2013a klapper

asli demi˙rgüç-kunt
World Bank
leora klapper
World Bank
Measuring Financial Inclusion: Explaining
Variation in Use of Financial Services
across and within Countries
ABSTRACT This paper summarizes the first publicly available, user-side
data set of indicators that measure how adults in 148 countries save, borrow,
make payments, and manage risk. We use the data to benchmark financial
inclusion—the share of the population that uses formal financial services—in
countries around the world, and to investigate the significant country- and
individual-level variation in how adults use formal and informal financial
systems to manage their day-to-day finances and plan for the future. The
data show that 50 percent of adults worldwide are “banked,” that is, have
an account at a formal financial institution, but also that account penetration
varies across countries by level of economic development and across income
groups within countries. For the half of all adults around the world who
remain unbanked, the paper documents reported barriers to account use, such
as cost, distance, and documentation requirements, which may shed light on
potential market failures and provide guidance to policymakers in shaping
financial inclusion policies.
Well-functioning financial systems serve a vital purpose, offering
savings, payment, credit, and risk management products to people
with a range of needs. More-inclusive financial systems—those that allow
broad access to appropriate financial services—are likely to benefit poor
people and other disadvantaged groups. For instance, access to formal sav-ings
and credit mechanisms may facilitate investment in productive activi-ties
such as education or entrepreneurship. Lacking such access, individuals
rely on their own limited, informal savings to invest in their education or
279

280 Brookings Papers on Economic Activity, Spring 2013
become entrepreneurs, and small enterprises on their limited earnings to
take advantage of promising growth opportunities. This can contribute to
persistent income inequality and slower economic growth.1
This paper benchmarks financial inclusion and explores country- and
individual-level variation in how adults around the world use formal and
informal financial products to manage their finances and plan for the future.
We define financial inclusion as the use of formal financial services, and
we investigate how patterns of financial inclusion vary across countries at
different levels of income per capita, and within countries at different lev-els
of relative income. Next, we examine the barriers to financial inclusion
and document the relationship between subjective and objective barriers to
access. Finally, we discuss examples of public and private sector–led initia-tives
in this realm and how better data can inform policymakers in shaping
financial inclusion policies. Although the literature and the data provide
suggestive evidence of market failures and of potential welfare gains from
greater financial inclusion, we emphasize that the role of our data is to help
policymakers better understand the existence of these failures, rather than
to advocate specific policy interventions.
Our paper contributes to a growing literature examining household finance
and, especially, the borrowing and saving decisions of households.2 Quali-tative
evidence from financial diaries demonstrates that poor people juggle
complex financial transactions every day and use sophisticated techniques
to manage their finances, irrespective of whether they use formal financial
instruments (Collins and others 2009). Evidence from field experiments
highlights that people with access to savings accounts or simple informal
savings technologies are more likely to increase consumption, productiv-ity
and income, and investment in preventive health, and to have reduced
vulnerability to illness and other unexpected events (Dupas and Robinson
2009, 2011, Ashraf and others 2010). Yet the evidence from field experi-ments
that increase access to microcredit shows more modest effects in
promoting investment and entrepreneurship, mostly for households with
existing businesses (Banerjee and others 2010, Karlan and Zinman 2010).
Until now, little was known about the global reach of the financial
sector—the extent of financial inclusion worldwide and the degree to which
1. See, for example, King and Levine (1993), Beck, Demirgüç-Kunt, and Levine (2007),
Beck, Levine, and Loayza (2000), Demirgüç-Kunt and Levine (2009), Klapper, Laeven, and
Rajan (2006), and World Bank (2008a).
2. For a detailed literature review see World Bank (2008a) and the references therein.
Campbell (2006) also provides an overview of the household finance field.

asli demi˙rgüç-kunt and leora klapper 281
groups such as the poor are excluded from formal financial systems. Sys-tematic
indicators on the use of different financial services were lacking for
most countries. The Global Financial Inclusion (“Global Findex”) database
provides such indicators, measuring how adults in 148 countries around the
world manage their day-to-day finances and plan for the future. The indi-cators
are constructed using survey data from interviews with more than
150,000 nationally representative and randomly selected adults over the
2011 calendar year.3 The individual-level data are publicly available online
and include over 40 indicators related to account ownership, payments,
saving, borrowing, and risk management.4
Consistent with previous findings, the Global Findex data show that the
vast majority of adults actively use financial products, formal or informal,
to manage their finances and plan for the future. We find that 75 percent
of adults worldwide use at least one of the financial management tools
included in the Global Findex survey, and half of all adults report having
an individual or a joint account at a formal financial institution. These
accounts are used for a wide range of purposes including receipt of wage
payments, government transfers, and remittances from family members
living elsewhere.
At the country level, the Global Findex data show sharp disparities in
the use of financial services between high-income and developing coun-tries,
confirming the findings of previous studies that show lower use of
formal financial services in developing countries (see, for example, Beck,
Demirgüç-Kunt, and Martinez Peria 2007 and Cull, Demirgüç-Kunt, and
Morduch 2013). For instance, the share of adults in high-income coun-tries
who are “banked” (have an account at a formal financial institution)
is more than twice that in developing countries.
At the individual level, the data also show significant variation in finan-cial
inclusion within countries across individual characteristics such as
income. Around the world, wealthier adults tend to make greater use of
formal financial services, even after one controls for other individual char-acteristics
and country fixed effects. For instance, in developing countries
as a group, adults in the highest 20 percent of income earners are more than
twice as likely to have an account as those in the lowest 20 percent.
3. The Bill & Melinda Gates Foundation funded three triennial rounds of data collection
through the complete questionnaire. The next data collection will be in 2014.
4. The database and the full questionnaire are available at www.worldbank.org/global
findex. Appendix A reproduces selected questions relevant to this paper. The questionnaire is
also available in 15 languages at go.worldbank.org/5XL9LXK6B0.

The Global Findex data set also includes novel cross-country data on
self-reported reasons for not having a formal account, making it possible to
identify barriers to financial inclusion. Moreover, the ability to disaggregate
the data by individual characteristics allows researchers and policymakers
to identify population groups that are excluded from the formal financial
system and to better understand what characteristics are associated with cer-tain
financial behaviors.
Worldwide, by far the most common reason for not having a formal
account, cited as the only reason by 30 percent of non–account holders,
is lack of enough money to use one. This speaks to the fact that having a
formal account is not costless in most parts of the world and that individu-als
with small or irregular income streams might view an account as an
unnecessary expense, given the relatively high cost. Other reasons com-monly
reported for not having an account are that banks or accounts are too
expensive (cited by 25 percent of adults without a formal account) and that
the nearest banks are too far away (cited by 20 percent).
We examine the percentage of adults who saved, in the sense of delib-erately
putting aside money for future use, in the past year, and find that
most saving in developing countries is done informally, even among adults
who have a formal account. Worldwide, 36 percent of adults report having
saved in the past year. Twenty-two percent of adults who reported saving
(formally or informally) said they did so using a formal financial institution
in the past 12 months. We also discuss informal saving and differences in
the mode of saving across different income groups. In developing countries,
for instance, 12 percent of account holders save using informal methods.
The use of informal, community-based saving methods (such as rotating
savings clubs) is also widespread, particularly in sub-Saharan African coun-tries
such as Cameroon, Kenya, and Nigeria.
We also find that most borrowing by adults in developing countries is
from informal sources. Globally, 9 percent of adults report having origi-nated
a new loan from a formal financial institution in the past 12 months,
while 23 percent report borrowing from family and friends. But in devel-oping
countries, adults are three times as likely to borrow from family and
friends as from formal financial institutions (25 percent and 8 percent,
respectively). In high-income countries, the most commonly cited purpose
of an outstanding loan is to purchase a home; emergency and health rea-sons
are those most frequently cited by adults in the developing world.
Finally, the Global Findex data set also provides new insight into the
results of recent initiatives to expand financial inclusion. For instance, in
Kenya 68 percent of adults in our sample report having used a mobile phone

in the past 12 months to pay bills or to send or receive money; of these, almost
two-thirds (41 percent of all adults) are otherwise unbanked. The spread of
mobile money products, the increasing proliferation of bank agents, and the
increasing movement toward dispensing government payments via formal
accounts all offer potential to significantly alter the ways in which people
manage their finances. Future rounds of data will allow us to document the
pace of change in these behaviors.
The rest of the paper proceeds as follows. Section I defines and sum-marizes
our financial inclusion indicators. Section II documents across-and
within-country variation in the use of formal and informal financial
services. Section III discusses self-reported barriers to financial inclu-sion.
Section IV discusses recent initiatives to expand financial inclusion,
and section V concludes.
I. Indicators and Methodology
The Global Findex indicators measure the use of financial services, which
is distinct from access to financial services. Access most often refers to the
supply of services, whereas use is determined by demand as well as supply
factors (World Bank 2008a). The Global Findex data can shed light on the
levels and patterns of use of different financial services both globally and
among different groups, such as poor people, youth, and women. But one
cannot assume that all those who do not use formal financial services are
somehow constrained from participating in the formal financial sector—
access and use are not the same thing. The role of policy is to broaden finan-cial
inclusion to reach those who are excluded because of market failures.
I.A. Indicators
The first set of indicators focuses on the ownership and use of an account
at a formal financial institution. For most people a formal account serves
as an entry point into the formal financial sector. Having a formal account
facilitates the transfer of wages, remittances, and government payments. It
can also encourage formal saving and open access to credit. Accounts are
also a simple and consistent metric that facilitates the measurement of finan-cial
inclusion across countries. Ownership and use of accounts are relatively
easy to define and observe, and basic checking and savings accounts are
fairly similar across countries.
The Global Findex survey includes several questions about accounts
that investigate the mechanics of their use (frequency of use, mode of access),
their purpose (receipt of payments from work, government, or family),

barriers to their use, and alternatives to formal accounts (mobile money).
Importantly, the survey’s account penetration indicator measures the
percentage of adults who have individual or joint ownership of a formal
account, defined as an account at a formal financial institution such as a
bank, credit union, cooperative, post office, or microfinance institution. It
includes those who report having a debit or ATM card tied to an account.
The second set of indicators focuses on saving behavior. Savings allow
individuals to smooth consumption, make large investments in education
or to start a business, and mitigate uncertainty and risk. The concept of sav-ing
is inherently more subjective than those of account ownership and use.
Individuals and cultures may have varying definitions of what constitutes
saving. We focus on the purposeful action of saving, surveyed by asking
individuals whether they have “saved or put aside any money” in the past
year. We collect data on general saving behavior, as well as on the use of
formal accounts and community-based methods to save. In doing so, we
highlight the distinction between deliberate saving, whether formal or not,
and the case where individuals simply consume less than their income.
Individuals may save in the latter case as well (perhaps using informal
means such as putting money under a mattress), but we are particularly
interested in the use of formal accounts for saving.
The third set of indicators focuses on borrowing. Most people need to
borrow money from time to time. They may want to buy or renovate a
house, to invest in education, or to pay for a wedding or a funeral. When
they lack enough money to do so, they turn to someone who will lend it to
them: a bank, a cousin, or an informal lender. In some parts of the world
many people rely on credit cards to obtain short-term credit. We gather
data on the sources of borrowing (formal and informal), the purposes of
borrowing (mortgage, emergency or health purposes, and the like), and
the use of credit cards.5
5. In a few instances, surveyors and their supervisors reported that respondents were
somewhat taken aback at the series of questions, given the personal nature of the topic. This
concern was particularly relevant in countries with large security risks, such as Mexico and
Zimbabwe, and in countries where personal finance is widely regarded as a private matter,
such as Cameroon, Italy, and Portugal. There were also reports from the field that the termi-nology
and concepts used in the survey were entirely new to some respondents. Although
efforts were made to include simple definitions of such terms as “account” and “debit card,”
the unfamiliarity and complexity of the topic were still reported to be a hurdle in several
countries, including Afghanistan, Cambodia, Chad, and rural Ukraine. Overall, however, the
rate of “don’t know” or “refuse” answers was very low. For the core questions (those not
conditioned on the response to other questions), “don’t know” or “refuse” responses made up
fewer than 1 percent of the total and no more than 2 percent in any world region.

I.B. Data Coverage
The Global Findex indicators are drawn from survey data collected over
the 2011 calendar year, covering more than 150,000 adults in 148 countries
that represent approximately 97 percent of the world’s population. The sur-vey
was carried out by Gallup, Inc., in association with its annual Gallup
World Poll. The Gallup World Poll has been used in previous academic
studies, mostly to study well-being and social capital. For example, Angus
Deaton (2008) uses Gallup World Poll questions on life and health satis-faction
and looks at the relationships with national income, age, and life
expectancy. Gallup World Poll questions are also used by Betsey Stevenson
and Justin Wolfers (2008) and by Daniel Sacks, Stevenson, and Wolfers
(2010) as part of their research to analyze relationships between subjective
well-being and income; by Bianca Clausen, Aart Kraay, and Zsolt Nyiri
(2011) to analyze the relationship between corruption and confidence in
public institutions; by Demirgüç-Kunt and others (2013) to study changes
in trust in banks in the wake of the global financial crisis; and by Stevenson
and Wolfers (2011) to examine trust in institutions over the business cycle.
As part of the World Poll, since 2005 Gallup has surveyed about 1,000
people annually in each of up to 157 countries,6 using randomly selected,
nationally representative samples.7 The target population is the entire
civilian, noninstitutionalized population aged 15 and above. Surveys are
conducted in the major languages of each country.8
Although the results obtained using the Global Findex data are broadly
consistent with those of earlier efforts, they differ in some nontrivial ways.
Three key differences between the Global Findex and other user-side data
involve the definition of an account and its use, the units of measurement
6. The worldwide aggregates omit countries for which Gallup excludes more than 20 per-cent
of the population in the sampling either because of security risks or because the popula-tion
includes non-Arab expatriates. These excluded countries are Algeria, Bahrain, the Central
African Republic, Madagascar, Qatar, Somalia, and the United Arab Emirates. Iran is also
excluded because the data were collected in that country using a methodology inconsistent
with that used for other countries (the survey was carried out by phone from Turkey). The
exclusion of Iran has a nontrivial effect on regional aggregates because its population is larger
and wealthier than the populations of most other countries in the Middle East and North
Africa. For example, account penetration in the region is estimated to be 18 percent when Iran
is excluded, but 33 percent when it is included.
7. In some countries oversamples are collected in major cities or areas of special inter-est.
In addition, in some large countries, such as China and Russia, sample sizes of at least
4,000 are collected.
8. For details on the data collection dates, sample sizes, excluded populations, and mar-gins
of error, see www.worldbank.org/globalfindex.

such as age cutoffs, and when the data were collected. Relative to other
demand-side data efforts, one significant advantage of the Global Findex
data is that they are consistent and comparable across countries. Two com-monly
cited cross-country user-side data collection efforts are the FinMark
Trust’s FinScope initiative, a specialized household survey in 14 African
countries and Pakistan, and the European Bank for Reconstruction and
Development’s Life in Transition Survey (LITS), which covers 35 coun-tries
in Central and Eastern Europe and Central Asia and includes several
questions on financial decisions as part of a broader survey. The Global
Findex country-level estimates of account penetration are generally insigni-ficantly
different from or higher than those of the FinScope surveys, perhaps
because of the difference in timing (most of the FinScope surveys were car-ried
out in the mid-2000s) and in the definition of an account (Global Findex
includes only accounts that can be used for both deposits and withdrawals).
The Global Findex estimates of account penetration are within 7 percentage
points of the LITS estimates for the majority of countries; the discrepancies
are perhaps explained by the fact that the LITS financial access question
is less descriptive than the corresponding questions in the Global Findex
survey.9 Compared with data collected from the providers of financial ser-vices
(financial institutions), the Global Findex data may fill a gap by going
beyond data collected only from regulated financial institutions and allow-ing
disaggregation of the data by demographic characteristics.10
I.C. Survey Methodology
The survey methodology for the Global Findex data is that used for the
Gallup World Poll. Surveys are conducted by telephone except in countries
where telephone coverage represents less than 80 percent of the popula-
9. The exact question in the LITS survey is “Does anyone in your household have a
bank account?”
10. On the provider side, the International Monetary Fund collects indicators of finan-cial
outreach such as the number of bank branches and automated teller machines (ATMs)
per capita and per square kilometer, as well as the number of loan and deposit accounts
per capita, directly from country regulators. These data sets are important sources of basic
cross-country indicators developed at a relatively low cost. Yet indicators based on data
collected from financial service providers have several important limitations. First, data are
collected from regulated financial institutions only and thus provide a fragmented view of
financial access. Second, aggregation can be misleading because of multiple accounts or
dormant accounts (see Beck, Demirgüç-Kunt, and Martinez Peria 2008 for a discussion).
Most important, this approach does not allow disaggregation of financial service users by
income or other characteristics. That leaves policymakers unable to identify those segments
of the population with the lowest use of financial services, such as the poor, women, or youth.

tion; in these countries the survey is conducted face to face.11 In most coun-tries
the fieldwork is completed in 2 to 4 weeks. In countries where Gallup
has conducted face-to-face surveys, the identification of primary sampling
units, consisting of clusters of households, constitutes the first stage of
sampling. The primary sampling units are stratified by population size,
geography, or both, and clustering is achieved through one or more stages
of sampling. Where population information is available, sample selection
is based on probabilities proportional to population size; where otherwise,
simple random sampling is used. Random route procedures are used to
select sampled households. Unless an outright refusal occurs, interviewers
make up to three attempts to survey the sampled household. If an interview
cannot be obtained at the initial sampled household, a simple substitution
method is used. Respondents are randomly selected within the selected
households by means of the Kish grid.12
In countries where telephone interviewing is employed, random-digit
dialing or a nationally representative list of phone numbers is used. In
selected countries where cell phone penetration is high, a dual sampling
frame is used. Random respondent selection is achieved by using either the
latest-birthday method or the Kish grid method.13 At least three attempts
are made to reach a person in each household, spread over different days
and times of day.
I.D. Data Weighting
Data weighting is used to ensure a nationally representative sample
for each country. First, base sampling weights are constructed to account
for oversamples and household size. If an oversample has been conducted,
the data are weighted to correct the disproportionate sample. Weighting by
11. Croatia, the Czech Republic, Estonia, Greece, Hungary, Poland, Singapore, and the
Slovak Republic are the only high-income countries included where phone coverage is less
than 80 percent.
12. The Kish grid is a table of numbers used to select the interviewee in each household.
First, the interviewer records the name, sex, and age of all permanent household members
aged 15 and above, whether or not they are present, and then numbers them starting with the
oldest and ending with the youngest. Second, the interviewer finds the column number of the
Kish grid that corresponds to the last digit of the questionnaire number, and the row number
for the number of eligible household members. The number in the cell where column and
row intersect determines the person selected for the interview. In countries where cultural
restrictions dictate matching interviewer and interviewee by sex, respondents are randomly
selected using the Kish grid from among all eligible adults of the interviewer’s sex.
13. In the latest-birthday method an interview is attempted with the adult in the house-hold
who had the most recent birthday.

household size (number of residents aged 15 and above) is used to adjust
for the probability of selection, as residents in large households will have
a disproportionately lower probability of being selected for the sample.
Second, poststratification weights are constructed. Population statistics are
used to weight the data by sex, age, and (where reliable data are avail-able)
education or socioeconomic status. Finally, approximate study design
effects and margins of error are calculated. The average country-level
margin of error for the account penetration indicator is ±3.9 percent. All
income group aggregates are also weighted by country population (aged 15
and above).
II. Individual- and Country-Level Variation in Financial Inclusion
In this section we discuss the main findings from our analysis of the Global
Findex database to highlight broad patterns in financial inclusion across
the globe. We focus on several indicators that we believe are particularly
important for understanding the financial behavior of adults, including the
ownership and use of formal accounts, the prevalence of formal and infor-mal
saving behavior, and the sources and purposes of borrowing. We first
examine country-level variation in account penetration across countries
and regions. Next, we focus on differences in the use of financial products
across individuals, and how disparities by individual characteristics vary
across countries. We also identify trends in account ownership such as fre-quency
and mode of use. We then discuss saving behavior. In particular, we
identify trends in the use of formal and informal methods of saving across
countries and across income groups within countries. Finally, we highlight
patterns in access to and sources of credit worldwide.
The variation in the data—pertaining to accounts, saving, and credit—
highlights differences in countries’ levels of financial inclusion. It also
emphasizes that the nature of the use of financial services, such as frequency
of account use or purpose of obtaining credit, not only varies across countries
but may be widely divergent within any given country as well. By focus-ing
on both within- and cross-country inequality, we identify patterns in the
data that may be useful to governments in informing their financial inclusion
strategies, and to private sector actors in new product development.
II.A. Accounts and Payments
explaining variation in account penetration Account penetration dif-fers
enormously between high-income and developing countries in the aggre-gate:
89 percent of adults in high-income countries, but only 24 percent in

Figure 1. Formal Account Penetration, by Country Income Group
Low
Lower-middle
Upper-middle
High
24%
20% 40% 60% 80%
low-income countries, report that they have an account at a formal financial
institution (figure 1). Globally, 50 percent of the world’s population—more
than 2.5 billion adults—do not have a formal account (figure 2). The major-ity
of these reside in developing countries.14 In several countries around the
world—including Cambodia, the Democratic Republic of Congo, Guinea, the
Kyrgyz Republic, Turkmenistan, and Yemen—more than 95 percent of adults
lack a formal account. Appendix B reports the percentage of adults with a
formal account in each country surveyed.
Why is account penetration high in, say, Denmark but almost negligi-ble
in Niger? GDP per capita accounts for much of the variation across
countries (top panel of figure 3): Denmark is among the world’s richest
countries whereas Niger is among the poorest. In most countries with a
GDP per capita of $15,000 or higher, account penetration is 90 percent or
higher.15 Indeed, regression analysis shows that national income per capita
14. According to the latest available data from the World Bank’s World Development
Indicators, there are 5.08 billion adults aged 15 and above worldwide.
15. Exceptions include Italy (with an account penetration of 71 percent) and the United
States (88 percent).
Adults with an account at a formal
financial institution (percent)
Country income groupa
Source: Authors’ calculations using Global Findex data.
a. Low-income countries are those with gross national income per capita less than $1,025 in 2011;
lower-middle-income countries, $1,026–$4,035; upper-middle-income countries, $4,036–$12,475; high-income
countries, $12,476 or more.
89%
57%
29%

Figure 2. Formal Account Penetration Worldwide
Percent of adults
with an account at
a formal financial
institution
0–15
16–30
31–50
51–80
81+
No data
is significantly associated with account penetration and accounts for about
70 percent of the variation among the world’s countries in the share of
adults with a formal account (column 1-1 in table 1). Country-level regres-sions
also show that whereas adults in low-income countries are 72 percent
less likely to have an account than adults in high-income countries, adults in
upper-middle-income countries are only 43 percent less likely (column 1-2).
We find significant difference in account penetration between adults in low-income
countries and in lower-middle-income countries and a significant
gap between adults in lower-middle-income and in upper-middle-income
countries (column 1-2, bottom rows).
National-level financial development, as measured by domestic credit
to the private sector as a percentage of GDP, is also significantly associated
with account penetration (bottom panel of figure 3), even when one con-trols
for GDP per capita (column 1-3 in table 1). However, large amounts
of credit—whether commercial or consumer credit—in a financial sys-tem
do not always correspond to broad use of financial services, because
credit can be concentrated among the largest firms and the wealthiest
individuals. For instance, domestic credit to the private sector amounts
to 112 percent of GDP in Vietnam, but only 21 percent of adults in that
country report having a formal account. Conversely, in the Czech Republic,
a country with relatively modest financial depth (domestic credit to the
private sector is 55 percent of GDP), account penetration is relatively high
(81 percent).

Figure 3. Formal Account Penetration, GDP per Capita, and Financial Developmenta
Account penetration and GDP per capita
Percent of adults with
a formal account
10 20 30 40 50
GDP per capita (thousands of 2000 dollars)
Account penetration and financial development
80
60
40
20
a formal account
80
60
40
20
40 80 120 160 200
Domestic credit to the private sector (percent of GDP)
Source: Authors’ calculations using Global Findex and World Development Indicators data.
a. Each observation represents 1 of 140 (top panel) or 130 (bottom panel) developing and high-income
countries.

Table 1. Country-Level Regressions Explaining Financial Inclusiona
Dependent variable
Independent
variable
Percent of adults reporting having an account at a
formal financial institution
Percent
reporting
formal
savingb
1-5
Percent
reporting
use of
formal
creditc
1-1 1-2 1-3 1-4 1-6
Logarithm of
GDP per
capitad
0.105***
(0.014)
0.122***
(0.014)
0.157***
(0.012)
Low-income
country (1)e
-0.436***
(0.106)
-0.316***
(0.027)
-0.063***
(0.014)
Lower-middle-income
country (2)e
-0.442***
(0.079)
-0.313***
(0.025)
-0.052***
(0.013)
Upper-middle-income
country (3)e
-0.325***
(0.053)
-0.276***
(0.024)
-0.039***
(0.013)
Domestic
credit to the
private sec-tor
(percent
of GDP)d
0.185***
(0.037)
Gini indexd -0.488**
(0.190)
Constant 0.203
(0.241)
-0.606***
(0.093)
-0.582***
(0.126)
0.407***
(0.017)
0.135***
(0.009)
No. of
observations
(countries)
134 134 123 110 139 139
R2 0.772 0.783 0.747 0.662 0.622 0.160
p value of
F statistic
H0: (1) = (2) 0.904 0.931 0.443
H0: (2) = (3) 0.015 0.127 0.310
Source: Authors’ regressions using Global Findex and World Development Indicators data.
a. Each column reports results of a single ordinary least squares regression. Standard errors are in parentheses.
Asterisks denote significance at the ***1 percent, **5 percent, and *10 percent level.
b. Percent of adults reporting having saved or put aside money at a formal financial institution in the past
12 months.
c. Percent of adults reporting having borrowed from a formal financial institution in the past 12 months.
d. Data are for 2011 or the most recent year available.
e. Dummy variable equal to 1 when the country is a member of the indicated country income group, and zero
otherwise. High-income countries are the omitted category.

Figure 4. Formal Account Penetration, by Country Income Group and Within-Country
Income Quintile
a formal account
90
80
70
60
50
40
30
20
10
Bottom quintile
Top quintile
High-income countries
Upper-middle-income countries
Lower-middle-income countries
Low-income countries
7 8 9 10 11
Logarithm of median income in dollars
These findings suggest that financial depth and financial inclusion are
related but ultimately distinct dimensions of financial development, and that
financial systems can become deep without delivering access for all. A more
formal investigation of the country-level determinants of financial inclusion
is beyond the scope of this paper, but the theme is explored further using
Global Findex data by Franklin Allen and others (2012).
account penetration across individual characteristics Beyond cross-country
variation, there is also significant variation in account penetration
across individuals within a given country. Examining account penetration
by within-country income quintile highlights differences between the poor
and the better off. The differences in slope from one segment to the next
in each of the lines in figure 4 indicate the differences in account penetra-tion
between income quintiles for the country income group represented
by that line—a rough measure of the gap in financial inclusion between
richer and poorer individuals at a given level of country income per capita.
Because the upper limit is 100 percent, there is little absolute difference
in the slopes between the dots for the high-income countries as a group. In
these countries, on average, poorer adults are not significantly less likely
than richer adults to have a formal account. But stark differences exist in
account penetration within most developing countries. In the upper-middle-

income countries the slope of the line is very steep, but relatively constant
across segments. The richest adults in these countries are more than twice as
likely as the poorest to have a formal account, with a gap of approximately
10 percentage points separating each pair of quintiles. The lower-middle-income
countries exhibit sharp differences between the poorest and the
middle class, as well as between the middle class and the rich, highlighted
by the kinks in the curve. In the low-income countries, account ownership
does not vary significantly across the bottom two income quintiles, but it
increases steadily as income increases further.
Two other results in figure 4 are striking. First, account penetration in the
poorest quintile in the high-income countries is 9 percentage points higher
on average than in the richest quintile in the upper-middle-income countries.
Second, account penetration in the richest quintile in the low-income coun-tries
is only 4 percentage points higher than in the poorest quintile in the
upper-middle-income countries.
We also estimate multivariate probit models using individual-level data
to test the relationship between account ownership and income quintile,
controlling for other individual characteristics such as sex, age, education,
marital status, household size, employment, and rural versus urban resi-dence.
The leftmost panel of table 2 reports marginal effects for the bottom
four income quintiles (the richest income quintile is the excluded category),
which show significant differences in within-country “financial inequality”
across country income groups. Although in all country income groups, adults
in the highest income quintile are significantly more likely to be banked, in
the high-income countries that difference is small: the poorest 20 percent
are only 5 percent less likely to have an account than the richest 20 per-cent,
whereas in the upper-middle-income countries the poorest 20 percent
of earners are 24 percent less likely, and in low-income countries the poorest
earners are 13 percent less likely.
These findings may be explained in part by differences in economic
inequality across country income groups. Indeed, we find a strong correla-tion
across countries (a correlation coefficient of 0.42) between inequal-ity
in the use of formal accounts and income inequality as measured by
the Gini coefficient (with higher values indicating a more unequal income
distribution). The correlation between these two measures of financial and
economic inequality continues to hold even when we control for national
income per capita (column 1-4 of table 1).
Consider the example of the United Kingdom and the United States.
These two countries have relatively similar GDP per capita and relatively
similar account penetration among adults in the top four income quintiles

Table 2. Individual-Level Probit Regressions Explaining Financial Inclusiona
Dependent variable and country income group
Has an account at a formal financial institution
Saved at a formal financial institution within last
12 months
Borrowed from a formal financial institution
within last 12 months
Within-country
income quintile Low
Lower-middle
Upper-middle
High Low
Lower-middle
Upper-middle
High Low
Lower-middle
Upper-middle
High
Bottom (1) -0.133*** -0.185*** -0.239*** -0.051*** -0.084*** -0.110*** -0.152*** -0.208*** -0.035*** -0.036*** -0.044*** -0.011
(0.009) (0.014) (0.019) (0.007) (0.010) (0.011) (0.012) (0.018) (0.013) (0.008) (0.010) (0.009)
Second (2) -0.113*** -0.148*** -0.177*** -0.037*** -0.075*** -0.090*** -0.116*** -0.137*** -0.038*** -0.034*** -0.036*** -0.012
(0.009) (0.013) (0.016) (0.007) (0.006) (0.013) (0.007) (0.013) (0.010) (0.010) (0.007) (0.009)
Third (3) -0.090*** -0.103*** -0.135*** -0.019** -0.049*** -0.070*** -0.085*** -0.087*** -0.033*** -0.018*** -0.034*** -0.000
(0.007) (0.011) (0.014) (0.008) (0.006) (0.008) (0.008) (0.013) (0.008) (0.007) (0.008) (0.007)
Fourth (4) -0.050*** -0.079*** -0.075*** -0.004 -0.032*** -0.045*** -0.052*** -0.032*** -0.021*** -0.014*** -0.022** 0.002
(0.008) (0.010) (0.010) (0.008) (0.006) (0.006) (0.010) (0.012) (0.007) (0.005) (0.009) (0.007)
No. of observations 25,369 34,144 35,820 30,681 25,369 34,144 35,820 30,681 25,369 34,144 35,820 30,681
p value of F statistic
H0: (1) = (2) 0.02** 0.00*** 0.00** 0.02** 0.34 0.01*** 0.00*** 0.00*** 0.72 0.75 0.27 0.90
H0: (2) = (3) 0.02** 0.00*** 0.00*** 0.01* 0.00*** 0.10* 0.00*** 0.00*** 0.41 0.02** 0.83 0.12
H0: (3) = (4) 0.00*** 0.02** 0.00*** 0.10 0.03** 0.00*** 0.00*** 0.00*** 0.11 0.48 0.09 0.73
Source: Authors’ regressions using Global Findex and World Development Indicators data.
a. Each column reports results of a single probit regression of the indicated financial inclusion measure (a dummy variable equal to 1 if the respondent meets the indicated criterion, and zero otherwise) on country fixed
effects, the respondent’s within-country income quintile (the top quintile is the omitted category), and the following individual characteristics: sex, age, age squared, rural versus urban residence, education, log of household
size, marital status, and whether employed. All regressions account for stratification and clustering in the survey design. Data are for 2011 or the most recent year available. Standard errors are in parentheses. Asterisks
denote significance at the ***1 percent, **5 percent, and *10 percent level.

Figure 5. Formal Account Penetration in the Poorest Quintile,
Selected High-Income Countries
Percent of adults without
a formal account
25.78
25
20
15
10 8.83
5 3.05 2.90
United States
Ginia = 37.8
United Kingdom
Gini = 34.5
Australia
Gini = 33.6
Canada
Gini = 32.4
Source: Authors’ calculations using Global Findex and Organization for Economic Cooperation and
Development (OECD) data.
a. A higher Gini index indicates greater income inequality.
(98 percent and 92 percent, respectively). But the Gini coefficient in the
United Kingdom is smaller than that in the United States, which may help
explain the sharp difference between the two countries in account penetra-tion
in the poorest income quintile (figure 5). In the United States 26 percent
of adults in this group report having no formal account; the correspond-ing
number for the United Kingdom is 3 percent. Such differences serve
to reinforce the hypothesis that although the correlation between income
per capita or income inequality and account penetration explains some
variation in the use of financial services, it by no means explains all of
it. Alternative explanations included differences in trust in banks and the
availability of alternatives to formal financial institutions. (A 2011 survey by
the Federal Deposit Insurance Corporation also found a large gap in account
penetration between rich and poor households within the United States.)
Bottom-quintile adults in the United States are also much less likely to
have an account than their counterparts in Australia or Canada—two other
countries with broadly similar economic development and legal traditions
to those of the United States, but with smaller Gini coefficients.
cross-country differences in the use of accounts Beyond the simple
ownership of formal accounts, data on the frequency and methods of use of

those accounts shed light on some stark differences between high-income
and developing countries. In developing countries, 10 percent of account
holders—more than 150 million people worldwide—maintain what can be
considered an inactive account: they make neither withdrawals from nor
deposits into their account in a typical month (although they may maintain
a positive balance). In contrast, only 2 percent of account holders in high-income
countries have an inactive account.
The majority of adults with a formal account in developing countries
make deposits or withdrawals only once or twice in a typical month.
In high-income countries, by contrast, more than half of account hold-ers
withdraw money from their accounts six or more times in a typical
month. ATMs and electronic payment systems (debit cards, electronic
bill payments, and the like) facilitate access to accounts. Indeed, adults
with a formal account in high-income countries report most commonly
using ATMs for withdrawals. Those in developing countries report most
commonly making withdrawals over the counter, in a branch of their
bank or at another financial institution.
People also have myriad reasons for maintaining an account at a formal
financial institution. Using a formal account to receive wages is most com-mon
in high-income countries, where 50 percent of adults report using an
account for this purpose, compared with 14 percent of adults in developing
countries. Relying on an account to receive payments from the government
is also most common in high-income countries, where 42 percent of all
adults (and 47 percent of account holders) report having used their account
for this type of transaction in the past year, compared with 6 percent of
adults in developing countries. Accounts are also used to send money to or
receive money from relatives by 8 percent of all adults (and 21 percent of
account holders) in developing countries.
II.B. Saving
Saving to cover future expenses—education, a wedding, a big purchase—
or to provide against possible emergencies is a universal practice. However,
not only does the propensity to save differ across and within countries; the
mode and the purpose of saving also vary. Globally, 36 percent of adults
report having saved (in the sense of deliberately setting aside money) in the
past year, although this ranges from 30 percent in low-income countries to
58 percent in high-income countries.
More interesting are the marked differences in how people save. A pro-portion
of adults who save do so using a formal account. But many others,
including some who own a formal account, turn to alternative methods of

Figure 6. Participation in Formal Saving, by Country Income Group
and Within-Country Income Quintile
a formal account
50
40
30
20
10
High-income countries Top quintile
Bottom quintile
Upper-middle-income countries
Low-income countries
Lower-middle-income countries
7 8 9 10 11
Logarithm of median income in dollars
saving. Worldwide, about one-fourth of adults report having saved at a
bank, credit union, or microfinance institution in the past year. This figure
ranges from 45 percent in high-income countries, to 24 percent in upper-middle-
income countries, to 11 percent in lower-middle-income and low-income
countries. The difference between high- and upper-middle-income
countries in the percentages of adults who saved formally is statisti-cally
significant, but there is no statistically significant difference among
developing-country income groups (column 1-5 in table 1).
Like account penetration, formal saving behavior also varies with indi-vidual
characteristics within countries. As figure 6 shows, in high-income
countries as a group, the share of adults who engage in formal saving rises
sharply with income in the bottom half of the income distribution, from
32 percent in the bottom quintile to 50 percent in the middle quintile, but
becomes much flatter in the top half, rising only from 50 percent to 56 per-cent.
This suggests that in high-income countries, individuals in the middle
class are significantly more likely to save formally than the poor, and only
marginally less likely to save formally than the rich.
The share of adults who save increases more linearly in upper-middle-income
countries: a gap of about 6 percentage points is seen between
each income quintile. Finally, in lower-middle- and low-income countries

there is almost no difference between the middle class and the poor in
the proportion of adults saving formally: for the lower-middle-income
countries the numbers are roughly 9 percent and 6 percent, respectively.
However, in both these groups of countries the rich are more than twice
as likely to save formally as the middle class: about 21 percent compared
with about 9 percent in the case of the lower-middle-income countries.
Probit estimations using individual-level data confirm these results. For
instance, in high-income countries adults in the poorest income quintile
are 21 percent less likely to save formally than adults in the richest quin-tile,
whereas in low-income countries the difference is only 8 percent
(compare the first and the last columns of the middle panel of table 2).
Saving behavior varies among account holders as well: even individu-als
who have a formal account may not necessarily use it to save. World-wide,
about 43 percent of account holders report having set aside money
at a formal financial institution in the past year; the figure varies rela-tively
little across country income groups. However, in many sub-Saharan
African countries, such as Liberia and Uganda, more than 65 percent of
account holders report saving formally. This suggests that in these coun-tries
the ability to save in a secure location may motivate individuals to
open and maintain a formal account. In contrast, in many countries in
Central and Eastern Europe and Central Asia, adults do not primarily use
their accounts to save: in this region fewer than one in six adults with a
formal account report having saved or set aside money using a formal
account in the past year. In Georgia just 3 percent of account holders (and
1 percent of all adults) report having saved using a formal account in the
past year. However, adults in this region are especially likely to use their
accounts to receive wages and government payments. This ability, rather
than the opportunity for saving, may thus be a key reason why these adults
own formal accounts.
Many adults, despite having a formal account, save solely using other
methods. These people, who might be classified as the “underbanked,”
constitute 12 percent of account holders worldwide. Individuals may
choose an informal saving method rather than use their formal account
because the costs of using their account are prohibitively high. Barriers
such as minimum balance and withdrawal fees and physical distance often
raise the cost of opening and maintaining a formal account. It is also pos-sible
that accounts set up by an employer or the government are not con-ducive
to saving. If that is the case, policymakers or commercial banks in
countries where greater financial inclusion is a priority could introduce
new products to encourage existing account holders to save in formal

institutions. Such products could be especially important in countries with
aging populations.16
In developing countries, savings clubs often serve as an alternative
(or complement) to saving at a formal financial institution. One common
form of such clubs is the rotating savings and credit association (ROSCA),
known locally as a susu in West Africa, an arisan in Indonesia, and a pan-dero
in Peru. These clubs generally operate by pooling the weekly deposits
of their members and disbursing the entire amount to a different mem-ber
each week. Although members generally do not earn interest on their
deposits as in a formal account, these clubs can provide members an oppor-tunity
to save.
Savings clubs and other community-based saving methods are widely
used in some parts of the world, particularly in low-income countries. In
sub-Saharan Africa 19 percent of adults report having saved in the past
year using a savings club or a person outside the family. Among just
those who report any saving activity in the past 12 months, 48 percent
used community-based methods. The practice is particularly common in
Nigeria, where ROSCAs are called esusu, ajo, cha, or adashi. In Nigeria
44 percent of adults (and 69 percent of those who save) report using a sav-ings
club or a person outside the family. Perhaps because of the widespread
use of this saving method, the share of Nigerians who report any type of
saving in the past year is equal to that in Canada and South Korea and far
higher than that in most other developing countries.
The popularity of savings clubs speaks to their advantages, but these
arrangements also have their downside. Their defining characteristic, infor-mality,
is accompanied by risks of fraud and collapse. Of course, formal
accounts are not immune to these risks, especially in many developing
countries where explicit government-run deposit insurance is absent or
inadequate. In addition, the cyclical nature of contributions and disburse-ments
in a ROSCA may be too rigid for some people. A fixed schedule may
not serve their need to deposit surplus income when available or to quickly
withdraw funds in an emergency.
Community-based saving methods and formal financial institutions are
not the only options for saving. A large share of adults around the world
who report having set aside money in the past year used neither a formal
financial institution, nor an informal savings club, nor a person outside the
family. Although the Global Findex survey did not gather data on other
16. See, for example, Chawla, Betcherman, and Banerji (2007), who provide an over-view
of the challenges of aging populations in Eastern Europe and the former Soviet Union.

alternative methods, they might include saving through asset accumulation
(such as gold or livestock) and saving “under the mattress.”17 These adults
account for 29 percent of savers worldwide and more than half of savers
in 55 countries.
II.C. Borrowing
In the Global Findex data, the overall rate of origination of new loans,
formal and informal, is fairly steady across country income groups and
individual characteristics. On average, almost one-third of adults in both
high-income and developing countries report having borrowed money in
the past year. However, measures of new (or rolled-over) household debt
are sensitive to the business cycle and other current economic factors, and
future rounds of data collection may yield significantly different estimates.
Moreover, the use of credit is sensitive to the tax, legal, and regulatory
environment of the country in question. For example, the provision of pri-vate
credit is higher in countries with better creditor protection and broader
credit information coverage (Djankov, McLiesh, and Shleifer 2007).
Beyond the overall rate of new borrowing, however, high-income and
developing countries exhibit little commonality in the sources and purpose
of credit. Individuals in higher-income countries are significantly more
likely to borrow from formal sources, such as banks or retail stores (col-umn
1-6 of table 1; see also figure 7). Those in lower-income countries are
more likely to use informal sources of credit such as family and friends. To
illustrate, in Finland 24 percent of adults report having borrowed from a
formal financial institution in the past year; in Ukraine only 8 percent report
having done so, and in Burundi only 2 percent. The pattern is reversed
with respect to the proportion of adults with informal credit: 37 percent of
adults in Ukraine and 44 percent in Burundi, but only 15 percent in Fin-land,
report having borrowed from family or friends in the past 12 months.
This propensity toward informal rather than formal lending is observed
in both low- and middle-income countries. Friends and family are the most
commonly reported source of new loans in upper-middle-, lower-middle-,
and low-income countries, but not in high-income countries (figure 8). In
low-income countries 20 percent of adults report friends or family as their
only source of new loans in the past year; only 6 percent report a formal
financial institution as their only source. Adults in poorer countries are also
17. Because of the sensitivity of household finances and the inhibitions brought about
by face-to-face surveys, the Global Findex survey did not probe deeply into the practices of
“under the mattress” saving in the home.

Figure 7. Origination of New Formal Loans Worldwide
Percent of adults
who have borrowed
from a formal
financial institution,
last 12 months
0–4
5–9
10–14
15–19
20+
No data
Percent of adults using
the indicated sourcea
Friends or family
Store credit
Bank, credit union,
or microfinance institution
Informal lender
35
30
25
20
15
10
5
High-income
countries
Upper-middle-income
countries
Lower-middle-income
countries
a. Respondents could report borrowing from more than one source.
Low-income
countries
Figure 8. Sources of New Loans, by Country Income Group

more likely to report having borrowed from an informal lender other than
a family member or friend in the past year. An important caveat to this
finding, however, is that social norms may have a significant effect on the
reporting of this type of borrowing.
The introduction of credit cards may affect the demand for and the use
of short-term formal credit. In high-income countries half of the adult pop-ulation
report having a credit card. Despite a surge in recent years, credit
card ownership in developing countries still lags far behind. Only 7 percent
of adults in low- and middle-income countries report having a credit card,
but there are some notable exceptions: in Brazil, Turkey, and Uruguay, for
example, the proportion of adults with a credit card exceeds 35 percent.
Given the widespread ownership of credit cards in high-income coun-tries,
adults in these countries may have less need for short-term loans from
financial institutions. This may help explain why the share of adults in
these countries who report having received a loan in the past year from a
formal financial institution is not particularly high. Indeed, if the adults in
high-income countries who report owning a credit card are included in the
share of those who report borrowing from a formal financial institution in
the past year (a measure that may not include credit card balances), that
share increases by 40 percentage points, from 14 percent to 54 percent.18
Here we focus on measures of borrowing activity that do not include credit
card ownership.
Within-country relative income is also associated with formal borrow-ing
only among developing countries (rightmost panel of table 2). On aver-age,
the difference in the origination of new formal loans over the past year
between the poorest and the richest income quintile in developing countries
is about 4 percent and statistically significant. Within high-income coun-tries,
in contrast, there is no significant difference across income groups on
this measure.
Just as the sources of credit differ across countries and individuals, so
do the purposes for which such borrowing is used. Data gathered in devel-oping
countries highlight that emergency and health needs are the most
common reason for having an outstanding loan (figure 9).19 Adults in the
18. The Gallup World Poll collects information on the ownership of credit cards but not
their use.
19. Data on the main purpose of outstanding loans were gathered only in developing
countries, because Gallup, Inc., enforces a time limit for phone interviews conducted in
high-income countries, limiting the number of questions that can be added to the core ques-tionnaire.
Respondents were asked to choose from a list of reasons for borrowing, so it is
possible that reasons not listed (borrowing to start a business, for example) are also common.

Percent of borrowers reporting
the indicated purpose
10
8
6
poorest income quintiles also commonly report emergency and health-related
loans. On average, in developing countries 14 percent of adults in
the poorest quintile had a loan for emergency or health purposes, compared
with 8 percent of those in the richest fifth of the population.
The data also highlight variation in the reasons for borrowing across
regions. In sub-Saharan Africa 8 percent of adults report borrowing to pay
school fees. In the developing world as a whole, outstanding loans for
funerals or weddings are reported by 3 percent of adults (figure 9), but such
loans are significantly more common in fragile and conflict-affected states
such as Afghanistan (where the figure is 29 percent), Iraq (13 percent),
Somalia (11 percent), and the West Bank and Gaza (11 percent).
Data on the use of mortgages show large differences between countries
at different income levels. In high-income countries 24 percent of adults
report having an outstanding loan to purchase a home; the corresponding
number in developing countries is only 3 percent. Even within the Euro-pean
Union the use of mortgages varies widely, with very low rates of use
in some of the new member states. For example, whereas 21 percent of
adults in Germany have an outstanding mortgage, only 3 percent in Poland
do (figure 10). Such differences may in part reflect cross-country differ-ences
in housing finance systems, such as in product diversity, types of
4
2
Home
construction
School fees Emergency or
health needs
Funeral or
wedding
Figure 9. Purposes of Outstanding Loans in Developing Countries

Figure 10. Mortgage Penetration in Europe
Percent of adults
with an outstanding
loan to purchase a
home or apartment
0–5
6–10
11–20
21–30
31+
No data
lenders, secondary mortgage markets, and degree of government participa-tion.
Studies have found that these factors may affect the availability of
loans to individuals (International Monetary Fund 2011). Collateral and
bankruptcy laws that define the legal rights of borrowers and lenders have
also been shown to affect housing finance (Warnock and Warnock 2008).
And to develop fully in the first place, a mortgage market requires the exis-tence
of formal property rights and an efficient framework to record owner-ship
of property (de Soto 2000).
III. Barriers to Financial Inclusion
Country income and individual characteristics clearly help explain some of
the differences in the use of financial accounts around the world. But what
do people themselves say when asked why they do not have an account?
The Global Findex survey, by asking more than 70,000 adults without a
formal account their reasons for not having one, provides novel data on the
barriers to financial inclusion. In this section we discuss each self-reported
barrier individually. Each represents a distinct dimension that policymakers
who are aiming to expand financial inclusion can address. We also exam-ine
these self-reported barriers by country income group and individual

Figure 11. Reported Reasons for Not Having a Bank Accounta
Not enough money to use
Too expensive
Family member already has account
Too far away
Lack of necessary documentation
Lack of trust
Religious reasons
10 20 30 40
50 60
Percent of respondents
a. Respondents could choose more than one reason. The lower bar for “Not enough money” refers to
the percentage of adults who reported only this reason.
characteristics. This allows us to document robust relationships between
subjective and objective assessments of barriers to financial access, even
when accounting for GDP per capita.
Globally, the most frequently cited reason for not having a formal account
is lack of enough money to use one (figure 11). This is the response given
by 65 percent of adults without a formal account, and 29 percent cited this
as the only reason (multiple responses were permitted).20 The next most
commonly cited reasons are that banks or accounts are too expensive, and
that another family member already has an account. Each of these was cited
by about a quarter of adults without an account. The other reasons reported
(in order of importance) are banks being too far away, lack of the necessary
documentation, lack of trust in banks, and religious reasons. On average,
each respondent chose 1.7 responses; the most commonly offered response
combined lack of enough money to use an account with a second barrier.
In low-income countries adults gave 1.91 responses, on average. Adults in
these countries were significantly more likely to cite distance, cost, docu-mentation,
and lack of money than were adults in other country income
groups. Lack of trust and someone else in the family already having an
account were more commonly cited in middle- and high-income countries.
20. Among all respondents, 12 percent chose none of the given reasons for not having
an account.

Figure 12. Subjective and Objective Measures of Cost as a Barrier to Account Access
Countries where the cost
to open an account isa
Negligible
Low
Medium
High
5 10 15 20
25 30 35
Percent of non–account holders in the country citing cost as a barrier
Source: Beck and others (2008).
a. As measured by the Annual Fees Account Index from the World Bank’s Bank Regulation and Super-vision
Database.
At first glance it may appear that the segment of the population for
whom lack of enough money is a concern is less likely to be bankable.
However, those who reported this reason are likely suggesting that, under
current circumstances, the costs of having an account outweigh its benefits.
It seems reasonable to assume that if individuals found it easier or cheaper
to use accounts, or if those accounts provided benefits such as the ability to
receive remittances or government transfers, then for some of these respon-dents
the costs associated with having an account would be outweighed by
the benefits.
Affordability is also an important barrier to account ownership. High
costs were cited by a quarter of unbanked respondents on average, and
by 32 percent in low-income countries, where fixed transaction costs and
annual fees tend to make small transactions unaffordable for large parts
of the population. (Fixed fees and other high costs of opening and main-taining
accounts often reflect lack of competition and underdeveloped
physical or institutional infrastructure.) Maintaining a checking account
in Sierra Leone, for example, costs the equivalent of 27 percent of GDP
per capita in annual fees alone. So it is no surprise that 44 percent of non–
account holders in that country cited high cost as a reason for not having
a formal account. Figure 12 shows that the proportion of adults citing cost
as a barrier to account ownership rises monotonically with actual costs
as measured by the Annual Fees Account Index from the World Bank’s
Bank Regulation and Supervision Database (Beck, Demirgüç-Kunt, and
Martinez Peria 2008).

The next most commonly cited reason for not having an account (offered
by 23 percent of respondents) was that another member of the family
already has one. Women were significantly more likely than men to give
this response, and adults in high-income and upper-middle-income coun-tries
(where relatives are most likely to have an account) were significantly
more likely than those in poorer countries to choose this reason. A recent
study (Hallward-Driemeier and Hasan 2013) shows that lack of account
ownership (and lack of personal asset accumulation) limits women’s ability
to pursue self-employment opportunities. Hence, although such voluntary
exclusion may be linked to individual preferences or cultural norms, it may
in some cases indicate a lack of awareness of financial products or lack of
financial literacy more generally.21
Twenty percent of unbanked respondents cited distance as a reason for
not having a formal account. The frequency with which this barrier was
cited increases sharply as one moves down the country income scale, from
10 percent in high-income countries to 28 percent in low-income countries.
Among developing countries there is a significant relationship between dis-tance
as a reported barrier and objective measures of providers such as bank
branch penetration. Tanzania, for example, has a large share (47 percent)
of non–account holders who cited distance as a reason for not having an
account, and the country ranks near the bottom in bank branch penetration,
averaging less than 0.5 bank branch per 1,000 square kilometers (according
to the 2010 World Bank Global Payment Systems Survey).
Documentation requirements for opening an account may also exclude
workers in the rural or the informal sector, who are less likely to have
wage slips or formal proof of residence. A significant relationship is seen
across developing countries between subjective and objective measures of
documentation requirements as a barrier to account use (figure 13); the
relationship holds even after we account for GDP per capita. Indeed, the
Financial Action Task Force has recognized that overly cautious safeguards
against money laundering and terrorist financing can have the unintended
consequence of excluding legitimate businesses and consumers from the
financial system. Accordingly, the task force has emphasized the need to
ensure that such safeguards also support financial inclusion, where greater
inclusion is a national goal.22
21. The institutional barriers to financial inclusion are further analyzed in Allen and
others (2012).
22. For more on documentation requirements and safeguards against money laundering,
see Yikona and others (2011) and Financial Action Task Force (2011).

Figure 13. Subjective and Objective Measures of Documentation Requirements
as a Barrier to Account Accessa
Percent of non–account holders
citing documentation as a barrier
40
35
30
25
20
15
10
5
1 2 3 4
No. of documents required to open a checking account
Source: Authors’ calculations using Global Findex and Bank Regulation and Supervision Database
(World Bank).
a. Each observation represents 1 of 37 developing countries.
Distrust in formal financial institutions is also a nontrivial barrier to
wider financial inclusion, and one that is difficult to address in the short
term. Thirteen percent of adults without a formal account cited lack of
trust in banks as a reason why they do not own an account (figure 11).
This distrust can stem from cultural norms, discrimination against certain
population groups, past episodes of bank failure or government expropri-ation
of banks, or economic crises and uncertainty. In Russia 38 percent
of non–account holders cited lack of trust in banks as a reason for not
having an account—approximately three times the share in developing
countries on average.
Finally, only 5 percent of unbanked respondents cited religious reasons
for not having a formal account, although the proportion is higher in some
Middle Eastern countries such as the West Bank and Gaza and in some
South Asian countries such as Pakistan. In these regions developing finan-cial
products compatible with religious beliefs (so-called Islamic finance)
could potentially increase account penetration.
These systematic data on self-reported barriers to the use of financial
services allow researchers and policymakers to understand the reasons for

nonuse and provide clues for the design of policy interventions. However,
such cross-sectional data cannot be used to determine the causal impact of
removing these barriers. Furthermore, since people often face (and report)
multiple barriers, addressing individual constraints may not necessarily
expand the use of accounts if other barriers continue to bind.
IV. Mobile Money, Branchless Banking, and Beyond
As documented in section II, there is a strong correlation between national
income and financial inclusion. However, policy innovations may still be
able to bring about more inclusive financial systems even at low levels of
income. The Global Findex database allows us to observe how public and
private sector–led initiatives might change how people engage with the
formal financial system.
The success of mobile money illustrates the transformative potential of
technical progress and innovation to promote financial inclusion. Mobile
money—sometimes considered a form of branchless banking—has allowed
people who are otherwise excluded from the formal financial system to per-form
financial transactions in a relatively cheap, secure, and reliable man-ner
(Jack and Suri 2011). Individuals using mobile money maintain a type
of account that allows them to make deposits and withdrawals through cash
transactions at a network of retail agents. They can then transfer money
or pay bills using text messages. Many mobile money accounts—such as
those provided by M-PESA in Kenya or GCash in the Philippines—are
not connected to an account at a financial institution, but the providers are
often required to store the aggregate sums of the accounts in a bank. Cus-tomers
are ordinarily charged a fee for sending money to others or making
a withdrawal from their account.
Mobile money has achieved the broadest success in sub-Saharan Africa,
where 16 percent of adults report having used a mobile phone in the past
12 months to pay bills or send or receive money (figure 14). The share of
adults using mobile money is less than 5 percent in all other regions, but a
few countries, including Haiti and the Philippines, are notable exceptions
to the pattern.
The degree to which mobile money is capturing the unbanked market
differs across countries. In Kenya 43 percent of adults who report having
used mobile money in the past 12 months do not have a formal account.
In Sudan the figure is 92 percent. This heterogeneity may reflect the varied
and fast-evolving regulations surrounding mobile money. When M-PESA
was launched in Kenya, it had no association with the formal banking sec-

Figure 14. Use of Mobile Money in Africa
Percent of adults who used
a mobile phone to pay bills
or send or receive money,
last 12 months
0–10
11–30
31–60
61+
No data
tor, and mobile banking customers there were exempt from the documen-tation
requirements imposed by banks. But governments are increasingly
favoring bank-led models in which mobile money providers have partner-ships
with or are formed directly through banks (Consultative Group to
Assist the Poor 2010).
In recent years the proliferation of branchless banking has also received
growing attention as a way to increase financial access in developing coun-tries,
particularly among underserved groups (see Mas and Kumar 2008).
One mode of branchless banking centers on bank agents, who often operate
out of retail stores, gas stations, or post offices. By capitalizing on exist-ing
infrastructure and client relationships, operators can expand financial
access in a more cost-efficient manner. Bank agents themselves can also be
mobile, making daily or weekly rounds among clients. Few account hold-ers
currently report relying on bank agents (whether over the counter at a
retail store or some other person associated with their bank) as their main
mode of withdrawal or deposit. But in several Asian countries—including

Bangladesh, Laos, Nepal, and the Philippines—more than 10 percent of
account holders already report using bank agents.
There is also enormous scope for the public sector to bring about trans-formative
change in how adults around the globe interact with the formal
financial sector. Increasingly, governments are using formal accounts to
disburse transfer payments. In Brazil the government allows recipients of
conditional cash transfers (as part of its Bolsa Familia program) to receive
payments via no-frills bank accounts, although many more choose to
receive payments via a virtual account that does not allow deposits or indef-inite
storage (Consultative Group to Assist the Poor 2011). Still, according
to Findex data, 20 percent of adults in Brazil report receiving government
transfers via a bank account, one of the highest proportions in the develop-ing
world. In India the government recently began depositing government
pension and scholarship payments directly into the bank accounts of almost
250,000 people in 20 districts. Officials plan to expand the program and
hope it will prevent corruption as well as expand financial access.23 The
data provide suggestive evidence that these types of reforms may have the
potential to dramatically expand the reach of the formal financial sector to
the poorest individuals.
V. Conclusion
For most people around the world, having an account at a financial insti-tution
serves as an entry point into the formal financial sector. A formal
account can encourage saving and open access to credit. It can also make
it easier to transfer wages, remittances, and government payments. Broad-based
access to accessible and affordable formal accounts is a hallmark
of an inclusive financial system, the absence of which can contribute to
persistent income inequality and slower economic growth.
Yet until now little was known about the global reach of the financial
sector and financial inclusion—the extent of account ownership and the
use of formal payments, saving, and credit—or about the degree to which
groups such as the poor are excluded from formal financial systems. Sys-tematic
indicators of the use of different formal and informal financial ser-vices
were lacking for most countries.
23. Gardiner Harris, “India Aims to Keep Money for Poor Out of Others’ Pockets,”
New York Times, January 5, 2013.

As the first public database of indicators that consistently measure
people’s use of financial products across countries and over time, the
Global Findex database fills a big gap in existing data on financial inclu-sion.
The data show wide gaps in account penetration between high-income
and developing countries and between the poor and the rich
within countries. Also, the data show variation in the use of formal and
informal saving and credit mechanisms. By enabling policymakers to
identify segments of the population excluded from the formal financial
sector, the data can help provide insights for the design and prioritization
of reforms.
a p p e n d i x a
Selected Questions from the Global Findex Survey
The text of this appendix is taken verbatim from the survey.
This next section is about banks and financial institutions. We are trying to under-stand
how people across the world use financial institutions and how available
they are to people. Please remember that all information you provide is com-pletely
confidential.
—Do you, either by yourself or together with someone else, currently have an
account at any of the following places? An account can be used to save money, to
make or receive payments, or to receive wages and remittances. Do you currently
have an account at [surveyor reads A and B]?24
1 Yes
2 No
A A bank or credit union (or other formal financial institution, where appli-cable,
like a cooperative in Latin America)
B A post office
—A debit card, sometimes called an ATM card, is a card that allows you to make
payments, get money, or buy things and the money is taken out of your bank
account right away. Do you have a debit card?
1 Yes
2 No
24. For all questions the choices of “don’t know” and “refused” are also included as pos-sible
responses (results not shown).

—A credit card is like a debit card but the money is not taken from your account
right away. You get credit to make payments or buy things, and you can pay the
balance off later. Do you have a credit card?
1 Yes
2 No
—In a typical month, about how many times is money deposited into your per-sonal
account(s)? This includes cash or electronic deposits, or any time money
is put into your account(s) by yourself or others. [surveyor reads 1 through 4 and
codes one response only]
1 0
2 1–2 times
3 3–5 times
4 6 times or more
—In a typical month, about how many times is money taken out of your
personal account(s)? This includes cash withdrawals, electronic payments or
purchases, checks, or any other time money is removed from your account(s)
by yourself or others. [surveyor reads 1 through 4 and codes one response
only]
1 0
2 1–2 times
3 3–5 times
4 6 times or more
—When you need to get cash (paper or coins) from your account(s), do
you usually get it . . . ? [surveyor reads 1 through 4 and codes one response
only; respondents can also answer that they do not withdraw cash (coded
as 5)]
1 At an ATM
2 Over the counter in a branch of your bank or financial institution
3 Over the counter at a retail store
4 From some other person who is associated with your bank or financial
institution
—When you put cash (paper or coins) into your account(s), do you usually do
it . . . ? [surveyor reads 1 through 4 and codes one response only; respondents
can also answer that they do not withdraw cash (coded as 5)]
1 At an ATM
2 Over the counter in a branch of your bank or financial institution
3 Over the counter at a retail store
4 From some other person who is associated with your bank or financial
institution

—In the past 12 months, have you used your account(s) to . . . ? [surveyor reads
A through D]
1 Yes
2 No
A Receive money or payments for work or from selling goods
B Receive money or payments from the government
C Receive money from family members living elsewhere
D Send money to family members living elsewhere
—Please tell me whether each of the following is a reason why you, personally,
DO NOT have an account at a bank, credit union or other financial institution.
[surveyor reads and rotates A through G]
1 Yes
2 No
A They are too far away
B They are too expensive
C You don’t have the necessary documentation (ID, wage slip)
D You don’t trust them
E You don’t have enough money to use them
F Because of religious reasons
G Because someone else in the family already has an account
—In the past 12 months, have you saved or set aside any money?
1 Yes [surveyor continues with next question]
2 No [surveyor skips to question 74]
—In the past 12 months, have you saved or set aside any money by . . . ? [sur-veyor
reads A and B]
1 Yes
2 No
A Using an account at a bank, credit union, or microfinance institution
B Using an informal savings club or a person outside the family (insert local
example)
—In the past 12 months, have you borrowed any money from . . . ? [surveyor
reads A through E]
1 Yes
2 No
A A bank, credit union, or microfinance institution
B A store by using installment credit or buying on credit
C Family or friends

D Employer
E Another private lender
—Do you currently have a loan you took out for any of the following reasons?
[surveyor reads A through E]
1 Yes
2 No
A To purchase your home or apartment
B To purchase materials or services to build, extend, or renovate your home
or apartment
C To pay school fees
D For emergency/health purposes
E For funerals or weddings
—In the past 12 months, have you used a mobile phone to . . . ? [surveyor reads
A through C]
1 Yes
2 No
A Pay bills
B Send money
C Receive money
a p p e n d i x b
Account Penetration by Country
Percent of adults with an account at a formal financial institution
Country
All
adults
Poorest
20%
Richest
20% Country
Afghanistan 9 0 20
Albania 28 7 43
Algeria 33 22 50
Angola 39 31 40
Argentina 33 19 55
Armenia 17 16 24
Australia 99 97 100
Austria 97 93 99
Azerbaijan 15 13 25
Bahrain 65 64 60
Bangladesh 40 33 54
Belarus 59 37 75
All
adults
Poorest
20%
Richest
20%
Belgium 96 92 96
Benin 10 5 24
Bolivia 28 12 50
Bosnia and
56 35 69
Herzegovina
Botswana 30 12 48
Brazil 56 33 71
Bulgaria 53 29 76
Burkina Faso 13 6 25
Burundi 7 3 23
Cambodia 4 0 12
Cameroon 15 14 22

Country
All
adults
Poorest
20%
Richest
20% Country
All
adults
Poorest
20%
Richest
20%
Canada 96 91 98
Central African
3 1 9
Republic
Chad 9 6 26
Chile 42 19 68
China 64 39 83
Colombia 30 9 62
Comoros 22 9 40
Congo,
4 0 18
Dem. Rep.
Congo, Rep. 9 1 20
Costa Rica 50 30 69
Croatia 88 75 94
Cyprus 85 76 89
Czech Republic 81 70 88
Denmark 100 99 100
Djibouti 12 4 34
Dominican
38 19 62
Republic
Ecuador 37 22 61
Egypt 10 5 25
El Salvador 14 1 32
Estonia 97 94 99
Finland 100 99 100
France 97 96 100
Gabon 19 4 38
Georgia 33 25 50
Germany 98 97 100
Ghana 29 17 61
Greece 78 75 85
Guatemala 22 8 52
Guinea 4 2 10
Haiti 22 4 49
Honduras 21 15 47
Hong Kong 89 78 98
Hungary 73 58 86
India 35 21 56
Indonesia 20 8 48
Iran 74 63 80
Iraq 11 5 13
Ireland 94 88 97
Israel 90 88 92
Italy 71 61 81
Jamaica 71 71 67
Japan 96 94 96
Jordan 25 16 33
Kazakhstan 42 30 55
Kenya 42 19 85
Korea, Rep. 93 86 94
Kosovo 44 24 59
Kuwait 87 86 90
Kyrgyz
4 1 11
Republic
Laos 27 16 27
Latvia 90 82 95
Lebanon 37 20 54
Lesotho 18 8 29
Liberia 19 3 41
Lithuania 74 66 87
Luxembourg 95 97 94
Macedonia 74 66 85
Madagascar 6 1 19
Malawi 17 9 36
Malaysia 66 45 82
Mali 8 4 18
Malta 95 93 96
Mauritania 17 7 43
Mauritius 80 66 94
Mexico 27 12 58
Moldova 18 6 36
Mongolia 78 68 89
Montenegro 50 34 67
Morocco 39 0 0
Mozambique 40 21 56
Nepal 25 15 39
Netherlands 99 98 99
New Zealand 99 100 99
Nicaragua 14 4 31
Niger 2 0 6
Nigeria 30 12 62
Oman 74 63 92
Pakistan 10 5 19
Panama 25 18 44
Paraguay 22 4 51
Peru 20 6 47
Philippines 27 4 54
Poland 70 60 82
Portugal 81 64 87
Qatar 66 47 80
Romania 45 25 69

Country
All
adults
Poorest
20%
Richest
20% Country
All
adults
Poorest
20%
Richest
20%
Russia 48 34 61
Rwanda 33 23 42
Saudi Arabia 46 32 51
Senegal 6 4 13
Serbia 62 47 70
Sierra Leone 15 4 30
Singapore 98 98 98
Slovak
80 66 85
Republic
Slovenia 97 92 100
Somalia 31 12 58
South Africa 54 35 78
Spain 93 91 92
Sri Lanka 69 52 87
Sudan 7 4 15
Swaziland 29 12 44
Sweden 99 99 100
Syria 23 20 28
Taiwan 87 77 90
Tajikistan 3 1 6
Tanzania 17 3 45
Thailand 73 64 87
Source: Global Findex.
Togo 10 2 18
Trinidad and
76 70 85
Tobago
Tunisia 32 14 63
Turkey 58 46 72
Turkmenistan 0 0 1
Uganda 20 7 37
Ukraine 41 21 59
United Arab
60 57 58
Emirates
United
Kingdom
97 97 97
United States 88 74 90
Uruguay 24 7 49
Uzbekistan 23 15 27
Venezuela 44 27 54
Vietnam 21 6 35
West Bank
19 8 34
and Gaza
Yemen 4 0 9
Zambia 21 8 50
Zimbabwe 40 22 63
ACKNOWLEDGMENTS We thank Franklin Allen, Oya Pinar Ardic Alper,
Thorsten Beck, Massimo Cirasino, Robert Cull, Pascaline Dupas, Maya Eden,
Tilman Ehrbeck, Michael Fuchs, Xavi Gine, Markus Goldstein, Ruth Goodwin-
Groen, Raúl Hernández-Coss, Richard Hinz, Jake Kendall, Aart Kraay, Alexia
Latortue, Sole Martinez Peria, Ignacio Mas-Ribo, Jonathan Morduch, Nataliya
Mylenko, Mark Napier, Douglas Pearce, Bikki Randhawa, Liliana Rojas-
Suárez, Richard Rosenberg, Armida San Jose, Kinnon M. Scott, Peer Stein, Gaiv
Tata, Jeanette Thomas, Klaus Tilmes, Asli Togan Egrican, Augusto de la Torre,
Rodger Voorhies, and Alan Winters for their valuable and substantive com-ments
during various stages of the project. The team is also appreciative of the
excellent survey execution and related support provided by Gallup, Inc., under
the direction of Jon Clifton. We are especially grateful to the Bill & Melinda
Gates Foundation for providing financial support that made the collection and
dissemination of the data possible. This paper was prepared with outstanding
assistance from Atisha Kumar and Douglas Randall. This paper’s findings,
interpretations, and conclusions are entirely those of the authors and do not
necessarily represent the views of the World Bank, their executive directors, or
the countries they represent. The authors report no potential conflict of interest.

References
Allen, Franklin, Aslı Demirgüç-Kunt, Leora Klapper, and Maria Soledad Martinez
Peria. 2012. “The Foundations of Financial Inclusion: Understanding Owner-ship
and Use of Formal Accounts.” Policy Research Working Paper no. 6290.
Washington: World Bank.
Ashraf, Nava, Diego Aycinena, Claudia Martinez A., and Dean Yang. 2010.
“Remittances and the Problem of Control: A Field Experiment among Migrants
from El Salvador.” University of Michigan.
Aterido, Reyes, Thorsten Beck, and Leonardo Iacovone. 2011. “Gender and
Finance in Sub-Saharan Africa: Are Women Disadvantaged?” Policy Research
Working Paper no. 5571. Washington: World Bank.
Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Cynthia Kinnan. 2010.
“The Miracle of Microfinance? Evidence from a Randomized Evaluation.” MIT
Bureau for Research and Economic Analysis of Development Working Paper
no. 278. Massachusetts Institute of Technology.
Beck, Thorsten. 2009. “FinAccess 2009: Trends, Analysis and Policy Conclusions.”
Consultant report for FSD Trust Kenya, Nairobi.
Beck, Thorsten, and Martin Brown. 2011. “Use of Banking Services in Emerging
Markets: Household-Level Evidence.” European Banking Center Discussion
Paper no. 2011-024. Tilburg University, the Netherlands.
Beck, Thorsten, Aslı Demirgüç-Kunt, and Ross Levine. 2007. “Finance, Inequality,
and the Poor.” Journal of Economic Growth 12, no. 1: 27–49.
Beck, Thorsten, Aslı Demirgüç-Kunt, and Maria Soledad Martinez Peria. 2007.
“Reaching Out: Access to and Use of Banking Services across Countries.”
Journal of Financial Economics 85, no. 2: 234–66.
———. 2008. “Banking Services for Everyone? Barriers to Bank Access and Use
around the World.” World Bank Economic Review 22, no. 3: 397–430.
Beck, Thorsten, Ross Levine, and Norman Loayza. 2000. “Finance and the Sources
of Growth.” Journal of Financial Economics 58, no. 1: 261–300.
Campbell, John. 2006. “Household Finance.” Journal of Finance 61, no. 4:
1553–1604.
Chawla, Mukesh, Gordon Betcherman, and Anne Banerji. 2007. From Red to
Gray: The “Third Transition” of Aging Populations in Eastern Europe and the
Former Soviet Union. Washington: World Bank.
Chen, Shaohua, and Martin Ravallion. 2010. “The Developing World Is Poorer
Than We Thought but No Less Successful in the Fight against Poverty.” Quar-terly
Journal of Economics 125, no. 4: 1577–1625.
Clausen, Bianca, Aart Kraay, and Zsolt Nyiri. 2011. “Corruption and Confidence
in Public Institutions: Evidence from a Global Survey.” World Bank Economic
Review 25, no. 2: 212–49.
Cole, Shawn, Thomson Sampson, and Bilal Zia. 2011. “Prices or Knowledge?
What Drives Demand for Financial Services in Emerging Markets?” Journal of
Finance 66, no. 6: 1933–67.

Collins, Daryl, Jonathan Morduch, Stuart Rutherford, and Orlando Ruthven. 2009.
Portfolios of the Poor: How the World’s Poor Live on Two Dollars a Day.
Princeton University Press.
Consultative Group to Assist the Poor (CGAP). 2010. “Technology Program Focus
Country: India.” Washington.
———. 2011. “CGAP G2P Brazil Country Report.” Washington.
Cull, Robert, Aslı Demirgüç-Kunt, and Jonathan Morduch. 2013. Banking the
World: Empirical Foundations of Financial Inclusion. MIT Press.
Deaton, Angus 2008. “Income, Health, and Well-Being around the World: Evi-dence
from the Gallup World Poll.” Journal of Economic Perspectives 22,
no. 2: 53–72.
Demirgüç-Kunt, Aslı, and Leora Klapper. 2012. “Measuring Financial Inclusion:
The Global Findex Database.” Policy Research Working Paper no. 6025. Wash-ington:
World Bank.
Demirgüç-Kunt, Aslı, and Ross Levine. 2009. “Finance and Inequality: Theory and
Evidence.” Annual Review of Financial Economics 1: 287–318.
Demirgüç-Kunt, Aslı, Leora Klapper, Peter Van Oudheusden, and Luigi Zingales.
2013. “Trust in Banks.” Working paper. Washington: Development Research
Group, World Bank.
De Soto, Hernando. 2000. The Mystery of Capital: Why Capitalism Triumphs in the
West and Fails Everywhere Else. New York: Random House.
Djankov, Simeon, Caralee McLiesh, and Andrei Shleifer. 2007. “Private Credit in
129 Countries.” Journal of Financial Economics 2, no. 84: 299–329.
Dupas, Pascaline, and Jonathan Robinson. 2009. “Savings Constraint and Microenterprise
Development: Evidence from a Field Experiment in Kenya.” Working
Paper no. 14693. Cambridge, Mass.: National Bureau of Economic Research.
———. 2011. “Why Don’t the Poor Save More? Evidence from Health Savings
Experiments.” Working Paper no. 17255. Cambridge, Mass.: National Bureau
of Economic Research.
European Bank for Reconstruction and Development (EBRD). 2011. Life in Tran-sition:
After the Crisis. London.
Financial Action Task Force (FATF). 2011. FATF Guidance on Anti–Money
Laundering and Terrorist Financing Measures and Financial Inclusion. Paris:
FATF/OECD.
Federal Deposit Insurance Corporation (FDIC). 2011. FDIC National Survey of
Unbanked and Underbanked Households. Washington.
Hallward-Driemeier, Mary, and Tazeen Hasan. 2013. Empowering Women: Legal
Rights and Economic Opportunities in Africa. Development Forum Series.
Paris: Agence Française de Développement, and Washington: World Bank.
Honohan, Patrick. 2008. “Cross-Country Variation in Household Access to Finan-cial
Services.” Journal of Banking and Finance 32: 2493–2500.
International Monetary Fund (IMF). 2011. Global Financial Stability Report:
Grappling with Crisis Legacies, September 2011. Washington.

Jack, William, and Tavneet Suri. 2011. “Mobile Money: The Economics of
M-PESA.” Working Paper no. 16721. Cambridge, Mass.: National Bureau of
Economic Research.
Johnston, Don, and Jonathan Morduch. 2008. “The Unbanked: Evidence from
Indonesia.” World Bank Economic Review 22, no. 3: 517–37.
Karlan, Dean, and Jonathan Zinman. 2010. “Expanding Credit Access: Using Ran-domized
Supply Decisions to Estimate the Impacts.” Review of Financial Stud-ies
23, no. 1: 433–64.
King, Robert G., and Ross Levine. 1993. “Finance and Growth: Schumpeter Might
Be Right.” Quarterly Journal of Economics 108, no. 3: 717–37.
Klapper, Leora, and Inessa Love. 2011. “The Impact of the Financial Crisis on New
Firm Registration.” Economics Letters 113: 1–4.
Klapper, Leora, Luc Laeven, and Raghuram Rajan. 2006. “Entry Regulation as
a Barrier to Entrepreneurship.” Journal of Financial Economics 82: 591–629.
Mas, Ignacio, and Kabir Kumar. 2008. “Banking on Mobiles: Why, How, for
Whom?” Focus Note no. 48. Washington: Consultative Group to Assist the
Poor.
Ratha, Dilip. 2006. “Leveraging Remittances for Development.” In Migration,
Trade, and Development, proceedings of a conference sponsored by the Federal
Reserve Bank of Dallas, The Tower Center for Political Studies, the Department
of Economics at Southern Methodist University, and the Jno E. Owens Founda-tion
(October).
Sacks, Daniel, Betsey Stevenson, and Justin Wolfers. 2010. “Subjective Well-Being,
Income, Economic Development and Growth.” Working Paper no. 16441.
Cambridge, Mass.: National Bureau of Economic Research.
Stevenson, Betsey, and Justin Wolfers. 2008. “Economic Growth and Subjective
Well-Being: Reassessing the Easterlin Paradox.” BPEA, no. 1: 1–102.
———. 2011. “Trust in Public Institutions over the Business Cycle.” American
Economic Review Papers and Proceedings 101, no. 3: 281–87.
Warnock, Veronica C., and Francis E. Warnock. 2008. “Markets and Housing
Finance.” Journal of Housing Economics 17, no. 3: 239–51.
World Bank. 2008a. Finance for All? Policies and Pitfalls in Expanding Access.
Washington.
———. 2008b. Global Purchasing Power Parities and Real Expenditures: 2005
International Comparison Program. Washington.
Yikona, Stuart, Brigitte Slot, Michael Geller, Bjarne Hansen, and Fatima el Kadir.
2011. Ill-Gotten Money and the Economy: Experiences from Malawi and
Namibia. Washington: World Bank.

Comments and Discussion
COMMENT BY
PASCALINE DUPAS The new Global Findex data set that Aslı
Demirgüç-Kunt and Leora Klapper are introducing in this paper is argu-ably
322
among the most important multicountry, repeated-cross-sectional
data sets being collected in this decade. It provides much-needed statistics
on the use of financial services around the world at a time when interest in
such services is peaking. Indeed, almost 40 years after Muhammad Yunus
made the first microloan—the first of many exciting developments in finan-cial
services for the poor—only now are we beginning to see concerted
research efforts to map the reach and effect of these tools on households
around the world.
Country-specific micro-level studies have suggested that financial inclu-sion
today may be much lower than what an informed observer would sup-pose
from the ubiquitous media accounts. For example, recent randomized
trials suggest that at best a quarter of households take up available loans
from microfinance institutions in India, Mexico, and Morocco (Banerjee
and others 2013, Crépon and others 2011, Angelucci, Karlan, and Zinman
2012). Dupas and coauthors (forthcoming) document that only 20 percent
of households in rural western Kenya have a bank account, and ongoing cen-suses
in Uganda and Malawi reveal comparable rates (see Dupas, Karlan,
and Robinson 2013). But such micro studies tend to be clustered in a few
countries or areas, and absent more wide-reaching data, it is difficult to
understand how representative and applicable these results are. Efforts to
date to provide more-exhaustive survey evidence have remained limited:
the FinScope survey sponsored by the U.K. Department for International
Development covers only 15 countries (14 of them in Africa), and the
European Bank for Reconstruction and Development’s Life in Transition
Survey covers only 35 countries in Europe and Central Asia.

comments and discussion 323
Given the lack of survey evidence, until the Global Findex data set was
introduced, the most extensive efforts to estimate rates of financial inclusion
worldwide had to rely on triangulation exercises between aggregate bank-ing
data from bank regulators and microfinance institutions (to get absolute
numbers of accounts, loans, and the like) and population counts. Thorsten
Beck, Demirgüç-Kunt, and Ross Levine (2007) focus on the formal banking
sector and estimate that across the 54 countries in their sample, the median
number of deposit accounts per 1,000 people is 529, and across a subset
of 44 of those countries, the median number of loans per 1,000 people is
80. Patrick Honohan (2008) builds on this effort and proposes a “compos-ite
indicator” of access to both formal and semiformal financial services.
This indicator is constructed from estimates of the number of bank accounts
and the size of deposits relative to the total population. These estimates are
generated as functions of the number of microfinance accounts and GDP
per capita, respectively; these functions in turn are based on correlations
observed in the few countries with enough available data.
The first thing that can be done with the Global Findex database is to
check the accuracy of such extrapolation exercises. Because the Honohan
(2008) estimates are from 5 to 8 years before the Global Findex mea-sures,
one should not expect a perfect correlation between the two, but
after downloading both sets of measures, I found the correlation to be
surprisingly high: 0.85 between Honohan’s estimate of “access to finan-cial
services” (Honohan 2008, table 2) and the share of the population
that “has an account at a formal financial institution” in the Global Findex
database (see figure 3 in the paper). What is more, further calculations
by Alberto Chaia and coauthors (2009) based on Honohan’s figures find
that (as their title states) “half the world is unbanked,” which is also a
key finding from the Global Findex. I found this high rate of consistency
across the two types of sources to be very good news: it means that esti-mation
exercises like those of Beck and others (2007), Honohan (2008),
and Chaia and others (2009) are relatively accurate in providing compa-rable
cross-country measures.
But gauging and analyzing covariates of cross-country variation can
take one only so far toward a better understanding of financial inclusion
worldwide. A key advantage of the Global Findex is that it provides infor-mation
on within-country distributions, as well as on basic individual-level
covariates of financial inclusion such as income, attitudes toward formal
banks, and self-reported reasons for using or not using a given financial
tool or service. The paper highlights a few of the many interesting pat-terns
that the data uncover. For example, the authors report that 13 percent

of unbanked adults worldwide mention lack of trust as a reason for not
saving with a formal institution. This suggests that reliability and qual-ity
concerns about the supply side, as also highlighted for the specific
context of western Kenya in Dupas and others (2012), are relevant in a
number of other countries, especially in Africa and South Asia. Another
stunning statistic from the Global Findex is the extremely low (6 percent)
rate of coverage with rainfall, crop, or livestock insurance among those in
developing economies whose livelihood is farming, fishing, or forestry.
Yet another interesting finding is that mobile money services are, at least
initially, disproportionately used by those already banked. (In Kenya, the
pioneer in terms of mobile money, 57 percent of mobile money users have
a formal bank account, compared with a population mean of 42 percent.)
Having access to such basic statistics will help shape the financial inclu-sion
research agenda for years to come.
The other extremely appealing feature of the Global Findex data is that
they are set to be collected triennially for at least three rounds. The first
round, analyzed in this paper, took place in 2011, and two rounds will fol-low
in 2014 and 2017. The timing is particularly fortuitous: the first round
was collected before the mobile money “revolution” really took hold: the
survey reveals that as of 2011, only 16 percent of African adults had ever
used mobile money, and fewer than 5 percent of adults in all other regions
had. By 2014 this percentage will likely have increased considerably. The
Global Findex data set will therefore provide a series of snapshots over
a tremendously exciting decade, during which the definition of financial
inclusion itself may change as new tools such as mobile phone–based sav-ings
accounts are further developed and adopted. Among other things, the
data will help advance research into how these new financial tools interact
with the more established tools and services.
The data set is a major advance, but there remains scope for improve-ment
in the next round of surveying. One such improvement would be
to add some measurement of “financial fragility.” In a recent Brookings
Paper, Annamaria Lusardi, Daniel Schneider, and Peter Tufano (2011)
examined this issue for U.S. households by looking at households’ capac-ity
to raise $2,000 in 30 days. The authors found that nearly half of the
households surveyed would probably not be able to do so. Adding a simi-lar
question to the Global Findex survey would enable researchers to
examine how financial inclusion correlates with financial fragility. In the
existing Global Findex survey, households are not asked whether they are
credit rationed, nor are they asked anything about the size of their current
savings. Asking people directly how much they have in savings may be too

Table 1. Survey Evidence on Financial Fragility in Kenya, Malawi, and Uganda
Percent of respondents
sensitive and prone to underreporting, but asking whether and how they
could access a given sum (which would have to be adjusted to the context,
for example by keeping the ratio to the local poverty line constant) would
be a great, if indirect, way to get a sense of how deep financial inclu-sion
is. In ongoing work, some colleagues and I asked such a question of
unbanked rural households in Kenya, Malawi, and Uganda between 2010
and 2012. The results, presented in table 1, show that most of the poor
households in our various samples have very limited savings, and that an
individual’s financial resources are to a great extent a function of the depth
of that individual’s social network.
The data reported in table 1 resonate with the present paper’s finding
that 65 percent of non–account owners mention “not having enough money
to use one” as one reason for not using a bank account (with close to
half of them reporting it as their only reason). Demirgüç-Kunt and Klapper
interpret this as evidence that “under current circumstances, the costs
of having an account outweigh its benefits” (emphasis in the original).
They also write that this finding “speaks to the fact that having a formal
account is not costless in most parts of the world and that individuals with
small or irregular income streams might view an account as an unnecessary
expense, given the relatively high cost.” I would like to qualify this inter-pretation.
Work that coauthors and I have done in Kenya and other parts of
Answera
“If you had an emergency that required [indicated
amount] urgently, where you would get the money?”
Kenya
(1,000 shillings
≈ $12)
Malawi
(1,000 kwacha
≈ $7)
Uganda
(10,000 shillings
≈ $5)
Would borrow from friends or
relatives
43 39 50
Would sell agricultural products 14 3 9
Would work more 14 21 9
Would sell assets 14 7 14
Would exclusively use savings 13 7 15
Would borrow from savings club 6 3 2
Would not be able to find the
0 18 0
money
Source: Household survey data collected by the author and Jonathan Robinson along with Anthony
Keats (Kenya, 2010), Dean Karlan, and Diego Ubfal (Malawi and Uganda, 2011) for ongoing projects.
a. Respondents could give more than one answer.

East Africa suggests that individuals reporting “not having enough money”
to use an account would not necessarily immediately start using accounts
provided to them completely free of charge. In Kenya, Jonathan Robinson
and I have found that very few bicycle-taxi drivers actively took up (that
is, made at least two deposits within a year in) accounts that they could
open at no cost to themselves, whereas about 40 percent of market ven-dors
did (Dupas and Robinson 2013a). More generally, only 18 percent of
a representative sample of unbanked households actively used accounts
that were free to open and maintain (Dupas and others 2012). Repli-cation
studies ongoing in Uganda and Malawi suggest rates no higher
than 30 percent. The fact that these accounts have withdrawal fees may
be part of the explanation for the low take-up, but many households do not
report fees as a barrier, instead simply stating that they do not have enough
money to save. But when provided with lockboxes (a simple metal box
with a deposit slit on top and a lock and key), the same Kenyan households
mentioned above made very regular deposits and saved in just 3 months
as much as would have taken them 18 months to save in an account. Thus,
households were in fact able to save more than they themselves thought
they could. This may be due to a feature that lockboxes offer that for-mal
accounts may not. In essence, access to these in-house savings tools
make people pennywise: they provide a place to store amounts that are
too small to warrant a trip to the bank, thus safeguarding funds that would
otherwise be kept close at hand and so be at risk of being frittered away
on unplanned small expenditures, such as sweets for the children or soda
for visitors.1
What this all means is that not having enough money to warrant a trip
to the bank to deposit it is itself a function of financial inclusion, defined
more broadly to encompass use of informal financial tools that facilitate
the day-to-day management of even very small sums, helping to grow them
into bankable lump sums. The stunning findings from the Global Findex
suggest that a better understanding of what type of tools can help unbanked
households save as much as they need in order to become “bankable” is an
important avenue for future research. Demirgüç-Kunt and Klapper have
provided the scientific community a much needed database and tool, and I
hope that they will make each wave of data easily available online through
a one-click download.
1. Dupas and Robinson (2013b) show that similar boxes enable households to reach a
given savings goal much faster.

references for the dupas comment
Angelucci, Manuela, Dean Karlan, and Jonathan Zinman. 2012. “Win Some Lose
Some? Evidence from a Randomized Microcredit Program Placement Exper-iment
by Compartamos Banco.” J-PAL working paper. Abdul Latif Jameel
Poverty Action Lab, Massachusetts Institute of Technology.
Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Cynthia Kinnan. 2013. “The
Miracle of Microfinance? Evidence from a Randomized Evaluation.” Working
paper. Massachusetts Institute of Technology and Northwestern University.
Beck, Thorsten, Aslı Demirgüç-Kunt, and Ross Levine. 2007. “Finance, Inequal-ity
and Poverty: Cross Country Evidence.” Journal of Economic Growth 12,
no. 1: 211–52.
Chaia, Alberto, Aparna Dalal, Tony Goland, Maria Jose Gonzalez, Jonathan
Morduch, and Robert Schiff. 2009. “Half the World is Unbanked.” Framing
Note. Financial Access Initiative. www.microfinancegateway.org/gm/document-
1.9.40671/25.pdf.
Crépon, Bruno, Esther Duflo, Florencia Devoto, and William Pariente. 2011.
“Impact of Microcredit in Rural Areas of Morocco: Evidence from a Random-ized
Evaluation.” J-PAL working paper. Abdul Latif Jameel Poverty Action
Lab, Massachusetts Institute of Technology.
Dupas, Pascaline, and Jonathan Robinson. 2013a. “Savings Constraint and
Microenterprise Development: Evidence from a Field Experiment in Kenya.”
American Economic Journal: Applied Economics 5, no. 1: 163–92.
———. 2013b. “Why Don’t the Poor Save More? Evidence from Health Savings
Experiments.” American Economic Review 103, no. 4: 1138–71.
Dupas, Pascaline, Dean Karlan, and Jonathan Robinson. 2013. “Expanding Access
to Formal Savings Accounts in Malawi, Uganda, Chile, and the Philippines.”
New Haven, Conn.: Innovations for Poverty Action. www.poverty-action.org/
project/0477.
Dupas, Pascaline, Sarah Green, Anthony Keats, and Jonathan Robinson. Forthcoming.
“Challenges in Banking the Rural Poor: Evidence from Kenya’s
Western Province.” In NBER Volume on African Economic Successes, edited by
S. Johnson, S. Edwards, and D. Weil. University of Chicago Press.
Honohan, Patrick. 2008. “Cross-Country Variation in Household Access to Finan-cial
Services.” Journal of Banking and Finance 32: 2493–2500.
Lusardi, Annamaria, Daniel Schneider, and Peter Tufano. 2011. “Financially Fragile
Households: Evidence and Implications.” BPEA (Spring): 83–134.
COMMENT BY
LILIANA ROJAS-SUÁREZ This paper by Aslı Demirgüç-Kunt and
Leora Klapper makes an important contribution to the literature by making
public, and providing the first analysis of data from, the Global Findex,
a new World Bank database comprising a variety of indicators on the use
of financial products by individuals around the world. The database was

constructed from 2011 survey data collected in interviews by Gallup, Inc.,
with selected adults in 148 countries.
Before Global Findex, the available data on the characteristics of pop-ulations
excluded from formal financial institutions remained scarce and
limited to a few regional efforts.1 Despite widespread recognition of
the welfare and efficiency benefits associated with improved financial
inclusion, and despite the large number of initiatives, public and private,
already in place around the world aiming to increase the percentage of the
population (households and firms) with access to financial services,2 cross-country
analyses faced severe constraints due to lack of comparable data.
In my view, one cannot overstate the importance of this new database.
Not only does it open up wide-ranging possibilities for future research, but
it also supports the efforts of policymakers, multilateral organizations, and
private donors, who now have a new tool to guide their policies and activi-ties
for improving financial inclusion. In this regard, the authors’ plans to
update the survey in a couple of years are of particular importance.
This paper is part of a series of analytical papers by the authors and
their colleagues that utilize the Global Findex database. The authors pres-ent
and analyze some key stylized facts derived from the survey, with a
significant focus on within-country differences in financial inclusion based
on individual characteristics. Their results are consistent with previous
(scattered) evidence, and from that perspective they validate a number of
policymakers’ concerns. For example, as expected, the authors find that the
percentage of individuals in developing countries who have an account at a
formal financial institution increases by income quintile; this is not the case
in most developed economies, where a large majority of the population at
all income levels have access to such financial products. Moreover, across
1. As the authors note, two of the best-known household surveys that include data on
the use of financial services are FinScope, a private sector initiative funded by the U.K.
Department for International Development, which collects data for 14 African countries and
Pakistan, and the European Bank for Reconstruction and Development’s Life in Transition
Survey, which covers 35 countries in Europe and Central Asia. To these may be added the
recent survey by CAF Banco de Desarrollo de América Latina (2011) that covers 17 major
Latin American cities.
2. Ongoing initiatives go beyond microfinance activities and include innovations to
improve the use of payments, savings, and insurance products. Two of the best-known initia-tives
are Kenya’s M-PESA money transfer service (operated by a mobile phone provider)
and the nonbank correspondent model in Brazil, which allows banks to reach remote popula-tions
through the use of existing nonbank networks, such as retail stores and post offices.
A common characteristic of these two initiatives is that they rely heavily on technological
advances in connectivity.

the developing world, the percentage of adult women with an account at
a formal financial institution is significantly below that for men, and this
gender gap persists across income levels (quintiles) within a given country.
The Global Findex data can be used for many other kinds of analysis. My
own research has already benefited from the availability of this new data-base.
Like the authors, I am interested in understanding the determinants
of the use of financial products, and I have gained some further insights by
focusing on the cross-country behavior of variables at the macro (aggre-gate)
level. These findings, which I will summarize here, complement the
authors’ results.
The existing literature allows one to identify four categories of obstacles
to financial inclusion at the country level, which affect either the demand
for or the supply of financial services or both (see Rojas-Suárez and
Gonzales 2010 and Rojas-Suárez and Amado forthcoming): socioeconomic
constraints, macroeconomic factors, characteristics of the operations of
the formal financial system, and institutional deficiencies. Here I will pres-ent
and discuss some simple correlations between these obstacles and one
of the Global Findex indicators of financial inclusion, namely, the percent-age
of the adult population with an account at a formal institution.
With regard to the first category, it is generally expected that coun-tries
that score high on indicators of social development, such as access
to high-quality health and education services, will also enjoy high levels
of financial inclusion. As discussed by Stijn Claessens (2005), financial
exclusion is often part of a broader social exclusion, which is related,
among other factors, to differences in education, type of employment, and
training. Income inequality has also been cited as a socioeconomic factor
influencing financial inclusion. Yet another factor, one that affects both
the demand for and the supply of financial services, is the percentage of
the population classified as middle class (see Rojas-Suárez and Amado
forthcoming for further discussion).
My figure 1 shows the correlation between financial inclusion and an
index of social development, constructed by equally weighting the first
two of the three components of the Human Development Index (HDI) of
the United Nations, which relate to health and education (the third relates
to income). The figure shows a strong positive correlation: as expected, the
developed countries display the highest values of both financial inclusion
and social development. The very high correlation between our social
development variable and GDP per capita (0.84) supports the authors’ find-ing
that the latter explains a significant part of the variation in account
penetration across countries.

Figure 1. Financial Inclusion and Social Development
Percent of adults with a formal financial account
Sources: Global Findex data and United Nations Development Programme (UNDP) data.
a. Constructed as a simple average of the health (life expectancy) and education components of the
2011 Human Development Index of the United Nations Development Programme. The line in the figure
is the best-fit line from a linear regression in which the percent of adults with a formal financial account
is the dependent variable.
Turning to the second category affecting financial inclusion, namely,
macroeconomic factors, I show in figure 2 the correlation between the
volatility of inflation (measured as the coefficient of variation of inflation
during 1990–2010) and financial inclusion. High volatility of inflation cap-tures
well the adverse effects of macroeconomic instability on the willing-ness
of the population to hold accounts in formal financial institutions.
In economies with very high and volatile inflation, depositors have expe-rienced
significant losses in their real wealth. It is therefore not surprising
that Argentina and Ukraine, both of which suffered from hyperinflation in
the 1990s, are among the countries with the lowest rates of financial inclu-sion.
By contrast, in Thailand, which has a history of low inflation volatil-ity,
the share of the adult population with an account at a formal financial
firm is similar to that in the developed countries.
The policy lesson is straightforward: stimulating demand for financial
services requires that individuals have trust that the real value of their pay-ments
and savings instruments will be preserved. If this trust is lacking,
90
80
70
Developing countries
60
50
40
30
20
10
0.3 0.4 0.5 0.6 0.7 0.8 0.9
Social development (1.0 = most developed)a

90
80 High-income countries
70
Ukraine
Source: Based on Rojas-Suárez and Amado (forthcoming) using Global Findex data and International
Monetary Fund (World Economic Outlook) data.
a. The line in the figure is the best-fit line from a logarithmic regression in which the percent of adults
with a formal financial account is the dependent variable.
not only will the use of financial services remain dismal, but deposits in
financial institutions will tend to be short-term as depositors stand ready to
withdraw their funds at the first sign of financial system difficulties.
A wide variety of characteristics pertaining to financial firms’ conduct
of their operations are included in the third category of obstacles to finan-cial
inclusion. Among these are inefficiencies in collection and information
processing, which may cause prohibitively high documentation require-ments;
insufficient numbers of branches, ATMs, points of sale, and other
forms of financial firms’ penetration, especially in small rural communi-ties;
and high administrative costs, which tend to increase the fixed costs
of extending loans and maintaining accounts. The authors have discussed
this type of obstacles extensively both in this paper and elsewhere (see, for
example, Allen and others 2012).
As an example at the country level, figure 3 shows the negative correla-tion
between financial inclusion and a commonly used measure of banking
system inefficiency: the ratio (in percent) of bank overhead costs to total
assets. As expected, developed economies display the lowest ratios.
Another potential obstacle to financial inclusion relates to the concen-tration
of the banking system. High levels of bank concentration may deter
60
50
40
30
20
10
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Coefficient of variation in annual inflation, 1990–2011
Thailand
Argentina
Figure 2. Financial Inclusion and Inflation Volatilitya

Figure 3. Financial Inclusion and Financial System Inefficiency
90
80
70
60
50
40
30
20
10
5 10 15 20 25
Bank overhead costs as a percent of total assets, 2009
Source: Based on Rojas-Suárez and Amado (forthcoming) using Global Findex data and Fitch’s BankScope
data.
a. The line in the figure is the best-fit line from a logarithmic regression in which the percent of adults
with a formal financial account is the dependent variable.
banks from lending to individuals and to small and medium-size enter-prises,
since there are no competitive incentives to assess the quality of
relatively riskier potential borrowers. Since the extension and repayment
of bank loans are usually conducted through deposits in bank accounts,
this argument could find support in a negative correlation between bank
concentration and the share of adults with a formal financial account.
However, recent studies have argued that, in a given country, the rela-tionship
between bank concentration and financial inclusion is strongly
affected by the quality of its institutions—the fourth category. (See, for
example, Claessens 2005 and Rojas-Suárez and Amado forthcoming.)
Financial systems can develop more fully and reach a larger segment of the
population in countries with adequate observance and enforcement of the
rule of law, political stability, and respect for creditors’ and debtors’ rights.
In particular, when contracts between creditors and debtors are observed
and enforced, depositors have a stronger incentive to entrust their savings
to banks and other formal financial institutions. How do bank concentra-tion
and the quality of institutions interrelate to affect financial inclusion?
In countries with weak institutions, where the enforcement of contracts is
very difficult, the oligopolistic power arising from a highly concentrated
banking system leads to greater discrimination against riskier borrowers
(who tend to be low-income individuals and smaller firms), and financial

inclusion suffers. Such discrimination is not as commonly seen in a more
competitive banking system.
Figure 4 illustrates these relationships. The top panel shows a nega-tive,
but low, correlation between bank concentration, measured as the
percentage of total system assets held by the country’s three largest
banks, and financial inclusion (the correlation coefficient is only -0.2).
However, a very different picture arises in the bottom panel, where the
bank concentration variable is adjusted by an additional variable measur-ing
the quality of institutions. This variable, called “rule of law,” is taken
from the World Bank’s Worldwide Governance Indicators and measures
agents’ confidence in and commitment to abiding by the rules of the soci-ety;
the quality of contract enforcement, police, and the courts; and the
likelihood of crime and violence. (I have rescaled the original variable
to range from zero to 100.) The adjusted bank concentration variable is
obtained simply by multiplying it by the rule of law variable.
Taken together, the two panels of figure 4 suggest that although bank
concentration might impinge on financial inclusion directly, it exerts its
most important effect through the quality of institutions. Two examples
will clarify this point. First, it is clear from the top panel that the devel-oped
countries are distributed across the whole range of bank concentra-tion,
and thus little can be said about any differences in concentration
between developed and developing countries (except that the negative rela-tionship
in the figure is driven by the latter). However, when the bank con-centration
variable is adjusted by the quality of institutions (bottom panel),
most of the developed economies migrate toward the upper right corner of
the scatterplot. For this group of countries as a whole, the relatively high
level of institutional quality seems a more relevant factor than bank con-centration
for understanding the behavior of financial inclusion.
Second, consider Malaysia and Nicaragua. These two developing coun-tries
share similar ratios of bank concentration (top panel), but the quality
of institutions is much higher in Malaysia than in Nicaragua. Therefore,
when bank concentration is adjusted for institutional quality, Malaysia lies
well to the right of Nicaragua. This is fully consistent with a higher level of
financial inclusion in the former country than in the latter.
Although correlations such as these provide valuable insights, fully
understanding the obstacles to financial inclusion requires a deeper analysis.
The first column of my table 1 reports results of an ordinary least
squares regression in which financial inclusion (as defined above) is the
dependent variable. Consistent with the discussion above, I include as
explanatory variables the five variables that appear on the horizontal axes

Figure 4. Financial Inclusion and Bank Concentration, with and without Adjustment
for Institutional Quality
Unadjusted
90
80
70
60
50
40
30
20
10
Malaysia
Nicaragua
High-income
countries
Developing
countries
30 40 50 60 70 80 90
Bank concentration (percent)b
Adjusted for institutional quality
90
80
70
60
50
40
30
20
10
Nicaragua
Malaysia
1000 3000
2000 4000 5000 6000 7000 8000 9000
Bank concentration rule of lawc
Source: Based on Rojas-Suárez and Amado (forthcoming) using data from Global Findex, Fitch’s
BankScope, and the World Bank.
a. The line in the figure is the best-fit line from a linear regression in which the percent of adults with
a formal financial account is the dependent variable.
b. Share of total banking system assets held by the country’s three largest banks.
c. The bank concentration variable in the top panel is multiplied by the rule of law measure from the
World Bank’s Worldwide Governance Indicators, rescaled from zero to 100 with higher values indicating
greater strength of the rule of law.

Table 1. Obstacles to Financial Inclusion: Implied Contributions from
Regression Analysisa
Contributionb (percentage points)
Independent variable
Regression
coefficient
Developed
economies
Emerging
markets
Other
developing
countries
Social underdevelopment -36.29*** -4.1 -8.2 -14.5
Bank concentration -0.29*** -17.3 -17.1 -17.8
Bank concentration × rule of law 0.004*** 19.2 10.1 7.1
Volatility of inflation -3.11** -2.9 -4.9 -4.0
Ratio of bank overhead costs to
assets -28.33 -0.8 -1.5 -1.3
Dummy for developed
economy 37.94*** 37.94 0 0
Dummy for emerging market 10.61** 0 10.61 0
Constant 55.3***
Source: Author’s regressions.
a. The dependent variable is the percentage of the adult population with a bank account in a formal
institution. Asterisks indicate statistical significance at the ***p 0.01 or the **p 0.05 level.
b. Measured as the regression coefficient for the indicated obstacle multiplied by that obstacle’s
average value in the indicated country group.
in figures 1 through 4. In addition, the regression includes two dummy
variables: the first identifies developed economies (again, those classified
by the World Bank as high-income countries), and the second emerging
markets (those classified as upper-middle-income countries).3 The regres-sion
was estimated for a sample of 116 countries.
Except for the coefficient on the bank overhead costs variable (which
had a p value of 0.18), all the estimated coefficients were statistically
significant. The regression’s R2 was 0.8359. For each country category, the
last three columns of table 1 show the implied contributions of the various
obstacles to financial inclusion, calculated by multiplying each variable’s
estimated coefficient by the variable’s average value.
To illustrate how to interpret the table, consider the group “other devel-oping
countries.” According to the table, if all factors affecting financial
inclusion identified in the regression were absent, this group would enjoy,
on average, a financial inclusion ratio (a share of adults holding a formal
financial account) of 55 percent (the value of the constant). With the obsta-cles
present, however, the predicted financial inclusion ratio for this group
3. The third group of countries (omitted in the regression) consists of all other develop-ing
countries (those classified as lower-middle-income and low-income countries).

of countries reaches only 25 percent. Similarly, absent the identified obsta-cles,
66 percent of the adult population in emerging markets would have
an account in a formal financial institution (the sum of the constant and the
coefficient on the dummy for emerging markets). However, because of the
obstacles, the predicted ratio is only 43 percent.
The most important conclusions to be drawn from the table are as
follows. First, the degree of social development matters greatly. The low
level of social development in developing countries on average, and to
a lesser extent in emerging markets, hampers financial inclusion through
both demand and supply factors. Second, only in the developed economies
does the high level of institutional quality at least partly offset the adverse
effect of bank concentration on financial inclusion (the sum of the implied
contributions for the bank concentration variable alone and for the bank
concentration × rule of law interaction is positive). In contrast, in emerging
markets and other developing countries on average, low institutional qual-ity
cannot counteract the financial exclusion effects of bank concentration.
Third, relative to the other factors, inflation volatility and banking ineffi-ciencies
play smaller roles as obstacles to financial inclusion.
My discussion here provides just a taste of the potential uses of the
Global Findex database and is intended as a complement to the authors’
analysis and their ongoing research. The authors are to be congratulated not
only for this paper but for their rich research agenda on financial inclusion.
references for the rojas-suárez comment
Allen, F., A. Demirgüç-Kunt, L. Klapper, and M. S. Martinez Peria. 2012. “The
Foundations of Financial Inclusion: Understanding Ownership and Use of
Financial Accounts.” Policy Research Working Paper no. 6290. Washington:
World Bank.
CAF Banco de Desarrollo de América Latina. 2011. Servicios Financieros para el
Desarrollo: Promoviendo el Acceso en América Latina. Reporte de Economía y
Desarrollo series. Caracas. publicaciones.caf.com/publicacion?id=1502.
Claessens, Stijn. 2005. “Universal Access to Financial Services: A Review of the
Issues and Public Policy Objectives.” Presented at the OECD-World Bank Fifth
Services Experts Meeting, Paris (February).
Rojas-Suárez, Liliana, and Maria Alejandra Amado. Forthcoming. “Improving
Access to Financial Services in Latin America: Policy Implications and Lessons
from Worldwide Experiences.” Washington: Center for Global Development.
Rojas-Suárez, Liliana, and Veronica Gonzales. 2010. “Access to Financial Services
in Emerging Powers: Facts, Obstacles and Policy Implications.” Paris: OECD
Development Center (March) www.oecd.org/dev/pgd/45965165.pdf.

GENERAL DISCUSSION While sympathetic to nonstandard ap-proaches
based on behavioral economics, information asymmetry, and
the like, Christopher Carroll wondered whether the problem of finan-cial
inclusion might not be better addressed using the textbook approach
that assumes perfect rationality. Looking at the list of countries with low-est
participation in the banking sector—Argentina, Greece, and Italy, for
example—was enough to suggest that many people around the world
might have good reason not to hold bank accounts. Financial inclusion,
Carroll thought, might turn out to be a good overall indicator of the quality
of a country’s institutions: low inclusion could correlate with the degree to
which a society and its institutions are dysfunctional. If that was the case,
efforts to increase the penetration of bank account ownership would not
address the underlying cause of noninclusion.
Donald Kohn added that any deliberate effort to increase financial
inclusion would surely bring in people who are not well educated and
who lack familiarity with financial products. Encountering a sophisticated
modern financial system for the first time, these individuals might not
understand the risks they are taking or, worse, might suffer exploitation.
Kohn observed that this danger was not limited to developing countries:
the subprime episode in the United States could be viewed as an attempt
at financial inclusion that ended badly. The U.S. response has been mainly
to increase disclosure and transparency, but there have also been propos-als
aimed, in a sense, at disinclusion by restricting the types of finan-cial
instruments available to the general public. Kohn asked whether the
authors had investigated whether the newly banked individuals in their
samples actually understood what they were getting, particularly on the
credit side, and whether their findings pointed to any measures that could
be taken to improve their understanding.
David Romer, also following up on Carroll’s comment, cited some
specific findings in the paper that could be interpreted as rational behavior
on the part of the nonincluded. For example, the paper reported that a large
fraction of the nonincluded chose not to have a bank account because they
had too little money to make it worthwhile. That seemed to Romer a plau-sibly
rational response. Others said that they had a relative with a bank
account, so that in effect they did have access to financial services even
though they were counted as excluded. Still others cited long distances to
the nearest bank branch as a reason for not having an account. Was that a
market failure, or was it an equilibrium outcome? These questions needed
to be sorted out, Romer argued, before useful policy interventions could
be proposed. Romer also asked the authors to clarify their distinction

between an individual’s explicit choice of how much to save and the stan-dard
definition of the same individual’s saving as simply income minus
consumption.
Ricardo Reis suggested that the present degree of inclusion in the
banking system in some countries might actually be too high, in the sense
that many people have bank accounts but lack access to other financial
services that are arguably more important to their welfare, such as insur-ance
against catastrophic shocks and vehicles for retirement saving.
Following up on Rojas-Suárez’s discussion and Kohn’s comment, Reis
reminded the Panel that in the United States for much of the 19th century,
a bank was an extremely risky place to keep one’s savings: the history
of banks in that largely unregulated era was rife with fraud. If bank
regulation today in some developing countries is comparably weak, that
argued against pushing for greater inclusion. On a similar note, Reis
observed that, according to the paper, 78 percent of Greeks still had
bank accounts in 2011, well into that country’s financial crisis. That
indicated to him a failure on the account holders’ part to appreciate the
risks they were taking.
For Benjamin Friedman, one of the paper’s most interesting findings
was that many people lack bank accounts because they do not trust the
banks. He suspected, however, that their distrust could arise for different
reasons. One might be the fear of bank fraud or recklessness that Reis
had mentioned, but another might be fear of government expropriation:
many Cambodians, for example, are wary of banks because they remem-ber
when the Pol Pot regime closed all the country’s banks overnight and
simply expropriated all the accounts. An entirely different potential moti-vation
was the fear that the bank would not keep one’s affairs secret. One
needed to distinguish among these different reasons before deciding what
sort of policy intervention was called for.
Michael Klein pointed out that people might choose to remain unbanked
for transactional reasons as well as because of lack of savings or for other
reasons. In countries where only cash is widely accepted, the transactions-related
features of a bank account—checkwriting and debit cards, for
example—would have little value.
Justin Wolfers remarked that the literature on financial access seemed
to take it as given that financial access was a good thing, and indeed, to
those who have it, it clearly is. But its benefits might be less obvious to
those who have never had it. Wolfers also identified what he saw as a pos-sible
problem with the authors’ empirical strategy. The paper claimed to
be measuring access to financial services, meaning the ability to use those

services if one wanted to, but what it really measured was an equilibrium
quantity: those who, given the equilibrium price of holding a bank account,
and taking into account all costs, actually used one. Just as not everyone
who has access to spinach at the supermarket eats spinach, so, too, not
everyone who has access to banking services opens a bank account.
Responding to the discussion, Leora Klapper answered Romer’s ques-tion
by saying that she and her coauthor were interested not in the quan-tity
of saving as such but in the behavioral decision to save: what people
did with the money that they were deliberately putting aside for a specific
purpose, such as for a large purchase or for retirement. What the authors
found was that these savings were typically placed in informal ROSCAs
(rotating savings and credit associations) or under the mattress. Since the
former are plagued by fraud and the latter is extremely unsafe, it seemed
to the authors natural to suppose that a formal bank account, or at least
a lockbox at a bank, was a better place to hold that money. Klapper also
noted that in some countries such as Bangladesh, microfinance institu-tions
are important sources of formal savings and credit, used by many
people who do not have bank accounts.
Klapper also acknowledged that the histories of some countries, includ-ing
some in Eastern Europe, provided their citizens ample reason to be
distrustful of banks, and that lack of understanding of bank terms and
conditions could result in exploitation or cause people to make poor use of
their accounts. As an example, the failure of small savers to understand the
concept of a minimum balance often leads to their savings being eroded.
Klapper therefore emphasized that greater financial literacy was a neces-sary
complement to sound regulation and consumer protection: it could
contribute not only to broadening the use of safer financial services, but
also to building trust in the financial system in countries with a history of
financial scandal and expropriation.
Aslı Demirgüç-Kunt sought to clarify that the paper was not claim-ing
or assuming that more people around the world should have bank
accounts. Rather, she and Klapper were interested in identifying the per-ceived
barriers to holding bank accounts among those who might want
to use them. They believed that if market failures are indeed preventing
broader access to bank accounts, the result was to limit people’s ability
to save for education or for other worthwhile purposes, and to limit busi-nesses’
access to capital for growth and expansion. In presenting their
findings to policymakers, they took pains to emphasize the importance of
responsible access. It was true that a majority of their respondents chose
not to hold bank accounts because they did not have enough money to

make it worthwhile, and some fraction of these presumably would see no
need for an account even if they had more money. But a large fraction of
the rest—30 percent of the total sample—identified other barriers such as
high costs, remoteness, or instruments whose design does not conform
with the potential user’s religious beliefs. These were issues that a well-crafted
policy might be able to remedy.

2013a klapper

More Related Content

What's hot (17)

Similar to 2013a klapper (20)

2013a klapper