Big banks-and-small-savers-gafis-project-report-dec2013

BIG BANKS & SMALL SAVERS
A new path to profitability
GAFIS Project Report
December 2013

“Banks can—indeed must
— be at the heart of any solution
to the potentially
destabilizing trend of
social and economic
inequality.
With their access to capital markets,
advanced technology, delivery
systems and public oversight, banks
offer a secure means for poor people to
accumulate wealth that can bring greater
economic security and social inclusion.
But establishing the kind of accounts
that poor and low-income
savers often need—low
minimum balance, high
transaction rates and easy
access—can be a significant
challenge to profitability and
sustainability.”
Chris Page, Chair
GAFIS Program Advisory Committee

GAFIS PROJECT REPORT | 1
Contents
Foreword ..............................................................................................................................2
Introduction..........................................................................................................................3
The report format.................................................................................................................5
1. Fostering institutional change.........................................................................................6
Drivers of institutional change.....................................................................................7
Financial inclusion in context: who are the banks’ clients?.......................................9
The structural shift: from margin to mainstream.....................................................10
2. Building the business case for a low balance account................................................14
Making low-balance accounts profitable..................................................................17
3. Gateways as customer acquisition and “high-touch” channels.................................20
Agents as a gateway for deposit taking.....................................................................23
Effects on the business case: activation and activity ...............................................28
4. How the poor save, informally and formally...............................................................30
The role of the bank in the client savings portfolio..................................................32
Conclusion: the journey to date and the way ahead.......................................................36
Imagining Proposition 2.0..........................................................................................37
The implications for other banks and financial service providers...........................38
References .........................................................................................................................40
Glossary...............................................................................................................................41
GAFIS project structure .....................................................................................................41
Annex A: GAFIS bank product features ...........................................................................42
Annex B: Stylized account assumptions: Proposition 1.0 ..............................................43

• Our five partner banks—Bancolombia,
BANSEFI (Mexico), Equity Bank (Kenya),
ICICI Bank (India), and Standard Bank
(South Africa)—have aggressively tested
new approaches to attract and better serve
poor savers. Candid and critical examina-tions
2 | GAFIS PROJECT REPORT
from the banks’ “champions” and
managers informed each other and shaped
the lessons and models reported here.
• The conceptualization, project
management, technical expertise and global
experience brought to this
assignment by the principals and field con-sultants
of BFA has been of
immense value. They have displayed
unparalleled support for the partner banks
while maintaining a clear and independent
perspective on the project outcomes.
• Members of the GAFIS Project Advisory
Committee (see page 41) contributed
diverse and valuable perspectives that
helped to shape our strategic approach and
assess progress through the project.
• We all acknowledge the often demanding
and complex administrative support
provided by RPA Associate Tejas Shah.
This report is a chronicle of our shared GAFIS
experience. We hope, at the end, you share our
optimism that opportunities exist to reduce barri-ers
and reach new levels of financial inclusion.
Chris Page, Senior Vice President
Rockefeller Philanthropy Advisors, Inc.
23 December 2013
Foreword
This report emerges at a time when public trust
in key institutions, and banks in particular, is at a
historic low, fueled by the recent global econom-ic
crisis and the perception that reckless bank-ing
practices were at fault. Such an assessment,
however, obscures the broader issue of rising
social and economic inequality in the world. It is
estimated that less than 1% of the world’s popu-lation
holds 41% of all its wealth. Meanwhile, just
over two thirds control a mere 3% of all wealth
and struggle to convert those assets into healthy,
safe and productive lives.
Banks can—indeed must—be at the heart of
any solution to reverse this potentially destabiliz-ing
trend. With their access to capital markets,
advanced technology, delivery systems and
public oversight, banks offer a secure means
for poor people to accumulate wealth that can
bring greater economic security and social inclu-sion.
But establishing the kind of accounts that
poor and low-income savers often need—low
minimum balance, high transaction rates and
easy access—can be a significant challenge to
profitability and sustainability.
As part of an ongoing commitment to increase
access to financial services among the world’s
poor, the Bill & Melinda Gates Foundation gener-ously
supported a four-year project (2010-2013),
Gateway to Financial Innovations for Savings
(GAFIS). Under the sponsorship of Rockefeller
Philanthropy Advisors (RPA) and the manage-ment
of Bankable Frontier Associates (BFA),
GAFIS engaged five commercial banks in a quest
to understand the factors that i) encourage or
inhibit poor savers; ii) affect institutional capacity
to serve them and; iii) contribute to a sustainable
business model.
For all of us, this has been a journey filled with
fresh insights, remarkable vision, and exemplary
commitment among all the participants.

Can large banks make low-balance accounts viable and profitable?
above and analyze how banks could change the
dynamic. GAFIS did so by identifying and then
supporting financial institutions in designing
and delivering savings products that are useful
and accessible to the poor—and which deliver a
sustainable business proposition. GAFIS worked
with five leading banks —Bancolombia,
BANSEFI (Mexico), Equity Bank (Kenya), ICICI
Bank (India), and Standard Bank (South
Africa)— that collectively hold more than $250
billion in assets and serve more than 77 million
customers.
The GAFIS journey (from 2010 to 2013) is one
mirrored in many banks around the world—a
journey from traditional, old-fashioned and often
exclusive banking practices, to 21st century,
inclusive retail banking. We dub this the journey
from banking Proposition 1.0 to Proposition 2.0
(see Figure 1). The most advanced GAFIS banks
are, at best, halfway through the journey, but the
GAFIS findings and commitment to change
inspire optimism that these oft-perceived “lum-bering
giants” can compete head to head with
more nimble, tech-savvy competitors and deliver
new value to low-income clients.
Introduction
What would make low-income
clients more likely to use bank
accounts to save?
How does the use of different
channels influence the outcome
for clients and banks?
People with low and moderate incomes save in
a variety of ways, as the seminal book, Portfo-lios
of the Poor,1 has shown. One way is using
a bank account. But the importance of a bank
account in a personal or household savings port-folio
varies greatly. Often, after some initial activ-ity,
clients let bank accounts go dormant; other
times, so long as the in-flow continues from
employers or government, the recipients make
one periodic withdrawal of all their funds, a
pattern labeled “dump and pull.”2 Banks across
the developing world encounter this low-balance
account profile among the so called mass
market of lower-income clients. 3
The Gateway to Financial Innovations for Savings
(GAFIS) project set out to answer the questions
Proposition 1.0
Targeted
products on a
pervasive platform
Proposition 1.5
Proposition 2.0
Undifferentiated
savings/transactional
account
Same basic savings
products, but increasing
dominance of agent
channel
Products and channels
differentiated by
client’s savings needs
1. Collins et al (2009).
2. That term is used inter alia in the description of South Africa’s Mzansi basic bank account (BFA 2009).
3. See Findex Note 01, 2012: most formal account holders in developing economies say they deposit and/or
withdraw in their account once or twice a month.
GAFIS PROJECT REPORT | 3
Figure 1: Big banks on the move
—with new channels and new
products
Robust large
scale agent
networks

4 |
Proposition 1.0 Proposition 1.5 Proposition 2.0
Savings Account No major changes Flexible
Peer learning, new analytical tools and external
accountability all supported change
management in each bank.
One indicator has been the creation of new
inclusive banking divisions within two of the
banks, with bottom-line responsibility for clients
in the inclusive segment. New products and
channels are now managed in a more integrated
fashion—with proactive CEO and board support.
In total, more than 4 million new GAFIS-linked
accounts have been opened during the project;
and poor savers use more than 543,000 of these
accounts.4
Development of an agent channel to origi-nate
and service accounts (Proposition
1.5) sharply improves the business case.
Even without changing the product offering, the
deployment of agents who can originate new
accounts and take deposits from clients radically
improves the business case for banks by displac-ing
more expensive branch-based transactions—
reducing the net loss per account per month
to -$1.02 (see Figure 5, page 30). The rollout of
an agent channel is a necessary condition for a
bank to sustain a mass account rollout, although
it is not sufficient in itself (hence it represents
only a half-step to Proposition 1.5). For the client,
access to a convenient agent network makes
it easier to deposit small sums as well as leave
funds in the account until needed.
Product Undifferentiated
Traditional low-balance bank accounts
(Proposition 1.0) are not close to
profitable. Basic bank accounts are the work-horse
product of retail banking today, but they
often don’t work well for the bank or for the
customer. For the bank, more than 80% of these
accounts typically have too low a balance to be
profitable on their own; and as many as 50% lie
dormant. Applying full costs, the “stylized” basic
bank account with a $30 balance loses the bank
$2.79 each month (see Figure 2, page 15). With
current cost structures, banks can only make a
profit on these accounts if the balances exceed
$800, a virtually unreachable average number
across a large segment of lower income clients.
Meanwhile, the customer often finds it incon-venient
and costly to make small deposits or
withdrawals through a bank branch structure.
Structural and business model change
precedes product innovation. Initially
framed as an issue mainly of product innovation,
improving the profitability of a basic account in
fact demands innovation in service channels,
internal structure and business model. Large
banks foster strong internal competition for
financial, human and IT resources. Marginal or
experimental products aimed at inclusion are
unlikely to succeed for long in the ongoing battle
for resourcing. But external support from the
GAFIS project helped to facilitate and buttress
the development of new inclusive propositions.
(subwallets)
Dominant
transaction
channel
Branch, ATMs Agents Mobile phone
Marketing Bought or Required Also sold;
with SMS outreach Mainly bought
4. Note that the total opened reflects only accounts linked to the GAFIS project, listed later in Table 5. See Box A for GAFIS definition of ‘poor’. ‘Savers’
were account holders with a positive balance in a GAFIS-linked account designed with specific savings features or else exhibiting savings
behavior of types defined in Section 4 within a general bank account.
GAFIS PROJECT REPORT

| 5
banks to identify potentially active clients, which
in turn increases the odds of cross-selling other
products.
By analyzing transactional data from the banks,
complemented by surveys and financial diaries
of clients, GAFIS identified three different
patterns of savings behavior—1) Spend down
slowly; 2) Accumulate; 3) Preserve, which
excludes other patterns not deemed savings
(e.g. so-called “dump & pull”). Low-income
clients today typically use informal products to
undertake different types of savings because
suitable bank products have not been available
or accessible. Broadening formal savings options
to support such a range of different types of
savings behavior is an important goal of
financial inclusion.
Banks that shift toward Proposition 2.0 are more
likely to endure and succeed as large retail banks
in the 21st century. In fact, the process of learn-ing
to serve low-income clients profitably with
new, flexible products should enhance a bank’s
ability to serve its mid- and upper-income clients
cost effectively. In the process, this shift could
change the face of large commercial and retail
banking.5
The report format
This report tells the story of the combined jour-ney
of the five GAFIS banks in the following way:
Section 1 highlights the changes banks have
made to their internal structures and processes
The GAFIS banks now collectively report some
25,000 agents, almost 10 times the number at the
start of the project, and five times the number of
their branches. GAFIS banks provide evidence
that an agent channel can play a significant role
in deposit mobilization for banks, not only in
processing payments or withdrawals. Building
a robust large scale agent channel is not easy,
however.
In addition to the agent channel, GAFIS banks
are also starting to use new communication
channels, sending out SMS messages with
reminders, information and offers. While GAFIS
banks’ use of SMS is still at an early stage, initial
evidence suggests that these messages can
stimulate higher activation rates in newly opened
accounts, further improving the business case.
At the same time, banks are battling with mobile
operators—which are rapidly emerging as com-petitors
for both agent channels and financial
services—for what they consider fair pricing on
channels used for transactional banking.
The retail bank of the future (which
embodies Proposition 2.0) combines
new agent and mobile channels with new
tailored and flexible products. A robust
agent channel enables banks to focus on tailor-ing
and targeting products to specific customer
needs. GAFIS banks are starting to combine
big data from their mass-market accounts with
qualitative survey insights to better understand
their low-income client bases. This data enables
5. Note that the concept ‘Banking 2.0’ has been used to frame a wider version of this proposition—see for example Peric (2013); here we use the
term ‘Proposition 2.0’ since the GAFIS framing comes from a more limited view, even if the implications are similar.

6 |
During the last three years, all GAFIS banks have
undergone considerable changes. All have expe-rienced
high turnover in staff: as one indicator of
this, only one bank had the same executive-level
GAFIS project champion at the end as at the
beginning; and only two the same project
manager. Similarly, only two banks had the same
CEO in 2013 as in 2009. A high level of staff turn-over
at all levels reflects, in some of the banks,
restructuring in response to market pressures;
and in others, strong competition for formal
banking skills in the wider financial sector.
While change in staff has been a common fac-tor,
the GAFIS banks (see Table 1) fit into three
categories based on different starting points and
drivers regarding financial inclusion:
State owned bank: BANSEFI of Mexico has a
direct mandate to provide financial services to
the poor;
Retail focused bank: Equity Bank of Kenya is
a multi-channel but mainly single-market retail
bank that has grown rapidly over the past decade
from its roots as a small building society to
become by far the largest retail bank (by number
of customers) in Kenya;
Commercial banking groups: Bancolombia,
ICICI Bank and Standard Bank are large, multi-channel,
multi-segment banking groups offering
a comprehensive range of products and services
across corporate, commercial and retail markets.
Across these clusters, the institutional change
observed during GAFIS has been most striking
among the latter group of large commercial
banks, where the change process is often
most complex.
to address the low-income market in a
meaningful way, and how they define
institutional success.
Section 2 analyzes the business case for typical
bank accounts, using a model stylized across
GAFIS banks to show why Proposition 1.0
accounts are unprofitable, and paths to
improvement.
Section 3 describes the emergence of new
agent channels, their role in account acquisi-tion,
activation and deposit mobilization—and
how the new cost structure radically improves
the business case for low-balance accounts.
Section 4 looks at demand, examines how
bank savings fit into low-income client
portfolios, how banks are using big data and
customer surveys to deepen their understand-ing
of client needs, and how this can inform
value propositions for low-income clients.
The Conclusion looks ahead to the coming era
of Proposition 2.0, when new channels deliver
new products to new, poor savers.
1. Fostering
Institutional Change

| 7
Country Colombia Mexico Kenya India South Africa
Ownership Private Public Private Private Private
Assets
$ billion 44 2 3 91 100
Rank in
country
1 27 3 2 1
(by assets)
Branches 1,002 493 159 3,350 619
Customers 7m 12m 7m 40m 11m
ROE 12.1% 5.7% 31% 12.9% 18.5%
Market cap
$ billion $12.1 N/A $1.5 $18.7 $19.2
Source: Bank financial statements, BFA analysis on most recent year end, mainly 2013.
was formalized in 2012 as a code that would
hold all financial sector players, including
banks, accountable for their efforts to pro-mote
black economic empowerment. These
requirements for private players are different
from the clear public mandate that BANSEFI
carries as a state bank in Mexico to serve
remote areas and underserved people.
• Regulatory compliance pressures are also
growing. In part as a reaction to the global
financial crisis of 2008, most countries have
been designing tighter regulations. The
Basel 3 Accord is now being implemented in
three of the GAFIS countries with potential
implications for savings products (described
further in the Business Case section below).
The greater emphasis on regulatory compli-ance
among major banks can absorb scarce
executive bandwidth and curb banks’
Drivers of institutional change
Powerful underlying forces have been driving
profound changes in retail banks worldwide over
the past decade or more:
• Government mandates to promote financial
inclusion are expanding. All the countries
in which GAFIS banks are represented now
explicitly aim to promote financial inclusion
as a national policy objective,6 although
only India has an explicit policy of directly
mandating the involvement of private banks
in defined segments (for lending). ICICI
Bank was first requested to deliver a three-year
financial inclusion strategy containing
detailed targets for deposit accounts with
Indian bank regulators in 2010, and this plan
was updated in 2013. In South Africa, the
Financial Sector Charter, first developed
as a voluntary industry agreement in 2003,
Table 1:
GAFIS Partner
Banks
6. The regulators in all five GAFIS countries are signatories of the Maya Declaration in which they commit to specific actions to promote inclusion:
see http://www.afi-global.org/maya-declaration.

8 |
In short, these drivers are changing the way
banks do business, especially at the low end of
the market. Banks must innovate, but financial
inclusion can be a double-edged sword: Offer-ing
the promise of growth in countries where tra-ditional
middle and upper retail banking markets
are considered saturated, combined with the
threat of regulatory or political disruption to their
business models if left unaddressed.
appetite for risk and innovation.
• The concept of “financial citizenship” is
taking root. The global financial crisis, to-gether
with other bank-related scandals, has
challenged the legitimacy of large banks and
the notion that they play a beneficial role in
society.
• Competition from outside the banking
industry is growing. Perhaps most signifi-cantly
for the GAFIS banks, technology and
regulatory changes have unleashed new
competition for their traditional client
markets. Equity Bank has faced fierce
competition from Safaricom’s M-PESA, the
world’s leading mobile-money service, fo-cused
on P2P transfers. Competition height-ened
further with the launch in 2012 of Sa-faricom’s
M-Shwari savings account (issued
by a local bank far smaller in retail pres-ence
than Equity), which also offers clients
access to small loans. Standard Bank and
Bancolombia have both seen smaller, more
nimble competitors, such as Capitec Bank
or Daviplata, respectively, make inroads into
traditional retail banking space. In India, the
regulator plans to license new banks, and
continues to consider telco-driven mobile
money solutions, in part to encourage more
competition in retail banking. Even BANSEFI
faces the possibility that its current exclusive
distribution of paying the Oportunidades
G2P scheme could end in the future.

| 9
Table 2: Proportion of mass market clients considered poor according to different metrics
Country Colombia India Kenya Mexico South Africa
GDP pc PPP,
2012 (WDI) $10,587 $3,876 $1,761 $16,731 $11,440
% of adults
with a formal
account
30% 35% 42% 27% 54%
(Findex 2011)
GAFIS banks: Share of mass market clients below:
Segment
measured
Mass market
(“Personal”) N/A Mass market
(“Ordinary”)
Sources: * Banks: GAFIS surveys. Note: Results for ICICI Bank are not shown because the GAFIS baseline
did not survey the actual client base; for Bancolombia and Standard Bank, the customers measured were
intentionally the “lower” socioeconomic part of the respective mass markets.
with a sizable proportion of poor and near poor.
Such a high proportion is to be expected for a
state bank like BANSEFI, which specializes in
government transfers, but even among large
commercial banks’ mass market segments
the proportion of poor customers ranges from
39-70%, according to the locally defined poverty
line (see Box A below for the GAFIS definitions
of poverty). If near-poor customers are included,
the proportion rises to 57-79%.8
For banks, basic transactional and savings prod-ucts
have cast the net widely, in the hope that
some features (such as salary deposit) may
Financial inclusion in context:
Who are the banks’ clients?
The level of formal banking penetration varies
across GAFIS countries (see Table 2). In South
Africa, more than half of adults have bank
accounts, while in Colombia, Mexico and India
the proportion of poor clients is more in the
30-40% range. The Kenyan percentage of 42%
includes a large proportion of non-bank mobile
wallets that constitute significant competition
to banks; adults with bank accounts are likely
closer to 29%.7
The client bases of GAFIS banks are diverse,
7. FinAccess National Survey, 2013
8. “Near poor” is defined here as 25% above the national poverty line (see Box A).
G2P
recipients
Mass market
(“Inclusive”)
National
poverty line* 39% N/A 47% 92% 70%
$2 PPP/day* 0% N/A 25% 46% 28%

10 |
(due to privacy restrictions).
In Mexico, the GAFIS-supported clients are
recipients of a major government program
(Oportunidades), which applies poverty target-ing;
and in India, a means test is applied by
ICICI Bank’s agent network partner (CASHPOR),
which ensures all clients are “poor.”
The structural shift: from margin
to mainstream
For banks, innovation is a complex and multi-faceted
process involving changes in resourc-ing,
structure and systems. Innovators have to
compete aggressively for resources with other
departments.
The three GAFIS commercial banks have all ad-opted
different approaches to the challenge of in-novating
within a large institution, a process well
described in Kosta Peric’s book, The Castle and
the Sandbox (2012). As the leader of SWIFT’s
Innotribe project to stimulate innovation in the
SWIFT community of banks, Peric uses the im-age
of the “castle” to depict an existing business
such as (but not limited to) a large bank. These
businesses have constructed elaborate systems
and procedures as part of their success as “con-servative
companies in established industries”
(part of the book’s subtitle). These systems are
often designed to withstand risk and change; to
innovate, they need to establish “sandboxes,” or
incubators, outside of the castle walls in which
open approaches to innovation can be tested.
Standard Bank has a long history of trying
different approaches to what it used to call the
mass market (see Box B ). In 2007, the Bank set
up a “sandbox” in the form of a dedicated Com-munity
Banking team outside of the retail bank,
ultimately increase usage and that the financial
service needs of some clients will grow with
their income over time. The starting point
for these partner banks mirrors that of
commercial banks in many countries of the
developing world.
Box A: GAFIS definition of poverty
Poverty lines are cut-off points separating the
poor from the non-poor, and there are a variety
of ways to measure them. In the GAFIS project,
in most countries, BFA implemented both a
baseline survey to understand the needs of poor
clients and also an endline survey to measure the
profile of new client bases.
To identify poor clients, GAFIS used national
poverty lines specific to each country, viz.:
• Colombia, COP 202,083 ($107) per capita,
per month
• South Africa, ZAR 762 ($76) per capita,
per month
• Kenya, KES 9,526 ($112) in urban areas and
KES 4,456 ($52) in rural areas
(monthly per capita).
Whenever possible, consumption measures
were preferred over income measures, as they
are typically superior indicators of well being,
smoothing for income fluctuations over the
year. For Kenya and South Africa, for example,
monthly household expenditure was estimated
by aggregating reported expenses for various
household categories. In Colombia, while a simi-lar
method was used for the baseline survey, an
income measure was used in the endline survey
(from the National Statistics Office in 2012), as
this was deemed the most reliable method based
on limited available access to existing customers

| 11
Banking (CB). CB was to focus on developing in-novative
ways of better serving the lower-income
segment. An important part was agent outreach:
For the first time, recruiting local merchants as
Standard Bank agents in the communities where
lower-income segments lived and often worked.
It was essentially a “sandbox” experiment in
terms of bank strategy and operational structure.
GAFIS started its work with SBSA at this stage.
In 2010, as part of a general bank-wide restruc-turing
driven by cost cutting, Standard Bank
integrated the Community Banking team as the
kernel of a new “Inclusive Banking” division.
This segment has responsibility for all bank
products, savings and loans, offered to the mass
market, which covers 70% of all the Bank’s cus-tomers
(those earning up to $800
per month).
The Inclusive Banking group was now required
to migrate all its accounts to the bank’s new SAP
core banking system and all marketing efforts
had to conform to long-established group stan-dards.
New product development was temporar-ily
frozen during this transition and most of the
original Community Banking team left the bank.
The “sandbox” had been relocated into the
“castle,” which restricted its freedom to innovate
in the same ways. However, a new leader took
over in 2011, with accountability for the bottom-line
contribution of this large segment to the
retail bank’s profitability. She was able to make
the internal case for resources, rolling out a new
suite of “Access” products in 2012. There was
also a mass rollout of around 7,500 agents, many
which reported to the investment bank respon-sible
for new products. Led by a non-banker,
the team experimented with the bank’s first
deployment of agents, now called Access Points
(previously called Bank Shops), and with mobile
banking products separate from the bank’s core
banking system.
Box B: From “the sandbox to the castle”:
Integrating financial inclusion into
Standard Bank’s core strategy
As one of the oldest and largest retail banks in
South Africa, Standard Bank has a long history
of outreach to what it used to call the “mass
market” segment. Beginning in the 1990s, as
apartheid crumbled, Standard Bank actively tried
different approaches to become more inclusive.
One of its first attempts was to create a new ring-fenced
bank, E-Bank, which evolved into E-Plan,
a transactional and savings product offered as
part of the main retail banking operations. This
became SBSA’s core deposit account offering
for the lower rungs of the formally employed
and salaried segment, serving around half of all
SBSA retail bank customers. In 2004, SBSA, to-gether
with the other large South African banks,
launched the Mzansi basic bank account, aimed
at low-income “informal” workers, using largely
ATM channels. Although the bank met its targets
for opening Mzansi accounts, this new product
was a mixed success in that there was little activ-ity
in the accounts compared to E-Plan. Mzansi
lacked an accessible channel to appropriately
serve this target market.
In 2007, SBSA formed a new unit outside of
its retail banking structures called Community

12 |
systems, which also led to freezes in product
development and ultimately delayed the rollout
of the bank’s new mobile-only product called
Ahorro a la Mano (ALM). With GAFIS support,
Bancolombia invested heavily in technological
adjustments to accommodate the ALM innova-tions
of over-the-air account opening through
mobile and cardless ATM transactions.
ICICI Bank’s organizational structure to
manage the lower-income segments has not
changed as profoundly in this period. The Rural
and Inclusive Banking Group (RIBG), headed
by a Senior General Manager reporting directly
to the Executive Director in charge of all Retail
Banking, already existed and continues to be the
line unit responsible for implementing the bank’s
financial inclusion strategy. RIBG is now revenue
generating and, while initially focused primar-ily
on opening new accounts, is now actively
promoting greater usage of opened accounts, in
part based on experience with its Apna accounts
(supported by GAFIS).
located in peri-urban and lower-income areas
where there are no bank branches. Inclusive
Banking managed to win internal approval for
new above-the-line and below-the-line market-ing
approaches, including TV and radio
mini-soap operas and guerrilla marketing (pic-ture
on prior page). The Inclusive Banking divi-sion
is now seen as integral to Standard’s retail
bank—a core part of the redesigned “castle.”
Bancolombia arrived at the same outcome via
a different route. After initially addressing inclu-sion-
related issues on a project basis, in 2012,
Bancolombia set up a dedicated financial inclu-sion
unit reporting to the head of the retail bank.
The adoption of financial inclusion as one of the
bank-wide strategic priorities under its new CEO
in 2011 required a redesign of the “castle.”
The new strategy brought a big paradigm shift.
In the past, low-balance accounts were con-sidered
a problem to be solved, rather than an
opportunity to be pursued. This change occurred
at the same time as a transition in core banking

| 13
marked in structural terms in part because the
internal structures of this smaller retail-focused
bank were simpler at the outset, the Bank was
already focused on moderate income Kenyans,
and also because their IT systems were more
recently implemented. As a result, they could
accommodate changes in channel and prod-uct
more easily. From the sometimes envious
perspective of the larger commercial banks,
Equity’s “castle” often looked during GAFIS
more like a “sandbox” in terms of its flexibility
and ability to embrace rapid change. However,
while the smallest of the GAFIS private banks by
assets, Equity is the fastest growing: from around
250,000 customers 10 years ago to around 7 mil-lion
today. It is also the most profitable measured
by return on equity (see Table 1 on page 7).
Equity demonstrates to larger banks the poten-tial
of a focused model, even in a competitive
domestic environment. However, such rapid
growth places inevitable pressure on organi-zational
capacity and even culture, as the staff
changes. By 2010, Equity senior management
introduced changes to reorient its branch and
operational managers to serve its lower-income
segment in the face of strong competition. It has
also had to manage the rollout of a new agent
channel, with substantial resource allocation
prioritized by the CEO. With GAFIS support, the
Bank established a new Research unit compris-ing
a half-dozen full-time employees focused on
understanding the financial needs and patterns
of the “common Kenyan.”
Originally, the financial inclusion accounts
offered by RIBG were hosted on a separate
version of its core banking system, in order to
handle large volume small-value transactions.
As a result, these accounts were not connected
electronically with the bank’s mainstream retail
channels. Off-line transactions were primarily
handled by agents (Business Correspondents,
or BCs). However, in 2012, ICICI began to en-able
the financial inclusion accounts to access
the same functionalities as accounts on the
mainstream core banking system. For example,
domestic remittances can now flow directly into
the GAFIS-supported (CASHPOR) Apna savings
account from outside (see picture above). This
IT-specific change underlines the process of
mainstreaming of the inclusion segment within
ICICI’s “castle.”
In all these three large banks, there is now a
well-resourced line department, headed by a
senior manager and reporting directly to a senior
executive in charge of the retail bank. These
departments are also responsible for bottom-line
financial results, measured on the same indica-tors
as other mainstream bank departments:
headline earnings, economic profits and return
on equity. Importantly, these inclusive banking
units all integrate segment, product and agent
channel responsibility in one group. This integra-tion
enables the banks to manage better the of-fering
of products through their agent channels.
Equity Bank has also undergone profound
changes. However, Equity’s changes were less

14 |
The GAFIS banks differ substantially in how they
measure costs. Only one bank undertakes full
activity-based costing, while the others generally
use hybrid-costing variants to assign transaction
costs to each service channel. Rather than simply
average out the (sometimes extreme) differences
among them, the GAFIS project developed a styl-ized
financial model of a (Proposition 1.0) typical
bank account, based on the profile of fees, costs
and usage that depicts the general case across
GAFIS banks (see Figure 2).10
Note that the “ACTA” model allows for dormancy
as a cost of (A)ccounts, since the median dor-mancy
rate across basic accounts was around
50% (with a range of 20% to 80%, depending in
part how dormancy is defined). Costs to open
accounts that subsequently go dormant are sub-stantial.
Adjacencies in the stylized model did not
include any cross-sell, since the evidence of this
happening at large scale was limited. The net
result, shown in red (in Figure 2), is the proposi-tion
that a stylized account with a $30 balance
and three transactions per month may lose the
bank $2.79 per month.
2. Building the business case for a
low balance account
Do banks make money offering low-balance
accounts? The short answer today is no, but it
does depend on how one defines and measures
‘making money’.9 In this section, we apply the
“ACTA” framework, developed by the Bill &
Melinda Gates Foundation in 2013, to address the
question of profitability of low-balance accounts
in GAFIS banks (see Box C).
Box C: The “ACTA” framework for
assessing account profitability
In 2013, the Financial Services for the Poor team
at the Bill & Melinda Gates Foundation published
a report called Fighting Poverty through
Payments, which frames a way of categorizing
the components of bank account profitability.
The framework recognizes four possible sources
of profitability from a financial account:
• (A)ccount. Revenues include fees charged
on the account itself less indirect costs such
as maintaining the account on a core IT
system and amortized cost of opening an
account;
• (C)ash handling. Revenues include fees
for cash-in, cash-out (CICO) transactions
(whether ATM, branch or agent) less the
channel costs to process these cash-handling
transactions;
• (T)ransactional. Revenues for other elec-tronic
transactions, whether charged to
client or to another, (e.g., a corporate payee
of bill payments from the account); and
• (A)djacencies. Income earned on the
deposit float at a bank, as well as cross-sell,
up-sell accounts.
9. GAFIS Focus Note 3, available on www.gafis.net, identified five
different levels at which the business case was made for banks to
introduce or sustain a product or segment.
10. To see the math calculations behind Figure 2, see Annex B.

| 15
Figure 2: Profitability for a Proposition 1.0 bank account (with $30 balance)
A C T A Net
the value, hence generating most of the adjacen-cy
float income. For Standard Bank’s AccessSave,
balances are even more concentrated. Only the
top 14% of account holders have balances above
$25, with an average balance of $350, which
represents 98% of total portfolio balance. But,
86% of opened accounts (including those even-tually
closed by the bank for non-activity) have
balances under $25, with an average of under $2,
very small even relative to monthly expenditures
in poor households. Other GAFIS banks show
similar dynamics depending on the breadth of
the market served. This skewed distribution
dramatically affects overall adjacency income
received on the float.
As shown in Figure 3, the highest average
balance in a GAFIS account category is
Standard Bank’s $151, for its AccessSave ac-count,
followed closely by Equity Bank’s $135 for
its Jijenge and School Fees accounts. Bancolom-bia
and Standard Bank figures reflect high ratios
of zero-balance accounts (a median of $0 for
each). ICICI Bank’s figures reflect a relatively flat
spread among its accountholders, with no
non-poor higher balances to bring up the
average or median.
-1.53 -1.43
0.50
0.00
-0.50
-1.00
-1.50
-2.00
-2.50
The GAFIS stylized model above reflects a “TA”
model in ACTA terms: (T)ransactions and
(A)djacencies make small but positive contribu-tions
but (A)ccount and (C)ash-handling costs
overwhelm these minor positive gains. In fact,
in some banks, high branch fees discourage
customers from transacting at the branch, and
they turn to ATMs (for withdrawals). To the
extent that banks can charge adequately for ATM
withdrawals, profits on cash-handling revenues
can become an important part of the business
model, although usually not sufficient to offset
the large loss in the (A)ccount category.
In the traditional business model for a basic
Proposition 1.0 account, a big issue is that
the float income (counted under (A)djacen-cies)
from the $30 balance falls far short of the
amount needed to cover relatively high costs
(reflected in the small $0.11 in Figure 2); indeed,
it would take an average balance $800 higher, all
else equal, for the account to break even. This is
far out of reach for most low-balance accounts
(see Figure 3).
In practice, savings balances vary greatly across
a portfolio of accounts. Typically, 20% (or less) of
the accounts carry more than 80% (or more) of
0.07 0.11
-3.00 -2.79
Net profit per account per month
Source: see Annex B

16 |
Figure 3: Balance levels across GAFIS banks for GAFIS-supported products
(positive balance accounts only)
license of the bank to pursue other lucrative lines
of business.
The stylized model also does not value pos-sible
income growth of the customer over time.
Equity Bank, for example, explicitly speaks of
“growing with its customers,” and can point to
high-income clients today who opened Equity
accounts when they were quite poor. Thus, large
banks with a strong emerging customer seg-ment
represent a kind of portfolio option on the
growth of this segment.
While these strategic considerations certainly
can improve the overall business case for offer-ing
low-balance accounts, the underlying issue
remains: Until the inclusive customer segment
is perceived as profitable, banks will struggle to
justify internal resource allocation. Because of
the sheer numbers involved, banks will be vul-nerable
to more focused competitors who can
cherry-pick profitable segments.
Source: GAFIS banks
ICICI’s
Apna-
Cashpor
Bancolombia’s
Ahorro a la
Mano
A common business case concern with pursu-ing
the business of low income clients is that the
bank’s overall average balance is likely to fall; or,
at worst, the lucrative “20%” of higher income
clients may be driven away by poor service as
the result of overcrowded branches, ATMs or call
centers.
One clear implication for the business case in a
Proposition 1.0 world is that a savings product
aimed exclusively at low-income people (for
example, at the extreme, a product which sets
low absolute caps on balances or otherwise
excludes the non-poor) may be destined to fail.
This simple framework does not capture the
strategic financial or non-financial adjacency
benefits that banks may receive from these ac-counts.
For example, banks may earn additional
business from government or corporates and, if
regulatory compliance requires this, then offer-ing
loss-making products cross-subsidizes the
Equity’s
Jijenge & School
Fees
Standard’s
AccessSave
BANSEFI’s
G2P

| 17
Bancolombia has recently done this with its
new product (see Box D). Standard Bank has
also revised its costing approach so that a higher
proportion of allocation is based on transactional
profile, not on the existence of an account which
would penalize less active low-balance accounts.
Box D: Bancolombia modifies its
costing approach
Bancolombia’s costing model is considered
the most advanced among the GAFIS banks. In
generating monthly income statements for each
client segment, product portfolio and banking
channel, the bank uses a full activity-based cost-ing
approach to address how to allocate fixed
overhead costs to products and divisions, a cen-tral
facet of bank costing models. The portfolio-level
view captures variable channel costs linked
to customer transaction volumes as well as
fixed costs allocated to product lines, including
a monthly amortized origination expense per
account. Bancolombia classifies each client as
profitable, unprofitable or break-even based on
his or her total contribution across all products at
the bank.
During the GAFIS-supported development of
Ahorro a la mano (ALM), a cardless product that
uses only mobile phones to transact, the bank
realized that the existing cost allocations would
burden the new product with unrelated expens-es.
Previously, Bancolombia had not differenti-ated
customer acquisition and servicing costs by
channel, since the majority of customers opened
accounts at bank branches. However, ALM’s
reliance on over-the-air account opening meant
Making low-balance accounts profitable
If the typical basic bank account is not profit-able
on its own, what are GAFIS banks doing
about it? A number of strategies can be used to
improve the business case:
Adjust costing methodology where
appropriate. Among GAFIS banks, there is no
correlation between the rigor of costing models
and the bank’s overall profitability, nor with
the effectiveness of its outreach to low-income
people. In fact, the banks vary widely in their
level of financial analysis. At one extreme, the
monthly profit-and-loss position is prepared at
the level of each individual customer (based on
regularly adjusted activity-based costing). At the
other, some banks initially did little or no product
or segment-level analysis (with no associated
channel or product-level costing), though GAFIS
subsequently supported some such costing
development. Others fall in between, preparing
portfolio-level profitability, based on some chan-nel
or transactional benchmarks.
All bank costing methodologies involve choices
with consequences for ex ante decision making,
as well as ex post measurement of outcomes.
Examples include the treatment of fixed (or
indirect) costs. Such costs—if truly fixed—are
arguably less relevant (even irrelevant) in
decision-making during time horizons less than
or equal to the time such costs are indeed fixed.
These discretional elements need to be aligned
with the long-term strategy of the bank, such that
the costing model provides relevant information
to inform decision-making about where to invest
resources and how to incentivize stakeholders.

18 |
Fixed costs of deployment are lower for agents
while variable costs (commission) are higher,
hence the agent channel comparative advan-tage
may be as much about strategic flexibility
in growing physical footprint (quick set-up
with low/nil sunk costs) as lower unit costs.
• Account origination. Whereas branch
origination costs among GAFIS banks range
from $15 to more than $25, depending on
marketing allocations, agent origination
costs range from $1 to $6, depending on
commissions and overhead allocation.
Note that agent channels are usually not expect-ed
to be profitable in their own right: Leading
models do not charge the client to deposit, while
the agent expects a commission. If the client
merely deposits more at the agent channel with-out
decreasing branch activity, then the business
case may even be weakened, unless balances or
fee-earning transactions (such as ATM withdraw-als
or electronic payments) increase as a result.
Increase revenue from e-transactions on
either or both sides of a payment. Some
transaction types, such as government benefits
or remittances, involve an electronic credit into
an account. This has two effects on the business
case. First, the e-credit may displace the need for
it would draw substantially less, if at all, upon the
bank’s sales force and related platforms. Banco-lombia
therefore re-calculated a lower monthly
unit cost for ALM accounts opened remotely
and serviced (for CICO) off-site at agents or
at an ATM without a card. This cost reduction
improved the business case for the new prod-uct,
and for offering savings and transactional
products to lower-income segments in Colombia
in general.
Use agent channels to reduce origination
and cash-handling costs. Because most
GAFIS banks have deployed agents only recently
on a widespread basis, their costing approaches
for agent transactions vary more widely than for
more established channels like ATMs or branch-es.
Nevertheless, it is possible to represent the
stylized effect of agent channels on the account
profitability calculation as follows:
• Cash-handling (deposits and withdrawals).
The cost of a deposit or withdrawal at an agent
depends on the commission paid, which is
typically less than $0.50 and may be as low as
$0.10. The cost is as low as a fifth of the cost of
a branch-teller transaction. While the cost may
not be cheaper than an ATM withdrawal trans-action,
the composition of costs is different.

| 19
able for the bank. As explained in more depth
elsewhere,11 the effect of Basel 3 across GAFIS
banks is likely to vary widely, depending on their
balance sheet structure. For South African banks,
which have relied more on wholesale deposits to
fund their assets, the effects of Basel 3 may make
retail savings more attractive. This may increase
the internal float rate paid by the bank Treasury
for retail savings relative to corporate deposits
(thus increasing this adjacency income to the
segment even at the same balance level). But
this effect may also be countered somewhat as
banks also bid up interest rates paid to custom-ers
to compete for retail savers, a direct cost to
the segment.
A combination of reducing costs of origination
and cash handling, combined with more income
(and lower costs) from electronic transactions
and higher average balances in the bank, will
start to improve the business case of basic bank
accounts. This is especially true when costing
methodology is aligned to the long-term strategy
in this segment. In practice, all GAFIS banks are
attempting a range of initiatives to address vari-ous
aspects of the ACTA framework (see Table
3). However, to realize this new vision, banks
need to reinvent the channels, or gateways,
through which they transact with clients.
the recipient to make more expensive branch or
agent cash deposits into the account. Second,
the bank may earn additional revenue from fees
charged to the payer (such as employer or gov-ernment)
for sending the money to the payee’s
account. Since the payee does not pay for this
inflow, these flows are treated here as a type
of adjacency of revenue from the broader
ecosystem in which the account holder lives
and works.
Exploit cross-sell adjacencies. Most GAFIS
banks consider cross-selling of other products,
especially loans, as an important part of their
strategy. One bank in particular has an explicit
target of doubling the number of bank products
used per client during the next five years. Yet few
have to date cross-sold other products to low
income account holders on a consistent basis.
Value additional ‘adjacencies’. Three GAFIS
banks are in countries where Basel 3 regulations
affecting banks were implemented in 2013—
India, Mexico and South Africa. Some of these
new rules may have an effect on the business
case for retail savings. Specifically, new liquidity
requirements will require banks to better match
the liquidity profile of their assets and liabilities.
Since retail deposits are seen as more diversi-fied
and stable than wholesale deposits, they are
weighted more highly than before in coverage
calculation and therefore become more desir-
11. See AFI blog post: http://www.afi-global.org/opinion/2013/5/21/porteous-implementing-basel-iii-what-will-it-mean-future-financial-inclusion.

20 |
Table 3: GAFIS banks’ diverse efforts to promote account-level profitability
accounts to recipients of various government
or donor programs. ICICI Bank was one of the
first private Indian banks to be accredited to pay
out Indian government cash transfer schemes in
various states.
Many banks have seen this approach as a safe re-tail
strategy to serve low-income customers in a
limited way, since it ensures a regular inflow into
the account and a regular fee from the payer. If
that fee is high enough, or if the bank is able to
charge the client enough for the cash withdrawal
(though typically not allowed for G2P), these
accounts can be an attractive proposition, even
with very low balances.
Indeed, GAFIS’ original intent was to promote
additional savings behavior arising from such
natural gateways—to move beyond the domi-nant
pattern for G2P of “dump & pull” behavior
“A”
(Account)
3. Gateways as customer acquisi-tion
and “high-touch” channels
Traditionally, a bank savings account has been
regarded as a “bought” product, in that the
customer chose when and where to open the ac-count
(although their choice of bank may have
been influenced by corporate marketing).
A major exception is when bank accounts are
required to receive payment flows; for example,
when a government agency or employer re-quires
a recipient to open an account if
s/he wishes to receive the money due. Opor-tunidades
recipients in Mexico are required to
open BANSEFI accounts to receive their benefits.
For BANSEFI, paying conditional cash transfers
(CCTs, a form of G2P) for the Mexican govern-ment
is a core business. At the start of GAFIS,
Standard Bank and Equity Bank already provided
“C”
(Cash handling)
“T”
(Transactional)
“A”
(Adjacency)
Bancolombia
Lowering origination
& maintenance costs
(agent, mobile,
cardless platforms)
Shifting CICO to
agents or ATM
Encouraging e-debits,
via mobile
Incentivizing balance
growth & product
cross-sell w/ insurance
feature & tiered inter-est
rates
BANSEFI
costs (scale of G2P
mandates)
Shifting CICO to
agents
G2P fees earned from
delivering e-credits Encouraging cross-sell
of billpay
Equity Bank
costs (agent
channels); aligning
costing methodology
Shifting CICO to
agents
Encouraging e-debits,
via mobile (Ordinary
a/c, not “pure” savings
accounts)
growth & relationship
deepening eventually
leads to cross-sell
ICICI Bank
costs (agent
channels, mobile
technology); alterna-tive
core platform
CICO exclusively via
agents
N/A for savings
product, except for
encouraging e-credits
(P2P, G2P, B2P)
Encouraging balance
growth by promoting
e-credits to fund a/c
Standard Bank
costs (agent channels,
mobile technology)
Shifting CICO to
agents
N/A for savings
product, except for
encouraging e-credits
(salary, etc.)
growth with discipline
and bonus interest

| 21
eficiaries to a payment provider linked to a small
bank, following a procurement process marked
by legal challenge and controversy. All benefit
payments now have to be routed into those new
bank accounts, limiting the ability of recipients
to choose their bank. While beneficiaries may
withdraw cash from the benefit account and
then redeposit part into savings accounts at their
former bank, this is costly and inconvenient, so
many of the Standard Bank clients’ accounts
have gone dormant. This potential gateway
was also effectively closed by a government
procurement decision.
BANSEFI, which currently benefits from
exclusive distribution on payment of the ma-jor
Oportunidades CCT program in Mexico,
has yet to see significant savings activity in its
accounts. Even when its agents are legally able
to take deposits (still the minority), the most
commonly observed behavior pattern remains
“dump & pull.” This is entrenched by rules of
the CCT scheme, which discourage savings, and
by many recipients unaware of the existence
of an account and its features: a 2011 survey of
BANSEFI account holders revealed that less than
30% knew the account could be used for sav-ings.
Even when they do, a further challenge is
that beneficiaries fear leaving money in the same
account (or even the same bank), thinking the
Government may decide that if they can afford to
save, the welfare payment is no longer needed.
Engaging clients through SMS messaging
Savings, which involves accumulation of money
over time, requires more than a one-time sales
process; it requires an ongoing relationship in
which the client has sufficient access to the bank
account to deposit small amounts. For the client
to be confident enough to leave balances in the
account, s/he needs to be able to make small
withdrawals easily and cost effectively. This
requires the proximity and convenience that
agent networks can bring (discussed in more
detail later in this section).
with one withdrawal. However, the experience
of GAFIS partner banks pursuing G2P as a low-cost
acquisition strategy has been very mixed.
The downside of relying on G2P payments is
that a government can change its entire payment
approach or drag its heels (see Box E). The bank
proposition of providing financial services based
on the payment of government benefits alone
is not, in fact, safe, but risky; and can be hard in
practice to move beyond the payment only.
Box E: G2P payments as a gateway
For three GAFIS banks, government social trans-fers
were considered to be a primary gateway to
attracting more savings. However, at the end of
GAFIS, the evidence of progress towards G2P as
a gateway for savings remains limited.
ICICI Bank and its agent network partner FINO
opened more than 1 million accounts in the
Indian State of Haryana. Starting in August 2011,
ICICI established a new model for servicing these
new account holders by offering more regular
village presence (several times per week instead
of once a month as before). This would allow
clients to leave money in the accounts (and even
deposit their own funds for safekeeping), since
they could withdraw as needed rather than only
on a defined monthly payday. However, just two
months later, the State Government stopped
making payments via banks and reverted back
to making cash payments via local government
structures. Without regular transfers, only 5% of
the new bank accounts opened remained active.
After the bank had invested to build out the ser-vicing
infrastructure, this gateway was in effect
closed for a long period.
At the outset of GAFIS, Standard Bank had more
than half a million account holders who received
monthly G2P cash transfers into their accounts.
These payees were originally identified as a
primary target for promoting savings behavior.
However, in 2011, the South African Social Secu-rity
Agency awarded a single contract for all ben-
12. See Bold, Porteous & Rotman (2012) for the business case of G2P based on analysis which included three GAFIS countries—Colombia, Mexico
and South Africa, although focused only on BANSEFI among the GAFIS banks.

22 |
days when banks did not even bother to col-lect
or verify accurate information since postal
delivery was too expensive or not possible. The
content of the SMS engagements aim to catalyze
initial deposits and build trust by confirming
agent deposit transactions or providing other
useful information.
Bancolombia enlisted a U.S.-based firm, Juntos
Finanzas, to help design and deliver SMS content
for its ALM account holders. A particular aim is
to create a high-touch 2-way “conversation” to
inform (building trust), encourage routine, and
remind, evolving over time to a form of social
(media) encouragement, and ultimately offering
the convenience of links to e-transfer functional-ity
or locating the nearest CICO point via GPS.
Standard Bank, with a new transactional product
aimed at the low-income segment, initiated Proj-ect
Activate, and now sends a choreographed
series of SMS messages to new account holders
during the 100 days following account opening.
This is deemed a clear success by the bank, with
inactive ratios cut by around a third year-on-year.
Project Activate is now being rolled out with the
AccessSave account.
However, in addition to physical proximity, there
are other ways in which banks can reach out to
touch their customers to encourage or promote
savings. In traditional mass-market savings ac-counts,
very little consideration has been given
to the ex post touch. This is in large part because
banks could not contact clients in the mass
market segment—in many cases, because of the
cost implications, they are even exempt from the
requirement of sending regular printed account
statements and often don’t collect accurate ad-dresses
(if clients even have these). Also, the cost
of a personal touch was considered prohibitive,
especially given the uncertain benefit.
Technology is changing this dynamic
in two ways:
• Using data analytics, banks can better identify
which customers to “touch” (for example,
those more likely to save, based on their previ-ous
transaction patterns) and how; and
• More than 90% of bank clients in GAFIS
surveys in South Africa, Colombia and Kenya
now have access to a mobile phone, which
enables SMS messaging (see Box F).
Box F: SMS as a new type of “touch”
Three GAFIS banks (Standard Bank, Bancolom-bia,
Equity Bank) are using SMS (text messages)
to communicate with their customers more
effectively. At Standard Bank and Bancolombia,
mobile numbers are automatically linked to the
GAFIS-supported savings account at opening. At
Equity, a mobile number is not required, but is
typically obtained. This is a big change from the

| 23
prevent or limit access. Another strategy is to
move into the mobile world: Bancolombia was
one of the first banks worldwide to acquire a
majority share of a mobile virtual network opera-tor
(MVNO), called Uff!, in 2012.14 An MVNO
does not have physical telco infrastructure but
rather leases access to the infrastructure of
others. Because the MVNO pays for wholesale
access to existing infrastructure, it may be able to
pass on better terms to its clients. However, even
MVNOs ultimately require fair access to existing
mobile infrastructure to succeed, so the role of
regulators remains important here, too.
Agents as a gateway for deposit taking
In 2010, regulators in most GAFIS countries had
only relatively recently authorized banks to use
agents to take deposits at all: this change
happened in 2006 in Colombia and India but
only in 2009 in Kenya and Mexico.
Recognizing the need to reduce costs of origina-tion
and cash handling, and empowered by the
new regulations, all the banks embarked on
ambitious plans to build new agent networks.
Collectively, these banks reported 2,600 agents
in 2010, but three years later now have nearly
25,000 (see Table 4). Four of the five GAFIS
banks already have substantially more agent
locations than branches.
Using agents, banks are also able to “sell” basic
accounts more than they could before. The agent
channel constitutes a low-cost account acquisi-tion
strategy, where, depending on the commis-sion
to the agent per account, agent-originated
accounts may cost a tenth to half of branch-orig-
Controlled experiments elsewhere by other
banks suggest that reminders can increase sav-ings
balances—although not by much in some
cases.13 The bigger effect may well come through
higher and sustained activation levels. It is too
early to judge the outcome fully at GAFIS banks,
although early results are promising.
While SMS communications for reminders and
balance enquiries show promise as part of an
activation process, mobile-initiated transactions
are still low at GAFIS banks relative to other
channels. Most are actively promoting mobile-enabled
accounts that allow the customer to pay
bills or make transfers to other mobile numbers.
However, most have experienced rising difficul-ties
with securing access to SMS or USSD mobile
channels. This is because, in some places,
banks have had limited access to these channels
because of competitive factors or because they
are not yet reliable; and in others, MNOs have
raised the cost of using these channels substan-tially.
Bancolombia, for example, faces a 600%
increase in the cost it has to pay to the domi-nant
MNO in Colombia (reflecting a shift from
fixed session costs to per SMS charges); ICICI
may see its charge at least doubled, as well as
a process challenge where the customer has to
pay for each message directly. Access to these 2G
channels will remain important for all financial
providers until most low-income customers have
smartphones with internet capability.
Banks can respond to these difficulties with
mobile channels in different ways. One is to
complain to the communications regulator that
dominant MNO positions are being abused to
13. Karlan 2013.
14. See Uff! website (in Spanish): www.uffmovil.com.

24 |
pared with branch-opened accounts. For Equity
Bank Ordinary Savings Accounts opened since
January 2011, when agent origination started,
the active ratio of agent-opened accounts is
around a quarter lower than the branch-opened
active ratio.
inated ones. However, in common with other
“sold” products like insurance, agent-originated
accounts also show lower activation rates,15
especially if the incentives of agents are based on
opening accounts rather than clients using them.
Standard Bank’s AccessSave accounts opened at
agents tend to be significantly less active com-
Table 4: Agent and cash merchant channel details and statistics
Bancolombia BANSEFI Equity Bank ICICI Bank Standard
Bank
Year of first
deploying agents 2006 2010 2010 2010 2009
Number of
approved agents 1,639 297 7,632 7,500 7,600
Number of agents
involved in origi-nating
accounts
None, but now
all can “assist
self-service”
OTA opening
of new ALM
account16
None All Most 100
Number of agents
providing cash
handling
In addition
to CICO, all
agents receive
cash for bill
payments
271 All All 7,500
Number of active
agents of all types 1,225 271 5,100 7,500 4,150
Source: GAFIS banks, 2013
Note: BANSEFI has many G2P payment points (several thousand), which service several million G2P accounts opened during the past few years, but these
payment points cannot take deposits and so are not counted here; and the new accounts serviced for payment distribution only are also not counted as part
of the GAFIS count of “new poor savers” (see Box I on page 36).
15. Meaning accounts into which a first deposit is made after opening.
16. Note: Bancolombia also has employees (called “PAMs”) that are similar to agents in that they rotate to multiple locations, including many CICO
agent locations, promoting products and originating accounts.

| 25
in agent transactions relative to the number of
agents (see Box H). For Equity, less than three
years after deploying agents, transactions
across agents exceeded the total number of
either ATM transactions or branch transactions
on a monthly basis. Deposits make up just over
half of agent transaction volume, but around
two-thirds of value transacted at agents; in other
words, agents have become a significant net
deposit-taking channel for Equity. Over an equiv-alent
two-year period to 2013, Bancolombia has
also seen the number of transactions at its agents
double, with agent deposits and withdrawals
virtually equal in value.
The convenience of agents can also encourage
more savings through deposits or retention. A
classic example of the agent in this role is the
susu deposit collector of West Africa, who liter-ally
makes daily rounds of clients to collect de-posits,
reducing transaction costs for clients but
also encouraging them to save.17 In the GAFIS
context, CASHPOR, an Indian MFI, has played a
similar deposit collector role (weekly) for ICICI’s
Apna savings accounts (see picture here and
Box G).
All GAFIS banks have taken a different approach
to developing agent networks over the past three
years. ICICI Bank and BANSEFI contracted with
intermediary institutions, such as CASHPOR or
FINO in India and DICONSA stores in Mexico, to
serve as agent network managers. This approach
enables banks to offer an agent channel with-out
acquiring or directly managing them, but it
can create vulnerability and dependence on the
network manager.
The other three GAFIS banks have sought to
build their own networks of agents, which takes
substantial resources and management skills.
Equity Bank has recorded the fastest uptake rates
17. For more background on the Ghanaian version of individual deposit collectors, see http://www.aaeafrica.org/start/the-susu-collection-system-in-
ghana-and-experiences-in-linking-formal-and-informal-financial-intermediaries/. See also the Care 2012 report Banking on Change which
describes linkages between a formal bank and village savings and loan associations.

26 |
As of October, 2013, around 150,000 (almost a
third) of CASHPOR’s credit clients had opened
ICICI savings accounts, with more than 110,000
of these active (with a positive balance). This is
a very high active ratio for India, where typically
less than 20% of “no-frills” (financial inclusion)
accounts are active. The bank charges customers
US$2 to open the account (the only product in
GAFIS to do this) as well as an annual fee (US$1)
for unlimited CICO transactions. Both fees likely
contribute to the high active ratio, and reduce the
chances of over-buying or over-selling.
The model mobilizes significant deposit activ-ity,
with an approximately 6:1 volume ratio of
deposits-to-withdrawals, although the average
withdrawal is much larger than the average depos-it.
ICICI also offers Fixed Deposit and Recurring
Deposit products through agents (though less than
5,000 accounts to date).
ICICI’s direct costs in this model come in the form
of revenue-sharing with CASHPOR, including
sharing margin on float balances (also unique
among GAFIS agent relationships), and per-trans-action
fees (also with ICICI’s Technology Service
Provider, EKO). These direct (mostly variable)
costs are covered by ICICI’s sharing of revenues.
Whether bottom-line profitability is achieved
depends on how the indirect (mostly fixed) costs
are allocated. For CASHPOR, the business case
appears adequate for now to sustain its agent role.
More importantly, as an NGO with a social mission
which cannot itself take savings, CASHPOR sees
that access to a savings account benefits its micro-credit
clients, and also helps them accumulate
money for loan-repayment purposes. This sense
of mission is tough to replicate.
Box G: ICICI Bank and CASHPOR: Mobilizing
savings through agents
To quickly and effectively establish its presence in
poor and rural areas of the State of Uttar Pradesh,
ICICI Bank partnered with CASHPOR, an estab-lished
micro-credit organization, which was also a
long-term commercial client of ICICI. CASHPOR’s
primary offering was micro-credit (not savings,
mainly as a result of regulatory restrictions) and
thus there was no inherent conflict in serving as an
agent for ICICI’s savings account. CASHPOR had
more than 500,000 rural clients, with roving loan
officers who could potentially double as ICICI
deposit-taking customer service points (CSPs).
While initially intending to help ICICI open a
remittance gateway, CASHPOR itself became a
“gateway” for new clients to the bank.
The CASHPOR micro-credit business follows a
traditional joint liability group lending model, in
which the loan officer visits multiple groups with
15-20 women clients each day. Each group meet-ing
lasts 45 minutes, with the first half dedicated to
micro-credit and second half to savings activities
(account openings, CICO and assistance with
balance inquiries).

| 27
Box H: Equity Bank: Rapidly scaling the
agent channel and agent transactions
Since early 2011, while the number of transac-tions
through Equity Bank branches and ATMs
has stayed relatively constant, the number of
agent transactions has increased dramatically.
It now exceeds the number through any other
single channel—and two thirds of the total
value of all agent transactions are deposits.
What has fuelled the rapid scale up at Equity?
Equity has benefited from the overall consumer
market in Kenya already being comfortable
with using agents and mobile phones for basic
financial transactions. However, chief among
the factors in the success of Equity’s agent
rollout is its agent selection process. The bank
has a history of serving small-business clients,
and branch managers were incentivized to
identify well-established businesspeople who
had repaid loans to the bank. These business
owners were selected as agents based on their
proximity to formal and informal markets, trans-portation
points and busy walkways. Later, other
agents were also selected based on their location
nearby the busiest branches.
Equity agents must set aside physical space near
or within the existing business. Preference is given
to those agents that can provide a modicum of pri-vacy
to customers; some agents have innovatively
used their marketing collateral to promote privacy.
Agents are also required to invest in marketing col-lateral,
POS machines and mobile phones to run
the business, demonstrating their commitment and
incentive to make the proposition work.
Today, more than 25% of the agents’ agent activi-ties
are profitable as stand-alone businesses. The
remainder are either near break-even, expect to be
profitable in the near future, or else consider the
increased foot traffic in their stores as a sufficient
reason to offer agency services. Over time, Equity
Bank has begun to implement a tiered franchise-type
system for managing agents. The most
successful agents identify and manage sub-agents,
which primarily conduct CICO services.

28 |
Effects on the business case:
activation and activity
The combination of agents in the field and SMS
messaging to the phone is changing both the
way banks originate accounts and the intensity
of their interaction with customers after account
opening. However, there are variations along the
spectrum of initial customer choice, which is
framed as “bought-sold-required” along one axis
in Figure 4; and with the other axis reflecting the
engagement after the account opening.
For “required” accounts, like the traditional In-dian
G2P no-frills accounts offered by ICICI and
other banks, ex post engagement by the bank
with the customer has traditionally been very
low, especially for those not receiving regular
G2P. BANSEFI’s full agents, open during shop
hours, have sought to offer higher touch, even
though, as Box E related, this alone has not been
enough to overcome other factors discouraging
savings and result in higher transactions
on the account.
On the other end of the spectrum, for “bought”
accounts such as ICICI’s Apna-CASHPOR prod-uct,
regular deposit collection in villages means
that the Bank remains in regular personal touch
through its agent. Bancolombia’s Ahorro a la
Mano account is also bought by customers who
open the account over-the-air on their mobile
phone and then deposit or withdraw cash via
agents or ATMs. The SMS campaigns described
earlier raise the ex post touch quotient for this
product, as they do for Standard Bank’s
AccessSave, which is more reliant on agents
to sell it.
Equity’s experience, alongside the microfinance
experience of CASHPOR for ICICI and the longer
experience of Bancolombia with cash-handling
agents, provides evidence that agents can be
part of a mass outreach strategy that includes
deposit mobilization. Agents can clearly play a
role in communities not previously served by
bank branches: many of Standard Bank’s Access
Points, for example, reach out into poorer peri-urban
communities.
A key business model question for banks is
whether the clients who use the agent channel
also use branches. If so, the business case based
on the cost reduction would be undermined to
that extent. One way to address this is to limit
the service channel at which new accounts can
be used (e.g., restricting branch deposits). But
even when the new clients can use any channel,
Equity’s experience so far suggests that a distinct
new segment of clients is emerging who mainly
or only use agents—of all Equity clients using
agents, Equity reports that around two thirds do
not use any other cash handling channel of
the bank.

| 29
Figure 4: Intensity of “touch” for bought, sold or required products
typical no frills accounts across banks in India.
However, we also observe instances of higher
dormancy resulting from agent origination.
Hence there is a need to follow more clients
using the new agent channels over a longer time
frame to understand better how the dormancy
profile changes over time, so we here assume no
net change in dormancy in calculating the shift
to Proposition 1.5.
These effects combine to improve the overall
account level proposition—losing just $1.02 per
month compared to $2.79 as before—but are
not sufficient in themselves to make this Propo-sition
1.5 profitable (see Figure 5). Proposition
1.5 alone is therefore not a final point, but rather
an interim step towards Proposition 2.0. This
will include more user-initiated transactions on
which fees are paid, as well as higher balances
and activity rates due to more flexible and useful
accounts. These further changes combine to
create at least a near break-even situation for the
account before any cross selling is contemplated.
Real profitability will depend on a bank’s ability
to understand its clients well enough to cross-sell
other products successfully.
How do Proposition 1.5 solutions using agents
change the business-case proposition from that
of Proposition 1.0 basic accounts? Since the
agent channels are relatively new, one cannot
yet be definitive about long-term effects on client
activity and usage profiles. However, indicative
evidence from across the GAFIS banks suggests
the following stylized facts so far:
Agents usually reduce opening costs relative
to branch-opened accounts (based on GAFIS
experience, we assume $6 rather than $15 in the
stylized model), although there may be a cor-responding
activation challenge as noted.
Agents reduce the cost of cash-handling transac-tions
to as low as a fifth of that in the branch,
although the fee structure to clients (whether
charges can be made for deposits) may still
mean that the bank cannot profit on cash-han-dling
alone.
The ex post touch, whether by SMS messaging
or by human agents, can improve activation
and usage rates—as shown in the rise in ac-counts
activated at Standard Bank following the
Project Activate SMS campaign, or in the more
than 70% active rate for ICICI’s Apna-CASHPOR
clients compared with below 20% or below on
Required
AccessSave

30 |
Figure 5: How Proposition 1.5 improves account-level profitability
dimensions: The relative value accumulated,
the period of accumulation (or duration), and
the client preference for liquidity or illiquidity.
The resulting three common patterns of savings
behavior (shown in their location across these
dimensions in Figure 6 below) can be character-ized
as:
Spend down slowly. Transactional balances
are retained for spending at a later point within a
relatively short spending cycle (such as a month)
and with a strong preference for liquidity;
Accumulate. Balances are built up over time in
small steps, often with illiquidity built in, such as
a requirement to regularly contribute and/or an
inability to access funds until a defined time
or goal;
Preserve. Balances that are relatively large are
held for a longer period with the emphasis on
preserving the lump sum until needed.
4. How the poor save, informally
and formally
Proposition 2.0 requires that banks tailor
products to match patterns of financial behav-ior
already exhibited by the client. This section
focuses on the learnings from GAFIS about how
to define and measure savings behavior, and
what this means for bank products in the context
of other products already in use.
For a relatively simple concept, savings is hard to
define. People save for a variety of reasons (such
as lifecycle, precautionary, income generating),
and use a variety of instruments. Clients may
use what banks call transactional (or debit card
or basic) accounts to accumulate balances over
time—that is, to save. And clients may never
accumulate any balances in what banks call sav-ings
accounts.
The portfolios of clients analyzed as part of
GAFIS demand-side research show at least three
types of savings behavior. These vary in three
CICO

Figure 6: Distinguishing 3 types of savings behavior in a portfolio
possible images with/without background
| 31
responds to transactional patterns based on
keeping cash in the home (under the mattress)
so that it is liquid and available for near-term
planned or unplanned expenditures. For a bank
account to accommodate this type of savings be-havior,
it must beat the convenience of the mat-tress,
even though it is already safer. Spend down
slowly bank savers may still withdraw all or most
available funds over a monthly or seasonal cycle,
but they do so in tranches so average balances
across the cycle are higher than the “dump &
pull” profile (withdrawal of full balance immedi-ately
upon arrival). This behavior type is a short
term form of savings, but relative to the dump
and pull pattern, reflects in higher average
account balances and more activity over time.
These observed patterns can all be carried out
using informal instruments which are widely
used in poor clients’ financial portfolios. For ex-ample,
poor clients in rural areas of developing
countries often use livestock, such as cows, as
instruments to preserve higher value, long-term
savings. They even have liquidity restrictions
since the cow must be sold to realize the savings.
While the informal instrument in this case also
has the advantage of an income flow (in the
form of milk production or offspring), and may
even hold its value or appreciate over time, the
cow could also die or be stolen, losing all value.
A formal financial institution wishing to attract
savings of this type literally has to “beat the cow”
as a value proposition for the low-income client.
Equally, “spend down slowly” behavior cor-

32 |
The role of the bank in the client savings portfolio
Some of the GAFIS products target one of these types of savings behavior (see Table 5), by offering a mix of incentives to
save and a range of associated liquidity (or illiquidity) options, while others allow customers to shape their own use of the
account.
Table 5: Bank account features & types of savings
Product features Savings behavior
targeted Savings incentives Liquidity restrictions
Ahorro a la Mano
(Bancolombia) Any
Offers free hospitalization
insurance if balance stays
above $175; tiered interest
ranging from 0.25 – 3.5%.
None: debits (withdrawals, inter-account
transfers and transfers/
payments to 3rd parties) can be
made anytime.
Apna account, Fixed
Deposit & Recurring
Deposit (ICICI Bank,
CASHPOR as agent)
Apna—Any, including
frequent incidence of
Accumulate or Preserve
Recurring deposit—
Accumulate
Fixed Deposit—Preserve
High interest offered on
all products, including the
basic Apna (4%) and much
higher (7%) for Fixed and
Recurring Deposits.
The basic Apna account relies on
weekly visits by MFI collector, a de
facto liquidity constraint in between
visits (can withdraw anytime from
nearest MFI branch, but less
convenient).
Fixed and Recurring Deposit
products offered, with minimum
thresholds lowered for rural poor.
AccessSave
(Standard Bank)
Any, esp. Accumulate or
Preserve
Bonus interest payment if
savings goal achieved.
Requires 7-day advance notice in
order to withdraw.
G2P recipient account
(BANSEFI) Spend down slowly Interest paid. None: debits (withdrawals) can be
made anytime.
Ordinary savings account
(Equity)
Any, esp. Spend down
slowly and Preserve
Interest paid only if balance
above $120.
None: debits (withdrawals, inter-account
transfers and transfers/
payments to 3rd parties) can be
made anytime.
Jijenge account
(Equity) Accumulate
Interest paid only with
balance above $120; bonus
interest if goal achieved.
Requires monthly commitment.
No withdrawal before the end of the
set term.
School Fees account
(Equity) Accumulate Interest paid on all
balances.
Four free withdrawals per year, high
charge thereafter.

| 33
While some account types seek to target one
specific type of savings behavior, open-ended
savings accounts can accommodate all three
types. Using transactional data from open-ended
accounts shows that basic bank accounts are in
fact used for all types of savings behavior—but
especially the “spend down slowly” and
“preserve” type. Regular accumulation behavior
is less common, since these accounts do not
offer strong incentives and penalties to reinforce
regular commitments in the way that accumulat-ing
savings groups do, for example.
How does the savings behavior shown in the
bank account fit into the wider portfolio of
instruments used by poor customers? Answer-ing
this requires the ability to link the profile of
the account to wider information obtained in
surveys of clients—using identifying numbers
that do not violate confidentiality requirements
around bank accounts. During the GAFIS project,
this intricate exercise was undertaken at one
bank in particular, yielding the average allocation
of all financial assets across six different savings
instruments, according to the savings behavior
type shown in the one bank account
(see Figure 7).
While the specific incentives or limitations may
be designed to support savings behavior, not all
are well understood by the clients. With Banco-lombia’s
new ALM product, for example, only
18% of the clients surveyed knew that there was
a free hospitalization insurance benefit if they
maintained a high enough balance. At Equity
Bank, the requirement to hold a minimum
balance of $120 to earn interest in the savings
account was not widely known. The seven-day
notice period for a withdrawal in Standard
Bank’s AccessSave was not universally under-stood,
as revealed by anecdotes of disappoint-ment
and surprise when clients were first denied
a withdrawal without giving the notice. We
noted earlier that, at GAFIS’ outset in 2011, most
of BANSEFI’s account holders did not realize the
account was a bank savings account.
Financial institutions that allow new clients to
test their products first can build up trust: for
example, by making balance inquiries cheap
and easy so clients can confirm their balances
frequently at first and understand the impact
of charges. This is a common pattern: Equity’s
Jijenge clients made frequent balance checks,
which are deliberately inexpensive, to help them
monitor balances and detect any surprises.
Figure 7: Allocation of financial assets held outside one GAFIS bank
Source: InFocus Note 3, BFA (2012)

34 |
the costs and intensity involved, financial diary
exercises are typically done on a relatively small
scale, requiring careful choice of participating
households to test different client profiles, as
well as caution in interpreting the results to apply
to a broader population. Financial diaries yield
indicative in-depth results but are usually not
statistically representative of the larger popula-tion
group. GAFIS undertook four financial diary
exercises to monitor bank client (and non-client)
portfolios over time (see Table 6).
Most of the GAFIS new product launches took
place relatively late in the project life (2012/3),
allowing limited time to track shifts in client port-folios.
Some diary exercises are still underway,
with additional co-funding to boost their scale,
and also extend their cycle beyond the limits of
the main GAFIS project. Despite these constraints
on results at this point, several observations have
emerged:
Several observations are apparent from
this exercise:
• First, even dormant customers may be custom-ers
of another bank, in which case the proposi-tion
to reactivate them is different than if they
had stopped using banks altogether;
• Second, clients who use their bank account
to spend down slowly seem to use the most
diverse portfolio of instruments over all;
• Third, informal savings clubs feature promi-nently
in all portfolios—even those using the
bank to preserve or accumulate, which was in
fact a small number in this sample.
Building such an integrated view of the client
from transactional “big data” as well as survey
data is still in its experimental stage with GAFIS
banks, but promises to yield valuable informa-tion.
It is possible that banks can link the ob-served
savings behavior in accounts with observ-able
client characteristics to improve targeting of
different types of customers. The account-level
profitability clearly varies with the balance and
activity level in the account, so that better iden-tifying
clients wishing to preserve usefully large
sums can become a profitable strategy to grow
a retail savings book; rather than cycling through
high transactors with low average balances.
How have the new GAFIS products changed
savings portfolios of the clients who use them?
Simple, one-off surveys rarely track financial
flows adequately or accurately enough to answer
this question. But the methodology of financial
diaries, popularized in the book, Portfolios of
the Poor, allows for repeated interaction with
the respondent over a prolonged period. Collect-ing
close to real-time data on flows helps build
an individual balance sheet profile. Because of

| 35
Diary
duration Sampling basis Date Diaries concluded
Purposive sampling across popula-tion
in 5 areas of country; aimed for
30 households with Jijenge users; 30
with School Fees users
Clients used Equity’s revamped Jijenge and
new School Fees accounts mainly because they
sought a separate place for savings oriented to-wards
a defined goal. For many, the appeal of the
bank account over the equivalent informal in-strument
(savings club) was its longer duration;
and compared to savings at home, bank saving
offered more privacy from family members
and friends.
As the GAFIS research shows, savings patterns
are diverse. Bank product design must either
target a specific savings behavior, or else be flex-ible
enough to allow clients to tailor their usage
of the account to meet different needs. The more
banks understand how savings products fit into
the wider portfolios of clients, the more they will
be entrusted with the portfolios of the poor.
Table 6: Financial Diaries studies of GAFIS bank clients
Location
Diary
sample
size
Kenya18 300 12 months
ICICI’s Apna clients started to save regularly in
their new accounts, although in small amounts.
After a year or so with an account, the ICICI
bank account held around 5% of the household’s
total financial assets (and 9% of the account
holder’s individual assets)—an important step
towards diversifying their portfolios from reliance
on informal savings products.
On the other hand, some of Standard Bank’s
new AccessSave clients saved as much as 50%
of their total financial assets in the new account.
Some of these clients previously kept the money
“under the mattress” in their homes. However,
they recognized that this was risky and that
AccessSave offered safety, convenience and
discipline, and appear to have shifted a portion
of their ‘home bank’ balances into the bank
as a result.
November 2013
India 89 7 months
Purposive, as follows: At project
inception, 60 held Apna account and
CASHPOR loan, 15 held a CASHPOR
loan only (no Apna), and 15 had
neither.
July 2013
Mexico19 180 10+ months
Sample includes 120 HHs who re-ceive
Oportunidades and 60 that are
poor but do not receive the ben-efit
for comparison in Oaxaca and
Puebla states and Mexico City.
December 2014
South
Africa 67 8 months
Purposive sampling of “Access”
target market, which by end of study
included 44 with at least 1 “Access”
product, including 14 with Access-
Save.
June 2013
18. The Kenyan financial diaries were co-funded by FSDK and the full results will be available via FSDK in 2014.
19. The Mexican financial diaries were co-funded to create a much larger sample started only in late 2013 so will yield insights only in 2014 outside of
the project timeframe.

36 |
Conclusion:
The journey to date and the way
ahead
The changes within GAFIS partner banks dur-ing
the project period (2010-2013) have been
profound. All five banks have made progress by
designing and launching appropriate new prod-ucts
and channels to serve low-income clients.
All make more use of information from diverse
sources to understand their client base and,
increasingly, to recognize segments within it.
All banks have built new agent channels and, in
some cases, they already contribute substantially
to the business.
Collectively, the GAFIS banks have opened more
than 4.2 million new accounts in GAFIS-linked
products in less than three years. More impor-tantly,
they now serve approximately 420,000
“new, poor savers” as a result of the project (see
Box I). This number is measured according to
the GAFIS project definition, which requires evi-dence
of both savings activity (870,238 accounts)
and the poverty status of the account holders
(543,119 new accounts meet both criteria); and
which also weights the level of attribution to the
project according to how directly GAFIS was
involved (reducing the 543,119 to 419,654). Even
with delays in launching new products at several
banks, these growing numbers provide early
indication that large banks can achieve scale
outreach with their new propositions.
Box I: The GAFIS project scorecard:
Desired outcomes and results
Desired Outcomes GAFIS Results
Significant learning dissemi-nated
to the sector about the
business case for delivering
useful savings for the poor,
particularly related to products
linked to identified gateways.
Four Focus Notes, as well as
blogs, local articles and numer-ous
talks. This report is itself
a key part of making findings
accessible to a wider audience.
At least two large formal
financial institutions have been
encouraged and assisted to
develop viable savings prod-ucts
for the poor.
Standard Bank, Bancolombia
and Equity Bank have launched
new savings products, while
the others have all taken steps
toward more viability of exist-ing
products. As important,
several have restructured their
retail banks to a position to
deliver inclusive banking well
into the future.
500,000 “new, poor savers” use
GAFIS savings products.
During the project, banks
opened more than 4.2 million
new accounts of types linked
to GAFIS, of which 870,238 are
“savers”. Of these, 543,119 are
also poor, while 419,654 are
“new, poor savers” attributable
to GAFIS.20
20. As of 31 October 2013.

Big banks-and-small-savers-gafis-project-report-dec2013

Big banks-and-small-savers-gafis-project-report-dec2013

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Viewers also liked

Viewers also liked (12)

Similar to Big banks-and-small-savers-gafis-project-report-dec2013

Similar to Big banks-and-small-savers-gafis-project-report-dec2013 (20)

Big banks-and-small-savers-gafis-project-report-dec2013