Finance & PSD Impact MAY 2018 The Lessons from DECFP Impact Evaluations ISSUE 48 Our latest note compares banking an agent or the branch in peri-urban Senegal. The Pros and Cons of Agent Banking: Evidence from Senegal Sinja Buri, Robert Cull, Xavier Giné, Sven Harten and Soren Heitmann Can agent banking bring large numbers branch (Branch); ii) Info, Incentives and told of poor, unbanked individuals and to open it with an agent (Agent), (iii) Info, households into the formal financial system? No Incentives and Branch (iv) Info, No Agents are local retailers that double as Incentives and Agent and (v) No Info. lower-cost alternatives to bank branches and enable customers to more conveniently 2. Results make deposits, withdrawals, money All individuals in the study were offered transfers, and payments of loans. Because of the same account irrespective of the location the convenience associated with their closer they were encouraged to open it, and neither physical and social proximity to underserved location had deposit or withdrawal fees market segments, agents may find it easier during the study period. In addition, the to reach poorer customers living farther selected study areas were close to and from formal bank branches, and can thus equidistant from agents and branches alike. deepen financial inclusion more cost- Of the 2,201 individuals in the study, effectively than the traditional banking 501 opened the new account. Providing model. Clients may, however, perceive that information about the account increased the agents will be more likely to share probability of opening the account by 11.2 transaction information with friends and percent, an increase of almost 100 percent relatives, making it harder to refuse their relative to the 11.7 percent of individuals in cash demands. Relatedly, agents may refuse the control group that opened one. Monetary large transactions due to security or liquidity incentives increased take-up by 5.5 percent, concerns. roughly half the effect of information. We find that visits to the agent, 1. The study including transport, waiting and face to face This study seeks to estimate the effects time with the agent, were around 10 minutes of access to agent banking on opening and shorter than visits to the branch. usage of a savings account in urban and Transactions costs are thus lower when peri-urban areas in Senegal. A randomly banking with agents. Because of the lower selected group of individuals were given transactions costs of banking with the agent, information about a savings account. Half of individuals encouraged to open the account the individuals were encouraged to open the with the agent should visit the agent more account at the nearest agent, while the other often and have higher savings balances. We half were encouraged to do so at the closest find that individuals directed to the agent branch. In addition, half of the individuals in made 1.4 more deposits and 1.5 more each of these two groups were also given a withdrawals (from an average of 2.5 monetary incentive to open the account. deposits and 3.2 withdrawals, respectively, Study participants thus fall into one of five among those directed to the branch) during groups: (i) offered information about the the 12 months after the opening of the account (Info), monetary incentives to open account. In addition, average balances 12 it (Incentives), and told to open it at the and 24 months after account opening across 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) all individuals directed to the agent are 18 trust develops over time, they may increase and 34 percent higher from a base of 38 and the size of transactions and rely more on 30 USD among individuals encouraged to them. We find however, no positive trend in open the account at the branch. the number or size of transactions with the We should also expect that if individuals agent (relative to transactions at the branch) trust either location equally and are and thus conclude that trust cannot explain unconstrained by the amount of the the simultaneous visits to agents and transaction at either location, they will only branches. bank at one location, thus minimizing Second, the agent may refuse large transactions costs. Yet, individuals withdrawals due to lack of liquidity, or large encouraged to open the account with an deposits for fears of a robbery. While these agent visited the branch as often as those concerns may be valid in other contexts, encouraged to open the account at the liquidity and technological constraints are branch. As a result, the overall increase in not binding for Microcred agents. Finally, the number of transactions comes from clients may be reluctant to make a large visiting the agent more often. In addition, transaction with the agent because of when visiting the branch, they made privacy concerns as the agent may talk to transactions of roughly 219 USD (median is friends and relatives of the client about the 74 USD) compared to transactions of around transaction making it harder for the client to 96.50 USD (median is 31 USD) when fend off cash demands. In contrast, at the visiting the agent. Individuals sent to the branch the transaction is kept secret. In this agent therefore visited agents more often case, individuals would choose the branch and chose the location depending on the size for large transactions and the agent for of the transaction they planned to make. In smaller ones. Focus groups with clients and contrast, individuals sent to the branch evidence from lab-in-the-field experiments visited the agent less often and did not chose confirm this hypothesis. the location based on the size of the transaction. 3. Policy Recommendations So why didn’t clients bank exclusively These results have the following relevant with agents thus minimizing transaction implications: costs? There are three possible reasons. 1) Individuals are willing to bank with agents First, clients may trust branch tellers more and to trust them with their money. than agents, at least in the beginning. 2) Experienced clients choose to transact with Microcred Senegal, our partner institution, is either the agent or the branch depending on well-established in Senegal but until the size of the transaction. recently has relied only on its branch 3) This choice suggests that banking with network and thus individuals were not agents involves a tradeoff: agents have familiar with Microcred banking agents. In lower transactions costs but also lower this case, clients may start by making small privacy. transactions with the Microcred agent and as For further reading see: Buri, S., Cull, R., Giné, X., Harten, S. and S. Heitmann “Banking with Agents: Experimental Evidence from Senegal”, World Bank Policy Research Working Paper 8417, April 2018 Recent impact notes are available on our website: http://www.worldbank.org/en/research/brief/finance-and-private-sector-impact-evaluation- policy-notes