Where Women's Patents Disappear

Women make up 35% of the U.S. STEM workforce. They are 12.8% of U.S. patent inventors. The 22-point gap does not appear all at once.

It opens at filing, widens at first office action, widens again at response, again at appeal, again at grant, and again at post-grant citation. Each stage individually looks small. The product is not small.

This page walks the prosecution funnel one stage at a time. At each stage we show the published data on the disparity, where it comes from, and what is known about why it persists. The cumulative picture is the patent gender gap as it actually exists, not as a single statistic.

This article presents published research data for educational purposes. It is not legal, policy, or investment advice. Patent gender statistics depend on name-based inference and methodology choices: see How we measure.

The funnel at a glance

Each row below is a stage at which the gender gap measurably widens. We expand each in its own section after the table.

1. Filing: the gap before the system sees the invention

The first leak in the funnel is not a USPTO decision at all. It is a decision made inside companies, universities, and labs about whose work gets written into a disclosure form and forwarded to counsel.

The USPTO's Progress and Potential 2020 update found that women were 12.8% of U.S. inventor-patentees on 2019 grants, while women were about 35% of the U.S. STEM workforce in 2023 (27% of the narrower science-and-engineering occupations). These are non-equivalent measures, different years, populations, and units, and grant-based inventorship is not a filing rate, so treat the roughly 22-point difference as a representation contrast, not a measured gap "at filing." It shows substantial underrepresentation in patent inventorship; it does not by itself locate the stage where the difference arises.

Structural causes are well documented. University tech transfer offices report consistently lower invention disclosure rates from women researchers, even when controlled for research output and field. Structural changes to the disclosure process are a promising response: after Washington University's WIT program introduced interventions of this kind (including pre-populated disclosure forms that list all lab members rather than asking the PI to nominate inventors), patent filings on behalf of women faculty rose 129% between 2011-13 and 2013-16. It is a single-institution before/after result, so it points to the approach rather than proving which component drove it.

2. First office action: the rejection-rate gap

Once an application is filed, the first signal from the USPTO is the first office action. A 2025 study published in Scientific Reports (Schuster & Goodman) analyzed millions of U.S. patent applications and found that applications listing women inventors have less favorable aggregate outcomes at several prosecution stages than otherwise comparable applications.

The differences persist after adjustment for many observed characteristics, and are smaller for applications filed through universities. This is an observational, stage-by-stage analysis: it does not compare the same invention under different inventor identities, but it is consistent with institutional resources and support mattering.

Pairolero et al. (American Economic Journal: Economic Policy, 2025) ran a randomized controlled trial at the USPTO among applicants without legal representation. Enhanced assistance improved patent-obtainment for both men and women, with the estimated treatment effect more than 12 percentage points larger for women, largely through successful examiner-amendment negotiations. It shows assistance can narrow the gap in this pro se setting; it does not establish the same effect for represented corporate applicants.

Whether examination itself contributes to the gap is genuinely contested, and the honest version says so. Gu, Lewellen, Mao and Qin (working paper, November 2025 version; 988,125 U.S. applications, 2001-2017) study the examiner's first-action decision. Immediate first-action allowance is uncommon (about 7.5% of applications), and women's applications are about 0.4 to 0.6 percentage points less likely to be allowed at first action after controls, a 5 to 8% relative decline. Examining solo inventors with common versus rare first names (a proxy for how readily an examiner might associate a name with gender), they find that gap concentrated among common-name inventors and close to zero among rare-name inventors, and they read the pattern as statistical discrimination by examiners. That reading is suggestive, but the rare-name test is an observational comparison, not a blinded experiment: no name was masked, no application examined twice, and rare-name inventors may differ in other ways.

A separate working paper (Avivi, 2025) reaches a more qualified result: controlling for application content with patent-text embeddings, it finds no average content-adjusted first-action difference between mixed-gender and all-men teams, while finding meaningful variation across individual examiners. Read the two together: whether there is a current, system-wide average first-action examiner effect, and how large, is unsettled.

3. Response and abandonment: where most of the gap actually lives

This is the largest single leak in the funnel.

Aneja, Reshef and Subramani ("Attrition and the Gender Patenting Gap," Review of Economics and Statistics) found that applications listing women inventors are 3.6 to 6.9 percentage points less likely to be continued after an early rejection. "Abandonment" here is an application outcome, not a personal decision by the inventor: an employer, assignee, tech-transfer office, founder, or counsel may control whether prosecution continues.

The same study estimated that differential abandonment explains more than half of the issued-patent gender gap within its sample, one of the largest measured contributors inside the post-filing process, though not a claim about every stage of the wider innovation system. The applications are being filed and reaching examination; many are simply not being continued through to grant at the same rate.

The UK IPO's 2026 Female Participation in Inventorship report (Brady Bowes, March 5, 2026) corroborates the same pattern in a different jurisdiction with cleaner team-level granularity.

4. Appeal: the additional 2.5% gap

For the subset of applications that reach a final rejection, the next decision point is whether to appeal to the Patent Trial and Appeal Board (PTAB). Jensen and colleagues at Yale found that women are approximately 2.5% less likely than men to file a PTAB appeal at this stage, controlling for technology and entity type.

The effect is small in absolute terms but it compounds. A final rejection followed by no appeal is a permanent loss. Several appeal-rate studies have proposed that the gap reflects both cost sensitivity (appeals are expensive) and confidence factors that interact with broader patterns in how inventors of different backgrounds engage with legal systems.

5. Grant and claim scope: even the patents that issue look different

For applications that survive to grant, the gap shifts from quantity to quality. The 2025 Scientific Reports analysis found that granted patents with female inventors have narrower independent claims on average than otherwise comparable patents with male inventors. Independent claim scope is an important determinant of a patent's enforcement value: narrower claims are generally easier to design around and harder to license, though commercial value also depends on enforceability, market coverage, validity, and remaining term.

The narrower-claim effect operates through prosecution: when examination is harder (Stage 2) and abandonment after rejection is higher (Stage 3), the remaining path to grant often runs through claim amendments that narrow scope. The patents that issue have, on average, been negotiated into a smaller commercial footprint to get there.

On ultimate grant, Gu, Lewellen, Mao and Qin report that all-women inventor teams are about 20.9 percentage points less likely to have a patent eventually granted than all-men teams (against a full-sample grant rate of 68.5%). Much of that is associated with technology and other observed characteristics: after narrow technology and application, inventor, applicant, and examiner controls, the gap shrinks to a residual 5.4 percentage points. That is an adjusted association, not a causal decomposition, and examiner decisions, prosecution support, applicant choices, and unobserved differences may all contribute.

Earlier peer-reviewed work is consistent. Jensen, Kovács and Sorenson (Nature Biotechnology, 2018; about 2.7 million applications) found less favorable acceptance and maintenance outcomes for women inventors, with the estimated acceptance disadvantage among solo inventors substantially smaller for rare first names (about 2.8%) than common ones (about 8.2%). Because names were not experimentally concealed, this supports, but does not prove, a gender-visibility mechanism.

Taken together, the lower grant rate and narrower granted claims mean that, on average, applications listing women inventors convert to less valuable patent assets after adjustment for many observed characteristics. The disclosure-to-asset conversion differs by inventor gender, though no single study isolates one cause.

6. Post-grant: citation, licensing, and the breakthrough penalty

Even after grant, the funnel is not done.

Jensen et al. found that patents whose inventors carry common female first names are cited approximately 30% less than otherwise comparable patents. The methodology is robust to several controls and points at a citation-side disparity that is independent of the underlying patent's technical content.

A 2026 study in PNAS documents the "breakthrough penalty" we name elsewhere in this series: applications from women-majority inventor teams that combine ideas in unconventional ways grant at substantially lower rates than otherwise comparable applications from male-majority teams. The grant-rate gap on unconventional applications reaches 14.9 percentage points. The most novel inventions from under-represented inventor teams are precisely the ones the patent system rewards least.

This stage matters commercially because forward citations and licensing revenue are how patents generate value over their 20-year term. A grant is not the end of the funnel. It is the beginning of the asset's commercial life. If the post-grant trajectory is shorter and quieter for women-involved patents, the cumulative gender gap in patent-generated wealth is larger than the gap in patent counts.

The cumulative arithmetic

Each individual stage looks tractable. A representation contrast here, a persistence gap there, a citation differential at the end. Considered separately, each one feels like a research finding that could be addressed with a focused intervention.

Considered together, the picture is sobering: disadvantages appear at more than one stage, and losses that survive one stage carry into the next. A caution on reading it, though. These estimates come from different studies, samples, years, units, and models, so they cannot literally be added or multiplied into one cumulative loss figure. Treat the funnel as an illustrative evidence map, not a single numerical cohort flow.

This is why interventions at a single stage rarely move the headline number on their own. Closing one leak is necessary but not sufficient. The interventions with the clearest measured effects (the WashU WIT program, the USPTO RCT, structural disclosure-form changes) tend to operate on the earliest stages and on the response-and-continuation stage.

What this means for organizations

For tech transfer offices, corporate IP teams, and patent counsel: the most consequential decisions about whose inventions become patents are made before the USPTO ever sees the application. Pre-populated disclosure forms, proactive outreach to under-represented researchers, and funded prosecution support after a first rejection are the three interventions with measured results.

For inventors: the difference between an abandoned application and a granted patent is most often persistence through the response cycle. Knowing the data on the response-and-abandonment gap is itself a small intervention.

For everyone else: the patent system was designed to reward inventors. Its current operation systematically rewards some inventors less than others for the same work. That is a structural problem, not an individual one. The data above is published. The interventions that close the gap are documented. The question is not what to do. It is whether to do it.

Sources

More from the patent gender gap series