Why Microfinance Can Fail Poor Women — and What Actually Works

Microfinance was never a foolish idea. Giving people who were excluded from conventional banking access to small loans can help them invest, smooth cash flow, and build businesses. But the promise became much larger than the evidence. Small loans were often presented as a direct route from poverty to entrepreneurship and empowerment, particularly for women.

The evidence is more complicated. Randomized evaluations have found that microcredit can increase borrowing and business investment without producing large improvements in household consumption, income, health, education, or women’s empowerment. A major evaluation of group-based microcredit in Hyderabad, India, for example, found more investment in existing businesses but no significant improvement in average consumption or women’s empowerment. citeturn0search0turn0search6

That does not mean every microfinance program fails. It means a loan should be treated as a financial tool, not as a poverty-reduction strategy by itself. For very poor households, the more important question may be what combination of capital, skills, savings, coaching, market access, and social protection allows income and assets to grow without creating a new repayment burden.

The Problem With the Simple Microfinance Story

The classic microfinance model assumes that a person with a small amount of capital can use it to start or expand a business, earn a return, repay the loan, and gradually build wealth. That can happen. But it depends on there being a viable business opportunity and enough flexibility for the borrower to survive periods when income is low.

Poor households often operate in unstable markets. A woman selling food may face a bad week because customers have no cash. A livestock business can be hit by disease. A small retailer can lose inventory to theft or a price shock. A seasonal worker may have income for only part of the year.

A fixed repayment schedule does not automatically adjust to these realities. The borrower still has to make the payment even when the business has not produced enough cash. In that situation, a new loan can become a way to repay the old one rather than a source of productive investment.

Repayment Pressure Is Not a Small Detail

Repayment design matters. A randomized study of microfinance clients in India compared traditional weekly repayments with a monthly schedule. Clients with monthly repayments were 51 percent less likely to report feeling worried, tense, or anxious about repayment and were more likely to report confidence in their ability to repay. They also reported higher business investment and income. citeturn0search5turn0search15

The lesson is not that weekly repayment always causes harm. It is that the timing of repayments can change the financial pressure placed on borrowers. A business needs working capital, and taking money out of a business too quickly can make it harder to build inventory or handle uneven income.

Good financial inclusion therefore needs to consider cash-flow patterns, not simply whether a borrower technically qualifies for a loan.

When Group Lending Creates Its Own Problems

Group lending was designed partly to solve a practical problem: people without conventional collateral could borrow through a group in which members helped reinforce repayment. Social relationships became part of the lending mechanism.

That structure can reduce lender risk, but it can also transfer pressure onto borrowers. When one member struggles, other members may feel pressure to help cover the payment or persuade the person to repay. During a community-wide shock, the problem can spread.

Research using repayment data from about two million microfinance borrowers during India’s 2016 demonetization crisis found that defaults were clustered within lending communities. The study also found evidence consistent with defaults spreading through informal community connections as well as formal joint-liability relationships. citeturn0search10turn0search14

Group lending is therefore not automatically exploitative, but it should not be treated as socially neutral either. Existing relationships can provide support, but they can also reproduce local power structures and place additional pressure on people who already have limited bargaining power.

A Loan to a Woman Is Not Automatically Women’s Empowerment

One of the most persistent assumptions in development finance is that lending to women automatically increases their control over resources. The evidence does not support such a simple conclusion.

Research on women’s lending in rural India found that loans taken in women’s names could be used to increase household assets and income while the women themselves did not necessarily gain ownership or control over those assets. The researchers described this as an empowerment paradox: household welfare could improve without women’s bargaining position improving. citeturn0search12

This distinction matters. If a woman takes a loan to purchase a cow but another household member controls the cow, the income, and the final financial decision, the loan may have helped the household without giving the borrower the economic power that the word ’empowerment’ suggests.

Programs that want to improve women’s economic position therefore need to consider asset ownership, decision-making power, income control, safety, mobility, and access to savings—not just the number of women who receive loans.

Why More Credit Is Not Always the Answer

The Hyderabad randomized evaluation provides an important warning against measuring success simply by loan uptake. Access to microcredit increased substantially in treatment neighborhoods. Existing businesses invested more, and some business outcomes improved. But average consumption did not rise significantly, and researchers found no significant changes in health, education, or women’s empowerment. citeturn0search0turn0search6

A separate randomized evaluation of group-lending expansion in Mexico found no evidence of transformative effects across a broad set of outcomes, including entrepreneurship, income, labor supply, expenditure, social status, and subjective well-being. citeturn0search4

These findings do not make microfinance useless. They show why it should be judged by outcomes rather than by the number of loans issued or the repayment rate.

What Shows More Promise for the Very Poor

For households living at the lowest levels of income, a loan may be the wrong first intervention. Someone with no productive assets, irregular income, food insecurity, and little ability to absorb risk may need a period of support before taking on debt.

This is where the ‘Graduation’ approach has produced stronger evidence. The model combines several interventions, commonly including a productive asset transfer, training, coaching, temporary consumption support, savings, and access to services such as healthcare.

A six-country randomized evaluation covering more than 10,000 participants found significant improvements across multiple measures. One year after the intervention ended, income and revenues were higher in every country, and household consumption was significantly higher in five of the six countries. The researchers concluded that the multifaceted approach produced sustained improvements in well-being and was cost-effective in five of the six sites. citeturn0search1

The India Evidence Is Particularly Useful

A randomized evaluation in Murshidabad, West Bengal, provides a useful example of how a broader livelihood program can work. The program targeted very poor households and combined a productive asset transfer with technical training, temporary consumption support, savings, regular household visits, coaching, and health information.

The long-term results were substantial. According to the Abdul Latif Jameel Poverty Action Lab, the program produced positive effects on consumption, food security, income, assets, productive activity, health, and income diversification. Positive effects on income and assets persisted years after the original asset transfer. citeturn0search8

The important point is not that every poor woman needs a goat, a cow, or a particular package of services. The lesson is that extremely poor households often face several constraints at the same time. Addressing only the shortage of credit may leave the other constraints untouched.

What About Cash Instead of Loans?

Some newer programs replace or complement productive-asset transfers with cash. Evidence from a randomized evaluation of a poverty-graduation program for ultra-poor women found that cash transfers combined with business training, mentoring, and savings produced positive effects on income, savings, asset accumulation, and food security in the short to medium term. citeturn0search11

Cash does not automatically work either. The design matters. Transfers can be useful when households face a capital constraint, but training, mentoring, savings, and market opportunities can determine whether that capital becomes a durable source of income.

The New Evidence Also Adds a Warning

Not every lighter and cheaper version of a graduation program produces the same results. A 2026 randomized evaluation in Ethiopia found that a lighter-touch model with a transfer, training, and savings produced some gains in financial inclusion, assets, and livestock income, but did not improve consumption and did not generally move households out of a poverty trap. Training and savings groups alone had minimal effects. citeturn0search2turn0search3

That finding is important for NGOs and funders. It suggests that removing the more expensive parts of a program may also remove some of the mechanisms that make it effective. ‘Cheaper’ is not automatically ‘better’ if the cheaper program no longer solves the constraints faced by the poorest households.

What Better Microfinance Can Look Like

Flexible repayment: Repayment schedules should reflect how borrowers actually earn money. Where income is irregular, rigid weekly payments can create unnecessary stress and working-capital pressure.

Savings before debt: Building a small financial buffer can help households manage shocks without immediately taking another loan.

Smaller or staged credit: Credit should grow with demonstrated business capacity rather than assuming that a larger loan automatically creates a larger business.

Asset and cash support for the ultra-poor: Households without productive assets may need grants or transfers before debt becomes useful.

Training and coaching: Technical and business support can help borrowers turn capital into productive activity.

Women’s control over assets: Programs should measure whether women control income and productive assets, not simply whether a loan is registered in a woman’s name.

Market connections: A loan cannot create customers. Programs should consider demand, supply chains, pricing, and market access.

Protection from over-borrowing: Providers should assess existing debt and avoid encouraging multiple loans that leave households dependent on refinancing.

What Should NGOs Measure?

A microfinance program should not declare success because repayment rates are high. A lender can achieve excellent repayment while borrowers remain no better off—or while they use new borrowing to manage old debt.

Better measures include changes in household income, consumption, savings, productive assets, business profits, debt levels, food security, financial stress, and women’s control over resources. Results should also be measured after enough time has passed to determine whether changes last.

The strongest programs should compare participants with an appropriate comparison group whenever possible. Randomized evaluations are particularly useful because they can separate program effects from changes that would have happened anyway.

The Practical Lesson

Microfinance works best when it is treated as one tool rather than a complete poverty solution. Credit can help a viable business grow. It can finance inventory, equipment, or working capital. But credit also creates an obligation, and that obligation can become harmful when income is uncertain or when the business opportunity is weak.

For the poorest households, the evidence increasingly points toward a broader model: provide the resources needed to start a productive activity, give people practical training and coaching, encourage savings, connect them to markets, and reduce the immediate pressure of poverty while the income source develops.

That approach is more expensive and more complicated than simply issuing loans. It is also closer to the actual problem. Poverty is rarely caused by one missing financial product. It is usually a combination of limited assets, unstable income, weak market access, low bargaining power, shocks, and inadequate support.

Conclusion

The strongest criticism of microfinance is not that small loans never help. It is that loans have often been asked to do a job they were never designed to do: solve poverty by themselves.

The evidence shows a more useful path. Flexible credit can reduce financial stress. Savings can provide resilience. Training and mentoring can improve the use of capital. But for people living in extreme poverty, integrated graduation-style programs—with productive assets or cash, coaching, savings, and other support—have produced much stronger evidence of sustained improvements in income, consumption, assets, and food security. citeturn0search1turn0search8

For NGOs and financial institutions, the practical goal should therefore be simple: stop measuring success by how many loans are distributed and start measuring whether women’s economic lives actually improve.

Selected Research and Evidence

Banerjee, Duflo, Glennerster & Kinnan (2015), American Economic Journal: Applied Economics — randomized evaluation of group-lending microcredit in Hyderabad, India.

Angelucci, Karlan & Zinman (2015), American Economic Journal: Applied Economics — randomized evaluation of group-lending expansion in Mexico.

Repayment flexibility randomized trial among microfinance clients in India, PLOS ONE.

Research on women’s lending, household vulnerability, and empowerment in rural India, World Development.

Banerjee et al. (2015), Science — six-country randomized evaluation of the multifaceted Graduation approach.

J-PAL — long-term evaluation of the multifaceted livelihoods/Graduation program in Murshidabad, India.

2026 Journal of Development Economics evidence on a lighter-touch graduation model in Ethiopia.