Finance & PSD Impact OCTOBER 2019 The Lessons from DECFP Impact Evaluations ISSUE 55 Our latest note looks at whether mobile money benefits households in poor and remote areas of Northern Uganda and finds large increases in non-farm self-employment and food security. Does Mobile Money Improve Livelihoods for Households in Poor and Remote Areas? Miriam Bruhn and Christina Wieser Digital financial services have spread During the roll-out, the professional rapidly in developing countries, enabling services firm activated 320 agents in areas in mobile phone users to make financial our study who undertook a successful first transactions on their phone, such as transaction. transferring money, buying airtime, and Airtel Money agents offer a variety of paying bills. services including cash withdrawals, cash Recent evidence suggests that mobile deposits, purchase of airtime, sending and money has helped households receive more receiving money, bill payment (such as remittances, smooth consumption in the face school fees and utilities), and payment for of shocks, and escape poverty in the long-run. goods and services. This evidence comes from areas were about 70 percent of households have mobile phones Study Design and Data and almost half or more receive remittances. With help from the Uganda Bureau of We designed a field experiment to study Statistics (UBOS), a sample of enumeration whether mobile money can also improve areas (EAs) was drawn from sub-counties in livelihoods in poorer and more remote areas. Northern Uganda that did not have an Airtel The experiment took place in rural areas of Money agent in 2015. To minimize potential Northern Uganda, where only 28 percent of spillovers of agents to control group EAs, the households owned a mobile phone and 15 selected EAs were mapped and grouped into percent received remittances. Areas in our clusters. A 0.5km buffer was drawn around sample also had low access to financial the boundary of each EA and EAs whose services, with the median distance to a bank buffers overlapped were grouped. Clusters branch being 25.2km. were thus at least 1km apart from each other. We randomly assigned 334 clusters of The Agent Rollout enumeration areas (EAs) to a treatment or a We collaborated with Airtel to roll-out control group, stratified by distance to a bank Airtel Money agents in Northern Uganda in branch. The Airtel Money agent rollout took locations not yet served by Airtel Money. place in treatment clusters between January Areas in our sample also had limited access and June 2017. The professional services to mobile money agents from other providers firm did not identify or activate agents in all (mostly MTN), with an average distance of treatment clusters, but close to half of these 8km to the closest agent. clusters received at least one agent. A professional services firm was hired to To analyze the impact of the agent assist with identifying potential agents for the rollout, we use data on about nine randomly rollout. This firm helped agents with the selected households in each EA, from a logistics of signing-up to become Airtel January 2016 baseline survey and a January Money agents and provided them with the 2018 follow-up survey. We also received necessary equipment, training, and marketing administrative data on mobile money materials. transactions from Airtel. Do you have a project you want evaluated? DECRG-FP researchers are always looking for opportunities to work with colleagues in the Bank and IFC. If you would like to ask our experts for advice or to collaborate on an evaluation, contact us care of the Impact editor, David McKenzie (dmckenzie@worldbank.org) Results shocks, such as a drought or flood, We analyze the effects of the agent rollout increasing the probability of taking work separately for areas close to a bank branch and reducing the probability of changing (less than 25.2km) and areas far from a bank the household’s diet. branch. All statistically significant effects • The agent rollout lowered the fraction of come from areas far from a bank branch. households with a very low food security • Here, the agent rollout led to cost-savings index from 63 percent to 47 percent. for remittance transactions. About 31 percent of remittance receivers in control Figure 2: Fraction of Households Working areas paid transportation costs for the in Non-Farm Self-Employment transaction (motorcycle or mini-bus taxis). This number decreased to 18 percent in the treatment group. • Similarly, thanks to the agent rollout, survey respondents were less likely to report high prices for remittance transactions or that the agent was far (figure 1). Figure 1: Problems with Sending or Receiving Money Policy Implications We conclude that mobile money can improve livelihoods even in very poor and remote areas. In line with previous studies, we find that mobile money increases remittance receipts. Moreover, in remote areas, mobile money also leads to non- negligible cost savings. In our sample, households who received remittances typically paid 4,000 Ugandan shillings (US$1) per transaction in transportation • Administrative data show an increase in costs, which represents about 10 percent of the likelihood of receiving peer-to-peer monthly per capita household expenditures. transfers due to the agent rollout. We do not find an effect of the agent • The agent rollout doubled the fraction of rollout on poverty. However, we measure survey respondents who work in non- follow-up outcomes only about six months farm self-employment (figure 2). after the agent rollout ended, whereas the • The agent rollout changed the way in previous literature finds a decrease in poverty which households respond to negative four years later. For further reading see: Wieser, Christina, Miriam Bruhn, Johannes Kinzinger, Christian Ruckteschler, and Soren Heitmann. 2019. “The Impact of Mobile Money on Poor Rural Households: Experimental Evidence from Uganda.” World Bank Policy Research Working Paper No. 8913. Recent impact notes are available on our website: http://www.worldbank.org/en/research/brief/finance-and-private-sector-impact-evaluation- policy-notes