Drawing on Paul Gompers' research covering 14,000 VC investments in 42,000 startups, this piece documents how homogenous venture capital is and what it costs. Women held steady at 8% of VCs from 1990 to 2016 despite rising women in STEM and business degrees, while Hispanic and Black representation sat at roughly 2% and 1% respectively. Partners from the same school had an 11.5% lower success rate on acquisitions and IPOs, and ethnically homogenous partnerships saw success rates 26 to 32% lower. Using a natural experiment based on partners having daughters, which raised female hiring 24%, Gompers found funds with at least one female partner returned 16 to 17% versus a 14 to 15% median, plus exit rates of 31% versus 28%.

This piece is useful precisely because it moves diversity in venture capital from a values argument to a performance argument backed by a genuine causal identification strategy, since simple correlations between diversity and returns could always be explained away by confounding factors that quietly skew such data. The daughters instrumental variable is a specific, teachable example of how to isolate a causal effect in a setting where randomized experiments are impossible, which makes this a useful reference for anyone studying applied causal inference as well as anyone making the business case for diversity to skeptical partners. LPs evaluating a fund's partner composition, and GPs building a new firm, both get concrete performance numbers rather than just an appeal to fairness alone.