Resilience through Contingency Loans in Savings Groups: Evidence from Mozambique

The intervention provides selected community savings groups with access to an external emergency loan, which the group uses to establish an emergency loan fund for its members. The fund is managed by the savings group and used to provide short-term loans to members who experience unexpected emergencies, such as illness, theft, fires, or other shocks. The emergency loan fund operates according to rules defined and enforced by the group. The intervention is implemented at the savings-group level.

2025-09-24

2026-06-01

The primary outcomes of interest in this experiment are:
- Emergency credit use: access to and use of emergency loans by household members.
- Coping strategies in response to shocks: households’ behavioral responses to shocks, including both harmful and adaptive strategies.
- Household consumption and food security: levels and stability of household consumption and food security outcomes.
- Household financial position: household savings, loans, and overall financial resources.

The primary outcomes are constructed from multiple survey variables.

- Emergency credit use is measured using indicators capturing whether a household member received an emergency loan, the amount borrowed, and the source of credit, with a focus on loans obtained through the savings group.
- Coping strategies in response to shocks are measured using survey questions capturing a range of household responses following a reported shock. These include potentially harmful strategies such as selling productive or household assets, reducing or skipping meals, or cutting education or health expenditures, as well as adaptive strategies such as increased labour supply, entry into work by previously non-working household members, or temporary migration of household members. These indicators will be analysed individually and, where appropriate, combined into summary indices distinguishing different types of coping responses.
- Household consumption and food security are measured using total household consumption expenditures on food and non-food items, as well as standard food security indicators, including the Food Consumption Score (FCS) and the Food Insecurity Experience Scale (FIES), constructed following established definitions.
- Household financial position is measured using information on current household savings, outstanding loans and debts, and financial transfers received. Where appropriate, these measures may be aggregated into summary indicators of household financial resources.

All primary outcomes are measured at the household level using baseline and endline household surveys, supplemented by high-frequency phone surveys.

The secondary outcomes of interest in this experiment are:

- Shock exposure.
- Asset ownership.
- Income and labour force participation.
- Social capital and informal support networks.
- Women’s economic empowerment.
- Psychosocial well-being.
- Savings group–level outcomes, including loan repayment and fund utilisation.

Secondary outcomes capture intermediate mechanisms and broader dimensions of household and group functioning that may be affected by access to emergency credit.

- Shock exposure is measured using indicators capturing the occurrence, type, timing, and self-reported severity of shocks experienced by households over the reference period.
- Asset ownership includes the number and value of productive and household assets owned by the household.
- Income and labor force participation are measured using information on household income by source and labor force participation.
- Social capital and informal support networks are measured using indicators capturing trust, perceived support, collective action, and households’ ability to rely on others for financial or non-financial assistance in times of need.
- Women’s economic empowerment is measured using survey questions on perceived agency in household decisions.
- Psychosocial well-being is measured using self-reported indicators of stress, esteem, emotional well-being, and perceived ability to cope with hardship.
- Savings group–level outcomes are measured using group financial records and group leader surveys and include indicators such as loan repayment rates, frequency and volume of saving and lending, and utilisation of the emergency loan fund.

Secondary outcomes are measured using baseline and endline household surveys, high-frequency phone surveys, group leader surveys, and savings group financial records.

Savings groups were randomly assigned to treatment or control using a computer-based procedure implemented in Stata. To improve balance, groups were first stratified by geographic location and then grouped into small blocks of similar groups based on baseline household and group characteristics, including shock exposure, financial resilience, group size, and gender composition. Within each block, approximately half of the groups were randomly assigned to receive the intervention and half to the control group, yielding an overall split of 51 treatment and 52 control groups.

Not available

Randomization was conducted using a reproducible computer-based procedure by the research team.

Randomization is conducted at the savings-group level.

Yes

103 savings groups

Approximately 1,030 individuals in the main household survey sample, plus data from four rounds of high-frequency phone surveys, with up to approximately 2,000 phone interviews per round across all savings groups.

Clusters: 51 savings groups in the treatment arm and 52 savings groups in the control arm.
Main household survey: approximately 510 individuals in treatment groups and 520 individuals in control groups.
High-frequency phone surveys: four rounds per group, with up to approximately 1,000 per round in both treatment and control arms.

Accounting for the clustered experimental design at the savings-group level (103 clusters, approximately 10 households per cluster) and an intra-cluster correlation of approximately 0.10, the minimum detectable effect size (MDE) for the main household survey outcomes is approximately 0.18–0.22 standard deviations at 80% power and a 5% significance level. This MDE applies to key primary outcomes measured in the main household survey, including the use of harmful coping strategies, the total value of emergency loans in the last 12 months, the Food Insecurity Experience Scale (FIES), and the self-reported probability of raising 5,000 MZN in an emergency. These MDEs should be interpreted as a conservative lower bound, as the study additionally collects repeated observations through four rounds of high-frequency phone surveys, which substantially increase the total number of observations and statistical power for outcomes measured at high frequency.