Description
In this detailed outline pertaining to now and later mentorship, investor ties and new venture performance in entrepreneurial.
Now and Later? Mentorship, Investor Ties and New Venture Performance in
Entrepreneurial Seed-Accelerators
Jorge Mejia and Anandasivam Gopal
Robert H. Smith School of Business
University of Maryland
College Park MD
20742
(jmejia, agopal) @rhsmith.umd.edu
ABSTRACT
While business accelerators remain understudied in the academic literature, there is growing
interest in understanding how accelerators work and where they provide value to entrepreneurs. In this
paper, we focus exactly on this question – we examine how mentorship and investor ties, two key aspects
observed across accelerators in general lead to positive accelerator outcomes and through them, to long-
term firm success outcomes for the start-ups participating in accelerators. Using the full cohort (n=105) of
an international accelerator, we follow the progress of the startups during the accelerated period and
continue to follow these startups for 15 months. We find that startups that participate more in mentorship
events have higher likelihood of achieving short-term outcomes during the accelerator, such as the release
of a prototype and generating revenue for the first time. Similarly, startups that develop more investor ties
during the accelerator survive and raise capital at a higher rate. Finally, we find evidence that certain
short-term accelerator outcomes also increase the chances of survival and investment. On the basis of
these results, we provide practical implications for start-ups as well as managers of accelerator programs,
in addition to theoretical contributions to entrepreneurship research.
Keywords: Seed-Accelerators, Accelerators, Firm-Survival, Mentorship, Investor Ties, Pre-entry
Experience, Entrepreneurship
2
INTRODUCTION
Consider a team of entrepreneurs at the earliest stage of firm formation: the concept of the product or
service is available, as are some technical and marketing resources within the team. However, whether the
entrepreneurial team will successfully make the transition to the next step, i.e. formulate a business plan,
engineer a viable product, acquire the first set of customers, and ultimately raise funding, is far from
assured. Prior research shows that more than half such new ventures will fail within five years (Aldrich
and Ruef 2006), some recently pegging the failure rates at over 90% (The R.I.P. Report, 2014). One
significant institutional form has emerged in recent years to address this problem facing early-stage
entrepreneurs – seed accelerators (Cohen 2013). While accelerators remain understudied in the academic
literature, there is concurrently increasing interest in understanding how accelerators work and the
mechanisms through which they provide value to entrepreneurs. In this paper, we focus on this question –
we study how mentorship and investor ties, two vital aspects observed across accelerators in general
(Cohen 2013), lead to positive entrepreneurial outcomes during the accelerator and thereafter, to long-
term positive outcomes for participating firms.
Seed accelerators (accelerators hereafter) are short-term programs for new ventures that are
characterized by three factors: a free and public online application to competitively select a cohort of
participating startups; a standard deal terms that typically exchange a startup’s equity for early-stage seed
investment; an intense number of activities to speed the development of the startup’s product and securing
future funding (Miller and Bound 2011). These characteristics also set them apart from business
incubators and angels, two alternative forms of early support (Cohen 2012). Accelerators were first
observed in 2005 (advent of Y-Combinator) and since then, the accelerator concept has expanded to
become an important part of the entrepreneurial ecosystem (Cohen 2013). The number of accelerators in
the US has grown from 51 in 2009 to over 200 in 2014 (Lennon 2013) and have collectively raised over
$5B in investment while creating over 16,000 jobs since (Seed-DB 2014).
Early case-based work suggests that accelerators provide value to entrepreneurs through three
specific mechanisms (Miller and Bound 2011; Dempwolf et al. 2014). First, the accelerator incentivizes
the relocation of the firm from its native ecology to that sponsored or hosted by the accelerator. While
3
firms operating in their own environments may be affected by local resources, and competition (Chandler
and Hanks 1994, Carroll and Khessina 2005), accelerators impose their own admittedly artificial and
accelerated ecology on entrepreneurs as a condition for support. The artificially created ecology
commonly imposed on all selected new ventures further offers the other two distinguishing mechanisms –
mentorship and access to network ties (Cohen 2013).
Mentorship, studied within organizational settings, has been shown to influence outcomes such as
turnover and organizational commitment (Payne and Huffman 2005, Carraher et al. 2008). Within
entrepreneurship, mentorship has been traditionally associated with business incubators (Amezcua et al.
2013) and angels (Mitteness et al. 2012). Early stage entrepreneurs may find mentors who provide them
with tacit support, knowledge and expert advice in their own environments. However, these mentoring
networks are heterogeneous and depend on how munificent local environments are (Castrogiovanni
1991). By comparison, the accelerator institutionalizes the accelerated provision of mentoring to all
teams, based on a standardized process model, thereby justifying the cost of participating in the
accelerator or the equity stake that the accelerator requests from the new venture. Additionally,
participating teams can mentor each other through co-location, which is a central process by which
accelerators add value to entrepreneurs during the accelerator program (Cohen 2013).
The third mechanism that accelerators provide is through the expedited creation of network ties
that provide entrepreneurs with hard-to-find resources, especially critical ones like capital, access to
markets, and human resources (Stuart and Sorenson 2005). The accelerator solves this problem by
providing, within its ecosystem, managed and expedited access to investors, angels and other accelerators
(Cohen and Hochberg 2014). Since both investors and entrepreneurs are located within the accelerator
ecosystem, the first-order information asymmetry problem is partially resolved (Shane and Cable 2002),
creating a positive environment for entrepreneurs to gain accelerated access to investors. A critical part of
accelerators, as opposed to incubators, is that they are of short duration. Therefore, participating firms
have strong incentives in creating valuable investor ties since these are beneficial even after the
accelerator ends and firms “graduate” from accelerators.
4
In as much as these three mechanisms have been cited as critical within accelerators, there is no
systematic evidence to show that new ventures that do invest in these experience positive short-term
accelerator-level outcomes. Cohen and Hochberg state as much: “While proliferation of such innovation
accelerators is evident, the efficacy of these programs is far from clear”, p2 (2004). Beyond short-term
outcomes, it is worth asking – do these programs ensure longer-term survival and success of the
“graduating” new venture? Prior work has argued that mentoring and access to investor ties increase the
odds of new venture success (Eesley and Wang 2014; Shane and Cable 2002) but this work is outside of
the accelerator context. Within accelerators, it is necessary from a policy and business perspective to
understand what relationship exists between these mechanisms and accelerator-specific as well as longer-
term outcomes. This forms the primary research question we address in this paper – what is the effect of
mentoring and investor tie formation within the accelerator on entrepreneurial outcomes of participant
firms in the short term (within the accelerator) and the long-term (after the accelerator ends)? Beyond
the role of mentorship and investor ties, we also ask: how do accelerator outcomes influence longer-term
outcomes for participant firms? If accelerators do indeed provide guidance in incentivizing firms towards
accelerator-level outcomes, a positive correlation between achieving short-term outcomes inside the
accelerator and longer-term entrepreneurial outcomes, such as firm survival, should be evident. This
forms our second broad research question.
We test our hypotheses using data collected in a unique manner within an accelerator. Consistent
with the notion that studying entrepreneurship requires embedding the researcher within the institutional
environment, we embedded ourselves in an accelerator as founders of a startup. From the unique
perspective gained from our role as participants in the accelerator for over 6 months, we collected
entrepreneur-specific data using surveys and social media. In total, we collected data on one entire cohort
of 105 teams hosted by the accelerator in the spring of 2013. In addition to the data collected from the
accelerator, we also tracked each startup in its post-accelerator phase for a period of 15 months to identify
longer-term outcomes. Using these data, we estimate econometric models addressing the impact of
mentorship and investor ties on accelerator-level and longer-term outcomes.
5
Since the accelerator we study utilizes the lean startup methodology model (Ries 2011), we
identify three accelerator outcomes associated with the lean startup approach – the development of a
minimum viable product (MVP), an official startup launch and a first sale. Our results clearly indicate a
positive relationship between accelerator outcomes and longer-term survival of the firm; those startups
that launch officially and experience a first sale during the accelerator are significantly more likely to
survive and raise funds at the 15-month point. In addition, we observe that firms that experience
significant investor tie formation attributable to the accelerator observe higher long-term entrepreneurial
outcomes, such as a higher likelihood of firm survival and investment at the 15-month period. Our results
also show that mentoring has a significant positive effect on short-term outcomes during the accelerator,
such as the development of a MVP, an official startup launch and a first sale by the new venture. Finally,
we observe that firms with previous entrepreneurial experience gain less from mentoring and the
formation of investor ties than inexperienced firms, indicating a significant moderating effect of prior
entrepreneurial experience.
Our work here makes several contributions to the entrepreneurship literature. First, we provide a
validation of the accelerator model by showing the relationship between accelerator-level outcomes and
those observed in the entrepreneur’s unconstrained ecology in the form of firm survival (Lall et al. 2013).
While we are careful to not make causal claims, we believe we can minimize the impact of the most
common criticism of such associational studies – that of unobservable startup quality. All startups in our
sample were chosen competitively, thereby ensuring that all firms in our sample have a level of baseline
quality. Therefore, our analysis only identifies variations in long-term outcomes associated with
variations within the accelerator-level factors such as outcomes and investor tie formation, accounting for
cohort, team-level variables and time, thus establishing a level of control typically not observed in such
studies. Perfect identification would require randomization, which is rarely ever possible in
entrepreneurship studies (Ruef et al. 2003); our approach provides cleaner identification in comparison.
Second, we provide evidence of the relationship between mentoring and short-term accelerator
outcomes. Studying the effect of mentoring on outcomes is hard when entrepreneurs are embedded in
their ecologies and receive more than simply mentoring (Rodan and Galunic 2004). The accelerator, via
6
contrast, works by relocating most entrepreneurs out of their natural ecology into a “new” environment,
where by definition mentoring ties, if any, are freshly established. The accelerator we study is particularly
suitable since it is not located in a startup hub such as Northern California or Boston. Therefore, by
necessity, most mentoring ties are new, with limited spillover from the entrepreneur’s original ties. Hence
it is more likely that we identify the true effects of mentoring on accelerator outcomes in our work.
Finally, we contribute to the small literature on accelerators by opening up the ‘black box’ of
processes within accelerators such as mentoring and investor tie formation. Accelerators have caught the
fancy of policy-makers and businesses alike in recent years for the implicit association between
accelerators and general success in entrepreneurship. Consider, for instance, the Startup America
initiative launched by President Obama in 2011 in partnership with TechStars. Such initiatives are well-
meaning but given the paucity of systematic research into the accelerator model (Cohen and Hochberg
2014), policymakers have few tools available to judge when certain activities lead to successful
entrepreneurial outcomes within the accelerator or in the long-term. Our work here represents one of the
first systematic analyses, to our knowledge, of outcomes associated with accelerators, linked to processes
established within the accelerator. As a first step, we review this literature next.
RESEARCH CONTEXT – STARTUP CHILE
Research Context
The setting for the analysis we provide in this paper is Startup Chile, a government-sponsored accelerator
located in Santiago, Chile. The program was created by the Chilean ministry of economy with the goal of
transforming Chile into an innovation and entrepreneurial hub for Latin America. The project started as a
pilot in 2010 with 22 startups from 14 countries providing $40,000 USD of equity-free seed capital to
develop a startup for six months. After the success of the pilot, Startup Chile (SUP) expanded to two
rounds per year in 2011, each round lasting 6 months (StartupChile 2014). As of 2013, the program offers
approximately $45,000 USD of equity-free seed capital, a one-year work visa, limited help getting into
Chile and finding accommodations, and access to the network of investors, mentors, and entrepreneurs
SUP has developed within and outside Chile for participating entrepreneurs. In exchange for these
resources, teams must remain in Chile while working on their startups for 6 months. Moreover, all startup
7
teams are required to engage in social impact activities in Santiago and in Chile, specified as part of the
program and applicable to all firms, whether Chilean or not.
Since 2011, SUP has received over 10,000 applications from 112 countries and has attracted
considerable media and entrepreneurial attention even outside the Chilean context. The application
process for the accelerator consists of a three-page online application with questions about the team, the
project, and status of the startup. According to the administrators at SUP, entrepreneurial firms are
accepted into a cohort based on two main criteria. First, an evaluation of the quality of team amounts to
50% of the application, as stated by SUP’s director:
In the world, there is a 85% failure rate in these types of projects [early-stage ventures], so we believe
the main criteria is to select the best possible teams, who in the worst of cases, will ultimately learn
from their failure (EMOL, 2014).
The other dominant factor driving acceptance of the startup relies on evaluation of the quality of team’s
business model, the project’s likelihood of success, and the size of the market (Startup Chile 2014). Once
teams arrive to Chile, SUP provides access to a working space for all the startups in downtown Santiago
where desks, meeting rooms, and sofas are provided. The program then gets started with all selected
startups doing 5-minute business ‘pitches’ to all the other startups. In terms of working locations, the
majority of the teams choose to work in the Startup Chile office space, but this is not mandatory. Many
Chilean teams, for instance, do not work at the Startup Chile offices.
As part of the terms and conditions of Startup Chile, every startup must fulfill a minimum quota
of social impact activities in Chile (Startup Chile 2014). Finally, the program ends with each cohort of
startups presenting their work during a high-profile Demo Day where the top 15 startups pitch their
businesses to invited star investors in one of the top venues of the city. All startups compete with each
other in several rounds for a place to pitch on Demo Day. In terms of outcomes from the program, Startup
Chile keeps detailed data of the social impact activities performed by the entrepreneurs. There is less clear
data available, however, on the business outcomes of the startups that have participated in the accelerator.
More importantly, there is even less evidence linking the resources provided by the accelerator to the
outcomes experienced by participating startups. First, we define the accelerator outcomes that are of
8
relevance to SUP, since different accelerators tend to emphasize different approaches to early-stage
entrepreneurship (Cohen 2013).
Defining Accelerator Outcomes
As startups advance through Startup Chile, there are two main checkpoints on the progress of
each startup. The first checkpoint is a meeting with an assigned financial advisor to agree on a set of
objectives for the startup while in the accelerator. These goals are heavily influenced by the Lean Startup
methodology introduced by Eric Ries (2011). The Lean methodology is pervasive in SUP and other
accelerators (Seed Ranking 2013) and is based on testing specific hypotheses, gathering early and
frequent customer feedback, showing early prototypes to prospects, and measuring success quickly
through an iterative process. Figure 1 shows the feedback loop, a central component of the ideology.
Since its introduction, both the popular (Lohr 2011; Stengel 2015; Tam 2010) and academic press
(Winkler 2014) have been interested in the Lean startup movement and its influence on entrepreneurs.
Central to the lean methodology are three key principles: First, replace months of planning and
research with testing specific hypotheses. Second, startups should engage in customer development to test
their hypotheses, i.e. direct contact with potential customers to gather feedback on the business model,
product features, and distribution channels. Third, the methodology recommends the use of agile
development (Beck 2001) to create minimum viable products (MVP), which represents the minimum
functionality or set of features within the product, allowing the firm to test the product in the market and
gather customer feedback, consistent with the second principle (Eisenmann et al. 2012).
Building on these entrepreneurial outcomes set by the startups in the accelerator, we introduce the
notion of accelerator outcomes. Differentiating accelerator outcomes from broader entrepreneurial
outcomes is necessary because these outcomes are emphasized and incentivized within the accelerator.
Additionally, typical measures of entrepreneurial outcomes, such as firm survival and funding, are
assured while in the accelerator. We define accelerator outcomes as the measurable lean startup
methodology business outcomes that may occur while participating in an accelerator. We introduce three
accelerator outcomes in this study: the completion of an MVP, the official product launch of a startup,
and a first sale, representing a significant increase in the viability of the firm. In addition to these
9
outcomes, we also identify longer-term entrepreneurial outcomes that are of interest to early-stage
startups; specifically, these are the survival of the firm 15 months after the accelerator program and the
amount of external funding received in those 15 months. As significant antecedent factors for these
outcomes, we discuss the role of mentorship and investor networking ties in the next section.
RESEARCH HYPOTHESES
Within Startup Chile, the creation of mentorship and investor ties arises as a result of multiple
interactions. Regarding mentorship, as a first step, participating founders mentor each other. There is a
high degree of collaboration among startup founders, and they give each other technical, programming,
and business feedback. Second, SUP encourages joining a ‘startup’ tribe, which are specialized mentoring
groups that meet every week to improve different aspects of the startup. Third, SUP invites entrepreneurs,
founders, investors, and VCs every week as speakers, who act as mentors and are available to startups as
resources. Beyond mentoring, other events are focused on raising capital and connecting founders with
investors. Here, the accelerator acts as a matchmaker between investors and startups, and it is common to
meet visiting investors every week, especially investors visiting from the U.S., Europe, and South
America. Finally, there are numerous social events, such as investor-sponsored parties. While
participating in these activities is highly encouraged, these are ultimately optional, allowing each startup
considerable agency. Hence, the amount of mentorship received and the level of access to the investor
network within SUP vary from startup to startup. We propose hypotheses for these relationships next.
Mentorship: Access to Knowledge and Information
A fundamental challenge for entrepreneurs is to accurately identify and shape the opportunity to be
pursued. Access to information about the value of available opportunities is a central part of this process.
Prior work in entrepreneurship suggests that social networks provide this valuable and privileged
information (Stuart and Sorenson, 2005). Our fieldwork here corroborates this argument and also
indicates why mentorship is valuable to entrepreneurs within the accelerator. To apply to an accelerator,
entrepreneurs need to provide an existing idea, which indicates that the entrepreneur has already
identified an opportunity. However, identification of the opportunity is not a single-time event (Dubini
and Aldrich 1991). Instead, the process of identifying the opportunity at the accelerator is a continuous
10
cycle of stating, testing, and refining the target opportunity. In accelerators, mentors play a key role in the
process of identifying the opportunity by providing feedback on the existing idea and by providing access
to private information to further refine the idea or ‘pivot’ into a similar but different business
model. Many of the teams we surveyed provided qualitative evidence of the role of mentorship in
identifying or pivoting to a more specific opportunity. One team remarked:
“Sure, we came in part for the funds provided, but we also came to SUP because we needed help
deciding where to focus. We were sure we had found the right opportunity in cloud security when we
applied, but by the time we started meeting our mentors, it was obvious our idea was too general. The
mentors helped us pivot into the more specific market of containerizing Dockers. Suddenly, instead of
sounding like every cloud security startup out there, we found a specific need that we could solve.” -
Tutum Cloud (A cloud services startup).
Beyond opportunity identification, a major challenge for entrepreneurs is the mobilization of resources to
provide a solution for the opportunity. In pursuing the opportunity, the literature argues that entrepreneurs
need tacit knowledge required to create a successful venture (Alvarez and Barney 2004). For example,
entrepreneurs need to be able to concisely deliver their idea and execution, commonly refereed to as a
‘business pitch’. Therefore, a second mechanism by which social networks affect the entrepreneurial
process is by providing more channels through which tacit information flows (Stuart and Sorenson 2005),
which in turn have been shown to positively affect entrepreneurial outcomes (Liles 1974; Klepper and
Sleeper 2005).
In our context, developing tacit knowledge in entrepreneurs is an area in which SUP mentors also
play a key role. Through continuous mentorship opportunities, entrepreneurs are coached to develop tacit
knowledge in the form of business pitching, marketing and branding strategies, product development, and
recruiting and managing resources. Entrepreneurs from a health IT (HIT) startup remarked as much:
“We got to Startup Chile with a great MVP, but we had no idea how to sell this product to physicians
or patients. After we were assigned a mentor with experience in health IT, we started understanding
how we should be approaching physicians, who were the key for the adoption of our product. For
example, we started focusing on security and addressing privacy concerns for patient data. Before we
joined SUP, we never thought security was a concern for our users.” -Medko (A health IT startup).
In summary, mentors in an accelerator can affect entrepreneurial outcomes by increasing the amount of
private information to identify the right opportunities and the tacit information available to organize and
manage resources in a new venture. In our context, where startups target the completion of certain
11
discrete accelerator outcomes and mentors are likely to emphasize these outcomes as well, mentorship
activities are likely to push the new venture towards these outcomes. Therefore, we propose:
Hypothesis 1a. Startups that engage in mentorship activities during the accelerator are more likely to
release an MVP during the accelerator.
Hypothesis 1b. Startups that engage in mentorship activities during the accelerator are more likely to
have an official product launch during the accelerator.
Hypothesis 1c. Startups that engage in mentorship activities during the accelerator are more likely to
generate the first sale (i.e. new revenues) during the accelerator.
Investor Ties: Access to Financial and Labor Capital
In the process of mobilizing resources to pursue entrepreneurial opportunities, networks are highly
influential for two reasons. First, investors use their network to become aware of investment opportunities
(Stuart and Sorenson 2005). Additionally, information asymmetry is a significant barrier to the
investment decision. Therefore, social networks offer information about the entrepreneur’s work ethic and
integrity, thereby helping access financial capital by helping reduce information asymmetry (Hallen
2008). Gulati (1995) further argues that embedding a transaction in an ongoing social relationship
motivates both parties to keep the relationship fair and generates a sense of obligation between the parties.
In general, there is consensus in entrepreneurship that social networks are important and affect
entrepreneurial funding and survival along different mechanisms. However, our question here is whether
such ties, instituted through the accelerator program, is successful in enhancing accelerator outcomes
(short-term) or do they manifest in the longer term?
Shane and Cable (2002), in a seminal analysis of direct and indirect ties, argue and show that
while direct ties increase trust and the quality of information that reaches potential funders, indirect ties
are useful in enhancing awareness among investors. However, their analysis stems from data from
investors rather than entrepreneurs and does not control for the entrepreneur’s prior experience. In
contrast, in an analysis of the Chinese VC market, Wang (2008) finds that while social ties are helpful in
allowing an entrepreneur to enter the focal VC’s consideration set, within the consideration set
entrepreneurs do not benefit directly from their social connections. This is consistent with the notion that
awareness of an entrepreneur may be distinguished from truly relevant information that leads to funding.
While accelerators are very useful in providing the first mechanism through their relatively shorter
12
program duration, they may not provide enough trusted information to directly influence accelerator-level
outcomes. However, firms that form such investor ties through the accelerator are likely to gain from
them in the longer run, i.e. after the accelerator program. In sum, the facilitation of investor ties helps
entrepreneurs find funding opportunities and increase the odds of survival in the long run but not so in the
short run of the accelerator’s program, which typically runs from 3 to 6 months (Cohen 2013).
Our fieldwork in Startup Chile further lends credence to the notion that social ties help
entrepreneurs raise capital by exposing them to investors and starting the trust creation process with
investors. Moreover, as the accelerator develops, many such entrepreneur-investor relationships
strengthen and the perceived risk of investing in one of these startups decreases for investors. One startup
stated:
“Since day one, the director [of Startup Chile] said that from now on we should always be on pitch
mode. I didn’t quite understand what he meant at the beginning, but after the first weeks there, I found
myself meeting a lot people and investors who were always asking what our idea was and who we were.
This was a radical change for our team as we never talked about our idea or product with strangers. In
fact, we tried to keep it as secret as possible. While the value of many of the social events sponsored by
the accelerator was in question at the beginning, after the first two months in the accelerator, we had
met more investors in these events than the last two years combined” – Arrively (A travel startup).
A second entrepreneur explained further:
“Once we developed relationships with a few investors in [CITY A], the biggest barrier to raise capital
was trust. How could they make sure we were not going to leave after the program was done? How
could we make sure these investors really had the funds and expertise to guide us? The sub-director of
the accelerator was essential in this process. After observing our work ethic for three months and how
we improved our MVP, [SUB-DIRECTOR] was able to speak about the integrity of our team to local
investors, which is what filled this trust gap.” Simple Crew (A mobile management startup).
We thus hypothesize that the investor networks provided in the accelerator increase the amount of
funding opportunities for the startup and thus its longer-term survival and funding. We focus on survival
and funding as two entrepreneurial outcomes of interest in the literature (Wicker and King 1989; Bates
and Servon 2000; Van Praag 2003; Bosma et al. 2004; Klepper, 2001; Shane and Stuart 2002). More
formally, we propose:
Hypothesis 2a. Startups that engage in investor-ties building activities during the accelerator are more
likely to survive in the longer-term.
Hypothesis 2b. Startups that engage in investor-ties building activities during the accelerator are more
likely to raise other early-stage investment in the longer-term.
Finally, one unanswered question with significant policy implications remains in this context: what is the
relationship between accelerator outcomes and long-term firm success? To our knowledge there is no
13
empirical evidence that these discrete accelerator outcomes (MVP, official launch, and first sale) are
indeed predictive of longer-term firm survival or funding. If accelerators truly help new firms emerge
from the “liability of newness” (Stinchcombe 1965), we should observe a significant positive association
between those startups that experience positive accelerator outcomes and those that survive and acquire
funding after the program has finished, indirectly providing evidence of the ‘accelerator effect’ (Miller
and Bound 2011). We therefore hypothesize:
Hypothesis 3a. Startups that accomplish accelerator outcomes (MVP, official launch, and first sale)
during the accelerator are more likely to survive in the longer-term.
Hypothesis 3b. Startups that accomplish accelerator outcomes (MVP, official launch, and first sale)
during the accelerator are more likely to raise early-stage funding in the longer-term.
EMPIRICAL STRATEGY
Collecting detailed data from within accelerators represents a challenge. Most accelerator cohorts are
relatively small (approximately 5-10 teams), which makes quantitative analysis infeasible. Furthermore, it
is generally hard to collect detailed data from teams within the accelerator and to observe the dynamics
that exist within from the outside, leading to the paucity of research on accelerators (Cohen 2013). In this
paper, we adopt a unique albeit different approach to collecting data. We embed ourselves in SUP as
founders of a startup and collect data through surveys and observation through the duration of the
accelerator. From the rare perspective gained from our role as entrepreneurs in the accelerator over 6
months, we collected entrepreneur-specific data tracking the progress of startups during the program. One
of the benefits of SUP was also in providing larger samples than usual – each cohort at SUP includes over
a 100 firms, allowing the possibility of quantitative analysis. Our tenure in the accelerator also allowed us
to interact with the entire ecosystem of the accelerator (i.e. entrepreneurs, staff, mentors, and investors)
and gain access to not only direct information from the teams but also the information provided by the
teams on social media sites (such as the firms’ Facebook, Twitter, and LinkedIn pages) as well as their
fundraising goals and activity data on entrepreneurial platforms, such as angel.co.
Sample and Data Collection
The Spring 2013 cohort of Startup Chile served as the research setting for this study. To test our research
models, we collected survey data from these startups. We developed the survey after an extensive review
14
of the literature and interviews with entrepreneurs, the staff members at SUP, investors, and
administrators. As far as possible, we used questionnaire items that have been widely tested in the
literature (described later in this section). Before the eventual use of the survey, we pilot tested the survey
with 20 entrepreneurs from the previous cohort of Startup Chile, the fall cohort of 2012. Based on the
feedback from the pilot and our interviews, we iteratively refined the survey instrument. These steps,
combined with the use of pre-existing scales, helped ensure face and content validity of our survey
instrument.
The relevant data used in analysis here was based on information collected at the end of the
venture’s stay at SUP, thereby allowing us to collect data on accelerator outcomes as well as perceptions
of how much the team had actually invested in mentorship and investor ties, as opposed to intention. Data
was collected in a single continuous wave as startups conducted the exit interview with SUP in the last
three weeks of the program at the co-working space and the administrative offices in Santiago, Chile.
While we initially attempted to collect survey data from all official members of the venture team (modal 3
members), or those engaged in a contractual relationship with SUP, this proved infeasible primarily
because by the end of the accelerator, some members of the team had relocated back to their home
country. We obtained a total of 175 questionnaires from the 105 firms representing the Spring 2013
cohort. Six surveys were discarded as they were from entrepreneurs from previous cohorts while ten more
were discarded because non-founder members, such as interns, helped complete them. In nine cases, we
had two founders from the same startup complete the survey while twelve firms had three founders from
the same startup complete the survey. For those firms with multiple respondents, we examined the
correlations on the key perceptual questions across multiple respondents and found correlation
coefficients of over 0.85, suggesting high levels of uniformity in responses. Therefore, for subsequent
analyses, the responses from individual respondents on the same startup were averaged out to form the
research variables. The dataset includes teams from 13 different nations, mean age of 28 years, from four
races, and with varying levels of education (spanning high school to doctoral degrees). A common
criticism of entrepreneurship research is the homogeneity of the respondent pool (i.e. white males from a
15
single state in the US); in contrast, here we capture a reasonably wide cross-section of racial and cultural
backgrounds as part of the accelerator. We describe the specific variable operationalization next.
Variable Descriptions
We used prior items from the literature on mentorship and network ties whenever possible. As
recommended in the mentorship literature, rather than employing a binary measure, mentorship was
measured using the multi-dimensional mentorship functions questionnaire (MFQ-9) developed by
Scandura and Ragins (1993). The MFQ-9 is a shorter version of the initial 20-item MFQ (Scandura,
1992). The MFQ measures three mentoring functions: career support, psychosocial support, and role
modeling with three questionnaire items each, for a total of nine items. The instrument was measured in a
5-point Likert scale with responses ranging from 1 (strongly disagree) to 5 (strongly agrees). The
questionnaire items used to measure mentorship, and all other such variables, are provided in Table 1.
Similarly, to measure the development of network ties for the entrepreneurs during the
accelerator, we adapted the measures used by Shane and Cable (2002), which measure direct and indirect
ties between investors and entrepreneurs, to the context of entrepreneurs in an accelerator. The direct and
indirect ties were measured with six items in a 5-point Likert scale with responses ranging from 1
(strongly disagree) to 5 (strongly agrees). The items captured the extent to which ventures were able to
form professional, informal or friendship-based relationships with investors through SUP.
In addition to these two key independent variables, we also control for several contextual and
demographic variables that have been found to be influential in increasing firm survival and rates of
funding; we also use these variables as control variables in our models of accelerator outcomes. Pre-
entry experience is measured using the scale from Dencker et al. (2009), which measures the
entrepreneurial team’s pre-entry knowledge of the business activity of the startup. Additionally, and
related to pre-entry experience, we adapt a measure of organizational capital using Hsu (2007) to measure
the number of startups previously founded by the founding team, the age (in months) of the startup, and
the average age (in years) of the founding team. Also following Hsu (2007), we collect the level of
education of the founding team (i.e. the number of members of the founding team possessing MBA and
PhD diplomas). We also take into account team composition variables, and we adapt the measures from
16
Campion et al. (1993) to control for the degree of heterogeneity, flexibility, preference for group work,
and the number of founding members of the startup. Finally, we also asked entrepreneurs to provide what
percentage of the seed funds provided by SUP were spent on operations (i.e. rent, web storage, cloud
services) and human resources (i.e. interns, salaries) – these represent two official accounting categories
that entrepreneurs are requested to report to SUP. Further details on operationalization are shown in Table
1 while summary statistics and correlations are shown in Table 2 and 3 respectively.
As a first step, we tested for the psychometric properties of the perceptual constructs to confirm
that they were of good quality. With respect to reliability, for multi-item reflective constructs, the
composite reliability was calculated and varied from 0.72 to 0.84 across constructs, thereby establishing
their reliability. All Cronbach’s alpha values obtained for the constructs easily exceeded the 0.70
threshold established in the literature. To test divergent validity, we verified that the average variance
extracted (i.e. the average variance shared between a construct and its measures) was greater than the
variance shared with other constructs in the model, i.e. other perceptual constructs here (Fornell and
Bookstein 1981). In addition, we performed exploratory factor analysis using varimax rotation and a
threshold eigenvalue of 1.0 to ensure convergent validity; all hypothesized items loaded appropriately
with high factor loadings on the underlying reflective construct, as shown in the first column of Table 4.
Finally, we ensured the factor structure of the reflective constructs through confirmatory factor analysis
using LISREL. All the hypothesized paths from the indicator variables to the hypothesized latent variable
were significant (p90%). We find strong agreement (over 95%) between observed survival on social
media and the email form. Therefore, we use survival noted from activity on social media as our measure,
with value 1 if the firm survives and 0 otherwise. In our cohort, 63%, i.e. 66 startups, still survive at the
end of the 15-month period.
Beyond survival, we also measure early stage capital investments received by the startup using
Angel.co, a platform designed to showcase and raise capital for startups. The use of Angel.co is highly
encouraged in SUP and the entire cohort actively used it throughout the accelerated period and during the
15-month period where we followed the startup. As with survival, we also confirm new investments using
the same short email format, where fellow entrepreneurs provide information about their current fund-
raising status
1
. Again, we find a strong correlation (approximately 0.9) between the reported measure of
1
We note here that such email and social media interactions are common amongst entrepreneurial teams
from the same cohort and indeed, represents one of the advantages of the data collection approach we adopt here. As
members of the accelerator’s programs, we are able to use these networks that are usually closed off to outsiders.
18
total investment in Angel.co and the email form (to the nearest 1,000). In the interest of consistency, we
use the information from Angel.co as our measure of capital investment. Approximately, 75% of our
cohort (78 startups) received early stage investment. The firms that do not raise any funds are excluded
from the analyses in which investment is the dependent variable. The average amount raised by these
startups in the 15-month period was approximately $225,000.
MODELS AND RESULTS
Using the data collected from SUP, we estimate econometric models addressing the impact of mentorship
and investor ties on accelerator-level and longer-term outcomes. We specifically estimate three models to
test our hypotheses. Model 1 tests the effects of mentorship and investor ties on short-term accelerator
outcomes, while Model 2 focuses on longer-term outcomes. Finally, Model 3 examines how accelerator
outcomes influence longer-term outcomes, to indirectly test for an ‘accelerator’ effect.
Model 1: Accelerator Outcomes
Our goal here is to estimate the effect of mentorship and investor ties on accelerator outcomes - MVP,
Launch and First Sale (FS). Since these outcomes are discrete, we fit a logistic regression model for each
accelerator outcome ? ? {???, ??????, ??}:
logit??
?
? = ?? +?? +?,
where ?
?
equals 1 when the outcome ? is accomplished during the accelerator period and equals
0 when the outcome is not accomplished during the accelerator period. Startups that accomplished
outcome ? before the accelerator period (observed occasionally) are therefore excluded from the models.
Specifically, we exclude 11 firms who have released an MVP before the accelerator, 9 firms who have
launched products officially before the accelerator, and 5 firms who have generated revenue before the
accelerator. Additionally, the vector X contains the variables corresponding to pre-entry experience,
education, team, resource allocation, mentorship, and investor ties as described above and Z includes
indicators of whether previous accelerator outcomes were accomplished before or during the accelerator
period. Logically, MVP occurs before the startup’s official launch, which should result in a first sale.
Table 6 displays the coefficient estimates and standard errors for each explanatory variable as
well as model fit indicators corresponding to each accelerator outcome. The model fit indices are suitably
19
high. More importantly, we find that Hypotheses 1a, 1b, and 1c are strongly supported; mentorship is
significantly associated with each accelerator outcome (p
In this detailed outline pertaining to now and later mentorship, investor ties and new venture performance in entrepreneurial.
Now and Later? Mentorship, Investor Ties and New Venture Performance in
Entrepreneurial Seed-Accelerators
Jorge Mejia and Anandasivam Gopal
Robert H. Smith School of Business
University of Maryland
College Park MD
20742
(jmejia, agopal) @rhsmith.umd.edu
ABSTRACT
While business accelerators remain understudied in the academic literature, there is growing
interest in understanding how accelerators work and where they provide value to entrepreneurs. In this
paper, we focus exactly on this question – we examine how mentorship and investor ties, two key aspects
observed across accelerators in general lead to positive accelerator outcomes and through them, to long-
term firm success outcomes for the start-ups participating in accelerators. Using the full cohort (n=105) of
an international accelerator, we follow the progress of the startups during the accelerated period and
continue to follow these startups for 15 months. We find that startups that participate more in mentorship
events have higher likelihood of achieving short-term outcomes during the accelerator, such as the release
of a prototype and generating revenue for the first time. Similarly, startups that develop more investor ties
during the accelerator survive and raise capital at a higher rate. Finally, we find evidence that certain
short-term accelerator outcomes also increase the chances of survival and investment. On the basis of
these results, we provide practical implications for start-ups as well as managers of accelerator programs,
in addition to theoretical contributions to entrepreneurship research.
Keywords: Seed-Accelerators, Accelerators, Firm-Survival, Mentorship, Investor Ties, Pre-entry
Experience, Entrepreneurship
2
INTRODUCTION
Consider a team of entrepreneurs at the earliest stage of firm formation: the concept of the product or
service is available, as are some technical and marketing resources within the team. However, whether the
entrepreneurial team will successfully make the transition to the next step, i.e. formulate a business plan,
engineer a viable product, acquire the first set of customers, and ultimately raise funding, is far from
assured. Prior research shows that more than half such new ventures will fail within five years (Aldrich
and Ruef 2006), some recently pegging the failure rates at over 90% (The R.I.P. Report, 2014). One
significant institutional form has emerged in recent years to address this problem facing early-stage
entrepreneurs – seed accelerators (Cohen 2013). While accelerators remain understudied in the academic
literature, there is concurrently increasing interest in understanding how accelerators work and the
mechanisms through which they provide value to entrepreneurs. In this paper, we focus on this question –
we study how mentorship and investor ties, two vital aspects observed across accelerators in general
(Cohen 2013), lead to positive entrepreneurial outcomes during the accelerator and thereafter, to long-
term positive outcomes for participating firms.
Seed accelerators (accelerators hereafter) are short-term programs for new ventures that are
characterized by three factors: a free and public online application to competitively select a cohort of
participating startups; a standard deal terms that typically exchange a startup’s equity for early-stage seed
investment; an intense number of activities to speed the development of the startup’s product and securing
future funding (Miller and Bound 2011). These characteristics also set them apart from business
incubators and angels, two alternative forms of early support (Cohen 2012). Accelerators were first
observed in 2005 (advent of Y-Combinator) and since then, the accelerator concept has expanded to
become an important part of the entrepreneurial ecosystem (Cohen 2013). The number of accelerators in
the US has grown from 51 in 2009 to over 200 in 2014 (Lennon 2013) and have collectively raised over
$5B in investment while creating over 16,000 jobs since (Seed-DB 2014).
Early case-based work suggests that accelerators provide value to entrepreneurs through three
specific mechanisms (Miller and Bound 2011; Dempwolf et al. 2014). First, the accelerator incentivizes
the relocation of the firm from its native ecology to that sponsored or hosted by the accelerator. While
3
firms operating in their own environments may be affected by local resources, and competition (Chandler
and Hanks 1994, Carroll and Khessina 2005), accelerators impose their own admittedly artificial and
accelerated ecology on entrepreneurs as a condition for support. The artificially created ecology
commonly imposed on all selected new ventures further offers the other two distinguishing mechanisms –
mentorship and access to network ties (Cohen 2013).
Mentorship, studied within organizational settings, has been shown to influence outcomes such as
turnover and organizational commitment (Payne and Huffman 2005, Carraher et al. 2008). Within
entrepreneurship, mentorship has been traditionally associated with business incubators (Amezcua et al.
2013) and angels (Mitteness et al. 2012). Early stage entrepreneurs may find mentors who provide them
with tacit support, knowledge and expert advice in their own environments. However, these mentoring
networks are heterogeneous and depend on how munificent local environments are (Castrogiovanni
1991). By comparison, the accelerator institutionalizes the accelerated provision of mentoring to all
teams, based on a standardized process model, thereby justifying the cost of participating in the
accelerator or the equity stake that the accelerator requests from the new venture. Additionally,
participating teams can mentor each other through co-location, which is a central process by which
accelerators add value to entrepreneurs during the accelerator program (Cohen 2013).
The third mechanism that accelerators provide is through the expedited creation of network ties
that provide entrepreneurs with hard-to-find resources, especially critical ones like capital, access to
markets, and human resources (Stuart and Sorenson 2005). The accelerator solves this problem by
providing, within its ecosystem, managed and expedited access to investors, angels and other accelerators
(Cohen and Hochberg 2014). Since both investors and entrepreneurs are located within the accelerator
ecosystem, the first-order information asymmetry problem is partially resolved (Shane and Cable 2002),
creating a positive environment for entrepreneurs to gain accelerated access to investors. A critical part of
accelerators, as opposed to incubators, is that they are of short duration. Therefore, participating firms
have strong incentives in creating valuable investor ties since these are beneficial even after the
accelerator ends and firms “graduate” from accelerators.
4
In as much as these three mechanisms have been cited as critical within accelerators, there is no
systematic evidence to show that new ventures that do invest in these experience positive short-term
accelerator-level outcomes. Cohen and Hochberg state as much: “While proliferation of such innovation
accelerators is evident, the efficacy of these programs is far from clear”, p2 (2004). Beyond short-term
outcomes, it is worth asking – do these programs ensure longer-term survival and success of the
“graduating” new venture? Prior work has argued that mentoring and access to investor ties increase the
odds of new venture success (Eesley and Wang 2014; Shane and Cable 2002) but this work is outside of
the accelerator context. Within accelerators, it is necessary from a policy and business perspective to
understand what relationship exists between these mechanisms and accelerator-specific as well as longer-
term outcomes. This forms the primary research question we address in this paper – what is the effect of
mentoring and investor tie formation within the accelerator on entrepreneurial outcomes of participant
firms in the short term (within the accelerator) and the long-term (after the accelerator ends)? Beyond
the role of mentorship and investor ties, we also ask: how do accelerator outcomes influence longer-term
outcomes for participant firms? If accelerators do indeed provide guidance in incentivizing firms towards
accelerator-level outcomes, a positive correlation between achieving short-term outcomes inside the
accelerator and longer-term entrepreneurial outcomes, such as firm survival, should be evident. This
forms our second broad research question.
We test our hypotheses using data collected in a unique manner within an accelerator. Consistent
with the notion that studying entrepreneurship requires embedding the researcher within the institutional
environment, we embedded ourselves in an accelerator as founders of a startup. From the unique
perspective gained from our role as participants in the accelerator for over 6 months, we collected
entrepreneur-specific data using surveys and social media. In total, we collected data on one entire cohort
of 105 teams hosted by the accelerator in the spring of 2013. In addition to the data collected from the
accelerator, we also tracked each startup in its post-accelerator phase for a period of 15 months to identify
longer-term outcomes. Using these data, we estimate econometric models addressing the impact of
mentorship and investor ties on accelerator-level and longer-term outcomes.
5
Since the accelerator we study utilizes the lean startup methodology model (Ries 2011), we
identify three accelerator outcomes associated with the lean startup approach – the development of a
minimum viable product (MVP), an official startup launch and a first sale. Our results clearly indicate a
positive relationship between accelerator outcomes and longer-term survival of the firm; those startups
that launch officially and experience a first sale during the accelerator are significantly more likely to
survive and raise funds at the 15-month point. In addition, we observe that firms that experience
significant investor tie formation attributable to the accelerator observe higher long-term entrepreneurial
outcomes, such as a higher likelihood of firm survival and investment at the 15-month period. Our results
also show that mentoring has a significant positive effect on short-term outcomes during the accelerator,
such as the development of a MVP, an official startup launch and a first sale by the new venture. Finally,
we observe that firms with previous entrepreneurial experience gain less from mentoring and the
formation of investor ties than inexperienced firms, indicating a significant moderating effect of prior
entrepreneurial experience.
Our work here makes several contributions to the entrepreneurship literature. First, we provide a
validation of the accelerator model by showing the relationship between accelerator-level outcomes and
those observed in the entrepreneur’s unconstrained ecology in the form of firm survival (Lall et al. 2013).
While we are careful to not make causal claims, we believe we can minimize the impact of the most
common criticism of such associational studies – that of unobservable startup quality. All startups in our
sample were chosen competitively, thereby ensuring that all firms in our sample have a level of baseline
quality. Therefore, our analysis only identifies variations in long-term outcomes associated with
variations within the accelerator-level factors such as outcomes and investor tie formation, accounting for
cohort, team-level variables and time, thus establishing a level of control typically not observed in such
studies. Perfect identification would require randomization, which is rarely ever possible in
entrepreneurship studies (Ruef et al. 2003); our approach provides cleaner identification in comparison.
Second, we provide evidence of the relationship between mentoring and short-term accelerator
outcomes. Studying the effect of mentoring on outcomes is hard when entrepreneurs are embedded in
their ecologies and receive more than simply mentoring (Rodan and Galunic 2004). The accelerator, via
6
contrast, works by relocating most entrepreneurs out of their natural ecology into a “new” environment,
where by definition mentoring ties, if any, are freshly established. The accelerator we study is particularly
suitable since it is not located in a startup hub such as Northern California or Boston. Therefore, by
necessity, most mentoring ties are new, with limited spillover from the entrepreneur’s original ties. Hence
it is more likely that we identify the true effects of mentoring on accelerator outcomes in our work.
Finally, we contribute to the small literature on accelerators by opening up the ‘black box’ of
processes within accelerators such as mentoring and investor tie formation. Accelerators have caught the
fancy of policy-makers and businesses alike in recent years for the implicit association between
accelerators and general success in entrepreneurship. Consider, for instance, the Startup America
initiative launched by President Obama in 2011 in partnership with TechStars. Such initiatives are well-
meaning but given the paucity of systematic research into the accelerator model (Cohen and Hochberg
2014), policymakers have few tools available to judge when certain activities lead to successful
entrepreneurial outcomes within the accelerator or in the long-term. Our work here represents one of the
first systematic analyses, to our knowledge, of outcomes associated with accelerators, linked to processes
established within the accelerator. As a first step, we review this literature next.
RESEARCH CONTEXT – STARTUP CHILE
Research Context
The setting for the analysis we provide in this paper is Startup Chile, a government-sponsored accelerator
located in Santiago, Chile. The program was created by the Chilean ministry of economy with the goal of
transforming Chile into an innovation and entrepreneurial hub for Latin America. The project started as a
pilot in 2010 with 22 startups from 14 countries providing $40,000 USD of equity-free seed capital to
develop a startup for six months. After the success of the pilot, Startup Chile (SUP) expanded to two
rounds per year in 2011, each round lasting 6 months (StartupChile 2014). As of 2013, the program offers
approximately $45,000 USD of equity-free seed capital, a one-year work visa, limited help getting into
Chile and finding accommodations, and access to the network of investors, mentors, and entrepreneurs
SUP has developed within and outside Chile for participating entrepreneurs. In exchange for these
resources, teams must remain in Chile while working on their startups for 6 months. Moreover, all startup
7
teams are required to engage in social impact activities in Santiago and in Chile, specified as part of the
program and applicable to all firms, whether Chilean or not.
Since 2011, SUP has received over 10,000 applications from 112 countries and has attracted
considerable media and entrepreneurial attention even outside the Chilean context. The application
process for the accelerator consists of a three-page online application with questions about the team, the
project, and status of the startup. According to the administrators at SUP, entrepreneurial firms are
accepted into a cohort based on two main criteria. First, an evaluation of the quality of team amounts to
50% of the application, as stated by SUP’s director:
In the world, there is a 85% failure rate in these types of projects [early-stage ventures], so we believe
the main criteria is to select the best possible teams, who in the worst of cases, will ultimately learn
from their failure (EMOL, 2014).
The other dominant factor driving acceptance of the startup relies on evaluation of the quality of team’s
business model, the project’s likelihood of success, and the size of the market (Startup Chile 2014). Once
teams arrive to Chile, SUP provides access to a working space for all the startups in downtown Santiago
where desks, meeting rooms, and sofas are provided. The program then gets started with all selected
startups doing 5-minute business ‘pitches’ to all the other startups. In terms of working locations, the
majority of the teams choose to work in the Startup Chile office space, but this is not mandatory. Many
Chilean teams, for instance, do not work at the Startup Chile offices.
As part of the terms and conditions of Startup Chile, every startup must fulfill a minimum quota
of social impact activities in Chile (Startup Chile 2014). Finally, the program ends with each cohort of
startups presenting their work during a high-profile Demo Day where the top 15 startups pitch their
businesses to invited star investors in one of the top venues of the city. All startups compete with each
other in several rounds for a place to pitch on Demo Day. In terms of outcomes from the program, Startup
Chile keeps detailed data of the social impact activities performed by the entrepreneurs. There is less clear
data available, however, on the business outcomes of the startups that have participated in the accelerator.
More importantly, there is even less evidence linking the resources provided by the accelerator to the
outcomes experienced by participating startups. First, we define the accelerator outcomes that are of
8
relevance to SUP, since different accelerators tend to emphasize different approaches to early-stage
entrepreneurship (Cohen 2013).
Defining Accelerator Outcomes
As startups advance through Startup Chile, there are two main checkpoints on the progress of
each startup. The first checkpoint is a meeting with an assigned financial advisor to agree on a set of
objectives for the startup while in the accelerator. These goals are heavily influenced by the Lean Startup
methodology introduced by Eric Ries (2011). The Lean methodology is pervasive in SUP and other
accelerators (Seed Ranking 2013) and is based on testing specific hypotheses, gathering early and
frequent customer feedback, showing early prototypes to prospects, and measuring success quickly
through an iterative process. Figure 1 shows the feedback loop, a central component of the ideology.
Since its introduction, both the popular (Lohr 2011; Stengel 2015; Tam 2010) and academic press
(Winkler 2014) have been interested in the Lean startup movement and its influence on entrepreneurs.
Central to the lean methodology are three key principles: First, replace months of planning and
research with testing specific hypotheses. Second, startups should engage in customer development to test
their hypotheses, i.e. direct contact with potential customers to gather feedback on the business model,
product features, and distribution channels. Third, the methodology recommends the use of agile
development (Beck 2001) to create minimum viable products (MVP), which represents the minimum
functionality or set of features within the product, allowing the firm to test the product in the market and
gather customer feedback, consistent with the second principle (Eisenmann et al. 2012).
Building on these entrepreneurial outcomes set by the startups in the accelerator, we introduce the
notion of accelerator outcomes. Differentiating accelerator outcomes from broader entrepreneurial
outcomes is necessary because these outcomes are emphasized and incentivized within the accelerator.
Additionally, typical measures of entrepreneurial outcomes, such as firm survival and funding, are
assured while in the accelerator. We define accelerator outcomes as the measurable lean startup
methodology business outcomes that may occur while participating in an accelerator. We introduce three
accelerator outcomes in this study: the completion of an MVP, the official product launch of a startup,
and a first sale, representing a significant increase in the viability of the firm. In addition to these
9
outcomes, we also identify longer-term entrepreneurial outcomes that are of interest to early-stage
startups; specifically, these are the survival of the firm 15 months after the accelerator program and the
amount of external funding received in those 15 months. As significant antecedent factors for these
outcomes, we discuss the role of mentorship and investor networking ties in the next section.
RESEARCH HYPOTHESES
Within Startup Chile, the creation of mentorship and investor ties arises as a result of multiple
interactions. Regarding mentorship, as a first step, participating founders mentor each other. There is a
high degree of collaboration among startup founders, and they give each other technical, programming,
and business feedback. Second, SUP encourages joining a ‘startup’ tribe, which are specialized mentoring
groups that meet every week to improve different aspects of the startup. Third, SUP invites entrepreneurs,
founders, investors, and VCs every week as speakers, who act as mentors and are available to startups as
resources. Beyond mentoring, other events are focused on raising capital and connecting founders with
investors. Here, the accelerator acts as a matchmaker between investors and startups, and it is common to
meet visiting investors every week, especially investors visiting from the U.S., Europe, and South
America. Finally, there are numerous social events, such as investor-sponsored parties. While
participating in these activities is highly encouraged, these are ultimately optional, allowing each startup
considerable agency. Hence, the amount of mentorship received and the level of access to the investor
network within SUP vary from startup to startup. We propose hypotheses for these relationships next.
Mentorship: Access to Knowledge and Information
A fundamental challenge for entrepreneurs is to accurately identify and shape the opportunity to be
pursued. Access to information about the value of available opportunities is a central part of this process.
Prior work in entrepreneurship suggests that social networks provide this valuable and privileged
information (Stuart and Sorenson, 2005). Our fieldwork here corroborates this argument and also
indicates why mentorship is valuable to entrepreneurs within the accelerator. To apply to an accelerator,
entrepreneurs need to provide an existing idea, which indicates that the entrepreneur has already
identified an opportunity. However, identification of the opportunity is not a single-time event (Dubini
and Aldrich 1991). Instead, the process of identifying the opportunity at the accelerator is a continuous
10
cycle of stating, testing, and refining the target opportunity. In accelerators, mentors play a key role in the
process of identifying the opportunity by providing feedback on the existing idea and by providing access
to private information to further refine the idea or ‘pivot’ into a similar but different business
model. Many of the teams we surveyed provided qualitative evidence of the role of mentorship in
identifying or pivoting to a more specific opportunity. One team remarked:
“Sure, we came in part for the funds provided, but we also came to SUP because we needed help
deciding where to focus. We were sure we had found the right opportunity in cloud security when we
applied, but by the time we started meeting our mentors, it was obvious our idea was too general. The
mentors helped us pivot into the more specific market of containerizing Dockers. Suddenly, instead of
sounding like every cloud security startup out there, we found a specific need that we could solve.” -
Tutum Cloud (A cloud services startup).
Beyond opportunity identification, a major challenge for entrepreneurs is the mobilization of resources to
provide a solution for the opportunity. In pursuing the opportunity, the literature argues that entrepreneurs
need tacit knowledge required to create a successful venture (Alvarez and Barney 2004). For example,
entrepreneurs need to be able to concisely deliver their idea and execution, commonly refereed to as a
‘business pitch’. Therefore, a second mechanism by which social networks affect the entrepreneurial
process is by providing more channels through which tacit information flows (Stuart and Sorenson 2005),
which in turn have been shown to positively affect entrepreneurial outcomes (Liles 1974; Klepper and
Sleeper 2005).
In our context, developing tacit knowledge in entrepreneurs is an area in which SUP mentors also
play a key role. Through continuous mentorship opportunities, entrepreneurs are coached to develop tacit
knowledge in the form of business pitching, marketing and branding strategies, product development, and
recruiting and managing resources. Entrepreneurs from a health IT (HIT) startup remarked as much:
“We got to Startup Chile with a great MVP, but we had no idea how to sell this product to physicians
or patients. After we were assigned a mentor with experience in health IT, we started understanding
how we should be approaching physicians, who were the key for the adoption of our product. For
example, we started focusing on security and addressing privacy concerns for patient data. Before we
joined SUP, we never thought security was a concern for our users.” -Medko (A health IT startup).
In summary, mentors in an accelerator can affect entrepreneurial outcomes by increasing the amount of
private information to identify the right opportunities and the tacit information available to organize and
manage resources in a new venture. In our context, where startups target the completion of certain
11
discrete accelerator outcomes and mentors are likely to emphasize these outcomes as well, mentorship
activities are likely to push the new venture towards these outcomes. Therefore, we propose:
Hypothesis 1a. Startups that engage in mentorship activities during the accelerator are more likely to
release an MVP during the accelerator.
Hypothesis 1b. Startups that engage in mentorship activities during the accelerator are more likely to
have an official product launch during the accelerator.
Hypothesis 1c. Startups that engage in mentorship activities during the accelerator are more likely to
generate the first sale (i.e. new revenues) during the accelerator.
Investor Ties: Access to Financial and Labor Capital
In the process of mobilizing resources to pursue entrepreneurial opportunities, networks are highly
influential for two reasons. First, investors use their network to become aware of investment opportunities
(Stuart and Sorenson 2005). Additionally, information asymmetry is a significant barrier to the
investment decision. Therefore, social networks offer information about the entrepreneur’s work ethic and
integrity, thereby helping access financial capital by helping reduce information asymmetry (Hallen
2008). Gulati (1995) further argues that embedding a transaction in an ongoing social relationship
motivates both parties to keep the relationship fair and generates a sense of obligation between the parties.
In general, there is consensus in entrepreneurship that social networks are important and affect
entrepreneurial funding and survival along different mechanisms. However, our question here is whether
such ties, instituted through the accelerator program, is successful in enhancing accelerator outcomes
(short-term) or do they manifest in the longer term?
Shane and Cable (2002), in a seminal analysis of direct and indirect ties, argue and show that
while direct ties increase trust and the quality of information that reaches potential funders, indirect ties
are useful in enhancing awareness among investors. However, their analysis stems from data from
investors rather than entrepreneurs and does not control for the entrepreneur’s prior experience. In
contrast, in an analysis of the Chinese VC market, Wang (2008) finds that while social ties are helpful in
allowing an entrepreneur to enter the focal VC’s consideration set, within the consideration set
entrepreneurs do not benefit directly from their social connections. This is consistent with the notion that
awareness of an entrepreneur may be distinguished from truly relevant information that leads to funding.
While accelerators are very useful in providing the first mechanism through their relatively shorter
12
program duration, they may not provide enough trusted information to directly influence accelerator-level
outcomes. However, firms that form such investor ties through the accelerator are likely to gain from
them in the longer run, i.e. after the accelerator program. In sum, the facilitation of investor ties helps
entrepreneurs find funding opportunities and increase the odds of survival in the long run but not so in the
short run of the accelerator’s program, which typically runs from 3 to 6 months (Cohen 2013).
Our fieldwork in Startup Chile further lends credence to the notion that social ties help
entrepreneurs raise capital by exposing them to investors and starting the trust creation process with
investors. Moreover, as the accelerator develops, many such entrepreneur-investor relationships
strengthen and the perceived risk of investing in one of these startups decreases for investors. One startup
stated:
“Since day one, the director [of Startup Chile] said that from now on we should always be on pitch
mode. I didn’t quite understand what he meant at the beginning, but after the first weeks there, I found
myself meeting a lot people and investors who were always asking what our idea was and who we were.
This was a radical change for our team as we never talked about our idea or product with strangers. In
fact, we tried to keep it as secret as possible. While the value of many of the social events sponsored by
the accelerator was in question at the beginning, after the first two months in the accelerator, we had
met more investors in these events than the last two years combined” – Arrively (A travel startup).
A second entrepreneur explained further:
“Once we developed relationships with a few investors in [CITY A], the biggest barrier to raise capital
was trust. How could they make sure we were not going to leave after the program was done? How
could we make sure these investors really had the funds and expertise to guide us? The sub-director of
the accelerator was essential in this process. After observing our work ethic for three months and how
we improved our MVP, [SUB-DIRECTOR] was able to speak about the integrity of our team to local
investors, which is what filled this trust gap.” Simple Crew (A mobile management startup).
We thus hypothesize that the investor networks provided in the accelerator increase the amount of
funding opportunities for the startup and thus its longer-term survival and funding. We focus on survival
and funding as two entrepreneurial outcomes of interest in the literature (Wicker and King 1989; Bates
and Servon 2000; Van Praag 2003; Bosma et al. 2004; Klepper, 2001; Shane and Stuart 2002). More
formally, we propose:
Hypothesis 2a. Startups that engage in investor-ties building activities during the accelerator are more
likely to survive in the longer-term.
Hypothesis 2b. Startups that engage in investor-ties building activities during the accelerator are more
likely to raise other early-stage investment in the longer-term.
Finally, one unanswered question with significant policy implications remains in this context: what is the
relationship between accelerator outcomes and long-term firm success? To our knowledge there is no
13
empirical evidence that these discrete accelerator outcomes (MVP, official launch, and first sale) are
indeed predictive of longer-term firm survival or funding. If accelerators truly help new firms emerge
from the “liability of newness” (Stinchcombe 1965), we should observe a significant positive association
between those startups that experience positive accelerator outcomes and those that survive and acquire
funding after the program has finished, indirectly providing evidence of the ‘accelerator effect’ (Miller
and Bound 2011). We therefore hypothesize:
Hypothesis 3a. Startups that accomplish accelerator outcomes (MVP, official launch, and first sale)
during the accelerator are more likely to survive in the longer-term.
Hypothesis 3b. Startups that accomplish accelerator outcomes (MVP, official launch, and first sale)
during the accelerator are more likely to raise early-stage funding in the longer-term.
EMPIRICAL STRATEGY
Collecting detailed data from within accelerators represents a challenge. Most accelerator cohorts are
relatively small (approximately 5-10 teams), which makes quantitative analysis infeasible. Furthermore, it
is generally hard to collect detailed data from teams within the accelerator and to observe the dynamics
that exist within from the outside, leading to the paucity of research on accelerators (Cohen 2013). In this
paper, we adopt a unique albeit different approach to collecting data. We embed ourselves in SUP as
founders of a startup and collect data through surveys and observation through the duration of the
accelerator. From the rare perspective gained from our role as entrepreneurs in the accelerator over 6
months, we collected entrepreneur-specific data tracking the progress of startups during the program. One
of the benefits of SUP was also in providing larger samples than usual – each cohort at SUP includes over
a 100 firms, allowing the possibility of quantitative analysis. Our tenure in the accelerator also allowed us
to interact with the entire ecosystem of the accelerator (i.e. entrepreneurs, staff, mentors, and investors)
and gain access to not only direct information from the teams but also the information provided by the
teams on social media sites (such as the firms’ Facebook, Twitter, and LinkedIn pages) as well as their
fundraising goals and activity data on entrepreneurial platforms, such as angel.co.
Sample and Data Collection
The Spring 2013 cohort of Startup Chile served as the research setting for this study. To test our research
models, we collected survey data from these startups. We developed the survey after an extensive review
14
of the literature and interviews with entrepreneurs, the staff members at SUP, investors, and
administrators. As far as possible, we used questionnaire items that have been widely tested in the
literature (described later in this section). Before the eventual use of the survey, we pilot tested the survey
with 20 entrepreneurs from the previous cohort of Startup Chile, the fall cohort of 2012. Based on the
feedback from the pilot and our interviews, we iteratively refined the survey instrument. These steps,
combined with the use of pre-existing scales, helped ensure face and content validity of our survey
instrument.
The relevant data used in analysis here was based on information collected at the end of the
venture’s stay at SUP, thereby allowing us to collect data on accelerator outcomes as well as perceptions
of how much the team had actually invested in mentorship and investor ties, as opposed to intention. Data
was collected in a single continuous wave as startups conducted the exit interview with SUP in the last
three weeks of the program at the co-working space and the administrative offices in Santiago, Chile.
While we initially attempted to collect survey data from all official members of the venture team (modal 3
members), or those engaged in a contractual relationship with SUP, this proved infeasible primarily
because by the end of the accelerator, some members of the team had relocated back to their home
country. We obtained a total of 175 questionnaires from the 105 firms representing the Spring 2013
cohort. Six surveys were discarded as they were from entrepreneurs from previous cohorts while ten more
were discarded because non-founder members, such as interns, helped complete them. In nine cases, we
had two founders from the same startup complete the survey while twelve firms had three founders from
the same startup complete the survey. For those firms with multiple respondents, we examined the
correlations on the key perceptual questions across multiple respondents and found correlation
coefficients of over 0.85, suggesting high levels of uniformity in responses. Therefore, for subsequent
analyses, the responses from individual respondents on the same startup were averaged out to form the
research variables. The dataset includes teams from 13 different nations, mean age of 28 years, from four
races, and with varying levels of education (spanning high school to doctoral degrees). A common
criticism of entrepreneurship research is the homogeneity of the respondent pool (i.e. white males from a
15
single state in the US); in contrast, here we capture a reasonably wide cross-section of racial and cultural
backgrounds as part of the accelerator. We describe the specific variable operationalization next.
Variable Descriptions
We used prior items from the literature on mentorship and network ties whenever possible. As
recommended in the mentorship literature, rather than employing a binary measure, mentorship was
measured using the multi-dimensional mentorship functions questionnaire (MFQ-9) developed by
Scandura and Ragins (1993). The MFQ-9 is a shorter version of the initial 20-item MFQ (Scandura,
1992). The MFQ measures three mentoring functions: career support, psychosocial support, and role
modeling with three questionnaire items each, for a total of nine items. The instrument was measured in a
5-point Likert scale with responses ranging from 1 (strongly disagree) to 5 (strongly agrees). The
questionnaire items used to measure mentorship, and all other such variables, are provided in Table 1.
Similarly, to measure the development of network ties for the entrepreneurs during the
accelerator, we adapted the measures used by Shane and Cable (2002), which measure direct and indirect
ties between investors and entrepreneurs, to the context of entrepreneurs in an accelerator. The direct and
indirect ties were measured with six items in a 5-point Likert scale with responses ranging from 1
(strongly disagree) to 5 (strongly agrees). The items captured the extent to which ventures were able to
form professional, informal or friendship-based relationships with investors through SUP.
In addition to these two key independent variables, we also control for several contextual and
demographic variables that have been found to be influential in increasing firm survival and rates of
funding; we also use these variables as control variables in our models of accelerator outcomes. Pre-
entry experience is measured using the scale from Dencker et al. (2009), which measures the
entrepreneurial team’s pre-entry knowledge of the business activity of the startup. Additionally, and
related to pre-entry experience, we adapt a measure of organizational capital using Hsu (2007) to measure
the number of startups previously founded by the founding team, the age (in months) of the startup, and
the average age (in years) of the founding team. Also following Hsu (2007), we collect the level of
education of the founding team (i.e. the number of members of the founding team possessing MBA and
PhD diplomas). We also take into account team composition variables, and we adapt the measures from
16
Campion et al. (1993) to control for the degree of heterogeneity, flexibility, preference for group work,
and the number of founding members of the startup. Finally, we also asked entrepreneurs to provide what
percentage of the seed funds provided by SUP were spent on operations (i.e. rent, web storage, cloud
services) and human resources (i.e. interns, salaries) – these represent two official accounting categories
that entrepreneurs are requested to report to SUP. Further details on operationalization are shown in Table
1 while summary statistics and correlations are shown in Table 2 and 3 respectively.
As a first step, we tested for the psychometric properties of the perceptual constructs to confirm
that they were of good quality. With respect to reliability, for multi-item reflective constructs, the
composite reliability was calculated and varied from 0.72 to 0.84 across constructs, thereby establishing
their reliability. All Cronbach’s alpha values obtained for the constructs easily exceeded the 0.70
threshold established in the literature. To test divergent validity, we verified that the average variance
extracted (i.e. the average variance shared between a construct and its measures) was greater than the
variance shared with other constructs in the model, i.e. other perceptual constructs here (Fornell and
Bookstein 1981). In addition, we performed exploratory factor analysis using varimax rotation and a
threshold eigenvalue of 1.0 to ensure convergent validity; all hypothesized items loaded appropriately
with high factor loadings on the underlying reflective construct, as shown in the first column of Table 4.
Finally, we ensured the factor structure of the reflective constructs through confirmatory factor analysis
using LISREL. All the hypothesized paths from the indicator variables to the hypothesized latent variable
were significant (p90%). We find strong agreement (over 95%) between observed survival on social
media and the email form. Therefore, we use survival noted from activity on social media as our measure,
with value 1 if the firm survives and 0 otherwise. In our cohort, 63%, i.e. 66 startups, still survive at the
end of the 15-month period.
Beyond survival, we also measure early stage capital investments received by the startup using
Angel.co, a platform designed to showcase and raise capital for startups. The use of Angel.co is highly
encouraged in SUP and the entire cohort actively used it throughout the accelerated period and during the
15-month period where we followed the startup. As with survival, we also confirm new investments using
the same short email format, where fellow entrepreneurs provide information about their current fund-
raising status
1
. Again, we find a strong correlation (approximately 0.9) between the reported measure of
1
We note here that such email and social media interactions are common amongst entrepreneurial teams
from the same cohort and indeed, represents one of the advantages of the data collection approach we adopt here. As
members of the accelerator’s programs, we are able to use these networks that are usually closed off to outsiders.
18
total investment in Angel.co and the email form (to the nearest 1,000). In the interest of consistency, we
use the information from Angel.co as our measure of capital investment. Approximately, 75% of our
cohort (78 startups) received early stage investment. The firms that do not raise any funds are excluded
from the analyses in which investment is the dependent variable. The average amount raised by these
startups in the 15-month period was approximately $225,000.
MODELS AND RESULTS
Using the data collected from SUP, we estimate econometric models addressing the impact of mentorship
and investor ties on accelerator-level and longer-term outcomes. We specifically estimate three models to
test our hypotheses. Model 1 tests the effects of mentorship and investor ties on short-term accelerator
outcomes, while Model 2 focuses on longer-term outcomes. Finally, Model 3 examines how accelerator
outcomes influence longer-term outcomes, to indirectly test for an ‘accelerator’ effect.
Model 1: Accelerator Outcomes
Our goal here is to estimate the effect of mentorship and investor ties on accelerator outcomes - MVP,
Launch and First Sale (FS). Since these outcomes are discrete, we fit a logistic regression model for each
accelerator outcome ? ? {???, ??????, ??}:
logit??
?
? = ?? +?? +?,
where ?
?
equals 1 when the outcome ? is accomplished during the accelerator period and equals
0 when the outcome is not accomplished during the accelerator period. Startups that accomplished
outcome ? before the accelerator period (observed occasionally) are therefore excluded from the models.
Specifically, we exclude 11 firms who have released an MVP before the accelerator, 9 firms who have
launched products officially before the accelerator, and 5 firms who have generated revenue before the
accelerator. Additionally, the vector X contains the variables corresponding to pre-entry experience,
education, team, resource allocation, mentorship, and investor ties as described above and Z includes
indicators of whether previous accelerator outcomes were accomplished before or during the accelerator
period. Logically, MVP occurs before the startup’s official launch, which should result in a first sale.
Table 6 displays the coefficient estimates and standard errors for each explanatory variable as
well as model fit indicators corresponding to each accelerator outcome. The model fit indices are suitably
19
high. More importantly, we find that Hypotheses 1a, 1b, and 1c are strongly supported; mentorship is
significantly associated with each accelerator outcome (p