Are Early Stage Investors Biased Against Women? Ewens & Townsend University of North Carolina at Chapel Hill & NBER NBER Entrepreneurship Working Group Meeting, December 2017 Discussion: Are Early Stage Investors Biased Against Women? 1-18
What does this paper do? Uses a proprietary dataset from AngelList covering fundraising startups with investor information Major advantage: observe a large set of startups that are trying to raise capital-so see pool of submitted & "accepted" projects. Observe investor actions such as "shares" "requests for introductions" "investment" Find that controlling for a battery of founder/startup characteristics: 1 Male investors are less likely to share, request introductions or fund female-founded startups. 2 Startups with female founders garner significantly more interest with female investors. 3 Suggest that evidence is consistent with taste-based discrimination and/or homophilistic preferences. Discussion: Are Early Stage Investors Biased Against Women? 2-18
Roadmap for Discussion 1 Estimating Equation: Alternative Specifications 2 Complementary Evidence: Gender Representation & the NBER s Summer Institute 3 Alternative early stage funding environment where gender may play a role: SBIR & STTR grants Discussion: Are Early Stage Investors Biased Against Women? 3-18
Identification From an investor perspective: where: ROI i = f (X, Z) (1) X= A set of observables (productivity characteristics) such as founder characteristics (education, experience) & project characteristics (sector, location.) Z= A set of unobservables such as founder & project quality, investor risk-tolerance, etc.. Discussion: Are Early Stage Investors Biased Against Women? 4-18
Identification Challenge Are investors screening on unobservable investment characteristics rather than gender? Eg. Projects founded by women are less profitable or of lower quality. If correlation is known then statistical rather than taste based discrimination? If gender is not correlated with ROI and unobservables, then evidence that gender is correlated with the investment decision may be evidence of taste-based discrimination. Discussion: Are Early Stage Investors Biased Against Women? 5-18
Estimating Equation: Benchmark Specification y i = α + β Female i + δ X i + ɛ i (2) y: startup level outcomes (shares, intros, investment) by gender Female: Female Founder-CEO X=vector of start-up and founder-level controls Discussion: Are Early Stage Investors Biased Against Women? 6-18
Estimating Equation: Benchmark Specification y i = α + β Female i + δ X i + ɛ i (3) N=17,780 startups 13.3% (2,365) funded, 27% gender-identified Investors who fund & have identifiable gender ( 640 startups) & with multiple investors Dependent Variable: Male investors=1,637; Female investors=193; Unidentified gender =535? No interest=15,532? Discussion: Are Early Stage Investors Biased Against Women? 7-18
Estimating Equation: Benchmark Specification y i = α + β Female i + δ X i + ɛ i (4) Coding the dependent variable: 1 Male Interest=1 2 Female Interest=0 3 Unidentified male and female Interest=0 4 No interest=0 Mapping multiple investors (1,830) to 640 startups. How? Dealing with start-ups that have male and female investor interest. How? Discussion: Are Early Stage Investors Biased Against Women? 8-18
Estimating Equation: Alternative Specifications Preserving panel structure and estimating at the start-up as unit of observation: y ij = α + β Female i + γ i + ɛ ij (5) Dependent variable: Continuous (Proportion of male interest), cross-sectional regression: y ij = α + β Female i + δ X i + ɛ ij (6) Discussion: Are Early Stage Investors Biased Against Women? 9-18
Estimating Equation: Alternative Specifications y i = α + β Female i + γ Female i Male Investor Dummy where: + δ Male Investor Dummy + ψ Missing Dummy + ɛ i (7) y i = Any interest. Female i =Controls for Quality? (β: Possibly statistical discrimination?) Male Investor Dummy=Controls for risk-tolerance, etc. (γ: Taste discrimination?) ψ: Selection? Is there something systematically different about observations with missing data? Discussion: Are Early Stage Investors Biased Against Women? 10-18
Chari and Goldsmith-Pinkham (2017) Document the representation of female economists on the conference programs at the NBER Summer Institute from 2001-2016. Average share of women authors rises from 18.5% (2001-2004) to 20.6% (2013-2016). Persistent gap between finance (14.4%), macroeconomics (16.3%) & microeconomics (25.9%) subfields remains. Conditional on submission rate of paper acceptance for women is statistically indistinguishable to that of men. Share of female authors comparable to the female tenure-track professors share, but is ten percentage points lower than the share of women assistant professors. Within conference program, we find that when a woman organizes the program, the share of female authors and discussants is higher. Homophilistic affiliation? Discussion: Are Early Stage Investors Biased Against Women? 11-18
Role of Organizer Gender Female Share it = α i + α t + Female Organizer it + ɛ it (8) (1) (2) (3) (4) (5) (6) Share Female Share Female Share Female Share Female Share Female Share Female Any Female Organizer 0.027 0.035 0.032 (0.020) (0.015) (0.016) Finance -0.040-0.039-0.022 (0.043) (0.007) (0.016) Micro -0.002 0.025 0.022 (0.029) (0.016) (0.017) Macro 0.067 0.072 0.064 (0.027) (0.026) (0.029) Observations 654 654 544 544 544 544 Year F.E. Yes Yes Yes Yes Yes Yes Program FE No Yes Yes No Yes Yes Field-Year FE No No Yes No No Yes Standard errors in parentheses p <.1, p <.05, p <.01 Discussion: Are Early Stage Investors Biased Against Women? 12-18
Variation across Programs Impulse and Prop. Mechanisms Asset Pricing Dynamic Equilibrium Models Monetary Economics Econ. Fluc. and Growth Capital Markets in the Economy Corporate Finance Real Estate Labor Studies Entrepreneurship Public Economics Crime Health Care Health Economics Children 0.0 0.2 0.4 Share of Female Authors Finance Macro/International Micro Discussion: Are Early Stage Investors Biased Against Women? 13-18
Alternative Settings: SBIR & STTR Grant Applications Discussion: Are Early Stage Investors Biased Against Women? 14-18
Alternative Settings: SBIR & STTR Grant Awards Discussion: Are Early Stage Investors Biased Against Women? 15-18
Alternative Settings: SBIR & STTR Grant Success Rates Discussion: Are Early Stage Investors Biased Against Women? 16-18
Alternative explanations unobservables (same startups garner more female investor interest, more favorable unobservables?) gender-specific screening/monitoring advantage (gender nuetral startups) different payoff distributions (high within-group correlation) non-financial considerations (male-male pairs underperform) Discussion: Are Early Stage Investors Biased Against Women? 17-18
Conclusion I really like this paper. Addresses an important question about potential gender bias in entrepreneurship. Takes on a difficult identification challenge. Encourage the authors to examine alternative specifications of their benchmark regression specification to make more definitive case for taste-based discrimination. Discussion: Are Early Stage Investors Biased Against Women? 18-18