The evidence · women & money
I spent a career in research before I built this platform, so I turned that same lens on women and money. Below is the evidence, drawn from primary sources and stated as honestly as the data allows, caveats included.
The short version: the gap is smaller than the headlines claim, it is not confined to poorer or less-educated women, and it is mostly about confidence, not knowledge.
This page is my working read of the literature as I go, not a settled review. I am still checking sources, so treat what you find here as thinking in progress rather than a finished conclusion.
No number appears here without a traceable primary source, and where a source has a commercial interest in the story, it is flagged. Each finding is stated as it is most defensibly stated, with the caveat the slogans tend to drop.
The confidence gap
Test women on money and they score lower. Look closer and a large share of that gap is women declining to answer, not women getting it wrong.
⅓
When researchers split the financial-literacy gender gap into its causes, about one third is explained by women’s lower confidence and roughly two thirds by an actual knowledge difference.
The caveatSo “the gap is just confidence” is wrong, and so is “women simply know less.” It is both, in known proportions.
SourceBucher-Koenen et al., “Fearless Woman,” Management Science (2024) ↗
−40%
On the standard three-question literacy test, women pick “don’t know” far more than men but give wrong answers at about the same rate. Force everyone to make a best guess and the gap in correct answers falls from roughly 20 points to 12, a drop of about 40%.
The caveatIt does not vanish. A real knowledge gap remains underneath the confidence effect.
SourceFederal Reserve, FEDS Notes (2024) ↗
½
Re-ask the same people six weeks later with the “don’t know” option removed and the three-question gap roughly halves. On the diversification question alone it fell from 27 points to 9.
The caveatThe single cleanest test of the confidence effect, run on Dutch panel data.
SourceBucher-Koenen et al., “Fearless Woman” (NBER, 2021) ↗
32% · 18%
Women are nearly twice as likely as men (32% versus 18%) to call their own investing knowledge “non-existent,” even as 71% of women now hold investments. Under-rating runs right alongside rising participation.
The caveatSelf-reported attitudes from an asset-manager survey.
SourceFidelity, Women & Investing Study (2024) ↗
53% · 43%
On the US Personal Finance Index, men answer 53% of the questions correctly and women 43%. The gap is widest on investing and risk, the topics where uncertainty invites a “don’t know.”
The caveatIt survives controls for income, education and marital status, so it is not just a demographics artefact.
SourceTIAA Institute & GFLEC, P-Fin Index (2023) ↗
The honest counterweight: on the OECD’s composite literacy score across 39 countries, the average gap is under two points. This is not a crisis of ability. It is a fixable gap in confidence. Source: OECD/INFE, 2023.
It is not just poverty
We tend to file this under a poor-woman or developing-country story. The data says otherwise: the gap reaches the wealthy and the highly educated, and in some cuts the youngest women most of all.
56%
Among married women with $250,000 or more in investable assets, 56% leave long-term financial decisions to their husbands, and 85% of them say it is because they assume he knows more. Millennial women defer more (61%), not less.
The caveatA wealth-manager survey of affluent married couples, heterosexual couples only.
SourceUBS, Own Your Worth (2018) ↗
~10% less
Female finance students, arguably the most financially educated women there are, showed no gap in knowledge or attitudes against their male peers, yet were still about 10% less likely to act on a highly advantageous financial opportunity.
The caveatA student sample, and the gap is behavioural, not a knowledge deficit. Hold education in finance itself constant and it still appears.
SourceInt. Journal of Educational Management (2025) ↗
47%
Fewer than half of women, 47%, feel confident discussing money and investing with a professional. Women are confident running the day-to-day budget; the reluctance is specific to investing, and it reaches into affluent households.
The caveatSelf-reported, from an asset manager.
SourceFidelity, Money FIT Women Study (2015) ↗
The emotions of money
The relationship with money is emotional first, and that is where behaviour is actually decided. It is the part the spreadsheets miss, and where I chose to start.
20–24%
Across 114 countries and 96,000 people, women are 20 to 24% more likely than men to experience financial anxiety, even after accounting for income and circumstance. The burden is not evenly shared.
The caveatCross-sectional data, measured during the pandemic.
SourceKadzima et al., Economics Letters (2025) ↗
4 scripts
Four largely unconscious “money scripts” formed in childhood, avoidance, worship, status and vigilance, predict adult income and net worth. What you believe about money quietly shapes what you do with it.
The caveatCorrelational; beliefs and outcomes reinforce each other over time.
SourceKlontz et al., J. Financial Therapy (2011) ↗
scarcity
Financial scarcity itself consumes mental bandwidth: the same person reasons measurably worse when money is tight than when it is not. Poor money decisions can be a consequence of stress, not a cause of it.
The caveatField and lab evidence; a large and much-discussed effect.
SourceMani, Mullainathan, Shafir & Zhao, Science (2013) ↗
The overconfidence paradox
What we call a confidence deficit in women shows up as a costly surplus in men: more trading, more fees, lower net returns.
45% more
Men trade their portfolios 45% more than women, and it backfires: that trading cut men’s yearly returns by 2.65 points against 1.72 for women. High turnover is the fingerprint of overconfidence, not of better information.
The caveatSingle-broker US data from the 1990s, but the pattern replicates widely.
SourceBarber & Odean, “Boys Will Be Boys,” QJE (2001) ↗
+0.4% / yr
Across 5.2 million real accounts over a decade, women’s returns edged men’s by about 0.4 of a percentage point a year, mostly by staying invested and not panic-selling.
The caveatFidelity is an asset manager with an interest in this story, and the method is not published. Read it as “at least as good,” not “far better.”
SourceFidelity, Women and Investing Study (2021) ↗
68% · 59%
Women held 68% of their investable money in cash against 59% for men. Over decades that caution costs more in lost compounding than trading fees ever save.
The caveatSelf-reported survey from an asset manager, now a decade old, though later data points the same way.
SourceBlackRock, Global Investor Pulse (2015) ↗
Knowing is not doing
The clearest finding in the whole field, and the one that most shaped how I designed the platform: around action, not lectures.
0.1%
A meta-analysis of 201 studies found that financial education, on its own, explained just 0.1% (one tenth of one percent) of the variation in what people actually do with money, and the effect faded within about two years.
The caveatThe precise claim is “small and decaying,” not “useless.” But information alone does not move behaviour.
SourceFernandes, Lynch & Netemeyer, Management Science (2014) ↗
rules > theory
In a trial with entrepreneurs, simple rules of thumb like “keep business and home cash in separate drawers” improved real financial behaviour, while formal accounting training did nothing at all.
The caveatHow you teach beat how much. It helped the least-confident the most.
SourceDrexler, Fischer & Schoar, AEJ: Applied (2014) ↗
Who holds the wheel
It is shaped by who has ever been handed the decision, and by whether you were given the chance to learn the thing in the first place.
48%
48% of women in couples, nearly half, say a spouse takes charge of long-term investing decisions, rising to 51% among millennials. A lot of what looks like low confidence is simply never having held the wheel.
The caveatA wealth-manager survey of high-net-worth couples, so not fully representative.
SourceUBS, Own Your Worth (2021) ↗
expertise, not gender
Women are more risk-averse investors on average, but the gap shrinks with wealth and knowledge, and among financial professionals it largely disappears. The caution tracks expertise, not gender.
The caveatSay “narrows,” not “reverses.” It is a tendency, not a rule.
SourceCroson & Gneezy, J. Economic Literature (2009) ↗
How you say it
The same fact, framed two ways, produces two different decisions. For talking about money this is not decoration, it is the mechanism, and it is why I care so much about language.
gain vs loss
The same choice described as a loss rather than a gain flips people’s decisions, because a loss is felt more sharply than an equal gain. Wording is never neutral.
The caveatThe lab classic. It shows frames shift a choice, not which frame sustains action.
SourceTversky & Kahneman, Science (1981) ↗
fit > force
A frame persuades when it matches the reader’s goal: growth and empowerment language lands with women in an aspirational frame of mind, security and protection language with those guarding what they have. There is no universal winner.
The caveatConsumer studies, not financial services, but the principle is robust.
SourceLee & Aaker, J. Personality & Social Psych. (2004) ↗
2×
In real pension communication, a “protect what you’re building” frame got about twice the engagement of a “grow your money” frame. When the job is simply to get someone to start, security can out-pull ambition.
The caveatPensions are an inherently security-flavoured topic, so this may not transfer everywhere.
SourceEberhardt et al., Int. J. of Research in Marketing (2021) ↗
The honest counterweight: concrete framing does not always win. For long-term goals like building wealth, an abstract, aspirational frame (the life you want) can beat concrete how-to detail. Match the frame to the time-horizon, not to a rule. Source: Rudzinska-Wojciechowska, PLOS ONE (2017).
It is happening again with AI
The confidence-and-emotion story is repeating in the newest technology. Women use AI less and feel warier, even as access levels out, which is exactly why I care about it.
~16 pts
Even in the same job with the same tasks, women are about 16 percentage points less likely than men to use ChatGPT at work. The gap is not about believing in the tool, women are just as optimistic. It is friction and “I don’t know how.”
The caveatDanish workers; the earlier working paper put the same-job gap nearer 20 points.
SourceHumlum & Vestergaard, PNAS (2025) ↗
13% → 33%
What drives the gap is risk perception, not skill or access. When the framing shifted toward what AI is good for, young women’s use jumped from 13% to 33%. Reassurance moves the needle more than another tutorial.
The caveatA UK preprint, and correlational, so read the lever as promising, not proven.
SourceStephany & Duszynski, preprint (2026) ↗
11% · 22%
Only 11% of women think AI will be good for society, against 22% of men, and women are twice as likely to expect a negative personal impact. Yet women and men now use chatbots at nearly the same rate, so this is an emotional gap, not an access one.
The caveatA US survey, and attitudes measured at one point in time.
SourcePew Research Center (2026) ↗
50% · 43%
Half of women, versus 43% of men, say using AI at work “feels like cheating,” and women report less encouragement from managers to use it. The barrier is permission, not capability.
The caveatA single-question survey, reported via a news writeup.
SourceCNBC · SurveyMonkey (2026) ↗
+34%
When less-experienced workers got an AI assistant, their productivity rose 34%, against 14% on average. The gains flow to those furthest behind, so closing the adoption gap is high-leverage, not cosmetic.
The caveatNot a gender study: one firm, customer support. The bridge to gender is who gains most.
SourceBrynjolfsson, Li & Raymond, QJE (2025) ↗
The honest counterweight: the trial gap is closing. US women and men now use chatbots at almost the same rate. What still lags is daily use, confidence, and enthusiasm, so the work is keeping women going, not just getting them to try once. Source: Pew Research Center, 2026.
The emotions of technology
The wariness I see around AI is the same one researchers measured around computers for decades, and it has always fallen a little harder on women.
23 countries
In a study across 23 countries, many had a technophobic majority, more than half of first-year students anxious or uneasy about computers, and women reported the more negative attitudes. Fear of the machine is measurable, widespread, and gendered.
The caveatA 1990s student sample, so the absolute rates are dated. The cross-national and gender pattern is the durable part.
SourceWeil & Rosen, Computers in Human Behavior (1995) ↗
equal skill
Given real online tasks, women and men performed equally well, yet women rated their own internet skills significantly lower. The gap was in perceived ability, not actual ability, and it shapes what a person is willing to try.
The caveatA single-region, early-web sample, though the underrating pattern has been widely replicated since.
SourceHargittai & Shafer, Social Science Quarterly (2006) ↗
4.0 · 3.5
On a validated AI-anxiety scale, women scored higher than men (4.0 versus 3.5), and that anxiety explained much of their cooler attitude toward AI. The barrier is felt before it is reasoned.
The caveatA single, modest-sized sample, and the effect is real but small. The direction is not universal across every study.
SourceRusso et al., Frontiers in Psychology (2025) ↗
The honest counterweight: the gap is apprehension, not disinterest. In the same research, women score higher than men on wanting to learn AI, so the work is to lower the dread, not to convince anyone it matters. Source: Uğur & Dursun, 2025.
Why women hold back
When you ask why the gap exists, the answers are rarely about skill. They are about risk, judgement, and who is told it is allowed.
13% · 6%
Reviewers shown identical work rated the woman who used AI 13% less competent, more than double the 6% penalty a man took for exactly the same thing. The fear of being judged for using AI is not imagined.
The caveatA pre-registered experiment in one field, software, using a stylised code-review task.
SourceGai, Hou, Tu & Acar, working paper (2025) ↗
+11%
Women rated AI about 11% riskier than men, driven by higher risk aversion and by having more to lose to automation. When the promised benefit to their own job was made certain, the gap shrank.
The caveatAttitudes on an opt-in panel in two countries, not observed behaviour.
SourceBorwein et al., PNAS Nexus (2026) ↗
30% · 37%
Fewer women than men said a manager had encouraged them to use AI (30% versus 37%), and women were likelier to worry it would look like cheating. Less permission, more guilt: the block is social, not technical.
The caveatA self-reported survey from an advocacy organisation, not peer-reviewed.
SourceLeanIn.Org (2026) ↗
What closes it
This is the part that made me want to build for it: the gains from AI flow to whoever is furthest behind, and confidence responds to being shown, not told.
40%
Giving people ChatGPT for writing cut the time they took by 40% and raised the quality by 18%, and the weakest writers improved the most, so the skill gap narrowed. AI is a leveller when people actually use it.
The caveatNot a gender study: a writing experiment with professionals. The bridge to gender is who gains most.
SourceNoy & Zhang, Science (2023) ↗
11% → 14.5%
A single one-hour classroom visit from a woman working in science raised the share of French girls choosing selective, male-dominated STEM tracks from 11% to 14.5%. One hour of a relatable role model moved a real choice.
The caveatIt measures field choice, not AI use, and the effect appeared only for the most male-skewed tracks.
SourceBreda et al., The Economic Journal (2023) ↗
100%
Every first-year woman in engineering given a female peer mentor stayed in the field a year later, against an 18% dropout for those given a male mentor and 11% for those with none. Same-gender support beat a generic programme, and the benefit lasted two years.
The caveatRetention in engineering, not AI adoption. The shared mechanism is belonging and peer support.
SourceDennehy & Dasgupta, PNAS (2017) ↗
Why it matters: if the biggest gains go to whoever is furthest behind, then closing the adoption gap for the women who hesitate is not charity, it is the highest-leverage move on the board.
The bigger picture
These findings sit beyond what any one platform can touch. I include them because they are why this work matters, and where it points next.
700M
Roughly 700 million women worldwide still have no financial account at all. Even after rapid progress, the absolute scale of exclusion is enormous.
SourceWorld Bank, Global Findex (2025) ↗
35% · 30%
Worldwide, 35% of men and 30% of women count as financially literate on a survey of 150,000 adults. The five-point gap holds in rich and poor countries alike.
SourceHasler & Lusardi, S&P Global FinLit / GFLEC (2017) ↗
22 → 0
India erased an account-ownership gap of more than twenty points in a decade, to near zero. Ownership is now equal, though women’s account usage still lags, so access is not the same as use.
SourceWorld Bank, Global Findex (2021) ↗
12 points
In Sub-Saharan Africa women remain twelve points less likely than men to hold any account. The gap is deeply regional: near-closed in South Asia, still wide across much of Africa.
SourceWorld Bank, Global Findex (2025) ↗
194,000
Kenya’s M-PESA mobile money lifted 194,000 households out of poverty, with gains concentrated among women: about 185,000 moved from subsistence farming into business. A usable tool plus agency, not a course.
SourceSuri & Jack, Science (2016) ↗
1–2%
All-women founding teams received about 1% of US venture funding in 2024, down from 2% in 2023. Teams with at least one woman did far better, near 20% of deal value.
SourcePitchBook, All In: Female Founders (2024) ↗
$30T
About $30 trillion in assets is shifting to US women this decade, and the most affluent women report being the most under-served by the financial industry. The audience is large, growing, and waiting to be taken seriously.
SourceMcKinsey (2024) · BCG (2020) ↗
No findings match that filter yet. Try “Everything” to see the full set.
The evidence kept pointing at the same thing: not a missing brain, a missing sense of permission, and a relationship with money that is emotional long before it is technical.
That is a solvable problem, and it is why I built Be Your Own CEO the way I did: to work on confidence and emotion through doing, not just reading.
See the platformA working reference, not an argument. Each finding gives the number as it is most defensibly stated, one or two plain sentences on what it means, and the caveat that keeps the naive reading honest.
The pension and wider wealth gap are deliberately left out. They are real, but they are a story about what has been done to women, and this is a page about what a woman can do as her own CEO.
Peer-reviewed Published in an academic journal or an NBER working paper.
Institutional Central banks, statistical offices and research bodies: Federal Reserve, OECD, World Bank, GFLEC.
Industry Asset managers and consultancies with a commercial interest in the story. Useful, but flagged, and cited to the underlying number.