The short answer

Repayment tells you that an instalment was paid. It does not tell you how it was paid, or whether the enterprise behind it is still earning. If your programme aims at sustained progress, add a small set of measures of continuity and recovery alongside your financial ones: days unable to trade, whether productive assets were restored, time to resume earning, what the household had to draw on, and whether participants found the support useful. The framework below is a proposal to adapt and test. It is not a set of measured results.

What repayment tells you, and what it leaves out

Repayment is a sound measure of what it measures. It is timely, already collected and comparable across branches. A livelihood programme should keep it.

It is also narrower than it looks. Regulation itself looks past the instalment to the household. RBI's directions on microfinance loans require each regulated lender to have a board-approved policy for assessing household income, and cap a household's monthly repayment obligations, across all its loans, at 50 per cent of monthly household income. The unit of concern is the household's capacity, not only the loan's performance.

An instalment paid on time after a disruption can come from very different places: enterprise income, savings, a relative, a new loan or the sale of something the household needed. The repayment record looks the same in each case.

Three pieces of published evidence show why the source matters.

  • Access to emergency money is not the same as easy access. The World Bank's Global Findex 2021 asked adults where they would find emergency money within 30 days and how difficult it would be. In developing economies, 87 per cent said they could reach some source, including savings, borrowing, family or friends, work, or selling assets. Only 55 per cent said they could do so without much difficulty.
  • How households cope can carry its own cost. Research published by CGAP, carried out by Freedom from Hunger with women in rural Burkina Faso, found that households mainly relied on savings kept at home, reduced food consumption and sold grain and livestock. That is one study in one setting, but it shows the kind of information a repayment record cannot hold.
  • Client-reported data shows strain that portfolio data may not. The summary of 60 Decibels' 2023 Microfinance Index, based on interviews with more than 32,000 clients, reports that 30 per cent still expressed concern about their ability to meet an unforeseen emergency expense.

Sector standards point the same way. The Universal Standards for Social and Environmental Performance Management, published by Cerise+SPTF, expect a provider to collect data on outcomes for clients and their households, covering both positive and negative changes, at least annually, and to analyse it by client segment such as gender, age, location and income level.

What to add: the ability to keep earning and to recover

A livelihood programme usually has an implicit theory: with an asset, a skill or working capital, a person's enterprise becomes more dependable. Testing that theory means asking what happened when something went wrong, because something usually does. Was the enterprise closed for two days or two months? Is the asset working again? What did getting back cost the household?

These questions respect the people you serve. They treat an enterprise owner as someone managing a business through ordinary setbacks, and they ask her directly about her own experience.

A proposed outcome framework for livelihood resilience

Five indicators to sit alongside your financial measures. Choose the two or three that fit your programme's purpose; a short set that is collected well is worth more than a long set that is not.

Proposed indicatorWhat it asksSuggested baselineHow it could be collectedMeasurement limitations
DowntimeOn how many days in the period was the enterprise unable to trade because of a disruption?The same question for the equivalent season before the programme beganA short question at an existing meeting or visit, or a simple calendar the owner keepsRecall weakens over months. Seasonality: compare like seasons. Planned closures and festivals need separating from disruptions.
Productive-asset restorationAfter damage or breakdown, was the asset repaired or replaced and back in use, and within how long?An asset record at the start: what it is, its condition, a photograph with consentA follow-up check against the asset record after a reported incidentFew incidents mean small samples. "Restored" needs a clear definition. A replacement bought with new debt counts as restored but carries a burden.
Time to resume earningHow many days passed between the disruption and the first income from the enterprise afterwards?The owner's account of the most recent earlier disruption, if anyAsked once, soon after the enterprise restartsRecall. Enterprises that never restart drop out of an average: report them separately. Report the median and the range.
Recovery burdenWhat did the household draw on to recover: enterprise income, savings, help from family, new borrowing, sale of assets, cutting essentials?What the household says it would rely on today, in the manner of the Global Findex emergency-money questionA short structured question with fixed options; never as a condition of serviceSensitive, so under-reporting is likely. Needs clear consent and the option to decline. Describes the household, not only the enterprise.
Participant usefulnessDid participants understand the support offered, did they use it, and would they use it again?Understanding at enrolment: can the participant say in their own words what the support is and is not?Two or three questions asked by someone other than the person who delivered the serviceCourtesy bias: people are reluctant to criticise a provider they depend on. Shows usefulness, not impact.

Report each indicator by segment where numbers allow, for example women and men, or trade and location. Always show how many people answered.

The limitations that apply to all five

Recall and wording. People forget dates, and survey words do not always mean what the designer intended. When CGAP tested financial health questions, about 30 per cent of respondents in India reported a shock in the survey, although the team's qualitative work suggested most households had experienced one. The team's own explanation was that the idea of a "shock" may not have been clear. Ask about concrete events, such as days closed or a machine not working, and ask soon after they happen.

Attribution. A change between baseline and follow-up is not proof that the programme caused it. Weather, prices, a family event or another scheme may explain it. Without a credible comparison group, describe what changed and say plainly that you cannot isolate why.

Small samples. In any one period only some participants will face a disruption. Percentages drawn from a handful of cases mislead. Use counts and short case descriptions until numbers grow.

Seasonality. A monsoon quarter and a festival quarter are not comparable. Compare the same season year on year.

Consent and data minimisation. Collect only what you will use, explain why, and let people decline without consequence. Keep resilience data apart from credit decisions unless participants have clearly agreed otherwise.

Using the framework in a programme

Settle four things before any data is collected.

  1. Purpose. Which gains is the programme trying to sustain? Choose the indicators that speak to those and leave the rest.
  2. Baseline. Record it before support begins. A baseline reconstructed afterwards inherits every recall problem above.
  3. Roles. Who asks, who records, who sees the answers, and who pays for the time this takes?
  4. Use. Decide in advance what you will change if downtime is long or recovery leans on borrowing. Data nobody acts on is a cost to participants with no return.

Measurement is one half of the work. The other half is the practical planning that gives the numbers a chance to improve: knowing what each enterprise depends on and who is responsible when it is interrupted. Our companion piece sets out a worksheet for that: productive assets and livelihood continuity.

Agreeing the measures of progress is part of how AVYA Nirantar scopes a pilot with lenders, funders and implementation organisations, together with the community, the continuity problem and the roles. Assessments used in that work are estimates. They are not predictions or credit decisions.

Questions people ask

Why is loan repayment an incomplete measure of a livelihood programme?

Repayment shows that an instalment was paid, not how. After a disruption the money may come from enterprise income, savings, family, new borrowing or the sale of assets, and the repayment record looks the same in each case. It also says nothing about whether the enterprise is still earning.

What should a livelihood resilience programme measure?

Alongside financial measures, consider five proposed indicators: days unable to trade, whether productive assets were restored, time to resume earning, what the household drew on to recover, and whether participants understood and used the support. Choose two or three that fit the programme's purpose.

Do these indicators prove that a programme works or reduces defaults?

No. They are proposed indicators, not measured results, and have not been validated. A change between baseline and follow-up does not show the programme caused it, and the indicators are not a credit score or an input to one.

What are the main limitations of measuring resilience?

Recall weakens over time and survey wording can be misunderstood; attribution is hard without a comparison group; few participants face a disruption in any one period; seasons are not comparable; and the questions are sensitive, so consent, the option to decline and data minimisation matter.

Sources

  1. Master Direction – Reserve Bank of India (Regulatory Framework for Microfinance Loans) Directions, 2022 — Reserve Bank of India (14 March 2022; page shows updates to 17 July 2025)
  2. The Global Findex Database 2021, Chapter 3: Financial Well-Being (section 3.1 Financial resilience) — World Bank (2022)
  3. The Role of Financial Services in Building Household Resilience — CGAP (research by Freedom from Hunger) (April 2016)
  4. Measuring Financial Health: Not as Easy as It Looks — CGAP (4 April 2017)
  5. The 2023 Microfinance Index (summary page) — 60 Decibels (2023)
  6. Universal Standards for Social and Environmental Performance Management — Cerise+SPTF (February 2022 (third edition))

Links open the source in a new tab. Policy wordings and regulations change; check the current version before relying on any detail.