ABSTRACT
M-PESA, a mobile phone-based technology for transferring money provides a gateway to formal financial services for populations who otherwise would not have access. This study re-examines Jack and Suri's 'Risk Sharing and Transaction Costs' paper. With a few minor differences, the results are robust to consistency tests and sensitivity analyses. Since rural households are expected to particularly benefit from M-PESA, the heterogenous effects have also been explored by comparing benefits across urban and rural residents. The findings reported here and in the original study provide strong empirical evidence that M-PESA has had a positive impact on people’s financial health. Such financial benefits derived from market-based mobile money innovations can be vital in combating world poverty.
Acknowledgments
This paper is part of the 3ie Replication Program, Replication Window 4: Financial Services for the poor, funded jointly by the International Initiative for Impact Evaluation (3ie) and the Bill & Melinda Gates Foundation. The author would like to thank the funders for their financial support. This study has benefitted from discussion with Professor W. Robert Reed, Benjamin DK Wood, Eric Djimeu and Scott Neilitz. I am grateful to the original authors, Professor William Jack and Tavneet Suri, who provided me with their code and data and were helpful in answering my questions.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1. The Sustainable Development Goals are a collection of 17 global goals to transform our world by the United Nations.
2. M-PESA: M stands for mobile and pesa is Swahili for money.
3. For additional materials including data and codes provided by original authors see: https://www.aeaweb.org/articles?id=10.1257/aer.104.1.183.
4. For the details of pure replication visit: https://www.3ieimpact.org/evidence-hub/publications/replication-papers/risk-sharing-and-transaction-costs-replication-study. Codes to replicate this study is available at: https://dataverse.harvard.edu/dataverse/3ie
5. There are two households with missing values in one round. I dropped these households (four observations) from the sample.
6. These data sets are available at: https://dataverse.harvard.edu/dataverse.xhtml?alias=mobilemoney
7. Total expenditures are defined as the sum of weekly, monthly and yearly expenditures.
8. It might be acceptable not to have monthly and/or yearly expenditures, but it seems unusual not to spend on foods. Households with zero or missing weekly expenditures are as follows: 1428005, 1433001 and 1280060 for zero values and 1603020, 1295003 and 112118 for missing values.
9. Due to the large frequency of zeros compared to the other values, the bar charts in the right-hand side of (b) are not obvious here.
10. The reason why I also report the results corresponded to the Tobit without controlling for left-censoring is to show that the estimated effects are not far from the original results, once the left-censoring issue is taken into account. It is worth mentioning that these estimates are not particularly large if they are compared to the average size: the mean values for ‘Number received’ and ‘Total received (square root)’ are 2.067 and 68.50, respectively.
11. Due to the limited coverage of cell phone towers and M-PESA agents, the residents of the north and northeast parts of the country were excluded from the sample.
12. Nairobi is the capital and largest city of Kenya.
Additional information
Funding
Notes on contributors
Nazila Alinaghi
Dr. Nazila Alinaghi is a Post-doctoral Research Fellow for the Chair in Public Finance at Victoria University of Wellington, New Zealand. Her research focuses on public economics, labour economics, and meta-regression analysis.