In a survey CGAP conducted in South Africa, roughly half of those surveyed said they preferred to deal face-to-face with a person rather than with an electronic device, even if the device is quicker. Interestingly, the responses were similar between WIZZIT customers and people who have a mobile phone but do not use it to conduct transactions.

Despite being satisfied with the mobile banking service, users still missed the human touch. Customer research conducted in South Africa pointed at a likely reason for this: having to deal with machine interfaces undermines people’s sense of control over the process. Indeed, a larger proportion of WIZZIT customers than nonmobile-enabled bank customers felt that they had insufficient control over their finances. Similarly, in one anecdote from South Africa, customers using ATMs for the first time checked their balances so frequently that they lost their entire balances to ATM fees.

The same research in South Africa also highlights the need to improve customer awareness of branchless banking and to educate customers about how it works and what it costs. Not understanding the technology is the single most frequent reason given for WIZZIT customers who have stopped using the service. Nonusers thought the cost of the service was on average 14 times more expensive than it really is.

These results demonstrate the importance of marketing and of balancing technology with human interfaces, both to improve awareness and understanding, as well as to improve perceptions of the service. Achieving this through a branchless model will be a challenge.

Most MFI-led branchless banking initiatives have been small pilots or have had only limited success. Even though MFIs have strong local knowledge, product development acumen, and the ability to manage small loans, most lack the stable core banking systems and specialized technical skill to implement branchless banking models or tap into existing platforms.

In the Philippines, an initiative to let customers of rural banks use G-Cash instead of cash to make deposits and repayments has been constrained in part by the poor quality of banks’ core banking systems. Based on interviews with experts in the field and observations from our own visits, CGAP estimates that the vast majority of the approximately 750 rural banks will need an IT overhaul or major upgrade to participate. In Kenya, an MFI that substituted group loan cash repayments with repayments in M-Pesa found a different problem. Group loan borrowers made fewer on-time repayments under the new system. Customers no longer attended the group meetings that had helped to keep up repayment pressure.

On the other hand, those relatively few MFIs that have the financial resources and skills to deploy branchless banking have been among the first movers. Microfinance banks, including Tameer Bank in Pakistan and Xac Bank in Mongolia, are developing their own mobile banking channels and are partnering with mobile operators to reduce delivery costs and to reach unserved urban and rural areas.

Another way MFIs may get involved is  as partners for banks seeking to expand their market among the unbanked. SKS Microfinance in India has developed a mobile banking initiative in partnership with Andhra Bank, in which customers use designated SKS banking agents to deposit money into Andhra Bank accounts and use a mobile phone to repay SKS microloans. Small MFIs and local community-based organizations can also play on the other side—as correspondents for other, larger banks. This ensures them a steady revenue stream in a synergistic relationship with the larger bank, as long as they target different population segments. An interesting case is the intent of the Andhra Pradesh State government in India to use up to 30,000 village organizations (local federations of self-help groups [SHGs]), to act as a cash agent for payment of social services, for SHG members under their umbrella, as well as for local banks.

Finally, MFIs are also tackling branchless banking as a group to overcome their individual limitations. In Ecuador, for example, the Red Financiera Rural association of MFIs and cooperatives is planning to contract a technology provider to build and maintain core banking systems and branchless banking channels on behalf of the group to minimize up front costs and the expertise needed inside each member organization. This sharing of technology costs and expertise has perhaps the highest potential to bring MFIs onto payment networks and allow them to take advantage of mobile banking and other delivery channels they cannot implement alone.

Most financial service providers see partnerships with businesses that have a substantial local retail presence as a key competitive strategy. They act to build their networks as quickly as they can to expand the pool of potential customers and attain local brand presence. The pace of agent sign-up is most dramatic in Brazil, where 95,000 agents have opened for business, leaving no municipality without a retail bank outlet. This agent network has directly led to the opening of more than 13 million bank accounts in the past five years.

Depending on regulations, agents can be used to open new accounts (signing up customers and conducting customer due diligence) or to conduct customers’ cash transactions (to deposit into or withdraw from an account, or to make or receive payments). Given the finding that most branchless banking customers do not build sizable deposit balances (per observation 3, above), most customer transactions do in fact entail a cash transaction. Many banks that want to enter into branchless banking have partnered with businesses that have many local outlets so that they can jump-start their agent networks, including mobile operators, post offices, and major retail chains:

• Mobile operators. Mobile operators run some of the largest national retail distribution networks to support prepaid card sales. This puts them in a strong position to lead or participate in mobile banking projects. For instance, five banks have partnered with SMART Communications in the Philippines, and Standard Bank in South Africa partnered with mobile operator MTN in South Africa.

• Post offices. Brazil’s Banco Bradesco purchased the rights to use the national post office network as a banking agent network. Bradesco created the Banco Postal subsidiary to trade on the trust that Brazil’s population has in the postal service and to differentiate from Bradesco’s branding as one of the leading private banks in the country. By May 2007, Banco Postal had an agent network of about 5,600 agents, two-thirds of which were post offices. The rest were retail outlets branded as “Bradesco Expresso” points.

• Major retail chains. Equity Bank in Kenya signed a deal in mid-2007 to use the Nakumatt chain of retail stores as its anchor banking agents, and WIZZIT has arranged to use the Dunn’s chain of about 400 clothing stores across small town South Africa to act as account opening locations. Where banks are unable to partner with large retail chains, or in rural areas where these chains have limited or no presence, banks often outsource the building and management of chains of agents to third-party agent management companies. Banco Popular in Brazil (the banking correspondent brand of Banco do Brasil) uses companies such as Net Cash in Sao Paulo State and the Brasilia Federal District and Pag Facil in Pernambuco to sign up, equip, train, and maintain agents on its behalf. Lemon Bank has no branches at all and relies on 16 agent management companies (including three that it purchased) to manage the majority of its 5,750 agents.

A bank’s ability to sign up agents in disparate locations depends on the national payments system rules and practices. Referring back to the Brazilian success case, a second legal provision spurred geographic coverage to such a stunning extent: an agent is legally able to deposit its excess cash in to its account with its sponsoring bank through the branch of any bank, at no extra cost, and without having to open an account at that bank. The situation is quite different in Colombia, for instance, where the bank with the largest network of rural branches, state-owned Banco Agrario, charges such high cash handling fees to other banks that those banks cannot profitably set up agents in remote municipalities. While Banco Agrario’s high cash handling fees may be justified by the high cost of operating in such remote locations, the result is that other banks are not able to use agents unless they set up their own branches nearby.

Based on our observations, it appears that being an early mover in creating an agent network confers three key competitive advantages:

• Early movers are able to partner exclusively with the businesses that have the largest number of local retail outlets, thereby patching together a sizable agent network relatively quickly. Subsequent entrants are likely to find it more difficult to assemble an agent network of their own, particularly in areas with few retail establishments. The number of agents or physical locations is an easy concept to differentiate advertising, and hence it becomes a self-sustaining advantage for early movers.

• Early movers with larger agent network scan negotiate more favorable agreements with utility companies and various government agencies to distribute or collect payments on their behalf. As noted earlier, most banks realize that payments (from customers to utility companies and lenders, and from governments to welfare and pension beneficiaries) is the first product likely to move through this channel.

• A bank that is first to introduce banking services in a given geography is likely to capture greatest market share among the local population. The general manager of Banco Popular in Brazil explained that putting Banco Popular agents in unserved neighborhoods gave the bank a presence and the start of a relationship with local customers. As these communities develop and become increasingly banked, Banco Popular would be the bank whose name they would remember the best.

Customers primarily make payments and send transfers through branchless banking channels, even when most branchless banking channels offer a broader range of services, including account opening, cash deposits, and cash withdrawals. Most customers either time their deposits to coincide with bill payments or cash withdrawals, leaving a near-zero balance in their accounts, or they do not open a savings account at all. Consider the following experiences:

• In Brazil, bill payments and the payments of government benefits to individuals comprised 78 percent of the 1.53 billion transactions conducted at the country’s more than 95,000 agents in 2006. CGAP research in Brazil found that, of the 750 people who responded to a survey in Pernambuco State, 90 percent reported using banking agents to pay utility and other bills, only 5 percent reported opening a bank account at the agent, and less than 5 percent said they had made a cash deposit in to their bank account at an agent.7 Indeed, 87 percent of those who had opened an account stated that they had done so just to receive welfare or salary payments.

• In Russia, more than 100,000 automated payment terminals have sprung up in the larger cities in recent years. One provider, CyberPlat, claims to have processed 1.2 billion transactions worth US$4.7 billion through the first three quarters of 2007 via its 70,000 “cash acceptance” points, mostly for prepaid air time, television, Internet, and other utilities.

• The average mobile banking customer of WIZZIT (a mobile phone banking provider in South Africa) bought air time with WIZZIT twice as often (2.6 times) as they withdrew funds from a branch or ATM (1.3 times), and five times as often as they made a money transfer (0.5 times).

Customers use payments and transfers rather than banking services in part because providers focus their marketing efforts on payments and transfers. M-Pesa advertises its service as “an affordable, fast, convenient, and safe way to transfer money by SMS any where in Kenya,” and WIZZIT’s slogan is “the easy way to pay.” Mobile operators, in particular, prefer marketing payments services rather than the ability to store value because payments services are a closer fit with their traditional revenue model (e.g., per minute or per SMS). Some mobile operators argue that if they did advertise the ability of their mobile banking services to take deposits, they would run afoul of the approvals they’ve received from banking regulators.

The predominance of payments services over savings also likely reflects the perceived relative value that each service brings to the economic lives of the poor. Using banking agents and electronic payments to pay utility bills takes less time than traveling to and queuing in a range of utility offices, thereby bringing very tangible benefits. Similarly, collecting a pension, remittance receipt, and welfare or salary payment is a strong driver for opening accounts.

On the other hand, the value proposition of saving money, particularly in electronic form, appears to be less strong. The former head of Banco Postal in Brazil reported that, in rural areas in particular, his team spent considerable effort trying to explain to customers why they should have a bank account at all.10 It seems that although branchless banking has brought formal banking services physically closer to many unbanked people, it hasn’t changed their perceptions of the value proposition of saving in formal financial institutions. When they receive a payment or a remittance, an overwhelming majority of people go to the agent to withdraw the full amount received.

We believe that, over time, as customers increase their use of branchless channels to make a broader range of payments, they will start to find more value in maintaining transactional or savings balances in their account. In the meantime, more research must be done to distinguish how customers feel about savings in general, about the benefits of saving in banks, and about the branch and branchless channels available to them.

The success of agents in Brazil—achieving 100 percent coverage of municipalities—hinged in no small degree on the fact that utility bill paying is considered a banking service and cannot be done at nonbank outlets. This created a natural captive market of transactions for new correspondents opening up in towns without prior bank presence, where previously residents had no choice but to travel to nearby towns to pay their utility bills. In other countries, such as Colombia, local stores may have collection contracts with utilities, and it has proven much harder for correspondents to seize the utility payments business upon entering the market.

Branchless banking can dramatically reduce the cost of delivering financial services to poor people

We believe branchless banking can offer basic banking services to customers at a cost of at least 50 percent less than what it would cost to serve them through traditional channels. Branchless banking helps address the two biggest problems of access to finance: the cost of roll-out (physical presence) and the cost of handling low-value transactions. This is achieved by leveraging networks of existing third-party agents for cash transactions and account opening and by conducting all transactions online. This sharp cost reduction creates the opportunity to significantly increase the share of the population with access to formal finance and, in particular, in rural areas where many poor people live.

The biggest cost saving is on transactions that can be done completely electronically, through mobile banking. In the Philippines, a typical transaction through a bank branch costs the bank US$2.50; this would cost only US$0.50 if it were automated by using a mobile phone (Asian Banker 2007).

The cost reduction from using agents rather than banks for remote cash transactions is equally dramatic. Banco de Credito in Peru estimates that a cash transaction at a branch costs about US$0.85, while the same transaction at an agent would cost US$0.32.4 Tameer Bank in Pakistan estimates that, in the Orangi slum of Karachi, the set up cost of a bank branch would be 30 times more than the set up cost per agent, which is about US$1,400. Monthly running costs average about US$28,000 for a branch, compared with US$300 for an agent, but also, a much larger share of monthly running costs is variable for an agent than for a branch.

We are looking into the different pieces of the value chain for delivering information and core banking systems through an application service provider (ASP) OR software as a service (SaaS) model. These functions may be performed by a microfinance institution (MFI), a national or regional microfinance association (MFI-A), a local IT service provider (ITSP), the ASP or SaaS vendor (Vendor), or another, new party.

ASP or SaaS models would seem particularly likely to fall short of customer expectations when it comes to support functions. One reason that MFIs are so dissatisfied with existing microfinance software vendors is that they provide poor quality support after the sale – and in particular that most of these vendors do not have local support providers in the countries in which their MFI customers operate. For example, a vendor from Ecuador may have customers in Peru but no on-the-ground support staff in that country.

When we talk of ASP or SaaS models, there is even less personal interaction between supplier and MFI consumer. Software and servicing post-implementation is touted as being completely remote. To ensure that ASP and SaaS vendors aren’t painted with the same brush as traditional MFI software suppliers, these vendors must pay particular attention to the customer experience post-sale. Support functions, as well as other critical “soft” pieces of the value-chain are asterisked for emphasis. These pieces seem relatively more likely to influence whether an MFI will sign up for the service and how satisfied with the service it will be post sign-up.

In most cases the functions above can be matched easily with one or another actor’s competencies or potential capabilities. But for several functions, none of the actors on the scene seem best positioned to assume responsibility. In general, these functions require a combination of technical and microfinance abilities that typically neither a vendor nor microfinance association possesses. Most functions in this category relate to offering the service to the MFI community – they include the design of packages, the “onboarding” of an MFI onto the service platform, and perhaps most importantly, the handling of responses to MFIs who request bug fixes or adaptations to the service.

It is early morning in Bukura, a small village in Western Kenya. The shop-keeper and his wife are preparing to open their small store, which sells household commodities such as flour and cooking oil. They also offer M-PESA services. There is already a queue outside. A group of about twenty villagers are crowding the entrance. “It is always like this,” the shop-keeper complains while pointing to the crowd. “Since we have become M-PESA agents we have no time to rest. This thing has even over-run our other business”. He then holds up a packet of sugar. “We have not sold any sugar in months. They only want M-PESA”. Not just the Bukura agent has seen a great demand for M-PESA services. Since its introduction in March of 2007, the M-PESA application has had great success all over Kenya. There are currently over 2.3 million registered users. Over 18 Billion Ksh had been moved through the system, via person-to-person transfers.

Some of the work that I have been doing makes several arguments as to why M-PESA has become so popular. Firstly, it is the young, male, urban migrants who are driving the uptake of services – customer adoption. These migrants are what innovation researchers call ‘early adopters’ of a technology. They are usually better educated and earn higher incomes than those in the village. Because these migrants are the senders, they can choose the channel for money transfer. They then influence recipients in the rural area—who are usually female, less educated and poorer—to also use M-PESA. This segment is referred to as the ‘technology laggards’. They are usually the last, and often the least likely, to adopt an innovation.

This research also notes some barriers to adoption. Both agents and customers complain of cash float problems, especially in the rural areas. Because the majority of transactions in the village are withdrawals, agents must maintain their cash float. They do this by making frequent trips to the bank. This can be problematic if the agent is not close to an urban centre, where most banks in Kenya are located. An agent in Malaha, a small village in Western Kenya, commented, “almost every day I ride my bicycle to Kakamega to top-up my float. This takes me almost three hours. I have to leave at 6am because I want to be there when the bank opens. I must then come back again and serve my customers”. When asked if there was any other means of transport to Kakamega, the agent shook his head. He said that he was several kilometres away from the main road. He also said that he could not afford to pay the 200 ksh fee for the matatu (shared taxi).

Despite these cash float problems, the majority of customers in both the urban and rural areas assert that they prefer M-PESA over other money transfer services. This means that M-PESA must be offering them some kind of substantial benefit. In Bukura, this benefit comes in the form of savings on transport. Customers do not need to travel into Kakamega, the nearest town, to access the service. One elderly farmer commented that “I can just walk from my shamba (farm) and get money. I don’t have to spend and go into town. If the agent does not have cash today, then I will come back tomorrow. It is cheaper to wait”. Finding strategies to manage the cash float problem will undoubtedly be one of the greatest challenges for Safaricom. For now, however, it seems like customers are willing to accept the inefficiencies of the service. It is, after all, cheaper to wait.

But if the POS is so good, why not go mobile? Especially in Kenya, home to M-PESA, an early pioneer in mobile payments and now topping 2 million registered users and 2500 agents across the country.

Jamii Bora seems ready. They have decent core banking software (Banker’s Realm), which should be able to handle a real time connection to Safaricom’s M-PESA platform for transaction processing and accounting. CGAP has said group lending may not always work with branchless banking. This seemed to be the case with Faulu, another MFI that worked with M-PESA when it first launched. But Jamii Bora has already gone to a system of groups nominating one member to carry in transactions. So there doesn’t seem to be a case that M-PESA will undercut the desire to attend group meetings, and through it the joint liability mechanism which deters delinquency.

In the end, M-PESA’s charges are too high to be economical for microfinance clients. So far, Safaricom has geared M-PESA’s fees to the remittance business. The KSh 30 (USD 0.48) it charges for a remittance up to KSh 2500 (USD 40.35) is quite reasonable compared to the post office’s PostaPay product, or even bus drivers who carry remittances. But microloan repayments are a different business, more similar to bill payments (set schedule, relatively small value of $5-10) than a money transfer (less frequent, larger amounts of $50 or more).

The average JB client makes a KSh 394 (USD 6.36) payment each week. If they used M-PESA to send in loan payments, it would cost KSh 600 (USD 9.69) over the life of an average 20 week loan. That’s equal to 69% of the interest paid on that loan! Another way to express the added cost is an increase to the interest rate paid: using M-PESA would be like raising the interest rate from 12.5% to 21% on the average Jamii Bora microbusiness loan. That’s costly.

But Safaricom might easily see a business case for slashing the fees it charges for bill payments, in order to bring in clients like microcredit borrowers, who will use M-PESA on a frequent basis. From the outside, it seems the average M-PESA client does 1-2 transactions per month. A Jamii Bora client would make at least 4 loan repayments per month. And once familiar with M-PESA, they may very well use it for other purposes.

Figuring out how to make M-PESA economical for MFI clients might deliver the kind of intensive user of whom Safaricom is in hot pursuit.