Entrepreneurship fuels job creation, innovation, and wealth—but women are sharply underrepresented among startup founders. Less than 20% of venture-capital-backed companies have even one woman on the founding team. The gender gap in entrepreneurship is persistent and problematic, and noncompetes widen that gap. This may seem unexpected, as neither noncompete contracts nor laws say anything about gender. But noncompetes make entrepreneurship more costly, more risky, and more penalizing for women. I studied this gender gap using confidential data from the Census and IRS, which lets us examine the entire population.

First, women start companies earlier in their careers than men, and have been paid less than men despite working in the same industry. This leaves them more financially vulnerable to a noncompete lawsuit, as the same per-hour attorney fee is more costly on a relative basis for someone with less cash in the bank.

Second, most new ventures fail. Those that succeed tend to be founded by those with experience in the same industry. But that is exactly what noncompetes block: using the skills and expertise you previously developed. Therefore, the risk of startup failure is higher when subject to a noncompete. Several studies in economics and psychology suggest that women tend to be more risk-averse than men. Because noncompetes add legal risk to the inherent business risk of entrepreneurship, these contracts discourage women from starting companies that draw on their skills and experience.

Third, even if someone decides to found a company, noncompetes make it harder to hire talent with relevant skills. Startups in California can draw on the full talent pool in the state, but a startup in Florida or Massachusetts must carefully consider whom they can hire without risking a lawsuit. This inability to marshal human capital reduces the chances of success. We know from many studies that women suffer stiffer penalties for failure. This is also true for failed entrepreneurs: Women who abandon their startups and return to employment at an existing company are paid less than men whose startups do not succeed.  

For all of these reasons, noncompetes curb the enthusiasm of women who want to found a startup but fear a lawsuit, either against themselves or the workers they try to hire. Perhaps not surprising is that female founders are less likely to put their former coworkers at risk of a lawsuit by hiring them in states where noncompetes are more tightly enforced.

Again, there is sure to be stern opposition to the ruling from existing companies who do not want their employees to leave and either join or start rival firms. But, as the FTC report states, there are several other tools including non-disclosure agreements and trade-secret litigation which can protect confidential information.

Noncompetes are a blunt instrument that favors incumbents over startups, as well as men over women. Implementing the FTC noncompete ban will serve to narrow the gender gap in entrepreneurship and usher in the wave of startup activity that policymakers worldwide seek. This is one part of California’s Silicon Valley that is easy to copy.

Matt Marx is the Bruce F. Failing, Sr. Professor and Faculty Director of Entrepreneurship at Cornell University, who has researched noncompete agreements in the workplace, finding that they disproportionately affect women.

More must-read commentary published by Fortune:

• I’m a venture capitalist, and here’s why I believe we need to guarantee everyone’s basic needs: The social floor is actually a trampoline that can propel our economy
• Gen Zers and millennials are switching jobs at an accelerating pace, and it’s paying off. Here’s where it can still go wrong
• $122 Thai delivery and $26 to-go coffees: New wage laws meant to help gig workers are backfiring big-time
• Second Chance Month is a failed promise for former inmates. Here’s how we can fix that

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.