Close
Empowerment

Women’s loans lift household income though their own profits stay flat

(19 min read)

A year ago this month, WEESA, the World Bank’s community of practice on women’s economic empowerment, ran a webinar titled “Who Gets the Capital? How Household Dynamics Shape Female Microentrepreneur Investments” and sent attendees to one paper: “Household Matters”, by Arielle Bernhardt, Erica Field, Rohini Pande and Natalia Rigol, published in American Economic Review: Insights in September 2019 (AEA record). WEESA is hosted by the World Bank’s South Asia Gender Innovation Lab and says it has more than 4,600 members in over 90 countries. We have not seen the recording or the three slide decks, which sit behind the members’ platform, so this article reads the paper and the evidence around it, not the talk.

The paper asks a plain question: when a woman gets a loan or a grant for her business, whose business grows? It matters to anyone who pays women directly, which is what Kilimora does through AgriKonnekt. This article reads it for what it means for the cooperatives we serve. Our companion piece, “Who pays when the rain fails?”, covers the other side of cash: who carries the loss when a shock hits.

Read this first. The experiments ran between 2005 and 2010 in Kolkata, Accra and three districts of Sri Lanka. They study urban microenterprises, not rain fed farms, and the loans and grants they test are not payments for verified climate practice. We use the paper as a lens for how to measure, not as proof of what AgriKonnekt payments do. Figures come from the January 2019 working paper draft we could read in full (PDF); the published version may differ slightly. “Not significant” means the effect cannot be told apart from zero.
Four numbers from one paper

Bernhardt, Field, Pande and Rigol, India study of 474 households with a female entrepreneur

+48%all household businessprofits, India≈0change in her own profitsin multi-business homes97.5%of other household firmsare run by men, India+81%her profits when hers isthe only business, India

Eight findings

  • A puzzle with a long history. Several experiments find that grants raise men’s business profits but not women’s.
  • Women often share a home with another business. Between two fifths and three fifths of the women in the three samples did, and women were about 1.5 times as likely as men to.
  • Count the household, and the gain appears. In India one loan feature raised all household business profits by 48% and household income by 25%, while her own business showed no measurable change.
  • The money went to the best opportunity at home. Where another member had a business, that business’s profits rose 44%, and 97.5% of such businesses were run by men.
  • Sole owners do well. Where hers was the only business, her profits rose 81% in India, 22% in Sri Lanka and 43% in Ghana.
  • The gender gap closes for sole owners. In none of the three samples could the authors reject equal returns for women and men.
  • Household income gains look alike. In Sri Lanka, gains for male and female recipients could not be told apart.
  • The “why” is still open. Earnings norms, care duties and capture all fit parts of the data. Only capture implies misallocation.

The puzzle: grants that seem to work for men only

The finding that started the debate came from Sri Lanka. De Mel, McKenzie and Woodruff randomly gave cash and equipment grants to small business owners and found that profits rose in men’s enterprises but not in women’s. Studies in Ghana, Uganda, Tanzania and Nigeria found the same pattern, with one exception. The usual explanations are that women’s money is taken by others, that women are less committed or more impatient, or that women sort into less profitable trades. The paper’s own review of that literature is below.

StudyGrantMen’s profitsWomen’s profits
Sri Lanka, de Mel and colleaguesCash or equipmentUp about 9% of the grant in two years; a 6 to 12% monthly return persists at five yearsNo increase on average, even five years on
Ghana, Fafchamps and colleaguesCash or in kindIn kind grants: +30 to 60%No increase on average
Uganda, FialaGrants or loans, some with trainingLoans plus training: +54%No increase in any arm
Tanzania, Berge and colleaguesBusiness grantsNo increase on averageNo increase on average
Nigeria, McKenzieBusiness grants0.16 standard deviation rise in a profit and sales indexNo increase
Uganda, Blattman and colleaguesBusiness grantsAbout +30%, held at four yearsNone at two years; +73% at four
Patience pays: the Uganda result

Change in business profits after unconditional grants, per cent, men against women

MenWomenAfter two yearsabout +30%about +30%no increaseno increaseAfter four yearsabout +30%about +30%+73%+73%

The exception is instructive. In Uganda, women’s profits showed nothing after two years and then jumped 73% by year four, level with men. A verdict taken after one season would have called the grants a failure for women. For Kilimora: AgriKonnekt pays in seasonal cycles. The framework’s promise of annual independent evaluations with control groups should be read across several years, and the first result should not be taken as the last.

Same loan, two ledgers

The India study is the cleanest test. In 2007, women microfinance clients in Kolkata, in 169 five member repayment groups, were offered individual loans of Rs 4,000 to 10,000 (about US$90 to 225). Some groups got the standard contract, with the first repayment due two weeks after disbursal. Others got a two month grace period, which eased cash flow and was otherwise identical. The authors analyse the 474 households that include a female entrepreneur.

One loan, two ledgers

India, multi business households. What happens depends on whose books you read

A loan with a grace periodgoes to a woman(Kolkata, 2007)Her own businessProfit change: not measurable(not significant)Other household businesses97.5% run by menProfit change: +44%The householdEnterprise profits +48%Household income +25%
The rupees that were hiding in the household

Effect of the grace period loan on average weekly profits, Indian rupees. Control group means: Rs 401 (her business), Rs 1,387 (all household businesses)

Her own enterpriseRs 167 (not significant)Rs 167 (not significant)All householdenterprisesRs 672 (1% level)Rs 672 (1% level)

Her own enterprise’s weekly profits rose by Rs 167, which cannot be told apart from zero. Profits across all household businesses rose by Rs 672, a 48% gain, and the difference between the two is significant at the 5% level. Household income rose 25%. In Sri Lanka, grants to women left their own enterprise profits unchanged but lifted household income by about 8%.

Household level gains when women get capital

Change relative to the control group, per cent

India: all householdbusiness profits+48% (1% level)+48% (1% level)India: householdincome+25% (1% level)+25% (1% level)Sri Lanka: householdincome+8% (10% level)+8% (10% level)

The lesson is about measurement. Reading only her own business would have said the capital did nothing. Reading the household says it did a good deal, somewhere else. The authors note that recent cash transfer and microcredit studies, which measure at household level, find that the recipient’s gender does not matter for household outcomes, among them Haushofer and Shapiro’s study in Kenya.

Where the money went

The authors split each sample by household type. In single business homes, her enterprise is the only place for the capital to go. In multi business homes, there is a choice. In India, where they could observe every household enterprise, the choice was clear: her own business showed no gain, and the other members’ businesses gained 44%.

Her own profits, by who else runs a business at home

Change in the woman’s own enterprise profits, per cent of the control mean

Another household member also has a businessShe is the only business ownerIndia−11% (not significant)−11% (not significant)+81% (5% level)+81% (5% level)Sri Lankano effect (not significant)no effect (not significant)+22% (10% level)+22% (10% level)Ghana+31% (10% level)+31% (10% level)+43% (5% level)+43% (5% level)
Who owns the “other” business?

India: 474 households with a female entrepreneur, and the owners of the other businesses in those homes

213 homes, 45%261 homes, 55%474households213 homes, 45%She is the onlybusiness owner261 homes, 55%Another member alsohas a business97.5%2.5%97.5%run by men97.5%Other businessesrun by men2.5%Other businessesnot run by men

Women face this choice more often than men. In Sri Lanka and Ghana, women entrepreneurs were at least 1.5 times as likely as men to live with a second self employed person, partly because fewer women are in the labour force, so households have more male run businesses to invest in.

Women are likelier than men to share a home with another business

Per cent of sampled entrepreneurs living with at least one other self employed person

WomenMenIndia55%55%Sri Lanka62%62%42%42%Ghana40%40%25%25%

For Kilimora: in a farming household the equivalent of a second business is a second plot, often the husband’s, or a livestock or trading enterprise run by another member. A woman paid for her work on her own plot may rationally put the money into the plot or business that earns most. If we only measure her plot, we will read a household investment as a failure.

When she runs the only business in the house

Is the gender gap real, then? The authors compare women who are the sole business owner with men. In Sri Lanka and Ghana, the gains are statistically indistinguishable, and in India the gap is large in size but also not statistically significant. The data cannot rule out some remaining gap, but they do not show one.

When she runs the only business, the gap closes

Women in single business homes, as per cent of the grant gain for comparison men. Dashed line = parity

Woman who is the only ownerSri Lanka(men sampled)94% of men’s gain94% of men’s gainGhana(men sampled)78% of men’s gain78% of men’s gainIndia(vs other household owners)61% of their gain61% of their gain

Why? Norms, time or capture

The experiments randomised the loans and grants, not the household’s business mix, so the authors are careful: their evidence on why is suggestive. They weigh three explanations.

ExplanationWhat it impliesWhat the paper found
Sorting by earnings norms. Men are expected to earn more, so women end up in lower return trades.Allocation is not “wrong”; policy should work on norms and on moving women into better sectors.Suggestive: husbands out earn wives in 89% of homes, and women with weaker earning husbands sort into lower return work.
Sorting by care and mobility norms. Childcare and limits on movement push women to home based work.Same as above: capital goes where the household’s best opportunity is.Mixed: women in both home types spend similar minutes on work and chores and give the same reason (flexibility) for their business.
Capture. Others in the household take the capital or the profits, or women hand them over in advance.The only explanation that implies households misallocate capital away from women.Not the main driver on this evidence: the gap appears with multiple businesses, not when she is the sole owner. The authors cannot rule it out.
The earnings norm in the Kolkata sample

Who earns more in the household, India

89%11%89%husband higher89%Husband earnsmore than wife11%Wife earnsmore than husbandWhere she does earn more,two thirds of women say thehusband is temporarily out ofwork, mostly through illness.Husbands’ wage income insingle-business homes is aboutdouble the business income ofhusbands in multi-business homes.
Different homes, different trades

Share of the woman’s enterprises or work pattern, India baseline, per cent

Another member also has a businessShe is the only business ownerPiece-rate work(low return)53%53%33%33%Retail ofnon-perishables28%28%51%51%Works from home70%70%54%54%
The chores are longer than the business day

Minutes spent yesterday by the woman, control group, India. Differences between the two home types are not statistically significant

Another member also has a businessShe is the only business ownerOn herenterprise165 min165 min131 min131 minOn chores andchildren207 min207 min219 min219 min

Why the distinction matters: if the pattern comes from earnings norms and care duties, the allocation may be efficient given the choices open to the household, and the fix is to change norms and open better trades to women. If it comes from capture, the household is misallocating, and the fix is bargaining power and control. The paper’s tentative reading leans to the former; the authors say clearly that they cannot settle it.

For Kilimora: AgriKonnekt’s individual payment accounts address capture at the input stage. They do not decide where she chooses to spend, and nothing here says they should. The right posture is to protect her control, measure household outcomes honestly, and avoid treating spending on a family plot as a failure by default.

Old data, still the best test we have

When each experiment ran, when the paper appeared, and where Kilimora’s own targets sit

200520102015202020252030Sri Lanka grantsMar 2005 to Mar 2007Mar 2005 to Mar 2007India loans (Kolkata)2007 to 2010 survey2007 to 2010 surveyGhana grants (Accra)Oct 2008 to Feb 2010Oct 2008 to Feb 2010Working paper (CEPR, NBER)Apr 2017Apr 2017Published, AER: InsightsSep 2019Sep 2019WEESA webinarNov 2025Nov 2025Kilimora gender frameworkJan 2026Jan 2026Kilimora 2028 targetsby 2028by 2028

The time budget behind the business

The chores finding in Kolkata connects to a wider pattern. UN Women estimates that women in Sub Saharan Africa spend 3.1 times as long as men on unpaid care and domestic work, against a world average above 2.5. In West and Central Africa, rural women can spend up to 12 hours a day on unpaid work, and 55% of women business owners in Africa spend between two and five hours a day on unpaid care. Uganda shows a gentler ratio, and Malawi a far harsher one.

How many times more unpaid care time women do than men

Ratio of women’s to men’s daily or weekly time on unpaid care and domestic work. Surveys of different years

Malawiabout 7×about 7×Benin5.3× (3 h 41 vs 42 min)5.3× (3 h 41 vs 42 min)Mali3.8× (21.6 vs 5.7 h a week)3.8× (21.6 vs 5.7 h a week)Tanzania3.5×3.5×Sub-Saharan Africa, average3.1×3.1×Rwanda3.0×3.0×Uganda1.2×1.2×

For Kilimora: AgriKonnekt’s USSD lessons need no data plan and can be taken on any phone, which is designed to cost little time. We should record minutes spent on plots, training and chores, before and after, so that the claim rests on measurement rather than design.

Kenya: parity at the account, gaps in what flows through it

Kenya is the region’s pioneer in mobile money, and its numbers show how the paper’s logic plays out nationally. The 2024 FinAccess survey found that 84.8% of adults have access to formal financial services, 84.1% of women and 85.7% of men, a gap of 1.6 points. The Central Bank credits mobile money for that near parity. Women’s digital financial usage rose to 82% in 2024 from 71% in 2021, per an AFI case study.

Behind the parity are gaps in value. Men made 61% of the value moved through mobile money, which is 5.3 trillion of 8.7 trillion as reported, and earned an average Sh47,571 a month against Sh34,847 for women. The World Bank’s 2025 Findex finds that in Kenya 32% of adults borrowed from their mobile money provider, yet women are 16 points less likely than men to borrow that way.

Parity at the account, gaps in what flows through it

Women’s level as a per cent of men’s, Kenya, FinAccess 2024

Has a financialaccount or product98%98%Average monthlyincome73%73%Value moved throughmobile money64%64%
Where the digital money goes in Kenya

Left: who moves the value in mobile money. Right: who took Hustler Fund loans, by wealth fifth

61%39%8.7 trnmoved in total61%Men’s share ofmobile money value39%Women’s share ofmobile money value35.8%33.5%30.7%69%top two fifths35.8%Richest fifthof Kenyans33.5%Second-richestfifth30.7%Other threefifths

The state backed Hustler Fund is a useful test. Its Cooperatives Secretary reported in May 2025 that it had disbursed Sh70 billion to about 25 million borrowers, 48% of them women. But a report on the KNBS Economic Survey 2025 says the richest fifth of Kenyans took 35.8% of the loans and the next fifth 33.5%. Having an account, in other words, is not the same as getting capital that fits your need.

For Kilimora: the Kilimora framework notes that women in East and Central Africa hold verified identity credentials at rates 15 to 25 points lower than men, which blocks them from KYC compliant mobile money. That is the front door to the account. The paper adds the question that comes after it: once the money arrives, whose opportunity does it serve?

The credit gap in African farming

For women in farming the gap is wider still. Estimates of women’s share of the agricultural labour force range from about half in Sub Saharan Africa, according to FAO figures cited by the Haller Foundation, to over 60% in Kilimora’s own framework, depending on source and year. Either way, they are central to the work and peripheral to the credit.

The credit gap in African farming, in four numbers

Estimates from FAO, the African Development Bank and the studies that cite them

<10%of farm credit goes towomen (FAO figure)5–10 ptslower credit access forfemale smallholders$42Bgender financing gap,Sub-Saharan Africa20–30%lower output from womenfarmers: the “crop gap”

A study of five Sub Saharan countries found male headed households on average 10.36% likelier to access agricultural credit, and that almost 88% of the gap could not be explained by measured differences such as location or socioeconomic status. The authors read this as structural barriers. The paper from Pande and colleagues adds a household level reason why a loan to a woman can look unproductive on paper while helping the family.

Measured differences explain little of the credit gap

Female and male headed households, agricultural credit access, five Sub Saharan countries

87.55%12.45%87.55%unexplained87.55%of the gap is not explainedby measured factors12.45%explained by measuredfactorsMale-headed households wereon average 10.36% likelier toaccess agricultural credit.Land ownership, age and viewsof credit widen the gap; theauthors point to structuralbarriers faced by female-headedhouseholds.

FAO’s response includes group based lending, financial literacy training and portfolio guarantees that share the risk with commercial banks. Grameen Foundation’s work in Kenya and elsewhere adds “intra household dialogues”, which aim to help families and communities redefine social norms and share power. We cite the approach, not its evidence, which we have not reviewed.

What it means for AgriKonnekt

Kilimora’s Gender Lens framework, effective January 2026, treats women’s participation as a binding system constraint. Over 60% of AgriKonnekt participants are women, paid individually by mobile money, and the framework sets 2028 targets for enrolment, active use, climate finance and cash for work. It also commits to annual independent evaluations, gender disaggregated data and publishing results whether positive, neutral or mixed. That is the right foundation. The paper suggests six additions to what gets measured.

AgriKonnekt’s 2028 targets for women

Women enrolled and paid individually, from Kilimora’s Gender Lens framework (effective January 2026). Targets, not results

Enrolled with individualpayment accounts7,5007,500Active users(about 70%, modelled)~5,250~5,250Climate financerecipients5,8005,800Cash-for-workrecipients2,5002,500
Question the paper raisesWhy it mattersWhat a cooperative programme can record
Who else in the household farms or trades?Women in multi business homes showed no gain in their own business but large gains in the household.A household roster at enrolment: plots and businesses, and who owns or runs each
Where did the payment go?Capital moves to the best opportunity at home, which is often not hers.Consent based follow up on spending, by whose plot or business it served
Whose ledger are we reading?Her own profit and household income can point in opposite directions.Report both, with a control group, before calling a payment ineffective
How long do we wait?Uganda’s women showed nothing at two years and +73% at four.A multi year evaluation window, not one season
Is it norms, time or capture?Only capture implies money is being misdirected.A decision making index, as in the paper, and household dialogue sessions
How much time does she have?Chores exceeded business time in both home types.Minutes on plot, training and chores, before and after

Two differences from the paper deserve plain mention. First, AgriKonnekt payments reward verified climate practice, so they are not unrestricted capital grants, and the household’s response may differ. Second, the studies are urban and several are 15 or more years old. A first season result from Murang’a, Nakuru, Kilifi, Kisumu, Meru or Kakamega should be allowed to say something different.

What we can and cannot claim

StatementEvidence status
Over 60% of AgriKonnekt participants are women, paid individuallyDesign statement on Kilimora’s framework page; no payments by gender report yet
Payments reach women’s own mobile accounts within 24 hours of funder releaseDesign claim; not independently audited
Individual accounts keep money in women’s handsDesign intent. The paper shows that who holds the money does not by itself decide whose business it feeds
Women’s household income risesNot yet measured; the framework commits to annual independent evaluations with control groups
5,000 women in formal climate finance by 2028; 5,800 climate finance recipientsBoth published targets; they are not the same measure
USSD lessons save women timeHours saved or spent not measured

What we would ask for

  1. Report women’s results at two levels, her own plot or business and the household, before calling any payment ineffective.
  2. Take a household roster at enrolment: who else runs a plot or business, and who owns it.
  3. Split results by single operator and multi operator households, as the paper does.
  4. Judge outcomes over several years, not one season.
  5. Measure minutes on plot, training and chores, before and after.
  6. Test household dialogue sessions alongside individual payments, with a comparison group.
  7. Do not treat spending on a family plot as failure by default, and ask whether a gap reflects norms, time or capture.
  8. Publish findings whether positive, neutral or mixed, as the framework promises.
Sources & notes

SOURCES: Bernhardt, Field, Pande and Rigol, American Economic Review: Insights, 2019; working paper draft of 11 January 2019; Yale EGC abstract; Harvard Kennedy School abstract, April 2017; Kilimora, Gender Lens Programming, AgriKonnekt, Closing the verification gap, Who pays when the rain fails?; Central Bank of Kenya, FinAccess 2024 findings; Alliance for Financial Inclusion, AFI Kenya case study; Kenya Times on FinAccess; Eastleigh Voice on the World Bank Findex 2025; UN Women, West and Central Africa care strategy, October 2024 and care toolkit, 2023; Nation, UNCTAD on unpaid care; TimesLive, Oxfam; FAO, gender and rural financial services; Addis Fortune; AllAfrica; Haller Foundation; Global Solutions Initiative brief; Asiseh and colleagues; The Star on the Hustler Fund; Business Quest. The WEESA event email (November 2025) names the webinar and links the paper; the recording and slides were not available to us and nothing here describes what the speakers said. Our own arithmetic: percentage effects for multi business women in India and Ghana, women as a share of men’s gain, Kenyan ratios, and the 30.7% remainder. The Hustler Fund wealth split, the Sh8.7 trillion total and several Africa wide estimates come from secondary reports. Chart notes: the Sri Lanka multi business effect is drawn as zero because the paper reports no significant effect and its coefficient is per 100 rupees of grant; none of the three single owner gaps against men is statistically significant; sector values are the paper’s mean differences added to the multi business means; Kenyan access figures are 84.1% for women and 85.7% for men, income Sh34,847 and Sh47,571 a month; the Benin and Mali care ratios are our own division. Kilimora statements about its systems and targets are the company’s own.