Improving well-being through mobile money: a replication study in Niger

ABSTRACT

Around the world, there has been an exponential growth in cash-based programmes as part of social policies. Innovative payment mechanisms, such as mobile money, could have benefits for recipients. An experimental study suggests that the use of mobile money, Zap delivery, might improve intra-household socio-economic dynamics on Niger. Our replication study evaluates the impact of Zap delivery in contrast to more traditional delivery mechanism. We examined original author´s findings and estimated similar results about increasing household diet diversity without decreasing their durable and nondurable assets. The heterogeneity evaluation by age groups suggests that the Zap delivery had a different impact on older beneficiaries than younger ones. In addition, we evaluated the robustness, which considers multiple imputation and Lee bounds analysis; as well as a nutritional evaluation of children under 5 years using anthropometric measures.

KEYWORDS:

• Impact evaluation
• replication study
• mobile money
• cash transfer program
• nutrition

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here

Notes

1. According to the National Institute of Statistics of Niger (2014), these 96 communities belong to the Department of Tahoua, capital of Tahoua region, one of the country’s seven regions.

2. See https://sites.tufts.edu/jennyaker/research-2/ for the data sets.

3. The communes – Affala, Bambeye, Barmou, Kalfou, Takanamat and Tebaram – are representative of the communes of the Department of Tahoua. In Appendix H, we present additional information on Niger’s demographics.

4. The HDDS is the aggregation of 12 food groups: cereals, white tubers and roots, vegetables, fruits, meat, eggs, fish and other seafood, legumes and seeds, milk and milk products, oil and fats, sweets, and spices (condiments and beverages).

5. We identified two statistical parameters of the variable amount spent on children’s clothing for festivals that had rounding differences from those estimated by the original authors.

6. The original survey was applied to women who received the intervention, but we found one male in the analysis. We suppose this is due to the absence of a female receptor or a typing error.

7. We also evaluated quartile grouping (Appendix D). We did not find major differences in relation to our behavioural grouping analysis.

8. In the original study, 96 villages were stratified by commune and villages and were randomly assigned to the cash, mobile or Zap interventions (32 villages per group).

9. Although it was not a pre-specified analysis, we analysed heterogeneity by polygamy condition, but did not find important differences.

10. In Malawi, Uganda and Zambia, the poverty gap ratio for households with elderly members is 6–20 percentage points higher than the average ratio.

11. The fact that fruit and vegetables are an expensive source of energy is an important constraint for poor households.

12. The original study suggests that reducing the transfer’s observability by m-transfer could also affect interhousehold sharing, leaving more income available for the household (Jakiela and Ozier 2015).

13. In Table MI, the original results have 60 statistically significant coefficients, while MI suggests that two coefficients lost their significance. This represents 3 per cent of all original significant coefficients from Tables A4 to A10.

14. We refer to increases or reductions of significance if p-value increases or reduces at least 0.01.

15. Some recent and influential studies that use LB analysis to test for robustness are Hidrobo et al. (2014), Kremer, Miguel, and Thornton (2009), Cunha (2014) and Drexler, Fischer, and Schoar (2014).

16. The original authors documented that there is no evidence of differential attrition across the experimental arms.

17. Children from 0 to 5 months were not included.

18. About programs with negative impact, in Nicaragua, children in coffee-producing households receiving conditional cash transfers saw greater declines in height-for-age Z-score measure with a − 0.27 coefficient after treatment (Dammert 2008). A program in Brazil found that 6 months after benefits began to be distributed, beneficiary children were 0.13 z-scores lighter (weight-for-age) than excluded children (Morris et al. 2004). Houngbe et al. (2017) found no evidence that unconditional cash transfers reduced the incidence of wasting.

19. Some replication studies included heterogeneity analysis as part of TCA. In this sense, Atanda (2019) replication study evaluated the moderating factors arising from the heterogeneous implementation of the biometrically authenticated payment system across districts and its influence.

20. According to Andrews and Buchinsky (2000), we estimated the coefficients using bootstrap. The statistical significance of coefficient is maintained using 1498 repetitions.

Additional information

Funding

This study was funded by International Initiative for Impact Evaluation (3ie) Replication Programme on Financial Services for the Poor and Bill and Melinda Gates Foundation .

Notes on contributors

JP Meneses

Jean Pierre Meneses is a Physician and Master of Science at UPCH, and researcher at PUCP. As a Physician, his research interests lie in neurology, neuroscience, nutrition and impact evaluation on vulnerable populations.

ET Ventura

Edgar Ventura Neyra is Master in Economics, Professor at the UPC School of Economics and researcher at PUCP. His primary field of research is social policy, and impact evaluation focuses on financial inclusion initiatives

OA Elorreaga

Oliver A. Elorreaga is an Economist at PUCP and Master of Science in Epidemiological Research Sciences at UPCH. His primary research interests include social policy, impact evaluation, as well as innovative interventions to aid marginalized populations.

C Huaroto

Cesar Huaroto is a PhD student of Economics at Pontifical Catholic University of Chile. He works on economic development, using applied econometrics to assess social policies on poverty and natural resources.

GG Aguilar

Giovanna Aguilar has a degree of Doctor of Economics from PUCP, and Master´s degree in Economics from PUC-Rio. She is Full Professor and researcher at the PUCP´s  Department of Economics. Her work focuses on microfinance, development issues and the design of financial inclusion strategies.

EP Beteta

Edmundo Beteta is Master of Arts of Economics at Georgetown University, Professor and researcher in Economics at PUCP, whose research includes health economics, government and public policy, along with evaluation methods.