literature on the Triple Helix or academic entrepreneurship (Corbett, Siegel, and Katz 2014; Link, Siegel, and Wright 2015) has barely covered this dimension.

Lastly, we would like to note that all senior level staff at support organizations had significant experience in the private sector or as entrepreneurs, and none had backgrounds involving traditional economic development agencies. Through our interviews, we found evidence to suggest that the support organization staff could connect entrepreneurs to potential mentors because they had a well-developed network and knew what kind of experiences different potential mentors had gone through. This allowed them to diagnose what kind of specific entrepreneurial challenge each entrepreneur was facing and associate which mentor would be most helpful in that specific context. It was not a simple assignment of mentors based on generic categories, such as industrial sectors or functions of business operation. This leads us to a significant policy implication: For the most successful mentorship of entrepreneurs, it is important for local support organizations to be led by those seasoned in the local system and networks of entrepreneurship. Knowing that potential mentors exist is probably insufficient, and inserting an outside coordinator to find local mentors is unlikely to be useful.