89804 Access to Finance Forum Reports by CGAP and Its Partners No. 6, June 2013 Financial Access 2012 Getting to a More Comprehensive Picture Oya Pinar Ardic, Kathryn Imboden, and Alexia Latortue a Acknowledgments This report was written by a team from CGAP and IFC. We are grateful that the IMF continues to conduct its annual Financial Access Survey. We are immensely appreciative of the thought partnership and review of Leora Klapper and Douglas Randall for Chapter IV, as well as for the precious advice of Marten Leijon and Scott Gaul. We also thank Nina Bilandzic and Jasmina Glisovic for their work on Chapter III. Goran Amidzic and Alexander Massara helped in analyzing the data from the Financial Access Survey. Camilo Tellez and Michael McCord provided the content and reviewed boxes on branchless banking and microinsurance—thank you. We greatly value the careful review and insights provided by Jeanette Thomas, Timothy Lyman, Peer Stein, Peter Wrede, and Vijayasekar Kalavakonda. Anna Nunan edited the report. Finally, the support of our partners is indispensable. This report was produced with the financial support from the Netherlands’ Ministry of Foreign Affairs to IFC and from AusAid to CGAP. © CGAP and International Finance Corporation, 2013 All rights reserved. CGAP 1818 H Street, N.W. Washington, DC 20433 USA Internet: www.cgap.org Email: cgap@worldbank.org Telephone: +1 202 473 9594 This volume is the product of the staff of the Consultative Group to Assist the Poor (CGAP) and staff of IFC (International Finance Corporation), a member of the World Bank Group. The material in this work is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. CGAP and IFC does not guarantee the accuracy, reliability or completeness of the content included in this work, or for the conclusions or judgments described herein, and accepts no responsibility or liability for any omissions or errors (including, without limitation, typographical errors and technical errors) in the content whatsoever or for reliance thereon. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of CGAP or IFC concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to CGAP at the address in the copyright notice above. CGAP and IFC encourage dissemination of their work and will normally give permission promptly and, when reproduction is for noncommercial purposes, without asking a fee. Permission to photocopy portions for classroom use is granted through the Copyright Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, MA 01923 USA. Contents Foreword 3 Executive Summary 5 CHAPTER I. Evolution of the State of Financial Inclusion— What Commercial Bank Data Tell Us 8 The State of Access to Insurance—Preliminary Data 17 CHAPTER II.  The State of Access to Finance by SMEs—An Update 22 CHAPTER III.  Putting Supply- and Demand-Side Data Together— CHAPTER IV.  What FAS and Global Findex Tell Us 27 Linking Financial Access Indicators to Economic Development and CHAPTER V.  Financial Systems Development 31 Annexes Principal Financial Inclusion Data Sources 37  The G-20 Basic Set of Financial Inclusion Indicators and ATMs— Latest Available Figures 38 FAS: Definitions and Data Availability 39 References 42 1 What Is FAS? The IMF’s Financial Access Survey (FAS) is the most compre- Monetary and Financial Statistics and its accompanying hensive source of global supply-side data comparable across Compilation Guide. countries and over time. The latest round of FAS was con- • In 2012 the FAS questionnaire was expanded to include ducted by the IMF Statistics Department in 2012 in collabora- times series data on credit unions and financial cooperatives tion with CGAP and IFC Access to Finance Advisory. and MFIs, with separate identification of small and medium FAS facilitates the analysis of trends in access to deposits, enterprises (SMEs), households, and life and nonlife insur- loans, and insurance by households and enterprises over time ance providers. and across countries, and by type of financial service provider. The FAS database includes geographic and demographic in- • FAS is one of the three global data sources for the G-20 dicators on access to and usage of basic financial services at Basic Set of Financial Inclusion Indicators endorsed by the an annual frequency for 187 jurisdictions, including all G-20 G-20 Leaders at the Los Cabos Summit in June 2012. economies, covering an eight-year period (2004–2011). FAS data are publicly available at http://fas.imf.org. IMF’s • FAS collects data from country financial regulators, primar- standard data query (eLibrary) and visualization (DataMapper) ily central banks. FAS’s methodology is based on IMF’s tools also include FAS data. Foreword W e have seen remarkable progress toward national Finance Corporation (IFC), and the World increasingly robust global financial inclu- Bank. The Basic Set integrates existing global data sion data architecture in a few short years. sets to track financial inclusion around the world. The IMF’s Financial Access Survey (FAS) and (See Annex 2 for the latest available figures for the the World Bank’s Global Financial Inclusion Data- G-20 Basic Set of Financial Inclusion Indicators.) base (Global Findex) represent major global and While the G-20 Basic Set aims to provide an es- comprehensive supply- and demand-side datasets. sential, yet simple picture of the financial inclusion At the national level, countries such as Brazil, Mexi- landscape, additional indicators are needed to pres- co, and Malaysia are investing heavily in defining fi- ent a more comprehensive view. The GPFI Data and nancial inclusion indicators that need to be collect- Measurement Sub-Group is developing additional ed, monitored, and analyzed to help advance indicators beyond the Basic Set to capture more ac- financial inclusion. Last year, the Irving Fischer cess, usage, and quality dimensions of financial in- Committee on Central Bank Statistics of the Bank of clusion. Progress thus far is creating even more de- International Settlements convened central bankers mand for increasingly detailed and segmented data. and other experts for its first meeting ever on finan- And this is a good thing. cial inclusion data. This report goes beyond simply measuring access National and global policy makers, funders, and to finance. It analyzes the underlying market dynam- private-sector providers all stand to benefit from ics and linkages to broader financial sector and eco- this progress. And while stakeholders may priori- nomic growth indicators. A better insight into the tize different kinds of data, there is general agree- drivers of access and the relevance of financial access ment that supply- and demand-side data are com- to broader financial architecture and the real econo- plementary, and both are important to create a my is critical for informed policymaking. meaningful picture of access and usage of financial As organizations focused on financial inclusion services globally. (See Annex 1 for the principal fi- policy and the on-the-ground development of ro- nancial inclusion data sources.) bust provider ecosystems, we find that our interests At the Los Cabos Summit in June 2012, the G-20 can and indeed must come together on behalf of the Leaders endorsed the G-20 Basic Set of Financial In- billions of poor people still without access to re- clusion Indicators (G-20 Basic Set). The G-20 Basic sponsibly delivered financial services. Credible, Set was developed by the Global Partnership for Fi- sound, and comprehensive data provide us with a nancial Inclusion’s (GPFI) Data and Measurement better foundation to ensure we are each playing our Sub-Group and its Implementing Partners the Alli- appropriate roles for “Financial Inclusion for All” to ance for Financial Inclusion (AFI), CGAP, the Inter- become a reality. Tilman Ehrbeck Peer Stein CEO and Director Director CGAP IFC Advisory Services | Access to Finance 3 Executive Summary H alf of the adult population around the world Expanded data coverage provides a broader does not have an account at a formal finan- and better understanding of market dynamics cial institution. And 75 percent of poor peo- This year’s report is an important step in leveraging ple are unbanked (Demirgüç-Kunt and Klapper the FAS data to provide an up-to-date and in-depth 2012). Yet, research shows that people have active view of the state of financial inclusion today, com- financial lives and need a range of services to take plementing other supply- and demand-side data advantage of economic opportunities and manage initiatives in data collection and analysis. Available and mitigate risks (Collins, Morduch, Rutherford, data have expanded every year, though they remain and Ruthven 2009). Expanding access to a broad far from complete. range of financial services to households and enter- As regulators work to meet the financial inclu- prises has never had such strong momentum among sion needs in their countries, they increasingly rec- national policy makers and global standard setters. ognize the importance of information on the supply To create inclusive financial systems that serve of financial services. Regulators are expanding their more people with a range of services at lower cost, data collection efforts in this area, beginning with a diverse set of providers, supporting financial in- data on the deposit-taking regulated institutions frastructure, and protective and enabling policies they oversee while also starting to include data will all be needed. As the financial landscape has from other financial service providers. This broader become more complex with an ever wider array of view of all institutions that serve poor people un- providers and delivery channels, the data architec- derpins improved policy design. ture for capturing financial inclusion is also evolv- Chapter I presents a trend analysis of the evolv- ing. To have a comprehensive view of access to fi- ing state of financial inclusion using commercial nancial services—payments, savings, credit, and bank data. Across all markets, not surprisingly, data insurance—data must be collected from more within the purview of the financial regulator, the sources, and on a variety of dimensions, including source for the IMF’s FAS, are of the highest quality access, usage, and quality. for commercial banks. Whether in the front-line of Each year, the Financial Access report aspires to service delivery or as partners to mobile network include data on an increasing number of financial operators or refinancers of microfinance institu- service providers, based on data availability. Finan- tions, commercial banks are likely to be an increas- cial Access 2011 was an exception that provided an ing part of the broad ecosystem of providers that overview of the supply-side data landscape with will help extend financial services to previously un- little fresh data. Financial Access 2012 builds on the served or underserved client segments. Moreover, work done in Financial Access 2009 and Financial previously specialized pro-poor financial institu- Access 2010 to provide new data on financial access. tions are transforming into commercial banks, and Using eight years of data (2004–2011) from the local commercial banks are showing new interest in IMF’s Financial Access Survey (FAS) in combina- reaching the base of the pyramid. tion with other relevant data, Financial Access 2012 The evolution in deposit and loan penetration further contributes to measuring and analyzing the shows a clear, albeit still nascent, recovery from the current state of financial inclusion. (See Annex 3 for financial crisis. Over the eight-year period of 2004– FAS definitions and data availability.) 2011, the number of deposit and loan accounts per 5 1,000 adults continued to rise steadily, though with are beginning to catch up from a low base. While very little growth in 2009 and 2010 and an upturn high-income countries experienced very little in 2011. Interestingly, the growth rate of deposit ac- growth, the average annual growth rate in the num- counts was slightly higher than the growth rate for ber of policies for the lower-middle-income coun- loan accounts, following a pattern of deposit pene- tries was 9 percent in 2011. Life insurance is the dom- tration increasing faster than loan penetration es- inant service provided (this is confirmed by sources tablished in 2008. Coming out of the financial crisis, complementary to FAS, such as the Latin America people’s willingness to save may have been greater and the Caribbean and Africa microinsurance land- than their willingness to take on loans. Also, the scaping studies). The 2008 financial crisis did not af- slow-down of lending in the crisis environment, es- fect life insurance policies, presumably because they pecially given the banking crisis in high-income are longer-term contracts by nature, though it did countries and greater risk aversion from providers, negatively affect nonlife insurance policies as well as may also have been a contributing factor. reduce insurance technical reserves. The rapid (and consistent) increase in the num- Chapter III includes an overview of access to ber of bank branches and automatic teller machines finance by small and medium enterprises (SMEs). over 2004–2011 has also helped deepen access to Access to finance for managing cash flows, funding commercial bank services across all regions of the investments, and insuring against risk is one im- world. However, there is a wide dispersion in de- portant barrier for SMEs’ growth, alongside non- posit and loan penetration across regions and coun- financial barriers such as infrastructure. For this try income quartiles. High-income countries have reason, in 2012 FAS included questions on SMEs over 10 times the deposit penetration as low- for the first time. Though still incomplete, FAS data income countries, and lower-middle-income coun- show that higher-income countries tend to have tries have almost three times the deposit penetra- more developed SME finance markets than that of tion of low-income countries. Differences in loan developing countries as measured by ratios of SME penetration follow a similar pattern. Deposit pene- finance volume-to-GDP and SME loan accounts- tration, by commercial banks, is the lowest in sub- to-total firm loans. In low-income countries, only a Saharan Africa. Though from a low initial base, small percentage of enterprise loan accounts are growth in 2011 in commercial bank deposit ac- held by SMEs. counts comes from low-, lower-middle-, and upper- Chapter IV analyzes the complementary rela- middle-income countries, with high-income coun- tionship between demand- and supply-side data tries stagnating. and what purpose they each serve. Twenty-twelve Notwithstanding the role commercial banks are was a milestone year for financial inclusion data: increasingly playing, nonbank financial institutions the enhanced FAS released in September 2012 pro- (NBFIs) contribute significantly to reach unserved vides the most extensive supply-side data available and underserved clients in many markets. Indeed, to date, and Global Findex released in March 2012 deposit-taking NBFIs are playing a more important offers the most extensive demand-side data to date role in deposit and loan penetration, and NBFI loan on a global scale. FAS and Global Findex are by de- penetration increased relative to that of commer- sign complementary, and not substitutes. Supply- cial banks everywhere in the world, with the excep- side data surveys such as FAS offer a relatively low- tion of high-income countries. cost means of data collection, with frequent and Chapter II provides trend analysis based on pre- comparable data that are viewed as highly credible liminary data on access to insurance. Based on still by national authorities. Demand-side surveys such incomplete coverage, FAS data show that high- as Global Findex offer rich information on the many income countries account for the vast majority of the dimensions of financial inclusion, from the per- global insurance market, historically as well as today. spective of individuals. Conducted annually as a However, the number of insurance policies has more written survey sent to financial regulators, the unit than doubled since 2004, and low-income countries of analysis for FAS is regulated financial institu- 6 tions. Conducted triennially through interviews Finally, Chapter V explores the links between fi- with individuals, Global Findex gathers data on the nancial inclusion and financial sector and macro- usage of financial services from regulated, unregu- economic variables. Although such analyses con- lated, and informal institutions. tinue to be works in progress, some interesting The two surveys generally tell similar stories of results have emerged. FAS data show that greater financial inclusion, though they do not necessarily financial inclusion (measured by deposit penetra- give the exact same number for data points at the tion) correlates with higher income levels (GDP per country level. For loans, it is to be expected that FAS capita and GDP per capita growth) and a reduction and Global Findex do not have similar results be- in income inequality. Higher financial inclusion is cause FAS asks for “all outstanding loans” while associated with less inequality, though a certain de- Global Findex asks for “all loans taken in the past 12 gree of financial access and usage and financial sec- months.” For deposits, the story is more nuanced. tor depth is required before inequality improves; Countries with low income levels and less devel- for a country with low levels of financial inclusion oped financial systems are more likely to have simi- and financial depth, inequality increases at first, lar FAS and Global Findex rankings. For many coun- then decreases as the financial system becomes tries where FAS and Global Findex tell divergent deeper and more inclusive. deposit stories, FAS data show greater inclusion. While the theoretical (and intuitive case) for Policy makers and regulators make use of FAS to linking responsible financial inclusion and financial understand the offer of financial services by institu- stability is strong, demonstrating empirical evi- tions under their purview. FAS can help provide an dence is a challenge. A growing body of literature understanding of the market structure, pointing to suggests a positive relationship between financial strategies to work with different kinds of financial inclusion and financial stability; however, empirical institutions to increase access. A deeper under- evidence does not yet confirm this. Although finan- standing of the profiles of users through Global Fin- cial stability overall has a low correlation with ac- dex can lead to more access-friendly policies, legis- cess, depth, and efficiency, financial access and fi- lation, and regulation, potentially targeting the nancial stability correlate better in low-income and groups that are most underserved or are priorities lower-middle-income countries, where access is- for the government. Providers, as well as donors sues are more acute. and investors, can deepen their understanding of Lastly, greater financial inclusion is associated client profiles and behavior via Global Findex, in- with more developed financial infrastructure, and a cluding client segments that are persistently under- sounder institutional and legal environment. A served. Both FAS and Global Findex can be used for stronger business environment is linked to greater benchmarking across countries and are endorsed deposit and loan penetration. data sources for the G-20 Basic Set. 7 I CHAPTER Evolution of the State of Financial Inclusion— What Commercial Bank Data Tell Us A n increasingly broad range of financial ser- A clear, albeit still nascent, recovery from vice providers of all legal statuses offers fi- the financial crisis nancial services to poor people. In more and The evolution of deposit and loan penetration tends more countries, previously specialized pro-poor fi- to move together both within country income nancial institutions are transforming into commer- groups and regional groups. Both deposit and loan cial banks, and local commercial banks are showing penetration growth rates—measured by the growth new interest in reaching the base of the pyramid. in the number of accounts or accountholders of de- Moreover, whether in the front-line of service deliv- posit/loan per 1,000 adults—slowed considerably in ery or as partners to mobile network operators 2008 and started to pick up after the financial crisis. (MNOs) or refinancers of microfinance institutions The growth rate of loan and deposit accounts per (MFIs), commercial banks are likely to be an in- 1,000 adults recovered in 2010, and even more creasing part of the broad ecosystem of providers strongly in 2011. The growth rate of deposit ac- that will help extend financial services to previously counts was slightly higher than the growth rate for unserved or underserved client segments. loan accounts, following a pattern of deposit pene- Notwithstanding the role commercial banks are tration increasing faster than loan penetration increasingly playing, nonbank financial institu- established in 2008 (Figure 1). This reversed a pre- tions (NBFIs) contribute significantly to reach un- vious three-year period (2005–2007) of loan pene- served and underserved clients in many markets. tration increasing more rapidly than deposit pene- Though still far from providing a complete picture tration. of nonbank providers, the 2012 FAS provides in- The world as a whole added about 48 deposit creased coverage of these types of institutions that accounts and 24 loan accounts per 1,000 adults in serve unserved and underserved people. 2011, which amounts to 1,314 deposit accounts and This chapter focuses on data on commercial 264 loan accounts per 1,000 adults. This suggests banks. Across all markets, not surprisingly, data that coming out of the financial crisis, people’s within the purview of the financial regulator, the willingness to save may have been greater than source for the IMF’s FAS, is of the highest quality their willingness to take on loans. Also, the slow- for commercial banks, allowing for technical down of lending in the crisis environment, espe- (econometric) analysis. The analysis in this chap- cially given the banking crisis in high-income ter tracks the evolution of commercial bank de- countries and greater risk aversion from provid- posits, loans, and commercial bank physical access ers, may also have been a contributing factor. as well as trends by country income group and re- The change in deposits-to-gross domestic prod- gion using FAS data. To complement the commer- uct (GDP) led that of loans-to-GDP, reversing the cial bank data, available data on NBFIs are pre- 2005–2007 trend of loans-to-GDP increasing more sented in Box 1. Another important development rapidly than deposits-to-GDP (Figure 2). The re- is the new ways in which MNOs and other service cent trend in the growth rate of outstanding loans- providers are rapidly reaching an increasing num- to-GDP and deposits-to-GDP may stem from one or ber of clients. Unfortunately, data collection ef- both of the following factors: (i) a greater variation forts have not yet caught up with the rapid expan- in loan and deposit volumes as a post-financial crisis sion of mobile-based financial services. See Box 2. effect (higher loan amounts in 2011) and (ii) the lim- 8 Box 1 The Growing Importance of Nonbank Financial Intermediaries In many parts of the world, specialized financial institutions Deposit-taking NBFIs are playing a more important role in with an explicit focus on poor people or specific segments of deposit and loan penetration. Burundi, for example, reported the unserved/underserved have emerged alongside com- a larger number of deposit accounts and loan accounts in cred- mercial banks. These NBFIs classified in FAS as “nonbanks” it unions than in commercial banks. Based on the latest data are either (a) “other depository corporations,” including cred- available in FAS, the ratio of commercial bank deposit accounts it unions and financial cooperatives, deposit-taking MFIs and to NBFI deposit accounts started to decline especially after the other deposit-taking institutions (which include saving and financial crisis around the world, except for low-income and loan institutions, building societies, rural banks, agricultural lower-middle-income countries. The role of NBFIs is even more banks, savings banks, and post banks) or (b) “other financial prominent in loan penetration. NBFI loan penetration increased corporations,” which are not deposit-taking.a relative to that of commercial banks everywhere in the world, These NBFIs often play an important role in the financial with the exception of high-income countries (see Figure B1.A). system, and many have explicit financial inclusion mandates. Total number of deposit-taking NBFIs in the world remained However, the financial regulator often has only partial purview more or less constant at around 40,000 throughout 2004– of them. For this reason, the data on NBFIs in FAS are not as 2011, while the total number of commercial banks declined complete as for commercial banks. This box provides an over- from around 16,000 in 2004 to 14,200 in 2011. As of 2011, view of the role of the NBFIs in financial inclusion, drawing on about half the deposit-taking NBFIs are in high-income coun- available FAS data, which quite certainly under-represents the tries. The remaining institutions are divided among upper-mid- role they play in many markets. dle-, lower-middle-, and low-income countries, at 16 percent, 14 percent, and 20 percent, respectively. FIGURE B1.A. Deposit and loan penetration of deposit-taking NBFIs (relative to commercial banks) Commercial bank deposits/NBFI deposits Commercial bank loans/NBFI loans 15 15 12 12 9 9 6 6 3 3 0 0 2004 2005 2006 2007 2008 2009 2010 2011 2004 2005 2006 2007 2008 2009 2010 2011 World Upper Middle Low World Upper Middle Low High Lower Middle High Lower Middle Note: Loan data for lower-middle-income countries do not have sufficient coverage to calculate pre-2007 commercial bank loans-to-NBFI loans ratio. See Annex for the classification of financial institutions used by FAS. a.  9 figure 1 figure 2 Percentage change in commercial bank Percentage change in commercial bank deposit and loan accounts (annual medians) deposit and loan volumes (annual medians) .15 .1 .08 .1 .06 .05 .04 .02 0 0 2005 2006 2007 2008 2009 2010 2011 2005 2006 2007 2008 2009 2010 2011 # deposits/1,000 adults # loans/1,000 adults Outstanding deposits/GDP Outstanding loans/GDP ited GDP growth, including lower GDP growth in and with inequality (the greater the income equali- 2011, indicating that a recovery of the financial sec- ty, the greater the degree of financial inclusion). tor may have preceded that of the real economy. The 2011 increase in outstanding loans and deposits to GDP may thus be explained by higher loan Deposits amounts and lower real GDP growth in 2011 (with the exception of a very slight increase in real GDP Over the eight-year period of 2004–2011, growth in sub-Saharan Africa [SSA], where the low deposit accounts per 1,000 adults rose steadily growth rate in 2010 in loan volume to GDP is attrib- Trend analysis shows renewed growth in deposits utable to higher real GDP growth rates, which were in 2011 (Figure 3). The number of deposit accounts not sustained in 2011). per 1,000 adults rose steadily, albeit with very little growth in 2009 and 2010, but with an upturn in Wide dispersion in deposit and loan penetration 2011. The number of depositors per 1,000 adults Across regions, South Asia (SA) and SSA experi- also increased over the period, but with a slightly enced the largest average increases in deposit pen- different trajectory: the number of depositors etration in 2011. A look at differences across coun- dipped in 2007 and then recovered from 2008 on- try income quartiles (country ranking by income) ward. On average, the number of depositors per indicates that high-income countries have over 10 1,000 adults increased at a faster rate than deposit times the deposit penetration as low-income coun- accounts per 1,000 adults, suggesting that overall tries and lower-middle-income countries have al- there are more people/enterprises in the formal fi- most three times the deposit penetration of low-in- nancial system. While the recovery from the global come countries. Differences in loan penetration financial crisis is still considered fragile (World follow a similar pattern. We shall see in Chapter V Bank Global Financial Development Report 2013), that both deposit and loan penetration are corre- the 2011 data are nevertheless encouraging in terms lated with GDP per capita (the greater the GDP per of penetration figures for deposits. capita, the greater the degree of financial inclusion) 10 Box 2 A New Frontier—Capturing Data on Branchless Banking and Mobile Money The Opportunity financial inclusion datasets that cover some aspects of branchless banking and mobile money. They are IMF’s Banks, MNOs, and other financial service providers FAS, the World Bank Global Payments Survey, and the are finding new ways to deliver financial services to GSMA Global Mobile Money Adoption Survey, which unbanked people. Rather than using traditional covers around 60 percent of the mobile money de- brick-and-mortar branches, they offer banking and ployments worldwide. payment services through postal and retail outlets, including grocery stores, pharmacies, seed and fer- Challenges tilizer retailers, and gas stations, among others. Vari- ous models of branchless banking through retail Even though some branchless banking data are al- agents have emerged: some led by banks and oth- ready captured in these surveys, a large gap still ex- ers led by nonbank commercial actors using infor- ists. For example, there is no consensus around how mation and communication technologies, such as the services are defined and distinguished in terms of cell phones, debit and prepaid cards, and card read- the channel used for the provision of financial services ers to transmit transaction details from the retail and the actual service provided. A lack of clarity in agent or customer to the bank and provide cash-in thinking about channels and financial services can and cash-out points for customers. lead to double counting of services. As branchless banking deployments start to In addition, the current surveys do not completely reach scale, it will be increasingly important to in- capture the landscape. The GSMA Global Mobile clude them in the wider data architecture to get a Money Adoption Survey, for example, provides deep complete picture of financial inclusion. Financial ser- information on mobile money deployments, but vice providers, the source of supply-side data, track does not capture card-based deployments that serve the services they deliver as part of their everyday the poor through agents. Moreover, the data are business. They should start systematically tracking publicly available only at the market level and not at numbers of registered and active branchless bank- the deployment level, as there is currently little incen- ing and mobile money agents, registered and active tive for MNOs to make their figures public. Likewise, mobile money customers, and other types of trans- the fact that MNOs are usually regulated by the tele- actions, such as person-to-person transfers or insur- communication regulator in many countries makes it ance premium payments made over a mobile plat- challenging for a survey such as the IMF’s FAS to cap- form. ture the full spectrum of branchless banking provid- Mobile money data are still missing from many ers. FAS goes to regulators who collect data on finan- datasets or are poorly represented, and there is a cial service providers such as banks and MFIs, but lack of common definitions and indicators. However, excludes other providers such as MNOs and third- efforts are underway to improve the situation. There party providers, including data from their respective are currently three global/multicountry supply-side agent networks. 11 Growth in 2011 in commercial bank deposit and 2011 in the number of deposit accounts in coun- accounts comes from low-, lower-middle-, and tries such as Azerbaijan (+28 percent), Panama (+16 upper-middle-income countries percent), Peru (+16 percent), and Venezuela (+17 The difference between high-income countries and percent). At the same time, deposit account penetra- the rest of the world in terms of commercial bank tion increased in low- and lower-middle-income deposits per 1,000 adults is striking (Table 1). Over countries, but there is still a large gap. While the 15.3 time, deposit accounts per 1,000 adults declined in percent increase in deposits in low-income coun- high-income countries from five-plus accounts per tries is impressive, it is based on a far lower starting adult to around four accounts per adult. The con- point, and low-income countries still lag far behind. tinuing economic and financial crisis in Europe Deposit penetration also varies greatly within seems to be a factor. The greatest growth in deposits the country income level categories (Figure 4). For is in the upper-middle-income countries, slightly example, there is very low deposit account penetra- passing the deposit penetration of high-income tion in the Democratic Republic of Congo (20 ac- countries. There was strong growth between 2010 counts per 1,000 adults) and Afghanistan (88 ac- counts per 1,000 adults). However, Kenya has 611 accounts, Bangladesh 539 accounts, and the Gam- figure 3 bia 320 accounts per 1,000 adults. Individuals in Number of deposit accounts per 1,000 adults low-income countries may use nonbanks to save (e.g., cooperative and community-based structures Commercial bank deposits (annual medians) in SSA). This may influence not only the compari- 1,000 son with higher-income groups, but also the disper- sion within the low-income category. 800 As with deposit account penetration, deposit 600 volumes to GDP show a consistent increase 400 across country income groups In a comparison of country income groups (Figure 200 5), the variation of deposit volumes to GDP within the same income group is larger for high-income 0 2004 2005 2006 2007 2008 2009 2010 2011 and upper-middle-income countries. This is partly # deposit accounts/1,000 adults # depositors/1,000 adults due to variation in GDP (including the financial cri- sis effect in these two income groups) and in part to differences in deposit volume. Low-income coun- Table 1 Deposit accounts/1,000 adults, by income groups Deposit accounts/ 1,000 adults World Low Lower Middle Upper Middle High 2004 1124 240 529 774 5205 2005 1162 238 623 908 5086 2006 1152 256 626 984 4975 2007 1170 256 656 1045 4765 2008 1200 275 696 1132 4127 2009 1208 248 752 1138 3912 2010 1266 288 802 1270 3897 2011 1314 332 863 1322 3878 Increase in 2011 +3.8% +15.3% + 7.6% + 4.1% 0.4% 12 figure 4 figure 5 Number of depositors per 1,000 adults Deposit volume per GDP across and within income groups 6 8 Natural log of number of depositors per 1,000 7 Natural log of deposit/GDP 4 6 2 5 4 0 3 -2 Low Lower Upper High Low Lower Upper High Middle Middle Middle Middle tries have low deposit volume relative to GDP be- Vietnam in the lower-income segment (with a cor- cause of (i) less developed financial sectors in low- responding lower number of deposits per GDP). income countries (less depth) and/or (ii) more use of nonbanks for deposits, notably in SSA. Loans Deposit penetration varies widely across and within regions Loan penetration varies considerably across SSA is the region with the lowest number of ac- income groups counts per 1,000 adults on average and has the As with deposits, there is a direct relationship be- greatest variation within the region. This is tied to tween loan penetration and country income levels a low-income effect within SSA (and income dis- (Figure 6):1 the higher the income level, the greater parity within the region), as well as differences in the number of loan accounts and the greater the use of commercial banks as opposed to nonbanks number of borrowers. for savings. For example, Burundi had 32 deposit accounts per 1,000 adults in commercial banks, Number of loan accounts per 1,000 adults while it had a total of 124 deposit accounts per continued to increase globally over 2004–2011, 1,000 adults in credit unions and financial coopera- with declines in 2009 and 2010 due to the tives, deposit-taking MFIs and other deposit-tak- financial crisis ing institutions. The trend line over the period is steepest for the The variation in deposit volume is the largest in upper-middle-income countries, with upper-mid- East Asia and the Pacific (EAP). This may be ex- plained by the large disparity in GDP per capita in Boxes show the observations between the 25th and the 75th 1.  EAP countries, e.g., Singapore, Hong Kong, and Ma- percentile, with the line in the middle of the box showing the median for each group. The extent of whiskers shows the nor- laysia in the higher-income zone (with a corre- mal range for each group, while observations that fall outside sponding higher number of deposits per GDP) and the normal range, if any, are represented as dots above (and Cambodia, Indonesia, Mongolia, Timor Leste, and below) the highest and lowest points of the whiskers. 13 figure 6 figure 7 Number of borrowers/1,000 adults in Commercial bank loan accounts 2011—Across and within income groups per 1,000 adults (by income groups) 8 Number of borrowers/1,000 adults (log) 600 6 500 400 4 300 200 2 100 0 0 2004 2005 2006 2007 2008 2009 2010 2011 Low Lower Upper High World Low Lower Middle Upper Middle High Middle Middle dle-income countries catching up with and passing times of crisis or credit retrenchment. In low-in- high-income countries in 2010 (Figure 7). For ex- come countries, short loan maturities mean that ample, the growth rate for loan accounts per 1,000 there is a higher loan turnover, and a decline in new adults was striking for Argentina (average annual loan issues would be reflected much faster. The growth rate of 22 percent for the period 2004–2011, shorter maturity of loans in low-income countries with no change in 2009, reaching 615 in 2011), Ven- makes FAS data closer to flow data in these coun- ezuela (28 percent per year, also with a small de- tries, although by definition, FAS data are stock data cline in 2009, reaching 494 in 2011), and Peru (16 (i.e., FAS asks for outstanding loans). Thus, the ef- percent annual growth, albeit with slower growth fect of a crisis on lower-income countries may be in 2009, reaching 247 in 2011). The upward slope is overstated in relation to effects shown in higher- less steep for lower-income countries, and the num- income countries. ber of loan accounts per 1,000 adults dips down- ward for low-income countries. Over the same pe- The post crisis upswing in the number of riod, percentage changes in SSA are high while the borrowers came a year earlier than the number of accounts per 1,000 adults in 2011 re- renewed increase in loan accounts mains very low (Rwanda, 9 up from 0.5; Burundi, The number of loan accounts per 1,000 adults con- 6.6, up from 2.6; and the Gambia, 37.5, up from 8.8). tinued to drop in 2010, whereas the number of bor- Differences in loan maturities across income rowers per 1,000 adults turned upward in 2010. The groups may help explain the differences in trends relatively fewer loans per borrower in 2010 (Figure displayed in Figure 7. In high-income countries, 8) suggest caution in the number of loans a borrow- loan maturities are longer, even up to 30 years for er is willing to take, or banks are willing to issue. mortgages, while in low-income countries, loan maturities tend to be much shorter, usually less Variations in loan penetration across and than one year. The longer loan maturities in high- within regions are striking, but as expected income countries mean that it takes much longer to High-income countries have the highest loan pen- register changes in the overall number of loans in etration and numbers of borrowers, while SSA has 14 figure 8 figure 9 Commercial bank loans (annual medians) Number of borrowers/1,000 adults in 2011—Across and within regions 250 8 Number of borrowers/1,000 adults (log) 200 6 150 4 100 2 50 0 0 2004 2005 2006 2007 2008 2009 2010 2011 EAP ECA HI LAC MNA SA SSA # loan accounts/1,000 adults # borrowers/1,000 adults the lowest. Less-developed regions have greater Despite the positive growth in low- and lower- dispersion within the region, SSA in particular. middle-income countries in both ATM and branch Overall, the results correspond to the income lev- networks throughout the period, there is still a els of regions (Figure 9). On average, the differ- large gap between lower-income and upper-mid- ences across regions are less in terms of outstand- dle- and high-income countries. The distribution of ing loans (as percentage of GDP), while dispersion ATMs is strikingly different among the country in- within regions is higher in developing countries. come quartiles. The world as a whole had 47 ATMs and 17 commercial bank branches per 100,000 adults in 2011. Low- and lower-middle-income Physical Outreach countries had 3.2 and 13.1 ATMs per 100,000 adults in 2011, respectively, while this figure is 76 in upper- Measured by the rapid (and consistent) in- middle-income countries and 123 in high-income crease in the number of bank branches and countries. automatic teller machines (ATMs) over 2004– There is considerable inequality across regions 2011, access to commercial bank services has and within regions in terms of commercial bank deepened across all regions of the world branches per 100,000 adults (Figure 11). Low- and Overall, the number of commercial bank branches lower-middle-income countries had 3.8 and 9.6 and ATMs increased steadily during 2004–2011. commercial bank branches per 100,000 adults, re- The trend lines between ATMs and bank branches spectively, in 2011. In contrast, upper-middle and are similar (Figure 10), with a slowdown in the cri- high-income countries had 26 and 34, respectively. sis years and picking up in 2010. Not surprisingly, SSA saw the largest growth in branch network, ATMs grew slightly faster than bank branches, as from a low base, with an average of 6.8 branches per ATM expansion is less costly in terms of infrastruc- 100,000 adults, with many countries having consid- ture. Rapid ATM expansion is linked to branch ex- erably fewer branches, including Malawi, Tanzania, pansion (i.e., a dual effect of more branches and Ethiopia, the Democratic Republic of Congo, and more ATMs per branch.) Sierra Leone. 15 figure 10 figure 11 ATMs and commercial bank Commercial bank branches per 100,000 branches (annual medians) adults, regional dispersion, 2004 and 2011 40 40 30 30 20 20 10 10 0 0 2004 2005 2006 2007 2008 2009 2010 2011 EAP ECA HI LAC MNA SA SSA ATMs/100,000 adults Com. banks branches/100,000 adults 2004 2011 16 II CHAPTER The State of Access to Insurance—Preliminary Data I nsurance is broadly recognized alongside pay- plete. Recent landscape studies on microinsurance ments, deposits, and loans as one of the four main help to provide a deeper understanding of the use categories of financial services. Insurance services of insurance in the low-income market. See Box 3. are helpful to manage and mitigate risk, which is es- pecially important for poor households and small High-income countries account for the vast businesses that are particularly vulnerable to shocks. majority of the global insurance market, With the help of insurance, individuals and firms can historically as well as today pass risks on to others, building resilience against di- On average, there were 21.5 insurance corporations sasters, whether personal or large scale. There is also per country in the world in 2011, which amounts to evidence that insurance can help improve invest- less than one (0.6) per 100,000 adults. In developed ment opportunities and returns, enabling individuals countries, this average is as high as 120 per country, and firms to borrow more and grow. corresponding to approximately nine insurance While informal risk transfer mechanisms are corporations per 1 million adults, while in SA, prevalent in many parts of the world, formalization where penetration is the lowest, it is four per coun- is seen as important to ensure the soundness of the try or six per 10 million adults (see Figure 12). The insurance and reinsurance industries, facilitate small island economies in Latin America and the risk-pooling across geographies, and provide ade- Caribbean (LAC) have a large number of insurance quate consumer protection.2 corporations relative to their population, e.g., Baha- Insurance data in FAS include the number of cor- mas (48 per 100,000 adults), followed by Grenada porations, number of policyholders, number of poli- and St. Kitts and Nevis (with 30 per 100,000 adults cies, and technical reserves. Information on policies each). With the exception of the Middle East and and technical reserves are available for life and non- North Africa (MENA), the number of insurance life policies both as a combined number and segre- corporations declined on average across all regions gated. Life insurance is any form of insurance whose and income groups throughout the period. The payment is contingent upon the death of the insured number of insurance corporations alone, however, (e.g., term life insurance) or whose payment is con- is insufficient to analyze the level of insurance de- tingent upon the survival of the insured to or beyond velopment in a market. a specified point in time (e.g., endowment insurance and annuities). Nonlife insurance includes provision The number of insurance policies increased of insurance services other than life insurance: ac- throughout 2004–2011 cident and fire insurance; health insurance; travel In 2011, on average, there were 709 insurance poli- insurance; property insurance; motor, marine, avia- cies per 1,000 adults in the world, and 269 insur- tion, and transport insurance; and pecuniary loss ance policyholders per 1,000 adults.3 The number and liability insurance. of policyholders more than doubled since 2004. While FAS is the major source of globally com- parable data on the number of insurance policies, The number of insurance policyholders refers to the number 3.  of life and nonlife insurance policyholders that include only policyholders and insurance technical reserves, nonfinancial corporations (public and private) and house- data coverage on insurance is still far from com- holds. A policyholder is a person or an entity that pays a pre- mium to an insurance company in exchange for the coverage provided by an insurance policy. Number of insurance policies 2.  The “Application Paper on Regulation and Supervision Sup- refers to the number of insurance policies held by nonfinancial porting Inclusive Insurance Markets” of the International As- corporations (public and private) and households. An insur- sociation of Insurance Supervisors (IAIS) is straightforward ance policyholder may have multiple insurance policies (life, on this point: “in the event that informal services exist, then health, property, etc.). Also, the FAS survey is not able to dis- formalisation is needed” (IAIS 2012, p. 6). tinguish between individual and group policies/policyholders. 17 figure 12 Number of insurance corporations Median number of insurance corporations, 2011 Median number of insurance corporations, 2011 150 60 100 40 50 20 0 0 EAP ECA HI LAC MNA SA SSA Low Lower Upper High middle middle The 2008 financial crisis negatively affected figure 13 nonlife insurance policies in particular Figure 13 shows that, throughout 2004–2011, over- Number of insurance policies over time all the number of total insurance policies and the number of nonlife policies per adults increased, Number of insurance policies/1000 adults 3,000 though both decreased following the crisis and then picked back up. In contrast, the number of life in- 2,500 surance policies per adult globally remained more 2,000 or less constant throughout 2004–2011 around one 1,500 policy for every two adults. As life insurance poli- 1,000 cies are longer-term contracts by nature compared to nonlife insurance policies, it is intuitive that the 500 nonlife insurance sector was most affected by the 0 2004 2005 2006 2007 2008 2009 2010 2011 financial crisis. Insurance policies/1,000 adults There is a big difference in insurance penetra- Life insurance policies/1,000 adults Nonlife insurance policies/1,000 adults tion between high-income economies and developing economies Although insurance penetration—measured by in- High-income countries have an average of 1,723 in- surance policies per 1,000 adults—almost doubled surance policies per 1,000 adults, while the low-in- in developing economies throughout 2004–2011, come countries have an average of 203. However, there is still a big gap between high-income econo- low-income countries are beginning to catch up. mies and developing economies. While there were While high-income countries experienced very lit- almost 2.5 insurance policies per adult in high-in- tle growth, the average annual growth rate in the come economies throughout the period, there were number of policies for the lower-middle-income only 0.5 policies per adult in developing economies countries was 9 percent. in 2011 (Figure 14). 18 figure 14 figure 15 Insurance policies per 1,000 adults: Insurance technical reserves across High-income vs. developing economies income groups Number of insurance policies/1,000 adults Insurance technical reserves (%GDP) 30 2,500 25 2,000 20 1,500 15 1,000 10 5 500 0 2004 2005 2006 2007 2008 2009 2010 2011 0 2004 2005 2006 2007 2008 2009 2010 2011 World High income Low income High-income Developing Lower middle income Upper middle income The 2008 financial crisis reduced insurance surance sectors and where the financial crisis hit technical reserves the hardest, a decline in insurance technical re- FAS defines insurance technical reserves as net serves-to-GDP in lower-middle-income countries equity of households in life insurance and prepay- (60 basis points) can also be observed. The effect on ments of insurance premiums and reserves low-income countries was in the form of stagnation against outstanding claims. As FAS does not cover rather than a decline. pension funds, net equity of households in pen- sion funds is not included here. In essence, insur- Life insurance dominates in terms of volumes ance technical reserves amount to the difference The median number of global life insurance poli- between the present value of expected future pre- cies made up 21 percent of total insurance policies miums and the present value of expected claims in 2011, up one percentage point from 2005. Despite and expense payments, as well as provisions for the small portion of total policies, the median level claims in course of settlement or expected to have of global life insurance technical reserves account- occurred but have not been reported. A major part ed for nearly 70 percent of total technical reserves of life insurance can be considered savings of in both 2005 and 2011. households.4 For the number of policies, policyholders, and Figure 15 plots the trend in insurance technical technical reserves, the proportion of life and non- reserves (relative to GDP) for the world, as a whole life insurance to total insurance varied greatly as well as for the four country income groups over across regions and income groups. Life insurance 2004–2011. Although the dip in 2008 is more visible policies as a percentage of total policies were high- for the world as a whole and for the high-income est in SA (63 percent) and SSA (49 percent) in and upper-middle-income countries (1.3 and 1.9 2011. Conversely, life insurance policies accounted percentage points, respectively) that have larger in- for only 14 percent of total policies in the MENA Note that a comparison of asset holdings instead of technical 4.  reserves would have be more relevant; however, data on insur- ance corporation assets are not available. 19 region for the same period (see Figure 16).5 Life in- surance policies ranged from 58 percent of total figure 16 policies in low-income countries to 20 percent in Life vs. nonlife insurance the upper-middle-income group (see Figure 5, up- policies by region (2011) per left panel). The preliminary data on insurance from FAS Graphs by region provide a useful starting point for understanding the dynamics of access to insurance globally. Yet more complementary data, such as asset holdings and subsegmentation by product type, are needed to draw deeper insights on the underlying charac- EAP ECA HI teristics of insurance markets globally and how in- clusive they are. 5. SA average uses data on percentage of life insurance policies in total policies from Bangladesh (93 percent) and Bhutan (33 percent) only; SSA average is based on Rwanda (79 percent), LAC MNA SA Seychelles (12 percent), and Mauritius (55 percent) only; and MENA average is based on Saudi Arabia (8 percent) and Tuni- sia (17 percent). Policies: Life Policies: Nonlife SSA Box 3 Microinsurance in Latin America and the Caribbean and Africa Two landscaping studies focusing on the low-income insur- • Around 90 percent of the organizations identified are for- ance market in LAC and in Africa complement FAS data and mally regulated providers, and they are the market leaders. also provide deeper insights on the development of insur- • The growth of insurance in the low-income market in LAC ance markets in the two regions, with a focus on the low-in- has happened largely without donors and regulatory induce- come market.* ments, which has given the region a different character than Africa or Asia. The result of this maturing insurance market is Key Findings from the LAC Microinsurance Landscape a positive mix of broader outreach, a greater variety of distri- Report bution, and a movement toward products that likely offer • Of the 45.5 million identified lives and properties covered, greater value to the low-income market. The positive devel- 71 percent had life insurance (excluding credit life), 53 per- opments there are likely to continue, with more insurers cent had some form of accident insurance, 35 percent had achieving profitability in working in low-income markets. credit life insurance, 23 percent had some form of health insurance, and 6 percent had property insurance. Key findings from the Africa Microinsurance Landscape • The five countries with the greatest number of insured Report (Mexico, Brazil, Colombia, Peru, and Ecuador) account for • The Africa landscape survey identified 44.4 million lives and just over 90 percent of all lives and property covered in properties covered at the end of 2011. More than 60 per- Latin America. Of 19 countries reporting microinsurance in cent of this coverage comes from South Africa. LAC, 55 percent of the people and properties covered • Although life products cover more lives than all other prod- were in Mexico and Brazil. ucts combined, most of the products reported were health 20 Box 3 (continued) products, due to the large numbers of mutuals and com- • There has been a significant increase in life coverage in munity-based organizations, primarily in West Africa. Africa, which is important, but not sufficient. Credit life in • Considerable regional differences in product outreach particular is often considered an entry product. But the exist: sector has not yet progressed from credit life to more complex products such as health and agricultural insur- – Southern Africa represents the majority of the popula- ance, both important for low-income people. tion covered by life and credit-life products because of the strong presence of funeral insurance. • New distribution channels, such as life insurance prod- – West Africa has the greatest number of people cov- ucts embedded into savings accounts and bundled into ered by health products, primarily because of strong mobile phone subscriptions, have helped to expand the donor promotion of group insurance policies through industry in terms of covered lives in the past two years. community-based health insurance programs in fran- These types of developments hold great potential to cophone Africa. dramatically increase coverage, though they also raise – East Africa has the greatest agricultural coverage important issues of consumer education, protection, and and, due to one large insurer, the greatest accident regulation. coverage. • Despite the large number of community-based organiza- • The region remains dominated by funeral coverage, the tions (77 percent of the organizations identified), regu- main driver of growth, with Southern and East Africa rep- lated commercial insurers account for 78 percent of the resenting the greatest number of lives and properties lives covered. The fact that commercial insurers are the covered. drivers of growth in the microinsurance industry has posi- tive implications for scale, profitability, and sustainability of microinsurance. FIGURE B3.A. LAC: Lives covered by types of FIGURE B3.B.Africa: Number of lives/ coverage properties by product 50 50 40 40 Lives and properties covered (including secondary) (MM) Lives cover, in millions 30 30 20 20 10 10 0 0 Total Credit life Life Health Accident Property 2008 2011 Agriculture Credit life Property Life & accident Health Source: McCord, Tatin-Jeleran, and Ingram (2012) and McCord et al. (2012). * The work for the two landscaping studies was executed by the Microinsurance Centre with support from the Multilateral Investment Fund for Latin America and Making Finance Work for Africa and the Munich Re Foundation for Africa (McCord, Tatin-Jaleran, and Ingram 2012; McCord et al. 2012). Both studies define microinsurance as “insurance that is modest in both coverage and premium levels based on the risks insured.” 21 III CHAPTER The State of Access to Finance by SMEs—An Update J obs are a central priority for policy makers in countries (Figure 17). Overall, higher-income coun- both developed and developing economies. tries tend to have higher ratios of SME finance vol- The focus on jobs has spurred strong interest ume-to-GDP (Figure 18, left panel) and SME loan in SMEs. At the end of 2010, total commitments of accounts to total firm loans (Figure 18, right panel), public funders to SMEs were around $24.5 billion suggesting a more developed SME finance market (Siegesmund and Glisovic 2011). The G-20 is also compared to developing countries. committed to improving access to finance for SMEs in developing countries, and the Global In low-income countries, only a small Partnership for Financial Inclusion (GPFI) has percentage of enterprise loan accounts are prioritized SMEs as one of its four priority topics.6 held by SMEs Recent studies suggest that SMEs contribute There are large variations across countries surveyed more to the employment share than large firms do, with regard to whether SMEs have a basic loan ac- and their contribution is larger in low-income count (Figure 19). For example, in India, Madagas- economies than in high-income countries (Ayya- car, and Georgia, a very small percentage of enter- gari, Demirgüç-Kunt, and Maksimovic 2011). Many prise loan accounts in commercial banks are held by barriers to SMEs’ growth exist, however. Some are SMEs (less than 20 percent). This is largely consis- nonfinancial barriers, including lack of infrastruc- tent with the recent evidence from the World Bank’s ture such as electricity.7 Access to finance for man- Enterprise Surveys. For example, the latest survey aging cash flows, funding investments, and insuring data from Madagascar (2009) indicate that 36 per- against risk is another barrier. Given the impor- cent of small firms and 46 percent of medium firms tance of SMEs to GDP and employment creation, identify access to finance as a major constraint.9 expanding SMEs’ access to formal banking services As mentioned previously, questions on SME ac- is critical. cess to finance were only recently added to the This chapter provides an overview of SME fi- FAS survey. Over time and as regulators improve nancing volumes provided by commercial banks us- reporting, FAS will have time series data on SME ing FAS data. Data on SME loan volumes were col- finance that will allow for a thorough trend analy- lected by financial regulators in only 37 out of the 187 sis. Box 4 is a first attempt to illustrate SME fi- economies surveyed by FAS in 2012 (Figure 17).8 As nance trends in Bangladesh, where the regulator 2012 was the first year FAS included questions on was able to provide historical data for the period SME loan volumes, the hope is that more countries 2004–2011 to FAS. will provide SME data in future rounds of FAS. Establishing common definitions and data The higher the GDP per capita, the greater standards for SMEs remains challenging the volume of SME lending There is no standard definition for an SME, and The share of SME loans as a percentage of total countries measure SMEs by different yardsticks. In loans in commercial banks varies greatly across addition, the heterogeneity of SMEs themselves— with highly varying sizes, levels of formality, capac- 6. http://www.gpfi.org/about-gpfi/sub-groups-and-co-chairs ity, and financial needs—makes it challenging to 7. See IFC (2012) for more information on the importance of standardize information across difference coun- nonfinancial services for SMEs. tries. This is compounded by the different levels of 8. Although SME definitions vary across countries, many central banks that submitted SME data to FAS used the World Bank definition to report the data. See Annex for further details. 9. www.enterprisesurveys.org. 22 SME lending by banks, % GDP (log) –2 0 2 4 50 40 10 20 30 0 4 Malta Ireland Korea, Republic of figure 18 figure 17 6 Thailand RWA Bosnia and Herzegovina MDG ZMB KEN GIN BGD Kosovo IRQ IND MNG Samoa IDN GEO Belgium KSV FJI 8 WSM TON BIH FSM MKD SLV THA NAM Macedonia, FYR RUS PER COL El Salvador MYS TUR VEN HUN CHL Malaysia URY Tonga MLT Real GDP per capita, $ (log) GRC 10 Greece SME lending as percentage of GDP BEL KOR IRL Macao SAR, China MAC Turkey Hungary SME lending by commercial banks (2011) Georgia 12 Bangladesh SME loan accounts at banks Micronesia, Fed. Sts. % NFC loan accounts, (log) Guinea 2 3 4 5 Chile 4 Russian Federation Peru Indonesia Namibia MDG 6 India Mongolia BGD IND Uruguay GEO Rwanda FJI 8 FSM MKD Fiji PER COL West Bank and Gaza DOM MYS HUN EST Colombia Venezuela, RB PRT Real GDP per capita, $ (log) 10 Iraq BEL Madagascar Zambia Kenya 12 23 knowledge, data, and measurement capacity within figure 19 different countries. Nonetheless, efforts to harmonize SME data SME loan accounts as percentage of definitions and practices to better measure, track, total firm loans in commercial banks (% nonfinancial corporation loan accounts and assess the state of access to finance for the in commercial banks) SME sector at the national and global levels are underway. The work of the GPFI Sub-Group on Portugal Estonia Data and Measurement is just one example. The Belgium Organisation for Economic Co-operation and De- Micronesia, Fed. Sts. velopment (OECD) has also developed an OECD Macedonia, FYR Malaysia Scoreboard that provides a comprehensive inter- Colombia national framework for monitoring SMEs’ and en- Peru trepreneurs’ access to finance over time (OECD Hungary Dominican Republic 2012). In addition, a number of ongoing efforts Bangladesh among development finance institutions aim to Brunei Darussalam streamline and harmonize SME definitions. Re- Fiji Georgia cently, IFC constructed a database of formal and Madagascar informal micro, small, and medium enterprises India (MSMEs) around the world to estimate global and 0 20 40 60 80 100 regional gaps in enterprise finance. Key findings of this study are summarized in Box 5. 24 Box 4 SME Lending Trends in Bangladesh Bangladesh is one of the few developing economies that pro- SMEs are important for growth and employment for the vided detailed responses on the SME-related questions of the Bangladeshi economy. SMEs constitute 90 percent of all indus- FAS survey, allowing for deeper analysis. Trend analysis of the trial firms in Bangladesh and generate 25 percent of GDP. Thir- FAS data indicates that the development of SME finance in ty-one million people are employed by SMEs, and SMEs con- Bangladesh has progressed significantly over time, with the tribute to 75 percent of household income (Abdin 2012). It is share of SME borrowers to nonfinancial corporation borrow- estimated that SMEs of 250 or fewer employees contribute to ers (enterprises other than financial institutions) increasing around 21 percent of employment in Bangladesh (Ayyagari, from 27 percent in 2004 to 50 percent in 2011. The number Demirgüç-Kunt, and Maksimovic 2011). The government has of SME loan accounts nearly tripled over the same period recognized the SME sector as an important driver of economic (Figure B4.A). development and employment creation, and also sees women- owned businesses as a priority (MIDAS 2009). FIGURE B4.A.Number of SME borrowers and number of SME loan accounts in commercial banks in Bangladesh Borrowers (1,000s) Loan accounts (1,000s) 800 1600 600 1200 400 800 200 400 0 0 2004 2005 2006 2007 2008 2009 2010 2011 2004 2005 2006 2007 2008 2009 2010 2011 Year Year Nonfinancial corporations SMEs Nonfinancial corporations SMEs Source: IMF FAS. 25 Box 5 Estimating the Global SME Finance Gap IFC recently conducted a study on the state of access to fi- • Value of credit gap: ~$0.9 trillion–1.1 trillion gap in credit nance for formal and informal MSMEs, based on data from for formal SMEs in developing economies (~26–2 percent of the World Bank’s Enterprise Surveys. The study defines current outstanding SME credit in developing economies) MSMEs as micro (1–4 employees), very small (5–9 employ- • Women-owned firms’ credit gap: ~63–69 percent of wom- ees), small (10–49 employees), and medium (50–250 employ- en-owned SMEs (5.3 million–6.6 million) in developing ees) enterprises. Key highlights from the study on SMEs are economies are either unserved or underserved, which as follow: amounts to a financing gap of ~$260 billion–320 billion. • Total number of SMEs: ~36 million–44 million formal • Regional variations: The gap relative to current outstanding SMEs globally and ~25 million–30 million formal SMEs in SME credit varies widely across regions, e.g., SSA and developing economies MENA require a greater than 300 percent increase in out- • Women-owned firms: ~31–38 percent (8 million–10 mil- standing SME credit compared to 7–8 percent and 25–30 lion) formal SMEs in developing economies have full or percent in East Asia, and Eastern Europe and Central Asia, partial women ownership respectively. • Total credit gap: 55–68 percent of formal SMEs (14 mil- lion–18.6 million) in developing economies are either un- served or underserved Source: IFC Enterprise Finance Gap Database 2011. The database will be publicly available on the SME Finance Forum at www.smefinanceforum.org. Methodology Note: Total number of firms (aggregate and different sizes) is based on data from national statistical offices, business registries, etc., and may include firms that are out of business. See IFC MSME Country Indicators (www.ifc.org/msmecountryindicators) and Kushnir, Mirmulstein, and Ramalho (2010) for details. World Bank Enterprise Surveys are conducted in developing economies, and not all countries are covered. Data for the noncovered countries are extrapolated based on regional averages. Based on the Enterprise Surveys database, IFC Enterprise Finance Gap Database develops four categories of constraint levels: (1) well-served; (2) underserved; (3) unserved; (4) no need. “Unserved” enterprises are those that are do not have a loan or overdraft but need a loan; “underserved” enterprises are those that have a loan and/or overdraft but have financing constraints. 26 IV CHAPTER Putting Supply- and Demand-Side Data Together— What FAS and Global Findex Tell Us T wenty-twelve was a milestone year for finan- late to each other, what they are best useful for, if cial inclusion data: The enhanced FAS re- and how they contradict each other, and if so why. leased in September 2012 provides the most This chapter seeks to help policy makers, practi- extensive supply-side data available to date, and tioners, and funders interested in financial inclu- Global Findex released in March 2012 offers the sion data better understand the two data sets, what most extensive demand-side data to date. they measure, and what they do not measure. It also Much attention—and high expectations—has compares the results of the two data sets, as part of been focused on the enhanced FAS and Global Fin- a matching exercise (analysis of the extent to which dex and how they can, together, assess the state of these two data sets match—and do not match—for a financial inclusion and help point to needed policy set of key variables) and offers hypotheses as to why reforms and market opportunities.10 Both data sets and when the data sets do and do not match. are international reference points in the evolving financial inclusion data architecture that can be used for benchmarking and are sources for the What is the difference between G-20 Basic Set of Financial Inclusion Indicators. FAS and Global Findex? Questions abound about how the two data sets re- FAS and Global Findex are by design 10.  Indicators developed from both FAS and Global Findex are complementary, and not substitutes expected to complement other sources, including thematic and country-led efforts. While FAS provides a supply perspective, Global Findex measures demand. One survey is not meant Box 6 to replace the other.11 Table 2 summarizes the dif- ferences between FAS and Global Findex on select Global Findex dimensions. FAS obtains data directly from financial regula- Global Findex is the new public demand-side tors through a comprehensive written survey. The database on financial inclusion developed by survey questions cover data that financial service the World Bank, with a 10-year grant from providers already report—or should be reporting— the Bill & Melinda Gates Foundation, and to central banks, other regulators, and supervisory implemented by Gallup, Inc., as part of the annual Gallup World Poll. The data are based agencies. Global Findex goes straight to the indi- on interviews with more than 150,000 nation- viduals using financial services through nationally ally representative adults in 148 economies. representative, individual surveys that are inter- Covering over 40 indicators, the database view based (face-to-face or on the telephone). Both provides insight into how people around the surveys share common features, such as compre- world save, borrow, make payments, and hensive country coverage allowing for cross-coun- manage risk. All indicators can be further re- try comparisons and public availability of data, fined by gender, urban, or rural residence, as which underpin the robustness of both FAS and well as age, education, and income. See Global Findex. www.worldbank.org/globalfindex. See IFC (2011). 11.  27 Table 2 Comparing FAS and Global Findex on key dimensions FAS Global Findex Unit of analysis Financial institution Individuals Frequency Annual Triennial Survey method Survey of financial regulators Interviews with individuals (face-to-face in (mostly central banks) developing countries, phone interviews in developed countries) Coverage Global Global Time series available: 2004–2011 Time series: Available in early 2015 Types of institutions Regulated only* Regulated, unregulated, informal Access indicators Yes No** Usage indicators Yes (limited) Yes Sampling None—administrative data Representative sample of 1,000 individuals in each country*** Products Deposits, loans, insurance Deposits, loans, payments, insurance (limited) Data type—Loans Stock (all outstanding) Flow (past 12 months)**** Comparative advantage Access provided by different Usage cuts of the data by age, gender, income, institutional types, broad measures education level, urban–rural of usage * Data on financial institutions that are not regulated by the central bank (primary financial regulator) may be available for some countries. ** Global Findex has data on how individuals use different channels to access financial services rather than the availability of these channels. *** Sample size is larger than 1,000 individuals for some countries. **** Data on loan series are flow, but data on loan purposes are stock. Supply-side data surveys such as FAS offer • FAS is conducted by the IMF and is linked to a relatively low-cost means of data collection, other IMF statistical efforts, such as the Interna- with frequent and comparable data that are tional Financial Statistics and the Financial viewed as highly credible to national Soundness Indicators. authorities • Coverage of semi-formal or informal providers • FAS uses administrative data and should provide of financial services is by definition weak, if in- the exact number of accounts and account hold- cluded at all, given the survey method. ers in an ideal situation. Administrative data, however, may have measurement and recording Demand-side surveys such as Global Findex errors. Dormant accounts and multiple counting offer rich information on the many dimensions are also persistent issues. of financial inclusion • It is possible to use FAS to measure access to dif- • Global Findex has detailed data on the users of ferent types of financial institutions, and to have financial services enabling a deep and nuanced broad measures of usage of savings, loans, and in- understanding of financial inclusion from the in- surance offered by different types of institutions. dividual perspective.12 It also includes subjective assessments of the barriers to access to finance • FAS has data on access points and urban–rural from the perspective of individuals. breakdown for access indicators. • Country ownership for supply-side data is See Allen, Demirgüç-Kunt, Klapper, and Martinez Pería 12.  strong, with national governments collecting (2012) for further details on the determinants of account and often easily validating the data. ownership and usage around the world using Findex data. 28 • Global Findex uses a sample of roughly 1,000 re- the two surveys to tell similar stories of financial in- spondents to estimate the value of the indicators clusion, they do not necessarily give the exact same for the whole country, a cost-effective way to cap- number for data points at the country level, even ture the financial inclusion story in a given country. after correcting for the differences in definitions of similar concepts to the extent possible. • Global Findex facilitates analyses of how differ- FAS and Global Findex do not have similar re- ent types of financial behaviors—both formal sults for loans. But this should not be surprising. and informal—fit together at the individual level. FAS and Global Findex are expected to differ fun- • Global Findex can generate an estimate of the damentally in terms of loans because FAS asks for percentage of account holders in the world. As “all outstanding loans” while Global Findex asks for with all individual-level surveys, the precision of “all loans taken in the past 12 months.” The differ- the estimate is affected by sampling design, ques- ences are expected to be particularly significant in tion wording, and response biases (especially re- countries where loans with maturities over 12 lated to individual recall and perceptions). Glob- months are available. As noted, FAS data are also al Findex indicators thus have standard errors susceptible to including dormant and multiple ac- attached, which define the range within which counts. Essentially, FAS has stock data for loans and the estimates would fall if the same exercise Global Findex measures flow. were to be repeated. For deposits, the answer is more nuanced. The analysis that follows focuses on comparing results • Ownership from national governments for de- on deposits across FAS and Global Findex. mand-side data conducted by an external party may vary from country to country. A number of countries are in the process of developing their Fifty of the 103 countries that are covered own demand-side surveys to have a deeper un- by both FAS and Global Findex (49 percent) derstanding of usage of financial services based have indicators on the usage of deposit on their country-specific conditions. accounts that match13 There are many ways to approach the comparison of the two datasets. To provide a simple and intui- FAS and Global Findex are used for different reasons tive comparison of the FAS and Global Findex re- sults that does not require a deep technical analysis, Policy makers and regulators make use of FAS to we simply ranked countries based on the levels of understand the offer of financial services by institu- usage reported against two variables: “account at a tions under their purview. FAS can help provide an formal financial institution” (Global Findex) and understanding of the market structure, pointing to “total depositors” (FAS).14 Countries were consid- strategies to work with different kinds of financial ered either a “match” or “no match” based on the institutions to increase access. A deeper under- closeness of their relative orders in FAS and Global standing of the profiles of users through Global Fin- Findex (with ±10 range in the ranks being consid- dex can lead to more access-friendly policies, legis- lation, and regulation, potentially targeting the groups that are most underserved or are a priority Since the surveys do not measure deposits in the exact same 13.  for governments. Providers, as well as donors and way, we had to construct a total depositor variable before investors, can deepen their understanding of client comparing the two. To construct the total depositor variable, we summed the 14.  profiles and behavior via Global Findex, including FAS depositor data for 2011 for four types of financial institu- client segments that are persistently underserved. tions in the survey (commercial banks, credit unions, finan- Both FAS and Global Findex can be used for bench- cial cooperatives, deposit-taking MFIs, and other deposit takers). Since most countries did not have all of these data, marking across countries. we substituted the 2010 data where available. In cases where deposit account data were provided but not depositor data, How do FAS and Global Findex compare to correct for multiple and dormant accounts, we divided the deposit account data by three (see Kendall, Mylenko, and at the country level? Ponce [2010, p. 22–24] and Table 16) and used it as an esti- Should FAS and Global Findex tell the same story at mate for depositors, essentially assuming that, on average, the country level? Though it is reasonable to expect each depositor has three deposit accounts. 29 ered a match).15 Beyond the initial evidence of a 49 are simply fewer suppliers of financial services. It is percent match, deeper analyses provide clues for also likely that there are fewer multiple accounts in the reasons for the 50 percent nonmatch between low-income countries. the two data sets. Weak financial data management within the financial institutions included in the FAS For many countries where there is not a match, responses could be a source of errors. Weak systems FAS data show greater inclusion likely lead to over-counting of depositors and de- More than half of the nonmatching countries had a posit accounts as a result of dormant, multiple, and higher number of depositors in FAS than in Global joint accounts. In the case of the Global Findex Findex. Thirty-three of the 53 countries (62 per- sampling design, question phrasing, the use of cent) in the “no match” group had a higher rank in phone interviews in developed countries instead of the FAS dataset than in Global Findex. Countries face-to-face interviews, and the general sensitivity with a higher Global Findex rank had twice the av- of asking about financial behaviors can also affect erage per capita GDP. These countries also had sig- results. There are studies on the effects of different nificantly more commercial bank branches per approaches to questionnaire administration on data capita. Furthermore, of the 12 countries where quality that suggest that data from different modes some or all of the Global Findex survey was con- of administration may not be comparable.16 ducted via telephone, 10 had Global Findex rank- ings that were higher than FAS. Countries with lower income levels and less- FAS and Global Findex have together consider- developed financial systems are more likely to ably improved the availability of data on access to have similar FAS and Global Findex rankings and usage of financial services. Great progress has There is a higher rate of matching for countries with been made, and as more information is available, less-developed financial systems measured as a per- there is interest in going even further. For example, centage of adults participating in savings clubs, per- policy makers, practitioners, and funders have indi- centage of population living in rural areas, percentage cated that they would like to see better firm-level of adults with loans from private lenders, and depth data. FAS added SME data only recently in 2011, of credit information. For example, the “match” and responses were not as complete as loan and de- group averages for percentage of adults participating posit account information from commercial banks. in savings clubs, percentage of population living in ru- The Enterprise Surveys provide data on developing ral areas, and percentage of adults with loans from countries only, and are conducted only every three private lenders are significantly higher than those of to four years. Also, there is interest in developing the “no match” group, while the average score on the deeper coverage surveys that can provide even depth of credit information for the “match” group is more nuanced usage data than what Global Findex significantly lower than that of the “no match” group. is currently providing, as well as information on fi- Confirming this, low-income countries had the high- nancial literacy, for example. est match rate among income groups, and SSA had FAS and Global Findex can help inform national the highest match rate among regions, with high-in- and global financial inclusion policy making. Com- come countries and Eastern Europe and Central Asia bining data from multiple sources can provide policy (ECA) having the lowest match rates. Why? There makers with more information on which to base deci- may be fewer discrepancies between supply-side and sion-making. For example, the use of spatial technol- demand-side markets in areas where financial sec- ogy together with demand- and supply-side data tors are not very deep and complex, and where there could offer a more comprehensive picture of the fi- nancial inclusion landscape, identifying geographic 15.  Global Findex accounts data are used together with the rele- areas where access is limited but where there is high vant standard errors in ranking the countries, allowing for a demand. This, in turn, can provide actionable obser- margin. Also, rankings are normalized to reflect different numbers of observations in the two data sets: while FAS had vations for financial inclusion policy makers and oth- 126 observations for “total depositors,” Global Findex had 148 er actors. Some countries such as Malaysia and Brazil observations for “account at a formal financial institution.” are in the process of developing financial inclusion See, for example, Kasprzyk (2005) on how measurement er- 16.  rors may differ for different types of data collection modes as indices that aggregate various dimensions of financial well as other measurement errors in household surveys. inclusion drawing on different data sources. 30 V CHAPTER Linking Financial Access Indicators to Economic Development and Financial Systems Development F inancial inclusion is important because it con- The correlations and trends that follow suggest tributes to improving poor people’s lives. It important possible effects of financial inclusion be- does so by providing the tools to manage cash yond the individual and business level. However, it flows, build assets, mitigate risks, and plan for the is important to caution that these effects are not future of families and businesses. necessarily evenly distributed across and within There is increasing evidence that inclusive fi- countries and may not manifest themselves directly nancial systems are positively correlated with in the lives of the most vulnerable and poor seg- broader financial sector development and growth ments of society. Also, the indicators used and re- as well as complementary to financial regulators’ lated measurement issues call for caution in draw- core goal of ensuring financial stability. The most ing definitive conclusions. In exploring the recent data from FAS and other sources help show relationship between financial inclusion and finan- the relationship of financial access to broader finan- cial stability, challenges exist in both the choice of cial sector development and to the real economy indicators used for financial stability and the avail- through correlations of selected financial inclusion ability of data, notably for lower-income countries. indicators (deposit and loan penetration) with oth- For example, the use of bank nonperforming loans er macroeconomic and financial sector variables. as a percentage of total bank loans is a commonly Deposit and loan penetration are also correlated accepted indicator for financial stability and is use- globally with indicators of economic development ful in that there is greater data availability than for as confirmed also in earlier Financial Access reports more sophisticated indicators, but this indicator and background papers, e.g., Gini coefficient, edu- can miss important dimensions of stability as it cation level, population density, road density, mo- does not capture the off-balance sheet operations of bile phone coverage (CGAP 2009; CGAP and the banks nor the operations of shadow banks. World Bank Group 2010; Kendall, Mylenko, and Ponce 2010; Ardic, Heimann, and Mylenko 2011). This chapter explores relationships among finan- Financial access, growth, and cial access and financial systems and economic de- the reduction of income inequality velopment parameters using commonly accepted and widely available indicators:17 (i) financial access Considerable work has been done in the past 10 as related to growth and the reduction of income in- years on the relationship between financial access equality; (ii) financial access as related to other fi- and income levels. Well-established literature nancial sector parameters, notably financial stabili- shows that the degree of financial intermediation is ty; and (iii) financial access in relation to financial not only positively correlated with growth but is infrastructure and the business environment. The generally believed to causally impact growth and complex relationship between financial inclusion reduce income inequality (Levine 2005, Demirgüç- and financial stability has stirred increasing interest Kunt and Levine 2008; and World Bank 2008). FAS in international financial sector policy discussions. data also show that greater financial inclusion cor- relates with higher income levels (GDP per capita The most commonly used indicators are access, which is 17.  and GDP per capita growth) and a reduction in in- measured by the number of ATMs per 100,000 adults; usage, which is measured by commercial bank deposits per 1,000 come inequality. adults; depth, which is measured by domestic credit to pri- vate sector (% GDP); and stability, which is measured by bank nonperforming loans (% bank loans). 31 Deposit penetration is positively and statisti- shows that as access (shown by ATMs/100,000 cally significantly associated with GDP per adults), usage (shown by loan accounts/1,000 capita and GDP per capita growth adults), and financial depth (shown by domestic FAS data show a number of important relationships credit to the private sector as a percentage of GDP) between deposit penetration and income levels and all increase, inequality (shown by the Gini coeffi- between deposit penetration and the growth of cient) first increases and then decreases. This find- GDP per capita. The number of commercial bank ing merits further exploration to determine possible deposit accounts per 1,000 adults is positively and policy implications in terms of defining appropriate statistically significantly associated with real GDP risk-based measures to foster greater financial ac- per capita and the growth rate of real GDP per cap- cess at earlier stages of financial development. ita, as well as with market capitalization (percent- age of GDP) and domestic private credit (percent- For a country with low levels of financial age of GDP), both indicators of financial depth. inclusion and financial depth, inequality Where data are available for SSA countries for increases at first, then decreases as the finan- the 2004–2011 period, the positive correlation of cial system becomes deeper and more inclusive real GDP per capita and number of accounts per Figure 20 uses three dimensions of financial devel- person is particularly noteworthy, especially for opment and associates them with income inequali- Burundi, DRC, Ghana, Liberia, Lesotho, Mozam- ty: (i) financial inclusion (access and usage), (ii) fi- bique, Rwanda, Tanzania, and Uganda. nancial depth, and (iii) financial stability (measured by the percentage of nonperforming bank loans). Higher financial inclusion is associated with The relationships of access, usage, and depth to in- less inequality, though a certain degree of come inequality are, to varying degrees, expressed financial access and usage and financial sector in inverted U-shapes. This means that a certain de- depth is required before inequality improves gree of financial sector size is required before in- Until recently, research on financial development equality improves. Wealthier segments of the popu- mainly used indicators of financial depth and stabil- lation benefit first; beyond a certain threshold, ity rather than financial access/financial inclusion. income inequality declines with financial develop- This is because country-level aggregates of access ment. For a country with low levels of financial in- to and usage of financial services were not available clusion and financial depth, inequality increases at at a large scale that is comparable across countries first as the financial system becomes deeper (as the and over time.18 The literature shows that financial wealthier segments are better positioned to access development under normal circumstances does not and use financial services), before decreasing, as the merely contribute to economic growth; it also di- financial sector becomes more inclusive. Overall, vides the growth more evenly.19 Poor households developed financial systems are associated with and enterprises, notably SMEs, leverage the oppor- less inequality.20 tunity of access to financial services into greater as- sets and higher incomes, nourishing their growth potential (IFC 2011). Financial access and financial FAS provides a large enough panel data set to al- stability low statistical analyses that use both the time and cross-country dimensions to explore the linkage of In recent years, a growing number of governments— financial inclusion to income equality. Figure 20 many of them lower-income countries with high levels of financial exclusion—have made financial 18.  Examples of earlier research include Beck, Demirgüç-Kunt, inclusion a policy priority, alongside the traditional and Martinez Pería (2007) and Honohan (2008); more re- focus on efficient financial intermediation within cent studies using data from CGAP/WBG Financial Access stable financial systems. Yet, there remain signifi- include Kendall, Mylenko, and Ponce (2010) and Ardic, Hei- cant gaps in the knowledge and capacity of many mann, and Mylenko (2011), and those using Global Findex include Demirgüç-Kunt and Klapper (2012) and Allen, Demirgüç-Kunt, Klapper, and Martinez Pería (2012). This analysis updates and expands on that of Jahan and Mc- 20.  See Jahan and McDonald (2011). 19.  Donald (2011). 32 figure 20 Financial development is associated with less inequality Access Usage 4.2 4.2 4.0 4.0 Gini coefficient, log Gini coefficient, log 3.8 3.8 3.6 3.6 3.4 3.4 3.2 3.2 –2 0 2 4 6 2 4 6 8 10 ATMs/100K adults, log Commercial bank deposit accounts/1000 adults, log Depth Stability 4.2 4.2 4.0 4.0 Gini coefficient, log Gini coefficient, log 3.8 3.8 3.6 3.6 3.4 3.4 3.2 3.2 1 2 3 4 5 6 –2 0 2 4 Domestic credit to private sector (% GDP), log Bank NPLs/bank loans, log policy makers seeking to bring about a shift to pro- financial intermediation, lowering of systemic risk) inclusion policies, balancing concerns of financial and at the micro level through labor market effects inclusion, financial stability, financial integrity, and and enterprise development (greater entry)—i.e., a financial consumer protection. Increased evidence livelihoods effect. Recent analyses suggest that a to understand and explain these linkages is critical virtuous circle between financial inclusion and fi- to help inform the thinking of national policy mak- nancial stability is created when other conditions ers, standard-setters, and emerging global actors. are present, and that an important factor is respon- Current thinking suggests that financial inclu- sible financial inclusion. sion—financial stability linkages—exist through a This section explores and contributes to the avail- number of channels and under a number of condi- able evidence on the relationship between financial tions associated with financial sector development.21 access and financial stability. While the theoretical Financial inclusion increases financial depth, with (and intuitive case) for linking responsible financial its effects on growth and income equality. This has inclusion and financial stability is strong, demon- effects at the macro level (increased savings, greater strating empirical evidence is a challenge. Linkages among inclusion, stability, integrity, and protection 21. See Cull, Demirgüç-Kunt, and Lyman (2012) and World Bank (2012). can be positive or negative. The goal for policy mak- 33 ers is to optimize linkages, which requires maximiz- sured for financial institutions and for financial ing synergies and minimizing trade-offs and other markets. Correlations between pairs of these four negative outcomes.22 Both the theoretical underpin- financial sector parameters for financial institu- ning and the initial empirical evidence suggest the tions show that correlations between access and need both to control for other factors affecting these depth and access and efficiency were statistically linkages and to identify factors required to optimize significant, whereas the correlation between access the linkages. These factors include, for example, ele- and stability was not statistically significant. ments of the enabling environment for finance, in- When the theoretical underpinning for the link- cluding measures to foster financial consumer pro- age is so cogent, why is there a lack of empirical tection and financial integrity. evidence? There are a number of other intervening factors to consider that have a greater influence on While a growing body of literature suggests a financial stability. Overall, financial stability, as positive relationship between financial inclu- measured by the Z-score weighted averages for ex- sion and financial stability, the empirical ample,26 is affected by many financial markets fac- evidence does not yet confirm this23 tors, and the financial inclusion market segment is a Current empirical evidence cannot prove a direct small piece overall. The lack of positive correlation correlation between financial inclusion and finan- may be due in part to a lack of solid data, but it may cial stability. Statistically, financial inclusion (as also mean that the relationship between financial measured by deposit account penetration in FAS inclusion and financial stability is not straightfor- data) does not correlate either positively or nega- ward. These results call for a deeper examination of tively with IMF Financial Soundness Indicators this relationship. and with indicators on financial stability included in the World Development Indicators (WDI) of the Financial access and financial stability corre- World Bank.24 late better in low-income and lower-middle- A similar finding is presented in the World income countries, where access issues are more Bank’s (2012) Global Financial Development Report acute (World Bank 2012, pp. 30–31) (GFDR) 2013, which frames four financial-sector Stability, as measured here by Z-scores, is not characteristics as descriptors of financial develop- strongly correlated with country income levels, as ment: depth, access, efficiency, and stability,25 mea- was highlighted during the global financial crisis, when the financial sectors of many middle- and 22. CGAP has initiated multiple country-level research exercises low- income countries were relatively isolated from on the linkages among inclusion, stability, integrity, and pro- the global turmoil and less affected by global liquid- tection (I-SIP), aimed at elaborating and refining the meth- odological approach for the exploration of these linkages, as ity shocks (World Bank 2012, pp. 30–31). At the well as bolstering the evidence base for policy approaches same time, some higher-income, higher-access likely to serve all four policy objectives. Work began with countries are linked to a number of factors that lead South Africa, prepared in the context of the GPFI First An- nual Conference on Standard-Setting Bodies and Financial to greater instability: lower capital requirements, Inclusion: Promoting Financial Inclusion through Propor- weaker regulations for nonbanking financial activi- tionate Standards and Guidance, Basel, 29 October 2012. ties, lax nonperforming loans responses and related http://gpfi.org/knowledge-bank/publications/issues-paper- provisioning, inadequate bank equity and provi- 3-financial-inclusion-pathway-financial-stability-under- standing-linkages sion, and weak incentives for the private sector to 23. See Cull, Demirgüç-Kunt, and Lyman (2012); Beck et al. monitor risks.27 (2010); GPFI (2012); Khan (2012); and World Bank (2012). 24. IMF Financial Soundness Indicators are available at http:// fsi.imf.org/ and World Bank WDI at data.worldbank.org. The “Z-score-weighted average for commercial banks” (also 26.  25. The Global Financial Development Report (World Bank 2012) called the “distance to default”) captures the probability of de- uses the following definitions for correlations among financial fault of a country’s banking system, calculated as a weighted systems characteristics related to financial institutions: for average of the Z-scores of a country’s individual banks. The depth, private credit to GDP (%); for access, accounts per Z-score compares a bank’s buffers (capitalization and returns) 1,000 adults, commercial banks; for efficiency, 100 minus lend- with the volatility of those returns and is defined as the sum of ing-deposit spread (%); and for stability, Z-score-weighted capital to assets and return on assets, divided by the standard average for commercial banks (called the “distance to default” deviation of return on assets. (See Čihák et al. [2012].) and defined as the sum of capital-to-assets and return on as- See Čihák, Demirgüç-Kunt, Martinez Pería, and Mohseni 27.  sets, divided by the standard deviation of return on assets). (2012). 34 A deep analysis of FAS data reveals some Country evidence on the inclusion–stability evidence for the inclusion–stability linkage linkage is still limited, but emerging country cases There are negative and statistically significant cor- confirms the inclusion–stability linkage. Kenya is relations between financial access (measured by one such example. Research on the Kenyan finan- number of loan accounts per 1,000 adults) and two cial sector (Beck et al. 2010) indicated that asset other indicators of stability: quality improved, liquidity positions improved, in- terest spreads declined (all three contributing to • Bank nonperforming loans/loans, which is an- stability), while outreach improved (in the case of other indicator of stability at the institutional Kenya, driven by mobile payment services). level. Loan penetration decreases as loan quality decreases—i.e., increases in nonperforming Recommendations for further empirical loans are associated with greater instability and research on the financial inclusion—financial less financial inclusion. stability linkages • Risk premiums. As loan penetration increases, The analyses of the relationships described contin- risk premiums (the difference between the prime ue to be works in progress. More statistical work to interest rate and the T-bill interest rate), an indi- quantify and collect data for factors that affect these cator of stability at the financial markets level, relationships is needed. The increasing availability decrease. This means that when the financial sys- of FAS and other data will contribute to this work. tem is more stable, financial inclusion increases. Priority areas of study include the following: • Isolating the effects of the global financial crisis On the other hand, there is also a negative correla- where lower access seems to have correlated pos- tion with bank capital/assets, meaning higher loan itively with stability, as the lower-income, low- penetration in markets with lower capitalized access countries were less affected. In countries banks. This may be explained by the fact that banks affected by the financial crisis, instability was in low-income countries have higher capital-to-as- high, and the crisis resulted in decreased access sets ratios (whether to meet regulatory require- as financial institutions became more risk averse ments or simple prudence), given regulatory re- and levels of financial intermediation declined. quirements and less sophisticated capital structures. This corresponds also to the fact that lower- and • Ascertaining the existence or nonexistence of sys- lower-middle-income countries in fact responded temic risk factors stemming from greater finan- more proactively than high-income countries to cial access. Available evidence suggests that the adopt more prudent regulatory frameworks in re- small loans associated with greater financial ac- sponse to the financial crisis (Čihák, Demirgüç- cess do not contribute to systemic risk, but are Kunt, Martinez Pería, and Mohseni 2012, p. 11). rather counter-cyclical and that the increasing volume of small deposits contribute to a stable do- More competitive banking sectors are associ- mestic savings base (J.P. Morgan and CGAP 2010). ated with greater stability and greater inclusion • Better understanding the qualitative nature of A lower interest rate spread indicates more competi- access, with a focus on what constitutes respon- tive banking sectors (as competition drives the sible access. For example, it makes sense that spread down) and therefore greater stability.28 FAS greater financial protection, a key element of re- data show that a lower interest rate spread is also as- sponsible finance, leads to less over-indebtedness sociated with greater financial inclusion—i.e., the overall, which in turn is important for stability. number of commercial bank deposit accounts per 1,000 adults is negatively and statistically significant- • Understanding (a) how the regulatory environ- ly associated with interest spreads. This means that a ment determines how access is managed, while higher deposit penetration is associated with lower ensuring financial stability, and (b) where a pro- spread and, in turn, with financial stability. Countries portionate regulatory and supervisory framework with more competitive banking sectors have higher can play a role in fostering the linkage is needed.29 deposit penetration and greater stability. This dimension is explored in Cull, Demirgüç-Kunt, and 29.  28.  See Schaeck, Čihák, and Wolfe (2006). Lyman (2012). 35 Financial access and financial Having deposit insurance in place is linked infrastructure to more deposits, but also to more loans Deposit insurance is an important component of fi- Financial infrastructure includes credit bureaus; nancial infrastructure. On average, countries that collateral registries; and payment, remittance, and have deposit insurance have more deposit accounts securities settlement systems—all of which are vi- per adult.30 The same holds true for loans. For ex- tal. When financial infrastructure is available, effi- ample, FAS data show that in 2011, the number of cient, and reliable, the cost of financial intermedia- deposit accounts per adult was over 50 percent tion falls. Financial products and services become more in countries with deposit insurance compared accessible to greater numbers of citizens. Lenders to those without. The number of loan accounts was and investors have greater confidence in their abil- 30 percent more per adult in countries with deposit ity to evaluate and guard against risk. insurance in 2011. The same pattern is not statisti- cally robust when deposit volume (percentage of Greater financial inclusion is associated with GDP) and loan volume (percentage of GDP) are more developed financial infrastructure, and compared across groups of countries that have and a sounder institutional and legal environment that do not have deposit insurance. This may mean A stronger business environment is linked to great- that deposit insurance is important for access, but er deposit and loan penetration. The relationship of not that important for depth. “Doing Business” indicators on Getting Credit (i.e., getting credit rank, credit legal rights index, credit The source of information for the analysis here is the Bank 30.  information index, private credit bureau coverage, Regulation and Supervision Survey by the World Bank, avail- able at http://go.worldbank.org/WFIEF81AP0. Another im- public credit registry coverage) to loan and deposit portant source of information and data on deposit insurance penetration (measured by number of loan accounts systems is the International Association of Deposit Insurers per 1,000 adults and number of deposit accounts (www.iadi.org). per 1,000 adults) is positive. 36 ANNEX 1 Principal Financial Inclusion Data Sources Figure A1.1 Principal International/Multi-Country Data Sources Broader coverage IMF FAS (IMF IFS) (IMF FSI) Global Findex WB/FinCoNet Global Financial Consumer Protection Survey Supply-side Demand-side WB Global Payment Systems WB Enterprise Surveys BIS Payment Systems WB LSMS WB Global Remittance Prices ECB HFCS WSBI research FinScope WOCCU annual statistical report ECB Access to Finance of SMEs The MIS (SAFE) State of the Microcredit WB CP/Financial Capability Summit Campaign OECD/INFE Measuring Financial Micro Insurance Centre Literacy Landscape of Microinsurance (WB Migration & Remittances) studies Financial Diaries Deeper coverage ( . . . ) = covers relevant financial sector data, but not explicity focused on access. Source: Adaptation from Bill & Melinda Gates Foundation. “The Measurement Challenge.” Note prepared for Global Savings Forum, 2010. 37 ANNEX 2 The G-20 Basic Set of Financial Inclusion Indicators and ATMs—Latest Available Figures Table A2-1 Account Loan from % SMEs with % SMEs Commercial Commercial ATMs ATMs at a formal a financial an account with an bank bank per 1,000 per 100,000 financial institution at a formal outstanding branches branches km2 adults institution in the past financial loan or line per 1,000 per 100,000 (% age 15+) year institution of credit km adults 2 (% age 15+) (5–99 employees) (5–99 employees) Year Latest available 2011 2011 Latest available 2011 2011 2011 2011 Source Global Findex Global Findex Enterprise Surveys Enterprise Surveys FAS FAS FAS FAS Data 148 148 128 128 160 160 150 150 Availability economies economies economies economies economies economies economies economies World 50 9 82 37 8 17.3 15.9 47.4 Regions High-income 90.5 14.2 84.2 46.8 31.3 31.0 57.5 90.5 OECD EAP 54.9 8.6 87.7 36.3 6.3 8.1 15.9 17.7 ECA 44.9 7.7 88.2 41.6 10.3 18.5 25.8 46.6 LAC 39.3 7.9 92.5 45.7 6.2 14.9 10.2 33 MENA 17.7 5.1 36.2 5.6 9 17.2 12.1 21.9 SA 33 9 80 28 21.6 8.3 17.2 5.8 SSA 24.0 4.8 86.6 21.2 0.9 3.4 1 4.5 Income Groups Low Income 23.7 11.4 84.1 20.5 1.2 3.8 1.2 3.2 Lower Middle 28.4 7.3 84.5 31.7 7.1 9.6 10.7 13.1 Upper Middle 57.2 7.9 92.3 42.9 8 25.5 17.3 75.6 High Income 89.5 13.8 90.9 50.3 27.1 33.7 69.5 122.9 Notes: Income group classification is based on World Bank Income Classification as of July 2012. EAP: East Asia and Pacific, ECA: Eastern Europe and Central Asia, LAC: Latin America and the Caribbean, MENA: Middle East and North Africa, SA: South Asia, SSA: Sub-Saharan Africa. 38 ANNEX 3 FAS: Definitions and Data Availability 1.  Definitions of Financial • Other deposit takers include all resident finan- cial intermediaries other than central banks, Institutions commercial banks, credit unions and financial The unit of analysis for the IMF’s FAS is financial in- cooperatives, and deposit-taking MFIs that meet stitutions. The classification of financial institutions the definition of ODCs. These institutions have in the IMF’s FAS is based on a functional approach.31 varying names in different countries, such as This approach emphasizes measuring access in savings and loan associations, building societies, terms of the type of financial service offered, such as rural banks and agricultural banks, post office deposit services, credit services, insurance services, giro institutions, post office savings banks, sav- and payments services. Financial institutions are ings banks, and money market funds. classified into two groups: “other depository corpo- rations (ODCs)” and “other financial corporations In this report, credit unions/financial coopera- (OFCs).” IMF’s Monetary and Financial Statistics tives, deposit-taking MFIs, and other deposit-tak- Manual provides details of this classification. ers are together referred to as nonbank financial ODCs include all deposit-taking institutions res- institutions (NBFIs). ident in a country other than the central bank: OFCs consists of a diverse group of resident fi- nancial corporations that provide financial services, • Commercial banks (banks) include all resident either through intermediation or auxiliary services, financial corporations and quasi-corporations and that do not issue liabilities included in broad that are mainly engaged in financial intermedia- money. FAS covers two major types of OFCs—other tion and that issue liabilities included in the na- financial intermediaries and insurance corporations: tional definition of broad money. • Other financial intermediaries (OFIs) include • Credit unions and financial cooperatives in- financial institutions that raise funds on finan- clude financial institutions that are owned and cial markets, but not in the form of deposits, and controlled by their members (customers), re- use the funds to extend loans, mainly to nonfi- gardless of whether they do business exclusively nancial corporations and households, actively with their members. competing with ODCs. OFIs include nondepos- • Deposit-taking microfinance institutions it-taking MFIs, which comprise formal (i.e., le- (MFIs) include institutions whose primary gally registered) financial institutions whose pri- business model is to take deposits and lend to the mary activity is microcredit. poor, often using specialized methodologies • Insurance corporations include financial insti- such as group lending. tutions that provide financial benefits to policy- holders and their survivors in the event of acci- 31.  Alternatives to the functional approach are the institutional dents, illness, death, disasters, or incurrence of approach (measuring access in terms of the types of institu- various or personal expenses. FAS disaggregates tions—commercial banks, credit unions, cooperatives, MFIs, insurance corporations into life and nonlife in- etc.), which was used by the CGAP/WBG Financial Access, and the product approach (measuring access in terms of the surance. type of product—debit cards, home mortgages, etc.). See Barr, Kumar, and Litan (2007) for details. 39 The commercial banks category is the broadest 2. Data Availability in FAS, given that any resident bank functioning as commercial banks that meet the definition of ODCs 2.1 Commercial banks and NBFIs are classified as commercial banks. Hence, the com- Figure Annex 3.1 displays the number of countries mercial banks category, in some countries, may in- that reported commercial banks and NBFIs (de- clude development banks or financial institutions fined as credit unions and financial cooperatives, that serve the poor, such as deposit-taking MFIs, deposit-taking MFIs, and other deposit takers). But depending on the financial activities these institu- not all the countries that reported having deposit- tions are engaged in. taking NBFIs reported the access and usage infor- This report is based mainly on data from com- mation for these institutions. mercial banks, but also includes a box on NBFIs Data availability on commercial banks is the best, (i.e., ODCs other than commercial banks). Chapter although about 15 countries did not report the num- II of the report discusses the state of access to in- ber of commercial banks, while a few others did not surance corporations. The state of access to finance report any commercial bank data besides the num- by SMEs is described in Chapter III, mainly relying ber of commercial banks. Twenty-four countries on commercial bank data. Data availability for each reported no credit unions or financial cooperatives, of these topics follows. while about 40 countries reported having no MFIs. figure A3.1 Deposit-taking financial development institutions Number of commercial banks Number of NBFIs 200 150 Reporting countries Reporting countries 150 100 100 50 50 0 0 2004 2005 2006 2007 2008 2009 2010 2011 2004 2005 2006 2007 2008 2009 2010 2011 Number of CUs & financial co-ops Number of MFIs 30 30 Reporting countries Reporting countries 20 20 10 10 0 0 2004 2005 2006 2007 2008 2009 2010 2011 2004 2005 2006 2007 2008 2009 2010 2011 40 2.2 Insurance Corporations Table A3.1. SME definitions by the World Bank A total of 139 countries reported some insurance Firm Size Employees Assets Annual Sales Loan Size data as part of FAS. The most commonly reported Proxies± series was the number of insurance corporations. In Micro <10 <$100,000 <$100,000 <$10,000 general, life insurance data were more widely re- Small <50 <$3 million <$3 million <$100,000 ported than nonlife insurance data. Data coverage is the lowest for the number of policyholders, broken Medium <300 <$15 million <$15 million <$1 million (<$2 million for some down by life and nonlife insurance. About 20 coun- advanced countries) tries reported data on the number of policyholders, ± Used by IFC. and about 30 reported on the number of insurance policies. Data availability on insurance corporations increases by year, with the most recent years having the highest number of reporting countries.32 Figure A3.2 2.3 SMEs Number of countries Reporting SME loan volume In 2012, the FAS questionnaire was expanded to in- data in 2011 clude data on SMEs for the first time. SME defini- Commercial Banks 37 tions vary across countries. While the FAS ques- tionnaire allows for the use of national SME Other Deposit Takers 9 definitions, it also provided the World Bank classifi- Others FIs 8 cation as guidance to the regulators. Table A3.1 lists MFIs (total) 5 the World Bank definitions. A firm must meet two Credit Union 5 of the three criteria on number of employees, as- sets, and sales volume to be classified as micro, small, medium, or large. IFC also uses loan size proxies as data on employees, assets, and sales vol- umes may not always be available. Since most gov- Table A3.2. Key SME indicators in FAS† ernments track and monitor data for SMEs as a Deposits Volume Outstanding SME deposits (% GDP) whole, data cannot be differentiated across differ- Number SME depositors (% NFC* depositors) ent enterprise size and types—i.e., the number of loan accounts for small vs. medium enterprises. SME deposit accounts (% NFC deposit accounts) Collecting data on access to finance by SMEs is Loans Volume Outstanding SME loans (% GDP) still a challenge for many countries. Although FAS Number SME borrowers (% NFC borrowers) includes all regulated financial institutions, the ma- SME loan accounts (% NFC loan accounts) jority of the regulators provided data on SME fi- nance only for commercial banks (37 countries). †The underlying data are collected for the other four financial provider categories: credit unions, MFIs, other deposit takers, and other financial intermediaries, but the indicators Figure A3.2 shows SME data availability in FAS listed here are calculated and disseminated only for commercial banks. across different types of financial institutions, and * Nonfinancial corporation. Table A3.2 lists the available indicators. ECA and upper-middle-income countries provided the most 32.  insurance data when analyzing the data by region and in- come group. High-income and MENA countries provided the least amount of data. This makes part of the regional analyses not feasible. 41 References Abdin, Md. Joynal. 2012. “The Development Imperatives of SMEs in Bangladesh.” 17 October. Available at http://ssrn.com/abstract=2162771 or http://dx.doi.org/10.2139/ssrn.2162771. Allen, Franklin, Asli Demirgüç -Kunt, Leora Klapper, and Maria Soledad Martinez Pería. 2012. “The Foundations of Financial Inclusion: Understanding Ownership and Use of Formal Accounts.” Policy Research Working Paper No. 6290. Washington, D.C.: World Bank. Ardic, Oya Pinar, Gregory Chen, and Alexia Latortue. 2012. “Financial Access 2011: An Overview of the Supply-Side Data Landscape.” Forum 5. Washington, D.C.: CGAP and IFC. Ardic, Oya Pinar, Maximilien Heimann, and Nataliya Mylenko. 2011. “Access to Financial Services and the Financial Inclusion Agenda Around the World: A Cross-Country Analysis with a New Data Set.” Policy Research Working Paper No. 5537. Washington, D.C.: World Bank. Ayyagari, Meghana, Asli Demirgüç -Kunt, and Vojislav Maksimovic. 2011. “Small vs. Young Firms across the World: Contribution to Employment, Job Creation, and Growth.” Policy Research Working Paper No. 5631. Washington, D.C.: World Bank. Barr, Michael S., Anjali Kumar, and Robert E. Litan. 2007. Building Inclusive Financial Systems: A Framework for Financial Access. Washington, D.C.: Brookings Institution Press. Beck, Thorsten, Aslı Demirgüç-Kunt, and Maria Soledad Martínez Pería. 2007. “Reaching Out: Access to and Use of Banking Services across Countries.” Journal of Financial Economics, 85: 234–66. Beck, Thorsten, Robert Cull, Michael Fuchs, Jared Getenga, Peter Gatere, John Randa, and Mircea Trandafir. 2010. “Banking Sector Stability, Efficiency, and Outreach in Kenya.” Policy Research Working Paper 5442. Washington, D.C.: World Bank. CGAP. 2009. Financial Access 2009: Measuring Access to Financial Services around the World. Washington, D.C.: Consultative Group to Assist the Poor/The World Bank. CGAP and the World Bank Group. 2010. Financial Access 2010: The State of Financial Inclusion through the Crisis. Washington, D.C.: CGAP and the World Bank. Čihák, Martin, Aslı Demirgüç-Kunt, Maria Soledad Martínez Pería, and Amin Mohseni. 2012. “Banking Regulation and Supervision around the World: A Crisis Update.” Policy Research Working Paper No. 6286. Washington, D.C.: World Bank. Collins, Daryl, Jonathan Morduch, Stuart Rutherford, and Orlanda Ruthven. 2009. Portfolios of the Poor: How the World’s Poor Live on $2 a Day. Princeton, N.J.: Princeton University Press. Cull, Robert, Asli Demirgüç-Kunt, and Timothy Lyman. 2012. “Financial Inclusion and Stability: What Does Research Show?” Brief. Washington, D.C.: CGAP. Demirgüç-Kunt, Asli, and Leora Klapper. 2012. “Measuring Financial Inclusion: The Global Findex.” Policy Research Working Paper No. 6025. Washington, D.C.: World Bank. Demirgüç-Kunt, Asli, and Ross Levine. 2008. “Finance, Finance Sector Policies and Long Run Growth.” Washington, D.C.: World Bank. 42 GPFI (Global Partnership for Financial Inclusion). 2012. “Financial Inclusion—A Pathway to Financial Stability? Understanding the Linkages.” Issues paper prepared for the GPFI 1st Annual Conference on Standard-Setting Bodies and Financial Inclusion: Promoting Financial Inclusion through Proportionate Standards and Guidance, Basel, 29 October 2012. http://gpfi. org/knowledge-bank/publications/issues-paper-3-financial-inclusion-pathway-financial- stability-understanding-linkages GPFI and IFC (International Finance Corporation). 2011. Strengthening Access to Finance for Women-Owned SMEs in Developing Countries. Washington, D.C.: GPFI and IFC. . Honohan, Patrick. 2008. “Cross-country variation in household access to financial services.” Journal of Banking & Finance 32 (2008): 2493–500. IAIS (International Association of Insurance Supervisors). 2012. Application Paper on Regulation and Supervision supporting Inclusive Insurance Markets. Basel: IAIS. IFC (International Finance Corporation). 2011. “Financial Inclusion Data: Assessing the Landscape and Country-level Target Approaches.” Discussion paper prepared for GPFI. Washington, D.C.: IFC. ———. 2012. “Why Banks are Increasingly Providing Non-Financial Services to Small and Medium Enterprises.” Washington, D.C.: IFC. J.P. Morgan and CGAP. 2010. “All Eyes on Microfinance Asset Quality.” Global Research. New York and Washington, D.C.: J.P. Morgan Chase & Co. and CGAP. Jahan, Sarwat, and Brad McDonald. 2011. A Bigger Slice of a Growing Pie. Finance & Development, Vol. 48, No. 3. Washington, D.C.: IMF. Kasprzyk, Daniel. 2005. “Measurement Error in Household Surveys: Sources and Measurement.” In Household Sample Surveys in Developing and Transition Countries, UN-DESA, Statistics Division, Studies in Methods, Series F No. 96. New York: United Nations. Kendall, Jake, Nataliya Mylenko, and Alejandro Ponce. 2010. “Measuring Financial Access around the World.” Policy Research Working Paper No. 5253. Washington, D.C.: World Bank. Khan, H. R. 2011. “Financial Inclusion and Financial Stability: Are They Two Sides of the Same Coin?” Mumbai: Reserve Bank of India. Kushnir, Khrystyna, Melina Mirmulstein, and Rita Ramalho. 2010. “Micro, Small, and Medium Enterprises Around the World: How Many Are There, and What Affects the Count? MSME Country Indicators.” Washington, D.C.: IFC and The World Bank. Levine, Ross. 2005. “Finance and Growth: Theory and Evidence.” In Philippe Aghion and Steven Durlauf, eds. Handbook of Economic Growth. Amsterdam: Elsevier. McCord, Michael J., Clémence Tatin-Jaleran, and Molly Ingram. 2012. “The Landscape of Microinsurance in Latin America and the Caribbean: A Briefing Note.” Washington, D.C.: IDB/Multilateral Investment Fund. McCord, Michael J., Roland Steinmann, Clémence Tatin-Jaleran, Molly Ingram, and Mariah Mateo. 2012. “The Landscape of Microinsurance in Africa 2012.” Eschborn: Making Finance Work for Africa and Munich Re Foundation. 43 MIDAS. 2009. “Women Entrepreneurs in SMEs: Bangladesh Perspective. Report prepared for the SME Foundation, Bangladesh.” OECD (Organisation for Economic Co-operation and Development). 2012. “Financing SMEs and Entrepreneurs 2012: An OECD Scoreboard.” Paris: OECD Publishing. Schaeck, Klaus, Martin Čihák, and Simon Wolfe. 2006. Are More Competitive Banking Systems More Stable? Working Paper WP/06/143. Washington, D.C.: International Monetary Fund. Siegesmund, Peter, and Jasmina Glisovic. 2011. “Estimating Funder Support for Small and Medium Enterprises (SMEs).” Washington, D.C.: CGAP. PowerPoint presentation. World Bank. 2008. Finance for All? Policies and Pitfalls in Expanding Access. Washington, D.C.: World Bank. ———. 2012. Global Financial Development Report 2013. Washington, D.C.: World Bank. 44 M “ uch progress was made in 2012 in mapping the landscape of financial inclusion. This revealing study would not have been possible just one year ago—and demonstrates how investment in robust supply and demand side data are helping policy makers and service providers understand the real state of access, quality, and usage of financial services. The opportunity now is to build on these global data sources to expand national data collection and use that is responsive to policy priorities and attuned to country contexts. Importantly, Financial Access 2012 demonstrates that while progress has been made, especially in basic access, there is still much to be done to reach poor individuals and SMEs. It also points to the relationship between financial inclusion and equitable economic development, which in the end, is what access to financial services is all about.” Her Majesty Queen Máxima of the Netherlands, United Nations Secretary-General’s Special Advocate — for Inclusive Finance for Development (UNSGSA)