NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Complementarities in microcredit and financial services interventions for financial inclusion and empowerment: evidence from randomized evaluation in South West Nigeria
Initial registration date
February 05, 2018
February 05, 2018 1:28 PM EST
Initiative for Evidence Based Development and Empowerment
Other Primary Investigator(s)
Additional Trial Information
The intervention being evaluated is the Amoye Digital Microcredit Initiative for Women Empowerment, in collaboration with a microfinance provider; Amoye Microfinance Bank, Ikere Ekiti Nigeria. The intervention offers microcredit alone and microcredit combined with digital banking services to clients. The key objective of study is to assess the effectiveness of the interventions, paying particularly attention to whether microcredit combined with digital banking services offers incremental impacts over what the demand side alone (microcredit) would have offered. The main identification strategy is a randomized control trial, involving random allocation of eligible beneficiaries into two treatments: microcredit only, microcredit + digital financial services, and a control group. The experimental design to examine spillover effects follows a clustered Randomized Saturation design. At least 75 clusters per block are required to generate sufficient power (min 0.80) and effect size (0.20) to examine the main research questions, leading to a total of 375 clusters. Findings will provide evidence informing Nigeria’s national financial inclusion and gender empowerment policy.
Alzua, Maria and Damilola Olajide. 2018. "Complementarities in microcredit and financial services interventions for financial inclusion and empowerment: evidence from randomized evaluation in South West Nigeria." AEA RCT Registry. February 05.
Alzua, Maria, Damilola Olajide and Damilola Olajide. 2018. "Complementarities in microcredit and financial services interventions for financial inclusion and empowerment: evidence from randomized evaluation in South West Nigeria." AEA RCT Registry. February 05.
The intervention being evaluated is the Amoye Digital Microcredit Initiative for Women Empowerment, in collaboration with a microfinance provider; Amoye Microfinance Bank, Ikere Ekiti Nigeria (AMfB). The intervention offers small microloan alone, and microloan combined with access to digital banking services to clients. The key objective of the complementary interventions is to enhance economic empowerment amongst women microentrepreneurs through increased financial inclusion, in line with Nigeria’s national policy of significantly reducing the population of women excluded from the financial system. The beneficiaries are women microentrepreneurs in Ikere community, Ekiti State Nigeria. The implementing bank is using the evaluation as part of a learning process and strategy to reaching the largely unbanked / underbanked women microentrepreneurs in Ikere community.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
The primary outcomes are at the beneficiary level for economic empowerment, and household level outcomes for reduction in household vulnerability.
The main indicators for economic empowerment include Financial inclusion; Decision making in hhold; Expenditure on hhold members; Contribution to hhold expenses (e.g. education, health), Peer networking / community participation; and Consumption expenditure (durable assets). Indicators for household vulnerability includes Household income; Savings; Child work; Food shortage in household; Health services demand; and Migration from household; and Copying with shocks. Indicators for institutional level outcomes include; Usage of digital banking services; Intensity of usage; Nature of usage; and Type of income generating activities.
Primary Outcomes (explanation)
Each of the primary outcomes will be constructed using scoring from the response to specific questions on the relevant variables in the survey:
(i) Financial inclusion will be constructed using scoring from the response to specific questions including, ownership of bank accounts, saving, borrowing, payments, use of digital banking and financial services, insurance. The higher the points scored the more financially included.
(ii) Economic empowerment will be constructed from variables such as contribution to household income, decision making in household, ownership of productive asset, contribution to household income/expenditure, and networking, community activities & self-confidence. The higher the score, the more empowered. Financial inclusion can also be included as an empowerment measure. (iii) Household vulnerability will be constructed from variables such as exposure to shocks and type, child work, health service demand, household asset, and food shortage in household. Lower scores indicated less household vulnerability. *Sometimes, financial inclusion can also be considered as an indicator for economic empowerment,
Secondary Outcomes (end points)
The secondary outcomes are mainly institutional design outcomes, such as interest rate, repayment, loan tenure.
Secondary Outcomes (explanation)
Secondary outcome indicators will be mainly for qualitative analysis, focusing main mainly on indicators for program design and implementation; and beneficiary perceptions of key findings.
The experimental designed is the randomized saturation (RS) design developed at the World Bank by Baird, Bohren, Mcintosh, and Ozler (2016). It is a two-level clustered-type randomization design where within-cluster assignment to treatment follows some degree of intensity (saturation). The RS design approach is adopted primary to enable us measure the nature and size of community level spillover effects potentially arising from the microcredit.
(i) Baird, S., J.O. Bohren, C. McIntosh, and B. Ozler (2016). Optimal Design of Experiments in the Presence of Interference. Poverty and Applied Microeconomics Seminar Series, World Bank.
Experimental Design Details
Following the suggestions by Baird, Bohren, Mcintosh, and Ozler (2016), who helped us review our design, the specific RS design to use is the partial population experiment, in which within-cluster treatment saturation at 50% is optimal. The PPE approach is selected for the following reasons;
(i) It allows identification of both the Intention-to-treat and spillover effects from treatment assignment on the untreated;
(ii) It minimizes the loss of power arising from trying to estimate the variance of treatment intensities;
(iii) The PPE at the ICC 0.25 best captures the implementation plans of the implementing bank; and
(iv) based on previous results, it is believed that much of the variations in outcomes will be between clusters more than within clusters, such that we might not see significant differences across smaller levels of saturation.
The randomization method will be done in office using STATA codes. The randomization procedure follows three-stages. The first stage involves random assignment of the entire Z clusters (women groups) into N overall treatment and Y control clusters, respectively. In the second stage, within each block, the N overall treatment clusters undergo further random assignment into equal-sized Treatment arm 1 (T1) (microcredit only, T1=N/2 clusters), Treatment arm 2 (T2) (microcredit + digital banking services, T2=N/2 clusters), Treatment arm 0 (T0) (pure control group, T0=Y clusters). The final stage involves within-cluster random allocation of beneficiaries into 50% saturation (treatment intensity) for both T1 and T2. Each of the randomization procedure will take place in each of the five fixed blocks that the community is divided.
Women groups (clusters) are the unit of randomization. As described above, the first two-stages of the randomization procedure are group level randomization. The third level will be within-cluster randomization for treatment saturation is an individual level randomization. Each of the randomization procedure will be undertaken in each of the five fixed blocks that the community is divided.
Was the treatment clustered?
Sample size: planned number of clusters
375 women groups, comprising 35 women groups within each block.
Sample size: planned number of observations
There will be at least 3,750 women microentrepreneurs. Each women cluster comprises at least 10 women members.
Sample size (or number of clusters) by treatment arms
Treatment arm 1 (T1) = 1,000 beneficiaries allocated into 100 women groups to receive microloan alone.
Treatment arm 2 (T2) = 1,000 beneficiaries allocated into 100 women groups to receive microloan + access to digital banking services.
Treatment arm 0 (T0) = 1,750 beneficiaries allocated into 175 women groups as pure control.
Total sample size=3,750
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The effect size to detect under the specific RS design adopted uses the Graphical User Interphase (GUI) software developed by Baird, Bohren, Mcintosh, and Ozler (2016: p.23). Household income was used as the main outcome. Given the necessary parameters, the minimum detectable effect for the Intention to treat (ITT), min_MDE_T and the minimum detectable effect for the spillover effects, min_MDE_S were obtained for the estimated sample size and estimates of within-cluster variation, ICC = 0.25, obtained from relevant outcomes from previous studies. Also, both effect sizes detectable were given equal weights (weight=1), the min_MDE_T and min_MDE_S are 0.18, respectively. These correspond to the optimal share of the clusters at ICC=0.25, where n=1,750 and n=2,000 eligible beneficiaries to the control and two treatment arms, respectively. Also, the estimated minimum effects sizes detectable were consistent with the expected 0.20 effect size under the general design.
INSTITUTIONAL REVIEW BOARDS (IRBs)
Ethics Advisory Committee, Ekiti State University Nigeria
IRB Approval Date
IRB Approval Number
Post Trial Information
Is the intervention completed?
Is data collection complete?