Introduction

This NBER Working Paper examines how state-level reforms that weaken non-disclosure agreements (NDAs) in cases of workplace sexual harassment affect the hiring and organisational decisions of venture-capital-backed (VC-backed) startups. Using a staggered difference-in-differences design and LinkedIn data on over 50,000 U.S. VC-backed startups from 2014–2022, the paper finds NDA reforms reduce female hiring by approximately 8%, concentrated among junior women and in small or male-dominated startups (p.2).

Sexual harassment and NDA reforms

A 2019 national survey reports that 81% of women and 43% of men have experienced sexual harassment or assault during their lifetime, and 38% of women and 14% of men have experienced it at work (p.7). Women filed 78.2% of the 27,291 harassment charges submitted to the EEOC over 2018–2021 (p.7). Before the reforms, 85% of victims never filed a formal legal charge and approximately 70% did not report incidents internally (p.4). Between 2018 and 2022, fourteen states enacted legislation restricting or voiding NDAs related to sexual harassment, culminating in the federal Speak Out Act of 2022 (p.9).

Sample construction and summary statistics

The final sample contains 52,607 unique VC-backed startups and 207,469 startup-year observations (p.12). About 59% of startup-year observations are in the fourteen states that enacted NDA-weakening laws (p.12). The average startup hires 2.93 workers per year, of whom 1.12 (34%) are women (p.12).

Female hiring after NDA-weakening laws

Startups in treated states hire 7.8% fewer women per year relative to control startups (p.14). The female share of hires drops by 1.2 percentage points — a 3.5% decline relative to the sample average of 34% — while male hiring shows no meaningful change (p.15).

On the extensive margin, treated startups are 2.8 percentage points less likely to hire their first female worker within one year of enactment, rising to 4.9 percentage points within four years (p.20). The decline concentrates among junior women — treated startups hire roughly 10% fewer junior women per year — while senior female hiring remains largely unchanged (p.22). Small startups hire 12.4% fewer women per year compared with 4.5% for large startups (p.22), and those with low pre-existing female representation show an annual reduction of 10.9%, while female-friendly startups show virtually no effect (p.23). The 2022 federal Speak Out Act produced a comparable reduction of approximately 14–15% in previously untreated states, supporting a demand-side interpretation (p.24).

Organisational changes after NDA-weakening laws

Treated startups promote more women to managerial positions, with the female share of promotions to management rising by approximately 3 percentage points (p.27). The male share of managerial departures rises by 2.7 percentage points (p.28). Departing male managers transition to lower-salary jobs (up 3.6 percentage points), lower-seniority positions (up 3.9 percentage points), and experience longer unemployment spells (up 5.1 percentage points) (p.28), suggesting many separations are involuntary. NDA-weakening laws also lower the probability of receiving any VC funding by 2.8 percentage points (a 9% decline) and reduce the number of financing rounds by roughly 12% (p.30).

Conclusions

While NDA-weakening laws reduce female hiring in the short run, they also trigger internal restructuring — promoting women into leadership and removing potential harassers from management. The findings suggest the policy problem lies not in transparency itself, but in the absence of complementary firm-level compliance infrastructure. Whether these organisational changes eventually translate into higher female hiring remains an open question.