The absence of women in the ranks of high-growth start-up entrepreneurs in the US is striking. According to Gompers and Wang (2017), only 10-15% of venture capital (VC)-backed start-up founders are women. The gap is smaller for entrepreneurs more generally; for example, 28% of incorporated entrepreneurs are women (Levine and Rubinstein 2017).
The gender gap in high-growth entrepreneurship has many causes (Ewens and Townsend 2020). But there is much anecdotal evidence that women face particular hurdles in raising VC funding, which is crucial for helping many early-stage, risky ideas to get off the ground. One survey suggested that 70% of female founders felt their gender negatively affected their ability to fundraise.1
In a recent paper, we examine a particular channel that might explain the gap: networking frictions (Howell and Nanda 2020). It is widely known that face-to-face connections and trusted referrals are important – if not primary – deal-sourcing methods for many top VC investors. Because more than 90% of VC investors are men (Gompers and Wang 2017), homophily in networking could disproportionately impact women if entrepreneurs do not get as many chances to interact with VCs of their own gender without an introduction.
On one hand, it is possible that exposing early-stage female entrepreneurs to more networking opportunities with VCs could help reduce the gender gap. That is, exposure to more VCs could help compensate for any baseline advantages that men have in meeting VCs. On the other hand, merely creating opportunities for connections may be insufficient if networking frictions prevent women from taking advantage of the exposure.
The Harvard Business School’s new venture challenge as a natural experiment
Rigorously studying network-related information frictions presents certain challenges. Social networks are endogenous, making it hard to separate the role of networking frictions from unobserved variables, such as whether an entrepreneur’s business model is a good fit with VC. We address this empirical challenge by employing exogenous variation in exposure to VC networks at Harvard Business School’s (HBS) New Venture Competition (NVC), which is a key gateway to VC-backed entrepreneurship after HBS.
In the first round of the NVC, each team is assigned to one of about 15 panels, each composed of about six judges. Having delivered a pitch to the judges and answered their questions, the participants are in a position to reach out to the judges after the competition, leveraging the connection to secure financing for their ventures. The competition does not, however, explicitly encourage such follow-up. The core dataset for our analysis consists of comprehensive team and judging information for HBS NVCs held between 2000 and 2015, comprising 964 participants.
Our research design exploits random variations in the number of VC judges across panels, which arises from the way judges are allocated to panels in the NVC. Specifically, assignment is random conditional on sector, which is to some degree taken into account. Our core empirical specification estimates the differential effect of having an additional VC judge on the panel for a male participant’s subsequent chances of VC-backed entrepreneurship relative to a female participant’s chances of VC-backed entrepreneurship.
Does having more VCs on a panel benefit women?
If lack of exposure to VCs among female participants were an important friction, we would expect to find, on the margin, that exposure to an additional VC on a panel would be more valuable to female participants. Instead, we find the reverse. The raw relationship shown in Figure 1 indicates that exposure to more VCs substantially benefit male participants, while this relationship is at most only marginally positive for women.
Figure 1 Relationship between number of VCs on panel and VC-backed entrepreneurship
Regression estimates find that relative to the effect among women, each additional VC judge on a panel increases the likelihood of men starting a VC-backed company after HBS by 25%. Placebo tests show that the result is not present among participants starting ventures with no VC financing or among participants joining VC-backed start-ups as employees. Also, judges on the panel in the same sector as the participant or with backgrounds besides VC – such as corporate executives, lawyers, or academics – have no differential effect by gender on VC-backed entrepreneurship (nor do they have an independent effect).
The results are robust to controlling for individual characteristics such as college major, stated interest in entrepreneurship when entering HBS, and entrepreneurship experience prior to attending HBS. Such covariates are known to be correlated with VC-backed entrepreneurship but are typically unobserved. Our main estimates also include a variety of competition controls, such as the number of ventures on the panel, the score the venture received, and whether the venture won. The results are further robust to including venture sector fixed effects, gender by sector fixed effects, gender by year fixed effects, and panel fixed effects. These controls are useful for precision and for establishing the robustness of our research design.
Why might exposure to VCs compound the gender gap rather than reduce it?
To shed light on the mechanism, we rely on survey evidence from NVC participants. The survey responses reveal that male participants are nearly twice as likely as women to proactively reach out to VC investors after the NVC. Conditional on reaching out, however, men and women report no difference in the rate at which VCs respond to their outreach or any difference in the degree to which VCs proactively reach out to them.
Each individual outreach to an investor by an entrepreneur has only a small chance of being ultimately valuable in terms of raising VC, but a fixed difference in the likelihood of outreach between men and women compounds across the number of VCs on a panel, making the difference in the benefit larger for panels with more VCs than for those with fewer VCs. Consistent with this, qualitative responses in the survey also suggest that because follow-up with judges at the NVC was not explicitly encouraged, women were more likely to have some reservations about leveraging the connection to discuss fundraising.
We do not find obvious evidence of explicit bias by male VCs against female participants in our sample. In addition to VCs responding to outreach equally by participant gender, we show that the private scores of VC judges are in fact slightly lower for male-led ventures than for women-led ventures. Also, while there are too few female VC judges to establish precise effects by a judge’s gender, we do not find evidence that female VC judges are differentially beneficial for female participants. Still, less observable discrimination may be at play, and the lack of outreach by women could reflect expectations of bias or harassment.
Implications of our results for policy
While our analysis is based on a sample that is quite particular, we believe the results have broader significance. If networking frictions matter in this already highly ambitious and well-connected sample of potential entrepreneurs, it seems likely that they will matter more generally in VC, contributing to the overall gender gap in VC-backed entrepreneurship.
From an economic perspective, the gender gap in VC-backed entrepreneurship is particularly worrying if it reflects systematic gender-related frictions that can cause high-quality entrepreneurs or ideas to go unfunded. VC is a crucial financing source for new ideas and technologies (Kaplan and Lerner 2010). Yet a relatively small number of VC firms and their investing partners account for a disproportionate share of the capital that VCs deploy. Frictions in the process through which these gatekeepers select and advance new ideas can have consequential effects on the types of ideas that are commercialised in the economy (Kerr et al. 2014).
Beyond shedding light on how the gender gap emerges early in the start-up lifecycle, our results have implications for new venture competitions and accelerators, which can design networking between entrepreneurs and investors in ways that facilitate the financing of the best (rather than simply the best-networked) ideas. Structural solutions that focus only on providing female entrepreneurs more exposure to VCs may not be enough to eliminate networking frictions. Instead, our results point to benefits from encouraging and potentially formalising networking opportunities between individuals, rather than assuming that people will contact each other independently.
Ewens, M and R Townsend (2020), “Are Early Stage Investors Biased Against Women?”, Journal of Financial Economics (forthcoming).
Gompers, P A and S Q Wang (2017), “Diversity in innovation”, NBER Working Paper (23082).
Howell, S T and R Nanda (2020), “Networking frictions in venture capital, and the gender gap in entrepreneurship”, NBER Working Paper 26449.
Kaplan, S N and J Lerner (2010), “It ain’t broke: The past, present, and future of venture capital”, Journal of Applied Corporate Finance 22(2): 36–47.
Kerr, W R, R Nanda and M Rhodes-Kropf (2014), “Entrepreneurship as experimentation”, The Journal of Economic Perspectives 28(3): 25–48.
Levine, R and Y Rubinstein (2017), “Smart and illicit: who becomes an entrepreneur and do they earn more?”, The Quarterly Journal of Economics 132(2): 963–1018.