Briefing Document: The design of startup accelerators
1. Introduction and Core Concepts
Rapid Growth of Accelerators: Startup accelerators are a recent but
rapidly expanding part of the entrepreneurial landscape. The paper notes
there are an estimated 2,000 accelerators globally, with a significant
portion located in the US. They are defined as limited-duration, cohort-
based programs that support startups through mentorship, education,
and networking, often providing seed capital and coworking space.
Distinguishing Characteristics: The key feature that separates
accelerators from incubators and other intermediaries is the "fixed-term
and cohort-based aspects." This differentiates them from more open-
ended incubator programs. The paper defines an accelerator as "A fixed-
term, cohort-based program for startups, including mentorship and/or
educational components, that culminates in a graduation event."
Heterogeneity of Programs: A critical point is that accelerators are
not homogenous. There's significant variation in their design, mission,
and impact. This variation has been largely overlooked by prior research,
which often treated all accelerators as the same. The paper seeks to
understand this variation and its implications.
Goal of the Paper: This paper aims to map the landscape of US
accelerators and analyze the variations in design features. It examines
the relationships between these design elements and the performance of
the startups that participate. The paper seeks to help stakeholders
understand how accelerators vary and how they impact startups and
ecosystems.
Accelerators as Laboratories: The paper proposes that accelerators
can act as "laboratories" for studying the entrepreneurial process and
understanding what new ventures need to thrive.
2. Literature Review & Research Gaps
Fragmented Research: Existing research on accelerators is "highly
fragmented" and has not yet formed a unified body of knowledge.
Conflicting Results: Some studies show a positive impact of
accelerators, while others find muted or even negative effects. This
highlights the importance of considering the heterogeneity of
accelerators. For example, studies have found that "even top startup
accelerators can be detrimental to accelerated firm performance."
Key Considerations: Studies suggest the impact of accelerators
depends on both the characteristics of the accelerator and its applicants,
making it "difficult to untangle when considering accelerators are not a
homogenous population."
Focus on Program Design: The paper argues that it’s essential to
move beyond simply assessing if accelerators work to exploring how their
design impacts outcomes. Prior studies often overlook the significant
variation in accelerators along multiple design features, which is
important for understanding their impact.
Ecosystem Role: Accelerators also act as "catalysts for wider ecosystem
development," bringing in resources and connections, and driving
investment in new local firms. They function as brokers in the developing
social graph of their region. Many founders chose location near to their
childhood home. Accelerator founders have a mix of regional
development aspirations and pecuniary gains.
3. Data and Methodology
SARP Dataset: The study uses data from the Seed Accelerator Rankings
Project (SARP), a database that tracks accelerator program graduates.
The authors also augment this data with other sources such as LinkedIn,
Venture Economics, and Crunchbase.
Multi-Dimensional Analysis: The study analyzes accelerators along
three key dimensions:
o Program Elements: Duration, funding, mentorship, cohort size,
workspace, etc.
o Sponsor Identities: Corporations, investors, universities,
governments, etc.
o Founder Backgrounds: Professional experience and education of
managing directors.
Regression Analysis: Linear regression models are used to examine
the correlations between these elements and the performance of startups
(e.g., funding raised, valuations).
4. Accelerator Emergence and Sponsors
Growth Over Time: The largest growth of accelerators in the US
occurred during 2011-2013, coinciding with the recovery from the 2008
financial crisis. Local governments also began to use the accelerator
model for economic development.
Geographic Diffusion: Accelerators have expanded geographically
beyond traditional entrepreneurship hubs to "areas with less activity."
While some are located in "key entrepreneurship hubs such as Silicon
Valley, Boston and New York," many can be found in places such as
Memphis, TN, Spartanburg SC, and Cincinnati, OH.
Founding MDs: Accelerators typically have between one and three
founding managing directors (MD) responsible for operations. The
average is 2.0 founding MDs. MD backgrounds vary considerably.
Sponsors: Sponsors provide financial and in-kind support. The largest
categories of sponsors are corporations (62%) and investors (57%).
Governments (34%), Non-profits (20%) and universities (16%) also
contribute. The average accelerator has 1.35 unique sponsor types.
o Investors as Sponsors: Investor-led accelerators focus on
improving startup quality and reducing costs to access startups,
often emphasizing cohort selection, education, and "demo days". Y
Combinator is cited as a good example of this model.
o Corporations as Sponsors: Corporations often use accelerators
to "scan the environment for new technologies and markets or
promote their products and services," sometimes providing access
to company resources and intellectual property. For example,
Disney's accelerator lets firms experiment with their characters.
o Academia as Sponsors: University-led accelerators aim to diffuse
university research into the economy through firm formation, or "to
develop student’s entrepreneurship skills." They may focus on
technology transfer or student capacity building. Arizona State
University's "Furnace" program, which commercializes technologies
developed in ASU labs, and MIT's "Delta V," which reinforces
entrepreneurial education for students, are good examples.
o Government as Sponsors: Government-led accelerators often
focus on regional economic development, including bringing in or
retaining skilled entrepreneurs. Government aims to promote job
creation. The Ark Challenge in Fayetteville, Arkansas, which
attempted to retain talent in the region, is used as an example.
5. Program Elements
Cohort Structure: The use of cohorts is a key design element. Cohort
structure attracts multiple key resources to the program (startups,
mentors, and investors). Key design choices around cohorts include: size,
human capital characteristics, industry diversity, and stage of
development.
Funding and Equity: Accelerators have different investment models,
often providing a small amount of capital in exchange for equity. The
mean minimum funding provided is $26K while mean maximum is $68k.
Equity stakes range from 0% to 15%, with a mean of 6.1%.
Mentorship: The provision of technical, business feedback, advice, and
social support is a key component. Mentors can be internal or external to
the program.
Education: Some programs provide formal education, while others use
an "a-la-carte" approach.
Workspace: 77% of accelerators provide workspace. The provision (or
not) of workspace can "generate major differences in terms of the social
and cultural impact of the program on startup firms."
Graduation Events: Most programs end with a "graduation event," like
a demo day, where startups pitch to investors.
6. Key Findings & Relationships
MD Background and Design: There are systematic relationships
between the professional experience of founding managing directors and
the choices made regarding the design of the accelerator, and its sponsor
organizations. Some professional backgrounds and sponsors are a more
"natural fit" than others.
Sponsor Type and Design: Founding sponsors of accelerators also
impact the design choices. For instance, venture capital-sponsored
accelerators often take more equity and provide less office space.
Ecosystem Fit: The geographic location of an accelerator, and its
surrounding ecosystem, are correlated to the professional backgrounds of
the founding managing directors and the types of sponsoring
organizations. Government-sponsored programs are more likely to be
located in larger cities, but with lower patent activity, while investor-
sponsored programs are associated with cities with higher patent activity.
Performance Associations:
o Investor-sponsored programs are associated with higher
amounts of capital raised post-graduation, and higher valuations.
o Government-sponsored programs show no significant increase in
the likelihood of raising significant funding relative to other sponsor
types and raise lower sums of capital.
o Corporate-sponsored programs are associated with higher
valuations, which suggests that corporations can act as a substitute
for capital by enabling the company to do more with less.
Longer programs are associated with higher performance for alumni
startups.
Smaller cohorts appear to be associated with better performance.
Higher equity stakes taken by the accelerator are negatively
correlated with performance of the companies, which is likely due to for-
profit accelerator models.
External mentorship was also shown to have a negative relationship to
performance.
Pre-Accelerator Funding and Revenue: The patterns of
design/performance are consistent, even when controlling for the level of
funding and revenue that startups had before entering the accelerator.
7. Future Research Agenda
Startup Process: Accelerators can be used to study the relative
contributions of ideas and founder capabilities.
Entrepreneurial Entry: Explore how accelerators influence who
becomes an entrepreneur and how ideas and teams co-evolve.
Treatment Effects: Develop better understanding of how different
accelerator design choices influence outcomes for startups.
Accelerator Business Models: Investigate the business models
employed by accelerator programs, as data has been challenging to
obtain.
Ecosystem Impact: Study how accelerators impact regional
entrepreneurial capacity and the conditions necessary for success.
Mechanisms of Impact: Investigate how accelerators impact their
stakeholders, and how mechanisms might vary for different ecosystem
programs.
8. Conclusion
Correlational Evidence: The paper provides "guideposts" for
researchers, policymakers, and practitioners, though the results are
correlational and not causal.
Tradeoffs: Startups should be aware of the tradeoffs between different
types of accelerators, such as investor-led programs that tend to increase
performance at the cost of equity.
Policy Implications: Policymakers should be aware that the higher
performance of investor-led programs can be fully captured by the
sponsors and equity holders.
Broader Ecosystem: It is important to recognize that accelerators are
just one type of intermediary in the entrepreneurial ecosystem.
Need for Taxonomy: Further cataloguing and creating a taxonomy of
entrepreneurship programs to compare the different types of programs in
the ecosystem is needed.
Future Research: Future research should focus on "systematically
understand[ing] how the spatial distribution of entrepreneurship support
programs alters the likelihood of entrepreneurial entry."
Key Quotes
"Accelerators are limited-duration programs that help cohorts of startups
build and launch their ventures."
"The definition of an accelerator becomes: A fixed-term, cohort-based
program for startups, including mentorship and/or educational
components, that culminates in a graduation event."
"… accelerators vary widely in their [design]…"
"While accelerators have proliferated quickly and startups are flocking to
such programs, research on this new organizational form is still
emergent."
"...accelerators can be viewed not only through their impact on individual
startups but also as a catalyst for wider ecosystem development."
"The cohort structure used to admit startups serves as one of the most
important design innovations introduced by accelerators."
"These results suggest that engagement with corporations might provide
a substitute for capital spurring startups to be able to achieve more (and
thus be worth more) with less capital inputs."
"Our preliminary results, while correlational rather than fully causal,
provide guideposts for researchers, policymakers and practitioners alike
as they seek to explore and act upon the impact of accelerators."
