38412 FocusNote NO. 37 SEPTEMBER 2006 SAFE AND ACCESSIBLE: BRINGING POOR SAVERS INTO THE FORMAL FINANCIAL SYSTEM Despite significant evidence to the contrary, many financial institution managers and policy makers do not believe poor people save money. They tend to assume that poor people are "too poor to save," that they prefer to consume rather than save excess The author of this Focus Note income, or that when they do save it is only to access a loan.1 is Rani Deshpande, a mico- finance specialist at CGAP. The numbers, however, paint a different picture. Research has repeatedly demon- The author thanks Brigit Helms strated that saving is central to poor people's economic management strategies. for her guidance in writing this Projects such as the Financial Diaries initiatives in India, Bangladesh, and South Africa; note; Rich Rosenberg, Ousa Sananikone, Jeanette MicroSave in eastern and western Africa; and studies by the International Food Policy Thomas, and Elizabeth Research Institute, for example, have documented savings practices among the poor.2 Littlefield for their comments; What is unclear is how well formal financial institutions satisfy poor savers' needs. and Mark Pickens and Although over 90 percent of adults in industrialized economies typically have accounts in Jasmina Glisovic-Mezieres for their editorial and research financial institutions, market studies in some areas of the developing world indicate pen- assistance. etration rates as low as 6 percent. Access to savings services for the poor thus varies widely. Understanding this variation and its root causes is the primary goal of CGAP's © 2006, Consultative Group to Assist the Poor Country-Level Savings Assessments (see Box 1). The assessments explore opportuni- ties and constraints encountered by formal institutions in meeting demand for deposit The Consultative Group to services among the poor.3 This Focus Note summarizes the findings from Country- Assist the Poor (CGAP) is a Level Savings Assessments in Benin, Bosnia, Mexico, the Philippines, and Uganda, consortium of 33 development agencies that support microfi- which suggest five strategies for improving poor people's access to savings services. nance. More information is Although these five countries were chosen for the assessments, in part, because of their available on the CGAP Web diversity, the findings of the assessments are strikingly similar in many ways. All five assess- site (www.cgap.org). ments found high demand for savings services but low usage of formal financial institu- tions, which lag behind informal savings mechanisms in fulfilling key client preferences. Institutional capacity and incentives are the primary factors behind the inability of for- mal financial institutions to compete with informal savings approaches. However, sup- portive industry infrastructure and policy frameworks also play important roles. 1Interviews conducted as part of CGAP's five Country-Level Savings Assessments, where review teams met with over 300 individuals knowledgeable about the financial system, indicate that these beliefs are widespread. 2For more information on these findings see Rutherford 2003, Ruthven and Kumar 2002, the Financial Diaries Web site (www.financialdiaries.com), and Zeller and Sharma 2000. 3Formal institutions are defined as those registered with the government. Generally, these include banks of all kinds; nonbank financial intermediaries, including microfinance institutions; and cooperatives licensed to take deposits. When not regulated or supervised, the latter are sometimes referred to as semi-formal. Building financial systems for the poor Box 1 Country-Level Savings Assessments: Purpose and Method CGAP's Country-Level Savings Assessment methodology examines the obstacles and opportunities for improving small deposit mobilization in a particular market, and helps identify strategies to address them. Savings Assessments can be used as an input into design of particular programs or policies, or commissioned by groups of stakeholders to build industry- wide dialogue on the subject. For more information on the assessments, go to www.cgap.org/savings. Improving small deposit services, therefore, requires consciously targeted small deposits, a phenomenon work at all levels of the financial system. observed in market after market. In Benin, the microfinance institution PAPME introduced volun- Low Usage Does Not Mean tary savings products in 2000 and saw deposits Low Demand increase over 20 times in five years even without an active marketing campaign. In Mindanao, the poor- The prevailing perception that poor people do not est region of Philippines, One Network Bank save may stem from the low penetration of formal acquired 23,150 new depositors in three months by savings mechanisms, the most common of which actively promoting an image of stability and good are accounts in financial institutions. In Mexico, management to rural clients. And in Bosnia, recent studies estimate that less than 25 percent of ProCredit Bank doubled deposits in one year simply the urban population, and as little as 6 percent of by improving staff knowledge of available savings the rural population, has any kind of account in a products. financial institution (World Bank 2005 and World Another indicator of demand for deposit services Bank 2001). In the Philippines, only 12 percent of is the extent of informal savings. A survey of rural survey respondents in small cities kept money in a Ugandans indicates that 89 percent save in cash, in bank (Micro-Enterprise Bank and Karlan, Ashraf, kind, or in informal savings groups (as opposed to and Wesley 2004). And in Uganda, only 10 per- only 12 percent who save in formal or semi-formal cent of rural residents report using a financial insti- financial institutions) (Pelrine and Kabatalya 2005). tution (Wright and Rippey 2003 and Pelrin and Even employees of financial institutions are reported Kabatalya 2005). Thus, low penetration rates of Figure 1 Demand for bank accounts financial institutions are evident regardless of among Bosnian survey respondents urban or rural distinctions. Evidence suggests an enormous market for deposit services among poor clients. In a few mar- kets, surveys have documented this demand. In 36% 38% Bosnia, for example, Gallup International (2003) found that almost 40 percent of survey respondents indicated that they wanted, but did not have, a bank account (see Figure 1). In a survey of Mexico City residents, demand for interest-bearing accounts was almost unanimous among the unbanked (World Bank 2003). Because such market data are rare in low-income 26% countries, demand often must be inferred in other Have an account ways. One indication of demand is the rapid Do not have account, do not want account Do not have account, want account growth in deposits among institutions that have 2 Table 1 Use of formal and informal savings mechanisms in Mexico (Percentage of respondents using respective mechanisms) Formal financial Informal financial Physical assets (represented by (Represented by (Grains, animals, banks) tandas) construction materials) TOTAL 27 20 29 Income level Very Low 10 11 42 Low 22 19 32 Medium 29 30 23 Medium-high 53 23 24 High 67 22 17 Source: Bolaño, Pilar Campos. 2005. El Ahorro Popular en México: Acumulando Activos para Superar la Pobreza. CIDAC and MiguelAngel Por- to operate their own rotating savings and credit asso- signal demand for formal deposit services. But to tap ciations among staff. that demand, formal services must offer clients a In the Philippines, one study suggests that two- more attractive value proposition than what they thirds of people in smaller urban cities keep their already can access informally. Whether or not they can savings at home (Karlan, Ashraf, and Wesley 2004). do so--at least with current service delivery models-- Informal savings are also widespread in Mexico, is an open question. where another survey shows that 50 percent of rural residents keep money at home. Respondents How do formal savings services compete also reported keeping over 90 percent of their sav- with the informal sector? ings in cash or in kind and less than 10 percent on deposit in a financial institution (World Bank To compete with informal savings mechanisms, 2001). In Mexico City, 28 percent of the unbanked formal services must be appropriate for poor savers, and 40 percent of the banked hold some financial satisfing key client preferences. Although these pref- savings outside of financial institutions (World erences vary from market to market, worldwide, Bank 2003). Nationwide in Mexico, the level of low-income savers tend to care most about accessi- usage of informal instruments increases as educa- bility and security. Accessibility can be seen as physical tion and income levels decrease (see Table 1). accessibility (proximity) and financial accessibility Perhaps the most compelling evidence of demand (affordability). Accessibility also can be defined as for deposit services comes from West Africa, where liquidity, although studies indicate that poor clients clients routinely pay roving deposit collectors to hold have both liquidity and illiquidity preferences (see their money. Just as the existence of moneylenders Box 2). indicates a demand for credit among the poor, the When formal and informal savings mechanisms are widespread use of informal savings mechanisms may compared on the key criteria of security, proximity, Box 2 Don't poor people want liquidity? Liquidity is often considered another key facet of accessibility and a priority for poor savers. Although this may be true for a certain set of clients--or a certain portion of most clients' overall economic portfolios--market studies demonstrate demand for a range of maturities. For example, a study in Mexico found that a wide range of maturities is among the top three product features demanded by rural savers (World Bank 2001). A nationwide survey in Uganda found clients prefer making withdrawals monthly or quar- terly, but they want to make deposits more frequently, suggesting a preference for illiquidity (Pelrine and Kabatalya 2005). This may be related to the reasons for which clients saved, which were skewed heavily in favor of school fees and medical expenses. These were also the top two savings goals in a study of urban low-income Filipinos. 3 and affordability, the reasons behind low usage of Philippines) is always accessible and carries no extra formal services are clear. costs to the saver in terms of time or travel. By contrast, formal deposit-taking institutions Proximity can be very far away indeed. Because precise aver- Informal mechanisms out compete formal financial age distances are difficult to calculate, the Savings institutions when it comes to both dimensions of Assessments used comparisons of branches, popu- accessibility. In terms of proximity, it is hard to beat lation, and surface area as rough proxies. Figures saving at home: cash stuffed into a mattress (or a 2a, 2b, and 2c illustrate the average number bamboo post in one's house, as is done in the of branches per population and square kilometer, Figure 2a Branches per million inhabitants (regulated deposit-taking institutions) 250 194 200 138 150 84 94 100 38 50 - Benin Bosnia Mexico Philippines Uganda Figure 2b Branches per square kilometer (regulated deposit-taking institutions) 0.030 0.026 0.025 0.020 0.016 0.015 0.010 0.005 0.005 0.005 0.005 - Benin Bosnia Mexico Philippines Uganda Figure 2c Branches/population/square kilometer (regulated deposit-taking institutions) 3,785 4,000 3,500 3,000 2,500 2,000 1,500 758 1,000 303 500 49 191 - Benin Bosnia Mexico Philippines Uganda 4 separately and combined, across the countries section of the Savings Information Resource Center studied. The tremendous variation in these num- (www.cgap.org/savings). bers shows little correlation to income level, with both Mexico (annual per capita GNI of almost Affordability $7000) and Benin (annual per capita GNI under Informal savings also tend to trump formal institu- $300) scoring similarly. tions in terms of price--even when they carry an Within countries, however, CGAP found a much explicit cost, as in the case of deposit collectors in stronger relationship between income level and the West Africa. Typically, deposit collectors visit each distribution of deposit-taking institutions. In the client daily and collect a fixed amount, deducting Philippines, the correlation between the number of one day's deposit at the end of the month as their subsistence poor and the number of people per service fee. On an average-balance basis, this branch of deposit-taking institutions in different amounts to a negative interest rate of 7 percent. regions was 0.68. In Mexico, the correlation Although this fee seems steep, for most clients, it between the number of people per branch and the is still a better deal than using either local coopera- tives or banks. The travel cost to use either formal level of marginalization in different states was 0.62.4 institution generally outweighs the interest rate Figures 3a and 3b illustrate this disparity in advantage, even without factoring in the value of the greater detail. (See pages 6 and 7.) In parts of client's time. Figure 4 illustrates that a typical small- Mexico City, for example, population per branch is town resident of Benin would have to save more less than 3,000. However, in certain districts of than 10,000 FCFA (USD 20) per month before Chiapas (Mexico's poorest state), this ratio exceeds using a cooperative made more financial sense than 100,000. Several districts have no branches of using a deposit collector. To make saving in a bank deposit-taking institutions at all. Similar situations worthwhile, the client would have to save approxi- were observed in Benin, Bosnia, the Philippines, and mately 100,000 FCFA (USD 200) per month. Uganda; maps showing the distribution of branches for deposit-taking institutions in these countries are 4 Marginalization was measured according to a composite indicator es- available in the Country-Level Savings Assessment tablished by the government. Figure 4 Net return on savings by mechanism (Benin) 10% 0% -10% Return -20% Net -30% -40% 0 0 0 ,000 ,00 ,000 2,500 5,000 ,000 10,000 25,00 50,000 100 250,000 500 1,000 1,500,00 2,000,000 2,500 5,000,000 Monthly Savings Banquier ambulant Cooperative Bank Note: Calculations are based on costs faced by resident of typical small town in Benin. Assumptions: (1) Banquier ambulant (deposit collec- tor) collects 30 deposits per month for a fee equal to 1 daily deposit; (2) client makes weekly trips to local cooperative at a per trip cost of 150 FCFA, cooperative pays no interest; (3) client makes monthly trips to the bank at a per trip cost of 5000 FCFA, bank pays 3.5% annual inter- est. See Helms, et al. 2005. 5 Figure 3a Mexico penetration map MEXICO FEDERAL DISTRICT POPULATION PER FINANCIAL INSTITUTIONAL BRANCH 02 07 MEXICO 06 16 11 08 03 MEXICO DELEGATIONS: 01 01 Alvaro Obregón 09 02 Azcapotzalco 04 03 Benito Juárez 04 Coyoacán 05 05 Cuajimalpa de Morelos 13 06 Cuauhtémoc 07 Gustavo A. Madero 08 Iztacalco 10 15 09 Iztapalapa 10 La Magdalena Contreras 11 Miguel Hidalgo 12 Milpa Alta 14 13 Tláhuac 14 Tlalpan 15 Xochimilco 16 Venustiano Carranza 12 MORELOS This map was produced by the Map Design Unit of The World Bank. The boundaries, colors, denominations and any other information shown on this map do not imply, on the part of The World Bank Group, any judgment on the legal status of any territory, or any endorsement or acceptance of such boundaries. POPULATION PER FINANCIAL 115°W 110°W 105°W 100°W 95°W 90°W 85°W UNITED STATES OF AMERICA INSTITUTION BRANCH*: Tijuana MexicaliMexicali Ensanada Sonoita Sonoita San San Ciudad Juárez Felipe Felipe Rio Nogales Nogales BAJABAJA G > 15.001 Agua Agua Prieta Prieta u RioBravGo rande 30°N 30°N CALIFORNIACALIFORNIA lf SONORASONORA STATE BOUNDARY o Yaqui C H I H U A H U A f Ojinaga Ojinaga 12.001 - 15.000 HermosilloHermosillo C a lifo ChihuahuaChihuahua Salado DELEGATION BOUNDARY 9.001 - 12.000 Guaymas Guaymas Santa Rosalia rn BAJABAJA ia Conchos LaredoLaredo Navojoa Navojoa C O A H U I L A CALIFORNIACALIFORNIA Fuerte Frontera Frontera SURSUR 6.001 - 9.000 NUEVO NUEVO Gulf of Mexico Loreto Loreto Los Mochis Los Mochis LEON LEON Matamoros Matamoros 25°N 25°N MonterreyMonterrey Torreón TorreTorreón SaltSaltíllollo Saltíllo SI 3.001 - 6.000 CuliacCuliacán Culiacán NAL DURANGO N La Paz A TAMAULIPAS TAMAULIPAS LOA Durango Durango ZACATECASZACATECAS CiudadCiudad VictVictóriaria Victória 0 - 3.000 Mazatlán MazatlMazatlán Cabo San Lucas SAN LUIS SAN LUIS ZacatecasZacatecas Tampico AGUASCALIENTESAGUASCALIENTES POTOSIPOTOSI QUERQUERÉTAROTARO QUERÉTARO NAYARITNAYARITLeAguascalientesAguascalientes San Luis San Luis PotosPotosí Potosí * The best information available from the sources cited was used to YUCATANYUCATAN Cancun Tepic Tepic rma VERACRUZVERACRUZ Merida GUANAJUATOGUANAJUATO GuanajuatoGuanajuato HIDALGOHIDALGO Cozumel create the maps, some unregulated small local institutions may not be Guadalajara Guadalajara QuerQuerétarotaro Querétaro 20°N PuertoVallerta DISTRITO FEDERAL DISTRITO FEDERAL Campeche PACIFIC JALISCOJALISCO MEXICOMEXICO PachucaPachuca TLAXCALATLAXCALA QUINTANAQUINTANA included. MEXICOMEXICO JalapaJalapa MoreliaMorelia CITYCITY Bay of Campeche ROOROO OCEAN TolucaToluca TlaxcalaTlaxcala Veracruz Colima Colima Chetumal COLIMACOLIMA MICHOACANMICHOACAN CuernavacaCuernavaca PueblaPuebla CAMPECHECAMPECHE PUEBLAPUEBLA TABASCOTABASCO Balsas Villahermosa Villahermosa Gulf of GUERREROGUERRERO BELIZEHonduras Notes: Number of financial institution branches as of 2004 and population data from Chilpancingo Chilpancingo Area of map Oaxaca Oaxaca CHIAPASCHIAPAS Usumacinta Acapulco OAXACAOAXACA TuxtlaTuxtla 2000 National Census. GutierrezGutierrez Tehuantepec Tehuantepec Sources: INEGI, Bansefi, FONAES, ABM, Caja Libertad, Caja Popular Mexicana, MORELOSMORELOS Puerto 15°N 15°N Escondido Gulf of GUAGUATEMALATEMALA Tehuantepec HONDURAS Tapachula Compartamos and FinComun. EL SALVADOR 115°W 110°W 105°W 100°W 95°W 6 Figure 3b Mexico penetration map N N N °03 °52 of °51 °W Cancun Cozumel Gulf be 85 Honduras A to N N AT AT O Chetumal HONDURAS EL ADORV A NI O ROOR BELIZE not Popular C U U QUINTANAQ AL SAL YUCATANY E A H C M E used P may and M Merida ETEMALATA A U W Mexico CAMPECHEC aso acinta Caja Usum 09 BRANCHES of Campeche O i mrehall GUAG was C VillahermosaV L S A altxu zerreitu 2004 A TuxtlaT GutierrezG Gulf R B Tapachula E D Campeche ATABASCOT PER of EFFEDERAL O AL of cited R A of O Bay Libertad, ATAROTÉR OTI AS C W°59 GL A RT X A cepetnauhe Gulf FINANCIAL institutions as TehuantepecT W° C huantepeceT E 59 U DI SI ALTLAXCALAT AS A QUERQQUERÉTARO HIDALGOH DISTRITOD soro Z AMERICA mata U apalaJalapaJ Veracruz X A OAXACAO S R AL CHIAP MatamorosM A C A alacxal Caja B acaxa OaxacaO Puerto PER INSTITUTION BOUNDARIES OF PIL Tampico R TlaxcalaT albeu BRANCH*: sources local E U A dadui M airiarótci E VERACRUZV acuhca PueblaP U Escondido PUEBLAP PachucaP ES CiudadC VictóriaVictV ognicnaplih L T TAMAULIPASAT O CI I YT ChilpancingoC A O S A branches X CITYC ass R W the O W° V N yerretno MonterreyM SI I S siuLLuis oratarotéreu QuerQQuerétaro E Bal E OL CHIAP TION CITIES BOUNDARIES 0 oder E n t íso MEXICOM R 01 small Census. ABM, E O 10 ST LaredoaL aretn aculo U LEONEL ULLUIS TolucaT acavanreu R CuernavacaC R A or NUEVON OT aSan o S PotosPPotosí N O otaujanau E O 100.000 50.000 L N U Acapulco MORELOSM LI FronteraF A INSTITUTIONAL SANS POTOSIP GuanajuatoG A GUERREROG - - A TIONAL Salado U olllolítla S C A A H SaltSSaltíllo C CAPIT BOUNDARIES from A ET O A arajaladau ailero MoreliaM O H CI MEXICO OF FINANCIAL FinComun. UNITED COAHUILAC C GuadalajaraG MICHOACANM 20.000 E E 100.001 - T T nóerro ZACATECASAZ sacetaca ZacatecasZ setneilacsaug O C a A E National GO AguascalientesA Lerma SIL milo MIL NO > 50.001 20.001 1 A A institution W°° TorreónTorreT TA BRANCH FONAES, and ColimaC O Juárez R 105 A JALISCOAJ COLIMAC W°°0 Granavo de A aganij N U R A POPULATION INSTITUTION OjinagaO Y cipe ST SELECTED MUNICIP ST INTERNA 105 Rio Br ognaru TI H U DurangoD A TepicT Ciudad Rio available unregulated A U HI 2000 H auhauhih Conchos DURANGOD NAYARITN S I N A L O A n S OT áltaza ET A O N UJ PuertoVallerta CI X CHIHUAHUAC ChihuahuaC Fuerte MazatlánMazatlM EIL A E N MEXICOM ST A financial nácailu A C U S GUANAJUATOG some a CuliacánCuliacC POPULA Bansefi, A W° aug teir of from U AguaA PrietaP Yaqui G 110 A MochisM AGUASCALIENTESA R sa Paz O aojova sihco so LosL N NavojoaN La selago °W O SONORAS olliso myau Lucas GuaymasG r n i a 110 San information maps, data Compartamos NogalesN mre o l if AI oter HermosilloH INEGI, a N LoretooL Cabo ationo C CI o f R Santa AJ R Rosalia lf A FI Number SonoitaS G u BAJAB OFIL U SURS A C FINANCIAL best the W° AI CALIFORNIAC N PA OCEAN 115 R ilacixe SanSna epile AJ A FelipeF BAJAB OFIL A MexicaliM CALIFORNIAC The W° * create included. Notes: population Sources: Mexicana, 115 Tijuana Ensanada 30°N 25°N 20°N 15°N Casas Espinosa las de Solistahuacán de Duraznal Pinal Carranza Cancuc Ruíz Agua el Nuevo Chico Gutiérrez de Juárez Andrés Cristóbal Fernando Juan Lucas Comaltitlán Corzo Nicolás Ocosingo Ocotepec Ocozocoautla Ostuacán Osumacinta Oxchuc Palenque Pantelhó Pantepec Pichucalco Pijijiapan Puablo Rayón Reforma Sabanilla Salto San San San San San Santiago Siltepec Simojovel Sitalá Socoltenango Solosuchiapa Soyaló Suchiapa Suchiate Sunuapa Tapachula Tapalapa Tapilula Tecpatán Tenejapa Teopisca Tila Tonalá Totolapa Tumbalá Tuxtla Tuxtla Tuzantán Tzimol Unión Venustiano Villa Villa Villaflores Yajalón Zinacantán 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 Guerrero Comillas Madero de Tenejapa de de Margaritas Rosas Concordia Grandeza Independencia Libertad Trinitaria Domínguez Corzo Ixtapangajoya Jiquipilas Jitotol Juárez La La La La La Larráinzar Las Las Mapastepec Maravilla Marqués Mazapa Mazatán Metapa Mitontic Montecristo Motozintla León de Comalapa Hidalgo de 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 ALA 60 61 62 63 64 65 Bosque Porvenir Chiapa Chiapilla Chicoasén Chicomuselo Chilón Cintalapa Coapilla Comitán Copainalá El El Escuintla Francisco Frontera Frontera Huehuetán Huitiupán Huixtán Huixtla Ixhuatán Ixtacomitán Ixtapa M ET 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 AU G Frontera Américas la Valle las 6776 de del Corzo de Ocampo de Albino Vista Acacoyagua Acala Acapetahua Aldama Altamirano Amatán Amatenango Amatenango Angel Arriaga Bejucal Bella Benemérito Berriozábal Bochil Cacahoatán Catazajá Chalchihuitán Chamula Chanal Chapultenango Chenalhó MUNICIPALITIES: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 2525 5555 Dominguez 3773 5005 de 1551 3553 ALAMETAUG 7117 7227 3663 7007 Comitan 211121 2882 3300 1111 80 2112 6116 0505 6006 1081 2626 3773 9669 071107 711117 2772 0220 111111 3333 9889 6556 1991 0 3388 110404 6886 6556 2992 6262 11110 7474 0440 8998 pachulaaT 3663 201102 Casas 30 1661 1103 4114 6666 31 41 1881 9009 1818 2222 8448 Cristobal 3443 las 6 1131 1141 9393 San de 0101 2332 4554 9119 01106 9009 TABASCO 8 7887 2424 4499 3030 0066 7878 4747 11118 2002 7575 4545 3993 4242 00 7979 1551 4994 4444 1001 7575 3223 4884 7667 3443 2121 9999 1771 9292 2525 90 5 8008 9797 8668 5885 1091 5959 7777 5353 3113 11115 4114 6 any GUTIERREZ Bank 11611 or 0770 Flores information 6996 llai World 1 V PACIFIC OCEAN other 10101 The territory, TUXTLA 50 any of 1051 any 4664 and part of alan 06 the boundaries. de status To on such 2288 legal of imply, denominations the 0110 04 not on Cintalpa Figueroa do 0 colors, acceptance KILOMETERS map or judgment this VERACRUZ on any 02 OAXACA boundaries, The shown Group, endorsement 7 These minimum amounts are equivalent to 12 days current generation of savers, with the expected and 120 days, respectively, of minimum wage in impact on client confidence. Still, the most thorough Benin. Given that many people in Benin earn far study of security in informal savings to date indicates less than the minimum wage, formal savings serv- that informal mechanisms are far riskier than even ices are beyond the reach of many poor people. unsupervised deposit-taking institutions (Wright and Formal financial institutions tended to price Mutesasira 2002). themselves out of the market for poor people's sav- Ironically, security often emerges as savers' top pri- ings across the markets studied by CGAP. The ority in surveys of low-income clients. Why, then, do "threshold costs" of basic deposit products--i.e., poor clients seem to prefer informal to formal sav- the amount of money needed for opening fees, ings? Behavioral finance suggests that familiar risks or minimum balances, and shares, in the case of coop- those perceived to be under an individual's control eratives--were often many times what poor people tend to be underestimated compared with risks that could afford. This was especially true in safer, regu- are unfamiliar or perceived to be not within a per- lated institutions, such as banks. son's control (Hertwig et al. 2004). The risk inher- In CGAP's analysis, "basic deposit products" ent in saving at home, with a trusted deposit were defined as the lowest-cost savings or transaction collector, or in livestock that one has been tending account in a given institution. Because of minimum since childhood, may thus be underestimated com- balance requirements for other products, these pared with the risk of saving in an unfamiliar institu- demand deposit or current accounts are often the tion with opaque management. Because important first or only type of account poor clients can afford. measures of security in a formal institution--such as CGAP's research found that threshold costs tend loan losses--are not directly observable, it is easy for to decrease as branch density (the number of clients to misjudge the relative security of informal branches per population) increases. Across the versus formal savings options. Given the widespread countries studied, the correlation between thresh- absence of systems to signal institutional soundness, old costs (measured as a multiple of minimum comparing the security of different formal institu- wage) and population per branch was -0.78 for all tions is also a challenge for customers. deposit-taking institutions. The correlation was stronger (-0.83) when only bank branches were Trade-offs between security and accessibility considered (see Appendix 2). Although based on a Often, financial institutions that are most accessible limited sample of countries, the results support the to poor clients--especially in rural areas--are less expectation that competition brings down price. secure (in terms of financial soundness and there- fore risk to savings) than institutions that target Security wealthier clientele. In the Philippines, for example, Although informal options have advantages over rural banks have higher nonperforming loan ratios financial institutions based on proximity and afford- than either thrift or commercial banks. In Uganda, ability, comparisons between informal and formal the average reported portfolio at risk (PAR) greater options based on security are less conclusive. Deposit than 30 days in savings and credit cooperatives collectors are rarely subject to regulation or supervi- (SACCOs) reach as high as 42 percent, compared sion, and reports of them absconding with client sav- with single digits for regulated institutions.5 ings are common. Saving at home carries its own However, these regulated institutions are seldom risks, such as theft or loss from fire or flood. But located outside primary and secondary cities. banks also fail. In fact, every one of the markets stud- ied by CGAP had suffered a banking system crisis in 5PAR 30 equals the outstanding balance of loans with any payment more the past 10­20 years--well within the memory of the than 30 days late, divided by the outstanding balance of all loans. 8 Reliable data are seldom available for institu- ries a high price tag, and institutions that operate in tions serving the poor. In Uganda, PAR data are poor and rural areas usually need to maintain a low available for only 147 SACCOs out of an estimated cost structure. total of 750­900 currently active in the country. In To be able to afford high-capacity management, the Philippines, even the total number of financial institutions, such as cooperatives, need to achieve cooperatives in operation is unknown. The only economies of scale. Some institutions achieve performance data available come from a small economies of scale by consolidating and becoming number of co-ops selected for a donor-funded branches of a single "mother cooperative." The technical assistance project because of their prom- consolidation of 60 independent cooperatives to ise for improvement. At the time of entry into this become Caja Popular Mexicana is an example of a project in 1997, their collective PAR 30 was 54.6 successful transformation. Between 2001 and 2005, percent (although by 2005 this had declined to Caja Popular Mexicana saw capital adequacy 8.76 percent). increase by three times, operating expenses halved, No single type of deposit-taking institution gen- delinquency fall by two-thirds, and assets and mem- erally offers low-income clients security, proximity, bership double. However, this transformation took and affordability. Therefore, clients tend to use some 15 years of work by the consolidating credit savings instruments whose advantages they can see unions and $4 million in donor funds for skilled, and that they can access. Proximity and affordabil- credible technical assistance providers. ity--features offered par excellence by informal Institutions that already have the capacity to mechanisms--are clearly visible to clients. Lack of safely manage deposits need to focus on lowering security is not. Poor savers may thus be sacrificing delivery costs to be able to serve low-income mar- what they most value in a deposit service--secu- kets. Advancements in technology have made indus- rity--by using informal mechanisms that provide try professionals enthusiastic about the potential of clear evidence of proximity and affordability but do developing technology-enabled channels to lower not necessarily provide security. Conversely, formal costs for delivering services, including deposits. institutions that offer secure savings opportunities Unfortunately, this type of cost-cutting itself may have difficulty penetrating the savings market for require substantial up-front investment. poor clients because they tend to be inaccessible to In Uganda, one micro-deposit-taking institution poor clients. estimates that it has invested $200,000 in develop- ing a system using smart cards and point-of-sale ter- Why Can't Formal Institutions minals (in addition to considerable investment by Compete? public and private donors). Despite this investment, the new system has yet to be rolled out at scale. Why do so few formal financial institutions offer Another strategy increasingly used by banks to both security and access to poor savers? CGAP's lower service delivery costs is to take advantage of Savings Assessments have identified a constellation retail infrastructure that is already in place. In of causes at three levels of the financial system. Mexico, one bank established mini-branches inside a network of appliance stores and was able to open Inadequate management capacity and 4 million savings accounts in two years--more high cost structure accounts than the entire "popular finance" sector, The skills needed to manage savings programs are with comparable average balances.6 different, and often more sophisticated, than those needed to run credit programs. Capable manage- 6The "popular finance" or finanzas populares sector in Mexico refers to ment, staff, and systems are crucial for mobilizing institutions, such as MFIs and cooperatives, that serve the traditionally savings. But high-capacity management often car- unbanked. 9 In the Philippines, Banco de Oro enables Lack of liquidity management mechanisms was cashiers at a major retailer to act as service points especially common among cooperatives, perhaps for the bank's main mass-market product, a reload- because of the fear of spreading risk from badly able payroll and debit card. In two years, this prod- managed to well-managed co-ops. The disincentive to uct mobilized $5.6 million in balances on 700,000 savings mobilization posed by excess liquidity in coop- cards in force (in a country with a working popula- eratives can be especially damaging for poor clients, tion of about 50 million). In both the Mexican and because co-ops tend to be more accessible than other Filipino cases, the bank and cooperating retailer are types of institutions. Better liquidity management owned by the same parent company. tools would make deposits more attractive. In Uganda, stakeholders are considering ways to Few outlets for excess liquidity place excess co-op liquidity in regulated institutions Excess liquidity in financial institutions limits incen- that offer safe, remunerative investments--or directly tives to mobilize additional deposits--especially poor into Ugandan government bonds, which offer attrac- people's deposits, which tend to be perceived a tive rates. Either option is less risky than encouraging priori as short term, unstable, and costly. Unfortu- these cooperatives to ramp up lending (especially nately, regulated financial institutions in many uncollateralized lending like microcredit) in light of markets already have more than enough deposits their high loan delinquency rates. relative to lending or investment opportunities, Unfair competition from "cheap and easy" especially in the short term. At the time of the wholesale funds Philippines CGAP assessment, for example, savings in banks were 150 percent of bank loans to the Unfortunately, deposit mobilization often is discour- private sector. In Benin, the proportion was roughly aged by wholesale financing from apexes (usually similar. In Bosnia, although the banking system was government owned). In countries as diverse as slightly under-liquid overall, excess liquidity was Uganda and the Philippines, donors and govern- concentrated in the short term (under one year), ments provide millions of dollars in wholesale financing to deposit-taking institutions that are with 60 percent of deposits against only 25 percent among the most accessible to poor clients (SACCOs of loans.7 in Uganda and rural banks in the Philippines). At the institutional level, excess liquidity may be Second-tier financing in both Uganda and the caused by a lack of suitable lending opportunities Philippines is at near-market rates, so the funds are (real or perceived). Especially among banks, demand not as cheap as they could be. Still, relatively easy for on-lending funds can be limited, particularly in access to wholesale funds has served as a disincentive countries where government debt is not an attrac- for financial institutions to pursue additional deposits tive investment opportunity. to finance portfolio growth. If these funds were Although particular institutions may be crowding out deposits in shaky institutions, this temporarily under- or over-liquid, in developed might be considered a benefit to savers. However, countries these imbalances are usually corrected by liquidity management mechanisms. However, many 7Nor were excess funds being soaked up by treasury bonds. In the developing-country markets lack these liquidity Philippines, these were returning around the same rate as inflation; in Benin, government bond issues were extremely uncommon; and in management mechanisms. Even Bosnia, which is, by Bosnia, the government is currently prohibited from issuing sovereign some measures, the most advanced financial system debt. 8 studied by CGAP, lacks overnight markets, repur- Foreign-owned banks in Bosnia can alleviate excess short-term liquid- ity by sending short-term deposits abroad to their parent banks in re- chase agreements, and other tools commonly used turn for longer-term lines of credit. But most financial institutions do by banks to manage liquidity.8 not have this option. 10 the logic of wholesale funders leads them to channel ally, access to the payment system is restricted to funds to the strongest, not the weakest, institutions. more highly regulated institutions and often requires Incentives are therefore dampened for the very insti- a hefty security deposit with the payments clearing- tutions best able to safeguard client deposits. house. Payment systems are, in fact, often owned by A common rationale for second-tier credit lines is large banks. These attributes tend to put access to the mismatch that sometimes exists between financial payments systems beyond the reach of small financial institution over-liquidity in the short term and under- institutions that cater to the poor. The inability of liquidity in the longer term. But often, short-term these institutions to make clients' savings accessible funds are much more stable than commonly thought in multiple locations or to provide payments services and could be used for longer-term lending. Occasion- makes depositing with them less attractive. ally, regulation also can underestimate the stability of However, a new breed of payments provider short-term deposits: in Bosnia, 95 percent of deposits promises to make payment services more widely under 30 days must be lent out for the same term-- available. Some of these providers use mobile phones a very limited market in a country with no overnight linked to bank accounts for domestic and interna- interbank lending. tional money transfers, purchases, loan repayments, and cash withdrawals. Often these transactions Incomplete payment systems involve a debit card and a network of point-of-sale Access to the payment system can greatly affect an devices located in financial institutions and retail institution's ability to capture deposits. Tradition- establishments (see Box 3) (Ivatury 2006). Box 3 Bypassing the payments system Banks are increasingly teaming up with retailers and mobile phone providers to offer financial services, including payments and deposits, to poor customers. In Brazil, six banks have built a network of over 26,000 banking correspondents: mer- chants equipped with card-reading devices that customers can use to open accounts and make transactions. To make a deposit, customers identify themselves by swiping their card and giving cash to the retailer. The bank transfers the corresponding amount from the retailer's account to the customer's account. The clerk keeps the cash received in the till and issues a receipt to the customer (see diagram below). Largely because of the increase in banking correspondents, by the end of 2003, banking services were available in all of Brazil's more than 5,600 municipalities. xxxxx Customer swipes card and gives cash to retailer Retailer issues receipt to customer XXXX Customer Xxxx Retailer xxxx Bank transfer funds Bank transfers to customer's funds from account Bank retailer's account Similar services are being introduced elsewhere. In the Philippines, Globe Telecom offers G-Cash, a service that lets sub- scribers load cash onto their mobile phones via affiliated merchants or Globe outlets, and then use it to make international or domestic money transfers, pay bills, or make purchases. The service is usually operated like a "virtual wallet," with no bank involved. However, Globe is partnering with a growing number of rural banks to enable clients to receive and repay loans via G-Cash. Mobile phone banking also is making inroads in South Africa, where a joint venture between Standard Bank and mobile phone provider MTN has acquired more than 100,000 customers since 2004. 11 Uneven regulation and supervision quirements discourage rural branches. A similar phe- Regulation and supervision, which can be highly nomenon is at work in Uganda, where investments restrictive for some types of institutions and loose required by the new Micro-Deposit Taking Institu- or nonexistent for others, can perpetuate a situation tions (MDI) Act have at least temporarily slowed where safe institutions have few incentives to pur- branch expansion among newly regulated MDIs.9 By sue poor depositors, and accessible institutions can- driving up costs, these regulations can limit safer in- not guarantee the safety of deposits. stitutions to wealthier areas. Worldwide efforts to Postal banks are a case in point. In many markets, combat money laundering and the financing of ter- they maintain some of the most extensive networks rorism are also threatening to make small transac- of service points for deposit taking, and they offer tions unviable for financial institutions.10 payment services. However, many postal banks are In addition, partnerships that could help safe not subject to banking regulation. Because they are institutions extend their outreach are often discour- considered part of the postal service, their assets are aged by financial regulators in the name of safety. In at risk of leakage into loss-making postal systems. Bosnia and Uganda, for example, regulators recov- This is the case with Caisse Nationale d'Epargne, the ering from recent systemic crises have prohibited postal savings bank in Benin that is in the process of agency relationships that would allow unregulated separating from the postal service. institutions to handle transactions on behalf of reg- Policy frameworks are often inappropriate for ulated ones. financial cooperatives as well. In Uganda, for exam- It would be irresponsible to suggest that regula- ple, SACCOs are not governed by dedicated legisla- tions like these be loosened without careful analysis tion. They operate under a variety of legal regimes, of the implications of such a change. Strict regula- including the Cooperatives, Companies, and NGO tion and supervision is, after all, part of the reason Acts. In many countries, they are not monitored by that safe institutions are safe. Nor is strict regulation appropriate financial regulators, but rather by agen- and supervision the primary bottleneck to extend- cies in charge of all cooperatives. In Uganda, this is ing quality deposit services to the poor. Evidence the Ministry of Tourism, Trade, and Industry, which for this can be found in countries like Uganda, is widely acknowledged to lack the capacity to super- where significant donor and government invest- vise the 1400 SACCOs currently registered. ments in creating a separate regulatory category for Where a dedicated agency for cooperatives does MDIs has not yet led to significant increases in out- exist, as in the Philippines, that agency typically is reach. charged with promoting cooperatives as well as Furthermore, there is a legitimate place for unsu- monitoring them, resulting in a conflict of interest pervised deposit-taking institutions in the financial that can weaken the supervisory function. system, specifically for those that take only deposits Delegating supervision to membership-based feder- from members and are too small to supervise cost ations also can produce a conflict of interest, because efficiently (where supervision costs are measured member institutions may seek to leave the federation against the institution's assets). Research in Uganda to avoid strict supervision. In Mexico, where multi- shows that even unsupervised institutions tend to ple federations have been set up to supervise coop- eratives, institutions have reportedly switched their be safer than informal savings options (Wright and memberships to avoid possible sanctions. Mutasesira 2002). For other types of institutions, regulation and su- 9The 2004 MDI Act allows a limited number of MFIs that meet high pervision can be so strict that it ensures the security performance thresholds to transform into regulated financial intermedi- of deposits, but limits outreach. In Mexico, banking aries supervised by the Bank of Uganda. regulations that mandate expensive infrastructure re- 10 For more information, see Isern et al. 2005. 12 Strategies for Supporting Small-Balance financial intermediaries must offer products that Savings Mobilization combine the security of well-run institutions with the proximity and affordability of informal instru- What can be done to help financial institutions pro- ments. For regulated institutions, such as banks, the vide both safe and accessible deposit services? key is developing less costly delivery mechanisms CGAP's Country-Level Savings Assessments iden- while maintaining high levels of depositor protec- tifies five strategies that could alleviate key demand- tion. Technologies such as mobile phones and and supply-side constraints. Although the impor- point-of-sale networks hold promise, but often tance of specific constraints varies from country to require substantial up-front investment. In the short country, the strategies developed are relevant to term, offering incentives for regulated institutions almost all of the five markets studied. to reach out to unserved markets may be an effec- Document the market. Generating solid data tive use of subsidies. For institutions that build on the size of the low-income market and client points of service in poor areas, incentives could preferences for deposit services provides institu- include a share of the government's payments busi- tions with a sound basis for business planning and ness, temporary exclusivity rights in certain areas, or product development. Making these data widely tax breaks. available and visible could contribute to a change in For less tightly regulated institutions, such as mind set that might encourage previously inacces- cooperatives, strengthening management is often sible institutions to reach out to poor savers with more important. Both types of institutions should deposit services. Donor-subsidized market studies use data on client preferences to design products can increase awareness and contribute to this attitu- that respond to specific client needs. Donors, train- dinal shift. The one market--Uganda--in which ing and technical assistance providers, policy mak- CGAP found ready acceptance of the proposition ers, regulators, and supervisors all have roles to play that the poor do save was also the market where in helping institutions make these upgrades. client preferences were the best documented. Align liquidity-related incentives. Finding out- Help clients make smart savings choices. lets for excess liquidity in strong institutions may Many clients find it difficult to accurately assess the heighten their incentive to offer deposit services to risks of different savings instruments, both formal poor clients. Helping institutions take and manage and informal. Consumer education and financial intelligent lending risks, for example in the micro- literacy training can help poor clients judge the rel- credit market, could drive deposit collection. ative risks of available savings options and use them Improved liquidity management mechanisms that to manage household finances more effectively. enable institutions to place excess funds into safe, Client education on how to assess a financial insti- remunerative investments also could increase their tution's safety is especially valuable. Client knowl- appetite for deposits. edge of the indicators of institutional soundness may At the very least, donors and governments should help alleviate some of their uncertainty about think twice about pumping more liquidity into entrusting their savings to formal financial institu- already over-liquid systems through cheap, easily tions. Such mechanisms could build on the rating accessible credit lines. There may be alternatives that systems often used by apexes, federations, and tech- better preserve incentives for deposit collection. For nical assistance projects and can be promoted to example, an emergency liquidity fund that "guaran- clients through public education campaigns. tees" short-term deposits could help financial Increase capacity in accessible institutions institutions transform them into longer-term assets. and lower costs in safe institutions. To compete In countries with very strict asset-liability regulation, more effectively with informal savings mechanisms, increasing the proportion of short-term deposits that 13 can be transformed into longer-term loans could be these institutions are not supervised--may be a another potential solution. A study of the true practical alternative. The emphasis on performance stability of short-term deposits in a given institu- data also could help build institutional capacity--a tional or country context should underpin both necessary counterpart to improved policy and effi- these strategies. cient industry infrastructure. Establish balanced regulatory and supervi- sory frameworks. Regulatory and supervisory Conclusion frameworks exert a strong influence over financial institutions' ability and incentives to provide secure Poor people worldwide want safe, accessible savings deposit services to low-income savers. In light of mechanisms. Formal financial institutions cannot past crises, they often exhibit a bias toward ensur- fulfill this demand until they can offer services that ing systemic stability. But the restrictions entailed are secure, affordable, and located where poor by this approach can inhibit outreach to poor clients live and work. In the meantime, poor people clients. Often, this pressures regulated institutions will continue to hide money at home, lock up assets to innovate outside the law. For such institutions, in animals, or entrust cash to informal providers-- permitting limited piloting of new products and with all the risks those mechanisms entail. Funders, delivery methods may be a constructive alternative policy makers, and financial institutions themselves to closing off all avenues for experimentation. must work at all levels of the financial system to Stronger monitoring is often needed for more align incentives and create the capacity for formal accessible institutions, especially for postal banks institutions to tap the demand for savings services. and large cooperatives that approach the definition These efforts translate into different activities, of institutions that take deposits from the public. depending on the local context. While offering Financial authorities often are reluctant to regulate incentives to banks to downscale in deposit mobi- a plethora of small institutions because of the costs lization may be the most promising strategy in one involved. As a compromise, some countries have or country, staunching the flow of subsidized on-lend- are considering delegating supervision to member- ing funds to cooperatives or strengthening postal ship-based federations of institutions. These efforts bank management may be more important in have been largely unsuccessful, and international another. Regardless of the specific action, safety and experiences may hold useful lessons. accessibility are the goals. If financial institutions For smaller institutions that cannot be cost- can offer these, they will not only increase access to effectively supervised, nonprudential systems that savings services for poor people, but also create involve collecting and posting performance data-- access to a fortune of deposits for themselves. as well as communicating clearly to members that 14 Appendix: Data on Average Product Costs and that, as the number of branches per population Branch Density increased, the average costs to open an account In the markets studied, average threshold costs decreased. The correlation was even stronger (-0.83) tracked closely to branch density. "Basic deposit when only bank branches were considered. Although products" were defined as the lowest cost savings based on a limited sample of countries, the results or transaction account in a given institution. support the expectation that competition between "Threshold" costs are the direct financial costs to institutions brings down price. open such accounts, comprised of minimum open- ing balances, fees, and any shares that must be pur- Table A-1 Threshold costs for basic deposit prod- ucts in institutions sampled, by country chased in the case of cooperatives. Table A-1 gives the range and averages of thresh- All figures in USD High Low Average Minimum daily wage old costs in institutions sampled by country. The Benin 946.97 9.09 157.43 1.71 institutions represent a cross-section of deposit- Bosnia 6.33 0 0.79 6.55 taking institutions catering to low-income people-- Mexico 88.50 0.88 34.73 4.04 usually mass-market-oriented banks, postal/savings Philippines 35.66 8.91 14.49 6.15 Uganda 28.29 5.66 9.86 0.11 banks, deposit-taking microfinance institutions, and financial cooperatives. For each country, daily Table A-2 Threshold costs as multiples of daily minimum wage is also given. minimum wage To study the relationship between threshold High Low Average costs and population per branch, average threshold Benin 554 5 92 costs for a range of institutions were converted into Bosnia 1 0 0 Mexico 22 0 9 a multiple of the local daily minimum wage. (See Philippines 6 1 2 Table A-2.) Uganda 253 51 88 Figure A-1 illustrates high, low, and average threshold costs and compares them against daily Table A-3: Correlations between measures of minimum wages by country. branch density and threshold costs When these multiples were compared to average High Low Average nationwide population per branch, the correlation Bank branches/million people 0.73 0.56 0.83 Deposit-taking institution 0.57 0.72 0.78 was -0.78 for all deposit-taking institutions, meaning branches/million people Figure A-1 Threshold costs for deposit products 554x $160.00 92x $140.00 $120.00 High $100.00 22x Low Unweighted avg $80.00 Min daily wage $60.00 251x Xx multiple of 9x 6x min daily wage $40.00 253x 0x 2x 88x $20.00 5x 1x 1x 0x 0x $- Benin Bosnia Mexico Philippines Uganda 15 FocusNote No. 37 References Bolaño, Pilar Campos. 2005. El Ahorro Popular en México: Acumulando Activos para Superar la Pobreza. Mex- ico City, Mexico: CIDAC and Miguel Angel Porrúa. Financial Diaries Web site. www.financialdiaries.com. Hertwig, Ralph, Greg Barron, Elke Webber, and Ido Erev. 2004. "Decisions from Experience and the Effect of Rare Events in Risky Choice." Psychological Science 15 (8): 534­9. Please share this Helms, Brigit, Rani Deshpande, Mark Pickens, and Nazaire Sado. 2005. Benin Country-Level Savings As- Focus Note with your sessment. Washington, D.C.: CGAP, August. colleagues or request extra Isern, Jennifer, David Porteous, Raul Hernandez-Coss, and Chinyere Egwuagu. 2005. "AML/CFT Regu- copies of this paper or others lation: Implications for Financial Service Providers That Serve Low-Income People." CGAP Fo- in this series. cus Note 29. Washington, D.C.: CGAP, July. Ivatury, Gautam. 2006. "Using Technology to Build Inclusive Financial Systems." CGAP Focus Note 32. CGAP welcomes Washington, D.C.: CGAP, January. your comments on this paper. Karlan, D. S., N. Ashraf, and Y. Wesley. 2004. Market Survey Report. Universities of Princeton and Harvard. Mimeo. All CGAP publications are Mareco Index Bosnia Gallup International. 2003. Omnibus Survey. Mareco Index Bosnia Gallup Interna- available on the CGAP Web site tional, June. at www.cgap.org. Micro-Enterprise Bank. Market research. Manila, Philippines: Micro-Enterprise Bank. CGAP Pelrine, Richard, and Olive Kabatalya. 2005. "Savings Habits, Needs and Priorities in Rural Uganda." Kam- pala: USAID/Rural SPEED, September. 1818 H Street, NW MSN P3 - 300 Rutherford, Stuart. 2003. "Money Talks: Conversations with Poor People in Bangladesh about Managing Money." Journal of Microfinance 5(2). Washington, DC 20433 USA Ruthven, O., and S. Kumar. 2002. "Fine-Grain Finance: Financial Choice & Strategy Among the Poor in Tel: 202-473-9594 Rural North India." Finance & Development Working Paper 57, IDPM, University of Manches- ter. Fax: 202-522-3744 World Bank. 2001. "Mexico Rural Finance: Savings Mobilization Potential and Deposit Instruments in Mar- ginal Areas." Report No. 21286-ME. Washington, D.C.: World Bank, June 4. Email: cgap@worldbank.org ------. 2005. "Broadening Access to Financial Services among the Urban Population: Mexico City's Un- banked." Washington, D.C.: World Bank. Wright, Graham, and Leonard Mutesasira. 2002. "The Relative Risks to the Savings of Poor People." Nairobi: MicroSave. Wright, Graham, and Paul Rippey. 2003. "The Competitive Environment in Uganda: Implications for Mi- crofinance Institutions and their Clients." Kampala: MicroSave/FSDU/Imp-Act, September. Zeller, Manfred, and Manohar Sharma. 2000. "Many Borrow, More Save, and All Insure: Implications for Food and Micro-finance Policy." Food Policy 25(2): 143­67.