After the last post’s surprising conclusion, this post dives into Are Early Stage Investors Biased Against Women? by Michael Ewensa and Richard R. Townsend, professors at two Californian schools. You can download the paper here to follow along.
Ewensa and Townsend take AngelList data and subject it to rigorous analysis to pinpoint gender bias. As the abstract summarizes: We find that male investors express less interest in female entrepreneurs compared to observably similar male entrepreneurs. And: Overall, the evidence is consistent with gender biases.
After discussing some introductory material pondering why women make up only 10-15% of startup founders, and covering some related literature, the article gets down to business with its AngelList data. We’ll assume for this post you are familiar with AngelList.
The paper looked at interactions within AngelList, specifically (i) “sharing” a startup’s profile with another person; (ii) requesting an intro to the startup founder who is raising capital, and (iii) investing in a startup (which, at the time, happened outside of the AngelList platform). The first two are measured from AL administrative data; funding is reported by the entrepreneurs. The sample is 17,780 companies, all with the first fundraising event from 2010 to 11/2015.
What did the paper find?
First that 15.8% of firms have a female CEO, 20.9% of firms have a female founder, and that 8% of the investors who do some sharing are female. Those data seem fairly consistent with other sources in the paper (Crunchbase) and external ones (e.g. a bit below the 2019-2022 levels from surveys of angel investing in the prior post, but similar). Male-led and female-led companies had similar characteristics (e.g. claiming traction, incubator graduation), though male-led companies sought notably higher amounts of funding; and male and female entrepreneurs had similar characteristics (e.g. age, work experience, educational attainment), though males had more prior founder experience.
The significant disparity comes when the data shows that men are more successful than women in terms of generating interest (p.17). More “being shared” (5% vs 3%), more often being asked for an intro (19% vs 16%), and more often funded (2.5% vs 1.5%). On all measures, therefore, female-led companies struggle vs “observably similar” male-led companies.
(Female investors reverse these behaviors, but male investors outnumber female so the offset is only partial.)
Do these results indicate gender bias? The paper considers two sets of explanations – ones where investors have gender bias, and ones where they do not. From a detailed review of potential ideas in both sets, and testing those ideas against the startups’ characteristics and performance data, the paper finds all the “do not” ideas unsupported and the “do” ideas plausible. Male investors seem particularly to disregard female entrepreneurs’ incubator affiliation credentials, give less weight to the claims of traction, and to “pigeonhole” women into female-centric industries.
One final analysis (section 6.2) suggests this bias reduces the more experienced the angel investors become – suggesting some of the bias is from “miscalibrated beliefs” that get better calibrated with experience, rather than “taste”.
Do these findings seem convincing? We’ll leave it to you to decide, but personally (this is Paul!) I’d note some issues but overall I’m convinced.
There are, naturally, objections to the data: AngelList is not a representative sample of early stage investors; the “outcome data” is from outside information not AngelList, and so (like other sources of information about funding early stage companies) likely poor; and 2010-2015 (with early adopters to AngelList and the dynamics of exits in that period) is perhaps not representative of today’s investing activity. We could also object about the “proof of outcome” – particularly that you can deduce a company failed based on whether its website is active (p.31) and whether “raised a follow-on round from a VC” is really a good “interim measure of startup success” (p.31).
Still, the conclusion that “our results appear consistent with some form of bias among male investors” (p.37) seem justified to us. It’s a clear reminder that, to achieve our goal of the best financial outcomes we can, investors need to stay aware of our biases, and to consider particularly those areas (like evaluating traction, which is subjective but could be made more quantitative) where those biases seem most concentrated.
In addition, a couple of unrelated fun nuggets stood out to me:
- The paper was based on data from before syndicates had taken off on AL, and so couldn’t really see into funding decisions, as those were done “offline”. Would repeating this investigation today with real AngelList investment data reinforce or undermine these findings?
- Only 2% of female angel investors in AngelList were associated with a known female-focused angel group of VC fund (p.21). I wonder if the same is true for male investors? And does this show the “market share” of explicit funds vs the general AngelList pool? If so, it’s interesting to see how much AngelList has done to “expand the market” of angel investors.