Measuring financial inclusion: explaining variation in use of financial services across and within countries.

Demirguc-Kunt, Asli ; Klapper, Leora

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 savings and credit mechanisms may facilitate investment in productive activities such as education or entrepreneurship. Lacking such access, individuals rely on their own limited, informal savings to invest in their education or 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 levels 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 initiatives 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) Qualitative 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, productivity 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 experiments 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 groups such as the poor are excluded from formal financial systems. Systematic 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 indicators 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 countries, confirming the findings of previous studies that show lower use of formal financial services in developing countries (see, for example, Beck, Demirguc-Kunt, and Martinez Peria 2007 and Cull, Demirguc-Kunt, and Morduch 2013). For instance, the share of adults in high-income countries 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 financial 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 characteristics 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.

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 certain 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 individuals with small or irregular income streams might view an account as an unnecessary expense, given the relatively high cost. Other reasons commonly 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 deliberately 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 countries 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 originated a new loan from a formal financial institution in the past 12 months, while 23 percent report borrowing from family and friends. But in developing 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 reasons 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 summarizes 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 inclusion. 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 financial 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 financial 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 saving 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)

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 survey 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 satisfaction 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 Demirguc-Kunt and others (2013) to study changes in trust in banks in the wake of the global financial crisis; and by Stevenson and Wolfers (201 l) 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 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 commonly 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 countries 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 insignificantly different from or higher than those of the FinScope surveys, perhaps because of the difference in timing (most of the FinScope surveys were carried 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 services (financial institutions), the Global Findex data may fill a gap by going beyond data collected only from regulated financial institutions and allowing 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 population; in these countries the survey is conducted face to face. (11) In most countries 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 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 available) 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 [+ or -] 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 informal 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 frequency 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 focusing 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 differs enormously between high-income and developing countries in the aggregate: 89 percent of adults in high-income countries, but only 24 percent in 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 majority 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 negligible 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 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 regressions 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).

[FIGURE 2 OMITTED]

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 controls for GDP per capita (column 1-3 in table 1). However, large amounts of credit--whether commercial or consumer credit--in a financial system 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 OMITTED]

[FIGURE 4 OMITTED]

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 penetration 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 countries 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 residence. 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 percent, 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 correlation across countries (a correlation coefficient of 0.42) between inequality 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 (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 penetration in the poorest income quintile (figure 5). In the United States 26 percent of adults in this group report having no formal account; the corresponding 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 holders 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 common 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 proportion 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 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 statistically significant, but there is no statistically significant difference among developing-country income groups (column 1-5 in table 1).

[FIGURE 6 OMITTED]

Like account penetration, formal saving behavior also varies with individual 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 percent. 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 quintile, 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 individuals who have a formal account may not necessarily use it to save. Worldwide, about 43 percent of account holders report having set aside money at a formal financial institution in the past year; the figure varies relatively 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 countries 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 possible that accounts set up by an employer or the government are not conducive 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 pandero in Peru. These clubs generally operate by pooling the weekly deposits of their members and disbursing the entire amount to a different member each week. Although members generally do not earn interest on their deposits as in a formal account, these clubs can provide members an opportunity 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 savings 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, informality, 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 disbursements 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 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 private 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 (column 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 Finland, 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 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.

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

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 population 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 countries, 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 borrowing only among developing countries (rightmost panel of table 2). On average, 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 countries, 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 developing countries highlight that emergency and health needs are the most common reason for having an outstanding loan (figure 9). (19) Adults in the 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.

[FIGURE 9 OMITTED]

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 European 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 differences in housing finance systems, such as in product diversity, types of lenders, secondary mortgage markets, and degree of government participation. 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 existence of formal property rights and an efficient framework to record ownership of property (de Soto 2000).

[FIGURE 10 OMITTED]

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 examine these self-reported barriers by country income group and individual 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.

[FIGURE 11 OMITTED]

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, documentation, 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.

[FIGURE 12 OMITTED]

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 respondents 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 maintaining 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, Demirguc-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 countries (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 distance 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)

[FIGURE 13 OMITTED]

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 expropriation 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 financial 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.