Entrepreneurship is influenced by many factors, including ones that may not seem obvious, like ethnicity and culture. Much has been written about diversity in the startup and venture capital worlds in recent years.
Yet there has not been a thorough analysis of the influence of ethnicity in startup teams because studying ethnic and racial backgrounds has traditionally been unwieldy, requiring many expensive surveys. Furthermore, only 3% of Fortune 500 companies make their diversity data accessible to the public, making studies increasingly difficult to conduct.
In a novel approach to studying ethnicity in startups, Kauffman Fellows in collaboration with MaC Venture Capital leveraged software to analyze the faces of startup executives to gather empirical answers. More specifically, the Kauffman Fellows Research Center analyzed the profile images of 260,000+ startup founders and executives in the United States using publicly available demographic models created by Clarifai (Figure 1).
The photos that had conclusive results (in this case, the algorithm was 70%+ confident of a person’s ethnicity) were placed into five perceived ethnic categories: White, Asian, Latinx, Black, and Other.
- We found that 79.2% of startup executives were White, 15.6% Asian, 2.6% Latinx, 2.1% Black, and 0.5% Other.
- The data show a large discrepancy between the Latinx working-age population in the United States (17%) and Latinx startup executives (2.6%), as well as the Black working-age population (13%) compared to startup executive representation (2.1%).
- Latinx and Black students have been earning bachelor’s and master’s degrees at record rates, but those gains are not translating into representation. In fact, U.S. startups need 6.7x more Latinx and 5.6x more Black executives to match working-age U.S. demographics.
Perceived Ethnicity Matters
We first evaluated perceived ethnicity at a high level to create a baseline. Our analysis of 260,000+ samples found the following breakdown shown below in Figure 2.
Our findings show a large discrepancy between the perceived ethnicity of executives in startups and the general demographics in the United States (Figure 3). According to the projections of the National Center for Higher Education and Public Policy and the U.S. Bureau of Labor Statistics 2018 report, currently about 17.5% of the working-age population is Latinx compared to 2.6% in startup executives. The Bureau of Labor Statistics also found that 11.8% of the working-age population is Black compared to only 2.1% of startup executives.
Putting our data and the U.S. working-age population side-by-side shows two clear trends (Figure 3). First, White and Asian groups are both over-represented in startup executive leadership. And second, the data show that both Latinx and Black groups are more than 80% under-represented.
Next, we explore education levels and type to see if education helps explain this difference.
Education Levels and Types Vary by Perceived Ethnicity
There is a large variance in education when segmenting by perceived ethnicity. Asian executives, for example, had the highest percentage of master’s and doctoral degrees in our sample. This finding agrees with national statistics, where 52.3% of Asian-Americans hold a bachelor’s degree or higher—the highest for any ethnic group in the United States. In fact, as shown in Figure 4, education seems to have helped Asian-Americans break into startups, but it does not seem to have the same effect for Latinx or Black graduates.
Moreover, we would like to highlight the changing education demographics over the last several decades. Figure 5, showing data from the National Center for Education Statistics (NCES) data shows college enrollment rates by ethnicity for the years 1980 through 2016.
The NCES data show two clear trends. First, that White Americans are becoming a smaller proportion of post-secondary degree holders. And second, Latinx and Black graduation rates have risen 350% and 55%, respectively (Figure 6). These significantly larger numbers of Latinx and Black Americans with degrees have not translated into similarly higher levels of representation within startups. Education can certainly help, but it does not provide the full picture.
Our results show that White and Asian ethnicities are over-represented in the population of startup executives. 79% of Crunchbase profiles had a perceived ethnicity of White, whereas Whites account for only 63% of the working-age population in the United States, making them 25% over-represented. Furthermore, 15.6% of profiles had a perceived ethnicity of Asian, whereas Asian Americans account for only 6% of the working-age population in the United States, making them 160% over-represented. There are also two significantly underrepresented groups: Latinx and Black.
Latinx employees make up 17% of the U.S. working-age population but only 2.6% of startup executives in our dataset, making them 85% underrepresented. It’s similar for Black constituents: they comprise 13% of the working-age population, but only 2.1% of startup executives, making them 82% underrepresented.
These numbers drop even further as we move to the highest echelons of Fortune 500 firms. As of 2018, there were only 3 black CEOs and only 1 out of 5 board members is non-White. This finding is particularly surprising because studies have found that improving ethnic diversity increases EBIT, a measure of corporate profitability.
No, we do not believe this is an educational pipeline problem
As mentioned, there has been a sharp increase in the number of Latinx and Black students earning post-secondary degrees (Figures 5 and 6). The data for master’s degrees is, even more, telling (Figure 7). From 1980 to 2016, the share of master’s degrees awarded to White students fell from 89% to 66%. At the same time, Latinx students grew from 2% to 10% and Black students grew from 6% to 14%.
The number of Latinx and Black students seeking all college degree types has grown, while the proportion of White students receiving similar accreditations has consistently decreased, yet the number of technology jobs going to Latinx and Black candidates has not grown.
As Jason Torres, Kauffman Fellow and healthcare angel investor, said: “If you do not have diversity early on, including on the founding team, then you will never have a diverse team. Diverse executives hire more diverse executives, so it’s critical to start early.”
There is enough diverse talent across the country— with the right education and experience—to make better hiring decisions. The “pipeline” is not the problem.
It’s time to put that myth to bed once and for all.
Part 2: Ethnically Diverse Startups Have Higher Returns, but Still Struggle with Access to Capital
In Part One above, we discuss the impact of the perceived ethnicity of the people behind startups, including the founders and executives. We found that Latinx and Black executives are woefully underrepresented in startup leadership.
Furthermore, we conclude that this underrepresentation is not a pipeline issue, given that Latinx and Black students have increased their educational attainment by 350% and 55%, respectively, over the last four decades.
Here, we examine the composition of firms as a whole. We utilized the same public Crunchbase profiles as we did in part one, fed into publicly available demographic models by Claraifai to better understand these 260,000+ samples.
This report focuses on the median return multiples (RMs) of startups that were acquired or completed an initial public offering (IPO), so that these findings are not influenced by failures or bankruptcies.
We evaluated tens of thousands of diverse teams. Our findings, along with the findings of other studies, show that ethnically diverse teams outperform homogeneous teams. Our findings are grouped according to the following definitions:
- White Founding Teams: Startup founding teams that only have founders with a perceived ethnicity of White
- Diverse Founding Teams: Startup founding teams that have at least one founder with a non-White perceived ethnicity
- White Executive Teams: Startup executive teams that only have executives (directors, vice presidents, or C-level execs) with a perceived ethnicity of White
- Diverse Executive Teams: Startup executive teams that have at least one executive (director, vice president, or C-level exec) with a non-White perceived ethnicity
- Perceived ethnicity has a large impact on fundraising, with White Founding Teams and White Executive Teams raising venture capital more often.
- However, when Diverse Founding Teams do successfully raise capital, they tend to raise more across all rounds (Seed – Series E). For example, Diverse Founding Teams raised 60% more venture funding than White Founding Teams in late-stage (Series E) private rounds.
- Diverse Founding Teams have higher returns when cash is returned to investors. Historically, they have earned a 3.26x median realized multiple on IPOs and acquisitions, compared to a 2.50x realized multiple for White Founding Teams, a 30% increase.
Ethnicity Across Startup Founders and C-Level Executives
It’s important to note the difference between the ethnicity of startup founders and startup executives. As mentioned above, an “executive” is someone holding a director, vice president, or C-level title. By contrast, a “founder” includes both solo founders and co-founders who turn an idea into a company. In this report, we discuss the role of founders as well as executives: both are important, so we set out to examine the differences between diversity at company formation and diversity in leadership roles as the company scales. The vast majority of startup founders (75%) and startup executives (81%) are White (Figure 8). Asians are the second most represented group of founders (18%) and executives (13%).
Companies with Diverse Perceived Ethnicity Have Better Financial Returns
The data are clear when it comes to venture capital: Diverse Founding Teams raise significantly more venture capital. As shown in Figure 9, the median Series B for Diverse Founding Teams was 17% larger than the median Series B for White Founding Teams ($12.9M vs. $11M). The difference becomes more pronounced by Series E: Diverse Founding Teams raise 60% more than White Founding Teams.
That said, Diverse Founding Teams have much more limited access to capital than White Founding Teams (Figure 10). Our data show that more than 75% of all rounds are raised by White Founding Teams. That means 3 out of 4 venture capital rounds are going to non- diverse teams, who will then go on and become repeat founders or venture partners if their companies are successful. In the later stages, the numbers are even worse: 80% of Series D and E rounds (4 out of 5) have all- White founders.
The data hold true for White Executive Teams, with 74% of Seed and 70% of Series A rounds being invested into teams with no diversity at the highest level (Figure 11). As a startup grows it has the opportunity to increase executive team diversity with new hires. That said, 55% of companies at the Series E stage still have no ethnic diversity in their C-suite (Figure 11 and 12).
When looking at acquisitions and IPOs, we find that Diverse Founding Teams have a slightly higher likelihood of completing an initial public offering (IPO) than acquisition: 21.7% versus 18.6% (Figure 13).
The data also show that Diverse Founding Teams generate higher median realized multiples (RMs) on Acquisitions and IPOs (Figure 14). Diverse Founding Teams returned 3.3x, while White Founding Teams returned 2.5x. The results are even more pronounced when looking at the perceived ethnicity of the executive team. Diverse Executive Teams returned 3.3x, while White Executive Teams only returned 2.0x. As mentioned above, we report realized multiples (RMs) only for successful startups that were acquired or went through the IPO process. We have not included companies that fail to reach a successful outcome.
Our Analysis Shows the Financial Harm of Bias in Startups
Our data show a clear difference between diverse and non-diverse teams. Ethnically diverse startup founding teams and executive teams both raise more money from and provide higher returns for investors.
These results are consistent with other published findings. A Boston Consulting Group (BCG) report found that companies with diverse executive teams had 19% more revenue than their non-diverse peers. A recent McKinsey study also found that ethnically and culturally diverse boards of directors were “43% more likely to experience higher profits.”
Previously published reports show that firms with teams having high racial and ethnic diversity are “35 percent more likely to have financial returns” above the industry median. In our opinion, diverse teams outperform other teams because they have broader viewpoints, have stronger networks, and more closely match their target customers. Specifically, according to the U.S. Census, the United States is roughly 77% White, 18% Latinx, 13% Black, and 6% Asian. It only makes sense to have startup leadership more closely align with those demographics.
However, we also find an alarming difference in access to capital between White and non-White founders. For example, 77% of all Seed and Series A rounds were raised by startups with zero ethnically diverse founders (Figure 10).
“If you’re looking to invest in the top 2-3% of companies, the pipeline doesn’t matter as much as selection bias,” said Miriam Rivera, co-founder and managing director of Ulu Ventures and Kauffman Fellow. “How can you not find diverse candidates and diverse founders when top universities have been accepting diverse cohorts for the last 30-40 years?”
This data, combined with the education findings from Part One, point to a clear selection bias. As in other professions, venture capitalists invest in people that look like themselves, have similar views, and have similar upbringings. The data also show that despite the lack of access to capital, diverse teams—when successful in closing a round—raise more money per round (Figures 9 and 12).
For example, White Founding Teams closed 81.4% of all Series E financing rounds while Diverse Founding Teams make up the remaining 18.6%. Despite having a fraction of the access, Diverse Founding Teams raised 60% more in their rounds (median $40M per Series E financing for diverse teams compared to $25M per Series E financing for all-White founding teams; see Figure 9).
We need to do a better job as an industry to increase diversity in early-stage venture funds.
As mentioned earlier, people invest in others who are similar to themselves (the “selection bias problem”). Today, 80% of investment partners at venture capital firms are White, compared with only 3% Latinx and 3% Black (NVCA). This helps explain why 77-80% of venture rounds go to companies with White Founding Teams. In our opinion, there is a strong disconnect between the people writing the checks and those who are working in the trenches to incubate and start companies.
We need ethnically diverse venture capital partners who understand the nuances of different industries, products, cultures, and problems. We believe that there is a moral argument and even a moral necessity to do so—and there is a clear profit argument to be made as well.
This research was funded by the Ewing Marion Kauffman Foundation. The contents of this publication are solely the responsibility of Grantee.
All the data used in the development of all charts and insights were obtained from Crunchbase private-market data. Specifications of the analysis include the following.
• Venture-backed rounds for private firms between 01 January 2000 to 31 December 2018 were included.
• All executives whose profile pictures were available on the Crunchbase dataset were analyzed using the Clarifai publicly available API. If the profile image of a person had a logo, multiple faces, or had a low-resolution image, they were skipped.
• The perceived ethnicity of an individual was determined using the flowchart described in Figure 1. Classification of images was accomplished using the publicly available pre-trained demographic model on the Clarifai platform, which is also owned by Clarifai. If the person’s perceived ethnicity could not be conclusively determined using the rules defined in the flowchart, then they were marked “unknown” and not included in our analysis.
• Ethnic groups are all perceived ethnicity
• Executives with “founder” (case insensitive) in the titles were tagged as ‘Firm Founders’.
• Executives with titles including “CEO”, “CTO”, “CFO”, “CIO”, “chief”, “COO”, “CRO”, “president”, “co-owner” and “CDO” (all titles are case insensitive) were marked as ‘C-level Executives’.
• The startup industry classification was based on the segment description listed in the dataset.
• College degrees were manually curated and classified based on the number of years and requirements for the degree. For people with multiple degrees, the above degrees were ranked and the top degrees are selected.
• The classifications of degrees used in this study include the following.
• Undergraduate degree or less:
containing degrees like High school, Certificate, Diploma, Bachelor’s
• Graduate degree or higher: containing degrees such as MBA, Master’s, Doctorate/Doctoral, Physician, Law,
• Education subjects, like degrees, were also manually curated and classified into the large buckets of “Engineering,” “Business,” “Science,” or “Other,” based on subject type listed in the dataset. Subjects are selected for the top degrees for each person. Examples are listed below.
• Engineering – Computer Engineering, Chemical Engineering, Information Technology, Mechanical Engineering
• Business – Business Administration, Marketing, Business, Business Management, International Business, Finance
• Science – Genetics, Physics, Mathematics, Psychology, Biochemistry, Medicine, Molecular Biology, Pharmacy, Neuroscience • Firms with at least one founder with a perceived ethnicity other than White are referred to as Diverse Founding Teams … All other firms were marked as White Founding Teams.
• Firms with at least one C-level executive with a perceived ethnicity other than White are referred to as Diverse Executive Teams. All other firms are included as White Executive Teams.
• Startup returns were calculated only for firms with positive exits (acquired or IPO). For these firms, the Realized Multiple (RM) was calculated using the money raised and the post-valuation including the dilution that occurs in each round.
Disclaimer for data: “Crunchbase’s dataset is constantly expanding, but there are gaps. A company may not have all the founders and chiefs listed on its Crunchbase profile or have the complete details regarding the executive’s education listed in detail.”