Scaling Mobile Money: Ignacio Mas and Dan Radcliffe, Bill & Melinda Gates Foundation FINAL: 31 May 2011
Scaling Mobile Money: Ignacio Mas and Dan Radcliffe, Bill & Melinda Gates Foundation FINAL: 31 May 2011
Ignacio Mas and Dan Radcliffe, Bill & Melinda Gates Foundation
FINAL: 31 May 2011
Abstract
Retail payment systems require scale to get off the ground and struggle to grow incrementally. This is
due to three factors: (i) Network effects: When it comes to payment systems, the value of joining a
network is directly proportional to the number of people already on it. (ii) Chicken-and-egg trap: In
order to grow, these systems must aggressively attract both customers and cash-in/cash-out merchants
in tandem. Otherwise, merchants will stop offering the service due to low transaction revenue and
customers won’t join the system because they can’t access a convenient outlet. (iii) Trust: Customers
have to become comfortable going to non-bank retail outlets to meet their cash-in/out needs and
initiating transactions through their mobile phones. Until a deployment serves a large number of
customers, people will lack trust in the new system because they know few who can vouch for it.
To overcome these barriers, mobile money deployments need to reach a critical mass of customers as
quickly as possible, lest they get stuck in the ‘sub-scale trap.’ To do this, they need to get three things
right. First, they must create enough urgency in customers’ minds to learn about, try, and use the
service. Second, they must invest heavily in above- and below-the-line marketing to establish top-of-
mind awareness of (and trust in) the service among a large segment of the population. And, third, they
must incur considerable customer acquisition costs (beyond marketing and promotion) to ensure that
their cash-in/out merchants are adequately incentivized to promote the service. Many deployments
around the world have potential to scale, but are stuck in the ‘sub-scale trap’ because their promoters
either under-estimate the investments needed to achieve scale or are reluctant to make these
investments because they can point to only one major success – M-PESA in Kenya.
Key words: Mobile money, access to finance, mobile banking, branchless banking, financial inclusion
An estimated 2.5 billion adults lack access to basic formal financial services.1 This situation arises
because banks do not find it economically attractive to deploy banking infrastructure and staff bank
employees in poor and rural areas. Branch penetration, for example, averages only two branches per
100,000 people in the poorest country quintile (compared with 33 in the richest). ATMs are even
scarcer, averaging only 1.3 per 100,000 people in the poorest quintile (compared with 67 in the richest).2
With no access to formal banking infrastructure, poor households, who have little money to begin with,
are left only with informal financial tools which are often risky, inconvenient, and expensive. While there
1
Electronic copy available at: http://ssrn.com/abstract=1681245
is considerable work to be done to develop products that better meet poor people’s financial needs,
financial exclusion is fundamentally a distribution problem.
In recent years, banks, mobile operators, and payment system providers have begun experimenting with
branchless banking models which promise to reduce the cost of delivering financial services by taking
small-value transactions out of banking halls and into local retail shops, where cash-in/cash-out
merchants, such as airtime vendors, petrol station attendants, and shopkeepers register new accounts
and convert customers’ cash into electronic value (and vice versa). One form of branchless banking,
called mobile money, seeks to fundamentally change the economics of retail banking in four ways:
First, it piggybacks off widely deployed infrastructure – the retail shops that exist in every village and
every neighborhood and the telecommunications networks that are rapidly being built in developing
countries3 – to extend financial services to large segments of unbanked poor people. Clients can convert
cash into electronic money (and vice versa) through these retail outlets and then use their mobile phone
to instantaneously send and receive money wherever they have cell coverage. The transaction
information is then communicated back to the telecommunication provider or bank. The customer
needs to visit a retail outlet only for transactions that involve depositing or withdrawing cash.
By leveraging this real-time communications system, neither the scheme operator nor the customer
needs to worry about stores running off with their cash because stores must pre-purchase electronic
value from the issuing bank or mobile operator. Hence, any cash exchanged between the customer and
the retail store is offset by an immediate, opposite transfer of value between the customer’s and store’s
accounts (not the issuer’s). The store and the customer thus engage in an instantaneous spot
transaction and neither the customer nor the bank faces any credit exposure to the store (i.e. the
transaction is settled by the time the customer walks out of the door).4 And this system can be
propagated cheaply because both customers and stores increasingly have mobile phones of their own.
There is no need to distribute bank cards among poor customers and point of sale terminals among
small shops since the mobile phone is functionally equivalent to these two pieces of equipment.5
Second, mobile money turns fixed costs (banking infrastructure, staff salaries) into variable costs
(paying merchant commissions based on transaction volume). This dramatically reduces the revenue
threshold needed to establish a viable transactional outlet – revenues no longer have to cover the cost
of building and maintaining a bank branch and paying staff salaries; they only need to be enough to
encourage the shopkeeper to promote the service alongside his other products.
Third, mobile money employs usage- rather than float-based revenue models for reaching poor
customers with financial services. Because banks make most of their money by collecting and
reinvesting deposits, they tend to distinguish between profitable and unprofitable customers based on
the likely size of their account balances and their ability to take on credit. In contrast, mobile operators
2
Electronic copy available at: http://ssrn.com/abstract=1681245
in developing countries have developed a usage-based revenue model, selling prepaid airtime to poor
customers in increments as low as 10 US cents, such that each transaction is profitable on a stand-alone
basis. This is the magic behind the rapid penetration of prepaid airtime into low-income markets – a
card bought is profit booked, regardless of who bought the prepaid card. A usage-based (i.e.
transactional) revenue model for financial services for the poor would make possible a true mass-market
approach, where all customers pay for themselves and banks no longer distinguish between profitable
and unprofitable customer segments.
Float (the cumulative amount stored in e-money accounts, which is in turn deposited in bank accounts)
remains an important part of revenue for mobile money schemes. But the important point is that the
profitability of the majority of customers is more likely to depend on the number of transactions they do
than on their saved balances. Interest on float would usually go to the mobile money provider, though
this is sometimes precluded by regulation as is the case with M-PESA in Kenya.
Fourth, mobile money harnesses customers’ tremendous need and willingness to pay to make remote
payments conveniently and securely. For most poor households, especially those in split families where
the head of household works in one part of the country and sends money to his family in another, it is
difficult to move cash over distances. However, once a customer is connected to an e-payment system,
(s)he can use this capability to instantaneously send and receive money from friends and family, store
value, pay bills and monthly insurance premiums, receive pension or social welfare payments, receive
loan disbursements, and make repayments. In short, when a customer is connected to an e-payment
system, her range of financial possibilities expands dramatically.
A Note on Terminology
We refer to e-money as monetary value that is recorded in electronic media, exchangeable for physical
cash at par value, and backed by liquid bank assets. The bank backing can be direct if the e-money is
issued by a bank, or indirect if it is issued by a non-bank entity as long as it holds an equivalent amount
of value in liquid banking assets.
E-payments are transfers of monetary value that occur entirely by electronic means, involving the
crediting and debiting of electronic accounts, whether these are bank deposits or e-money.
Cash merchants are non-bank retail outlets that exchange e-money for physical cash and vice versa
instantaneously, on-demand, for a fixed commission. This may or may not be the exclusive business of
these stores, and they may or may not have a direct contractual link with the e-money issuer.
Mobile money is a loose term for an e-payment system that is based on e-money issued by a non-bank
player (typically a mobile operator), and which is combined with a dense network of cash merchants
numbering typically in the thousands.
3
Developing a Mobile Money Ecosystem
Mobile money is not a one man show. It requires stitching together a complex web of actors who
collectively generate the large volumes of transactions needed to fuel the system. Figure 1 below depicts
the four core players that need to be connected in a successful mobile money ecosystem.
First, there are the end-users, who do 1:1 (or Figure 1: The mobile money ecosystem
person-to-person) transactions between each
other, whether sending money among friends
and family or for business transactions. Then
there are the bulk users, such as governments,
firms, and utility companies. These players
conduct 1-to-many transactions (salary,
pension, and social welfare disbursements) or
many-to-one transactions (utility payments,
loan repayments, tax collections). And, finally,
there are the merchants. These players come in two forms. Some use mobile money purely as a way to
accept electronic payment (instead of cash) from their customers when they sell products and services.
Others use their mobile money account to offer cash-to-electronic value conversion services (i.e. cash-in
and cash-out) to end-users. These are called “cash merchants.”
Each player plays an essential role in the ecosystem: end-users create network effects since the system
becomes more valuable to customers when they can send and receive money to (or from) more people;
bulk payers create regular transaction volume and pull more users into the system; merchants who
allow end-users to buy goods and services electronically reduce the amount of costly cash-out that
customers need to do in order to meet daily expenditures; and cash merchants allow customers to
convert physical cash into electronic value and vice versa.
All these players must be brought onto the system in large numbers and largely in parallel. This is a big
undertaking, and requires convincing a range of actors – people, shops, corporations, and governments
— that enough players will get on board (in technical terms we call this a multi-sided platform).
M-PESA in Kenya is by far the most successful mobile money deployment. Since its commercial launch in
March 2007, it has been adopted by 14 million customers (corresponding to more than 80% of
Safaricom’s subscriber base) and processes more transactions domestically than Western Union does
globally.6 The service now reaches 70% of Kenyan households and 50% of all unbanked households.7 US
$415 million per month is transacted in person-to-person transfers, equal to 17% of Kenya’s 2009 GDP
on an annualized basis.8 There are more than five times the number of M-PESA outlets (28,000) than the
4
total number of postal outlets, PostBank branches, commercial bank branches, and ATMs in the
country.9
The experience outside Kenya, however, has not been as spectacular. As shown in the table below, since
the launch of SMART Money in the Philippines in 2003, at least 79 mobile money deployments have
been launched across 43 developing countries.10
While some deployments have gained early traction in the market, such as MTN MobileMoney in
Uganda, Tameer Bank / Telenor’s easypaisa in Pakistan, and Vodacom’s M-PESA in Tanzania, none have
scaled at (or near) the level experienced by M-PESA in Kenya. This has prompted us to take stock of
where we are in the evolution of mobile money and to sharpen our focus accordingly. We assess the
state of play in mobile money as follows:
5
Awareness: The CGAP Technology Program and the GSMA Mobile Money for the Unbanked
(MMU) initiative have been instrumental in generating awareness and interest in branchless
banking through their grant making facilities, market research, convenings, and in-country
technical assistance. One might even argue there is now too much hype, leading to the risk of
unfulfilled expectations. That said, there is still need for considerable investment in research,
learning events, and technical support to build the industry’s knowledge base.
Proof of concept: Many deployments around the world have shown that customers are ready to
take up the service, the technology works, and stores are qualified and can be motivated to offer
cash in/out services. There isn’t need for additional investment in pilot projects to demonstrate
proof that mobile money can work at small-scale.
Proof of scalability: As noted above, M-PESA in Kenya has shown that mobile money can be scaled
up to become a self-propagating ecosystem, with customers, merchants, and institutional players
(banks, billers, employers) eager to join the platform in equal measure.11
Proof of replicability: Kenya provided ‘fertile grounds’ for mobile money due to many factors,
including demand-side conditions (a high percentage of split families which fuelled demand for
domestic transfers; poor alternatives for sending and receiving money), supply-side factors (a
dominant operator with 80% market share; reasonably extensive bank branch infrastructure to
support merchant liquidity management), and regulatory factors (a central bank who took the
objective of financial inclusion to heart).12 The task now is to determine whether mobile money
can scale in countries where some of these factors may be weaker.
Proof of financial service delivery: While the early evidence suggests that access to a safe and
convenient merchant-based e-payment system increases household welfare,13 e-payments is not
the end goal. We instead view e-payment systems as the ‘transactional rails’ on which a broader
range of financial services can “ride.” M-PESA today is still largely a payments system. In May 2010,
Equity Bank and M-PESA announced a joint venture, M-KESHO, which permits M-PESA users to
move money between their M-PESA mobile wallet and a prudentially-backed, interest-bearing
Equity Bank account. While several hundred thousand customers have opened M-KESHO accounts,
only a fraction of these are actively being used and it is unclear how aggressively Equity Bank and
Safaricom are promoting this jointly-branded product. There is thus considerable work to be done
to show that special financial products which take advantage of the unique features of an e-
payments platform can be developed, and that these products can be marketed effectively to poor
people. This can happen either because a scheme promoter shifts its emphasis from payments to a
broader range of services, or because banks connect to an e-payment platform and assume the
roles of product development and marketing (as in the case of M-KESHO).
Speed to Scale
A fundamental characteristic of mobile money systems is that they struggle to achieve scale
incrementally. This is due to three factors: (i) Network effects: The value to a customer of a payment
6
system depends on the number of people actively using it – the more people on the network, the more
useful it becomes. While network effects can help a scheme gain momentum once it reaches a critical
mass of customers, they can make it difficult to attract early adopters when there are few users on it.
(ii) Chicken-and-egg trap (2-sided market): In order to grow, mobile money systems have to attract both
customers and stores in tandem. It is hard to sell the proposition to customers while there are few
stores to serve them, and equally hard to convince stores to sign up while there are few customers to be
had. Thus, schemes need to drive both customer and store acquisition aggressively. (iii) Trust:
Customers have to become comfortable going to non-bank retail outlets to meet their cash-in/out needs
and initiating transactions through their mobile phones (or other point-of-sale device). The best way to
build trust in the system is to reach critical mass quickly so that existing customers become the prime
mechanism for drawing in new customers.
At first, all these elements work against a deployment. The benefit to a customer of joining the system
is minimal when few others are connected (network effects) and the merchant network is not
sufficiently dense and geographically dispersed to meet their (and their recipients’) cash-in/out needs.
Meanwhile, merchants remain reluctant to tie-up scarce working capital investing in electronic float14
because they don’t yet see enough demand from customers (chicken-and-egg trap). And customers lack
trust in the system because they know few people who can vouch for the service. We’ve termed this
period the ‘sub-scale trap’ – when a deployment has launched but remains stuck at a sub-scale level of
customers, with all these elements working against it. Most mobile money deployments around the
world are stuck in this trap.
It’s not all doom and gloom, however. Once a system reaches a critical mass of customers, all these
elements start working in its favor. Customers want to sign up because their friends, family, and
employers are already sending and receiving money through the system. Meanwhile, stores eagerly
apply to become cash merchants because they see its money-making potential; customers eagerly join
the system as outlets open up in their neighborhood; and existing customers start pulling in new
customers by showing their friends and family how to use it. And the first deployment in a market to
reach this critical mass of customers has a distinct first-mover advantage as the challenger may have
difficulty peeling off customers from a well-established payment network. M-PESA is the only mobile
money deployment that has reached this tipping point where the system became self-propagating.15
The upshot is that mobile money deployments need to get from zero to critical mass as quickly as
possible, lest they get stuck in the sub-scale trap. To do this, they need to get three things right: (i) they
must create enough urgency in customers’ minds to learn about, try, and use the service; (ii) they must
invest heavily in marketing to establish top-of-mind awareness of the product in a large segment of the
population; and (iii) they must incur considerable customer acquisition costs (beyond marketing and
promotion) to ensure that their cash merchants are adequately incentivized to promote the service.
These are the topics addressed in the rest of this paper. Appendix 1 includes a list of key business model
elements that help drive scale. While this list is not exhaustive, it may provide a useful starting point for
evaluating the scalability of a mobile money deployment.
7
What is a ‘Critical Mass’ of Customers?
Identifying when a deployment reaches a self-sustaining critical mass of customers is difficult. As a
benchmark, we suggest that a scheme has demonstrated scalability and sustainability when it meets
two criteria. First, it has at least 10 times the number of cash-in/out outlets (i.e. branches) of any bank
in the country. This indicates that the deployment offers a compelling value proposition to customers
based on physical proximity to a transactional outlet. And, second, the scheme is able to generate at
least 50 transactions per outlet per day. This suggests that merchants are motivated to promote the
service and the retail channel will propagate organically, based on a demonstrated effect from
successful merchants (50 transactions at around US 10 cents commission per transaction would
contribute $5 of daily revenue to the store).
From a customer’s perspective, mobile money is a screwy system. It asks poor customers, many of
whom have never interacted with a formal financial institution, to get comfortable with three totally
new elements: going to a local shopkeeper to meet their cash-in/out needs; initiating transactions
through their mobile phone; and trusting that their payments are processed accurately by the bank
and/or mobile operator at the back-end. Faced with this complexity, the customer has every incentive to
hold off on trying the new service until their friends and family have first tested it and deemed it safe.
Delays like this pose a direct threat to the viability of a system that must reach a critical mass of
customers quickly. To overcome this barrier, schemes should tailor their marketing and messaging
campaigns at promoting the “Wow Factor” – the service which solves such a big cash pain point for
customers that they are willing to try this screwy new system today.
In many environments, the promise of instantaneous remote payments offers the needed “wow factor”
to get people to try a new mobile money service: “Last week I paid 5% to a hawala agent to send money
from Delhi to my wife and kids in rural Bihar. It took five days to get there and, last year, an agent ran off
with the entire amount. And this advertisement tells me I can send my money instantaneously, for less
than 5%? That’s something I want to try today!” That’s the wow factor.
The customer’s mental calculation hinges on several underlying factors which make remote payments
(especially domestic remittances) a good foundation on which to build mobile money propositions:
First, as indicated above, domestic payments address a key pain point of people living in a cash
economy. The need for domestic payments is often large, whether it is spurred by migrant labor
remittances, informal support networks of friends and family, entrepreneurs’ commercial transactions,
or bill payments. And it is particularly difficult for poor people to move cash over distances. M-PESA, for
example, bet that the best way to get people to try out the new system was to promote the benefits of
8
being able to instantaneously ‘send money home’. They surmised that, once customers are using the
system to move money around, they will eventually work out the value of using the system to store
balances. Safaricom also bet that customers would be able to quickly compute their willingness to pay
for instantaneous e-payments, based on the typical costs they incur when making payments through
existing channels, in terms of charges (official fees plus informal charges and bribes), access costs (the
cost of traveling to/from the nearest payment outlet plus foregone wages due to time spent traveling
and queuing) and the probability of mishaps (delay or outright loss).16
Savings, on the other hand, does not have that kind of immediacy. Families that have stored cash under
their mattress for generations may not see an immediately compelling need to move some of their
savings into a new service that they don’t quite understand. Some may also feel they have more
workable informal alternatives (money guards, rotating savings groups with friends and neighbors) for
storing money than for sending and receiving money. Hence, they may have a harder time convincing
themselves of the need to move their informal savings into the new system.
Second, because mobile payments are completed in real time, customers can verify the security of the
system instantaneously by calling the other party to confirm receipt. With this possibility of immediate
feedback, customers do not need to understand how the service works – they need only check that the
payment went through. It takes much longer to build trust around savings services because savings is an
inter-temporal service. A customer can check her savings balance day after day, but she never knows
whether it is truly safe. Thus, starting with payments allows customers to build trust by completing
several basic transfers. And, gradually, this trust can be extended to savings and other inter-temporal
products.
Third, starting with payments allows scheme operators to direct scarce marketing dollars towards a
more easily addressable segment – remittance senders. These are more likely to be richer, more
financially aware, more urban, and more exposed to a wide array of mass marketing media such as TV
and billboards. Under this ‘follow the money’ marketing logic, early messaging can be crafted to target
the most frequent senders (domestic migrants, daily wage laborers), who can then be relied upon to
draw their poorer, more rural circle of payees into the service.
Fourth, domestic payment systems can be integrated into (rather than displace) households’ existing
informal savings and insurance networks. We see financial services operating at three levels:
1. Among a social circle of someone’s friends and family (intra-family loans, ‘parking’ money with a
trusted neighbor, migrant remittances)
2. Among informal financial service providers (pawnshops, moneylenders, deposit collectors)
3. Among licensed financial institutions (banks, microfinance institutions, credit unions).17
While there are strong arguments for shifting financial services from the informal to formal (from 2 to
3), we mustn’t neglect the importance of preserving and possibly enabling the social network which
connects households (level 1). This remains the most basic level of financial protection for poor families.
An efficient domestic payment system can improve people’s ability to manage their financial lives by
9
extending the reach of these social networks to friends and family who are more distant. Rural families,
for example, can call on more distant relatives in times of need and labor migrants can better time their
remittances to suit the needs of their families back home.18
Fifth, as clients conduct electronic transactions, they develop a financial transaction history that can
potentially be used to evaluate credit risks and to up-sell other high-value financial services, such as
savings and insurance.
Of course, every market is different and what works in one context may not work in another. It’s
incumbent upon the scheme operator to conduct market research to understand what service solves
such a big pain point that potential customers are willing to try the new system today.
As mentioned earlier, from a customer’s perspective, mobile money is a screwy system. Even after a
scheme promoter has identified the ‘wow factor” that will help drive adoption, it must somehow explain
what mobile money does, why it’s better than the alternatives, and how it works, all while reassuring
them that indeed it can be trusted to work. And all this must be achieved without direct contact with
the customer. And it must be done quickly to ensure that cash merchants receive enough transaction
commissions to continue to promote the service.19
The best way to build trust in (and establish top-of-mind awareness of) this system is through aggressive
above-the-line (TV, radio) and below-the-line (stage plays, banner ads) marketing campaigns. The TV
and radio advertisements signal that the service is legitimate (not a ‘fly-by-night’ operation), while the
road shows and banner advertisements explain the product’s benefits, demonstrate how to use it, and
tell customers precisely where (s)he can avail of the new service. Over time, as people become more
aware of the service, TV and radio can give way to basic store branding at retail outlets, supported with
a few large billboards. And, as customers become familiar with how to use the service, it becomes less
necessary to support hands-on outreach activities. In short, after an initial ‘big push’ investment in
marketing to reach a critical mass of customers, a scheme’s marketing expenses begin to taper off as the
system begins to self-propagate.
A key driver of M-PESA’s success was Safaricom’s early investment in above- and below-the-line
marketing (as much as $10 million) during the first two years of its deployment (helping it reach 6.2
million customers at the two-year mark). To overcome the significant trust and awareness barriers
associated with the new service, it invested aggressively in promoting the M-PESA brand until network
effects began to turn in its favor as new customers begat more customers.20 Over time, Safaricom was
able to scale back its marketing spend as the system became self-propagating, relying predominately on
word-of-mouth advertising and the support of cash merchants.
10
3) Greasing the Channel (Merchant Push)
Establishing top-of-mind awareness of the new service is only half the battle. Retail cash merchants
perform the all-important customer-facing functions that determine a deployment’s success or failure.
Because mobile money is an intangible service in a store that is cluttered with physical goods, merchants
need to actively promote mobile money or customers won’t know it is available. Moreover, new
customers may require some initial handholding until they are comfortable using the service. Retail
merchants must therefore be willing to spend time performing first-line customer care. And merchants
must tie up precious working capital in order to maintain sufficient stocks of cash and e-money for
conducting cash-in/out transactions.
Retail outlets will not perform these costly and time-consuming functions unless they are adequately
compensated for doing so. Hence, schemes must incur considerable customer acquisition costs (beyond
marketing and promotion) to “grease” the merchant channel. Safaricom, for example, pays its channel
US ~$0.95 per customer registered.21 While these commissions have cost Safaricom a hefty $12.6 million
to-date, they provide a powerful customer acquisition incentive for stores, especially during the early
days of a deployment when there are few customers and (hence) relatively little transaction
commissions to be had. Also, by giving merchants some initial cash-flow before they start earning
significant revenues from transactions, scheme operators mitigate some of the initial risk borne by
merchants who must tie-up precious working capital to maintain sufficient stocks of cash and e-value.
Because store revenues are dependent on the number of transactions they facilitate, operators must
also carefully manage the ratio between merchants and customers, lest they depress individual
merchants’ transaction revenues. When operators recruit too many merchants before launch, there
won’t be enough business to go around, causing merchants to defect. This can create a vicious cycle
which can quickly jeopardize a deployment because, when merchants stop holding float, customers
become frustrated because they can’t find a liquid outlet. And, as customers stop conducting
transactions, the remaining outlets defect as their transaction volumes start to drop. To avoid this
unhappy scenario, scheme promoters should carefully maintain a balanced growth in the number of
outlets relative to the number of active customers, resulting in an incentivized and committed merchant
base.22
11
but to do this they enlisted parallel channel intermediaries either directly or by hiring a third-party to
set up this channel. These dedicated intermediaries then identified, trained, and managed the
network of cash merchants (many of which also sell airtime). If mobile money appears in the same
stores as airtime cards, it may not be because the same distribution channel is being exploited;
rather, it may be an ‘artifact’ stemming from the fact that most stores sell airtime anyway.
The main reason mobile operators have failed to engage their airtime wholesalers for mobile money
is that mobile money commissions tend to be much lower than airtime commissions and customer
demand for mobile money is not yet proven. Also, airtime intermediaries may sense that the new
product bears the seeds for cannibalizing their airtime sales because, if it becomes successful, people
will top-up their airtime directly from their mobile account rather than from scratch cards. And,
finally, airtime intermediaries need only manage their down-stream outlets’ e-value balances, while
mobile money intermediaries must manage their down-stream’s e-value and cash balances. Given all
this, the airtime intermediary may opt to stick with what it knows and sells well.
Does this obliterate the mobile operators’ supposed channel advantage in building mobile money
systems? No – it just makes the opportunity less directly graspable than was assumed in the early
stages of the industry’s evolution. While mobile operators may not be able to convert their entire
airtime channel, they still have an advantage relative to most other players in that they thoroughly
understand the channel logistics around the product (most of which carry over from airtime), as well
as the needs and motivations of the various channel players. But they still face the challenges of
identifying a set of ‘hungry’ players who are keen to promote mobile money and constructing the
right set of incentives for them.
Up to this point, we have stressed the importance of marketing, branding, and the need to define an
early use case that will encourage customers’ to learn about, try, and use the service. But these activities
create a set of customer expectations – on convenience and liquidity – which need to be fulfilled every
time the customer goes to a store and wants to transact. It sounds paradoxical, but the success of
mobile money depends largely on how well it can handle physical cash. This is particularly true if it is
aimed at the unbanked, for whom everything happens in cash. They need bridges between the cash
economy and the electronic money world they are being wooed into. Those bridges are the retail stores
acting as cash merchants. How well those bridges work – how reliably these stores are able to meet
customers’ liquidity needs – will in the end determine how customers judge the convenience and
trustworthiness of the mobile money system.
To ensure that customers can access physical cash and electronic money when they need it, schemes
must build a dense, multi-layered distribution channel which can effectively perform the following
activities:
identifying, screening, and training new shops;
12
providing the necessary store signage and maintaining ongoing delivery of promotional materials;
supervising store activities and ensuring consistent application of brand guidelines;
distributing commissions across stores; and (above all)
giving stores convenient mechanisms to frequently rebalance their stocks of cash and e-money.
These are challenging tasks as they have to be done at the same time across a large number of
geographically disperse outlets. The logistical challenge is compounded as the number of outlets grows.
Much has been written on this topic and we will not repeat that analysis here.23 For the purposes of this
paper, suffice to say that scheme operators need to invest heavily in recruiting, training, and monitoring
their retail cash-in/out networks, especially during the initial roll-out when the service is new to both
merchants and customers.
The first similarity is that they both had a very specific and powerfully targeted the proposition. At
PayPal’s inception, the main use case was thought to be P2P payments, with the typical scenario being
friends settling the dinner tab with each other (one person puts the meal on their credit card, the
others just email him their share of the bill). Then PayPal discovered that its service was being used to
pay for goods on online auctions, of which eBay was the market leader. PayPal recognized that eBay
had the capacity to drive viral growth, since a few thousand sellers advertising the availability of
PayPal payments could promote the service with millions of buyers. PayPal was therefore able to
devote all their marketing and product development efforts to make business easier for eBay sellers, a
nicely compact set of super-users with defined needs. PayPal positioned itself as the friend of the
micro-seller.
This is similar to the “send money home” positioning of M-PESA. M-PESA too started with a fairly
generic payments proposition, but identified a sub-segment of users to whom this was going to solve
the biggest problems. M-PESA counted on them to drive the growth of the system virally with
payment recipients. In the case of PayPal, eBay sellers had a much larger leverage or viral effect,
because for every eBay seller there may be hundreds or thousands of potential eBay buyers who will
have noticed the PayPal payment option. Consequently, PayPal needed to spend a lot less on end-user
marketing than Safaricom did.
On the other hand, PayPal faced a constant fight with their ecosystem host, eBay, once eBay realized
that some of the value from their customers was going to PayPal. As the owner of the platform, eBay
sought to derive significant advantage from integrating its own payment engine into its marketplace
website. But in the end PayPal won out because they focused unrelentingly on customer growth and
getting the network effects on their side before eBay could catch up. Their mad scramble for growth
paid off. Ultimately PayPal knew their continued success was going to be dependent on eBay not
shutting them out of their auction website entirely, so they sought increasingly to diversify from eBay
1
The analysis of the early development of PayPal in this box is based on Jackson (2004).
13
auctions as the primary business driver, and eventually sold out to eBay.
The second similarity between PayPal and M-PESA is that both were willing to incur huge losses to
get to critical mass quickly. In its first 9 months (in 2000), PayPal lost $92 million (vs. revenues of $6
million), amounting to a loss of $23 for each of the 4 million customers they acquired over the period.
And over the first 18 months, PayPal accumulated $137 million in losses, or $14 for each of the 10
million customers they acquired.
PayPal’s early losses were due to three factors. First, they offered a sign-up bonus of $10
(subsequently reduced to $5) to whoever referred a new customer. Referrals tended to came from
eBay sellers, so they had every incentive to promote PayPal alongside their own products through
eBay. eBay sellers played an equivalent promotional role as the retail agents did for M-PESA. Second,
in the early days customers funded their PayPal account from their credit cards, and PayPal absorbed
the merchant fees on the card. So PayPal lost money not only each time they acquired a new
customer, but also each time these customers used PayPal. This cost is equivalent to the agent
commissions on M-PESA deposits, reflecting the cost of funding customer accounts. Third, PayPal had
escalating fraud costs, which at one point reached 2.4% of volume transacted. This was linked to the
use of credit cards as a funding mechanism: fraudsters could pump money out of stolen cards through
PayPal, and then the legitimate owner of the card would trigger a reversal of the charges, causing a
loss for the merchant in question -- which happened to be PayPal.
A significant difference between the two, though, is that M-PESA had a workable business model from
the beginning on the strength of its P2P revenue stream, whereas PayPal had to shore up their
business model in the years after its launch. PayPal was willing to ramp up such losses because their
number one objective was creating a large base of users to get the network effect going. After 18
months, once they had over 10 million customers and no competitor was going to catch up with them,
they started focusing on how to make money, or at least not to lose so much of it. They increased
revenues by creating a new kind of “premium” business account for eBay sellers, while keeping basic
“consumer” accounts free. They got their super-users to start paying for their account, partly by
upgrading these accounts with useful innovative features and partly by cajoling frequent users. They
also contained costs by shifting the primary funding mechanism from people’s credit cards (which
incur merchant fees) to their bank account, from which they could transfer funds for free through the
automated clearinghouse (ACH) capability. They also became better at spotting suspicious transactions
and trading patterns, and were able to reduce fraud in half.
How then should a scheme promoter design an investment strategy that will help it avoid the sub-scale
trap? Once a scheme has completed technical trials and gained confidence in the deployment’s long-
term business model, it must dive into the deep end, making big early investments in marketing,
merchant commissions, and merchant training in order to overcome the three barriers which threaten
these systems — network effects, chicken-and-egg trap, and lack of trust in the new system. Otherwise,
the positive networks effects associated with payment systems never kick in.
14
Mas and Radcliffe (2011) categorize the reasons that may explain Safaricom’s success with M-PESA in
terms of country factors, service design factors and execution factors. But a key ingredient of M-PESA’s
success was Safaricom’s heavy investment over the first two years of its deployment in platform costs,
customer acquisition commissions, TV and radio marketing, and aggressive merchant acquisition and
training. Safaricom hasn't released any figures on its investment in the M-PESA roll-out, but it may have
been as much as $25 - $30 million over the first two years. In other words, Safaricom didn't grow M-
PESA through tepid, incremental steps because, when it comes to mobile money, gradualism likely leads
to failure. After small pilots involving less than 500 customers, Safaricom launched M-PESA nationwide,
making heavy up-front investments so it could reach a critical mass of customers in a short time-
frame. Many deployments have potential to scale, but are stuck in the “sub-scale trap” because their
promoters either under-estimate the investments needed to achieve scale or are unable or reluctant to
make these investments because they can point to only one major success – M-PESA.
15
Appendix 1: Key Business Model Elements that Drive Scale
Element Hypotheses Questions
Identifying the pain point- Remote payments is the optimal Willingness to try- What is the pain point being addressed and how immediate is that need? Will the system rely
"gateway" product because it (i) has the highest customer primarily on remote payments (e.g. domestic remittances; bill payments) or proximity payments (paying for bread at
willingness to pay (and more importantly to try) because it the local store)?
solves an immediate pain point for customers, (ii) helps Willingness to pay- Does the pricing correspond to the pain points (highest price = biggest pain point)?
Early revenue build trust in the system because customers can verify in Size of the need- Is there a ready pool of payments (remittances, bill payments, G2P payments) that can drive
driver real-time when their transaction has been successfully considerable transaction volume through the system (>50 transactions per agent per day)?
processed, and (iii) draws into the system more urban, tech Barriers to adoption- Does the service encourage take-up and usage of the system by permitting free registration,
savvy, and cash-flush early adopters who are more easily free deposits, and charging no monthly fees (while charging on transfers and withdrawals)?
marketed to and who can encourage remittance receivers to Building customer trust- Does the service provide an instant and tangible confirmation of the transaction (e.g. a
register. confirmation SMS, written confirmation in an agent log)?
Heavy marketing (pull)- To reach a critical mass of
Ambition level- How many customers and agents does the scheme plan to register in 6 months, 1-year, 2-years?
customers you need (i) a strong brand to build trust in the
Brand relevance- Is the scheme promoter's brand well-known among the target population? What is it known for?
system, (ii) a clear, simple message focused on the primary
Clear, simple message- What is the scheme's marketing message? Is it clearly connected to a pain point?
need the service is addressing, (iii) a national above-the-line
Heavy up-front marketing- What is the scheme's marketing plan in the first year (e.g. direct marketing to individuals;
marketing campaign (TV, radio) to establish top-of-mind
Breaking the above-the-line marketing to large segments of the population)? What is the marketing budget per target customer?
awareness among large segments of the population, and
chicken & egg Greasing the channel- What is the scheme's customer acquisition strategy (e.g. through agents only, third-party
(iv) below-the-line marketing activities like canopies and
problem sales forces, existing customers)? What does the scheme pay agents for new customer acquisitions? What
street plays to educate customers on how they stand to
commission rates do master agents and sub-agents receive for transactions conducted, and how do these compare
benefit from the service and how to use it.
to airtime commissions?
Greasing the channel (push)- To reach critical mass, scheme
Barriers to sign-up- What steps are required to sign up a new customer (e.g. KYC reviews, documentation, time
operators will have to incur substantial customer acquisition
delay, SIM swap, remote versus on-site KYC)?
costs (beyond marketing and promotion).
Agent selection- The optimal stores for promoting an Agent selection- What criteria will the scheme promoter use to select cash-in/out agents? What steps are required
mobile banking scheme are (i) located in areas where for an agent to register (application form, KYC, training, cash bond)?
people transit/congregate, (ii) have the financial capacity to Ongoing management- How will the scheme ensure consistent customer experience across stores, in terms of
Active maintain adequate liquidity, and (iii) are seen as pricing, store branding, and customer service (e.g. monthly on-site supervision)?
channel trustworthy in the community. Channel management roles- How will the scheme divide channel management functions (agent identification,
management Consistent customer experience- Channel managers must screening, enrollment, training, supervision, liquidity management, distribution of agent commissions) across its
conduct ongoing onsite supervision of agents to ensure channel partners?
consistent customer experience throughout the agent Liquidity management- How will the scheme ensure that stores maintain adequate cash and e-value balances (e.g.
network. directly through bank branches, through master agents, other methods)?
Feeling for the business- Can the team easily answer all the questions listed above? Can they justify significant
Capacity - The operational team should be laser-focused on
departures from the M-PESA model? What other schemes have they thoroughly analyzed?
solving a major customer pain point and have a thorough
Top-level focus on marketing/channel ramp-up execution- Is there a dedicated team working on the project? Are
Execution understanding of the business model and critical factors for
key internal decision-makers (i.e. those with control over budgets and the channel) committed to ensuring proper
capacity success.
execution of the scheme's early stage marketing and channel ramp-up activities.
Top-level commitment - Key internal decision-makers must
Readiness to invest- In orders of magnitude, how much time and money do key decision-makers expect to invest in
be fully committed to building the scheme at scale.
the scheme before reaching break-even? Do they have detailed financial projections?
16
References
Alexandre, Claire, Ignacio Mas, and Dan Radcliffe (2010). “Regulating New Banking Models that can
Bring Financial Services to All.” Challenge, Vol. 54, No. 3, May/June 2011.
Alliance for Financial Inclusion [AFI] (2010). “Enabling mobile money transfer: The Central Bank of
Kenya’s treatment of M-PESA”
Beck, Thorsten, Asli Demirguc-Kunt, and Maria Soledad Martínez-Peria (2007). “Reaching Out: Access to
and use of banking services across countries,” Journal of Financial Economics 85 (1): 234–66.
Collins, Daryl, Morduch, Jonathan, Rutherford, Stuart, Ruthven, Orlanda (2009). Portfolios of the Poor:
How the World’s Poor Live on $2 a Day, (Princeton: Princeton University Press).
Financial Access Initiative (2009). “Half the World is Unbanked.” Focus Note.
Financial Sector Deepening Trust *FSDT+ (2009). “FinAccess National Survey 2009: Dynamics of Kenya’s
changing financial landscape.”
Frederik Eijkman, Jake Kendall, and Ignacio Mas (2010). “Bridges to Cash: the retail end of M-PESA.”
Savings & Development, Vol. 34, No. 2, forthcoming.
Davidson, Neil and Paul Leishman, GSM Association (2010). “How to build, incentivize, and manage a
network of mobile money merchants.”
Davidson, Neil and M. Yasmina McCarty, GSM Association (2011). “Driving customer usage of mobile
money for the unbanked.”
GSM Association (2009). “Wireless Intelligence Database.” Online database, available at
www.wirelessintelligence.com.
GSMA Association (2010). “MMU Deployment Tracker.” Online database, available at
www.wirelessintelligence.com/mobile-money
Heyer, Amrik and Ignacio Mas (2009). “Seeking Fertile Grounds for Mobile Money.” Enterprise
Development & Microfinance, Vol. 22, No. 1, March 2011.
Jackson, Eric (2004). PayPal Wars.
Mas, Ignacio and Olga Morawczynski (2009). “Designing Mobile Money Services: Lessons from M-PESA.”
Innovations, Vol. 4, Issue 2, Spring (Boston, MA: MIT Press).
Mas, Ignacio (2009). “The Economics of Branchless Banking” Innovations, Volume 4, Issue 2, Spring
(Boston, MA: MIT Press).
Mas, Ignacio and Amolo Ng’weno (2010). “Three keys to M-PESA’s success: Branding, channel
management and pricing.” Journal of Payments Strategy & Systems, Vol. 4, No. 4, December 2010.
Mas, Ignacio and Dan Radcliffe (2011). “Mobile payments go viral: M-PESA in Kenya,” Journal of
Financial Transformation, September (forthcoming).
Mas, Ignacio (2008). “M-PESA vs. G-Cash: Accounting for their Relative Success, and Key Lessons for
other Countries.” CGAP, unpublished, November.
Mas, Ignacio (2008). “Realizing the Potential of Branchless Banking: Challenges Ahead,” CGAP Focus
Note 50 (Washington, DC: CGAP, October).
Mas, Ignacio, and Kabir Kumar (2008). “Banking on Mobiles: Why, How, for Whom?” CGAP Focus Note
48 (Washington, D.C.: CGAP, June).
17
Mas, Ignacio and Jim Rosenberg (2009). “The Role of Mobile Operators in Expanding Access to Finance,”
CGAP Brief (Washington, D.C.: CGAP).
Morawczynski, Olga & Mark Pickens (2009). “Poor People Using Mobile Financial Services: Observations
on Customer Usage and Impact from M-PESA.” CGAP Brief.
“M‐PESA Key Performance Statistics” and various financial reports available on www.safaricom.co.ke.
Suri, Tavneet and William Jack (2011). “The Economics of M-PESA.” Unpublished manuscript, available
at http://www.mit.edu/~tavneet/M-PESA-Final.pdf.
Tarazi, Michael and Paul Breloff (2010). “Nonbank E-Money Issuers: Regulatory Approaches to
Protecting Customer Funds.” Focus Note 63 (Washington, D.C.: CGAP).
1
FAI (2009).
2
Beck, Demirguc-Kunt and Martínez-Peria (2007).
3
Mobile penetration in Africa has increased from 4 percent in 2002 to 51 percent today, and is expected to reach
72 percent by 2014; see Wireless Intelligence (2010).
4
For further analysis on the regulatory implications associated with branchless banking systems, see Tarazi and
Breloff (2010) and Alexandre, Mas and Radcliffe (2010).
5
For further analysis on the role of mobile operators in promoting branchless banking schemes, see Mas and
Rosenberg (2009) and Mas and Kumar (2008).
6
M-PESA customer data is as of December 31, 2010 (www.safaricom.co.ke). Population figures are from the
United Nations (2010) (http://data.un.org/CountryProfile.aspx?crName=Kenya). Safaricom subscriber data is as of
9.30.10 (http://www.cck.go.ke/resc/statistics/SECTOR_STATISTICS_REPORT_Q1_1011.pdf).
7
Suri and Jack (2011).
8
M-PESA transaction flows are as of July 31, 2010. Kenya GDP figure is from the World Development Indicators
database, World Bank (July 2010).
9
Central Bank of Kenya (http://www.centralbank.go.ke/), Kenya Post Office Savings Bank
(http://www.postbank.co.ke/), and Safaricom (http://www.safaricom.co.ke/index.php?id=1073).
10
GSMA (2011); includes deployments launched prior to December 31, 2010.
11
For further analysis on the factors which led to M-PESA’s success, see Mas and Ng’weno (2010).
12
For further analysis on the country factors which contribute to a deployment’s success or failure, see Heyer and
Mas (2009). For further analysis on the regulatory treatment of M-PESA, see AFI (2010) “Enabling mobile money
transfer: The Central Bank of Kenya’s treatment of M-PESA”
13
Suri and Jack (2011) find that households who have access to M-PESA and are close to an agent are better able
to maintain their level of consumption, and in particular food consumption, in the face of negative income
shocks, such as job loss, livestock death, harvest or business failure, or poor health.
14
Merchants must keep a stock of cash and electronic value on hand so they can meet customers’ cash-in/out
needs. Hence, they must invest in working capital before they can start processing customer transactions.
15
In FY2010, M-PESA generated $94 million in revenues, equal to 9% of Safaricom’s total revenues
(www.safaricom.co.ke).
16
For further analysis on M-PESA’s service design, see Mas and Morawczynski (2009).
17
For further analysis on the informal and semi-formal financial services used by poor households in developing
countries, see Collins et al (2009).
18
For further analysis on the ethnographic evidence of M-PESA’s impact on informal social networks, see
Morawczynski and Pickens (2009).
19
For further analysis on how to market mobile money services, see Davidson and McCarty (2011).
20
A survey of 1210 users in late 2008 revealed that 70% of survey respondents claimed they had first heard about
M-PESA from advertisements, TV or radio; see FSDT (2009).
21
Safaricom initially paid customer registration commissions entirely up front. They now split the commissions —
50% when a customer signs up and the other 50% after the customer makes her first deposit — to ensure that
stores are incentivized to sign up customers who are more likely to conduct transactions.
22
For further analysis on how to build a merchant network, see Davidson and Leishman (2010).
18
23
For further analysis on how to build and manage a scalable merchant network, see Davidson and Leishman
(2010) and Eijkman, Kendall and Mas (2010).
19