mBanking Panel 2 – Building a Viable Agent Network

The second panel of the CGAP roundtable on mobile banking for the bottom billion included three panelists: Nick Hughes with Vodafone (the architect of Mpesa), Carl Johan with the Maldives Monetary Authority (MMA) and Sam Kamiti of Equity Bank, Kenya.


The key interface between the electronic world and cash-based world is the agent network which is responsible for handling the cash. Branchless banking (aka mobile banking) requires the outsourcing of cash transfers to these distributed networks of agents such as small shops. One important question is how to make the compensation model viable for these agents?

The moderator of the panel, Mark Pickens of CGAP showed the results of a small study carried out on Mpesa agents. While I believe this is exactly the kind of study necessary to better understand the cost benefit analysis (CBA) of agent participation, the study in question only drew on a sample of 20 Mpesa agents (!).

The study suggests that the number of transactions follows a Gaussian (or normal) distribution with a mean of 105 transactions per day providing an average of $10.7 daily commission. At -1 standard deviation, there are 58 transactions with $6.6 in daily commission, which is 2-3 times the daily wage. At +1 standard deviation, there are 152 transactions with $14.7 in daily commission.

Mark emphasized the point that an agent can only maintain so much money in float; about $250 for one Mpesa agent he spoke with. The agent therefore needs to visit his closest bank several times a day, which presents additional costs of doing mBanking. Partly as a result, the agent receives more profit by selling non-mobile products. In other words, one of the biggest challenges that an agent faces is getting liquidity, which is why the agent in question does not see Mpesa as the main driver of his sales and profit.

To provide a comparative analysis of Mpesa, Equity and mBanking in the Maldives, Mark Pickens compared the following variables for each initiative respectively: Agents, Transactions/Day, Value/Day, Number of Clients, Agent Networks.


The first panelist, Nick Hughes from Vodafone, played an instrumental role in the design of Mpesa. It was interesting to note that Mpesa never positioned itself as mobile banking:

All we did is ask, do you need to send money home? To move money around? If you try to introduce customers to mobile banking, don’t talk about banking, talk about need and address need, ie, the functions of mobile banking. We got 4.5 million subscribers not because we asked asked, ‘do you want to do mobile banking?’

Agents are critical because they serve as for cash-in, cash-out points. It is important to make it as convenient as possible for customers to take cash in/out. So Mpesa’s 4,000 agents basically act as human ATMs. In setting up mBanking, one should first concentrate on getting that agent network set up. The business case should not just be transfer of stock; one needs to think more broadly. For example, by offering Mpesa in a shop, the owner consequently gets more customers coming to the shop, potentially purchasing more items as a result.

The second panelist, Carl John from the Maldives Central Bank, noted that the average island dweller has no access to banking. There are no bank branches on islands with less 500 people and  only one bank that goes around once a month. There are only 41 ATMs in all of the Maldives and only 38 banking branches agents. Mobile banking thus provides some important opportunities for the country.

Carl pointed out that the endgame for the mBanking initiative in the Maldives is a cashless society. This means identifying new agents, such as small stores or basically anyone accepted and trusted by the local population. The agent network in the Maldives will also need to  handle checks and the Central Bank will treat ATMs as part of the agent network.

There are 3 types of agents according to Carl:

  1. Mobile/handset only agent (boat operator, fisherman), can download statement from website;
  2. Working level agent (printing statements services, provide cash in/out);
  3. Check-enabled agent (using scanners to submit images to central bank for clearing);

In closing, Carl emphasized that the Maldives’ mBanking system is not e-money, but a banking system.

Sam Kamiti of Equity Bank, Kenya, emphasized the need to focus on the Bottom of the Pyramid and to demystify banking. Equity’s approach is to only charge clients for transactions. They invested heavily in ATMs which is why Equity’s is the largest ATM network in Kenya. Equity also introduced points of service (POS) to provide services beyond payments of goods and services. The POS also charge a smaller transaction fee for cash back than ATMs.

During the Q&A session, one question addressed how best to extend agent networks further, and how to make it worthwhile for all agents along the value chain?

Nick of Vodafone replied as follows:

We need to give aggregators the ability to move funds around without having to go to a bank; the more you can avoid having to go a bank the better. We need to follow the money; if you can remove cash entirely from the marketplace, then you don’t need agents. For example, Vodafone provided a school in Kenya with a Mpesa account so parents could pay directly when they noticed that parents would take out large sums from their mBanking accounts several times in a row just to pay school feels. The take-home point? Follow the money.

Carl emphasized the need to for banks to be responsible for nominating agents and for providing sufficient revolving credits to these agents. Attracting more ATM deployers is also key to further extend agent networks.

Sam pointed out that petrol stations and marketplaces prefer to use less cash; handling large amounts of cash presents problems for these existing networks. To this end, creating synergies with pre-exisitng networks for the purposes of mBanking can provide mutual benefits.

Another participant during the Q&A session asked whether Vodafone had any plans to extend Mpesa to countries where Vodafone is not presently operating?

Nick replied yes since any phone operator can take Mpesa without needing the Vodafone footprint or infrastructure, such as Roshan in Afghanistan. Vodafone provides the platform, Roshan recruits the customers. In the past, mobile operators would go at mBanking alone. The new trend sees mobile operators forming partnerships with banks and other groups. One of the consequences, or necessary conditions, of such partnerships is that systems must be made fully interoperable.

Nick also pointed to worrying developments on the regulatory side of the equation. He gave an example in India that will almost certainly hamper possible partnerships between mobile operators, banks, etc: “A new regulation in India stipulates that no agents can be further than 5km from a bank branch.” In another example, the India Reserve Bank now requires inter-operatbility within 6 months of operation. “This completely stifles innovation and discourages start-ups with new ideas from taking any risks. The key to the future is interoperability, it’s how we survive, but regulation can seriously set us back.”

The problem is that it is particularly difficult if not impossible to scale up mBanking without partnerships between telecom companies and banks. The enabler (such as Vodafone) and the core finance provider (banks), need to find a way to share the profits/costs. Any regulation about what kinds of agents can be used forces a change in strategy from mobile operators.

In response to a question on improving agent training and financial literacy for end users, Nick emphasized the critical need to employ third parties to train agents. “It is vital that agents do their job well in order to establish trust with customers. We also need to move very quickly if we see agent behavior that is not sanctioned; this is absolutely essential.” In conclusion, Nick pointed out that the variation in in-country economic growth and mBanking is less a matter of technology and education and training.

Another Q&A question addressed the issue of tight regulatory control slowing down innovation and contrasted this with the current response to the global financial meltdown which calls for increasing regulations. Where does one draw the line? How does regulation effect agents, their business, Know Your Customer (KYC) procedures, customer protection issues, business models/cases?

Nick took the first shot at the question. It is important a contracts are in place with agents. Mpesa does not charge a fee for registration, but does require ID verification in order to check for fraud, terrorism finances, money laundering, etc. Agents need to collect the KYC data very carefully, “this is something that the Central Bank of Kenya is doing very well.” Agents should be awarded commissions when they bring on a new client, but the initial KYC must be carried out by these agents, with the full KYC done by Vodafone.

According to John, KYC procedures are rather limited in the Maldives because of the hundreds of islands. Agents try and check person ID cards. A client that supplies the most basic KYC data is allowed to get the first basic level of mBanking service. They can upgrade to additional services if they provide additional KYC data. Further upgrades require that they be issued a card.

Other services provided (with additional KYC verification) include Islamic banking. On this note, the Maldives Central Bank allows banks to define their own products/services. It is then up to individual banks to establish the degree of differentiation they want to spur competition.

Sam of Equity Bank concluded the panel discussion by noting that the cost of compliance is generally rather high, which means that expanding to marginal agents is rather difficult. Agents need to be provided with a business model that clearly identifies high returns.

Patrick Philippe Meier

One response to “mBanking Panel 2 – Building a Viable Agent Network

  1. Pingback: Mobile Banking for the Bottom Billion: An iRevolution? « iRevolution

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s