A NEW BANKING MODEL FOR AFRICA: Lessons on digitization from four years of operations By Christian Rodriguez, Julia Conrad, Gisela Davico, Susie Lonie, Lesley Denyes Table of Contents Executive Summary............................................................................................................ 5 Overview............................................................................................................................ 6 Study Background............................................................................................................... 8 Lessons Learned.................................................................................................................. 11 (I) Planning a Digital Financial Service (DFS) rollout............................................................. 11 Objectives .............................................................................................................................. 11 Regulatory environment ........................................................................................................ 12 Business case ........................................................................................................................ 12 Approach .............................................................................................................................. 15 Roll-out................................................................................................................................. 16 Technology ............................................................................................................................ 17 (II) Managing a DFS........................................................................................................... 18 Agent recruitment ................................................................................................................. 18 Network growth.................................................................................................................... 19 Agent incentives .................................................................................................................... 21 Smart tools ........................................................................................................................... 22 Liquidity management ........................................................................................................... 24 (III) Promoting DFS usage and activity ............................................................................... 26 Change management ............................................................................................................ 26 Customer activity .................................................................................................................. 27 Product design ...................................................................................................................... 30 (IV) Ensuring DFS sustainability.......................................................................................... 30 Sustainability ....................................................................................................................... 30 Geographic expansion ............................................................................................................ 33 Innovation ............................................................................................................................ 33 Savings mobilization............................................................................................................... 33 Present progress and outlook............................................................................................. 34 Conclusions....................................................................................................................... 39 Annex ............................................................................................................................... 41 2 TABLES Table 1: DFS definitions........................................................................................................................ 4 Table 2: Agent liquidity management solutions observed as part of study.............................................. 25 FIGURES Figure 1: Population above 15 years-old in millions, 2017......................................................................... 8 Figure 2: Gross National Income per capita 2017 (Atlas Method, in US dollars)......................................... 8 Figure 3: Life expectancy at birth in years.............................................................................................. 8 Figure 4: DFS market overview as of December 2017.............................................................................. 9 Figure 5: DFS approaches of study institutions as of December 2017...................................................... 10 Figure 6: Median customer and agent activity rates over course of the study......................................... 13 Figure 7: Average cash-in and cash-out amounts in US dollars over course of the study........................... 13 Figure 8: Agent median cash-in versus cash-out shares over course of the study.................................... 14 Figure 9: Branches (right y-axis), non-HQ staff and agent network size (both left y-axis) over data reporting period......................................................................................... 16 Figure 10: Monthly agent network growth rates over the operational lifetime...................................... 20 Figure 11: End-month values of active customers per active agents in 2017 ............................................ 20 Figure 12: End-month values of average transaction volume per active agent in 2017 in US dollars............................................................................................................................ 21 Figure 13: End-month values of average commission income volume per active agent in 2017 in US dollars............................................................................................................................ 22 Figure 14: Example of a daily operations dashboard.............................................................................. 23 Figure 15: Share of DFS transactions over data reporting period ........................................................... 26 Figure 16: Customer registration and activity (in 1,000s) over data reporting period .............................. 28 Figure 17: Shares of DFS transactions as of December 2017.................................................................... 34 3 Table 1: DFS definitions1 DFS Definition A network of banking agents. More generally, an agent is any third-party acting on behalf of a financial institution or a non-bank institution to deal directly with customers, under contractual agreement. Here, agents are contracted by a bank (bank agents) to provide services on their behalf, most importantly cash-in and cash-out. Agents may (if permitted under local regulations) engage sub-agents to carry out activities on behalf of the financial institution. • Proprietary versus non-proprietary (third party): A proprietary agent network is a network the financial institution recruits, equips, manages, and brands itself. A non-proprietary (third party) agent network is a network that is owned, equipped, managed, and branded by an external partner, such as a Mobile Network Operator or other payment network.2 AGENT • Exclusive versus non-exclusive: An exclusive agent serves only one Digital Financial Service provider. An NETWORK agent who serves more than one DFS providers is non-exclusive. • Dedicated versus non-dedicated: Dedicated agents only conduct DFS business for one or multiple providers. More common though is that agents also run other kinds of businesses, known as non- dedicated agents. • Static versus roving (mobile): A roving (or mobile) agent is (part-time) moving through villages, market areas, etc. It is a “doorstep service” that serves customers where they live and work rather than making them find the nearest static agent outlet. The service is especially applicable to the collection of daily savings from market traders. Mobile banking (or m-banking) is a service provided by a bank or other financial institution that allows its customers to conduct conventional financial transactions remotely using a mobile device such as a simple feature phone, smartphone or tablet. • USSD versus application: Unstructured Supplementary Service Data (USSD) technology is similar MOBILE to Short Messaging Service, but, unlike SMS, USSD transactions occur during the session only. USSD BANKING is accessible from any type and model of mobile phone and therefore is currently the best available technology to deliver mobile financial services to low-income customers. Mobile banking via application requires a smartphone or tablet with internet access – which low-income populations often still lack. Internet banking (or e-banking) is an electronic payment system that enables customers of a bank to conduct INTERNET a range of financial transactions through the financial institution’s website. BANKING A debit card is a bank card that allows bank account holders to pay third parties directly from their account DEBIT CARD balances electronically. If the available funds on the account are insufficient, the transaction is not completed. Unlike a debit card, a prepaid card is not linked to a bank account. Generally, when a customer uses a prepaid PREPAID card, he or she is using money that has been loaded onto the card in advance. CARD 1 See AFI “Digital Financial Services Basic Terminology” 2016: https://www.afi-global.org/sites/default/files/publications/2016-08/Guideline%20Note-19%20 DFS-Terminology.pdf 2 See the Mastercard Foundation & BFA “Alternative Delivery Channels for Financial Inclusion: Opportunities and Challenges in African Banks and Microfinance Institutions” 2016: https://mastercardfdn.org/wp-content/uploads/2018/06/BFA_ADC_FIpaper_April2017-Accessible.pdf 4 EXECUTIVE SUMMARY The ongoing Fourth Industrial Revolution is a technological By the end of the study period, the participating institutions transformation that is changing the way we live, work and had overcome many external and internal challenges and were communicate. It is altering every aspect of our society and successfully running their digital channels. Channel maturity economy, including the financial sector. In Sub-Saharan Africa, varied among institutions. Five of the financial institutions 44 percent of the population subscribed to mobile services in expanded their original DFS scope to explore other service 2017. By 2025, the number of subscribers is expected to grow offerings. Lessons from the study centered around four main to 52 percent, and 87 percent of those subscribers are expected themes: strategy and business case, staff buy-in, data, and to have mobile broadband access.3 Results of the 2017 Global DFS management. The research team also documented the Findex survey reveal significant progress in financial inclusion impact of DFS on outreach and banking operations. driven by a new generation of financial services accessed through mobile phones and the internet. Still, with 57 percent Study results show that digital strategies are dynamic, of its population lacking any form of bank account, Sub- requiring constant readjustment based on client feedback Saharan Africa remains the region with the greatest potential and changes in market conditions. Although all FIs had a for the adoption of Digital Financial Services (DFS).4 compelling business plan and strategy before developing their DFS, these had to be constantly fine-tuned and From 2014 to 2018, IFC and the Mastercard Foundation adopted to successfully grow their businesses. The original conducted a longitudinal study with nine partner financial DFS assumptions, in particular around client outreach and institutions (FIs) in seven Sub-Saharan African countries uptake, had to be adjusted to the realities of financial services to understand if Digital Financial Services are a viable providers. Moreover, a digital strategy requires internal strategy for these institutions to expand financial access. support from staff, and a financial institution has to define More precisely, the study explains the strategic objectives ways to overcome their clients’ initial resistance and fear of of the institutions and describes how they planned and going digital. The study also revealed that most successful implemented their DFS rollouts. The study aimed at extracting DFS implementations used strong data-driven approaches to lessons on implementing internal change management, as monitor and assess DFS operations and they are using those well as measuring the impact DFS has on business growth, insights to refine products and services, thereby improving sustainability, outreach and adoption. The study is unique customer service and the overall experience. That said, given as it identifies valuable benchmarks on DFS implementation that most FIs in the study started their DFS offering from that fit the banking context, which differs from many scratch, internal capacities needed to be built and the costs studies in the existing literature that focus on MNO-led DFS for that development were often higher than anticipated. The implementations. It serves as a guide for target-setting and institutions also had to learn how to assess and work with strategic engagement of FIs seeking to implement or scale external partners (i.e. mobile network operators, technology digital channels. The study also differentiates itself by focusing companies, among others). For several of the participating FIs, on many institutions over multiple years and countries. managing partnerships was a challenge. With respect to agent banking, FIs saw the importance of prioritizing quality over quantity, and providing the right incentives to network agents. 3 See GSMA “The Mobile Economy Sub-Saharan Africa 2018”: https://www.gsma.com/mobileeconomy/sub-saharan-africa/ 4 See World Bank Global Findex Database 2017 “Account (% age 15+)”, retrieved from: https://globalfindex.worldbank.org/ 5 OVERVIEW From 2014 to 2018, IFC and the Mastercard Foundation conducted a research study on the implementation of Digital Financial Services with nine financial institutions across The Longitudinal Study on Digitizing seven markets in Sub-Saharan Africa. The objective was to Distribution Channels understand implementation strategies, the effect of digital channels on business models, and the impact of digital The Partnership for Financial Inclusion is a $37.4 million transformation on the institutions. This report shares the joint IFC-Mastercard Foundation initiative to expand findings and best practice insights gained from the research. and advance Digital Financial Services in Sub-Saharan Africa. Under this Partnership, the Longitudinal Study is All nine FI partners successfully implemented digital channels a four-year cross-country applied research project that over the study period, some starting earlier than others. documents the experience of nine financial institutions Although all the institutions faced significant external in seven Sub-Saharan Africa countries. In identifying challenges (e.g. regulation, market context), as well as internal best-practices, the study also supports the FIs towards challenges (e.g. staff resistance, technology incompatibility), implementing digital delivery channels (agent networks, by 2018 the first FIs had outsourced significant shares of their mobile banking, debit cards, etc.) as part of their business transactions to the digital channels. The institutions with operations. more mature channels are reporting that they are on their way to reaching financial sustainability and are already covering The study employs a mixed-methods research approach, agent commission costs with generated fee income. Several combining quantitative and qualitative data collection institutions are expanding the initial strategic scope of DFS to at the institutions with structured interviews and user explore additional DFS offerings. feedback: The study identifies a number of lessons from FIs with • Qualitative questionnaires: Between Q2 2014 established digital channels and presents best-practices for and Q1 2018, a research team conducted four financial institutions planning new digital channels. The key rounds of visits at the nine bank partners in Sub- findings of the study are: Saharan Africa. Structured interviews guided all meetings with the FI management and department • Digital channels provide a viable business model for representatives and were later adapted to the microfinance Institutions (MFIs). individual context of each participating institution. • Digital channels are effective for reaching new customers These interviews touched on multiple aspects of and expanding financial inclusion. DFS implementation, such as strategy, regulation, • FIs should draw on strategies, best-practices and partnerships, markets and products, distribution, benchmarks that are appropriate for the banking sector technology, risk management, business model, and to successfully implement digital channels. Business other operational considerations. models and benchmarks established from other sector actors, such as mobile network operators (MNOs), are not • Quantitative data: Eight institutions that replicable in a traditional financial sector context. implemented digital delivery channels provided • Change management strategies are vital to support quarterly data on outreach, transaction flows, internal and external organizational changes. staffing, fee revenue and operating costs of the respective channels. Although nine participated in the study, one institution delayed its DFS implementation to Q4 2017 and comparative data was therefore not collected at this time. All data was collected from January 2015 to December 2017, with a later reporting start of FIs that launched their DFS after January 2015. An analysis of the data describes how digital channel implementations have evolved and affected the overall business of institutions over time. The team also conducted an assessment of how initial digital channel business model assumptions had materialized by the end of the study timeline. 6 • Additionally, the team interviewed agents and • Research Report Aligning Expectations - The customers from the FIs, as well as savings and/or loan Business Case for Digital Financial Services: groups to incorporate the perspectives of DFS users. This research report provides a complete set of DFS financial modeling benchmarks, based on the study’s The study has produced a number of intermediate findings. It is addressing one of the main challenges publications, in addition to this report, which focus for the industry to date, which is the lack of on specific areas of business strategy, best-practices sufficient information on appropriate benchmarks. and risk areas that financial institutions face when Realistic expectations for a DFS solution are built implementing Digital Financial Services: from a digital strategy to guide development of the business case and a financial model to assess the • Breaking Free of the Branches - Microfinance long-term viability of the project. and Alternative Delivery Channels in Sub- Saharan Africa: This field note is a recap of findings • Changing Change Management - Adapting from the first round of interviews the research team Internal and External Culture in Times of conducted after most FIs had launched their DFS Digital Transformation: As for the effects of with plans to fully integrate the new services into DFS implementation on staff and customers, the the institutions’ operations. Major challenges to a study shows that managing change is not easy successful DFS rollout in this initial phase included and requires dealing with emotional responses and regulation and partnerships – which delayed the initial resistance. This field note discusses internal project in many cases – staff capacities, technical and external aspects of a digital transformation limitations as well as risk and agent liquidity and how institutions should address these in their management. change management strategy. • Turning FI Digital Strategies into Reality: This field note shares key strategic lessons from the study that FIs should consider before embarking on a DFS strategy: First, greater outreach has shown to be faster and easier to achieve than deposit mobilization, and strategic objectives evolve over time. Second, overall buy-in of staff is key, particularly when staff is in direct contact with clients. Third, agent network planning should focus on high-quality active agents5 vis-à-vis a large number of agents. Lastly, the study found that the cost of an agent transaction is 25 percent less than the cost of a branch transaction. 5 An active agent has completed one value transaction within 30 days. As transaction counts cash-in or cash-out from customer account, P2P payment, bill payment, etc. Balance inquiries, PIN resets and other transactions that do not involve the movement of value do not qualify as value transaction. 7 STUDY BACKGROUND Over the past ten years, the introduction of mobile money and The study sought to examine how digital channels impact agent banking has transformed the financial sector in Sub- the ability of FIs to expand access, scale and outreach Saharan Africa and greatly contributed to the expansion of the without relying on traditional branch infrastructure as well financial inclusion rate from 23 percent in 2011 to 43 percent in as the impact of DFS on overall FI operations. It identifies 2017. The promise of new technology and innovative business both the efficiencies and challenges that these services models has attracted a range of market actors to the evolving bring to such institutions. The study extracts lessons Digital Financial Services space in attempts to find sustainable learned and establishes best-practices that can be used ways of serving the mass market. and adapted by other institutions in different markets, providing benchmarks around performance, costs and The objective of the Longitudinal study was to understand if benefits of DFS implementations for the microfinance Digital Financial Services are also a viable strategy for FIs to sector as well as the general banking industry (Figure 1 reach financially excluded populations. to 4 give a short overview of general – and DFS market statistics on the study markets). Cameroon DRC Madagascar Nigeria Rwanda Senegal Tanzania Figure 1: Population Cameroon abobe 15 Madagascar DRC years-old in millions, 2017 Nigeria Rwanda Senegal Tanzania Population Below Poverty Line (using $1.90 a day as per 2011 PPP) Figure 1: Population abobe 15 years-old in millions, 2017 Population Below Poverty Line (using $1.90 a day as per 2011 PPP) Tanzania Tanzania Senegal Senegal Rwanda Rwanda Nigeria Nigeria Madagascar Madagascar DRC DRC Cameroon 0 20 40 60 80 100 120 140 160 180 200 Cameroon Note: Using poverty headcount ratio at national poverty lines (percent of population); reporting years are 2009 for Nigeria, 2011 for Tanzania, 0 Madagascar 2012 for Senegal, 20 and DRC,402013 for Rwanda, 60 2014 for 80Cameroon. 100 120 140 160 180 200 Figure 2: Gross National Income per capita Figure 3: Life expectancy at birth in years 2017 (Atlas Method, in US Dollars) 65,7 National Figure 2: Gross $2 100 Income per capita Figure 3: Life expectancy at birth in years 2017 (Atlas Method, in US Dollars) 66,7 $1 370 $1 240 67,1 65,7 $2 100 $910 $720 53,4 66,7 $460 $400 $1 370 $1 240 65,9 67,1 $910 $720 53,459,6 $460 $400 58,1 65,9 59,6 58,1 Figure 4: DFS market overview as of December 2017 Note: Reporting years are 2015 for Tanzania, 2016 for Senegal, Figure 4: DFS market overview as of December 2017 Madagascar, DRC, Cameroon and Nigeria, 2017 for Rwanda. 8 65,7 $2 100 66,7 $1 370 $1 240 67,1 $910 $720 53,4 $460 $400 65,9 59,6 58,1 Figure 4: DFS market overview as of December 2017 Senegal Rwanda Mobile subscriptions: 50% Mobile subscriptions: 46% Mobile internet users: 23% Mobile internet users: 21% M-money accounts: 32% M-money accounts: 31% FI accounts: 20% FI accounts: 37% Nigeria Tanzania Mobile subscriptions: 49% Mobile internet users: Mobile subscriptions: 38% 24% Mobile internet users: 14% M-money accounts: 6% M-money accounts: 31% FI accounts: 39% FI accounts: 21% Madagascar Cameroon DRC Mobile subscriptions: 27% Mobile subscriptions: 49% Mobile subscriptions: 35% Mobile internet users: 12% Mobile internet users: 26% Mobile internet users: 14% M-money accounts: 12% M-money accounts: 15% M-money accounts: 16% FI accounts: 10% FI accounts: 27% FI accounts: 15% Notes: Mobile subscription is the market penetration rate for unique subscribers calculated as total subscribers at the end of the period and expressed as a percentage share of the total market population. Unique subscribers are the total unique users who have subscribed to mobile services at the end of the period, excluding M2M. Subscribers differ from connections such that a unique user can have multiple connections. Mobile internet is the share of total unique users who have used internet services on their mobile device(s) at the end of the period. Mobile internet services are defined as any activity that consumes mobile data (i.e. excluding SMS, MMS and cellular voice calls). M-money account - and Financial institution account ownership only consider adult population, i.e. population of 16 years and older. Data sources are GSMA Intelligence database and World Bank Global Findex database 2017. The nine institutions that participated in the study represent a digital channel has been operating for more than three years) variety of characteristics with regards to market share, sector or young (if the operational life is less than three years). If not maturity and regulatory frameworks. Notably, DFS strategy otherwise stated, all data presented in this report does not and operational time frames differ as each institution started include DFS in pilot stage. Most of the banks that participated channel implementation at different times and progressed in the study have also developed or plan to develop mobile respectively. By the end of the study, all agent network banking systems to complement the agent banking channel. solutions were live at various stages of maturity: two network Some institutions have adapted DFS solutions to popular, solutions were more than five years old, three were operating informal, market solutions, e.g. mobile savings and loan for about three years, two were in their initial roll-out phases, repayment collection or integration of banking agents into and two were in pilot stages. For this report, we have classified group lending. the digital implementations in two categories: mature (if the 9 Figure 5: DFS approaches of study institutions as of December 2017 Rwanda Senegal Agent Network Agent Network • propriety & static, since Q3 2014 • propriety & static, since Q2 2013 • “agent kiosk”, since Q4 2017 Mobile Banking Mobile Banking • USSD, since Q2 2013 • app only, in planning Tanzania Nigeria Agent Network Agent Network • propriety & static, since Q2 2015 • via 3rd party provider, since Q3 2015 Mobile Banking Mobile Banking • USSD & app, since Q4 2014 • USSD & app, since Q3 2015 Debit Cards • since Q4 2016 DRC Madagascar Agent Network • propriety & static, since Q2 2017 Mobile Banking, Internet Banking, Agent Network Agent Network Debit Cards propriety, static & roving, since 2011 • propriety & static, pilot since • in planning Mobile Banking Q4 2017 USSD & app, since Q3 2017 Prepaid Cards Cameroon Agent Network • propriety & static, since Q1 2015 Mobile Banking • app only, in planning Agent Network • with cash-less branches, proprietray & static, since Q4 2016 • mobile collectors, since Q3 2016 Mobile Banking • in planning 10 LESSONS LEARNED Over the course of the four-year research period, the ethnographic research, surveys, and/or analysis of secondary team has extracted many valuable lessons, actionable data. It may not rely solely on the institution’s marketing or recommendations, as well as performance benchmarks on research department, and can include frontline cashiers, DFS deployment in the banking context. This final research loan officers and other roles. Depending on the availability of report gives a comprehensive summary of key learnings resources, market research activities may also be partly or fully as of the end of the study. It has been written to provide outsourced to an external company or consultant. practical guidance to institutions leveraging DFS for greater financial inclusion and is organized according to these four With regards to internal factors, an institution should consider phases of DFS rollout: reviewing the following areas: (I) Planning • The human resources capacity of the institution to put (II) Managing in place adequate project management skills, market (III) Promoting usage and activity research, technology, and data analysis that is necessary (IV) Ensuring sustainability to implement a DFS solution. • The ability of the institution to transform itself. The digital (I) LESSONS LEARNED: PLANNING project might require staff to learn new skills or become A DFS ROLLOUT familiar with basic DFS concepts. The ability to train and retrain staff requires implementing capacity building programs, communication strategies, and change Defining DFS objectives and a business management plans to guide existing staff through the case through analysis of internal factors digital transformation. and the market environment • The durability of legacy infrastructure and its readiness It is important to define appropriate and compelling business to support new digital channels. The capacity of the IT objectives that Digital Financial Services can deliver, e.g. department must be assessed in terms of the enterprise reaching customers in new geographic areas, building better systems and IT infrastructure necessary to support operational efficiencies, mobilizing deposits, or enhancing the digital channel in terms of core banking systems, customer experience. Those objectives should be backed by a integration services, hardware (servers, mobile devices, realistic and achievable business case.6 Point-of-Sale devices), and network communications, for example. Costs related to improving existing systems or To achieve this, an institution has to conduct a thorough acquiring new technology should also be considered, as analysis of external factors, such as market research on these can become significant CAPEX and OPEX items. potential and existing customers and the overall market environment where the DFS is to be deployed (i.e. size, Only a few FIs in the study conducted comprehensive market competition, country infrastructure, regulatory framework); research activities, encompassing both internal and external as well as internal factors such as staff capabilities and factors, for the design of their DFS strategy and business legacy systems. The results of this exercise should help in case. The research helped those FIs to have more realistic the formulation of recommendations and assumptions for expectations in terms of potential market uptake as well as the initial DFS strategy design and work planning stage. For client needs and expectations for the digital channel. Most example, market research results on (potential) customer institutions in the study relied heavily on internal information pains can ensure that (1) the DFS responds to a particular in combination with client focus groups, competitor analysis market need, (2) the DFS value proposition is in line with what or mystery shopping. Only one institution externalized the the market expects, and (3) that the market size assumptions market research to a specialized company. Some institutions for the financial modeling are built on the best available also leveraged publicly available information about mobile information and a solid rationale. All assumptions and their money services for building their DFS business case. Using rationales should be well documented in the DFS business information from mobile money services proved not useful for case.7 building assumptions for the DFS business case as it tended to mislead FIs in terms of digital channel size, customer The market research effort is a vital component when building acquisition growth, and transaction volume. a DFS strategy. It can include different quantitative and qualitative techniques, drawing on institutional databases, in-depth customer and staff interviews, focus groups, 6 See the Mastercard Foundation & IFC “Field Note #7 Turning MFI Digital Strategies into Reality” 2017: https://www.ifc.org/wps/wcm/connect/67a1ee9e-9f95-4baa-8430-2a101ca77a9e/MFI+Digital+Strategy+Field+Note_8.pdf?MOD=AJPERES 7 See the Mastercard Foundation & IFC “Aligning Expectations - The Business Case for Digital Financial Services” 2017: https://www.ifc.org/wps/wcm/connect/5a322011-52b6-4b52-b4d5-6175039e551d/MFI+Longitudinal+Study_Digital_FA.pdf?MOD=AJPERES 11 A careful assessment of the regulatory environment is an important aspect of the DFS planning stage An important aspect of the environment assessment for the Until receiving this license, the institution could not continue DFS planning stage is a careful review of the relevant banking recruiting new agents or launching new products on this regulation commonalities in the country or region where the channel. Many of the operating agents were also not bank operates. As part of this exercise, it may be worthwhile to complying with the new strict requirements that were put in contrast the country’s relevant banking and DFS regulations – place as part of the background check and they were forced if they exist – with regulations from countries that are known to close. In this period, agent network expansion was hence to have enabling DFS environments. This can help in identifying halted and the institution even reduced network size. challenges and gaps (an overview on the regulatory contexts of the study countries and their challenges is added in the The study found that digital financial inclusion thrives within annex). It can also help when seeking regulatory permission for enabling regulatory frameworks and when stakeholders a digital service delivery model, especially in underdeveloped collaborate closely with policy makers. Only a few of the markets where the institution may be breaking new ground. study countries had existing regulations for bank-led DFS It is important to be well prepared when doing so, as in many deployments when the institutions started their DFS journeys: instances the regulator may not have sufficient knowledge Rwanda, Senegal, and Tanzania. In other markets, the FIs had itself to be able to adequately assess a proposed initiative. to deal with legal voids and lack of clarity. In those kinds of situations, taking a pro-active approach and directly engaging For example, one of the institutions in West Africa that was with regulators can prevent delays in implementation. studied sought the regulator’s approval to integrate its mobile banking service with a leading mobile money platform to One FI proactively addressed all concerns regarding the allow customers to move money between their bank accounts implementation of digital channels before the regulator raised and mobile wallets. The regulator rejected this request based them. By doing so, it gained the Central Bank’s trust to be the on existing regulation that does not allow banks to “issue first institution allowed to develop and pilot a bank-led agent e-money.” However, the law was not clear about the scope network in the country. In other cases, such as Rwanda and of the term e-money and a case where e-value is only moved Senegal, regulators become more progressive about Digital between a bank account and a mobile wallet. The lack of Financial Services after better understanding how digital shared understanding between bank and regulator delayed channels work. Similarly, in some countries like Nigeria, the the account-wallet integration for more than two years. issue of agent exclusivity became a factor to consider when expanding an agent network. Different agent paradigms are Three of the institutions in the study encountered delays depicted in Table 1. in obtaining approvals and licenses from the regulatory authorities for their agent banking services. One regulator Defining a realistic DFS business case and requested an individual background check of each agent goals that fit the operating environment candidate, which substantially slowed down agent recruitment processes and agent network growth at the Most of the institutions participating in the study were first beginning of the rollout phase. This caused considerable movers. They were the first bank – or even banking institution – damage to the institution’s implementation plan as agents developing a DFS offering in their respective markets. This was that had already been recruited and trained were unable to true for the institutions in Cameroon, Democratic Republic start their business and lost motivation or even eligibility to of Congo (DRC), Madagascar, Nigeria, Rwanda and Senegal. become banking agents. During the waiting period, some As such, they were forced to rely on their own assumptions or agents were contracted by competing mobile network the experience of other ecosystem players, primarily mobile operators to join their networks and lost interest to become network operators, to guide the development of strategies bank agents. and business models. This would include the number of agents needed, how many customers an agent would serve, and One institution only received regulatory consent for a pilot what average transaction sizes would be, for example. They with four agents. Despite this limitation, the FI proceeded and did not have benchmarks available adapted to the realities built a network with more than 400 agents within 14 months. of financial institutions. While providing some rudimentary The bankruptcy of a leading bank in the same country then guidance, such an approach does not consider differences increased the Central Bank’s concerns about the sector as a in the business nature and culture of different ecosystem whole and the regulator became increasingly cautious and players. To help fine-tune assumptions that would be more risk-averse regarding new innovations. Control measures valid for financial institutions and to better understand the and bureaucratic barriers were raised, blockading further discrepancies between various types of market actors, the expansion of the bank’s agent network. The regulator required research team compared the initial financial projections for an individual background check of each new agent, conducted agent banking deployments of the study institutions with the by the regulator itself. Additionally, the Central Bank required actual realities to date. The team found that: that financial institutions intending to operate agents set up a Banking Operations Intermediary and apply for a license first. 12 a) Customer and agent registrations do not translate into activity.8 While customer and agent acquisition targets set in the financial models were often largely met or even exceeded, activity rates have remained far below assumptions. Figure 6: Median customer and agent activity rates over the course of the study 89% 81% 76% 73% 67% 63% 63% 62% 62% 53% 47% 41% 30% 31% 32% 26% 23% 21% 20% 20% 18% 18% 19% 10% Q1 2015 Q2 2015 Q3 2015 Q4 2015 Q1 2016 Q2 2016 Q3 2016 Q4 2016 Q1 2017 Q2 2017 Q3 2017 Q4 2017 Customer activity Agent activity Linear (Customer activity) Linear (Agent activity) b) Customer transaction targets were out of reach. The c) Customer transaction patterns differ from the MNO research team found that institutions reached only around 16 model. The team noted two main differences between percent of the projected agent cash transaction volumes and customer transaction patterns at MNOs and banks. 17 percent of the value. This finding goes hand-in-hand with I. Since bank-led agent networks are primarily used as unmet expectations on customer and agent activity rates. outlets for the repayment of loans, transaction amounts tend to be higher than those at MNOs. In Q4 2017, the median cash-in amount of the study institutions was $146 and cash-out was $117. For MNOs in Sub-Saharan Africa, GSMA reported an average transaction amount of approximately $16 for December 2017.9 Figure 7: Average cash-in and cash-out amounts in US dollars over course of the study $166 $163 $145 $146 $131 $125 $118 $115 $117 $116 $120 $117 $109 $111 $111 $106 $111 $110 $107 $100 $97 $97 $101 $85 Q1 2015 Q2 2015 Q3 2015 Q4 2015 Q1 2016 Q2 2016 Q3 2016 Q4 2016 Q1 2017 Q2 2017 Q3 2017 Q4 2017 Cash-in amount Cash-out amount 8 An active customer has completed one DFS value transaction within 30 days. As transaction counts cash-in or cash-out from account, P2P payment, bill payment, etc. Balance inquiries, PIN resets and other transactions that do not involve the movement of value do not qualify as value transaction. 9 See GSMA “2017 State of the Industry Report on Mobile Money” 2018: https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2018/05/GSMA_2017_State_of_the_Industry_Report_on_Mobile_Money_Full_Report. pdf; in 2017 the average mobile money transaction amount in SSA was at $16, the global average was at $17.5 13 II. As of December 2017, the median cash-in share in to (a) an overestimation of fee income from cash- the transactions performed at agents of the study out transactions generated on the channel, (b) an institutions was 76 percent (i.e. only 24 percent of underestimation of efforts and costs related to cash transactions were cash-outs), which differs liquidity management support for agents. With more from the more balanced transaction mix the unequal float and cash flows, sustaining a sufficient industry observes for mobile money agents.10 When float and cash balance is more challenging for banking defining assumptions for financial modeling using agents. MNO experiences, this misconception may lead Figure 8: Median cash-in versus cash-out shares over the course of the study 11% 14% 20% 21% 20% 22% 28% 22% 22% 25% 24% 24% 89% 86% 80% 78% 75% 76% 76% 72% 78% 80% 79% 78% Q1 2015 Q2 2015 Q3 2015 Q4 2015 Q1 2016 Q2 2016 Q3 2016 Q4 2016 Q1 2017 Q2 2017 Q3 2017 Q4 2017 Cash-in share Cash-out share d) Agent networks are smaller, but agent activity than quantity. These factors contributed to higher agent rates are higher. Expectations on required agent activity rates for the participating FIs vis-a-vis the larger management efforts were hence met. The agent MNO networks. Expectations on the level of agent networks of participating institutions are small – between commission and management costs were largely met or 200 and 1,500 agents – and concentrated around banking even exceeded, despite projections on network activity branches. The number of active agents varied from 100 to and transactions volume being out-of-reach.11 900. Even the largest network of the study institutions, which has 1,471 registered agents of which 887 are active, Another important lesson is that financial modeling should is small compared to standard MNO agent network not be seen as a one-time exercise. New benchmarks that sizes that are on average ten times larger. The rate of were developed as part of this study can be used to develop active agents for the FIs was around 60 percent however, further iterations of business models as needed. Institutions compared to less than 40 percent for MNO networks. This should reassess original plans and make regular revisions indicates that the financial institutions outperformed based on adoption, usage, and actual growth of the digital the MNOs in terms of management and stimulation of service throughout its lifespan. their networks, with a model building on quality rather 10 See GSMA “2017 State of the Industry Report on Mobile Money” 2018: https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2018/05/ GSMA_2017_State_of_the_Industry_Report_on_Mobile_Money_Full_Report.pdf 11 For more information on the findings of the analysis on initial agent banking projections and FI benchmarks, please see the Mastercard Foundation and IFC “Aligning Expectations: The Business Case for Digital Financial Services” 2017: https://www.ifc.org/wps/wcm/connect/5a322011-52b6-4b52-b4d5- 6175039e551d/MFI+Longitudinal+Study_Digital_FA.pdf?MOD=AJPERES 14 Exploring hybrid approaches to adapt DFS act as risk mitigation under this lending methodology, as to existing customer habits and cultural group members guarantee each other’s loan repayments. context Traditionally, all payments are done in cash during in-person group meetings, where deposits are physically handed Digital channels are said to facilitate access to financial to the loan officer who repays in bulk at a bank branch. services and enhance customer experience at low cost. This model requires the physical presence at the meeting There are, however, a number of factors that pose challenges of both the borrowers and the loan officer. At one of the to market uptake, including a general lack of trust in the institutions being studied, group borrowers were offered financial sector, self-exclusion, customer stickiness to the option to either repay at an agent outlet or at the roving informal financial services, and fears about adopting new agent who was participating in the weekly group meetings. technologies. When interviewing bank customers in the To preserve key elements of the traditional methodology, field, the researchers found that many people distrust new the institution mandated all group borrowers to confirm digital services because they are afraid their money may ‘get their on-time repayment during the meetings by handing lost’ in the digital system due to issues of connectivity or in the transaction receipts from the agents or by using the human error. Additional reasons for not choosing to adopt roving agent at the meeting. Another institution, which digital services were a lack of awareness of the DFS value also had a history of relying on group lending, decided to proposition and self-exclusion, i.e. some of the interviewees move cautiously with DFS so as not to disrupt the existing believed that they were not sufficiently technology savvy, dynamics of its group clients, which account for 80 percent financially educated or affluent to use such services.12 of the overall customer base. Consequently, it initially limited the services it offered at pilot agents to account A key factor to customer adoption of DFS is trust, but it opening, savings collection, cash-out, and transfers for takes time to build trusting relationships with customers. individual customers. Loan repayments via agent outlets FIs operating in generally difficult market environments were not allowed in this first phase, and still require group or where digital technologies and digitization are still borrowers to participate and repay during regular group new phenomena may find it particularly challenging to meetings. Individual group members can still open separate build clients’ trust in digital services. Institutions that accounts and use agents for the purpose of individual are traditionally targeting customers at the base of the savings. At the end of the study period, the institution was income pyramid and that are heavily reliant on traditional planning to add a loan repayment option for individual microfinance strategies (group lending methodologies, loans at agents. The repayment of the group loan, however, frequent face-to-face interactions with customers, etc.) will remain exclusively within the group meetings. are advised to move cautiously and to carefully assess the readiness of the existing customer base or target segments As a test, some of the study institutions tried to incorporate to move into the digital era. The experience of the study common financial practices and market peculiarities into institutions shows that FIs will need to invest time and the DFS strategy. For example, the use of so-called susu capacities in training, marketing, and handholding of collectors for daily savings is common in many West African clients to support their learning processes and to build trust countries. This informal service relies on mobile collectors in using the new services. circulating in markets and communities to collect regular contributions from clients, usually for a monthly or weekly It may also be wise for institutions to consider a gradual flat fee. One of the study institutions had incorporated the approach for the implementation of DFS. Adopting a step- susu mechanism into its business operations even before by-step strategy to sensitize staff and customers alike could going digital, using roving staff to collect daily savings entail hybrid models where face-to-face interaction with from business customers in market areas. The institution customers is at least partly preserved at the early stages then digitized the process of this mobile collection service, of the digitization effort. In this sense, agent banking supported by a proprietary agent network that is centered models can serve as hybrid model and are particularly around cash-light branches. Its agents focus primarily on interesting for microfinance institutions that heavily rely collection of loan repayments and cash-out. The digitized on face-to-face interaction between staff and customers. collection service, which featured instant confirmation of Instead of branch staff, agents conduct digital transactions deposits via short message, has proved popular and has led on behalf of customers or are able to offer face-to-face to a significant rise in deposit mobilization and consequently support to customers when conducting DFS transactions improved the capital structure of the bank. After 18 months, on their own. Similarly, it may be fruitful to experiment the mobile collection accounted for around $15 million in with models that tailor DFS offerings to customers’ existing cash, equivalent to about 40 percent of the overall deposit financial habits and cultural peculiarities. For example, portfolio. some of the institutions that participated in the study have tried hybrid models that marry DFS with group lending methodologies. Groups’ social cohesion and peer pressure 12 Similar results were also observed in research performed by the Mastercard Foundation and IFC regarding perceptions and attitudes towards mobile money in Cameroon, DRC, Zambia and Senegal. See “A Sense of Inclusion - An Ethnographic Study of the Perceptions and Attitudes to Digital Financial Services in Sub-Saharan Africa” 2017: https://www.ifc.org/wps/wcm/connect/15e6158a-8e52-444b-9103-391547cb1730/ IFC+A+sense+of+Inclusion+DFS+Ethnographic+Study+2017.pdf?MOD=AJPERES 15 In markets where the use of innovative technologies was – in particular tellers and loan officers – felt threatened by more common and where a DFS ecosystem had already been agents and regarded them as competitors who could make established, the research showed that institutions found it their roles in the institution redundant. Branch staff, who easier to move their customers onto digital channels. Even used to be in direct ownership of the customer relationship, so, client education, effective marketing and well-targeted would therefore refuse to promote the adoption of the new products and services were critical in ensuring that digital service, leading to limited DFS uptake. Additionally, a digital channels are differentiated from existing offerings in the project requires putting clients in the center, a difficult market. task for FIs that normally are product-centric. Therefore, having a customer-centric approach requires a mindset Full engagement is needed to successfully change for most staff members of the organization, and it roll out a DFS requires mechanisms for properly gathering and addressing customers’ feedback. To be successful, an institution’s journey into digital In fact, the research shows that some of these concerns are transformation requires a shift in mindset of staff and legitimate. Figure 9 shows that out of the eight reporting clients. Most of the financial institutions in the study faced institutions deploying DFS solutions, five either reduced internal and external resistance to digitization. Internally, the number of their staff outside of the headquarters over staff felt threatened by the new digitization push as it the course of the study period or they kept staff size more promised to bring better efficiencies and cost reduction. or less stagnant. At two institutions, the number of non- It created fear among staff as positions and salaries were headquarter staff rose. perceived at risk. For example, the FIs noted that branch staff Figure 9: Branches (right y-axis), non-HQ staff and agent network size (both left y-axis) over data reporting period BANK 1 BANK 2 BANK 3 1000 25 1000 25 500 18 800 20 800 20 400 17 600 15 600 15 300 16 400 10 400 10 200 200 5 200 5 100 15 0 0 0 0 0 14 BANK 4 BANK 5 BANK 6 2000 25 250 40 800 50 20 200 40 1500 30 600 15 150 30 1000 20 400 10 100 20 500 5 10 200 50 10 0 0 0 0 0 0 BANK 7 BANK 8 600 50 300 20 500 40 15 400 200 30 300 10 20 200 100 5 100 10 0 0 0 0 Staff non-HQ Branches Agents Notes: Data reporting start dates vary; agent networks in pilot stage are included; for two institutions reporting is incomplete 16 It is still important to foster an understanding among staff adopted key performance indicators (KPIs) based on DFS – particularly branch staff – that ‘going digital’ is necessary, performance, and they have assumed responsibility for and it is important to accurately communicate that message the management of local agents. Some staff have been to all employees. Not doing so can create challenges for moved from headquarters to the branches to better the digital project and even put its success at risk. One guarantee branch oversight and support in areas outside study institution that did not properly communicate the the capital. Most branch staff were also moved to more advantages and requirements of the implementation of generic roles as Customer Service Agents, giving the bank an agent network to its branch staff struggled to get the more flexibility to quickly allocate resources where business digital channel off the ground. Loan officers did not have needs arise. The financial institution has also introduced any incentives to encourage customers to use the digital new training opportunities for staff, including, for example, channel but remained focused on loan recovery. Similarly, coaching staff on how to enroll clients at agents. Younger branch managers considered the integration of agents to staff have been offered opportunities in leadership roles, branch operations a low priority. The value proposition of and there has been a positive impact on staff morale and cashing out at agents (time savings, convenience) and a engagement. Since the implementation of these changes, detailed instruction on how to use the digital channel were the institution has seen an increase in the number of cash also not integrated into the financial literacy content of the transactions handled via agents, from 25 percent in May bank. 2017 to 31 percent in December 2017. The research further shows that it is important to leverage Friendly technology and a data-driven senior management when sensitizing staff to the value of approach increase the chances of success the digital venture, particularly when addressing branch staff that has the most direct interaction with customers. Technological readiness can make or break a digital Management should clearly communicate that ownership channel. All of the institutions in the study experienced of the new digital channels lies with the entire organization severe technological challenges during DFS planning and not headquarters alone. While it is the role of (and later rollout), including, among others, difficulties headquarters to manage the harmonization of digital and in selecting appropriate technology providers, delays in traditional channels, staff located at branches or in the field technology delivery, difficulties integrating DFS systems should be responsible for promoting digital services in their to the core banking systems, and insufficient network daily interactions with customers, e.g. by proposing agents strength and coverage. Reliance on Global System for as faster and easier option for cash services, or explaining Mobile communications (GSM) and internet connectivity is how agents can facilitate the account opening process. a major risk to the successful delivery of DFS in developing Internal communication should also center around how the and emerging markets. Consequently, it is advisable that digital channels benefit staff in their daily functions, e.g. FIs interested in launching DFS take note of the following how DFS enables loan officers to devote more time to loan during the planning stage: promotion, origination and recovery instead of collecting loan repayments.13 • Perform a proper due diligence or a selection process of the technology providers to use. This Even if an institution does not get communications or includes asking for references from previous clients. incentives structures right from the start, it may be possible to adjust them at a later stage, albeit at a cost. One • Assign enough time for system integration institution in the study that went through the digitization testing. This type of testing should be both technical process and concurrently made fundamental structural and functional. On the technical side, one should ensure changes, failed to define a unified communication plan that that all systems are using the right data exchange and could address the emotional aspects of change and help communication protocols. On the functional side, one to integrate the digital channel into the daily operations. should ensure that users will have the right experience and For many of its employees, the initiative to ‘go digital’ was receive the right responses. perceived as a threat. They were afraid they would lose their jobs to agents and the new mobile banking application. • System testing should also look at transaction reconciliation. Reconciliation testing should include In 2017, the financial institution mentioned above, began to developing different scenarios on the type of transactions implement measures designed to increase staff acceptance going through the system with the respective exceptions. of the DFS project. One immediate action taken by the FIs should aim for transaction reconciliation processes FI was a revision of incentive structures at its branches. that are fully automated and take less than 24 hours. The FI transferred ownership of the agent network from headquarters to the branches. The branches have since 13 See also the Mastercard Foundation & IFC “Field Note 8: Changing change management: adapting internal and external culture in times of digital transformation” 2018: https://www.ifc.org/wps/wcm/connect/93567f5c-6eb5-4e23-bc13-ee0c55f3eabc/IFC+MCF+Field+Note+8_ DFS+Change+Management+MCF.pdf?MOD=AJPERES 17 • The planned DFS technical infrastructure should Another important technological factor to keep in mind have communication redundancy in place. This when planning a DFS is the selection of user-friendly addresses risks regarding network connectivity and applications and devices. It is important to include an strength. For example, one of the FIs in the study uses assessment of the technical sophistication of your existing agent Point-of-Sale devices that support two or three or targeted customers in the upfront market research. SIM cards to mitigate risks related to connectivity. Upfront market research efforts to investigate this aspect were uncommon among the institutions of this study, The use of new technologies brings new risks to the which has possibly led to the adoption of technologies business, and it is advisable to put in place a risk that mismatched customer readiness and their capacities management framework that identifies such risks upfront in adopting the services. For example, a mobile banking as well as puts in place processes and measures to monitor application that is only compatible with smart phones will and mitigate risks. The study institutions have experienced prevent the large share of existing USSD users in a customer people and system risks that relate to the adoption of DFS: base from adopting the service. a. People risks related to agents: All institutions reported having experienced various kinds of agent The digitization of banking operations should also entail fraud. Agents or agent assistants may split transactions investments into building internal capacities for Business (the most common type of agent fraud reported by Intelligence (BI), including the recruitment or training the FIs), or abuse system downtimes and the lack of of staff to conduct collection, integration, analysis, and customers’ understanding of the system. Agents may presentation of business information such as customer also accept cash for deposits from customers but delay transactional data. This is worthwhile for the later design the actual completion process of the transaction in the of new products for the digital channel. For instance, two system to a later time. As a result, customers do not institutions have developed and successfully introduced receive an immediate confirmation of the transaction or nano loan products based on scoring systems built on receive the confirmation with a delay, spurring doubts client transaction data and credit histories that are solely about the reliability of the service. distributed via agents. b. System risk related to manual control systems: Most of the banks in the study still use manual control mechanisms for their DFS. This limits the capacity II. LESSONS LEARNED: to effectively monitor the service and raise flags e.g. MANAGING A DIGITAL FINANCIAL when suspicious transactions or irregularities arise – SERVICE especially when the service is scaling up. Effective management and close supervision of DFS One FI decided not to invest in an automated risk control operations are crucial for the success of the digital service. system for agent fraud and transaction irregularities when This is especially true when implementing a digital channel launching their DFS and has been using a manual, excel- for the first time. When the institutions in the study based tool for fraud control. When the network of agents first decided to go digital they all generally started by expanded, the DFS team found that the whole team had implementing agent banking. Hence, most of the findings to spend three to four days a month manually checking all in this section relate to the management of an agent agent transaction files. Still, the manager estimated they network. would only detect around 10 percent of all agent fraud actions causing the bank considerable overspending on Managing an agent network starts with a agent commissions. The bank is currently reviewing this strategy to identify and recruit the right process and plans to invest in an automated control system agents in the near future. An early investment into data-driven solutions for channel monitoring and management offers An agent network does not rely on a digital interface alone, better security and transparency on channel activities but also on human capacity. By employing agents to interact and should pay off once the DFS is scaling up. The team directly with customers, often as the first contact point for found that other institutions have benefited from early customers, an institution outsources a considerable part investments into centralized, automated structures for of the customer relationship to a third party and this will DFS management and monitoring. Two FIs worked with have an impact on the bank’s image. The service quality at a technology provider to design a dashboard for instant agents is therefore of utmost importance. Research shows monitoring of DFS operations and KPIs. that the quality of a customer’s first-time experience with a new channel has great impact on his usage and activity going forward. In an interview, one agent officer said: “If a customer tries the service a first time and it is not working, he will never try it a second time.” 18 Recruitment is critical to agent quality. All the eight In some markets, regulators intervened in the agent institutions in this study that rolled out agent networks recruitment process. In those cases, the regulators put in place sets of agent selection and/or agent profile specified, for example, that agent candidates must meet criteria to guide the recruitment task. These policies and a set of criteria before being approved for the role, or criteria evolved over time, based on requirements defined that candidates undergo mandatory background checks by the regulator as well as the experiences and assessments that were to be performed by regulators prior to starting made by the banks. Poor agent selection can lead to poor operations. This slowed down agent recruitment and agent performance, reputational risk, regulatory risks and impacted institutions’ growth plans for their agent eventually financial losses. networks. From the interviews the research team held with agent The study also found that the recruitment of qualified managers at the participating FIs, the following attributes agents was particularly challenging in two situations: were identified as success factors for agent profiling and recruitment:14 • In rural areas, limited knowledge of technology and limited education presented challenges. This required higher efforts and investments in financial education, digital literacy and training. What characterizes top performing banking agents? • In relatively mature DFS markets, where multiple service providers attempt to attract agents to their 1. They are previous banking clients with a proven track services, FIs are under pressure to build an attractive record of more than two years of loyalty to a bank. value proposition and incentive scheme to differentiate themselves from other market players. 2. They may be exclusive or non-exclusive*, i.e. agents may only serve one DFS provider or multiple Manage agent network growth to balance providers. Non-exclusive agents tend to adapt to internal capacities and customer demand their roles easier and faster because they are already familiar with similar businesses. For the institutions in the study, agent network growth 3. Retailers tend to make good agents. decelerated over time. The collected quantitative data shows that agent network growth, in terms of the number 4. Evidence also shows that women tend to make good of recruited agents, is, on average, four times faster in agents. the first operating year of an implementation than in the 5. They should have a good reputation in the local second year (see Figure 10). In the third year, monthly community; they are trusted and known to be network growth usually falls below 5 percent, arriving at reliable. a growth average rate of just 1.4 percent in the fourth year of operations. This shows that bank-led agent networks do 6. At their outlets, agents should clearly display their not grow indefinitely and are much smaller than the type marketing and pricing materials. run by MNOs. The data also suggests that bank-led agent networks reach a saturation point after some years of 7. They should have flexible opening hours, preferably operations (year four in the case of this study), most likely outside the opening hours of bank branches. due to two factors: internal capacity to manage and grow the network and/or market demand. In addition, over time 8. They should be located in areas where a lot of cash it became apparent during the study that FIs with large transactions are conducted, such as near markets agent networks but low agent activity rates are not cost- and mechanics’ streets. effective. Agent recruitment and setup is expensive, and a financial institution might not be able to recover these costs 9. The agents and their assistants should have a level of if the agents do not have sufficient business thereafter. The digital literacy and knowledge to be able to correctly institutions realized the importance of having an agent offer digital services on behalf of a bank. network that responds to evolving market demand, with 10. They are well supervised and monitored to sufficient business for each of the network agents. ensure they are delivering quality service. Bank representatives should visit each agent at least once every two weeks. * If the regulator does not forbid agent exclusivity, as it is the case in Senegal and Nigeria. 14 Top performing agents are characterized by high transaction volumes and the ability to reliably manage their float and liquidity requirements. 19 Figure 10: Monthly agent network growth rates over the operational lifetime 20,9% 5,1% 3,2% 1,4% 1. Year 2. Year 3. Year 4. Year Notes: First year growth measured for all networks > 25 agents The study also collected information on the required human Figure 6 shows that agent activity levels at the institutions resource capacities to internally manage the network. in the study have been rising over time, while the share of The team found that in Q4 2017 the number of agents active customers versus registered customers has been managed by a single agent manager varied significantly declining. Figure 11 puts these numbers in relation, showing among the institutions (between 17 and 55). On average, the average number of active customers per active agent at a single agent manager oversaw 26 registered agents. This the end of each month for six of the agent networks in the number decreased to 14 when only counting active agents, study (the blue lines). The agent network utilization rate varying between 7 and 21 among the institutions. For all has ranged from 40 active customers per active agent up to institutions, the number of agents being managed by a over 100 customers per active agent. As of December 2017, single staff manager was rising, hinting at a development both the average and median calculated for this indicator of efficiencies in agent management over time. stood at around 80 and has, more or less, remained at this level over the course of the last year, revealing a more or less Finding the right pace for business expansion that stable customer density per agent. corresponds to market demand can be a challenge. This is also true for agent networks. It is important that agents have sufficient business and can deliver high quality services when enlarging the network. Monitoring customer and agent activity as well as transaction flows can help predict situations when there are either too few agents for the market, or too many. Figure 11: End-month values of active customers per active agents in 2017 140 120 100 80 60 40 20 0 Jan 17 Feb 17 Mar 17 Apr 17 May 17 Jun 17 Jul 17 Aug 17 Sep 17 Oct 17 Nov 17 Dec 17 Average of study MFIs Median of study MFIs 20 Figure 12: End-month values of average transaction volume per active agent in 2017 in US dollars $600 $500 $400 $300 $200 $100 $0 Jan 17 Feb 17 Mar 17 Apr 17 May 17 Jun 17 Jul 17 Aug 17 Sep 17 Oct 17 Nov 17 Dec 17 Participating MFIs Average of study MFIs Median of study FIs Finding the right pace for agent network growth is more other DFS providers. Only an attractive business case challenging and is influenced by developments in the and incentive structure will ensure that the right agents market and in regulation. The study pulls the following are recruited; and that recruited agents show sufficient recommendations for agent network expansion from the devotion and commitment to your business in order to experience of the participating institutions: foster a loyal customer base. • When FIs build an agent network they should plan When interviewing agents, the team found agents are to have strong growth within the first two years. This especially sensitive to the business case and commission growth should then slowdown in subsequent years. compensation that an institution is offering. One agent shared a sophisticated table he had developed that showed • While human resources needs are high when starting the commission he would make for each transaction type the agent network (most likely due to high staff with each specific service provider he has been working involvement in agent recruitment), study results for. When, for example, a customer approached him for indicate that capacity needs for network management a $20 cash-out, he would refer to the table and if the decrease with time. commission for the requested provider was lower than for the competitors, he would attempt to convince the • FIs should not rely on MNO experiences when defining customer to use the service provider that paid him the growth and size assumptions for their agent networks. highest commission. If he only had limited cash available, FIs, and banks in general, tend to serve more specific he may even decline to serve the customer and wait for a market segments than MNOs. more lucrative service request. The example illustrates how savvy agents can allocate their resources to the service • Close monitoring of agent density and customer provider that is personally most profitable for them. This is utilization can help to identify recruitment needs or especially relevant for non-exclusive agents in competitive market saturation points. It is recommended to use on- environments. In less established markets, agents also time data visualization tools to monitor agent activity, appreciate training opportunities, the increased customer the ratio of active customers per active agent, average traffic the DFS service brings them, and the reputational transaction volumes per agent, etc. In addition, it is also gain and networking opportunities in their communities. worthwhile to monitor density of agents per population or area, separated by regions or districts, i.e. number The data suggests that the business case of agents has of agents per 1,000 square kilometers or per 10,000 become increasingly attractive over the course of the study population. period. The median number of monthly cash transactions per active agent has been rising, as also indicated in the The most attractive agent incentive previous section. It stood at 126 in Q1 2015 and had doubled structure makes the best agent network to 244 in Q4 2017. The median value of monthly cash transactions rose 2.8 times in the same time. The value of cash transactions at agents has hence increased faster To employ and keep high quality agents it is important to than the volume, which could indicate that customers provide them with the right incentives. This is especially are becoming increasingly trusting of the service and are true in more mature markets, where the majority of agents transacting higher amounts at agents. In some markets, recruited by FIs are non-exclusive and work with one or agents reportedly earned as much as $200 in commission 21 fees per month. Overall, agent income has been rising, This shows there is a convincing value proposition to across most of the institutions. In 2017 alone, the median prospective agents, which should be clearly communicated monthly income doubled from $44 in January to $90 in by financial services providers building or expanding agent December. It is reasonable to expect that transaction networks. One FI guarantees new agents a minimum volumes and commission income will rise further in the income for the first three months in operation, something future, since all institutions in the study have expressed institutions may consider in order to make the agent case plans to expand product and service offerings at agents. even more compelling in the recruitment process. Figure 13: End-month values of average commission income volume per active agent in 2017 in US dollars $200 $180 $160 $140 $120 $100 $80 $60 $40 $20 $0 0 Jan 17 Feb 17 Mar 17 Apr 17 May 17 Jun 17 Jul 17 Aug 17 Sep 17 Oct 17 Nov 17 Dec 17 Participating MFIs Average of study MFIs Median of study MFIs Proactively support your agents and agents and the performance of their agent networks (see leverage smart tools for DFS management Figure 14). The management system most often uses two dashboards: Especially when agents are freshly recruited, agent support 1. Daily Operations Dashboard provides a daily update of and management staff should take a proactive approach a bank’s savings and loan portfolios with automated and regularly inquire about their agents’ well-being and alerts when operational risks arise. Metrics can be feedback, if feasible in person. When launching the agent customized to operational needs, but also include KPIs banking service, the FIs participating in the study tended on transaction volumes, commissions and fees, agent to adopt the same feedback cycle for all their agents. Later, activity, suspicious and potential fraud activities, DFS Average of study MFIs Median of study MFIs it proved reasonable to reduce the frequency of feedback enrollment, failed transactions, and the geographic from agents as they became accustomed to the service. spread of operations. Some of the institutions have developed standardized questionnaires that their agent supervision staff use. If 2. Monthly Strategic Dashboard offers a longer-term, systematically consolidated, the information collected strategic view. It was developed to provide an overview during these visits can help the institution monitor agent of the customer lifecycle, including how the use of satisfaction levels, quality of equipment and material at services and products evolve (e.g. branch transaction agents, and other issues or concerns that arise in their daily volume versus agent transaction volume), and customer business. This will ultimately help to identify support needs adoption and usage of DFS. and effectively allocate resources. Agent management should also include monitoring Also, a step-by-step approach is recommended. The two customer traffic at agents, if financially feasible, through an banks started with some basic visualizations and then built early investment into an automatized dashboard solution. the dashboards with increasing sophistication over time. Two FIs in the study, for example, were early adopters of “next generation” DFS management systems, acquiring and implementing BIME, a visualization tool to help optimize operations. It enabled the institutions to develop interactive dashboards tailored to survey the operational needs of the 22 Figure 14: Example of a daily operations dashboard A B C A Information on customer and agent counts, transaction volume B Transactions by day (left), transactions by week (right). These different views highlight peak weekday demand mid-week and low weekend demand when viewed by day (left); with similar cycles over the year with weekly averages (right). C Top performing agents by transaction volume and by transaction number, top performing branches by volume 23 FIs need to understand the different • The belief that agents can handle this challenge on facets of agent liquidity management, their own, but agents don’t always have resources and prioritize agent liquidity monitoring, and capacities to do so. provide adequate daily support tools • Lack of reporting and monitoring tools that have ability When the research team interviewed agents and customers to assess quality of agents in the network, such as in the field, one of the main complaints concerned liquidity visualizing transaction peaks and valleys. management. Broadly, liquidity management can be defined as the sum of activities that ensure an agent has • The lack of reliable means to monitor the negative enough cash and float to service customers. impact of liquidity constraints; i.e. how many customers are sent away by agents due to liquidity constraints. To exemplify: a customer enters an agent outlet with the FIs report that agents use “insufficient liquidity” as an request to deposit $50 into his bank account. He hands excuse not to serve the customer when, for example, over $50 in cash to the agent, who transfers the equivalent they are actually hoping to use existing liquidity for a amount from his agent float account to the customer’s more lucrative transaction type or service provider (in bank account. The customer receives an electronic SMS or case the agent is non-exclusive). paper receipt as confirmation and the agent has earned a commission revenue. Similarly, a customer can request a • The costs and capacities required to invest in liquidity withdrawal at the agent outlet. In this case, an equivalent management tools. of the requested amount is transferred from the customer’s account to the agent float account from which the agent The study revealed two main reasons why agent liquidity hands it out utilizing his cash on hand. In both cases, the management has turned out to be a major issue. First, the sum of money at the agent remains the same, it is only the nature of bank-led DFS is different from the MNO model. The cash/e-float composition that changes. A liquidity shortage majority of cash transactions at bank agents are deposits, arises when a series of customer transactions deplete the with a net of cash coming in to the agent account. In agent’s cash or e-float. In this kind of situation, agents may December 2017 the share of cash-ins in the total transaction have to turn customers down wishing to perform a cash- volume of the study institutions varied between 48 percent out (turned down if agent has insufficient cash on hand), a and 97 percent. The average stood at 76 percent, i.e. 76 cash-in or a transfer (turned down if agent has insufficient percent of all cash transactions at agents were deposits. In e-float). Agents and financial institutions thus deal with value, the share varied between 55 percent and 96 percent, two types of liquidity management; cash and e-float. with an average of 79 percent for deposits. The cash streams of bank agents are thus not balanced, as is reported to be In a perfect situation, all agents of a network should have the case for MNO agents. For the FIs, this meant that it sufficient liquidity (cash and e-float) to serve every customer has been more difficult than expected to sustain sufficient transaction request at all times. If customers are repeatedly e-float. Second, the institutions in the study mostly target turned down due to lack of liquidity, they may conclude that SMEs and individuals in the lower income range, which the DFS is not reliable and refrain from using the service in usually reside in rural and remote areas, where the majority the future. This is especially the risk with new customers. of agents are also located. Supporting the physical liquidity Poor liquidity management practices therefore have direct management of such agents is time consuming and costly impact on business results, DFS uptake and usage, agent (e.g. delivery of cash against e-float) and requires putting in income and motivation, and consequently the success of place appropriate liquidity support tools. the digital channel as a whole. Despite these considerations, the team found that the FIs Over the course of the study, the research team observed did not prioritize agent liquidity management in a timely the FIs and agents adopting and refining various kinds and sufficient manner. Three factors contributed to this: of liquidity management tools. Table 2 summarizes these according to type, addressed needs, partnership requirements, process length, and costs. 24 Table 2: Agent liquidity management solutions observed as part of study POPULARITY LIQUIDITY PARTNERSHIP PROCESS ASSOCIATED COSTS FOR COSTS FOR PROCESS AMONG ADDRESSED NEEDED LENGTH RISKS* AGENTS FI STUDY FIs Cash & e-float None Agent visits nearest Multiple Low for FI; high High (time and Low High BRANCH bank branch to hours for agent travel costs) exchange float – cash Cash & e-float Yes, if Agent calls superagent, Multiple Low for FI; low Low to High Medium SUPERAGENT externalized superagent visits agent hours for agent; high medium for superagent (if fee per transaction) Cash & e-float None Agent contacts Multiple High for FI; low Low (given None High AGENT EXCHANGE other agents, agents hours to medium for agent is close exchange float – cash agent by) Cash & e-float MNO Cash: Agent conducts Multiple None for FI; none Low to Initial High, but can B2W / W2B15 partnership(s) B2W transfer; agent hours if cash for agent medium investment be difficult to calls MNO superagent delivery, (fee per for negotiate with to receive cash delivery automated transaction) integration MNO e-float: agent conducts if W2B W2B transfer e-float None Agent applies for Automated Low for FI (credit Low to None to high High OVERDRAFT FACILITY overdraft risk); medium Low none for agent (interest expenses) e-float None Situations of liquidity Automated None for FI; Low Initial Low BIG DATA** constraints are none for agent investment predicted by algorithm; for agent receives alert development * For example, risk of losing money, robbery. **Still in testing phase. Many of the institutions experienced difficulties deciding on Several institutions reported the emergence of informal the right liquidity management solutions. Often, solutions solutions among agents. At one institution, it was a known were tested, implemented, abandoned, and relaunched again fact that certain high-volume transaction agents acted as in a refined version. The decision to drop one solution for super-agents to smaller agents in their area, requesting a another has not necessarily yielded improved results. Only one small service fee in cash from the agents for each completed institution has outsourced the task to an external provider for transaction. At another institution, agents located close agent liquidity management support. While this super-agent to the bank’s branches frequently received e-float transfers aggregator has proven successful in urban areas, it was poorly from other network agents to deposit in cash into their bank equipped to deal with the challenges in rural regions where accounts. costs were substantially higher. When the institution faced internal budget cuts the service was completely abandoned and internalized, leading to challenges also in the cities. 15 Bank-to-wallet (B2W) and Wallet-to-bank (W2B) transactions allows bank account owners to move funds between their account and their mobile money wallet. 25 III. LESSONS LEARNED: PROMOTING DFS USAGE AND ACTIVITY DFS are able to redefine a business, but changes faced by staff and customers have to be managed carefully Figure 15: Share of DFS transactions over data reporting period BANK 1 share as of Dec 2017 - 39.7% BANK 2 share as of Dec 2017 - 37.5% BANK 3 share as of Dec 2017 - 60.9% 0-100% BANK 4 share as of Dec 2017 - 74.2% BANK 5 share as of Dec 2017 - 43.3% BANK 6 share as of Dec 2017 - 36.9% 0-100% BANK 7 share as of Dec 2017 - 30.9% 0-100% Notes: Data reporting start dates vary; for two institutions reported data includes transactions on mobile banking For some of the participating FIs, the digital channel offered The changes in business direction, however, created different an opportunity to reinvent themselves. Previously, all the perceptions and reactions among staff and customers. To institutions were credit-oriented microfinance banks, focused acknowledge and mitigate such negative sentiments, FI primarily on offering lending products, such as individual and management should implement a change management group loans. With time, the digital channel created means strategy that presents the benefits of the digital channel and to better understand customers’ needs and to begin offering recognizes the challenges in order to create understanding, other financial and non-financial products and services, acceptance, and commitment. To arrive at such a strategy, such as savings, government-to-person payments, and bill the FI should: payments. The ability to serve a broader market with multiple products fundamentally challenged the original identity of • Listen to and document the potential fears of staff some of the FIs, raising questions regarding ‘what we do’, regarding the new channel. ‘who we are’, ‘how we work’, ‘who do we serve’ and ultimately • Analyze those fears to identify the drivers behind them ‘where do we want to go’? As a result, some FIs in the study and develop ways to address or mitigate the risks. fundamentally changed their business identities. Most of the FIs now market themselves as multi-product and/or multi- • Design communication and training activities that service institutions. support the change and enhance staff understanding of the channel, highlighting the benefits the channel can bring to them. 26 • See change management as an open iterative process that The introduction of digital channels changes the touch points allows for two-way communication with staff. between the FI and its customers. In some cases, it makes face-to-face interactions less frequent while increasing the • Engage staff in the project, for example by asking staff ability for clients to transact at their leisure. It also broadens to pilot a service and to include digital KPIs for all staff, the variety of touch points between the FI and customers, among other measures. All staff should feel part of the requiring FIs to ensure a unified user experience across change and take responsibility for its success. interfaces (for example, agent and self-service through mobile applications). It is thus key to success that responses from During pilot and roll-out of the digital channels, FIs can clients to these changes are incorporated into the concept, implement specific communication initiatives to prepare prototype, and final version of the digital channel. Using staff for the cultural change; newsletters, videos, briefings, digital tools for collecting customers’ feedback (e.g. chat bots, and Q&A sessions with staff are some of the activities that online platforms, social media) on the new service can also management should consider. Appointing a charismatic help create increased literacy and acceptance of the service “champion” who is well-regarded within the institution to among customers by creating a broader digital engagement transmit core messages can also be useful for obtaining buy- with them. in throughout all staff levels. There is a delicate balance between the use of enhanced Successful FIs in the study also implemented early incentive technology and the expansion of financial inclusion. Modern structures for branch staff based on agent performance. technologies, such as smartphones, are still not available Organization-level performance KPIs that incentivize to much of the poor in developing markets. Focusing on harmonious integration between branches and the agent developing sophisticated technology can thus create a network help to mitigate staff fears about potential job losses mismatch between advanced technology and the FI’s and serve to enhance the credibility of messaging about customer base, backfiring in terms of poor uptake. It is thus the change process. Three of the participating FIs created important to have a solid understanding of the customer operational KPIs for branches linked to the performance of the profiles and segments that the institution serves and would digital banking channel. Two FIs introduced targets for field like to reach. For example, one of the FIs in the study created a staff on deposit mobilization and usage of the digital channel. mobile application only available on smartphones based upon One institution provided incentives to customer relationship the prediction that smartphones would soon overtake USSD managers to support the agent network as well as branches; technology. As a result, a large portion of the FI’s customers part of their salary was fixed, and the variable element was that were using simple feature phones were excluded, linked to the transactional performance of branches and affecting adoption and activity rates of the DFS overall. agents. At one of the study institutions, branches competed with each other based on the numbers of customers and transactions using the digital channel. 27 Customers easily enroll on new digital channels, but keeping them active requires effective marketing and handholding Figure 16: Figure 16: Customer registration and activity (in 1,000s) over data reporting period BANK 1 BANK 2 BANK 3 1000 5 400 800 4 300 600 3 200 400 2 200 1 100 - 0 0 BANK 4 BANK 5 BANK 6 150 150 100 100 100 80 60 50 50 40 200 200 20 - - - BANK 7 100 Registered and active DFS customers 80 60 Registered DFS customers 40 20 0 In general, customer acquisition grew during the time of • FIs’ initial focus on mainly registering clients: As early the study. All FIs were successful in acquiring and increasing adopters of DFS, several FIs in the study initially focused customers. For example, when looking at the number of on registering clients following the MNO approach of customers registered for the agent banking service, three doing mass registrations. However, once registered it institutions benefited from implementing the digital channel has been more difficult to engage customers to become as it helped to considerably enlarge their user bases. These active users. For most of the institutions, newly registered three FIs have evolved from a mono-product to a multi- customers did not initially transact at all or stopped product and multi-service approach, using the new channel to transacting after the first or second transaction. offer more value added services to customers. Consequently, they have increased the share of the total customer base • FIs offered too few services: Most FIs only provide the registered for agent banking, reaching 47 percent, 50 percent, minimum channel product offering, cash-in and cash- and 53 percent respectively, at the end of 2017. out. There was a lack of technology to collect savings from customers, and very little promotion of such services. It For other participating FIs, all banking customers are was a challenge to motivate active usage without a multi- automatically eligible to use agents without separate service offering. registration. However, the percentage of active users of agent banking has varied, fluctuating between a low of 14 percent • Customers did not trust the technology: Another to a high of 74 percent in the same period for the participating important factor for improving activity rates is to build FIs. None of the FIs reached 100 percent agent banking trust in the DFS. All of the agent banking deployments activity rate of their registered customers. in the study initially encountered difficulties with the technology and that created reconciliation and other There are a variety of explanations for the low level of transactional problems. It often led to an initial perception customer activity rates, ranging from customer perceptions by customers that the service was not reliable, which and trust of the service to provider operational capacity and adversely affected the adoption of the service. service offering: 28 This can be very damaging in the long run, and FIs should accompanied branch customers to agent outlets to show only rollout when they are sure the service is working them how to use the services. Uptake may be boosted by properly. increasing efforts to broaden and deepen financial literacy. In many cases, agents mentioned that customers claimed not • FIs did not adequately promote, market and to understand the DFS. In the case of agent banking, some explain the services: Most FIs in the study spent very FIs provided training refreshers to agents during their regular little -- when compared to the rest of their operational monitoring visits, in particular those agents that had been expenses – on customer acquisition, marketing and recently enrolled. promotional activities. Initially, FIs focused on doing mass acquisition and promotional campaigns that described Observations from the study highlight that it takes time the digital service. However, some mass promotions, e.g. to build customers’ trust in DFS. It has proved especially field activations, radio and newspaper ads, and flyers, challenging in markets where Digital Financial Services were did not clearly state the benefits or the value of the DFS a new phenomenon and implementing FIs were first movers. for customers. Some of these marketing campaigns FIs operating in this type of market needed to invest more in also failed to communicate how to use the service, in marketing, training, and handholding of clients to support the particular those services that were linked to a mobile learning processes and to build trust in the digital channel. In banking solution. markets where the use of DFS was more common, FIs tended to have an easier task building trust in their digital channels. As these challenges emerged, the FIs in the study adopted a Even so, effective marketing and promotion are critical to variety of approaches in response. FIs quickly learned that it ensuring the FI differentiates their offering from existing ones. was not beneficial to focus only on registering clients but that they needed to make promotion and marketing more When interviewing customers in the field, the research team effective. They looked at ways to better communicate and found that some customers distrust digital services because promote the value of the channel and to improve customers’ they lack information about the value proposition of agent technological and financial literacy. Three institutions, for banking or because they are afraid their money will “get lost in example, developed specific marketing tools that showed the system” in case of connectivity issues or human error. Self- how their digital services work rather than just describing the exclusion also evidently plays a role in keeping potential users benefits of using the service. away from enjoying the benefits of Digital Financial Services as some of the interviewees said they did not believe they had The approaches used to promote the services also effected the technical knowledge, financial education, or affluence to activity. For example, below the line (BTL) campaigns using use DFS. one-on-one approaches such as brochures and pamphlets generally proved more effective than above the line (ATL) Across all surveyed markets, lack of trust in agent banking initiatives using mass media. The BTL campaigns were more was seen as a major barrier to the adoption of DFS. To address effective because they helped customers not only through the this, the recruitment process should identify suitable agents registration process, but also to perform initial transactions that are able to address customer concerns. A good practice, while receiving support and information from field promoters. demonstrated by several of the FIs, is to recruit agents among A BTL approach also tends to be more effective at directly the existing customer base or agents who have a previous targeting the intended audience, low income users and the record of working with MNOs. Having trustworthy agents unbanked. For example, one of the institutions conducted also translates into branch trust - and even greater belief into market research that showed that the FI and its digital services the FI’s overall brand. As a way to reduce branch-dependency were largely unknown to the target market. It thus launched a and to create more trust through increased usage, two BTL campaign, including awareness and educational activities, institutions started charging fees for transactions at branches, that resulted in an increased response from the target market thus providing financial incentives for its customers to use about the FI’s digital offering. ATL campaigns, on the other DFS and build experience with the service. hand, have worked better with high-end customers, who, in this case, were not the intended audience of most of the FIs Another way FIs built trust in the channel was to use SMS participating in the study. as a proof of transaction. Users value text messages that confirmed their DFS transactions. One of the institutions has Secondly, FIs also adapted their digital channels to respond partnered with an external company that develops digital to particular customers’ needs. By highlighting this fact in marketing tools to engage customers via SMS, to further build marketing initiatives, FIs gave customers more compelling trust and ultimately increase usage of DFS. It has also been reasons for transacting through the channel. noted that providing agents with marketing materials (same colors and similar logo) in local languages facilitates service Third, FIs realized that customer onboarding needed the recognition by customers and increases adoption. In the case support of the marketing department and other branch staff. of one FI, the agent network branding eventually developed Such support included campaign materials, sales drives, calls a stronger appreciation among customers than the actual to customers to inform them of agent banking benefits, and name of the institution. redirection of customers from branches to agents for faster service. Field staff of one participating FI, for example, also 29 Designing customer-centric products and For the FIs in the study, the incorporation of value added services to distribute exclusively through products centered around the customers and their needs, and the digital channel helps to drive usage helped to ensure adoption and stickiness. and adoption It is important to note that the use of data is key to the ability As part of the digital transformation, a financial institution’s to develop digital services responding to customers’ needs. key business processes – such as loan evaluation, Data and information not only come from market research, disbursement, and repayment, as well as savings account but also from staff, in particular field staff such as cashiers opening and collections – need to be reviewed, improved, and loan officers who know and interact with customers and remapped according to the needs and potential uses of daily. It helps to have systems in place that enable the FIs to the new digital channel. Some new procedures will have to continuously improve services based on user feedback and be created, including procedures for customer enrollment, operational data. One institution analyzed transactional reporting, and fraud prevention. data in combination with agent enrollment information to understand and determine the profiles of successful agents. It is important that FIs keep users in mind when making these The use of mystery shopping, surveys, and focus group process improvements and changes. Users range from the discussions can also contribute valuable customer feedback institution’s staff, to agents and customers. Understanding on the channel. and including the user’s perspective allows the institution to design processes and services that are intuitive, thus IV: LESSONS LEARNED: facilitating change and supporting channel adoption. In order to this, FIs will find it helpful to: ENSURING DFS SUSTAINABILITY FIs need to understand the role of DFS • Identify the primary users of a process and invite them to participate in process development and to provide in the broader business and prioritize feedback on how to simplify or improve it. sustainability over instant profitability As the FIs in the study moved from planning to pilot and • Map out changes and discuss change with users to verify then to rollout, a common question asked at the board level that revised processes meet the expectations of users. and by senior management was “when will the DFS become • Pilot the revised processes with actual users before the profitable?”. The question, however, assumes that a banking revised processes are rolled out to gather additional channel by its nature is an income source when that may feedback and make necessary adjustments. not be the case. A channel can help generate value in other areas. The question should be rephrased to “when will the All FIs in the study showed interest in creating specific digital channel become self-sustainable,” i.e. generate enough financial products to be distributed exclusively through digital income to cover its operating expenses. Even that may only channels. But approaches differed. Some institutions followed reveal a partial appreciation of the value the channel brings to a traditional approach for product development, focusing on the overall business. designing products internally and then testing those products with customers. For example, for five institutions, the The development of the DFS business case and the financial traditional approach led to either the abandonment or low model to assess the long-term viability of the channel must usage of new digital product initiatives because customers be guided by realistic assumptions. As a nascent industry, did not find them valuable. Other FIs followed a more human- one of the main challenges for building the business case centric design approach for developing digital financial and a financial model for a digital banking channel has been products. For those institutions, the first step to designing the lack of sufficient information on industry benchmarks. digital products was to understand who the target customers Benchmarks are important to craft assumptions about the were and to learn about their motivations, fears, pain points, eventual performance of the channel. Most institutions and existing usage or perception of technology and financial participating in this study were first adopters in their services. A second step was to segment customers for tailor- respective markets, with some even having to guide their made products and services. Initially, the FIs segmented markets in the development of a DFS industry. customers in terms of products (e.g. loan customers versus Digital first movers are often forced to rely on their own savings customers) but then started to segment customers by assumptions or the experience of other ecosystem players, channel usage (e.g. agent banking users versus branch users), MNOs in the case of the nine participating FIs, to guide the and eventually segmenting customers by needs that could be development of initial business models. This approach misses, matched by products or services. to some extent, the differences in the business nature and While most of the institutions started with basic cash-in/ culture of the different players of the DFS ecosystem. For cash-out services, with time, the product offerings on the example, the MNO business model for mobile money agent digital channel evolved to include value added services such as deployments calls for a mass-market approach, whereas transfers, bill payment, nano loans, etc. These new products financial institutions look to serve specific market niches and services also provide opportunities to add direct revenue through their agent banking channels. This is particularly streams to cover operational expenses and help to make the the case when developing assumptions related to outreach, channel self-sustainable. customer transaction patterns such as transaction size and frequency, service offerings, and revenue generation. 30 Interviews conducted and information collected for this institution with an agent network in operation for more study led us to the conclusion that many FIs followed than three years and with a high number of transactions “conventional wisdom” in believing that digital channels flowing through the channel had a commission expense – mainly agent and mobile banking – were inexpensive to representing over 60 percent of the channel’s overall implement and inherently cost-effective because they lacked operating expenses. For other study institutions, the physical and human infrastructure required to expand commission expenses tend to be the second or third through traditional bank branches. However, many of the biggest expense category after technology and agent participating institutions came to realize that although management costs. upfront investments for digital channels were not as large as for branches, other recurring operational expenses, such • Technology: This refers to any ongoing fee, such as as commissions, technology platform maintenance, and software licenses or acquisition of hardware, including channel management could represent significant ongoing point-of-sale systems or tablets. Technology operating costs for the institution. expenses includes any type of annual maintenance fees for the digital channel software as well as any functionality To quantify the initial investment for a digital channel, there improvements that require a development fee. are four main cost categories to take into consideration: • Human Resources for Agent Management: This • Technology: Any investments related to the acquisition cost includes the salaries and incentives for the staff to of the hardware and software necessary to offer the manage the digital channel, including increased HR costs digital channel. Investments in technology platforms can for customer service support, as well as technical and be considered as capital expenditures on a balance sheet transaction rectification support. Any incentive given rather than as direct expenses on the income statement. to branch staff to promote usage of the digital channel should also fall under this expense category. • Human resources: Initial expenses for building the DFS team. This could cover head hunting fees for scouting a • Marketing: This refers to the cost of running above-the- digital channel champion and training existing personnel line and below-the-line marketing campaigns to promote to develop internal capacities. Institutions should also the adoption and usage of the digital channel. consider the costs of any consultant work to help in the design of the strategy or initial set-up of the channel. In the case of agent banking, operational expenses are mainly driven by the number of active agents. This activity plays an • Market research: This relates to expenditures for market important role in calculating the amount of commission research activities to better understand users’ attitudes as expense, as well as in the level of the resources necessary to well as existing DFS usage patterns. This could also include manage the agent network. Furthermore, keeping agent upfront market research for the design and prototyping of management costs low is important. This can be achieved new digital products. All institutions in the study invested by having good quality agent networks (i.e. agent activity in market research prior to launching a digital service. rates over 60 percent) and keeping the agent management team small. Furthermore, it is important to note that all FIs • Legal fees: Any expenses related to obtaining regulatory in the study shared the strategic objective of reducing branch approvals and licenses. This might include the cost of operational expenses by implementing agent networks. The creating a separate company to host the digital channels, data from the study shows that the FIs were able to reduce as in the case of countries that require a financial branch operational expenses by reducing the size of the team institution to have a separate company to run an agent needed to handle daily transactions as well as reducing the banking business. Three financial institutions in the number of new branches needed to reach new geographies. longitudinal study had to set up different companies to host agent banking networks to comply with local In terms of the income, it can be categorized in two groups: regulations. Additionally, these fees can include the cost direct and indirect. Direct income refers to the fees that an of processing contracts for each agent, bill payment institution can charge customers for transacting through aggregators, partnering with MNOs, etc. the digital channel. The financial institution should consider market conditions such as what competitors or similar DFS In the case of recurring operational expenses, the study providers charge to set up a pricing structure that can both found that the expenses categories are similar to the initial contribute to covering the operational costs of the service and investments. Using as a base the agent banking model, the foster customer uptake and usage. potential operational expenses for a digital channel could include: In the study, the participating financial institutions with an agent network generated direct income from the following • Commissions: Some digital channels such as agent transactions: banking, automated teller machines, and card payments, require FIs to pay a commission for each transaction • Cash-out fees: All institutions with an agent banking done through a third party. For the eight participating channel in the study introduced such fees. For those institutions with an agent banking channel, commission institutions not charging for cash-ins or not offering expenses paid to agents played a key role in quantifying additional fee-based services, the cash-out fees represent operational costs. The data from the study shows that the gross of the fee income generated by the channel. commission expenses tend to be one of the largest costs of an agent banking channel. As a case in point, one 31 • Cash-in fees: One institution in the study charges • Deposit mobilization: Through deposit mobilization, customers to deposit money into their accounts. Although banking institutions can benefit from the difference charging customers for cash-in is not common in similar between the lending and savings interest rates - also services (i.e. MNO’s mobile wallets), for this institution known as intermediation - and use a cheaper source this measure significantly contributes to covering the of funds to support lending activities. By including channel’s main operational costs and to reach break-even. intermediation in its financial calculation, an FI can strengthen profitability and the business case for a digital • Transfers: Fees charged for person-to-person transfers, service. either over the counter or through a deposit to an account, are another way to generate fee income from a digital • Interest income from the digital channel: Although channel. Four institutions in this study offer the service via many institutions have difficulties attributing savings agents, but fees generated by this service represent only a mobilization to the agent banking channel, all institutions small portion of the channel’s overall income. report that most cash-in transactions are done for loan repayment purposes. However, none of the financial • Payments: Two institutions in the study offer some institutions allocate a portion of the interest income from type of payment service (i.e. bill payments, school fees, their agent portfolios to the profitability of the agent government-to-people, payroll) through their agent banking channel. The logic behind allocating a portion of networks. In most cases, the fees received for these the loan portfolio’s interest income to the agent banking payments represent a marginal portion of the income channel is to recognize the contribution of this digital generated by the agent banking business. channel to the delivery of loans. In two of the participating FIs, the agent banking channel established itself as the • Account Opening: A portion of the income fees generated main outlet for loan repayments. It can be inferred that for the account opening should be allocated to the digital without the channel, servicing those loans might not be channel if new client acquisition, account opening, and possible or could be costlier for the institution. KYC collection is completely done through the digital channel. At the time of the study, none of the institutions • Cross-selling of services: The ability to cross-sell other were generating revenue by opening accounts. However, products should also be considered as a potential indirect two institutions in the study were exploring options to income for a digital channel. For example, the distribution charge customers for account opening through agents. of nano loans through an agent network generates interest income to the financial institution. Part of that • Other fees: Financial institutions can consider charging interest income should be allocated to the agent network, customers to perform balance checks, view mini- especially when it is the main channel for distributing and statements, or for general account maintenance. Two servicing those loans. institutions in the study charge customers a small fee to check account balances at agents. None of the institutions The calculation and analysis of indirect income streams for the in the study are currently charging maintenance fees for digital channel should be used in the context of understanding accounts that are directly linked to the agent banking the profitability of the channel and its progress towards channel. break-even. However, the indirect income calculation might not be used for formal accounting purposes, such as putting Quantifying indirect income can be a challenge because it together financial statements or reports to regulatory bodies. requires financial institutions to look beyond the standard income sources of the digital service as well as being able to Additionally, it is important to emphasize that income is driven quantify indirect income attributable to the digital channel. by the number of customers adopting and using the service. Most FIs in the study reported difficulties identifying indirect Therefore, it is essential to address operational problems that income sources as well as calculating and deciding on the might hinder the adoption and usage of the channel. portion of overall income to allocate to the digital service. Based on information from the nine participating FIs, it is possible, however, to identify three main potential sources of indirect income for the agent banking channel: 32 Digital channels allow FIs to operate in • Designing new services to cross-sell through the locations that would have been too costly digital channel: Two FIs decided to explore another to serve with branches approach to reach channel self-sustainability by cross- selling nano loans. The nano loan product is exclusively The study shows that six financial institutions in the study offered through agents. These loans are generating experienced some level of benefits from the channel, including enough interest revenue to cover a good portion of cost savings and growth opportunities. The agent banking agent network expenses – including agent commissions channel contributed to cost reductions in the following areas: – and pave the path to making the agent channel self- sustainable. • Cash-handling: Three institutions reported that the agent network contributed to a reduction in costs It is important to communicate and explain to customers and associated mainly with transportation, insurance and agents the reasons for price changes. Agents should be seen dealing with counterfeit bills. as an ally when communicating fee changes and FIs should ensure agents are well trained to respond to customers’ • Branch operational expenses: Due to less client traffic questions about new fees. As an example, one FI in the study at branches, four institutions were able to reduce the – after revising its fee structure and facing adverse client number of tellers, which lowered the operational costs of reaction – decided to hold open forums with agents to present branches. the reasons for changes in fees and listen to customers’ • Transactions: One of the FIs found conducting a concerns. transaction at an agent was 25 percent less expensive than The business case for savings mobilization at a teller. Another realized a 17 percent cost savings per transaction when using agents for deposit mobilization. is yet to be proved, but evidence suggests a positive impact It is also important to note that agent banking has significantly Savings mobilization was a strategic motivation for the contributed to the growth - in terms of clients and portfolio - implementation of a DFS for most of the institutions. From the of five of the institutions in the study. Three institutions in the study we know that deposit transactions for loan payments study stopped relying on physical branch market expansion are most common. However, it is important to also forecast because they found agent banking to be a more cost-effective and track cash-in transactions for savings-only purposes way to improve both market penetration and expansion. to assess the effects in savings mobilization of the digital channel. Although the potential to mobilize savings through a Innovation can be a driver to reach digital channel is theoretically high, due to system limitations, channel sustainability that potential could not properly be quantified for most of the Innovation has been key for reaching channel self- financial institutions in this study. sustainability. From the study, the two FIs that are currently covering their operational expenses have both been innovative Two institutions tracked the portion of cash-ins for savings. and unafraid of challenging conventional wisdom: One institution did not see any direct increase in savings mobilization for agent banking users over the course of the • Questioning the reasons for not charging for study. The other, however, reported a seven-fold increase cash-ins: One of the two institutions currently charge from January 2017 to December 2017. The same institution also customers for doing cash-ins at agents. Agent cash-in experienced a 17 percent cost savings per savings transaction transactions are traditionally free for customers; but they through the digital channel. In addition, the total savings can generate a large operational expense due to the need portfolio more than doubled after the launch and roll-out of for paying agents’ commissions for this service and costs digital channels at three FIs. associated with liquidity management. Although the initial reactions of customers to the new cash-in fees were negative and led to a decline in transactions of around 30 percent, the FI decided to continue with the fee and roll it out to all its agents. After more than a year with the new cash-in fee and improving the way this new fee is communicated to clients, the FI is seeing the bold strategy bear fruit by reaching operational break-even of the agent banking channel for the first time in the last quarter of 2017. 33 PRESENT PROGRESS AND OUTLOOK This section zooms into each of the participating institutions 2. FIs are developing new products and services to distribute individually in order to summarize their DFS journey and via their new channels (utility bill payments, nano suggest what lies ahead. The team also found some common loans, savings collection, etc.). FIs are also developing trends among all the participating FIs: partnerships to offer these products and services. 1. All of the institutions fully launched their first digital 3. Two institutions are on the way to channel self- channels, and all of the channels are operating and sustainability – mainly achieved through fee income from expanding. In some cases, these channels established new products, revision of fee structures, etc. as the main distribution channel for the FIs as per the transaction volume handled by the channel (see Figure 4. FIs are also developing partnerships to integrate digital 17). Many of the FIs are planning or already piloting further channels with other DFS solutions (e.g. mobile wallets, digital delivery channels. payment aggregators). 5. The banks are testing adopted DFS approaches (agent kiosks, roving agents). Figure 17: MFIs' shares of DFS transactions via digital channels as of December 2017 39,7% 60,9% 37,5% 74,2% 43,3% 36,9% 30,9% AccèsBanque Madagascar AccèsBanque Madagascar started planning their agent AccèsBanque’s future strategy revolves around three pillars: banking channel several years ago, however, it took over two digital channels, digital processes, and user experience. years to gain regulatory approval to launch. That delayed their It is investing in both a data warehouse to build business plans and inflated their costs. At the end of 2017, they piloted intelligence capability, and in social media and call center the channel with 10 agents, and in 2018, it was officially channels for customer support. The new digital strategy also launched under a new wholly owned subsidiary, targeting foresees growing into a full-service bank, including digitization existing SME clients to act as agents. Since the launch is fairly of back-end processes such as loan applications and Customer new, it is difficult to determine the long-term sustainability Relationship Management (CRM). This effort will link all of the channel, although it has been well received by the customer data to help the bank better understand and 90,000 customers of the bank and is viewed by management serve its customers as well as reduce the risk profiles. Many as key to their long-term growth and competitiveness in the facets of this strategy are grounded in the agent network market.During the wait for approval, AccèsBanque took the as the main digital channel for customer interactions. time to develop a risk management strategy and a customer The data and digitization will allow the bank to become engagement strategy that will support its initiative going customer-centric; to better service both the customers and forward and ensure steady growth with a viable offering. the agents, to build meaningful products, and to provide efficient and responsive service. 34 AB Microfinance Bank Nigeria AccessBank Tanzania AB Microfinance Bank Nigeria began offering digital financial AccessBank Tanzania launched its digital service at the end of services in 2015, providing debit cards, a mobile application, 2014, and has continuously improved and extended the service cash-in/cash-out via a third-party agent network, and since. The original customer offering included a USSD-based correspondent banking to allow customers to repay loans mobile account and an agent network for cash-in/cash-out. and withdraw cash through other banks. Central to the Banking agents are not permitted to register new customers deployment is a partnership with a local supplier, eTranzact, in Tanzania, so the bank created the position of Customer which provides the whole channel infrastructure including Marketing Officers (CMO) to recruit new customers directly software, switching, card solutions, and an agent network in the field. The introduction of transfers between its mobile for the bank’s digital services. Technically, its accounts are also account and the largest MNO wallets in Tanzania allowed integrated with the eTranzact “Pocketmoni” wallet, allowing AccessBank customers to use MNO agents, greatly increasing customers seamless access to mobile wallet transactions. agent access. More recently, group savings, mobile-accessed loans, and agent overdraft facilities have been piloted, among AB Microfinance Bank is a nationally licensed microfinance other initiatives. bank with about 93,000 active customers. AB Microfinance Bank sees its key competitors as being the low-end players in AccessBank Tanzania recognizes that it faces direct the mainstream banking sector. The banking regulator has a competition from a number of banks actively deploying clear financial inclusion mandate and the bank’s commercial mobile and agent banking. Nevertheless, more than 55 banking competitors are also starting to invest in this sector, percent of its 62,000 active customers are registered for the targeting the same small business customer base. services and more than 60 percent of the bank’s transactions are now conducted via mobile app and agents. While bank Overall the organization has found digital transformation to branches remain important for high value customers, opening be a positive experience of strategic importance in order to new branches is expensive, and AccessBank is now focusing remain competitive, even though uptake has been lower than on extending its geographical reach by growing its agent forecasted, especially for the debit card product. Nevertheless, network, with limited branch expansion. Agent recruitment about 21 percent of its customers are active on the mobile has been greatly improved in 2017 by a change in regulation, application and more than 40 percent of the bank’s financial and agent applications no longer need direct approval from transactions are now conducted out of the brick-and-mortar the central bank. The institution is in the process of more than branches. Central to the growth of the DFS is the delivery of doubling the size of its agent network. It also plans to create the group omni-channel strategy, which allows customers a special category of agents that can collect registration to use the mobile, online, agent, and branch channels documentation from potential new customers and collect interchangeably from the same account. As AB Microfinance loan application documentation from existing ones. Following Bank prepares for this, it is also working to support the pilot, selected agents are being given access to short-term infrastructure such as process efficiency and improvements in overdrafts to assist with float management. In addition, the CRM, web design, call center, and data infrastructure. Better CMOs are being replaced by Relationship Officers that will customer understanding and segmentation is needed. To combine the roles of CMO and loan officer. meet that need, new business intelligence (BI) staff has been recruited and a firm has been engaged to work on data-based Tanzania has seen several high-profile cases of DFS fraud analytics for scoring. Work is underway to deploy point of sale recently, and the institution’s strategy to protect its customers (POS), to pilot cashless e-branches, to develop financial credit and its business from such incidents involves creating a scoring, and create an overdraft facility for its customers. comprehensive program of staff training and raising customer With several years of customer data on file, the marketing awareness of safe practices. There are also plans to automate team plans to build future campaigns around key customer controls and key business processes that are currently being segments by making more use of digital marketing, such as done manually. pro-actively promoting the benefits of core products and digital channels. Product development is a priority for AccessBank Tanzania, and several new initiatives are planned. A new mobile wallet Risk management is an important area of development and platform is in development that will add smartphone access the bank’s risk strategy is moving from a credit risk focus to a to the current USSD channel, allowing customers access to a wider range of categories. In particular, new fraud prevention full suite of account management functionality. AccessBank policies are being implemented, including a drive for greater admits that this solution will initially focus on more affluent automation of some processes. Aware of the risk associated smartphone users but it is expected to spread as ownership with a single IT and infrastructure supplier, new partnerships grows. A digital lending service is under development, based are being sought. upon automated credit risk assessment. The “Kikundi” group lending account is being rebranded and relaunched with improved targeting on key savings groups such as church and company groups. 35 Advans Cameroon FINCA DRC Advans Cameroon has adopted a digital approach since 2016, FINCA DRC was founded in 2003 and, with more than 270,000 to support its strategic objectives of mobilizing deposits and customers, is the largest microfinance institution in DRC. increasing satisfaction of its more than 70,000 customers. In 2011, it was the first institution in the country to pioneer To improve clients’ service accessibility, the agent network agent banking and it now has a relatively mature agent is deployed across the country and provides cash-in/cash- network of more than 1,300 agents, which is still growing. out, account information, and transfers between accounts, The agent network is currently responsible for more than 80 all focusing on high-quality partners as well as customer percent of its transaction volume. Agents can perform cash- experience. Operations are performed in real time using a in/cash-out, loan repayments, bill payments, and transfers secure digital application linked to Advans’ core banking between accounts. Agents are given POS devices to process system. In line with its strategic objectives, complementary transactions. FINCA has used field staff to sign up clients, “outlets” have been deployed in remote areas to increase collect biometrics, and create personal relationships during clients’ access. These outlets handle customer registration, the recruitment process, but soon will leverage the agents to loan applications, and other administrative tasks normally also perform these processes. managed at branches. Outlets do not handle cash, which is managed through the network of agents. In addition, to A major component of FINCA’s mobile banking strategy is to encourage customers to frequently save, “mobile collectors” develop a mobile banking solution that can be accessed by visit clients in the field on a daily basis to collect savings USSD and smartphone app. Users will be able to pay bills and deposits, using the same digital application as agents. They transfer funds to FINCA and mobile wallet accounts. A new perform a service similar to traditional “susu” collectors. mobile savings and loan service is being created for existing customers and as a means of attracting new customers. Advans’ innovative digital approach has proven successful in This activity will be supported by improved performance growing the number of savers and the frequency of deposits. and monitoring tools, increased staff training, and a new The agent network now accounts for a substantial proportion monitoring and alert system that is currently in development, of the institution’s mobilization of deposits. More than two years after the launch of agents, almost half of the institution’s As MNO wallets become popular, some have been integrated overall cash transaction volume is conducted through the with FINCA accounts, opening up a wider range of agents agent network and mobile collectors. As a result, Advans is to customers for cash transactions. In addition, this allows reviewing the size of its branch network. In order to reduce FINCA to develop a micro savings and loan product for M-PESA costs, Advans may diminish the size of some branches and users in partnership with Vodacom. The product will be similar rely on alternative channels for cash services in the future. In to M-Shwari in Kenya. The first step will be to launch a nano addition, further geographic extension will be through the savings product, followed by micro-loans of increasing size as development of mini-branches or cashless outlets only. Both the automated customer risk scorecard algorithm is validated. models are relying on the agent network. All employees of For larger loans, FINCA sees this as a recruitment tool, advising the institution are fully engaged in the change process, and applicants to visit a local branch. For the last 2 years, FINCA aware that it benefits the general efficiency of operations, has been acting as Superagent for the MNO wallets, assisting also offering the opportunity for personal development with their agent liquidity management. Many FINCA agents through a new scope of work linked to these innovations. also offer MNO wallet services and currently need to keep a To both, promote the new services and effectively engage all separate float for each. The multiple floats add to the agents’ employees, “operation flotte” involves all the institution’s staff liquidity challenges. FINCA has identified an opportunity to having to undertake monthly door-to-door field activities. act as the holding account for all agent services so that agents Finally, Advans’ digital roadmap also includes integrating with would need only one float that could be used in real time for local MNO wallets to provide transfers between accounts any network, or for FINCA transactions. and MNO wallets (so called “bank to cash”), although this has suffered some delays for regulatory reasons. These DFS LAPO Microfinance Bank Nigeria activities are central to the organization’s engagement to LAPO Microfinance Bank is a leading Nigerian MFI with over increase client accessibility and satisfaction with Advans’ 400 branches in 34 of the 36 states in Nigeria. Its customer services, to improve productivity, and to reduce costs. As its base of over three million is split between individuals and digital financial services grow, Advans is planning to update group members who benefit from the bank’s credit and the tools and processes for risk monitoring. savings products. With ambitious growth targets, LAPO started an agent banking pilot in May 2017 to determine how Advans is becoming increasingly customer-focused and DFS can support its business targets. Through the service, recognizes the need to better understand the new savings which was recently rolled out, agents are able to perform activity that its DFS activity is bringing about. In addition, cash-in/cash-out and register new customers for mobile it wishes to grow its client-base of young entrepreneurs, banking. As a result of the pilot, LAPO’s focus has moved from especially women. Quarterly market studies and analysis is quantity to quality, concentrating on agents that are able to conducted in order to better understand customer behavior, represent the bank’s business to a high standard and provide needs, and preferences. excellent customer service. Agent quality controls for its 238 agents will be implemented and new processes developed to monitor performance. 36 LAPO recognized that field staff need to be more involved in the Senegal affiliate. To increase agent traffic and income for the agent network than they were in the pilot. The reporting both parties, “bank-to-cash” is planned, allowing customers to line of agent relationship officers has changed to the branches send electronic vouchers that can be cashed out by recipients and area officers, and eventually zonal Agent Network without a bank account. Agents will also be able to offer over- Managers will be recruited. Understanding the importance the-counter bill payments for utility bills. As a consequence of of partnerships, LAPO is creating a partner manager role to agent expansion and the rising popularity of agents, Baobab target agent recruitment of specific retail categories such as Madagascar has decided not to open more branches, and petrol stations and the Nigerian postal service. In addition, existing branches may be rationalized or turned into agent LAPO is seeking to partner with the payment processing management hubs. company Interswitch to enable bill payment collection. Baobab Madagascar accounts are now integrated with the The pilots clearly identified the need for staff education about Orange Money wallet, substantially increasing customer DFS, and several training sessions have already taken place in access to cash-in/cash-out; and connection with the other the Benin training center; these are being extended to create MNO wallets is in progress. In addition, there have been a formal training program. LAPO is also developing online ongoing discussions with Orange about partnering to provide training via an e-learning platform. A pro-active change a micro- savings and loan service like M-Shwari in Kenya, management process is being developed, with branches and a contract for collaboration has been signed. A mobile involved in agent management and staff reassured that their banking app for smartphones is on the institution’s roadmap jobs are not at risk. but as a lower priority. The group has received funds to develop advertising on Facebook, the only social media currently used LAPO’s future developments include bill payment via in Madagascar. The option of trying once again to introduce Interswitch, and provision of MNO airtime top-up. Based cash-in fees (as charged by Baobab Senegal) is also under upon customer feedback, a debit card linked to a savings consideration, but management is cautious because the fees account is being developed from which customers can access were poorly received last time they were tried. their loans, among other things. Baobab (MicroCred) Senegal Baobab (MicroCred) Madagascar Baobab Senegal has a similar DFS implementation strategy to Baobab Group is a credit-focused network of financial its Madagascar sister company, launching the “Baobab” agent institutions targeting the MSME sector. Baobab Madagascar banking network, then changing its company name to Baobab was one of several Group companies to launch the “Baobab” and adopting a long-term digital strategy. One key difference agent banking network. It has proven so successful that the is that Baobab Senegal is used by the group as an incubator whole group and its subsidiaries have rebranded as Baobab. for new ideas. The bank is committed to a customer-centric The strategic driver for Baobab Madagascar was to extend digital strategy that provides core services more efficiently access to new deposit accounts cost-effectively. This was to its more than 300,000 customers; extends its reach via accomplished using a shared technology platform, linked agents; and uses DFS to provide new services. The network’s to agent laptops and tablets with biometric customer 500 agents are linked to the nearest branch, and the branches identification. Agent liquidity was facilitated by providing are closely involved in the agent management process. agents with MNO wallets. When the regulator froze agent expansion for several months, the bank focused on improving Because of profitability issues, Baobab Senegal took the bold agent quality, resulting in increased agent usage and decision to start charging customers for deposits at agents. It productivity despite a halt to network growth. At the end of was the only institution in the study to do so. That decision led 2017, almost half of the active customer base of 53,000 was to a short-term drop in usage, though volume recovered over actively using the agent points, accounting for 43 percent of time, and by the end of 2017, more than one third of the bank’s the total cash transaction volume. The recently launched cash transaction volume went through agents. The experience “Taka” nano loans that can only be accessed via agents are also led Baobab Senegal to launch a successful agent overdraft also increasing the popularity of the agent channel and the facility that was originally designed to offset the agents’ additional loan income offsets a significant proportion of the reduction in income but ultimately allowed agents to serve agent management costs, helping the digital channel to get more customers. With an improved commercial footing the close to self-sustaining. institution is developing a number of new initiatives. A team of designers has been recruited to work in the field mapping Baobab Madagascar intends to pursue its digital strategy customer behavior and identifying ways to improve the user with a focus on achieving profitability and launching new experience for new and existing products and services. digital products. A priority is to grow the agent network and thereby increase geographical coverage. This includes the creation of exclusive agent kiosks using lessons from a pilot at 37 Bill payments are to be added to the agent OTC services and 2019, more than 70 percent of its group customers’ loan the bank is in negotiations to partner with a local payment repayments were collected via agents. aggregator for utility bill payments. That is expected to attract new customers rather than be a major source of revenue. Urwego has updated its strategy and is taking a step-by- Automated credit scoring is being piloted prior to roll out, step approach to improving its processes with the goal of and this is happening in parallel with the development of an converting all group loan disbursements and repayments to improved data warehouse, both for this project and to inform mHose. Addressing internal buy-in for the digital strategy has CRM activities. Planned SME overdraft facilities and flexible been a key component of making mHose successful. To enable guarantees will also benefit from this project. Recognizing a smooth transition, agent liquidity needs to be improved and the importance of partnerships in this sector, discussions are Urwego is considering a number of options to facilitate this. As underway with retail banks and MNOs with wallets to enable agents take a more prominent role in service delivery, Urwego account transfers. Senegal has a large diaspora, and it is hoped has empowered branch staff in the management of mHose, that the major money transfer organizations will integrate scaling down the role of the head office. New customers with Baobab Senegal as a destination for remittances. currently have to enroll at a branch, but going forward, agents will be able to register them immediately unlocking The institution has successfully piloted so-called “agent some of the bank’s services while KYC is being fully processed. kiosks,” i.e. exclusive and dedicated agents in small Baobab Along with improvements in DFS, Urwego has retooled its outlets focusing on customer registration. Agent kiosks use fundamental credit origination and recovery systems and the PULSE application via tablets, which is substantially procedures. faster than the account opening process at branches. It is also developing a mobile application for its customers. Urwego Bank (UOB) Rwanda Urwego Bank, formerly UOB, is a Rwandan microfinance bank offering a full suite of banking services to its more than 300,000 customers. Over the course of this study, HOPE International, a Christian faith-based nonprofit organization, acquired majority shareholding in Urwego from Opportunity International, a non-profit development organization. Urwego provides a USSD mobile banking channel and an agent network of 158 active agents branded mHose. The primary objectives for the digital channel are furthering financial inclusion, removing cash handling from loan officers, recruiting new customers, and driving efficiency. In February 38 CONCLUSIONS The main objective of this study was to understand if Digital Hence, to obtain the buy-in of staff and overcome fears, the Financial Services could help financial institutions expand participating FIs started to put in place different change their markets and provide financial services to excluded management approaches that included better communication segments of the population. It also sought to examine scale with staff, training opportunities for employees, and and outreach of digital channels as well as their impact on the empowering branch staff to have responsibility for the success institution’s overall operations. For this purpose, the study of the DFS. worked closely with nine FIs in seven countries in Sub-Saharan Africa to document their experiences in the design, piloting Data and rollout of their digital offerings. In the documentation The most successful DFS implementations in the study of these experiences, the research team was able to identify used strong data-driven approaches to improve customer important lessons for other FIs exploring the use of DFS. Those service experience, design better products and/or to monitor findings center around four main areas: operations. While most of the institutions started with basic cash-in/cash-out services, the FIs understood the need to offer Strategy and business case a wider range of digital products to expand customer adoption The study demonstrates that digital strategies should not be and increase the volume of transactions. For this purpose, the static and must be responsive to customer feedback as well as use of data played a key role, in particular for the development changes in market conditions. The strategic objectives of the of products that needed a fully digital experience. Additionally, participating FIs evolved over time as the institutions gained a three FIs employed data to better monitor operations. more comprehensive view of the benefits of digitization. For example, one of these FIs experimented with agent segmentation and predictive tools using transactional data to Initially, the strategy for implementing digital channels better support agent liquidity management, which in the long focused on increasing the institutions geographic footprints term, helped to improve customer experience. and expanding their client bases – especially in rural areas – while keeping the cost of expansion low. However, DFS management the path towards those goals was not always clear. For example, financial institutions that were aiming to expand Most of the FIs in the study started their DFS offering from beyond brick-and-mortar did not initially realize the level of scratch. Agent banking was the most common Digital investment they needed to make in technology and personnel, Financial Service deployed in the study. For these deployments, and how those costs would affect the DFS business case. FIs needed to build the internal capacities (people, processes That underestimation was due in part to FIs’ misperceptions and systems). The task of building these capacities was not regarding business growth and the lack of peer information easy for most FIs. Also, the costs for these capacities were necessary to build a business case. Furthermore, as the digital largely underestimated, in particular DFS staff management ventures matured new areas gained more relevance, such costs. Another challenge identified in the study was finding as focusing on improving operational efficiencies, providing the appropriate liquidity management approach. That was a better customer experience, and offering a wider range of trial and error process for the FIs. financial and non-financial services. In terms of agent banking, the study noted the importance Nevertheless, the findings from the study show that at least of finding the right size of the network. Most FIs in the study three FIs have become successful with their DFS operations, have learned to prioritize quality over quantity. The most demonstrating that Digital Financial Services are financially successful agent networks in the study are not necessarily the viable and can help FIs grow their businesses by reaching new largest. The agent networks that focus on high quality active clients, increasing deposits, and reducing operational costs. agents rather than high numbers of agents have proven to be However, the remaining FIs in the study are still fine-tuning the most financially viable. This is well demonstrated by the their strategies. case of one financial institution. Despite a number of external factors that forced it to close 25 percent of its agent network, a Internal buy-in strategic decision to focus on agent quality resulted in its DFS business continuing to grow. The FIs in the study realized the importance of obtaining the buy-in of staff for the digital venture. The institutions noted that they could have a good strategy, but this would be worthless if the strategy was not supported internally. All of the participating FIs faced different degrees of internal resistance to their digital channels. The most common resistance came from branch staff, who in some cases felt as if their jobs were threatened by the digital transformation. Some staff fears were justified as FIs did not always properly communicate the rationales for the changes or how the DFS could positively impact their daily tasks. 39 Additionally, agents must be properly trained and rewarded portfolio for most of the participating FIs grew during the for their efforts otherwise they tend to not stay with the FI time of the study. Some of this growth seems to have a or do not properly sell the digital service. Hence, FIs need to similar trend as the growth of DFS. However, this relationship ensure that there are sufficient clients visiting each outlet and was not properly proved and quantified because of system that the commissions that agents earn are competitive. limitations. Only two institutions in the study have data that allows for tracking the savings trends of customers, with one The study also intended to document the impact of savings bank showing a great potential for saving mobilization as well mobilization on DFS. The results on this aspect are mixed. as a reduction in the costs of managing those savings. First, the potential to mobilize savings through a digital channel is in theory high. For example, the size of the savings 40 ANNEX: REGULATORY OVERVIEW OF STUDY COUNTRIES RELEVANT BANKING DFS REGULATION CHALLENGES REGULATION • FIs supervised and regulated by the • Mobile money regulation (2011): Strict bank-led model Bank of Central African Countries Only credit FIs are habilitated to MNOs and Fintechs need to partner (BEAC) with powers delegated issue mobile money after approval with banks in order to provide mobile to the Central African Banking from the BEAC money services. Although the law Commission (COBAC) • MNOs are specifically regulated does not forbid the delivery of mobile • MFIs are members of the National and supervised by the savings and mobile loans, mobile MFI Association (ANEMCAM). Telecommunication Regulator money services offered by banks • Three categories of MFIs with (ART) are still limited to money transfers, specific governance rules: • Multi-banking (2014): Conditions airtime purchase, and bill payments. o SACCOs and Credit Unions for mobile money providers to o Deposit-taking MFIs make their mobile banking systems Limited mobile money regulation CAMEROON o Credit-only MFIs interoperable via a dedicated multi- The mobile money regulation provides • Foreign transactions are prohibited banking structure (switch) limited guidelines regarding the for MFIs. • BEAC suspended all mobile money activity of MNOs in the space. It transfers outside the CEMAC zone does not define the types of mobile in June 2017. products which are allowed, nor does it specify KYC requirements over mobile money clients. No chapter is dedicated to the regulation of agent activities. Prudential norms represent a burden to finance SMEs MFIs have mostly short- / medium- term financial resources. • FIs regulated and supervised by • BCC released a regulatory Challenges according to IMF: Banque Centrale du Congo (BCC) framework on e-money: (Directive • Banking Law to be revised soon • Five categories of FIs: #24, 2011): It allows non-banks for strengthening prudential o Banque to set up a subsidiary to provide regulation and empowering BCC o Coopérative d’épargne et de e-money services. (liquidation of banks in bankruptcy) crédit • E-money is provided by MNOs • BCC is generally weak with serious o Caisse d’épargne institution (considered as “Financial gaps in terms of AML/CFT o Financière spécialisée Companies” licensed and regulated • National Payment System is being o Société financière by the Banking Law. strengthened but remains weak, • Two categories of MFIs: • Money transfer companies have notably in terms of guarantees o Micro-credit Enterprises (cannot extensive networks and are widely • The regulatory framework for collect savings) used (a barrier for DFS in a cash- e-money ought to be reviewed to DRC o Microfinance Companies (can based economy). improve competition and improve collect savings) • MFIs are authorized to distribute customer protection. • Only Microfinance Companies are electronic money. subject to prudential regulation of BCC. • BCC manages a credit bureau for commercial banks, there is no credit bureau for MFIs. • KYC principles are reinforced with stronger emphasis on incoming and outgoing transfers than deposits 41 RELEVANT BANKING DFS REGULATION CHALLENGES REGULATION • FIs supervised and regulated by • Financial service providers can The new mobile money framework the Commission de Supervision directly offer mobile money is not fully implemented; Bancaire et Financière (CSBF) of services upon formal agreement persistence of the old informal the Central Bank of Madagascar from the CSBF. mobile banking system (Banky Foiben’i Madagasikara) • Non-financial service providers A new law (Law 2016-056) is • Banks are members of the have to create a separate entity regulating mobile money and mobile Professionnal Association of Banks licensed by the CSBF. money providers as the first legal (APB). • Mobile money providers cannot framework for mobile money ever • MFIs are member of the directly deliver services which are implemented in Madagascar. Prior Professional Association of MFIs not defined by the law. to this law, providers had started (APIMF). • Mobile loans and remunerated processing mobile money transactions • MFIs can be cooperatives, savings are explicitly and offering mobile money products credit-only and deposit-taking following informal instructions from MADAGASCAR institutions. the CSBF. These instructions imposed • MFIs cannot operate in foreign an exclusive-agent model, whereby currencies, issue cheque books, agents could only operate in the name perform foreign money transfers, of one financial provider. Financial and exchange. providers then started to operate mobile money through Banking Operations Intermediary and exclusive agents. The new law does not impose any restriction as far the activity of agents is concerned. 79 percent of the population lives in rural areas where financial access is limited. Coupled with a limited awareness of mobile services in general, this poses a challenge to the development of DFS. 42 RELEVANT BANKING DFS REGULATION CHALLENGES REGULATION • The banking sector, including MFIs, • Slow uptake of DFS: CBN’s 2009 • Over 60 percent of Nigeria’s is regulated and supervised by the “Guidelines on Mobile Money population is residing in rural and Central Bank of Nigeria (CBN) Services in Nigeria”: long excluded remote areas, most affected by • In 2012, CBN launched National MNOs from providing mobile financial exclusion. Financial Inclusion Strategy, which money set targets to reduce financial • Now there are 21 licensed mobile Challenges identified by the exclusion to 20 percent by 2020. money operators. Sustainable and Inclusive Digital • Biometric Bank Verification Financial Service (SIDFS) initiative of Numbers (BVNs) were introduced the Lagos Business School:16 by CBN but responsibility of (i) Consumer Constraints to DFS implementation with the banks – Adoption: economic factors, lack of which requires significant financial financial knowledge and awareness, investments lack of access to digital devices; • Three-tiered KYC regime enables (ii) Insufficient market financial service customers understanding on provider side: who are unable to satisfy all many are misguided by market size, KYC requirements to still access limited knowledge about unbanked financial services, albeit with segments NIGERIA limited transaction thresholds (iii) multiplicity of ID systems and their low enrollment numbers (iv) Low consumer protection and need to build safer DFS ecosystem, including pricing, complaints and dispute resolution, cybercrime, data protection (v) Insufficient interoperability on supply side, e.g. lack of a unified agent framework or regulations guiding agents’ provision of financial services (vi) Lapses in the telecommunications infrastructure, especially the expansion of telecoms infrastructure to rural locations, or locations that don’t yet have a proven business case for commercial investment 16 See “DFS in Nigeria: State of the Market Report (2016)”: https://www.digitalfrontiersinstitute.org/resources/item/dfs-in-nigeria-state-of-the-market-report-2016 43 RELEVANT BANKING DFS REGULATION CHALLENGES REGULATION • Banking sector (including • Rwanda’s regulatory environment Sector supervision and microfinance Institutions) is has been identified as one of microfinance practices regulated and supervised by the the drivers of enhanced digital The role played by the BNR and RCA National Bank of Rwanda (BNR) financial inclusion as it enables (Rwanda cooperative Agency) in the various entities (including bank and supervision of the microfinance sector Three categories of MFIs: nonbank formal providers) to offer has to be clarified to avoid overlaps. • Credit-only MFIs have to comply mobile financial services. microfinance bank performances with simplified prudential norms • Agent banking is permitted and should be included in the national defined by the BNR both mobile operator-led and microfinance sector statistics to • SACCOs with a value of bank-led mobile financial services take into account their important deposits higher than RWF20 are permissible models, subject to and growing role in the industry million ($27,000) and Limited licensing by the NBR. (microfinance sector - excluding Corporations providing saving • Banking agents can accept microfinance banks - accounts for 5.9 and credit services are required deposits, conduct cash-out percent of total financial sector asset to operate under the rules and services, and process a few versus 66.6 percent for bank sector). prudential norms defined by the transactions. Although nonbank BNR. agents originally could conduct Interoperability • SACCOs with a value of deposits only cash-in/cash-out operations, • In 2014, the National Bank of RWANDA below RWF20 million ($27,000): the list of permissible activities has Rwanda issued a policy on Deposit taking SACCOs - Governed expanded over time. interoperability highlighting by laws on saving and credit • E-money agents are permitted to the long-term vision for cooperatives provide cash-in/cash-out services interoperability in Rwanda but it • Umerenge-SACCO (the biggest and to conduct account opening. hasn’t been put in place yet. Sacco’s network in Rwanda with • In addition to issuing e-money, • In 2017, Access To Finance Rwanda initially 416 branches) consolidation non-bank e-money issuers may launched a call for proposals process aimed at consolidating engage in an array of services to develop a business plan for U-SACCOs at the District Level (payments, transfers, cash-in/out, interoperability of DFS as mobile • The consolidation is expected to savings, loans, insurance, etc.). money operators are currently commence in 2017 and will be a working under a closed-loop, precursor to the establishment not allowing transactions across of an Apex Cooperative Bank for different mobile operators. the SACCO network to further enhance financial strength of the Other network as well as service delivery • The current microfinance law is not (especially as a channel to the really favorable to MFIs registered national payment system) as “limited by shares” versus cooperatives. 44 RELEVANT BANKING DFS REGULATION CHALLENGES REGULATION • FIs Regulated and supervised by • Agency banking regulations are in • Decentralized Financial Systems Banque Centrale des États de place for banks (but are reported to Law Impedes access to finance for l’Afrique de l’Ouest (BCEAO) be very constringent) SMEs • Financial Intelligence Unit, called • Non-banks (i.e. MNOs) authorized • MFIs’ revenues from non-savings CENTIF, focuses on AML/CFT issues to issue e-money if they obtain and credit-related services cannot • The Observatoire de la Qualité des the license to operate as e-money surpass 5 percent of their total services bancaires, the first one in issuer revenues which is perceived by the UEMOA, has been launched • E-money issuers must partner with MFIs as a limitation to grow DFS a licensed FI to offer loan / savings • Money transfers by MNOs cannot product directly be done outside the WAMU • Bank do not need license to • There are few regulatory measures become e-money issuers to effectively enforce client • MNOs become more independent protection. from banks and flexible in developing their mobile money Agent banking offerings (i.e. second-generation • There is no regulation for Agent DFS, such as credit, savings and Banking for MFIs. insurance) • i.e.: it is unclear what financial SENEGAL • MNOs are regulated by l’Autorité de institutions can do to use the agent régulation des télécommunications networks of over-the-counter (TIC) et des postes du Sénégal providers to collect deposits and (ARTP) reimburse/disburse credit CGAP, 2016: • The annual interest rate caps on credit offered by banks and MFIs of 15 percent and 24 percent, respectively, may hinder the development of digital credit – when compared to the high interest rates charged in Kenya. • Some regulatory aspects remain unclear or incomplete, including know-your-customer requirements, identification, agent banking regulation, and access to the USSD channel. 45 RELEVANT BANKING DFS REGULATION CHALLENGES REGULATION • FIs Regulated and supervised by • DFS has a favorable environment, KYC implementation: Central Bank, Bank of Tanzania, but outlook is uncertain Consumer access to financial services (BOT) • There are clear regulatory is moderately favorable. National • Fragmented financial sector, guidelines for mobile money identity cards have been issued by involving over 50 banking regulation: the National Identification Authority institutions, over 100 MFIs and o Mobile Money is governed by “The (NIDA) since September 2016, but 5,500 SACCOs National Payment Systems Act coverage still appears to be low, with 2015: Electronic Money Regulation.” an estimated 47 percent of adults o Agent Banking is governed by with no ID. For KYC purposes, banks “Guidelines on Agent Banking for are instructed to “verify customer’s Banks and Financial Institutions, identity using reliable, independent 2017”: gives detailed eligibility source documents, data or rules for potential agents as well information.” as ongoing agent due diligence and audit requirements, agents Competitive Environment: are not permitted to accept, The competitive landscape is issue or otherwise deal in cheque advanced and MNO-led. Strong transactions competition between the 3 main • Interoperability: is mandated MNOs has driven DFS innovation in (for P2P transfers) and has been Tanzania. Mobile money usage has live for some time. In 2014, grown to rival Kenya, but in a more Tanzania became the first country competitive environment. Banks have been slow to offer DFS, partly due to TANZANIA to introduce an industry-led interoperability scheme between regulatory constraints in the early mobile money wallets for P2P days, and much of their presence is via domestic remittances. Consumers partnerships with MNOs. can transfer money between the four leading mobile wallets at no extra charge. Interoperability is currently based upon bilateral agreements between MNOs and financial institutions. • Distribution: Tanzania has a larger mobile money agent network than Kenya, and much of it is controlled by three main agent aggregators: Selcom, Maxcom and Cellulant. These organisations also provide their own OTC services at their agent outlets and the OTC market is thriving with an estimated 4.7 million users of remittance and bill payment services. • Tanzania tends to be an early adopter of DFS innovations and a wide range of DFS services are available and have become popular. AUTHORS Christian Rodriguez Christian is a digital financial services specialist with the World Bank Group. He has over 15 years of experience working with financial institutions in Africa, Asia and Latin America in the design and implementation of digital banking solutions. Julia Conrad Julia is an Operations Analyst at IFC’s Financial Institutions Group, based in Dakar, Senegal. She has been working in the financial inclusion and banking space in China and across Sub-Saharan Africa for the last three years. Gisela Davico For 13 years, Gisela has been an advocate for the development of better policies for financial inclusion as strategies to achieve greater social well-being. She was the research lead for IFC’s MFI Longitudinal study. Susie Lonie Susie Lonie is one of the creators of the M-PESA money transfer service in Kenya and Tanzania and an IFC consultant. She consults on digital financial services in emerging markets. In 2010 she was awarded “The Economist Innovation Award for Social and Economic Innovation” for her work. Lesley Denyes Lesley Denyes is the Program Manager of the Partnership for Financial Inclusion and a DFS Specialist with IFC Advisory Services in Africa. She has worked in the sector for the last 15 years; specifically in the areas of business modeling, financial analysis, strategic planning, product development and channel management for DFS across Asia and Africa. CONTRIBUTING AUTHORS Soren Heitmann at IFC also contributed to this report. April 2019 Search “Partnership for Financial Inclusion publications” or visit www.ifc.org/financialinclusionafrica for more information