Report No. 27773-BR Brazil Access To Financial Services February 19, 2004 Brazil Country Management Unit Finance, Private Sector and Infrastructure Department Latin American and the Caribbean Region Document of the World Bank CURRENCY EQUIVALENTS Currency Unit:Real (R$) December 2000 US$1=R$1.95 December 2001US$1= R$2.3 1 December 2002 US$ 1=R$3.55 December 2003 US$ 1=R$2.90 WEIGHTS AND MEASURES FiscalYear of Borrower January 1-December 31 ABBREVIATIONS AAA AtividadesAnaliticase Consultivas Analytic and Advisory Activities ABN AMRO AlgemeneBankNederland-Amsterdam-RotterdamBank Algemene BankNederland-Amsterdam-RotterdamBank ABRACOOP Brasileirapara o Desenvolvimento BrazilianAssociationof Cooperatives do Cmperativismo ACEP Agence de Creditpour /'Enterprise Privee(Senega/) PrivateEnterpriseCreditAgency AGAPE AsociacidnGeneralparaAsesorarPequeiias GeneralAssociationfor AdvisingSmall Business ASA Associagaopara Progress0Social Associationfor SocialAdvancement ATMiPAE Post0deAtendimentoBandrio Eletrbnico Automatic Teller Machine BAAC Bancoparaa agriculturae cooperativas agriculturaisde Jailindia Bankfor Agricultureand AgriculturalCooperativesof Thailand BANCOOB BancoCooperativodo Brasil CooperativeBank of Brazil BANEFE BancoSantander SantanderBank BANSlCREDl BancoCooperativoSlCREDl CooperativeBank SlCREDl BASA Bancoda Amazbnia Amazonia Bank BB BancodoBrasil Bankof Brazil BCB BancoCentraldo Brasil Central Bank of Brazil BIS Bancolnternacionalde CompensaGBes Bankfor InternationalSettlements BKD BadanKreditDesas Badan KreditDesas BNB Bancodo Nordeste NortheastBank BNDES BancoNacionalde DesenvolvimentoEconbmicoe Social NationalBank for Economicand Social Development BNDESpar BNDESParticipapjes NationalBank for Economicand Social Development - EquityBranch BOVESPA Bolsa de Valores de Sa0Paulo Stock Exchange of Brazil - Siio Paulo BRAC ComitbDoAvango RuralDeBangladesh BangladeshRuralAdvancement Committee BRDE BancoRegionalde Desenvolvimentodo ExfremoSul RegionalDevelopmentBankfor the ExtremeSouth BRI BancoRakyatlndonesia BankRakyat lndonesia BVR Associagao de BancosPopularesAlem&s e Raiffeisen Associationof GermanPopular Banksand RaiffeisenBanks CAM CentrodeApoyo a la Microempresa MicrobusinessSupportCenter CAMELS Sistemade RatingsBandrios Bank Rating System CARD Centroparaa agriculfura e o desenvolvimento rural Center for Agricultureand RuralDevelopment CAS Estrategiade Assistincia para o Pais CountryAssistance Strategy CBTAL EmprestimodeAssistincia Jecnicaa Modemizaqaodo BancoCentral Central Bank ModernizationTechnicalAssistance Loan CDB Certificadode DepositoBancario BankCertificateof Deposit CDI Certificadode Deposifolnterfinanceiro InterbankCertificateof Deposit CDs Certificadode Deposito Certificate of Deposit CEAPE CentrodeApoio aosPequenosEmpreendimentos SupportCenterfor Small Business i CEF CaixaEconbmica Federal NationalSavings Banks CERUDEB BancoCentenayde DesenvolvimentoRural Centenaly RuralDevelopmentBank CETIP Centralde Custodia e de Liquidago Financeirade Titulos CentralCustodyand Settlementof Securities CGAP GrupoConsultivo deAuxilioaosPobres ConsultativeGroup to Assistthe Poorest CMAC CabsMunicipalesdeAhorroy Credito MunicipalSavingsand LoansOffices CMM Coporacion Mundialde la Mujer Women's World Corporation CMN ConselhoMonetarioNacional NationalMonetaryCouncil CNH CNHCapital CNH Capital COCECRER-RS Cooperativa Centralde CreditoRural do RioGrandedo SulLtda RuralCreditCooperativefrom Rio Grande do Sui COFIDE (DFC) Copora@o Financeirade Desenvolvimento Development FinancialCompany COFINS Contribui@oparaFinanciamentoda Seguridade Social Contributionfor the Financingof SocialSecurity CPF Cadastrode PessoaFisica Cadastre of individuals CPI hdicede Pregosao Consumidor Consumer PriceIndex CPMF Contribui@oProvisoriasobreMovimentagaoFinanceira FinancialTransactionsTax CPR Cedula de ProdutoRural Billof RuralProduct CRAC CabsRuralesdeAhorroy Credito RuralSavings and LoansOffices CRESOL Cooperativa de CreditoSolidario MutualCreditCooperative CVM ComisSaode Valores Mobiliariosdo Brasil BrazilSecurities& Exchange Commission DAl Alternativasde Desenvolvimentolnc. Development AlternativesInc. DGRV Confedera@oAlema de Cooperativasde Credito GermanConfederationof CreditCooperatives DI Deposit0Interfinanceiro InterbankDeposit DiEESE Departamentolntersindicalde Estatisticase Estudos InterTrade UnionDepartmentof Statisticsand Socio-Economic Socioeconomicos Studies EDPYME Empresade Desarrollode la Pequeriay MicroEmpresa Small and Micro BusinessDevelopmentCompany ETENE 0EscritorioTecnicodeEstudosEconbmicosdoNordeste TechnicalOfficeof Economic Studiesfor the Northeast FACS PesquisadaAnalise e da Concorrdncia de Empresas FirmAnalysis and CompetitivenessSurvey FAEJ FundodeApoio ao EmpreendedorJoseense Support Fundfor Joseense Businessmen FAEP FundodeApoioao EmpreendimentoPopular Support Fundfor PopularEnterprise FAMPE FundodeAvalas Microempresase Empresasde PequenoPorte LoanGuarantee Fund for Microand Small Enterprises FAT FundodeAmparo ao Trabalhador Worker Assistance Fund FDI lnvestimentoDiretoEstrangeiro ForeignDirectInvestment FED FundacionEcuatoriana de Desarrollo EcuadorianDevelopmentFoundation FFP FondoFinancieroPrivado PrivateFinancialFunds FGC FundoGarantidorde Creditos CreditInsurance Fund FIE Foment0a lniciativasEconomicas(Bolivia) Furtheranceto EconomicInitiatives FINAME Agincia Especialde Financiamentolndustrial SpecialAgency of IndustrialFinancing FINCA Fundagaode para o Auxilio Comunitariolnternacional The Foundationfor InternationalCommunity Assistance FINEM Financiamentoa Empreendimentos Enterprise Financing FiNSOL FinanceiraSolidaria SA FMM FundacionMundoMujer World Women's Foundation FONDESiF Fondode DesarrollodelSistema Financieroy deApoyoa/ Sector Fundto the Developmentof the FinancialSystem and Support to the Productivo - Fundode Desenvolvimentodo Sistema Financeiroe de Productive Sector apoioa0 SetorProdutivo FONSET Forum NacionaldosSecretariosde Trabalho NationalForumof Labor Secretaries FSTAL SegundoEmprestimodeAssistincia Tecnica SecondTechnicalAssistanceLoan FWWB Friendsof Women's World Banking GCI lnspegaoGlobalConsolidada GlobalConsolidatedInspection GDP ProdutolntemoBrut0 Gross Domestic Product GNI ProdutoNacionalBrut0 GrossNationalIncome GTZ SociedadeAlemi de Cooperagao Tecnica Deutsche Gesellschaflfur TechnischeZusammenarbeit- German IAF Fundagolnteramericana societv for technicalco-operation Inter-AmericanFoundation IBGE lnstitutoBrasileirade Geografiae Estatisfica BrazilianInstituteof Geography and Statistics IBRD Bancolnternacionalparaa Reconstru@oe Desenvolvimento InternationalBankfor Reconstructionand Development ICR Relatorioda ConcluSaoda Execugio ImplementationCompletionReport IDB Bancolnteramericanode Desenvolvimento Inter.-AmericanDevelopmentBank IFC CoporagioFinanceiralnternacional InternationalFinanceCorporation IIC Coporag6olnteramericanade lnvestimento InterAmericanInvestmentCorporation ILO Organizago lnternacionaldo Trabalho InternationalLaborOrganization IMF FundoMonetariolnternacional internationalMonetary Fund IOF lmpostosobre OperagdesFinanceiras Taxon FinancialOperations IPCA indicede Pregosao ConsumidorAmplo IPEA lnstitutode PesquisaEconomicaAplicada Instituteof Applied EconomicResearch LACIF FundoDe lnvestimentoao DesafioAmerica Latina LatinAmerican Challenge InvestmentFund LlBOR Taxa de Juros lnterbancariode Londres London IntetbankOffered Rate LTN letras do TesouroNacional FederalTreasury Bills MDlC Minisferiodo Desenvolvimiento, lndustriae Comercio Exterior Ministryof Development,Industryand Commerce MFls lnstituigdesDeMicrofinance MicrofinanceInstitutions MIS Sistema Delnforma@oDa Gerdncia Management InformationSystem MPME Micro, PequenaeMediasEmpresas Micro,Smalland MediumsizedCompanies NABARD BancoNacionalparaaAgriculturae o DesenvolvimentoRural NationalBank for Agricultureand RuralDevelopment NAFIBO NacionalFinancieraBoliviana NationalBolivian FinanceCompany NCUSIF FundoDo Seguro Da ParteDa UniioDe CreditoNacional NationalCreditUnion Share InsuranceFund NGO OrganizagoNio-Governamental Non-profitGovernmentalOrganizations OCB OrganizagaodasCooperativasBrasileiras Organizationof Cooperativesof Brazil OECD Organizagoparaa cooperag6o eo desenvolvimentoecondmicos Organizationfor EconomicCooperationand Development ORGAPE Organizagiode Apoio aos PequenosEmpreendimentos Organizationof Supportto the Small Enterprises OSCIP Organizagoda SociedadeCivilde lnteressePliblico Civil SocietyOrganizationwith a PublicInterest PAA PosfoAvanpdo doAtendimento Advanced ServiceOutpost PAB PostoAvanpdo Bandrio Advanced BankingOutpost PAC PostodeAtendimentoCooperativo Cooperative Services PACRE PostoAvanpdo de CreditoRural RuralCreditOutposts PAP PostodeArecadago e Pagamentos PaymentPosts PCO Postode Comprade Our0 PCPP Prcgramade CreditoProdutivoPopular Programof PopularProductiveCredit PCR Programa de Capitalde Risco Venture Capital Program PDI Programade Desenvolvimientolnstitucional InstitutionalDevelopmentProgram PFSECAL EmprestimoProgramaticoparaAjuste ao SetorFinancier0 Programmatic FinancialSector Adjustment Loan PIS/ PASEP Programade lntegragio SociaVProgramade Formago do Patrimdnio Programof Social Integration/Programof Formationof the Patrimony do SewidorPliblico of the Public Server PLANFOR PlanoNacionalde Qualificagaodo Trabalhador NationalPianfor Worker Qualification POF Pesquisade Orpmentos Familiares Periodicconsumerexpendituresurveys PPV PesquisaSobrePadr6esde Vida Householdsurvey PROCERA-FNE Programade CreditosEspeciaispara Reforma Agtdria - FundoConstitucionalde LandReform Program - NortheastConstitutionaland Financial Fund Financiamentodo Nordeste PROES Programade lncentivoa Redugo do Setor PliblicoEstadualna Reductionof State Level Participationin Banking Activities AtividadeBandria PROLEITE Programade lncentivoa Mecanizagio,ao Resfriamentoeao Transporte Granelizadoda Produgode Leite PRONAF ProgramaNacionalde FortalecimentodaAgriculturaFamiliar NationalProgramto StrengthenFamilyAgriculture PROPASTO ProgramaNacionalde Recuperagiode PasfagensDegradadas Nationalprogramof Recoveryof DegradedPastures PROSOLO Programade lncentivoao Us0 de Corretivosde Solos Programof Incentivefor the Use of Soil Products PROTRABALHO Programade Promqio do Emprego e Melhoriada Qualidadede vida Programto Promote Employment and Improvethe Quality of Lifeof do Trabalhador Workers RlTS Redede lnformagdesparao Terceiro Setor Third Sector InformationNetwork ROA Retornoemrecursos Returnon Assets RTGS Sistema de Transferencia de ReSeNaS Real-timeGross Settlement SAL EmprestimoparaAjustamentoEstrutural StructuralAdjustment Loan iii SBEF Superintendinciade Bancose EntidadesFinanceiras SBS Superintendenciade Bancay Seguros Superintendencyof Bankingand Insurance SBS , Servigospara PequenosEmpreendimentos SmallBusinessService SCMs Sociedadesde Creditoao Microempreendedor MicroCreditCompanies SDI hdicede Depedinciaao Subsidio Subsidy DependenceIndex SEBRAE Servigo BrasileirodeApoio As Microe PequenasEmpress BrazilianServicesto Support Micro and Small Enterprises SELIC SistemaEspecialde Liquidagrioe Custodia OvernightInterest Rate SEWA Associago de MulheresProfissionaisLiberais Self-EmployedWomen's Association SHG GruposdeAutoAjuda Self-HelpGroups SiCOOB Sistemadas Cooperativasde Creditodo Brasil Brazilian Credit CooperativeSystem SlCREDl Sistemade CreditoCooperativo CreditCooperativeSystem SME Empresapequenae media Smalland MediumEnterprise SPC Secretariade PrevidbnciaComplementar Secretariatof ComplementarySocialSecurity SRO OrganizagoAuto-regulada Self-RegulatoryOrganization TA Assistincia Tecnica TechnicalAssistance TJLP Taxa de Jurosde 1ongo Prazo LongTerm Interest Rate TR Taxa Referencial Reference Rate UAD UnidadeAdministrativaDesmembrada Separated AdministrativeUnit UMBNDES UnidadeMonetariado BNDES BNDESMonetaryUnit UNICEF Fundolnternacionalde EmerginciadasNagdes Unidospara Crianps UnitedNationsInternationalChildren'sEmergency Fund USAID Agincia GovernamentalAmericana US Government Agency WBES PesquisadoAmbientede NegcjcioMundial World BusinessEnvironmentSurvey WDI lndicadoresde DesenvolvimentoMundial World DevelopmentIndicators woccu Conselho Mundialde Cooperativasde Poupanp e Credito WorldCouncilof Credit Unions WWB Women's World Banking Country Director: Vinod Thomas LeadEconomist Joachim von Amsberg Sector Director: Danny M.Leipziger Sector Manager: Fernando Montes-Negret iv This study was conceived and undertaken under thejoint leadership of the Brazil Country Unit of the World Bank, led by Gobind T. Nankani and Vinod Thomas, Country Directors and Joachim von Amsberg, Lead Economist, and the Financial Sector Unit of the Latin America and the Caribbean Region, led by Fernando Montes-Negret, Sector Manager, and Danny Leipziger, Director. The project has received extensive support from the Bank's Financial Sector Board, in recognition of the importance of access to financial services for reducing poverty and raising welfare. This work is the outcome of a process of collaboration of an extensive team of persons, from within the Bank, from Brazil, and external consultants, ledby Anjali Kumar, Lead Financial Economist, Latin America and the Caribbean Region. Scholars in Brazil who collaborated and participated with this effort include, notably, Professor Rosane Mendonqa and students from IPEA and NESO/UFF (Rio de Janeiro) in particular Cristine Campos and Daniel Santos. Professor Mendonqa also helpedcoordinate a field survey, undertaken by the firm Sensus (Belo Horizonte), led by Ricardo Guedes. The team benefited from the guidance of Professor Ricardo Paes de Barros, IPEA. Core backgroundpapers were prepared by Armando Castelar Pinheiro (IPEA, Rio de Janeiro), the firm of Thomas Felsberg and associates (Slo Paulo), and a team headed by Professor Ricardo Leal and students from the COPPEAD business school (Rio de Janeiro). Contributions were also received from MoysCs Kessel, formerly of the Central Bank of Brazil. The study was also enriched by discussions and contributions from the Comunidade Solid6ria and SEBRAE and with Jaime Mezzera and Henry Jackelen of the Brasilia offices of the L O and UNDP. The team met with numerous financial institutions in Brazil and close discussions were maintained during the process with financial sector associations, notably FEBRABAN, the association of banks, and also ANFAC and ABEL, which represent the factoring and leasing industries. Particularly valuable collaboration was extended by the team of Beatriz Azeredo da Silva of the BNDES Bank, and Lara Goldmark, consultant to BNDES. World Bank and IFC staff and consultants contributing to this report included, notably, Thorsten Beck, McDonald Benjamin, Thomas Glaessner, Mario Guadamillas, Margaret Miller, Nataliya Mylenko, Susana Sanchez, Sophie Sirtaine and Jacob Yaron. Contributions were also provided by Alfred0 Ebentrich, Lydie Ehouman, Daniela Klingebiel, Peer Stein and Robert Vogel. Significant background papers were prepared by a team from Shorebank Advisory Services (Chicago and Washington). Ricardo Gonqalves and Adam Parsons contributed technical notes and participated in the initiation, preparation and finalization of all parts of the study. Micky Ananth providedextensive support with document preparation. Thoughtful comments from a number of Bank colleagues as well as external persons are reflected in this study. Reviewers and advisors in the Bank have included Mauricio Carrizosa, Robert Christen, Francisco Ferreira, James Hanson, Aart Kraay, Soledad Martinez, Steven Schonberger, Shahid Yusuf and Dimitri Vittas. Comments and contributions have also been provided by Marguerite Berger and Rogerio Studart of the IADB and Ashoka Mody and Kenichi Ueda of the IMF. Professor Robert Townsend of Chicago University provided valuable insights on alternative frameworks for analysis. Professor AndrC Urani of Instituto de Estudos do Trabalho e Sociedade - IETS was a special advisor to the project and Professor Marcel0 Neri of the Fundqlio Getulio Vargas provided helpful comments on Brazilian data sources at the outset of the study. V The key counterpart agency for the project has been Brazil's Central Bank. Initial endorsement for the study came from the Central Bank Governor Am'nio Fraga and discussions of the final report continued with Central Bank Governor Henrique Meirelles, Afonso Bevilaqua and Eduardo Loyo. Director SCrgio Darcy and his office, including Clarence Hillerman Jr. and Marden Soares, provided guidance throughout, especially on microfinance and on financial regulation in Brazil. Subsequent discussions on themes of financial access within the Central Bank included especially Eduardo Lundberg and Mircio Nakane. Discussions with the Ministry of Finance were held with Marcus Lisboa and Otaviano Canuto. The report benefited immensely from intensive reviews by technical teams within the administration, including notably Wagner Guerra Jr., Roberto Shoji and OtAvio Damaso at the Ministry of Finance and Amaro Gomes at the Central Bank. Early versions of this study have been presented and discussed at seminars and workshops in Washington and Brazil. In Washington, early results from Brazil have been compared with ongoing parallel work in Mexico and Colombia, organized by Tova Solo of the Latin America region, FPSI unit. Events organized inBrazil have included an IETS/World Bank seminar on inequality and social justice at Rio de Janeiro, held in tandem with an IFC workshop on microfinance, a workshop which included themes of urban poverty, by the Cities Alliance at SHo Paulo, in October 2003, and a Central Bank workshop on microfinance. And recently results of this study were shared with an audience in Delhi, India, drawing comparisons between India and Brazil on financial access. Valuable messages were received from participants at these events, which have also been reflected. We would like to take this opportunity to thank all the persons in government, financial and academic communities who have provided their time, thoughts and contributions to the team and to the study. The views and findingsexpressedinthis report are exclusively those of the World Bank. vi Tableof Contents TABLE OF CONTENTS PageNo . ACKNOWLEDGEMENTS ........................................................................................................................... v EXECUTIVE SUMMARY .......................................................................................................................... XV PART 1. ASSESSING ACCESS .................................................................................................................. 1 1 . .......................................... 1 1.1 Introduction:The Importanceof Access................................................................................................................... GROWTH.POVERTYREDUCTIONANDACCESS FINANCIALSERVICES TO 1 1.2 5 1.3 The Metricsof Access: Challenges of Measurement............................................................................................... IssuesExamined: Structure of ProposedReport..................................................................................................... 7 2. .............................. Trendsand Distributionsof Bank Services: Branchesand Service Points .............................................................. BASICSUPPLY SIDEMEASURESOFFINANCIALSERVICESOFFEREDINBRAZIL . 9 2.1 9 2.2 Ownership: Publicversus Private Provisionof Bank Services............................................................................... 12 2.3 InternationalComparisonsof Bank Services.......................................................................................................... RegionalDistributionof FinancialServices............................................................................................................ 12 2.4 17 2.5 Econometric Analyses of Factors Associated with Bank Service Supply............................................................... 18 3. DEMANDSIDEMEASURES OF ACCESS EXTRACT . FROMAN ENTERPRISE SURVEY ......................... 21 4. DEMAND MEASURES SIDE OF ACCESS SURVEY OF URBANINDIVIDUALS . ........................................ 22 4.1 Survey Results Accessto FinancialInstitutions................................................................................................... . 24 4.2 Survey Results.Depositsand Savings Behavior.................................................................................................. 25 4.3 Survey Results .Loansand Credit Services.......................................................................................................... 26 4.4 Survey Results .Paymentsand MoneyTransmission Services ............................................................................ 28 5. ANALYSISOF SURVEYRESULTS. FACTORSASSOCIATED WITHACCESS? WHAT ARE .....................29 5.1 The Impactof Locationon FinancialAccess.......................................................................................................... 30 5.2 The Role of DifferentInstitutionsand FinancialAccess ......................................................................................... 32 5.3 The Role of RespondentCharacteristicsand FinancialAccess............................................................................. 35 6. DETERMINANTS ACCESSANECONOMETRIC OF . INvESTIGATION ................................................... 37 6.1 38 Determinantsof InstitutionChoice.......................................................................................................................... Determinantsof the Volume of Credit Requestedand Approved........................................................................... Determinantsof Access to FinancialServices ....................................................................................................... 6.2 39 6.3 39 7. SUMMARY OF FINDINGSAND POLICY IMPLICATIONS ........................................................................ 41 7.1 OverallFindings.SynthesizingSupply and Demand Side Findings.................................................................... 41 7.2 Implicationsfor Policy Directions............................................................................................................................ 42 ANNEX 1.1 FINANCIALMARKETS WELFARE-ENHANCEMENT AND ................................................... 44 ANNEX 1.2 NOTE ESTIMATION:URBAN FINANCIAL ON SURVEY ....................................................... 47 PART2 . EXPANDING MICROFINANCE ................................................................................................ 51 1 . MICROCREDIT. ACCESS AND POVERTY: A NEW PARADIGM? ........................................................... 51 Vii Brazil:Access to FinancialServices 2 . MICROFINANCE ..................................................................... 57 2.1 Evolutionof the PoliticalDebateand the LegalFramework................................................................................... IN BRAZIL: EVOLUTIONAND STATUS 58 2.2 Establishmentof CrediAmigo -A Major Player in Microfinance............................................................................ 60 2.3 Evolutionof the Government'sRole -BNDESand SEBRAE................................................................................ 63 2.4 Microfinancein BrazilToday 69 2.5 InstitutionalCapacity of Brazil's Microfinance Performanceand Impact............................................................. - CurrentStatus-........................................................................................................ 73 3 . MICROFINANCE INBRAZIL: CONSTRAINTSAND CHALLENGES ........................................................ 77 3.1 MicrofinanceProducts and FundingSources 77 3.2 MicrofinanceOrganization - Linksto Banks........................................................................................................... 79 ......................................................................................................... 3.3 PublicBanksand CompetitionIssues .................................................................................................................... 83 3.4 MicrofinanceMethodology -Principlesof Good Practice...................................................................................... 84 3.5 OtherConstraintson Expansion............................................................................................................................. 85 4 . MICROFINANCE REGULATIONAND SUPERVISION: FUTUREPOLICY OPTIONS ................................ 85 4.1 ShouldMicrofinancebe Regulatedor Supervised?............................................................................................... 85 4.2 SpecificPrudentialNorms: MinimumCapital.Capital Ratios and Credit Risk ....................................................... 87 4.3 SupervisionIssues - Who ShouldSupervise?........................................................................................................ 89 5 . THECREDIT COOPERATIVEMOVEMENT ITSCONTRIBUTIONSTOACCESS AND .............................. 90 5.1 The PrincipalPlayersin CooperativeFinance........................................................................................................ An Overviewof the CreditCooperativeSector in Brazil......................................................................................... 90 5.2 91 5.3 Size. Compositionand ServicesOffered................................................................................................................ 93 5.4 Penetrationof Brazil'sCredit Cooperatives:InternationalComparisons................................................................ 95 5.5 FinancialStructureand Performanceof BrazilianCreditCooperatives................................................................. 96 6 . THELEGAL. REGULATORY SUPERVISORYFRAMEWORKFORCREDIT COOPERATIVES AND ........102 6.1 102 6.2 Supervision........................................................................................................................................................... LegalFrameworkand PrudentialRegulation....................................................................................................... 103 6.3 ConceptualIssuesin the Regulationand Supervisionof CreditCooperatives.................................................... 104 6.4 Suggestionsfor Brazil........................................................................................................................................... 106 ANNEX 2.1 MICROFINANCE ANDTECHNOLOGY .............................................................................. 108 ANNEX 2.2 BNDESMFIPORTFOLIO ANALYSIS ............................................................................. 114 PART3 . DOWNSCALING PRIVATE BANKS ........................................................................................ 117 1. INTRODUCTION ANDS ~ R ......................................................................................................... Y 117 2 . DEPOSIT SERVICES SMALL CLIENTSAND SPECIAL SAVINGS . ........................................................ 120 2.1 120 The Client's Perspective- Demandfor BankAccountsand DepositServices.................................................... Deposits Below R$5,000 ...................................................................................................................................... DepositStructure-An Overviewof AggregateDeposits..................................................................................... 2.2 122 2.3 124 2.4 New Possibilitiesfor Small SavingsOtherthan Deposits - Retail Bonds ........................................................... 126 3 . CREDIT SERVICES PROVISIONOF SMALL LOANS . .......................................................................... 126 3.1 The Provisionof Small Loans............................................................................................................................... 127 3.2 SmallSize TransactionsversusTransactionswith Low IncomeSegments......................................................... 129 4 . ENTRY REQUIREMENTS.PRICESANDTRANSACTION COSTS ........................................................... 130 4.1 Entry Requirements. Client's Perspective.................................................................................................... the 130 4.2 Pricingof BankingServicesfor Small Enterprises............................................................................................... Pricingof BankingServicesfor Individualsand Households................................................................................ 132 4.3 135 5 . DELIVERYCHANNELS BRANCHES. . ATMs AND CORRESPONDENT BANKING ............................... 136 Viii Tableof Contents 5.1 BankingCorrespondents...................................................................................................................................... Brickand MortarInstitutions.Bank Branches..................................................................................................... 137 5.2 137 5.3 BankAutomaticTeller Machines.......................................................................................................................... 140 6. NEWTECHNOLOGIES . USEOFTHE INTERNETAND PHONEBANKING ........................................... 142 6.1 Telephone Banking............................................................................................................................................... 143 6.2 InternetBanking................................................................................................................................................... 146 7. DOWNSCALINGOFBANKS ELSEWHERE WHAT CANBANKS . DO? .................................................. 148 7.1 Downscaling- A New Imagefor a DifferentCommunity?................................................................................... DownscalingThrough Partnerships...................................................................................................................... 149 7.2 152 7.3 Downscaling -SuggestedGood Practice............................................................................................................ 153 A"~x3.1BANKDOWNSCALING-ONEBANK'SVIEWPOINT ......................................................... 156 PART4. PARTNEIUNGNON-BANKS .................................................................................................. 165 1. INTRODUCTION .................................................................................................................................. 165 2. ........................................................................................................................................ 169 Brazil'sFactoringRegime:Regulation.Supervisionand Tax Treatment............................................................. Factoringin Brazil.Evolutionand Development.................................................................................................. FACTORING 2.1 169 2.2 176 2.3 Factoring .Summary and Key Issues.................................................................................................................. 178 3. ............................................................................................................................................. 180 3.1 Origin. and Conceptand Evolution....................................................................................................................... LEASING 180 3.2 184 3.3 The Scopefor Extensionof Leasing..................................................................................................................... Legaland SupervisoryFrameworkand Tax Treatmentof Leasing...................................................................... 185 4. CONSUMERFINANCE FORINDIVIDUALSAND TRADE FINANCE ENTERPRISESFOR ......................... 187 4.1 Trade Financeand Vendor Credit........................................................................................................................ ConsumerFinanceCompanies............................................................................................................................ 187 4.2 189 PART5. CHANNELINGRURALFINANCE .......................................................................................... 191 1. INTRODUCTION: SPECIALIZEDFINANCE ANDDIRECTED CREDIT INBRAZIL ................................. 191 2. AN OVERVIEW OFRURALFINANCE BRAZn ................................................................................ 194 2.1 The Basic Modeland Volumesof OverallCredit.................................................................................................. IN 194 2.2 PrincipalPlayers:Demandfor RuralFinance....................................................................................................... 195 2.3 AgriculturalInsurance........................................................................................................................................... PrincipalSources: Credit Supply......................................................................................................................... 196 2.4 203 3. AN ANALYSIS OFTHEPRESENTSYSTEMOFAGRICULTURALCREDIT ............................................ 206 3.1 Output Levelsand Credit...................................................................................................................................... 206 3.2 A Disaggregationof Subsidiesand Estimationof SubsidyDependence............................................................. Outreach............................................................................................................................................................... 208 3.3 211 4. .................................... 212 4.1 Oldand New Approaches: GeneralPrinciples..................................................................................................... DESIGNINGRURALFINANCIAL SYSTEMS PRINCIPLESAND EXPERIENCE . 212 4.2 First Principlesof the NewApproach-An Efficient EnablingEnvironment......................................................... 214 4.3 The LegalFrameworkfor ContractEnforcement - RuralCollateral ..................................................................... 215 4.4 InformationalBarriers........................................................................................................................................... 216 4.5 Assessing the Meritsof Direct Interventionsin RuralCreditand Savings............................................................ 217 4.6 DevelopingInsuranceServices in RuralAreas .................................................................................................... 218 ix Brazil: Access io Financial Services 5. CONCLUSIONS AND RECOMMENDATIONS .................................................................................................. 220 ANNEX 5.1 THE SUBSIDY DEPENDENCE INDEX:CALCULATION RATIONALE AND ......................... 223 ANNEX 5.2 OLDAND NEWAPPROACHES RURALFINANCE:GOALS TO AND PRINCIPLES .............225 ANNEX 5.3 THENEWAPPROACH RURALFINANCE: TO INDONESIA'S BRI-UNIT DESAS ..............227 PART 6 . INSTALLING INSTITUTIONALINFRASTRUCTURE ............................................................... 229 1 . CREDITOR RIGHTS. SECURITY INTERESTSAND ACCESS CREDIT TO .............................................. 231 1*1 231 1.2 The Judicial Enforcementof Creditor Rights........................................................................................................ Legal Protectionto Creditor Rights....................................................................................................................... 233 1.3 Differential Impactson Small Borrowers.............................................................................................................. 236 1.4 Ongoing GovernmentMeasuresand Suggestionsfor Further Reform................................................................ 236 2 . SECURITY INTERESTS ......................................................................................................................... 238 2.1 Institutions and Services Requiredfor Securing Credit Interests......................................................................... 239 2.2 Cost of Public Deedsand Registries.................................................................................................................... 241 2.3 Search Processesat Registries........................................................................................................................... 242 2.4 Other Obstacles Faced in the Registration Process............................................................................................. 245 2.5 Suggestions for Simplifyingthe Creation of Security Rights ................................................................................ 246 3 . CREDITINFORMATIONAND CREDITREPORTING ............................................................................. 246 3.1 Sources of Credit Informationin Brazil................................................................................................................. 247 3.2 Brazil's Credit ReportingIndustry - International Comparisons............................................................................ 251 3.3 Cost Differentials in Credit Reporting................................................................................................................... 254 3.4 Legal and RegulatoryIssues................................................................................................................................ 255 3.5 Suggestionsfor Policy: Strengtheningthe Credit Reporting Industry................................................................... 257 PART 7 . ENLISTING THE GOVERNMENT .......................................................................................... 259 1 . GOVERNMENTPOLICY AND ACCESS: MACRO REGULATORYAND ENVIRONMENT ........................ 262 1.1 The Government's Borrowing Requirement......................................................................................................... 262 1.2 Taxation................................................................................................................................................................ 263 1.3 Directed Credit...................................................................................................................................................... 264 1.4 Labor Market Policies........................................................................................................................................... 265 1.5 RegulatorySegmentation..................................................................................................................................... 265 2 . PROACTIVEGOVERNMENT POLICIES: SUPPORTIVE MICROECONOMICMEASURES ...................... 265 2.1 Legislative Provisionsto ExpandOutreach.......................................................................................................... 267 2.2 Monitoringand InformationTracking.................................................................................................................... 272 2.3 Financial Support.................................................................................................................................................. 273 2.4 InstitutionalSupport.............................................................................................................................................. 279 2.5 EnhancingCompetition in the Provisionof Financial Services............................................................................. 282 3 . CONCLUSIONS .................................................................................................................................... 283 APPENDIX . STATISTICAL TABLES ..................................................................................................... 285 REFERENCE ......................................................................................................................................... 339 X Table of Contents LISTOFTABLES Table 1 1 Brazil.IncomeDistributionin InternationalComparison................................................................................................. . Table 1.2 Depthof FinancialMarkets - Braziland 2 Table 1.3 BranchDensityacross BrazilianRegions...................................................................................................................... 13 Other Emerging Economies (199912000)......................................................... 3 Table 1.4 Provisionof Bank Servicesacross Regions.................................................................................................................. Table 1.5 Provisionof BankServicesacross Municipalities(1996) .............................................................................................. 16 Table 1.6 Publicand PrivateProvisionof Bank Servicesacross Regions.................................................................................... 16 17 Table 1.7 Brazil-BankExplains 18 Table 1.8 Brazil Table 1.9 Brazil What Explains Publicversus PrivateBank Branch ServicesacrossMunicipalities?........................................ -- What Branch Density: InternationalComparisons........................................................................................... Bank Branch Services acrossMunicipalities? (All Banks)....................................................... 19 20 Table 1.10 Table 1.11 Indicatorof Access :Access to FinancialInstitutions.................................................................................................. FinancingConstraintsin Brazilin InternationalComparisons..................................................................................... 21 24 Table 1.12 27 Table 1.13 Access to Loansand Credits Access to Loansand Credits....................................................................................................................................... 28 Table 1.15 EconometricResults:Determinantsof Volume of Credit Requestedand Approved................................................... Table 1.14 EconometricResults - Determinantsof -Reasonsfor Access to FinancialServices-Select Results............................................ Loan Refusals....................................................................................... 39 39 Table 1.16 Table 1.I 7 EconometricResults: The Probabilityof UsingPublic Banks for Real Estate Purchase............................................. EconometricResults:The Probabilityof UsingPublic Banks for Access, Depositsand Credit .................................. 40 40 Table 2.2 BNDES Programof Supportto Microentrerprises(1997-2002).................................................................................... Table 2.1 Brazil- CreditCooperativesand MicrofinanceInstitutions(1997-2002) ....................................................................... 53 Table 2.3 Main MicrofinanceProvidersin Brazil(end2001)......................................................................................................... 64 70 Table 2.4 MicrofinancePenetration Table2.5 Brazil:Selected PerformanceIndicatorsof MainMicrofinanceInstitutions................................................................... - Braziland Other LatinAmerican Countries(2001)............................................................. 71 Table 2.6 LoanQuality Indicators.SelectedSampleof MicrofinanceInstitutionsin Latin America.............................................. 73 Table2.7 OperationalEfficiencyIndicators,SelectedSampleof MicrofinanceInstitutionsin LatinAmerica ............................... 74 Table2.8 Comparisonof Average LoanSize across LatinAmerica ............................................................................................. 76 76 Table2-10 COFIDE Financingto MicrofinanceInstitutionsby December2001 ........................................................................... Table2.9 Sources of Fundingand InterestRatesChargedby Main BrazilianMicrofinanceInstitutions...................................... 79 Table2.11 Capital Requirementsfor RegulatedMicrofinanceInstitutions.................................................................................... 85 Table2.12 Numberof Cooperativesin Brazil, by Type of Cooperative......................................................................................... 88 Table2.13 Distributionof RuralCredit by Purposeand LendingInstitution.................................................................................. 92 Table2.14 CresolMembershipby Farm Size and Annual Income, 1999..................................................................................... 94 Table2.15 Braziland Other Countries:Credit Cooperatives(2001) (US$ m and percent)........................................................... 95 Table2.16 Evolutionof WOCCU- AffiliatedCredit Unionsin Brazil (US$m and percent)............................................................. 96 97 Table 2.17 Classificationof Assets by Levelof Risk: Banksand Select Cooperatives(%) .......................................................... Table 2.18 Key FinancialIndicatorsfor the SicrediCooperativeSystem.................................................................................... 99 Table 3.1 Brazil- Structureof - Table 3.2 Three Main Reasonsfor Wantinga BankAccount...................................................................................................... Depositsby Type and Institution AverageDepositI Client (R$ 000) (December2001)........... 100 122 Table 3.3 Brazil:One Year of Treasury Direct Retail BondSales (January2002to January 2003) Table3.4 Share of Brazil'sCommercialBanks in the SmallLoan Marketin terms of LoanValue (R$000, November2001).... 126 ........................................... 125 - November2001) ..........128 128 Table 3.6 The Importanceof Small Loansfor the 10 LargestBrazilianBanks (Valueand %share, November2001) .............. 129 Table 3.5 Recent Evolutionof Bank Sharesinthe Small Loan Market (%shares. December2000 Table3.8 Brazil-Surveyon Accessto FinancialServices Reasons for Not Havinga BankAccount ..................................... Table 3.7 Non Bankswith a Significant MarketShare inthe SmallLoan Market(R$000 and percent.November 2001) ........130 131 Table3.10 InterestRates, OutstandingStockand New Loansto Consumersby LoanType (Free Credit) ............................... Table3.9 Feesand Requirementsfor Sight DepositServicesat Key BrazilianBanks(2002) ................................................... . 133 Table3.11 Constraintsto EnterpriseOperationsand Growth,an InternationalComparison...................................................... 134 135 Table3.12 Average TransactionCosts by Type of Banks in Brazil Table3.13 TransactionCostsComparison,Banksand Non Banks in Brazil.............................................................................. -Dataon all Banks ............................................................ 136 136 Table 3.15 Correspondentsin Brazil(2001and 2002)................................................................................................................ Table3-14 Types of DistributionChannelsUtilized by DifferentBankGroups .......................................................................... 137 138 Table 3.17 ATM Fleet in the USAand in Brazil........................................................................................................................... Table3.16 Geographic Coverageof the BankingSystemand the PostalNetwork.................................................................... 139 Table3.18 SouthAfrica - DownscalingThrough Partnershipsby CommercialBanks............................................................... 141 150 Xi Brazil: Access io FinancialServices Table 3.19 Peruand Bolivia: Banksand Finance Companies in Microfinance(US$ millionand %)........................................... 151 Table 3.20 Other LatinAmerican Banksand Finance Companies in Microfinance .................................................................... 152 Table 3.21 Sources of External Financingfor Peruvian MicrofinanceInstitutions(Dec2000-nuevos soles m)........................ Table 4.1 FinancialSystemAssets By InstitutionalType........................................................................................................... 152 165 Table 4.2 Brazil- Finance, Leasingand FactoringCompanies (December 2001) (R$ billion) 167 Table 4.3 Brazil Table4.4 Differencesbetween FinancialIntermediationand Factoring...................................................................................... - FactoringCompaniesin ................................................... SelectedStates and Regions(2001)........................................................................ 170 174 Table 4.5 Braziland Select Other Countries - Comparisonof Table4.6 Summaryof Characteristicsof Financialand OperationalLeasesin Brazil................................................................ FactoringTurnover (US$ million,2000)..................................... 176 181 Table 4.7 Braziland Other countries Table 5.1 Brazil Aggregate Creditto the RuralSector (1969-2002) (Constant2001 R$ millions)........................................... 183 Table 5.2 Brazil:AgriculturalLandholdingsare HighlyConcentrated......................................................................................... - - PresentValue of LeasingContracts(US$ billion, 1999)................................................. 195 196 Table 5.3 Brazil - Sources ........................................................................ 197 Table 5.4 PRONAFProvides Low-CostLoansto Family Farms................................................................................................. 200 of Financefor RuralCredit(1995-2002)(R$ million) Table 5.5 Agricultural Insuranceis MinimalRelativeto the Total InsuranceMarket(2001)........................................................ - Agricultural 204 Table 5.7 Agricultural ProductivityRose Evenas FormalAgriculturalCredit Declined............................................................... Table 5.6 Brazil Insurance:Basic Informationand Volume of PremiumslLosses.................................................. 205 Table 5.8 Brazilian Farm Units,Area, GrossValue of Productionand Total Financing.............................................................. 207 209 Table 6.1 Frequencywith which privatepartiesresortto justiceto postponeobligations,by area of law Table 6.2 Likelihoodof judges' decisionsbeingpoliticizedby type of cause.............................................................................. ................................... 234 Table 6.3 Brazil: Registry Offices in State Capitalsand Two major cities in Siio Paulo.............................................................. 235 239 Table 6.4 Notarial Feesfor Draftinga Public Deedof Mortgagein the Stateof Silo Paulo........................................................ Table 6.5 RegistrationCostsfor Deedsand Documents in the State of Siio Paulo.................................................................... 242 243 Table 6.6 Brazil Table 6.7 Cross-CountryComparisonsof PrivateCredit Bureaus: Key Characteristics............................................................ - Major Credit InformationProviders(2002)...................................................................................................... 249 253 Table 6.8 Cross CountryComparisonof PrivateCredit Bureaus: ConsumerAttention Issues ................................................... Table 6.9 Braziland Other Countries:Cost of a Credit Report ................................................................................................... 254 255 LISTOFFIGURES Figure1 1 Brazil:Growthof FinancialInstitutions( Nos., 1993-2002).......................................................................................... . - Evolutionof 10 Figure 1.3 Brazil- Ownershipof Bank ServicePoints.................................................................................................................. 12 Figure1.2 Brazil FinancialServiceOutlets(1994-2002)......................................................................................... 11 Figure 1.4 Brazil Figure 1.5 Brazil- Municipalitieswith no services ........................................................................................................................ -BranchDensityby State.................................................................................................................................. 14 15 Figure 1.6 Braziland Other Countries - Population 17 Figure 1.7 Depositand Savingbehavior Figure 1.8 Role of Locationin FinancialAccess............................................................................................................................ 31 -Survey of Urban Individuals....................................................................................... Servedper Bank Branch............................................................................... 26 Figure 1.9 The Roleof Incomein DifferentMeasuresof FinancialAccess................................................................................... 37 Figure2.1 Brazil:Growthof Loan Portfolioand Client Baseof Main MicrofinanceProviders....................................................... 61 Figure2.2 GrameenBank. Bangladesh;Self-HelpGroups:India;and Top 10 MFls: Bolivia 72 Figure2.3 Comparisonof InterestRates Chargedon Loanswith other Institutionsin LatinAmerica .......................................... - Growth........................................ Figure2.4 Capital, Depositsand Loansof the Sicoob and Sicredi CooperativeSystems. 1998-2001......................................... 75 Figure2.5 Cooperativesin Brazil,by Type of Cooperative,as of 2001 ........................................................................................ 91 Figure2.6 Evolutionof Profitsand Equityfor the SicoobCooperativeSystem........................................................................... 93 Figure3.1 Brazil InterestRatesfor SpecialSavingsDepositsand Term Deposits(1999-2002)............................................. Figure3.2 Brazil Structureof AggregateDeposits -- 101 120 Figure3.3 Brazil 123 Figure3.4 Brazil Structureof Sightand SpecialSavingsDepositsfrom R$1,000 to R$5,000 (December2001) -- Structureof Depositsbelow R$5,000 by Type, Institution, Nos. of Clientsand Value (December2001)..... 121 - December2001 ....................................................................................... ..................... Figure3.5 Creditoperationswith nonearmarkedfunds (R$ billionJuly 00 to Dec 02)................................................................ 124 135 Figure3.6 Brazil - RegionalDistributionof 141 Figure3.7 Braziland Other Countries - BankTransactionsand ATMs Relativeto Populationand Income................................................................ Cellular Phone Costs .............................................................. 143 Figure 3.8 Teledensityand GDP per capita and Telecommunications - Braziland - Braziland Other Countries.......................................... 144 Figure3.10 Braziland Asia -On-LineInternetBanking............................................................................................................. Figure3.9 MainTelephoneline, Cellularand Internetpenetration Other Countries................................................ 144 147 Figure3.11 Braziland the USA - InternetUse and Online Banking............................................................................................ 148 Tableof Contents Figure4.2 Evolutionof the Portfolioof FactoringCompaniesAssociatedwith ANFAC .............................................................. Figure4.1 Brazil. to Companies and Individualsby Instrument(%)............................................................................... Loans 166 171 Figure4.3 FactoringCompanies - PortfoliobySector (2000-2001)........................................................................................... 171 Figure4.4 Brazil- Evolutionofthe 172 Figure4.5 Brazil ....................................... Figure4.6 Brazil Types of Indexationin LeasingContracts(1998-2003*)................................................................................ 182 Figure4.7 Evolutionof LeasingOperationsby Sector (1993-2003*).......................................................................................... -- Evolutionof"Purchase Factor". InterestRatesand Spreads (%p.m.) (Jul 1994-Feb2003)........................ LeasingPortfolioand Operationsby Type of Asset (1990-2003*) 182 Figure4.8 Brazil: FinanceCompanies- Role in Consumer Credit............................................................................................. 183 Figure5.1 Trends and Compositionof DirectedCredit (2000-02).............................................................................................. 188 192 Figure5.2 Brazil: Nominaland Real InterestRatesin RuralCredit (%) (1995-2001) ................................................................. 194 Figure5.3 Credit Flowsto Agriculture-Reductionswith Fluctuation(1969-2002)........................................................................ Figure5.4 Brazil: Flowof Funds under the NationalRuralCredit System .................................................................................. 195 196 Figure5.5 Sourcesof Fundsfor AgriculturalCredit-IncreasedObligatoryLending................................................................... 197 Figure5.7 Agricultural ProductionSoars Even as FormalAgriculturalCredit Declines .............................................................. Figure5.6 Brazil- RuralCreditfrom PRONAFper monthand fiscalyear (R$ millions)............................................................. 200 207 Figure5.8 The Volumeof Credit is Correlatedwith the Value of LandPrices............................................................................ 208 Figure5.9 MostAgriculturalCredit is Capturedby a Minorityof Farmers ................................................................................... Figure5.10 Spread Between the SELIC Rateand the ControlledAgriculturalInterestRate has Narrowed............................... 209 211 Figure5.11 IntegratedCreditand InsuranceProducts:Suggestionsfor EfficiencyGains........................................................... Figure6.1 Brazil:Spreadsin Bank Intermediation...................................................................................................................... 218 Figure6.2 RegistrationCostsof Mortgages inthe Stateof Si0 Paulo ....................................................................................... 230 Figure6.3 The Useof Credit Registriesby Different Entities...................................................................................................... 247 243 Figure6.4 InternationalExamplesof Credit RegistryInformation............................................................................................... Figure7.1 Brazil:The Ratios of Creditto GDP and Debtto GDP Over Time ( January 1995-March2003).............................. 252 262 Figure7.2 Brazil:Shares of Securitiesin Bank Assets and MoneySupply................................................................................. 263 Figure7.3 Brazil:Trends in Spreads and ReserveRequirements( January 1995-March2003) ............................................... 264 Brazil: Access to Financial Services xiv Executive Summary BRAZIL: ACCESS TO ~ i ~ A SERVICESA ~ ~ C ~ PRINCIPAL CONCLUSIONSAND EXECUTIVE SUMMARY 1. Financial markets constitute a significant part of a broad group of factor markets, including also land and labor markets, which are the basic institutions underlying the effective functioning of the economy, and the production and sale of its goods. Financial exclusion reduces the potential welfare of individuals and the productivity of enterprises inan economy. Effective participationinfinancial markets and other factor markets, which are different from normal (product) markets is a precondition for effective participation in the economy. Access of disadvantaged groups to financial markets is thus of strategic importance for social and economic development and social inclusion. Market failures in these markets have particularly detrimental effects on economic productivity and social benefit, thus these markets are typically closely regulated. But regulation in turn generates risk of regulatory failure, and in fact many regulations may hinder access to the poor. These are the reasons for analyzing financial access, and reviewing the role of public policies to promote access. 2. This study therefore seeks to evaluate present levels of access to financial services and government policies adopted which impact upon access. Based on these findings, it explores options for increased future access to financial services in Brazil. The first section of this summary highlights the core conclusions to emerge from the study which would have implications for government policy. The second section provides a guide to the thematic scope and content of the study, and the third describes the findings, conclusions, and recommendations of each section. 3. The overarching message to emerge from this study i s that increased financial access would be promoted by sound overall macroeconomic and financial sector policy. Beyond that, the Government could and should undertake regulatory reforms to enable financial markets to function more smoothly, and undertake targeted policies to improve access. However, care should be taken to ensure that such targeted policies let the excluded groups participate efficiently in financial markets. This would direct the focus towards a review of incentives rather than public financing of special programs. 4. Sound macroeconomic policy refers first to the reduction of the government's borrowing requirement, which will enable more borrowing by individuals and enterprises in the economy and second, to the reduction and harmonization of taxes on the financial sector, which will reduce the burden on overall intermediation and also reduce current opportunities for regulatory arbitrage across different segments of the financial system. Beyond that, efforts should be made to ensure that competition i s maintainedinfinancial markets. 5. Regulatory reforms which would stimulate outreach and impact on disadvantaged groups would require first a move away from expensive special public financing which often fails to meet target groups, towards improved regulatory regimes which stimulate access. Indeed, present subsidies on some costly special programs (e.g., those for the rural sector) could have a phased reduction. Subsidies, mandatory lending targets, and measures which seek to curb market prices could perversely lead to reduced access over time. Examples of regulatory reforms would include a review of Brazil's high regulatory and reporting requirements for microfinance, compared to what may be merited by its present non-deposit taking nature. Reliance on relatively low cost credit could be gradually weaned, though alternative sources of funds may be needed and deposit taking could be considered once the industry adopts mature microcredit practices. In microfinance, a series of new regulatory initiatives have been adopted, many positive. However many other recent programs which rely on subsidies, lending limits and xv Brazil: Access to financial Services mandated lending, could limit the development of some credit markets. Care should be taken to not overburden financial institutions with the task of intermediation of special programs of government assistance, which could detract from the establishment of orthodox credit practices. In the banking sector, several new regulations have been recently adopted which are moves in the right direction, including the simplification of processes for opening accounts, expanding the scope for correspondents to financial institutions, and establishing basic accounts. Bank branch opening could also be simplified. However, microlending by microfinance institutions such as OSCIPs and SCMs could find their activities restricted by controls on onlending interest rates and new policies in this area should be monitored in terms of cost and impact. Innon-banks, there is a fragmentation of regulation insome areas, such as factoring (which i s outside the defined scope of financial sector activities), which may otherwise help to contribute to the sector's development. Such policies could extend beyond financial institutions themselves to the overall infrastructure for financial intermediation. A first area here relates to creditor rights, and regulatory reform can focus on attention to procedural delays inreachingjudgments in credit disputes and examining the rationale of tax write-offs on uncollected small claims. To enhance credit reporting, regulations limiting information to five years, restricting the sharing of information and limiting the use of positive information can be reevaluated. 6. Targeting and programs of special outreach imply not so much the public financing of programs as their appropriate design, to ensure their outreach to and impact upon desired groups of the population. For example, geographical targeting should go beyond efforts to place a service outlet ineach broadly defined geographic area (e.g., municipality), and focus rather on the identification of deprived neighborhoods, which could include inner city areas within broader urban municipalities, through systematic tracking of deprivation. Appropriate designs would include community outreach strategies to target customers with limited access, image differentiation, adopting new lending methodologies and technologies, strengthening credit reporting, easing the use of secured credit and the offering of incentive compatible services. Basic accounts and `lifeline' services can be reinforced with government support to employees or banks to open such accounts, the use of such accounts to deposit government transfers, as well as careful monitoring of the costs of basic service packages. 7. Incentive compatible support can take forms such as startup support, initial tax breaks or matching grants, partial risk or credit guarantees support for the development of community finance institutions, and enhancing regulatory measures to increase disclosure of practices towards access. Inthe context of microcredit institutions there are now a series of recognized sound practices for microcredit which could boost its sustainability, and microfinance entities which can demonstrate the incorporation of such good practice should be the recipients of funding support for institutional development or expansion. Encouragement can also be offered for the establishment of formal or informal partnerships and institutional support programs, between banks, non-banks or even non-financial entities (such as factoring companies) and microfinance. Finally, judicial education as well as programs of financial education and literacy are also key. 1. PRINCIPALFINDINGSAND CONCLUSIONS -TWENTYKEYPOINTS 8. Concern about financial exclusion increased over the last decade in Brazil, with the decline in the number of banks since the latter 1990s. It is pointed out that only around 60 million of Brazil's population of 176 million have bank accounts, or around a third of its population. Meanwhile not only is the cost of credit highbut bank spreads are arguably among the highest in the world. The study points to a series of factors which affect volumes and costs of financial intermediation. It emphasizes that despite the absence of simple remedies, there are a series of areas in which actions can be taken which together would help to expand access and lower its costs. This summary draws attention to twenty key findings and areas for future action. xvi Executive Summaw 9. First:Despite concerns regarding diminishingnumbers of banks over the last decade, this study finds that there is no evidence of a trend decline in access to bank services, in terms of numbers of banking service outlets, although such services, measured in traditional terms, may have stagnated. Yet, comparing Brazil with other countries at similar levels of development, Brazil is not underbanked in terms of bank branch presence. 10. Second: It is true that there are wide regional disparities in bank service provision, but such regional disparities can be significantly ascribed to differences in population density and income. Indeed, disparities in financial access can be at least as significant between neighborhoods within a city, as between regions of the country. Locational emphasis which i s based on the importance of providing a basic service outlet in each municipality may need to be refocused and supplemented by a locational policy which looks more at deprived areas, which may well be identified at a smaller and more disaggregated level than municipalities. Moreover, even if properly identified, the provision of a service outlet in itself is not sufficient to ensure access by lower income groups, without more explicit design features to reach the excluded. 11. Third: Initial measures designed to expand access adopted over the last few years, especially for the microfinance and cooperative sectors and later for banking correspondents were successful and pointed towards new modes of access to financial services. More recently, a number of new initiatives were taken which have relied more on mandated lending and controlled interest rates for target groups. The mainstay of policies to expand access remain traditional, focusing on the allocation of credit, frequently at low interest rates, with considerable reliance on large public banks to support this mandate, and some recent measuresare also inthis tradition. 12. Fourth: Traditional policies to expand access which emphasize the quantitative rationing of credit, at low interest rates, based on low cost sources of funding, and administered substantially through Brazil's public banks, have a high cost. Conservative estimates suggest that concessions and support could amount to several billion reais. The cost of some special programs, such as the PRONAF program alone, are estimated at R$1.1 billion. As illustrated by the analysis of rural finance programs, many such programs fail to reach intended beneficiaries, and tend to be captured by a small number of the better off, to the detriment of broad-based access. Inagriculture, which i s the beneficiary of many special programs, the largest 2 percent of borrowers receive 57 percent of loans, while the smallest 75 percent of borrowers receive only 6 percent of credit. Programs which propose ceilings on lending rates for target sectors or clients are costly, as measured by the volume of credit provided and by the interest rate differentials which are eventually borne by society. Additional costs are the crowding out of intermediation at market rates. It is recommended first that such programs, including some recently introduced special programs for microcredit, be closely monitored in terms of cost. The regressive incidence of some directed credit interventions may be compounded if they are funded through broad based and popular programs such as the FAT funds, which imply lower returns to the workers, who are the intended beneficiaries of the FAT fund. 13. Fifth:One form of alternative measuresto the traditional programs include new instruments to offer possibilities for market-based expansion of services. For example, in agriculture, new forms of index-based yield insurance, or loans combined with put options on crop prices, and the combination of different instruments (e.g., credit and insurance) should be explored. The costs of individual programs, and implied subsidies, need to be systematically tracked. Additionally, there is scope for gradually phasing in reductions in quantitative allocations (e.g., earmarked funds for housing and rural lending) and for allowing rates to more closely reflect market rates. Successful programs for microcredit, such as the CrediAmigo program, which has achieved outreach to the poorer segments, demonstrate the possibility of onlendingat market rates. New forms of technology-based microfinance can also be considered. xvii Brazil: Access to Financial Services 14. Sixth: Measures to expand the role of the banking system as a whole in the area of access are valuable, but care should be taken that they are established on a sound footing. Brazil's banks provide the bulk of its financial services and remain the mainstay of all intermediation. As the study shows, Brazil's private banks are already servicing small clients and scale of transaction may be less of an issuethan costs of basic services. Surprisingly, private banks dominate the provision of current and deposit accounts to small account holders. Survey results suggest that public banks do not demonstrate a clearly dominant role in terms of access, except with regard to some types of services (e.g., payments services). Recently simplified procedures for opening accounts impacting upon low income clients are welcome; especially simplifications in the proof of identity and income. The possibility of alternatives to taxpayer numbers could also be considered. Such measures could gradually be extended to all bank clients. Many obligatory costs of maintaining bank branches could be examined (e.g., security aspects, bank hours and staffing obligations, etc.). 15. Seventh: Recently introduced measures for the provision of `lifeline' banking with basic services, for lower income persons, is also a benefit and has been successfully introduced in many countries. However, the costs involved in providing such services should be carefully examined, to ensure that participating institutions' operating costs are covered, and monthly fees for such basic accounts be priced at affordable, non-zero levels. However in most countries such accounts do not usually include automatic access to loans, or low interest loan features. The government can encourage the establishment of such accounts, for example by channeling public payments through such accounts, by offering incentives to employers or their financial institutions for the establishment of such accounts for payroll purposes, and by initial tax breaks to financial institutions for setting up such accounts. 16. Eighth: The government could adopt disclosure based requirements to promote financial servicesfor the underserved.Tools such as the Community Reinvestment Act and the Home Mortgage Disclosure Act, when appropriately combined with disclosure requirements, have been very valuable in limiting discriminatory practices as well as inexpanding access. Especially if combined with systematic tracking of banks records in different categories of lending, possibly in partnership with the private bankers' association, such moral suasion has been found to be very effective in other countries. The government could also consider measures such as partial guarantees or credit insurance for loans provided by banks to excluded groups. In contrast to interest rate subsidies, such measures are incentive compatible with the increase of private credit to such groups, and would still encourage the prudent use of loanablefunds. Finally, the government couId also consider a review of banking sector competition,particularly in areas such as access topayments networks, based on recent experience with such investigations in countries such as the UK. 17. Ninth:Private banks could go further towards `downscaling' or expanding their services to the underserved, through efforts which make use of new lending methodologies and technologies, with appropriate incentives and support from the government. Such lending methodologies include for example practices such as image differentiation, partnerships with local or community based organizations, including in some cases, partnerships with microfinance organizations. The specialized subsidiaries set up by banks such as Unibanco for this purpose, are an example, and these features are to be included in the new Banco Popular. Efforts should be made to ensure that such organizations, while user friendly, follow sound banking principles. The experience of other countries with Community Development Finance Institutions can also be considered in this regard. New technologies such as risk- assessment or credit scoring models may also be useful in evaluating small clients, as would many new technologies which permit easier operation in remote areas or field locations, used successfully elsewhere, as detailed in the study. Although technologically sophisticated, Brazil's banks efforts in terms of technology, such as internet banking, are largely geared towards the better-off. Startup assistance for adopting such technologies could be considered. xviii Executive Summary 18. Tenth: Brazil's encouraging progress with the expansion of its erstwhile limited microfinance sector in recent years, albeit supported by funding from the BNDES bank and from the BNB, has shown in the CrediAmigo program that both outreach and efficiency can be achieved, and that microlending can approach sustainability on a cash flow basis. Onlending rates adopted by most of the private M F I s until recently were freely determined and approached market rates. Considerable international experience has demonstrated that interest rate caps are unnecessary and may lead to inefficient use of such funds and obstruct spontaneous expansion of microfinance. Rather, the incorporation of good lending practices and methodologies by such MFIs, for example, exploring solidarity group lending, gradual loan size increase, repayment beginning with immediate effect, etc., is suggested. 19. Eleventh: Looking at the funding side, many of Brazil's new microfinance institutions have enjoyed privileged access to BNDES funding at the TJLP rate. A phased and gradual increase to market levels would be desirable, at least to the DIrate, as inCrediAmigo. However a sudden significant increase could adversely affect the microfinance sector, given that the majority of M F I s are still deemed to be non-profit or civil society operations, and could reverse the gains of recent years. The impact of recent escalations in funding costs by loan threshold needs to be kept under review. Convergence of size thresholds and interest rates combined with gradual transition and guidance on the incorporation of good lending practice, would be desirable. 20. Twelfth: Even in the absence of recent measures, expansion of microfinance on a significant scale would be difficult, unless large scale partnerships with formal financial institutions are considered or alternatively, with a radical change in services offered, to include deposit taking. This could be contemplated as a future possibility, once there is evidence of maturity in the industry, to expand the funding base, but new regulations would need to be introduced for attendant increases in risk, which would include particularly new models of shared, delegated or independent supervision. Additionally, the rationale for present multiple regulatory windows for different institutional forms of microfinance should be examined. Harmonizing such regulations in the future would be desirable. 21. Thirteenth: Looking at the cooperative sector, it is observed that cooperatives have a less restricted funding base than microfinance, despite erstwhile membership restrictions, due to their ability to accept deposits. The recent easing of membership restrictions on new cooperatives to permit `open' structures, although admittedly for more sparsely populated and financially served areas, requires additional supervision and oversight. While recent regulations include some appropriate safeguards, it should be noted that membership restrictions are usual even in advanced countries and provide a form of `reputational collateral' to contain the risks of such institutions, especially in view of constraints in supervisory capacity. 22. Fourteenth: Untilrecently, leverage and liquidity restrictions may have been more important for cooperatives than membership restrictions. Recent measures introduced in June 2003 have partially alleviated these. The establishment of a liquidity facility for credit cooperatives is suggested, to reduce opportunity costs relative to commercial banks backed by the FGC deposit insurance fund. While there may be indirect access to such liquidity through participation in cooperative banks, more direct support would be desirable. It i s noted that some cooperatives are also assuming the role of transferring funds for government sponsored agricultural programs. If this becomes the practice on a significant scale, it could steer away the cooperative sector from good risk management practices, unless this is clearly a fee-based administrative operation with no credit risk for the cooperatives concerned. 23. Fifteenth: The contribution of non-banks to access so far has been based to some degree upon regulatory arbitrage opportunities with the banking system, and they have suffered from varying degrees of lack of legal clarity. The most promising area for attention in this context i s the contribution of non- financial companies, such as factoring companies, which are already sizeable in terms of the provision of xix Brazil: Access to Financial Services credit to small enterprises, and have made promising contributions in other countries. A first step to take in this regard would be the clarification of the fragmented legal framework for this activity, and passage of a pending law for its consolidation. Wider access to finance through commercial bills and debentures, as well as easing international factoring through permission to set up foreign currency accounts is also suggested. 24. Sixteenth: Credit to small borrowers is impeded by difficulties in loan recovery, due partly to `judicial activism' and also to difficulties in the use of collateral. Recommendations in this context are awareness programs for judges, combined with the reinforcement of self-executable instruments used for small borrowers, for example, the overdraft facilities offered over the ATM or the duplicatas of small firms. Legal recognition of these debts (which at present is not possible due to their lack of signature and computerized entry) could be explored. An examination of the possibilities of changing the procedural codes to speed up judgments is also advised, as well as the opening of special (small claims) courts for small credits. Finally, current tax write offs against profits on uncollected small claims can be evaluated in a wider context as these may be sending perverse signals to both borrowers and financial institutions for debt repayment. 25. Seventeenth: Additionally, the infrastructure for perfecting security interests use can be strengthened by the adoption of electronicfiling, retrieval and indexation systems at all registration offices in the country and the establishment of networksfor the interlinkage of registries. The present five day time and costs taken to conduct a search and certification could both be reduced by a selfguided web search. Deadlinesfor turnover timefor registries of deeds and documents could be adopted, and documentation requirements for secured transactions in real estate could be simplified. Additionally, efforts should be made to address the regressive tariff structures of registries and notaries public, and the extent to which notarial and registration services are still operating on a franchise basis could be examined through the offices ofjudicial oversight. 26. Eighteenth: Further improvements in credit registries and credit reporting will help to ease lendingto small borrowers with limitedcredit history. Brazil already has effective core systems, but these could be strengthened by more positive information (e.g., on utilities payments), and some easing on restrictions on the sharing of such information. There could also be a reconsideration of current time restrictions on information which limit credit data to five years. Brazil is an outlier inthis respect. Utility tariffs could also be structured to provide incentives for good payment records. 27. Nineteenth: It is of key importance to point out finally that the ultimate success of policies to expand access are predicated upon an appropriate macroeconomic environment, which has a profound influence on overall access. All the microeconomic interventions discussed above must be viewed against this backdrop. The government's large borrowing needs impact negatively upon private credit and the high taxation of the financial system, partly to finance such needs, i s another deterrent. High reserve requirements add to implicit taxation. Moreover, the impact of taxation i s not uniform across the financial system which leads to opportunities for regulatory arbitrage across different financial entities. Such regulatory arbitrage also occurs with regard to different forms of financial institutions, where some institutional forms grow popular in response to regulatory constraints in other areas of the financial system. 28. Twentieth: There nevertheless i s a role for more proactive government policies towards access at a micro level through the creation of a more supportive enabling environment in terms of regulation, monitoring and disclosure and selective support. Yet their eventual success will be the greatest if they are backed by fundamental changes in overall approach to access through broad-based financial sector reforms which limit constraints the price and quantity of credit and reduce also the overarching role accorded to a small number of public institutions intheir delivery. xx ExecutiveSummary 2. OVERALL INTRODUCTION AND SCOPEOFTHE STUDY 29. Brazil's financial system is by far the largest in Latin America. Beyond its sheer size, the overall depth of financial intermediation in Brazil, at almost 140 percent of GDP, i s greater than its large neighbors in the region such as Mexico or Argentina, despite their higher average per capita income. Financial depth and stability are increasingly recognized as contributing to poverty reduction through growth and crisis prevention, but beyond this, it can be argued that a more broad based distribution of financial services would raise welfare and productivity. Individuals can meet unusual or unexpected demands for income, or safeguard against periods of low income or unexpected fluctuations in income. Access to savings and borrowings could also have longer term welfare implications, permitting people to borrow when young, for building human capital, and then save for retirement when they are older. For producers, access to credit for fixed or working capital enables an increase in production possibilities benefiting not only the producer but affecting economy-wide productivity, employment and growth. Financing has featured prominently as a constraint for small and medium-size enterprises in some investigations. Some studies indicate that Latin American firms find access to financial markets to be the major obstacle to expansion. 30. Limits to financial access are pointed out by the Central Bank around 60 million or a third of Brazilians are estimated to have bank accounts. Nearly 30 percent of all municipalities, or 1,680 municipalities, out of a total of some 5,600 municipalities, have no bank branch. The 1,400 or so credit cooperatives offer basic financial services only to about 1.5 million Brazilians, mostly in urban areas. But expanding the supply of financial services to underserved segments of society can pose particular difficulties for financial intermediaries such as banks, due to limited information, highrisks and high unit costs. The challenge therefore is to identify ways of expanding access which acknowledge the risks and costs involved but suggest measures to mitigate their impact. This is the fundamental objective of the present study, which attempts to assess constraints to access and their determinants, in different segments of financial markets, and then identifies alternative public and private choices for improved access. 31. There are multiple concepts of the undersewed - first, the poorer segments of society, or second, those in specific geographic regions, which may or may not combine characteristics such as remoteness or sparseness of population with economic backwardness. Third, specific communities (racial groups, migrants or minorities) may be disadvantaged in terms of financial access. Additionally, the criterion of small size is often applied, particularly to micro or small scale entrepreneurs in contrast to individuals. Measures of access and actions to expand access vary depending on which groups of the underserved are being referred to. In the present study, poverty is the primary consideration though small size (for producers) and location are alsofactors which are considered. The study pays greater attention to factors which directly affect access. Thus indirect impacts on access, operating via their effect on the overall depth of financial services (for example, the potential `crowding out' of private credit provision by government borrowing, or the impact of financial sector taxation) are not investigatedindetail. 32. There are also a series of possible indicators of financial access, and the report discusses the tradeoff between the relevance and certainty of conclusions which may be drawn regarding levels of access, and the ease of obtaining information. Typically, more disaggregated information i s harder to get, and lends itself less easily to international comparison, than more aggregated data, which however is less conclusive. The study points to simple disaggregations, such as the `unbundling' of financial services - for example, money transmission, savings, and credit services, and demonstrates their application. 33. Brazil's government and Central Bank have been concerned about financial exclusion, particularly in the post Real plan period, when mergers or closures of a number of banks, both private and public, occurred, leading to concerns about contractions in bank services, especially in geographically remote regions. To what extent has there really been a contraction in bank services, and i s there, as a xxi Brazil: Access to Financial Services consequence, an overall decline in access? To what extent are persons' needs for financial services met? And what explains present levels of access?These are some of the questions addressed inPart 1of this study, which sets the stage for the rest of the analysis, by undertaking a broad overview of current patterns of access. It looks first at patterns of supply of financial services, through aggregate data on financial institutions and their branches, and then at the demand for financial services largely from the perspective of urban consumers, based on the results of a survey undertaken for this report, and briefly, also from an enterprise perspective. Inparticular, policies adopted towards the expansion of access by the government are evaluated. 34. The government has also been increasingly aware of the relatively underdeveloped provision of financial services by smaller players such as cooperatives and microjinance, which have been very successful elsewhere in Latin America, and where current penetration in Brazil i s low compared to its neighbors. Moreover, microfinance has increasingly been hailed in global communities as the new solution to financial exclusion. As a result, several steps have been recently taken towards expanding potential outreach for such institutions, and more such measures are planned. How successful has this been and how muchfarther can it go?What i s the best mix of policies to adopt for the future expansion in the supply of financial services from these sectors? These issues are addressed in Part 2 of this study, which launches an analysis of the supply of financial services through different institutional segments of the financial market. 35. Particularly in view of the dominant role of banks in Brazil's financial system, Part 3 of the study examines the role of private banks with regard to financial access. Especially in view of the reduction in bank numbers and some increase in concentration over the latter 1990s, questions may justifiably be raised regarding the impact of these trends on banking outreach. Beyond its analysis of the private banking system's present role in terms of outreach to the underserved, this section also draws extensively upon international experience with regard to the provision of private bank services for underserved communities. The study also investigates the potential for expansion of the role of non-bank financial systems such as finance and leasing companies and factoring, which have contributed to access successfully elsewhere (Part 4). 36. Traditionally, Brazil's government has attempted to address issues of financial access through its publicly owned financial institutions, which have been accorded a social obligation to attend to the disadvantaged segments of society. The large federal banks, as well as a number of state owned banks, have been responsible for outreach, not only through the extension of their own resources but also through the administration of earmarked budgetary funds and payroll deductions (such as the FAT). To the extent that some of the sectors served by such public financial institutions are known to face a number of market imperfections which may make a case for public intervention, Part 5 explores one such market segment; rural finance. Even if some intervention is warranted, there are issues of appropriate design. The extent to which these programs may have achieved their own objectives of large scale outreach towards the poor is a first issue for debate. It investigates the efficiency implications of present programs of outreach, as well as other new alternatives. 37. Following the investigation of service providers, Part 6 of the study investigates relevant elements of the institutional infrastructure for financial services. Such factors impact particularly on the cost of provision of financial services, and hence, on bank spreads. Some of the issues examined here are the implications of current legal processes for creditor rights and contract enforcement, registries for securedcredit and their potential impact on the use of collateral, as well as credit information systems and the applicationof credit scoring processes. 38. The study concludes with a discussion of the potential role of policy makers inthe expansion of access through different instruments and institutions. The study points out that macroeconomic factors xxii Execufive Summary such as large public sector deficits or high taxation significantly affect financial services. Policies impacting upon other market segments too can importantly affect access, for example, labor market policies, as well as policies towards competition in the banking sector. However, while it may be difficult for a government to act upon these in the context of the financial system there i s a more immediate role for the government to play, through policies which focus on the structure of financial system incentives, impacting upon the distribution of credit and other services. Based largely on experience in third countries, Part 7 discusses options for a more proactive microeconomic role for the government with regard to financial access. 3. SPECIFIC FINDINGS AND SUGGESTEDACTIONS ININDIVIDUAL AREAS 3.1 Assessing Access 39. Core analysis in this part of the study rests first upon an investigation of aggregate supply side information on banks and other financial institutions, based largely on Central Bank data, followed by a demand side investigation of individuals' use of financial services, based on the World Bank's survey data. 40. The analysis of aggregate supply-side data suggests, first, that despite contractions in bank numbers, delivery of financial services has not declined over the last decade but has remained largely static. There have been some expansions of new forms of service delivery points, such as ATMs and banking correspondents, and also, an increased presence of some small financial institutions, including cooperatives and microfinance institutions, although these are also numerically small. Moreover, by international comparison, Brazil may be underbanked compared to developed countries in terms of bank branches but it has performedat least as well or better than its peers among developing or middle income countries. 41. It i s true that spatial inequalities in the delivery of financial services are significant among Brazil's regions and have been persistent over time. But much of the spatial difference in the provision of banking services can be explained by regional differentials in population and income. Once we correct these factors, the differences are significantly eroded. 42. Next, the supply analysis looks at the differential contributions of public and private banks. Underlying the relatively static overall picture in terms of financial service provision, there has been a significant shifi in composition of service providers with a declining share for public banks. Private banks, while relying extensively on new delivery mechanisms such as ATMs, are also surprisingly more branch intensive than federal banks. 43. Econometric analyses of the determinants of bank service provision, usinga range of indicators of access, find first that for all banks taken together, per capita income tends to be positively associated with broad measures of access such as the numbers or density of banking institutions or the values of the loan and deposit services they provide. Banks (public and private) have an urban bias and tend not to provide services inrural areas. Nevertheless the agricultural community appears not to be underserved. 44. Finally, analyses which distinguish between public and private banks find some differences in the factors associatedwith the provision of services of these two groups: by some measures public banks may provide more services to the poor, but private banks appear to offer more services to agriculture and to small enterprises. In many respects the behavior of the two i s broadly similar; both tend to have a positive association with richer economic areas and both have some urban bias. Especially in thinly served areas, they tend to be substitutable, with either private or public services dominating. xxiii Brazil:Access to Financial Services 45. The second part of the overview of patterns and determinant of access is based on an investigation of the demand side - the extent to which individuals use financial services and the extent to which such services meet their needs. This analysis is based on a survey of 2,000 adult individuals, located in 9 cities and 2 metropolitan areas, undertaken for the purpose of this study. Urban areas were selecteddue to the fact that around 80 percent of the population is classified as urban, as well as the ready availability of a sample frame. A series of basic facts regarding the current availability and use of financial services emerged from this survey, and these were then correlated with individuals' socio economic characteristics. In terms of current levels of access, illustrative findings are that forty three percent of individuals surveyed have a bank account. The use of non-bank financial institutions such as cooperativeskredit unions or microfinance institutions was very limited (only 4 percent of responses). However, a considerably larger number of persons (who are not necessarily account holders) made use of financial institutions, through payment outlets of banking correspondents, especially through the lottery shop chain of the Caixa Econ8mia Federal; one of the two large public sector banks. An inquiry about the relative importance of different financial institutions revealed that almost half of the respondents to this question (47 percent) found the lottery shop centers to be their most important institution for financial transactions. Private bank branches are second (27 percent) and public bank branches are third (19 percent). Surprisingly, aside from the lote`ricas, the majority of respondents (58 percent) for whom banks were the primary financial institution use private banks. Most accounts are limited; only 61 percent of account holders have access to checking facilities. 46. Only around two thirds (64 percent) of those who did not have an account indicated an interest in having one. Among the voluntary reasons for not having an account, high fees clearly dominate. Individuals' lack of funds, lack of documentation and references and difficulties in opening an account also mattered. By contrast, factors such as hours of operation or distance, which suggest physical inconvenience, were not important. With regard to deposit and savings behavior, more than half the respondents (54 percent) have their largest deposits in private institutions and 41 percent have their largest deposits in public institutions. Other financial institutions, including cooperatives, accounted for less than 2 percent of responses. Returns on deposit accounts have low price elasticity; for over two thirds of respondentsthe most important reasoncited for making deposits was `security'. Not surprisingly loan and credit services are much more sparingly used. Only 15 percent of respondents claimed to have applied for credit in the past 12 months, and only two-thirds of those respondents had their applications accepted. Apart from formal loans, around a quarter of persons had access to a debit card and a fifth had a credit card. Regardingpayments services, cash was overwhelmingly the standard medium of payment for all types of transactions and respondents; 77 percent of respondents used cash for all transactions. There is surprisingly little variation by type of payment. Surprisingly, 92 percent of utility bills are paid in cash. The only expenditure categories where cash use is lower is consumer durables; 77 percent for electro domestic goods and 80 percent for furniture. 47. Further analysis were undertaken of associations between observed patterns of access, variously measured, and other parameters. Three broad groups of associations are examined. Given the efforts of providing access at all locations in the government's policies towards access, associations between differentials in access and location are first evaluated. Location i s considered at two levels; regions of the country as well as location in terms of neighborhoods or areas of a city. Second, since public financial institutions have been given a significant role to play in meeting the needs of financial access of a large cross section of the population, their role in this regard is examined. Third, the importance of individual characteristics in terms of income, wealth, education, employment, etc., in explaining levels of access, is investigated. 48. With regard to the importance of location, the survey of individual financial behavior supports the supply side finding of the importance of location in determining access, but demonstrates that location in terms of the micro characteristics of an area or neighborhood can be as important a xxiv Executive Summary discriminator for access as regions of the country. While a first look at the supply of financial institutions points to wide regional differentials in the provision of services, these are considerably reduced when we correct for differentials inregional GDP and for geographical size. 49. Regarding the roles of public and private financial institutions, the survey of users, as in the case of the aggregate supply analysis, indicates that there i s some association between lower income groups and public banks. However the user survey also showed that use of public banks varies by type of service. Thus, public banks are popular for payment services (due to the outlets of the Caixa Econamica Federal), for all socio economic groups, and also dominate the provision of housingcredit, but largely for the better off groups. For deposit taking and for credit, both the privileged and less privileged socioeconomic groups, broadly measured, have some preference for private banks, and with increases in income both increase their proportional use of private banks. 50. Finally, the survey strongly corroborates thefindings of the supply side analysis regarding the importance of income, at a per capita level, and points also to the significance of a range of socio economic characteristics, such as education, and (for credit services), wealth, in determining financial access. The importance of socio economic characteristics in access to financial services suggests that there may be problems of information, or of enforcement, which cause lenders to emphasize such information. 51. What are the implications of these findings, from the perspective of broad policy directions to adopt in addressing access related issues? The analysis suggests a number of thematic directions which could help to guide policy choices. 52. First, there could be implications for policies regarding location. The analysis suggests that although location i s important and there do seemto be some regions and locations which are underserved, much of the difference with regard to location can be attributed to levels of economic activity and sparseness of population. Thus, policies which are focused on the geographical equalization of services may not, on their own, be adequately effective in targeting access. This i s a significant comment in the context of Brazil where major efforts to expand access have focused on ensuring that there is at least one financial point of service (branch, service post or correspondent), in each municipality. Such policies cannot ensure that users of financial services in such locations are indeed the poorer segments, indeed the better off persons even in these neighborhoods may be more successful in achieving financial access. 53. Another important inference with regard to locational policies i s that regions of the country or even municipalities, are perhaps not the best points of focus for the expansion of access. Rather, location defined at a micro level in terms of neighborhoods, with service expansion targeted at areas orparts of a city with specific microeconomic characteristics (e.g. high concentrations of low income housing), should be targeted. 54. Eventually, factors related to location, though initially significant in the analysis, lose significance once socio-economic factors such as income and education are included in the analysis. Additionally, other socio-economic characteristics, such as employment related variables, gender and role in household were also show to be significant in the determination of many measures of access. One implication of this i s that these factors have assumed importance in the absence of sufficient direct information on the financial behavior and creditworthiness of individuals. Wealth and collateral were also important in determining access to credit, as these may help with loan recovery. This finding could also suggest that in the absence of adequate enforcement mechanisms for financial claims, the factors noted above could serve as indicators of financial reliability. The implications are therefore that policies which can directly expand information on clients would expand access, such as the sharing of `positive' xxv Brazil: Access to Financial Services information in some forms of credit registries. Equally, procedures which streamline the use of guarantees,secured credit, bankruptcy or recovery could also help to expand access. 55. The importance of income-related factors in determining access emerges as a central and consistent theme. A basic deduction is that access will be increased by overall growth oriented policies; thus improved financial distribution in addition to financial deepening is associatedwith growth. Perhaps a more significant corollary for policy makers i s that if access is so significantly influenced by income, then targetedpolicies, which particularly address lower income groups, are likely to be important in addressing access. What does this imply inpractice? That policies such as for example, "lifeline" or basic accounts, affordability of minimum packages of financial services, possibly special criteria related to documentation requirements for low income persons, or special financial products designed for low income persons, could be importantfor raising access. 56. Next in importance to income in terms of individual characteristics associated with access is education. One suggestion to follow up on this finding is that programs of financial education and awareness may be importantfor low income persons,even if overall educational policies, just as overall growth raising policies, may lie outside the scope of the financial sector policymaker. 57. Finally, when we analyzed the type of financial institution, given the broad range of substitutability, apart from the delivery of some specialized services where either public or private banks seem to dominate (e.g., public banks for payments or housing), a suggestion is that greater competition in remaining specialized services would be desirable, for example more private services in housing or in payments. A level playing field for public and private institutions would be a prerequisite. 58. Although the analysis did suggest that persons with lower levels of income tended to use public banks proportionately more, it also showed that along a broad spectrum of services, the roles of these institutions could be largely substitutable. Thus for both public and private banks, deposit and credit services are available to lower income individuals, but as persons become better off, they tend to prefer private services for both types of transactions. This also suggests that private banks could do more to refocus their products, image and services in terms of outreach towards the less well ojJ Proactive suggestions for how banks might achieve this, and what the government could do to support such efforts, are discussed inParts 3 and 7 of this study. 3.2 ExpandingMicrofinance 59. As the preceding section showed, Brazil's financial system so far has been largely bank dominated, and the development of other forms of financial institutions, especially microfinance institutions or credit cooperatives, has remained emasculated. Yet during the last ten to fifteen years, excitement has grown worldwide regarding the potential of such institutions, as vehicles for lending to the poor. Today, microfinance institutions (MFIs) are reaching out to over 9 million borrowers and 29 million savers, in over 53 countries worldwide. The importance of microcredit for the self employed and for small entrepreneurs, and thus for jobs, is also increasingly recognized. Microentrepreneurs of Brazil account for the vast majority of all firms and contribute substantially to employment and GDP. According to recent estimates, 98 percent of Brazil's 4.1 million micro and small enterprises account for 45 percent of formal employment and more than 60 percent of urban jobs. Directly or indirectly, they provide the primary source of income to almost 60 million people and generatenearly 20 percent of GDP. 60. A second group of financial institutions catering to these needs is the community of credit cooperatives. Credit cooperatives were first formed a century ago in the southern states of Brazil, and are a mature alternative to banks in many countries. Both deposit and loan services are offered. Like xxvi Executive Summarv microfinance, formal procedures and the use of formal collateral are usually absent; because the group's knowledge of its members provides the coinsurance neededinplace of guarantees. 61. Regarding Brazil's microfinance sector, the study finds that there was a remarkable acceleration in Brazil's microfinance sector from the end of the 1990s, due to a series of factors. First, there was new political support for the sector, leading to a series of legal and regulatory changes favoring the expansion of microfinance, introducing new institutional forms; the OSCIPs (civil society organizations) and the SCMs, (microcredit societies) and expanded funding flexibility. Prior to 1999, microfinance in Brazil could only be extended by NGOs, (non-profit, non-government organizations), constrained in their funding to donations, and limited technically in lending practices to unrealistic ceilings on interest rates under Brazil's Usury law. Second, a major new large scale microcredit experiment was launched; the `CrediAmigo' program, which incorporates many principles of good microfinance practice. And third, active government financial backing was extended to emerging microfinance institutions, through credit offered by the BNDESbank. 62. Despite this rapid acceleration, microfinance penetration in Brazil today remains low, especially compared to neighboring countries. The portfolio growth rate of individual small M F I s (Le. microfinance lending other than the BNB's CrediAmigo operation) has not been remarkably high, much of the growth spurt was due to the establishment of new firms rather than rapid expansion of existing firms. Growth indicators for the portfolios of MFIs supported by BNDES suggest nominal annual growth of 22 percent a year; much lower for larger M F I s . The slow growth of Brazil's independent M F I s despite apparent large unmet demand is paradoxical. Some factors responsible could include the difficulties encountered by independent MFIs in expanding client outreach without a branch network or savings products. It may also be noted that the presence of some so-called microfinance institutions in Brazil (run mainly by municipal governments to achieve social objectives such as employment) provide highly subsidized credit which could affect the expansion of market orientedmicrofinance. 63. In terms of performance, indicators of loans overduefor the BNDES-supported MFIs have been average (possibly with some trend increasefor BNDES supported M F I s ) and loan provisioning may have been somewhat low. But financial sustainability has been good, perhaps due to low cost funding. Efficiency indicators in terms of loans per loan officer though low have been rising. Impact indicators in terms of loan size suggest that outreach is reaching small client segments. 64. MFI expansion has been significantly motivated by public sector support, through the substantial presence of the public banks BNB and BNDES. Microfinance institutions also rely substantially on relatively low cost government lines of credit extended at below comparable market rates. Conversion to the new institutional forms, such as SCMs, although increasing access to wholesalefunding, is rendered less attractive by the implications of new regulatory requirements, for the provision of both non-prudential information as well as information on prudential ratios. There are also tax implications which would subject SCMs to the full range of financial sector taxation. Brazil's MFIs are subject to regulatory and reporting requirements which may be more rigorous than comparator models, in view of their non deposit taking scope of activities. However should Brazil contemplate the introduction of savings products or deposit taking, such regulations would need to be retained and further fine tuned. 65. In terms of suggestions for the future, it may first be noted that if Brazil's microfinance sector i s to proceed along a sound expansion path in the future, its institutions could make more extensive use of recognized best practice methods. Although extensively used at CrediAmigo, small M F I s do not appear to have followed these principles. xxvii Brazil: Access to Financia/Services 66. From June 2003, the government has introduced a series of new measures to support microfinance. While the broad direction of support is welcome, measures based on interest rate ceilings for microlending or the directing of bank resources towards microfinance (with the alternative of placement as unremunerated reserves) have a negative impact on financial markets. Znterest rate ceilings on microlending squeezeprofit margins and discourage private microcredit. Znternational experience on microcredit demonstratesthat demandis affected relatively little by lending rates, other features such as personal contact, flexible terms, and phased early repayment are more important ingredients of sustainable microcredit. Directed bank lending is an implicit tax on the jinancial system. Such measures could have a negative effect on the spontaneous growth of the microfinance sector. The real costs of the new programs, for banks and microfinance institutions as well as the government need to be carefully monitored. 67. Next, both to attract more wholesale bank funding, and to better incorporate good microfinance practice in its MFIs, Brazil could consider organizational forms which allow partnerships with large banks, which could help with training loan officers, tracking quality indicators and incorporating good practice. The new microfinance subsidiary at Banco do Brasil (Banco Popular) i s a first step. The BNDES bank could also consider a more indirectform of support to MFZs than its present model, through the provision of wholesale credit to private commercial banks which then liaise with self-help groups either directly or through NGOs. Another possible way to attract commercial banks to microfinance, inan initial phase, as in the NABARDmodel in India, could be through the use of current direct credit limits for rural and housing lending for appropriate microfinance schemes. Care should be taken to ensure that such partnerships are transparent and cost-effective. 68. Brazil's microfinance sector could also contemplate deposit taking in the future, once sufficiently mature, not only to expand its funding base if large scale roll-out is to be contemplated, but also to offer needed savings products and expand client interface. Brazil does not allow any form of deposit taking for its MFIs, and wholesale fundingfrom the banking system has been limited. 69. Microfinance in Brazil may need a separate regulatory window, but the reasons for the current multiple regulatory windows of ONGs, OSCIPs and SCMs may be driven partially by the need to counter other regulations which need overhaul, such as the Usury law. In the future, a simplification of this framework may be desirable. Harmonizing (and perhaps eventually integrating) the framework for regulated MFZs with the banking system may also be considered. 70. Within current regulatory norms, SCMs have very low entry capital requirements bothrelative to other countries and relative to Brazil's own banking system. This could pose a problem for supervision if numbers of SCMs expand significantly - although this may be unlikely in the new microfinance environment introduced in 2003. Conversely leverage ratios appear relatively high, even compared to other countries, which frequently use the same capital adequacy ratios as for banks. If separate regulatory windows are maintained, models of shared, delegated or independent supervision could be considered. 71. With regard to credit cooperatives, the Brazilian credit cooperative system made important advances, particularly since the mid-l990s, due especially to the permission to allow central credit cooperatives to participate as owners in cooperative banks. These enabled the rapid expansion of credit cooperatives combined with growing professionalism in cooperative management, information, accounting, staff training and incentives, and internally administered prudential standards. Incentives to individual cooperatives to federate have been high as leverage ratios for federated cooperatives were set at much higher levels than individual cooperatives. Inthe past this may have mitigated the extremely low start-up capital requirements, even lower than for microfinance institutions. xxviii Executive Summary 72. Federated or central cooperative members benefit through an expanded range of financial services (such as intemet banking, credit cards, CPRs, insurance and custodial services) in addition to more accessible loan and savings products, and in most cases they have been able to count on more reliable prudential oversight of the resources they have invested in cooperatives. 73. Despite such advances, it i s noted that some cooperatives, especially rural cooperatives, continue to serve to channel directedcredit to the rural sector, often at below market rates, sometimes at unattractive spreads. This undermines members' sense of ownership and reduces incentives to repay. To compensate, there have been special dispensations in terms of prudential requirements. Policies such as those that doubled the leverage permitted in the event of financing usingpublic rural creditprograms such as PRONAF reinforced incentives that undermine the cooperative spirit and the sustainability of the cooperatives. The lifting of this facility with new regulations introduced in June 2003 was a welcome move. In addition, consideration should be given to reviewing the extent to which the relative market positions of Banco do Brasil versus smaller rural credit cooperatives explain their relative shares of the spread on PRONAF loans, and what measures can be taken to better relate spread to credit risk and administrative expenditures on such public transfers. Care should be taken to ensure that the channeling of governmentsupport does not dominate the activitiesof cooperatives. 74. One constraint on the expansion of the cooperative sector has been the high liquidity requirement, considerably higher than prudential norms advocated by international standards.. Although new regulations introduced in 2003 bring the capital requirements of this sector in line with the banking sector, higher liquidity requirements remain. Other prudential regulations recently introduced could improve the soundness of the sector but will raise supervision resource requirements. Non-prudential disclosure based advances on the regulatory side comprise the inclusion of cooperative clients with loans above R$5,000 in the Central Bank's credit information system (Central do Risco), and reviews of cooperatives' Boards of Directors. As in the case of microfinance, over time, the harmonization of financial regulation for the credit cooperative sector with other financial institutions is desirable and recent changes have contributed towards this. Such harmonizationwould promote competition and reduce perceived higher risk associatedwith the lack to access to a liquidity facility and deposit insurance. 75. Another constraint on expansion which has frequently been alluded to is membership restriction, both geographic and by occupation. Relaxation of this constraint has been recently introduced, especially for remote areas. It should be noted that constraintson membership exist in most credit union systems in developed countries, and are deemed to provide a safeguard based on collective responsibility or `reputational collateral'. The recent relaxations must be carefully monitored interms of their impact on the safety and soundness of the cooperative system. In this context, the higher minimum capital requirements for the new open admission cooperatives are appropriate. 76. Regarding oversight, there i s a need to maintain a prudent balance between direct and delegated oversight of credit cooperatives. The present system of delegation appears to be beneficial though if the sector grows further formalized delegation may be needed and a choice of appropriate framework for this will need to be selected. The setting up of a special supervision unit for these institutions, following the regulations of 2003, i s a welcome move. Additionally, the collection and maintenance of further information on a regular basis should be strengthened. This should include information on the performance of microfinance institutions and credit cooperatives and determinants of such performance (for example, data on administrative cost structures in different geographic contexts, riskprofiles of clients by location and type of enterprises, etc.). 77. Finally, for both the microfinance and credit cooperative sectors, it i s suggested that a number of technological innovations are available which can greatly enhance productivity and delivery capacity and reduce risk in microfinance transactions. These include particularly the use of devices such as PDAs, xxix Brazil: Access to Financial Sewices handheld computers and smart cards, which help credit officers in the field to undertake evaluations of applications, registration of client information, synchronization of field and head office information, which can ease microlending. The applications of such new technologies to microfinance are discussed with examples from other countries in an annex to Part 2, and an exploration of possible adaptations in the Brazilian context i s suggested. 78. While it is hoped that these measures will help to increase the contribution of microfinance and credit cooperatives, their role will remain small at least for some time, especially for the microfinance sector. A big thrust for expanded access could not easily be shouldered by such entities. Especially inthe case of microfinance, large scale expansion would be difficult without partnerships with banks, public or private, and if government access to privileged credit is to be avoided, deposit taking will need to be seriously considered, with its attendant risks and safeguards. The expansion of client interface is another reason for the consideration of this option. The cooperative sector, by contrast, is already much larger than microfinance, and is already in a position to mobilize deposits. Thus from the scale perspective it has larger potential. 3.3 Downscaling Private Banks 79. Brazil's 175 private banks represent overwhelmingly the majority of its 191 commercial banking institutions in numbers, and also today in the provision of broad aggregates of financial services. Their role in deposit and lending activity has grown to a dominant position over the last decade, from 42 percent of total deposits in 1994 to 52 percent in 2001, and from 44 percent of total loans to 56 percent over the same period. Four of these private banks are substantially larger than the others: Bradesco with 9 percent of total banking assets in June 2001, Ita6 with 7 percent, Santander with 6 percent (via their acquisition of Banespa) and Unibanco with 5 percent. Nevertheless, Brazil's two large public retail banks - Banco do Brasil and Caixa EconGmicaFederal still account for close to 40 percent of deposits (20 and 19 percent respectively in December 2001) and three federal banks (including the wholesale bank BNDES, Banco do B r a d and Caixa EconGmica Federal) account for around 31 percent of total lending (13,6 and 11percent of total bank loans respectively). 80. To what extent have these private banks already `downscaled' their services; that is, contributed to the provision of financial services for the poorer segments of Brazil's society? This issue is investigated in Part 3 of the study, in a number of dimensions; the provision of small deposit and small loan services; minimumdeposits, costs and fees; and delivery channels. The role of new technologies in the provision of financial services to smaller and less affluent clients, is discussed, assessing the present use of such technologies in Brazil and offering examples from other countries. Finally, the experience of banks in other countries which have successfully achieved outreach to less affluent clients is described in an annex. 81. The analysis reveals, first, that private commercial banks in Brazil already play a significant role in the provision of financial services to small clients. Small transaction size per se does not appear to be an absolute barrier for private banks in entering this segment. However, there is a high degree of segmentation in the types of banking services which private banks wish to provide. On the deposit side, they are more likely to enter the possibly more lucrative segments of the market, such as sight deposits, which are unremunerated or special savings deposits, where remuneration i s much lower than the prevailing interbank rates. This appears to be the case despite the directed credit requirements placed on these categories of deposits. Private banks are not keen to serve small clients' term deposit needs, which are remunerated at rates close to the interbank rate, presumably because its advantages compared to safe assets such as government securities declines. Thus, transactions-based needs and needs for safe stores of value are met, though needs for returns and investments are served less. There i s considerable variation xxx Executive Summary between private banks' individual business plans in terms of outreach, whether they are wholly national or partly foreign-owned. 82. One important corollary is that public banks are not the principal providers of services to small account holders; however they do provide services in this segment to those account holders who are more interested in investments, and returns through term deposits, compared to those whose needs are based on transactions or on safely maintaining the value of their savings. And evidence suggests that low income clients do not show a strong preference for public banks. 83. Findings suggest that fees and costs matter. Looking at the prices of banking transactions, although fees for a basic package of banking services do not appear high, they could nevertheless constitute a significant proportion (one to two percent per month) of the income of a less well off individual. In terms of institutional variation of transactions costs, although average fees for private and especially foreign banks may be higher than federal banks, the difference i s not significant looking at comparable public and private banks. Some large private banks are as competitive in terms of prices and service range as federal banks, and in other instances, some federal bank service fees may be high. Transaction fees appear to be a greater obstacle for personal banking services than for the provision of credit to enterprises; high interest rates are the most important factor for enterprises, especially small enterprises. Another finding i s that some specific banking services appear to be more uniformly available across different banks (such as ATM use and debit card facilities) while others are less likely to appear in basic packages of services without extra charges, and may be more competitively priced across institutions (checking facilities, overdraft facilities, provision of printed statements, and debit orders). 84. The choice of delivery channels for banking services, and also the choice between banks and non-banks, appears to be affected by regulatory requirements which raise the costs of establishing banking services, especially bank branches. The growth of correspondent banking has been partly in response to these, but correspondents also help to spread fixed costs. In future, competition in correspondent franchises and further broad basing of their services (e.g. including government bond distribution) could help to ensure their competitive edge i s maintained. But also, prices of other types of services such as bank card networks may also be affected by limited competition in their provision. Differentiated solutions are requiredto address these issues. 85. New technologies based on phone connectivity (telephone and internet based) are widely used inBrazilbut not for services to the lower income segments. Teleconnectivity inBrazil is relatively low in average terms, although there are many well-to-do who use internet banking. Many countries including Brazil are making a conscious effort to encourage more universal access to and use of telecommunication services, through nondistortive targeted subsidies, and several examples of good practice are available. 86. New measures have beenrecently introduced which require banks to expand microlendingfrom their sight deposit resources, up to an obligatory ceiling of 2 percent - alternatively these funds go towards raised reserve requirements. This i s an implicit tax on the financial system. Other measures such as disclosure requirements, guarantees, or initial tax breaks, are more widely used internationally. Another new measure recently introduced, the basic or minimum account, i s however widely used in many countries, although it is rarely offered on a free basis, as inBrazil. 87. There are a number of additional good practices which have been identified by banks in other countries to promote the downscaling of their business, which could be considered by banks in Brazil. A first group of such practices depends on the development of appropriate products; which apart from appropriately priced basic accounts include employer-sponsored `payroll' accounts, and the adoption of technologies such as credit scoring, which may aid outreach to small clients. A second approach rests upon the development of partnerships, with microfinance institutions, or with other partners such as xxxi Brazil: Access to FinancialServices community organizations, and a third rests upon image differentiation. Financial training of the target population would be greatly beneficial and can be delivered incombination with outreach programs. 88. Finally, there are many measures in the overall environment which could help the banking system in terms of outreach. These are discussed in Part 6, which deals with institutional infrastructure, and in Part 7, which discusses how the government could help outreach by easing entry requirements for new bank clients, by reducing current regulatory arbitrage between different segments of the financial sector, through a number of measures which reduce overhead costs imposed upon the banking system, and through programs of active support for bank outreach. 3.4 Partnering Non-Banks 89. Although statistics on the financial system indicate a heavy concentration of assets in the banking system (between 90 and 95 percent), these data may overemphasize the importance of the banking system. Banking system assets as representedbelow consolidate data for Brazil's multiple banks, and to the extent that departments of these banks also engage in non-bank activities, these would be reflected under bank assets. Moreover, some activities which provide finance to consumers and producers, in the form of trade finance, commercial credit or through factoring, are deemed to be commercial activities and thus not included in the statistics above on financial intermediation. For example, the discount of invoices, included in credit data, amounted to around R$6 billion in 2001, while the estimated factoring portfolio at the end of 2001 was R$27 billion, and factoring represented an estimated 2.3 percent of Brazil's GDP. Credit cards receivables are only included inthe statistics above to the extent that these may be refinanced by banks. Finance companies' portfolio was 0.4 percent of GDP, amounting to R$4.4 billion in value, compared to an aggregate consumer credit portfolio of R$70billion for the entire financial system. And Brazil's outstanding leasing portfolio represented 1percent of GDP at end 2001, with a portfolio of R$11.6 billion. In many countries, non-bank activities, especially trade finance, factoring, and the leasing of equipment, have been valuable sources of financial access. Part 4 of the study evaluates their present and potential contributions to access. 90. Broad findings indicate that weak bankruptcy and secured transactions laws which complicate the use of collateral in Brazil create incentives for forms of financing (factoring, leasing, trade finance) other than straightforward credit. These forms of financing reduce risk by securing repayments against accounts receivable, or by maintaining title, and using self-executable legal instruments. Additionally, financial institutions offering such services do not face many of the high costs of entry of banks, and some non banks such as factoring and finance companies are better equipped to handle the highoperating costs for small scale clients. As such, they could have an important role, although factoring and trade financing only provide working capital and cannot meet investment needs. Leasing could potentially contribute to investment finance but has been little usedfor this purpose. 91. The contributions of these sectors have been below their potential, in part because their own legal underpinnings are fragmented and sometimes challenged as to their legitimacy. Factoring does not have a law governingits services and leasing laws arefragmented. The treatment they have received in civil litigation has sometimes been perceived as arbitrary. In both, there remains confusion between these activities and normal bank credit. Thus in the case of leasing it is sometimes concluded that leases are essentially similar to loans; also distinctions between financial and operational leases are sometimes unclear; in the case of factoring, the discounting of invoice or post-dated checks i s considered at par with the purchase of receivables. Until a mature distinction regardingthese novel contracts is established,it will be hard to expand thegrowth of these industries in Brazil. xxxii ExecutiveSummary 92. Looking more specifically at individual segments, the greatest potential appears to lie in the relatively overlooked area of factoring, which makes a sizable real contribution today in terms of finance and services to small enterprises and could be strengthened and legitimized to extend its outreach. At present, the distinction between genuine factoring and the discounting of invoices or checks i s not sufficiently clear in practice. Consolidation of the dispersed legal framework of the factoring industry could contribute towards this. A proposed draft law on factoring has been pending in Congress for years. Its final passage would be beneficial to the industry. The legal framework should also clarify distinctions between recourse financing and non-recourse financing by factors. Financing with recourse though not forbidden has not been recognized in legal jurisprudence. Lack of recourse in cases of fraudulent transactions by clients is costly for factoring companies. Also on the regulatory side, the tax treatment of factoring needs investigation. At present, it is subject to all financial transaction taxes as well as to commercial taxes. A case has beenpending inthe Federal Supreme Court to resolve this. 93. Permitting factoring companies wider access to finance would be desirable for example through the issue of commercial bills or debentures, currently not possible for those factoring entities not constituted as shareholding companies. This could be eased if the proposed draft law i s passed. And international factoring could be eased by the relaxation of certain restrictions against holding foreign currency accounts. 94. The leasing industry has fluctuated markedly in Brazil in the past decade, Regulatory arbitrage in the tax system has considerably influenced its development and contributed to its boom, especially in consumer vehicle leasing, in the latter 1990s. Further changes in the tax system together with the impact of devaluation on dollar indexed loans, and also, judicial activism which tended to shelter the lessee led to its subsequent contraction and to the move away from individual leasing by lessors. 95. Leasing is little used for investment finance at small firms; it i s an activity mainly for large firms, where the obsolescence of equipment and/or contracting of maintenance for operational leasing makes this an attractive option for investment financing. For individuals, leasing has been used largely by the better off, primarily to acquire vehicles. Leasing is largely beyond the domain of the less well off segments of consumers, or small firms, and thus today of limited immediate importance in terms of expansion of access. The relatively slow growth in leasing of machinery and equipment has also been ascribed to the availability of alternative sources of finance (such as the FINAME credit lines of the BNDESbank), which are longer interm(3 to 5 years). 96. The industry may also benefit from clarifications in its legal and tax framework. The legal framework for leasing is scattered and for many purposes leasing falls under the Civil Code. A dedicated leasing law i s under preparation and i s due to be submitted to Congress. This will be important interms of clarifying the framework for leasing operations. The tax advantagesof leasing may be difficult for smaller firms to realize as this would require the use of more complex tax forms (`lucro real') which permit itemized deductions, instead of the simplified tax forms (based on `lucropresumido'). The tax regime for leasing i s also complicated by the treatment of residual value. If low, leases are sometimes not recognized but treated as loans. The industry i s seeking legal clarification through a pending suit and its outcome, will considerably influence the extent to which the industry regains its attractiveness. There also appears to be a need to clarify and correct accounting practices with regard to leasing. InBrazil, financial leases are booked as operational leases, thus allowing capital goods acquisition without raising the gearing (indebtedness) ratio. However, this accounting treatment i s unorthodox. 97. Partnerships betweenfinancial intermediaries and commercial companies could constitute a feasible alternative to promote operational leasing for smaller enterprises. This would be promoted by more competition in the industrial segments which could constitute the lessees. xxxiii Brazil; Access to Financial Services 98. Finally,finance companies are an important alternative for consumer credit at low and medium income levels, which have the advantage for the consumer of easier access than bank accounts. Independent finance companies are a very small market segment today as most finance companies are part of multiple banks and wider financial conglomerates. For financial institutions such as banks, finance company subsidiaries offer an avenue for product differentiation as well as lower cost funds (as there are no reserve requirements). However, the implication is that access to funding is tight for those finance companies which operate independently, as they are not deposit taking institutions. Most of their services could today be provided by multiple banks. 3.5 ChannelingSpecialized Finance 99. A large part of the issue of financial access which has occupied the attention of policymakers has been the problems of access to specialized financial services, in sectors where it has been particularly difficult to attract private finance. Two areas of the financial system which stand out in this context are rural financial services, and access to housing finance. A substantial part of the system of `targeted' or `earmarked' credit in Brazil arose from the need to address these specialized credit needs. Part 5 examines the evolution of one of these areas; rural finance. It evaluates the rationale for and cost of the present system, and discussesthe extent to which it meets goals of outreach and sustainability. 100. It has been pointed by many that rural finance faces special difficulties and thus merits special solutions. There are first geographical issues of access in remote rural areas. These are combined with typically high rates of poverty, lower population densities, isolated markets, highly covariant risks, limited opportunities for risk diversification, sharp seasonal variations in income and in the demand for and supply of financial resources, and lack of traditional collateral. All these factors result in high transaction costs and higher perceived risk in providing financial services in rural areas. It is argued that as a result, credit to rural markets is `rationed', in the sense that lenders tend to limit quantity, even if prices are flexible. 101. Brazil's response to the challenge of rural finance has been similar to that of a number of countries; direction of credit, with a combination of quantitative controls (in the form of mandatory lending to agriculture) and price controls (lending at below market rates of interest). In Brazil as in many other countries, selective or directed credit has been channeled largely through public financial entities, either directly or through onlending arrangements with private financial entities. Brazil's largest banks, the Banco do B r a d and the Caixa Econbmica Federal, have served as the primary vehicles for channelling credit to the rural and housing sectors respectively. There are also mandated private sector programs for directed credit: 25 percent demand deposits (unremunerated by law), must be devoted to rural finance, and 65 percent of passbook savings deposits are earmarked for housing loans. Directed credit shrank from mid-2000, as part of the govemment-financed restructuring of the federal banks that removed a substantial volume of directed credit from their books in exchange for government debt. But despite the decline from about 54 percent of total credit inJune 2000, to about 38 percent inMarch 2002, directed credit remains high. 102. The analysis finds that the Brazilian rural financial system today i s far from good practice: it has limitedoutreach at highcost instead of mass outreach at low cost. Most rural households still have no access to formal financial services. Credit remains highly concentrated, with three-quarters of all clients receiving less than 6 percent of credit while the top 2 percent of contracts accounted for 57 percent of all credit. Public sector programs and institutions play a predominant role in the national rural credit system, and these have `crowded out' the private provision of credit at market rates. Freely provided private sector resourceshave declined to as little as 5 percent of formal rural credit. xxxiv Executive Summary 103. To move away from the present situation, Brazil needs to implement a gradual but credible transition program to market determined interest rates on all rural credit and savings. While ovemight removal may create at least a transitory decline in the allocation of resources to the rural sector and may not be politically acceptable, (i)there could be increases in interest rates on the most heavily subsidized programs (notably PRONAF, whose enormous timely repayment rebates result in highly negative nominal interest rates). (ii) Another area where phased rates could be applied is with regard to the 25 percent directed lending requirement. (iii) Instead of "equalization payments", commercial banks could compete for competitively priced administrative fees for administering subsidized funds. Preference would be given to those who require lower `equalization' or lower fees. 104. In parallel, remaining subsidies could be explicitly monitored and budgeted and refocused away from interest rates to transaction costs. Subsidies have largely been captured by well-to-do farmers and non farming corporations, for whom there is no justification for a credit subsidy on either equity or efficiency grounds. The net impact has been to inflate rural land prices as subsidies are capitalized into the value of land, and to promote a more regressive rural income distribution. In view of the fact that ultimately public funds finance programs such as both CrediAmigo and PRONAF, there is an urgent need to apply a more uniform approach in applying lending interest rates in such programs. 105. Brazil could try to address information and insurance constraints to reduce risks in rural lending. There is good potential of advancing a range of interrelated financial instruments for risk management; price risk management for agricultural commodities, area based index insurance and warehouse receipts. These instruments allow farmers to increase their specialization and adopt higher risk, higher return production strategies. Ina wider context, strengthening rural finance requires attention to the same issues as those affecting overall financial intermediation: addressing issues of high spreads, high administrative costs, and relatively high profits for banks operating in Brazil versus in comparable economies, combined with highfinancial sector taxation. 3.6 Installing InstitutionalInfrastructure 106. Substantive laws regarding the legal and judicial protection of creditor rights in Brazil are not very different from those in other French civil law countries, and are generally perceived as adequate by creditors. However legislative procedures are protracted, with recourse to successive appeals or injunctions, which diminishthe value of legal protection. Moreover,judicial decisions on credit disputes are perceived to be pro-debtor, reflectingjudicial social activism, sometimes to the disregard of what is established in the law or the contract. Consequently jurisprudence and patterns ofjudicial behaviour play a role as or more important than the law itself in regulatingcredit disputes. 107. Especially for small loans, judicial processes are avoided as far as possible, due to the uncertainty, expense and time consuming nature of judicial processes. Instead, extra-judicial collection through collection agencies is relied upon, despite tax advantages in proceeding through the court system. The use of self-executable credit instruments which can help to limit court procedures are also popular. However these are more common for larger contracts than for small retail loans. Small claims courts cannot be used for this purpose either, because only individuals can initiate causes in these courts. Instead, banks rely on collection agencies and the enforcing power of black lists, through which access of borrowers to credit or even the use of checks is effectively denied. 108. However the most popular consumer loan products are not self-executable instruments (the overdraft facility or "cheque especial"). This i s largely due to the formality of requirements for the recognition of a contracted loan (including signatures and witnesses). The ease of extending informal xxxv Brazil: Access to FinancialServices consumer loans through ATMs counterbalances for financial institutions the loss of their self-executable nature. Provisions for loan losses are reflectedinhigher spreads. 109. Potential areas for reform which could strengthen creditor rights would include first of all efforts to work with the judges and society at large to show that decisions that benefit a specific borrower in a specific case have broad repercussionsthat harmborrowers as a whole, at a wider level. As a political process, judicial activism must be dealt with inthe political arena, by building a constituency in favour of the strict enforcement of credit contracts. Judicial education is additionally desirable for the purpose of establishing a better understanding of specific financial instruments, such as self-executable instruments and Ckdulas, to establish more uniformity injurisprudence. 110. Additionally, the selfexecution of instruments used for small borrowers in the modern context can be reinforced, for example, the overdraft facilities offered over the ATM or the duplicatas of small firms. In keeping with new technology, legal recognition of these debts (which at present is not possible due to their lack of signature and computerized entry) could be explored. 111. A promising area of reform in the short term is inchanging theprocedural codes to speed up judgments, an initiative that would likely be supported by the magistracy. Judges largely agree that judicial slowness could be substantially decreased with a reform of procedural codes to reduce the possibilities of postponing a decision or of appeal processes. Further, to reduce costs and time for judicial execution of debt, the Special (small claims) Courts could be opened to small credits. These courts, which judge exclusively low-value disputes, operate with much lighter procedures, do not require the parties to be representedby a lawyer, and their decisions cannot be appealed. For that either the law that created the "Juizados Especiais" would have to be amended, to allow banks and other financial institutions to act as plaintiffs, or a new law would be necessary to create special claims courts able to judge small debt collections. The proposed new courts could be designed to combine pro-debtor services as well as the settlement of small loans, inorder to make them socially and politically acceptable. 112. Finally, current tax write offs against profits on uncollected small claims can be evaluated in a wider context as these may be sending perverse signals to both borrowers and financial institutions for debt repayment. International practice in the design of more incentive compatible tax treatment for small claims can be investigated. 113. The use of collateral is also rendered difficult by current procedures for the creation and perfection of security rights. There are a series of different registries, and few locations are interlinked. Systems of registration are in some cases time consuming, when the manual transcription of large deeds and documents is needed, and systems of retrieval are complicated by the lack of electronic filing and indexing. Finally costs of filing tend to be regressive, impacting disproportionately on small transactions. These difficulties impede the development of secured transactions. It i s also alleged that despite recent improvements there are still a large number of registries which were awarded to persons on a franchise basis, sometimes, passed down through generations, without the application of appropriate certification criteria. 114. Measures which could lead to an improvement in the process include the adoption of electronic filing, retrieval and indexation systems at all registration offices in the country; the establishment of networksfor the interlinkage of registries,initially if necessary on a pilot basis within cities, regions or state, and eventually, across regions and across the different types of registries required for different types of security interests (e.g., pledges versus mortgages, currently under different registries). There is a possibility of legal constraints on such networking, and any modifications which may be required in the legal framework, at both the federal or state levels, which could be necessary to accompany such a process of networking of registries should be explored. Also, deadlinesfor turnover timefor registries xxxvi Executive Summary of deeds and documents could be adopted, which will shorten the time taken for these processes and may encourage the adoption of modem technologies. And in parallel, documentation requirements for secured transactions in real estate could be examined, to evaluate the scope for simplification or consolidation. The present five day time and costs taken to conduct a search and certification could both be reduced by aself-guided web search. 115. Efforts should be made to address the regressive tariff structures of registries and notaries public, which at present imply disproportionately high costs for smaller transactions. One solution may be to remove the caps on the upper end of the scale and if necessaryuse the proceeds to lower transaction costs for small scale transactions. Finally, the extent to which notarial and registration services are still operating on afranchise basis could be examined through the offices ofjudicial oversight and the extent to which this is likely to remain a longer term problem could be assessed. Measures can be investigated which could be adopted within the present legal framework to ameliorate the near term impact of such arrangements. 116. Especially in view of the difficulties of executing collateral and establishing creditor rights, improved credit reporting may well be a more important route for Brazil to enhance the security of loans to small borrowers. In many countries including the USA, experience has shown that there was an association between increased credit to small entrepreneurs, and the introduction of credit reporting. Brazil already has effective core systems, both public and private, for credit reporting, and competition in the credit reporting industry is developing. Nevertheless there are some areas for improvement. A first is the limited extent to whichpositive information is used incredit histories, which in tum i s partly due to legal restrictions on the sharing of such information, and practical difficulties in establishing information sharing arrangements. There are signs that this is now evolving within the context of trade associations (for example, for all finance companies, through a new initiative, Acrefi Positivo), and such initiative should be strongly supported. Provisions in the bank secrecy law which refer to such restrictions as well as guidelines on notification, could be reevaluated against international norms to ensure appropriate balance between protection of private information, and the sharing of information which may help all borrowers. There could also be a reconsideration of current time restrictions on information which limit credit data to five years. Brazil is an outlier in this respect. 117. To enhance available positive information, regulations related to the dissemination / collection of payment data from utilities could also be examined, to try to enable data from these sectors to enter credit reports. Utility tariffs could also be structured so that service providers can take into account the credit standing and capacity to pay when providing services. It could be important to recognize the value of such positive information for low-income consumers who may lack a formal credit history. 118. In parallel, efforts could be made to prepare consistent strategic directions for the Central de Risco, at Brazil's central bank, including data which it would collect. Access to other public databases such as CADZN,which is currently limited, could also be examined. 119. These measures would need to be combined withjudicial educationloutreach programs on the benefits of credit reports, and records of case law in this regard could be prepared to promote the consistent application of current and new laws. Outreach programs for consumers, should also be launched, especially for low income consumers, so that they are aware of their rights, understand more fully how credit works and also understand their obligations. Consumer protection norms could be evaluated against international practice, and some areas where current practice may fall short of such norms, for example with regard to notification of access of a credit record and the taking of adverse actions based on such records, limits on the marketing of information, or changing erroneous data, could be investigated. xxxvii Brazil: Access to Financial Services 3.7 Enlisting the Government 120. To what extent will markets eventually solve the problem of financial exclusion, without government intervention? As the preceding chapters show, the role and presence of the government is pervasive. The nature of its regulation of financial entities significantly affects the services they can offer, and the appropriateness of these services for the financially excluded. Many observers believe that the government needs to take on a more proactive role, ranging from financial support to the indirect establishment of appropriate incentives to formal financial institutions to expand access. Some observers point out that governments have a particularly significant influence on banks due the latter's particular status. Banks' strategic economic position coupled with their government granted authorizations to operate and guaranteed deposit insurance renders them to be more socially responsible than ordinary undertakings. But social responsibilities may conflict with private profit. To what extent should the government exercise its influence or authority over the banking system to achieve social objectives? These issues are broadly discussed in the concluding chapter, based largely on the experience of major developed countries. 121. As a prelude, it is pointed out that financial access cannot be considered independent of the overall depth of financial services, and this depth i s influenced by a spectrum of government policies which may include those which lie outside the financial sector. The first section of the present chapter therefore examines the influence of such policies. These include, first and foremost, overall macroeconomic policy which has a major influence on financial system deepening. Macroeconomic stability, with sound fiscal management helps to establish the environment for inflation-free growth and provides an appropriate environment for financial system deepening. The government's taxation and public borrowing requirements are a central part of this environment which impacts upon the financial system. Second, government policies ina range of regulatory areas also impact upon the financial system. For example, bank branch openings are affected not only by regulations pertaining to banks and other financial intermediaries but also, prevailing competitive forces and labor laws. These factors affect interest rates and spreads and hence the costs of financial intermediation. 122. In the context of Brazil today, it is clear that macroeconomic policies with high borrowing requirements, financed though the issue of domestic debt, held largely by the financial system, has contributed to the `crowding out' of lending to the private sector. This has been combined with high taxation of the domesticfinancial system, both explicit and implicit (through highreserve requirements) and a consequent disincentive towards financial transactions and financial deepening. There has also been a high proportion of directed credit in total credit, which has tended to raise the cost of freely allocated credit. 123. Inevaluating the contribution of different segments of the financial systemto access, it became evident that there has been an uneven impact of regulatory policy across different parts of thefinancial system, with consequent `regulatory arbitrage' between its various segments. This explains at least a part of the relative `success' of some segments of the system, compared to others. For overall success in improving access, greater harmonizationof suchregulatory impact i s desirable. 124. Regulatory policy in other areas impacts indirectly upon the financial system, notably, labor market policies which add a considerable burden to hiring costs in the formal financial sector, thus providing incentives for the development of financial sector segments outside this system. Another example, not detailed here, i s competition policy. Increased attention to competition in some formal financial sector segments may broaden the base of support for expanded access. 125. Given the macroeconomic and regulatory setting, the second part of this chapter suggests that nevertheless policies favorable towards the expansion of access to financial services can be, and have xxxviii Executive Summary been, adopted in a number of countries. Based largely upon the experience of developed countries, especially the U.K.,the U.S.A. and Canada, a series of policies favoring access are investigated. Broad lessons to emerge from the cross-country examples are: 126. Government support, through legislation or enabling regulation which takes the form of consumer protection, (e.g. for the prevention of discrimination or predatory practices, for better disclosure of lending practices, and for the regular collection of quality data on these practices) are all critical to inform good policy and to alter the behavior of institutions. Many developed countries have a series of periodic reporting requirements on parameters related to financial access, which Brazil could examine. In many cases, these have been combined with analytic support from academic research, or research departments of the central banks or ministerial agencies of the countries concerned. Such policies towards disclosure apart from encompassing measures of access, can also help with the appropriate targeting of access -as inthe example of the Index of Multiple Deprivation of the UK. 127. Laws onfinancial disclosure,and the implied moral suasion, are powerful tools in shaping the behavior of financial institutions, especially if there i s an alignment of financial and non-financial incentives to reinforce legislation. Provisions for enhanced disclosure on loans to specific communities or sectors have been powerful tools in enhancing access, as demonstrated by the Community Reinvestment Act of the USA. Many countries have since tried to emulate this (e.g. South Africa) and in others, the threat of CRA type legislation has led to enhanced voluntary efforts by private banks to expand both disclosure and access (e.g. in the UK and Canada). UK banks will begin this summer (2003) to publish statistics on community lending. 128. As recently recognized in Brazil, there is a case for obligatoryprovision of someform of basic banking services; this is a widespread practice. There are many examples of such programs, from the "lifeline" (minimum) service provision bank laws in many states of the USA, to the basic account requirements in Canada.UK banks have also begun to provide basic accounts though without formal legal requirements. Such basic accounts emphasizefirst of all access to money transmission mechanisms and thepayments system, and second, depositfacilities. Services such as check writing and credit need not form a part of a basic account. Credit facilities offered with Brazil's new basic accounts should remain under review. The pricing and operating costs of such accounts also need to be constantly monitored, and establishing a zero price account may become expensive over time. Electronic funds transfers for federal payments and `First Accounts', as launched inthe USA, could accelerate the reduction of the "banked'. Brazil also proposes to transfer federal payments to such accounts. 129. Encouragingpartnerships, first of all, of formal financial institutions with community-based credit providers to employ alternative delivery channels can be very productive. There are Community Development Finance Institutions in some of the country cases discussed (UK), which have a very different role from traditional public development banks. These agencies are typically small and not-for- profit, and they provide a dedicated link between the recipient of financial services and the government. Second, sound partnerships between the public and private sectors, built around shared understandings of their different roles and what each brings to any transaction, are also valuable. Such institutional support needs to be involved in the efforts of financial outreach. These include not only business administration and fund administration agencies but also, non-profit agencies and community investment agencies who work closely with target communities and can assess both their needs and their eligibility. 130. In terms offinancial support, most successful examples of credit support for products that serve disadvantaged populations entail indirect support by government, where private credit providers still share in the risk. The US Community Development Financial Institutions Fundi s an example. The Small Business Administration's Microloan Program i s another, where the SBA makes funds available to community-based lenders (intermediaries) which, in turn, onlend to eligible small businesses and xxxix Brazil: Access to Financial Services individuals. The local provider handles all underwriting and decision-making. Such initiatives can be combined with policies to promote access by financial institutions to secondary credit markets. Such activities can be designed to complement rather than compete with activities of the private sector. 131. Indirect regulatory policy impacting uponfinancial sector competition can also impact access, even if this is not immediately apparent. Brazil's banking structure, dominated by a handful of institutions, bears some resemblance to that of the UK, where a recent investigation of competition issues particularly with regard to its big four banks revealed for example, the payment of service charges for access to money transmission. The impact of competition on other areas of access to financial services has also been detailed and pointedout in other countries too. A full evaluation of the status of competition inthe provision of `downscaling' inthis context is suggested. 132. Finally, the role ofprograms of financial education and literacy are also key; in this context programs which offer incentives for family support to financial planning are noteworthy. XI Part 1. Assessing Access PART 1. ASSESSING ACCESS "Access to financial markets is important for poor people. Like all economic agents, low-income households and microenterprises can benefitfrom credit, savings, and insurance services. Such services help to manage risk and to smooth consumption ....and allow people to take advantage of profitable business opportunities and increase their earnings potential. Butfinancial markets, because of their special features, often serve poor people badly...Since poor people often have insufficient traditional forms of collateral (such as physical assets) to offer, they are often excludedfrom traditional financial markets.. . transactioncosts are often high relative to the small loans typically demanded by poor people. And in areas where population density is low, physical access to banking services can be very diflcult ..." WORLDBANK,WORLDDEVELOPMENT REPORT^^^^-^^^^ "The area offinancial system regulation and organization will increasingly befocused on thefollowing aspects:. ......creation and improvement of financial instruments and activities including those intended to broaden and cheapen credit.." ... it is a priority to improve the regulation of mechanisms that broaden the access of the population to thefinancial system. ~ ~" HENRIQUE CAMPOS MEIRELLES,GOVERNORTHE CENTRALBANKOFBRAZIL,JANUARY 2003 DE OF 7, 1. GROWTH, POVERTY REDUCTIONAND ACCESSTO FINANCIAL SERVICES 1.I Introduction: The Importance of Access 1. Economic growth, by raising the total income of society, i s a primary vehicle for improvement in human well beingand the reduction of poverty. The quality of growth can be further enhanced by directly targeted education, health and poverty outcomes. To achieve these goals "poor people must be empowered to take steps to improve their lives, and governments must assist them by ensuring that they can obtain the services they need".' Targeted poverty interventioni s especially important in Brazil, which has one of the most unequal income distributions of the world (Table 1.1); significantly higher than the Latin American average, and higher than other upper middle-income countries. The income share of the lowest quintile of the population in Brazil is significantly below the average for middle income countries.2 ' Millennium Development Goals, 2002, World Bank. These goals have been embraced by the World Bank, the community of the United Nations, and by the Government o f Brazil. * The most recent census numbers (end-2000) do not show a significant change in the Gini coefficient. This is consistent with cross-country studies showing the persistence of Gini coefficients over time. Brazii: Access to Financial Services Brazil's government is keenly aware of these issues and has committed itself to reducing Brazil's rate of extreme poverty by 50 percent by the year 2015.3 Table 1.1 Brazil Income Distributionin InternationalComparison - 60.5 2.5 LatinAmerica & Caribbean uuu Note: Country averages are calculatedacross the most recent observation per country. Sources: Datafor individual countriesare for differentyears and sources (Deiningerand Squire (1996)and Lundbergand Squire (2001) GDP per capita data are from the World DevelopmentIndicators(WDI), World Bank, Staff estimates. 2. How can the financial systemof Brazil contribute to the reductionof poverty and inequality? It is increasingly accepted that greater financial system depth and soundness contributes to broad-based economic growth with poverty red~ction.~Deep and efficient financial markets promote investment and total factor productivity growth through their role in selecting and monitoring projects, diversifying risks, reducing asymmetries of information, improving resource allocation, and encouraging the optimization of scale, time frame and technology. It has been shown that in the case of Brazil, the possible effect of increased productivity on growth could be large.5 Strong financial systems help absorb shocks. In neighboring Argentina and Mexico, shallow domestic financial markets have magnified international financial market turbulence, impacting adversely on the poor. Conversely, weak financial systems and resultingfinancial crises entail huge fiscal costs that crowd out social spending and especially if financed by inflation, disproportionately impact upon the poor. Inthe case of Brazil, the crisis of 1995 is estimated to have had a fiscal cost of about 13 percent of GDP.GMoreover, badly managed financial crises cause severe disruption to economic processes, erode capital stock, erase confidence in the banking sector and set back poverty reduction for long periods. Recent crises in Mexico (1995), Ecuador (1998) and Argentina (2002) have significantly increasedthe incidence of poverty.' 3. Apart from the large absolute size of Brazil's financial system, by far the largest in Latin America, the overall depth of financial intermediation in Brazil i s greater than its large neighbors in the region (Table 1.2). The share of financial assets in GDP in Brazil is relatively high, at almost 140 percent of GDP, compared to 81 percent in Argentina and 68 percent in Mexico.8 However its achievements in terms of financial depth are modest compared to some large East Asian economies such as China, with financial assets /GDP of almost 200 per cent despite a per capita income of less than a fifth of Brazil; or Defined to include persons living in households with per capita income of less than R$65 per month. At this extreme, Brazil had a poverty headcount in 1999 of 22.6 percent. See Brazil: Attacking Brazil's Poverty. World Bank Report No 20475-BR. March 31 2001. Discussions of the association between growth and poverty reduction may be found in Dollar and Kraay (2001). See (MOP PFSECALI) for a discussion of this issue. The discussion o f the association between growth and the financial system can be traced through King and Levine (1993);Rajan and Zingales (1998), Levine, Loayza and Beck, (2000),and World Bank (2000).See also IMF working paper by Holden and Prokopenko (2001), for a well- documented survey of the literature. `The McKinsey Global Institute (1998). fiscal costs o f banking crises have been as much as 50 percent o f GDP inthe 1999 Indonesiacrisis (World Bank, Financefir Growth,2000). See also, Honohan and Klingebiel (2000) and Caprio and Klingebiel (1999). * De Ferranti, and Perry, 2000, "Securing our Future in a Global Economy". Using a measure of financial assets as constructed in Table 1below, for 1999/2000. Page2 Part 1, Assessina Access Malaysia, with financial assets to GDP of 370 percent, although its per capita GDP i s also below that of Brazil; US$3,390 compared to around US$4,350 in Brazil in 2000. And Brazil exhibits one of the highest interest spreads in the world and the maturity of loans i s very low, mostly less than one year. Moreover, the banking system has remained highly profitable, at least since 1999, and well capitalized, with capital ratios between 8.4 to 9.5 percent. Table 1.2 Depth of FinancialMarkets Braziland Other Emeraina Economies (1999120001 - Sources:World DevelopmentIndicators; IMF InternationalFinancial Statistics; IFC E&&gi;g - Markets Database; BIS. 4. But beyond the association of financial depth and stability with growth and crisis prevention, can the financial system also accelerate the reduction of poverty through targeted interventions designed to broaden the distribution of financial services? Brazil's unequal distribution of income and wealth may contribute to the structure of its financial sector. On the one side, the Brazilian financial system is the largest in the region and offers a broad variety of services to its clients. On the other side, according to estimates, around 58 million Brazilians have bank accounts, or a third of the total population, while some 80 million, or half of the total population is considered "bankable".' Central Bank statistics show that nearly 30 percent of all municipalities have no bank branch (Appendix Table A1.2). There are seven states (out of 27), mostly in the North, where the share of municipalities with no bank branch is over 70 percent. About 1,680 municipalities, out of a total of some 5,600 municipalities, have no bank branches (Appendix Table A1.2). The 1,400 or so credit cooperatives offer basic financial services only to about 1.5 million Brazilians, mostly in urban areas. The proportion of Brazilian households with access to more sophisticated financial services-such as mutual funds or insurance products-is low. Brazil's government is aware of these constraints on access and since 1999, has accelerated its search for means to expand access." InDecember 2002, there were 58.3 million savings accounts and 45.6 million current accounts inBrazil's banks, according to data maintained by the deposit insurance fund, FGC. FEBRABAN, the association of Brazil's banks, points out that estimates for current account numbers could be higher, since FGC excludes accounts indeficit. Assuming that some persons have only a current account while some others have only a savings account, the figure of 58.3 is thus a lower bound on the total number o f bank account holders. loFinancial exclusion is not aproblemo f emerging economies alone. Even today (2003), the FederalReserve estimates that almost 13 percent of families inthe United States do not have a checking account, and that almost 60 percent of that group are low-income, nonwhite or Hispanic, or both. According to one source, an estimated 8 percent of the population is unbanked in England and Wales and 15 percent for Scotland as a whole. Inanother investigation, HMTreasury UK (1999), reported that approximately 1.5 millionUK households use no financial services. An estimated one in every six adults in Britain, representing up to 15 percent o f households have no current account, and hence no access to money transmission services. Formal saving is largely non-existent among low- income people. Between 32 percent and 37 percent of people over the age of sixteen have no savings. (How People on Low Incomes Manage their Finances, ESRC seminar, Westminster, December 2002). Financial exclusion is also a problem across transitional countries. A third of all Hungarian households do not have a banking relationship, including a quarter of all households in Budapest, and 45 percent in outlying villages (Michael Stegman (2003)). By Page 3 Brazii: Access to Financial Services 5. The improvement of access to financial services should help both consumers and producers to raise their welfare and productivity. Individuals can insure themselves against periods of low income or unexpected fluctuations in income, and maintain their consumption standards through the accumulationof financial savings.I2 For example, for rural farmers, savings constitute insurance, to protect themselves against periods of drought or crop failure. Savings also provide a vehicle for future expenditure needs, whether expected (for example for special family occasions, or for the purchase of significant assets such as a home) or unexpected. Access to savings and borrowings could also have longer term welfare implications, permitting people to borrow when young, for example for education or for other physical or human capital, and then repay and save for retirement when they are 01der.l~ 6. For a producer, access to credit for fixed or working capital enables an increase in production possibilities which can have far-reaching implications not only for the producer but for patterns of employment, occupational choice and even economy-wide productivity and growth. Financing constraints have been shown to feature prominently among the constraints of small and medium-size enterprises in some investigation^.'^ Yet other studies indicate that Latin American firms find the difficulty of access to financial markets to be the major obstacle to the expansion of their business activities, ahead of factors such as macro instability, taxes, and street crime.I5 One study of Latin American countries shows that access to financial services for smaller enterprises directly impacts poverty, due to the disproportionately large number of persons employed by such enterprises. Survey evidence on 15 countries in Latin America shows that the microenterprise sector accounts for 56 percent of all earners but 70 percent of the region's poor eamers.I6 In Brazil, an estimated 98 percent of registered firms are enterprises with less than 100 workers and account for about 45 percent of registered workers." 7. But expanding the supply of financial services to underserved segments of society can pose particular difficulties for financial intermediaries such as banks. Limited provision of financial services cannot be simply interpreted as an unwillingness to provide such services. Lending to some segments, especially the very poor, may be very risky, as there may be a real difficulty inrepaying, or temptation to not repay, with households at the margins of financial and cash flow resources. Persons with informal or irregular employment may face real difficulties servicing loans. The possibility of such problems imply that it is important for financial intermediaries to get information on their prospective clients, to assess these standards, Brazil's achievement o f bank accounts for at least a third of its population, which implies a much higher proportion offamilies with bank accounts, are far from poor, especially inview of its income distribution, albeit low compared to developed countries. I1The microeconomic underpinnings of the welfare implications of improved access to financial intermediation are discussed in Annex I. 1which points out that intermediation enables inter-temporal choices inconsumption and investment for the individual, allows the determination of the cost of capital and hence helps guide investment to its most productive use, and permits the social reallocation of savings from low to high productivity uses thus raising social welfare. l2The role of savings and borrowing in protecting consumption against unexpected shocks, first discussed by Milton Friedman (1957) in the `permanent income hypothesis' has since been extensively tested empirically, as discussed in Bond and Townsend (1996). l3The `life cycle' hypothesis as an explanation for savings and borrowing behavior, discussed in Ando, Modigliani and Brumberg in a series of articles in the 1950s and 1960s.See Mayer (1972), and Romer, (1996) for reviews. 14Hallberg (2001), `A Market-Oriented Strategy For Small and Medium-Scale Enterprises.' IFC Discussion Paper No.40, International Finance Corporation, Washington D.C. l5Galindo (2001), `Creditor Rights and the Credit Market: Where do We Stand?', Working Paper 448, Inter- American Development Bank, Washington D.C. I6Westley (2001), Inter-American Development Bank, Washington, D.C. SEBRAE (ServiGo Brasileiro de Apoio hs Micro e PequenasEmpresas) website (January 2004) http:llwww.sebrae.com.brl. Page 4 Part 1. Assessing Access their creditworthiness, but such information may be difficult to obtain reliably and costly to collect.'* Such difficulties can imply that even potentially good clients are underserved, and sometimes, entire communities may face limits on credit which cannot be increased in volume even by raising interest rates." Additionally, there are costs associatedwith the provision of financial services, and if the value of the services provided is small, or services are to be provided in sparsely populated regions, it may be difficult to cover such costs. Costs of administration may be high due to the need for intensive interaction with and education of clients. Maintaining accounts may be costly as poor persons' deposits (which can provide interest incomes to banks) may be low, while transactions needs (which are costly to provide) may be high. Such issues arise in all countries and are not unique to BraziLZ0There may also be country specific factors for example due to regulatory requirements which could impact on costs associated with the provision of specific financial services. 1.2 Issues Examined: Structure of Investigation 8. The challenge therefore is to identify ways of expanding access which acknowledge the special factors which tend to limit the expansion of financial services to underserved segments of the population and to suggest measures to mitigate their impact. This i s the fundamental objective of the present study. The study first attempts to assess the extent to which constraints to access operate, and to identify the nature of these constraints, at the level of different types of financial institutions and indifferent segments of financial markets. The study then seeks to identify alternative strategies for improved access and their implied trade-offs. Based on this, the study discusses public and private choices which could enhance the availability or reduce the cost of provision of such services, consistent with sound financial practices. 9. Financial exclusion-the inability to access necessary financial services in an appropriate form-can result from access difficulties relating to conditions, prices, or marketing of financial services, or from self-exclusion by marginalized populations, often in response to negative experiences or perceptions." Although often implicit, there are many conceptually distinct and sometimes overlapping definitions of groups which are underserved. First of all, the poorer segments of society are usually identified to have disproportionately low access to financial services, and the poor can be definednot only in terms of income but also in terms of wealth and assets. Other criteria used are those in specific geographic regions, which may or may not combine characteristics such as remoteness or sparseness of population with economic backwardness. These criteria have been extensively used in Brazil, for example, as discussed further below. Third, specific communities of persons may be identified to be disadvantaged in terms of financial access. These could for example constitute communities of specific racial groups, migrants or minorities. Additionally, the criterion of small size is often applied, particularly to producers, so that micro or small scale entrepreneurs are often identified to face special difficulties in accessingfinancial services, especially credit. Sometimes, more than one such characteristic may apply to a particular `underserved' segment of the population. Measures of access and actions to expand access vary depending on which groups of the underserved are being referred to. In the present study, poverty is These difficulties, referred to as problems o f `moral hazard' and `asymmetric information' have been extensively discussed by economists. Besley (1994) provides a survey. I 9Such theories of `credit rationing' have been discussed for example by Stiglitz and Weiss (1981), Williamson (1986). Besley (1994) and Murdoch (1999) have discussed this specifically inthe case of communities such as small farmers and microfinance providers, and Bond and Townsend (1996) have raised the possibility of such rationing with regard to minority communities in the inner city Chicago area in the USA. 2o For example a discussion o f these issues in the context o f South Africa may be found at the South African Banker's Institute. See Extending Appropriate Financial Services to Low-Income Communities at http:llwww.ahi.co.za!current/. ''Sinclair,Stephen P. "Financial Exclusion: An Introductory Survey," CRSIS briefing, Edinburgh College of Art, Heriot Watt University. Page 5 Brazil: Access to Financial Sewices the primary consideration though small size and location are also factors which are considered. Given the wide array of factors affecting access, selectivity in the analysis has been essential. The study pays greater attention to factors which directly affect access. Thus indirect impacts on access, operating via their effect on the overall depth of financial services (for example, the potential `crowding out' of private credit provision by government borrowing, or the impact of financial sector taxation) are not investigated in detail. 10. The balance of the present chapter (Part 1 of this study) sets the stage for the remainder of the study, by undertaking a broad overview of current patterns of access, based on a series of alternative measurements, and investigating possible explanations for observed patterns. Section 2 of this chapter summarizes basic findings regarding present patterns of supply of financial services, in terms of the establishment of financial institutions and their branches. It examines trends insuch services over the past ten years, based on countrywide published data, and undertakes an analysis of the factors which might explain the existence, density, and depth and type of financial services provided in different areas. The underlying question is, what factors can be associated with the provision of services by financial institutions? Next, sections 3 to 6 try to look at the demand for financial services relative to availability, especially from the perspective of individual consumers based on a survey of urban individuals undertaken specifically for the present study. Particular attention i s given to an evaluation of broad government policies towards access. Finally, Section 7 summarizes the findings of this chapter. 11. The following sections of the overall study (Parts 2 to 5) examine in further detail the supply of financial services to the underserved, assessing the actual and potential contributions towards access of different forms of financial institutions. Beginning with a range of microfinance institutions (including credit cooperatives) which have been increasingly regarded internationally as promising instruments for the expansion of access, the study examines why the development of microfinance in Brazil has been relatively limited (Part 2)? Next, given the predominance of the banking sector in Brazil's financial system, the study examines its present and potential contributions (Part 3), and also investigates the potential for expansion of the role of non-bank financial systems such as finance, factoring and leasing companies (Part 4). Part 5 explores the structure of special market segments important for issues of access focusing particularly on rural finance.23 12. Followingthe investigationof service providers, Part 6 of the study investigates relevant elements of the institutional infrastructure for financial services. Such factors impact particularly on the cost of provision of financial services, and hence, on bank spreads. Part 6 of this study therefore examines implications of current legal processesfor creditor rights for small borrowers, registries for securedcredit, and credit information systems. 13. The study concludes with a discussion of the potential role of policy makers in the expansion of access through different instruments and institutions (Part 7). Brazil exemplifies the interplay of macro and institutional determinants in limiting the size of the overall credit market. Pronounced market instability, high real interest rates, the crowding out effect of large public deficits, high explicit and implicit financial sector taxation including high reserve requirements all impact upon the overall shallowness of Brazil's financial markets. While accepting the significant role of these factors, the present study attempts to focus rather on microeconomic issues affecting the distribution of credit and other 22Morduch(1999) describes the increasinginterest in microfinanceas a vehicle for addressingthe needsof the underserved. 23Housingfinance issues, also investigated duringthe study, have beenomitted from this reportdue to their detailed treatment in a parallelBank study (2002)`Housing Policy inBrazil: MaintainingMomentumfor Reform', Report No. 24335 BR. Page 6 Part 1. Assessing Access services, and explores the possibility for a more proactive microeconomic role for the government with regard to financial access.24 1.3 The Metries of Access: Challenges of Measurement 14. Ifaccess to financial services is to be systematically measuredor monitored and if the results of policy initiatives are to be tracked it is important to define standardized measuresand indicators of access. Conceptually, a series of alternative indicators may be conceived of, each with potential advantages and drawbacks, which could be applicable in different circumstances. Simpler measures, which can be constructed based on easily available statistics, and which could be usable for purposes of monitoring over time, or for cross-country comparison, have conceptual limitations in their interpretation. More sophisticated indices of access can require the collection of specialized data which would make them difficult to construct or to use for comparative purposes. Some alternative possible approaches are discussed below. 15. A first simple group of indicators of access is institutional presence - Le., the supply of financial institutions or service points, for the delivery of financial services. This could refer to a count of the numbers of different types of financial institutions (banks, non-banks, microfinance institutions, etc), or the number of branches, service posts, ATMs, etc., of such institutions. Statistics on such institutions are usually easily available, not only on an aggregated basis for the country as a whole, but also per geographic unit, for example at the level of individual states or municipalities. At least at the aggregate level, cross-country comparisons can be undertaken. At the level of geographic units within the country, such statistics can provide indicators of service availability by region or area. This i s the measure of access most extensively used in publisheddata inBrazil today. 16. One limitation of these measures is that in their simplest form they do not indicate the extent to which services provided meet demand for services, for example, in terms of numbers of inhabitants of a given area. Simple modifications can be introduced by presenting the indicators as ratios, eg, bank branches per unit of population, or per given geographic unit. Other limitations are that not all institutions provide the same levels of service; some may be larger or more comprehensive in terms of service provision. Another extension to the measure therefore i s the volume of deposits, credits, assets or net worth of financial institutions in specific areas - also already undertaken in some Brazilian statistics. More difficult to address is the limitation that the presence of a financial institution in a given area gives no indication of which types of clients it serves. Even in a relatively poor area of the country, a bank may choose to focus its services on the richer segments of the population. Another drawback is that with the change of technology today, institutional presence may become less important for the delivery of financial services, which can be undertaken remotely through the phone or web, without the need for physicalpresence. 17. More detailed supply side indicators of access attempt to look at volumes of services provided to particular categories of clients. Different types of financial services can be examined; deposits, credit as well as other transactions such as payments and money transmission, the provision of card facilities, etc. If the target groups of underservedpersons (low income / poor clients) could be identified directly, such indicators would be very valuable. However, it may be difficult for banks (and perhaps more so for less formal financial institutions) to have reasonable information on client income. As a proxy, the size of financial transactions i s often used as a proxy - Le., small deposits or small loans. But, this can clearly misrepresent client income, for obvious reasons. Such indicators could perhaps more easily be applied to identify services provided to disadvantaged communities, or groups with specific socio-economic characteristics, such as gender, color, migrant status, etc. 24 Some evidence inthis regard is presentedinPadillaand Requejo (2000) and Galindoand Micco (2001). Page /7 Brazii: Access to Financia/ Services 18. Measures of the extent to which the demand for financial services is met are more difficult to construct than supply side indicators. Not only is unfulfilled demand implicitly conditional on a specific service price, but it also depends on other conditions of service (such as distance, convenience, etc.,) as well as on the risk characteristics of the individual or enterprise. Broad proxies of measures of credit needs for enterprises, especially working capital credit needs, can however be constructed, based on turnover, raw material needs and other data. Credit needs for fixed capital, being ad hoc, are more difficult to proxy. Individual credit needs, or the needs of microenterprises which do not separatepersonal and business needs are harder to assess. Needsfor savings or deposits or other financial services also pose difficulties. In such cases, survey based indicators of individuals, households or enterprises, are a starting point. Undertaking such surveys in tandem with ongoing periodic statistical surveys such as census or other periodic surveys, or the Bank's household surveys, can help to provide data on socio-economic characteristics and a reliable sampling frame. The assessment of financial constraints is typically based partly on client perceptions as well as on information such as refusals to provide services, e.g., refusals to open an account or to provide credit. It should not be forgotten however that there may be valid reasons for such failures to obtain financial services, based on risk or cost. 19. Finally, broad-based measures of access could also be constructed based on the pricing of financial services. Clearly, the critical variable in this context is the interest rate. While interest rate levels would clearly affect overall volumes of access and hence financial depth, interest rate structures can focus more specifically on access issues. A first broad measure in this regard is the average spread between borrowing and lending rates, an issue which Brazil is keenly aware of." Disaggregations of spreads and examinations of the composition of spreads can also lead to the monitoring of trends in the individual subcomponents of spreads. Concerns about access could also be more specifically tracked by looking at the interest rate structure in greater detail; for example, prices of consumer credit, housing loans, rural credit; and also by examining the degree to which price discrimination may operate across different segments, and assessing the presence or absence of market segmentation. 20. Inpractice, measureswill have to be adapted to the end use for which they are intended. Simpler and more aggregate measures may be all that is feasible for cross-country comparisons, or even country wide or interregional comparisons, but more detailed measurescould be undertaken where institutional or user data permit. Thus, a series of `nested' measures could be constructed based on simpler uses (such as broad comparisons) or more sophisticated uses (targeted policy interventions). In the medium term, if access i s to be tracked systematically, it would be important for agencies such as the government to require the systematic provision of data on access indicators.26For the purposes of the present study, a series of different measures have been used, based largely on availability, with more aggregate supply side measures in the present chapter and more disaggregated supply measures at the level of the types of financial institutions in subsequent chapters. Demand side survey based data have been used extensively for individual consumersand in a more limited way for enterprises. 25 As studiedextensively in the three consecutiveCentral Bankreports on interest rate spreads "Juros e Spread Bancbrio no Brusil". A partialanalysisof some specific factors affectingspreads is undertakeninPart6 of this '`study.theUK In for example, bank credit to regions broadly indicatedas `deprived,' basedon a standardizedindex of deprivation,must now be documented(seePart 7 of this study). Part 1. Assessina Access 2. BASIC SUPPLY -SIDEMEASURES OF FINANCIALSERVICES OFFERED INBRAZIL 2.1 Trends and Distributionsof Bank Services: Branches and Service Paints 21. A first broad analysis of intertemporal as well as interregional or spatial aspects of the evolution of access to financial services, based largely on Brazil's Central Bank's data on access to financial institutions (mainly banks, their branches, and other points of service) reveals the following key points:" + Despite contractions in bank numbers, aggregate delivery of financial services overall has not declined over the last decade but has remained largely static. 3 There have been some expansions of new service delivery points, and growth of some institutions, including coops and microfinance firms, but these are quantitatively still nascent. 3 In terms of international comparisons, Brazil may be underbanked compared to developed countries in terms of bank branches but it has performed at least as well or better than its peers among developing or middle income countries. 3 Spatial inequalities in the delivery offinancial services are significant among Brazil's regions and have beenpersistent over time. But, much of this difference arisesfrom regional income differentials. + Underlying the relatively static overall picture in terms of financial service provision, there has been a significant shift in composition of service providers with a declining sharefor public banks. + Private banks, whilerelying extensivelyonnew delivery mechanismssuchasATMs, are also more branch intensive thanfederal banks. Municipalities with a larger share of agriculture and small firms are actually better served by private banks. 3 Econometric analyses of the determinants of bank service provision, using a range of indicators of access, find first that for all banks taken together, per capita income tends to be positively associated with broad measures of access such as the numbers or density of banking institutions or the values of the loan and deposit services they provide. Banks have an urban bias and tend not to provide services in rural areas. Nevertheless, the agricultural community appears not to be underserved. + Finally, analyses which distinguishbetweenpublic andprivate banksfind some differences in the factors associated with the provision of services of these two groups: by some measures public banks may provide more services to the poor, but private banks appear to offer more services to agriculture and to small enterprises. In many respects the behavior of the two is broadly similar; both tend to have a positive association with richer economic areas and both have some urban bias. Especially in thinly served areas, they tend to be substitutable, with either private or public services dominating. 22. First of all, at the broadest level, there has been a contraction in the number of banks over the last decade, in Brazil's bank-dominated financial system from 267 in 1993 to 194 a decade later, in 2002 (Appendix Table Al.1). This was accompanied by some decline in the number of bank branches, from 17,400 in 1994 to 16,000 five years later in 1998. These figures largely reflect the rationalization in the banking industry following inflation stabilization and consequent contraction in banks' float income, together with stricter prudential regulations on entry and operation, and later, the privatization of state banks (Appendix Table A1.2). 27Significant contributionsto this section havebeen madeby Thorsten Beck (2002). Page 9 Brazil: Access to Financial Services Figure 1.1 Brazil: Growthof Financial Institutions ( Nos., 1993-2002) AllBanks Leasing ",,,,,bers 8. FinancecompaniesBanks ha= declined in ...coops Credit Cooperatives the share of banks in financial assets h a w increased - 1,600 % shares ...But 100 remains stable 1,400 1,200 1 ._.---...._ .__.. _..- 1,000 lo 150. 1 200 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 1993 1994 1995 1996 1997 1998 1999 2000 2001 Source:Central Bank of Brazil.(See Appendix Table AI 1) 23. Data also suggest continued reliance on the banking system for service provision. There has been little diversification in the overall system of provision of financial services in Brazil over the past decade, in terms of asset values, with 92 percent of assets in the banking system in 1993, and 93 percent in 2001. A limited role is played by non-banks such as finance companies or leasing companies and this role has not shown any trend increase. Cooperatives have increased in absolute numbers from around 800 in 1993 to over 1300 in 2002, and in assets, but this has been too small in absolute terms to affect bank dominance (Appendix Table Al.1). 24. The apparent decline in bank services i s reduced if banking portfolios of multiple banks are counted - banking portfolios declined only from 334 to 296 between 1998 and 2001. And overall numbers of bank branches resumedgrowth after 1998, to reach 17,050 by 2002, so that over theperiod of a decade, branch numbers have been relatively stable (Appendix Table A1.2). Recovery of the numbers of branches may reflect specific conditions governing bank mergers and buyouts especially for state- owned banks, and increasedattention to income from credit operations. '' 25. The decline in bank branch numbers, moreover, has been accompanied by an expansion of new types of service points, especially towards automated teller machines. ATMs have increased from 3,500 to 22,500 between 1994 and 2002, principally through the expansion of individual networks (from some 2,800 to over 20,000) and some smaller expansion of group-operated networks (from around 570 to 2,100 over the same period). More recently there has also been a striking increase in the numbers of 'correspondent' bank arrangements to some 16,500 by 2002, almost as numerous as bank branches (Appendix Table A1.2). Apart from credit cooperatives, there has also been a growth in microfinance corporations (SCMs), where six were established in 1999, around 26 were inoperation at end 2002 and an estimated 36 were close to formal start up.29 28 Bank mergers or purchases, especially with regard to the privatization o f state banks under the PROES scheme, were frequently conditional on the maintenance of branch service provision. *')Statistics for microfinance entities do not include those outside the formal financial system (see Chapter 2). Page f0 Part 1. AssessinaAccess payment and deposi financial institution. private banks, nota 26. Looking beyond national aggregatesto the municipality level, the numbers of municipalities with no bank branch services increased significantly between 1994 and 1997, from 1,137 to 1,749. But thereafter, there was no further decline in municipalities without bank branches, with some reduction in this number to 1,665 by end-2002. Meanwhile the provision of banking service outposts (PAAs)grew and municipalities with no branchesbut with PAAs doubled between 1998 and 2002 from 334 to 654.30 Figure 1.2 Brazil Evolutionof FinancialService Outlets (1994-2002) - 25 - 2,ooo - /- 20 1 , 5 5 - 15 i 1 w - z d 10 5 5 5 - 0 1994 1995 1996 1997 1998 1999 2ooo 2001 2002 1 -Wncip 1994 19951996 1997 1998 19992ooo 2001 2002 w/rx)services(1) Wh PAA Source Central Bank of Brazil (See Appendix Table AI.2). 27. In all, the data suggest that although there has been no marked decline in access to financial services over the past decade, the overall picture i s relatively stagnant, barring some qualitatively significant new institutional segments (cooperatives and microcredit firms) and quantitatively more important development of new delivery channels. These positive signs have accelerated since 1999. 30Meanwhile, there was an increase in the numbers o f municipalities, from 5,011 in 1994 to 5,597 by 1998 and further to 5,658 by 2002. Brazil: Access to Financial Services 2.2 Ownership: Public versus Private Provisionof Bank Services 28. Behind the aggregate numbers, there have been significant changes in ownership composition of bank branches and other service points, with a huge decline in the proportion of public bank branches, which is largely responsible for the faltering overall numbers of branch services. Branches of public banks fell from some 9,000 in 1995 to around 6,700 in 2001 (Appendix Table A1.3). Meanwhile, branches of private and foreign banks almost doubled from some 5,400 to over 10,000 over the same period. Public banks rely to a greater extent on branches for delivery; ATMs accounted for only 31 percent of their outlets at the end of 2001, in contrast to almost 50 percent for private domestic banks (including banks with minority foreign ownership). 29. An analysis of differences in balance sheet structure of banks by ownership category - federal, state, private and foreign - also reveals that private domestic banks have less assets, loans and deposits per branch than federal banks, i.e., private banks have larger branch networks, relative to their size, compared to public banks. Federal bank branches appear to be larger, as measured by the numbers of employees. Figure 1.3 Brazil Ownershipof Bank Service Points - Public bank branches declinedwhile private and Private domestic banks are relying more on foreign bank branchesgrew... , ... 10 AlMs than branches 1 49 6 - 7 i;1 6 - Note, Private domestic banks include those with foreign participation. Numberscorrespondto December 2001 data unless stated otherwise. Source: Central Bank of Brazil, 30. These indicate that private banks have made great strides in client outreach, using delivery systems which are lower cost and use newer technology (see Part 3). Wholly foreign owned banks however had the lowest proportion of ATMs and the highest branch reliance, reflecting possibly their concentration in the highnet worth segment.31 2.3 Regional Distributionof FinancialServices 31. Undertaking a spatial analysis of financial access next, it is clear that there are significant variations in service provision across different regions of the country. Access to financial services is unevenly distributed geographically, with sparse services in the poorer parts of the country. Banking 3'These changes in composition are also associated with changes in the delivery of different types of financial services. Public banks although with a low share in banking system net worth (a third in 1996), accounted for almost 60 percent of credit operations and over 60 percent of deposits in that year. By December 2001, the public banks' share of system net worth had declined to 16 percent and its share in credits to 25 percent, but it still accounted for 43 percent of all deposits (Appendix Table A1.4). Foreign banks' shares in credit and net worth grew from around 10percent to some 31percent over the same period. Page 12 Part 1. Assessina Access facilities are concentrated in large southeastern states. Around 57 percent of all municipalities in the Northeast and inthe Northand nearly 20 percent of the populationinthese states have no access to a bank branch. Nationwide, 30 percent of all municipalities and 6 percent of the population, have no access to formal financial service^.^' There are seven states (out of 27), mostly in the North, where the share of municipalities with no bank branch is over 70 percent. About 1,665 poorer and arguably less populated municipalities, out of a total of some 5,660 municipalities, have not a single bank branch. 32. Associating levels of financial services supplied with regional population, area and GDP or per capita income reveals some interesting findings. Even when bank branches are present, services are more constrained in the Northand Northeast. The population per bank branch was nearly three times as highin the North as in the South of Brazil in 1996, and over thrice as high by 2000. Between 1996 and 2000, branch numbers declined in the North and Northeast, while they grew in the Southeast. However, branch numbers had also declined in the South and Center, in the first period, and branch numbers in all regions improved by 2002. But considering regional GDP differentials, the variation decreases substantially. GDP per bank branch in the North i s less than twice as high as in South, the region with the lowest GDP per bank branch. The smaller number of branches in the North and Northeast therefore at least partly reflect lower income levels and hence less potential business. Table 1.3 Branch Densitv across Brazilian Reaions North 3,907 90.5 87.6 17,501 23,161 645 557 580 6,644 3,O 2,327 Southeast 8,774 71.1 68.4 8,090 8,098 8,281 8,942 9,361 99 South 3,450 167 Central West 1,262 1,273 Source:Branch data, Central Bank of Brazil. GDP data. IPEA, population and area data: IBGE 33. The relation between average per capita income at the state level, and the existence of bank branches, is further illustrated, in Figure 1.4 at the level of individual provinces. As the first part of the figure shows, there is a broad positive association between bank services and GDP per capita, whether measured in terms of bank numbers, branch density per capita or branch density per unit geographic area. 34. Some states which appear to be outliers by one measure conform according to other measures. For example, although Sao Paulo appears to be more intensively branched than predicted by income, if branch density per numbers of persons is the measure of access this i s corrected; Le., the greater population density of Sao Paulo helps to explain the proportionally higher number of branches than expected on the basis of its average income. If branch density is measured in terms of geographical area, the positive relation overall between per capita GDP and branch density remains. However, in this case the state of Rio de Janeiro is a positive outlier; i.e., services are high, relative to its small size. 32Not includingcorrespondents ( Appendix Table A1.2). Page '? Brazil: Access to Financial Services Figure 1.4 Brazil Branch Density by State - 1.Bank Branch(NOS.) 6,500 ; 5,500 - 20 Paul0 4,500 3,500 -i RIO Grandedo Norte paran rd 1 2. BranchDensity(per 100,000 persons) Rorama 85 70 0Espirito 55 Santo 7 SantaCatarina Rio deJanero 3. BranchDensity(per 1,000km2) 1 40 0EspiritoSanto Distrito Feder Rio deJanecro "0 35 / PI Rio Grandedo SUI Marmha --, , -5 Acre VY/IZIRoi I 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 x),OOO 11,000 P,OOO y3,OOO 14,000 15,000 GDP per capita Note: The size of each point represents State GDP. Source: IBGE. Page 74 Part 1. Assessing Access 35. A few states such as Rio Grande del Norte (NE)or Espirito Santo (SE) remain positive outliers by all of these measures- that is, bank services provided are greater than predicted by per capita incomes by all measures. B y contrast, some others have less than expected services, even after factoring income differentials, by all these measures. They include for example, Acre, Amazonas and RondBnia (N), Bahia (NE),Minas Gerais (SE) andRioGrandedo Sul(S). What is interesting here is that some of the better off provinces are also included among states where services are lower than may be expected on the basis of overall economic size and population. Many states do better than expected on some measures of density and worse by others. Thus, both Mato Grosso and Mato Grosso do Sul (both CW), although apparently underserved according to geographical measures of branch density (per 000 square km) are not underserved if their relatively thin populations are taken into account (branch density per 100,000 persons). 36. However, these diagrams also suggest that for some of the northern and northeastern states which are clustered in the lower left quadrant, a few have lower levels of access than expected on the basis of population density, even once income is factored in. Those states which appear underserved on the basis of population per 100,000 persons include Alagoas, Bahia, Cearb, Maranhlo, Paraiba, Pemambuco and RondBnia. Thus although regional branch service differentials seem to be significantly related to income levels, access i s poorer than the income analysis alone would suggest in these areas. 37. Finally, the state-level analysis also shows that there i s a negative association between per capita GDP and the number of municipalities per state without service. Figure 1.5 Brazil Municipalitieswith no services - %share of Municiwlities with no Services Piauio 0Tocantinso Rio Grandedo Norte Paraibao Ro;am O A w a 65 Acre * 0 - Amonas 0 Maranhso Distrito Federal 35 - 20 - 5 - -x) 4 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 x),OOO 11,000 P,OOO l3,OOO 14,000 15,000 GDP per capita Note.The size of each point represents State GDP. Source: IBGE. 38. If instead of looking at numbers of institutions, we look at the volumes of services provided in terms of credits and deposits, and adjust these for income differentials, these may provide a more representative measure of financial access. Results of such measures regarding regional variations in the provision of financial services still indicate substantial differences, and the magnitude of variation between regions is especially pronounced again for the North and Northeast especially compared to the southeast (Table 1.4). However, the variance in per capita credits and deposits is much larger than in creditddeposits relative to GDP. Thus, per capita differentials in regional access to financial services are thus not only associated with but also greater than per capita income differentials. This could be a Brazil: Access io Financial Services reflection of prevailing inequalities in income distribution, so that highmean incomes could nevertheless coexist with large proportions of the regionalpopulationat a much lower median income. Table 1.4 Provisionof Bank Servicesacross Reaions North 3,907 21.39 6.19 820 238 Northeast 19.69 24 14 500 613 Southe South 7,692 Centra 6,559 Note Br ded from the Central West region. Source: Banco Central, IPEA and IBGE. 39. Such income differentials are more apparent looking at smaller geographic units, and thus not surprisingly, variations in the provision of bank services across municipalities is higher still (Table 1.5). Credits and deposits as a share of municipal GDP range from zero to over 1,000 percent. Similarly, credits and deposits per capita range from zero to R$85,000 (for credits) and R$45,000 for deposits. The large number of municipalities without any bank services explains why the average for the whole sample i s so much lower than the aggregate for Brazil. Source Estimatedfrom data from the Banco Central. 40. Finally, combining the examination of the regional provision of services with the nature of bank ownership, it is seen that the provision of bank services across regions varies significantly between private and public banks (Table 1.6). While public banks have dominated in the North and Northeast, private banks dominated in the Southeast. This could reflect the profit orientation of private and the "social" orientation of public banks. Data at the municipal level suggests some complementarity in the branch structure of public and private banks. In 1996, 1,603 municipalities had just one branch, either private or public, while 428 had two, three or four branches, entirely private or entirely public. In 2000, 1,393 municipalities had just one branch, either private or public, while 397 had two, three or four branches, entirely private or entirely public. This also suggests that there has been a pattern of segmented delivery of private and public banks, with limited competition between these segments in the more thinly served areas. Page f6 Part 1. Assessing Access Table 1.6 Public and PrivateProvisionof Bank Servicesacross Regions Note:Brasiliais excludedfrom the CentralWest reqion. IS Source: Banco Central, IPEA and IBGE 2.4 International Comparisonsof Bank Services 41. It is clear that compared to developed economies, Brazilian bank branches serve a much higher number of inhabitants per branch. While an averagebank branch in Germany serves a population of 1,479 people and an American branch serves an area with 3,568 people on average, Brazilian branches served an average population of 9,331 people in 1996 and 10,356 in 2000. Figure 1.6 Braziland Other Countries PopulationServed per Bank Branch - Population per branch, a Regional Comparison Population per branch, an International Comparison 16,000 7 14.888 18.000 17 50 I 17,459 14,000 16 000 14,000 12,000 12,000 10,000 10,000 8,071 8,000 8,000 6,835 6,000 6.000 4.000 4,000 2,000 2,000 0 North NortheaSt Southeast South Central 0 West Brazil CololibiaMexico Chile India Indonesia USA Japan G e r m y Frarce Source World Bank staff estimates. 42. But comparisons with other developing countries suggest that Brazil is not obviously under- branched (Table 1.7). Brazil has more bank branches relative to population than other Latin American countries and India and Indonesia, two large developing countries. Measured relative to GDP, its bank branch intensity is lower than Colombia, India and Indonesia, which are all poorer than Brazil, but the two latter are much more densely populated. Branch intensity relative to GDP in Brazil is higher than in Mexico or Chile, although both are richer than Brazil. Additionally, as alluded to earlier (see page 3 above), the overall proportion of bank account holders to population in Brazil of at least a third of total population (implying a much higher proportion of banked families), i s respectable, compared to statistics on other emerging and developed countries (e.g., 13 percent of persons in the USA, 15 percent of households in the UK,and a third of all households inHungary). 43. Measured according to average geographic area covered per branch, Brazil appears at first to be under-banked relative to all other comparator developing countries except for Chile. However, this is easily explained by the low population density of Brazil and the vast areas that are not populated. An econometric estimation of predicted branch intensity compared to actual intensity for Brazil and other countries, correcting for differences in branch intensity explained by differences in GDP, total population, and geographic size finds that Brazil is the only developing economy that i s not under-banked by this standard, since its actual branch intensity is higher than predicted. Page r: Brazii:Access lo Financial Services Table 1.7 Brazil - Bank Branch Density: InternationalComparisons India 448 14,888 6.7 44 -0.00 Indonesia u s 683 12,547 8.6 110 -0.07 33 087 3 568 118.0 117 -0.17 Japan 67.3 6 0.22 Germany 25,724 1,479 38.1 6 0.86 France 24,434 2,331 57.0 22 0.14 Note. Residual branchintensityis the deviation of actual from predicted numberof branches, derived from a regression of the numberof branches on GDP, land area, and populationfor 28 countries Positive numbers indicate that actual branch number exceed predictedvalues Sources:World Bank staff estimates, basedon Hawkinsand Mihaljek (2001)and World Development Indicator data. 2.5 Econometric Analyses of Factors Associated with Bank Service Supply 44. What explains banks' decisions to locate their services in some municipalities rather than others? Attempts were undertaken to associate three different supply side measures of access to banking services (including first, the presence or absence of a bank branch; second, branches per million inhabitants; and third, credits or deposits as a ratio of GDP) with economic characteristics including income per capita, overall economic size of the region in which banks operate, in terms of GDP per capita, and locational characteristics: population density, shares of rural and agricultural population, and the share of small firms to GDP (Table 1.8). 45. For all measures of access used here, except the existence of a bank branch, per capita income was positively associated with access. Thus, banks try to locate in regions where people are better Municipalities with higher overall GDP also tend to have better bank access.34There is also a clear result, across all measures, that areas with lower rural populations are more likely to have a bank branch. These results, indicating urban bias in the provision of bank services, are very significant. However, high population density i s negatively associated with branch existence and branch density, and positively associated only with the provision of deposit services. This could suggest that some more densely populated areas, such as those which are urban and poor are not particularly likely to attract bank services. The share of small firms in a municipality is positively associated with the establishment and density of bank branches, but there i s no clear relation with volumes of credit or deposits. 33I t is possible that there have been policy driven pressures to maintain at least one service outpost in each municipality, and these pressuresdominate other considerations. Also on economic grounds, the presence or absence of a branch is linked to overall GDP but that branch density i s linked to per capita income may nor be surprising. An area may need to have a certain minimum critical economic mass (as represented by GDP) in terms of client potential, to make i t worth installing a bank branch there, but decisions to further increase bank presence may depend more on average incomes, and also, as discussed further below, on the entire pattern of income distribution. 34Access measuredby branch density is an exception; but here overall association is weak. Page 18 Part 1. Assessing Access Table 1.8 Brazil What Explains Bank Branch Servicesacross Municipalities? (All Banks) - GDP(R$ million) Agricultural share 0.00 (0.93) 0.15 (0.52) 0.08 (0.16) -0.03 (0.46) Small firm share -0.01 (0.93) -0.01 (0.91) No. of observ 4,958 58 4,958 Note: Existenceof a branch is a dummyvariable equal to one if a municipalityhas a branchand zero otherwise. Branch density is the number of branches per million persons The regressionscontrolfor randomstate-levelerror terms P-values in parentheses indicate significance; .Ol>p>O indicates significance at the 1%level; .05>p>.01indicatesresultssignificant at the 5% level. Higher p values indicate weaker association Sources:IBGE, IPEA, Central Bank of Brazil. 46. Finally, the share of work force in agriculture does not show any significant relation with the level of bank services, in contrast to the negative association for rural indicators. This may reflect the series of policy incentive schemes offered for agricultural lending, discussed further in Parts 3 and 5 of this study. These results indicate that while rural areas are indeed undersewed, agriculture is not a sector that is obviously neglected by thefinancial sector. 47. Next, going beyond the aggregate measures of financial service availability examined above, further analyses were undertaken to separately examine the factors affecting the provision of financial services by private versus public banks to see if bank ownership affects the findings regarding the determinants of access (Table 1.9). 48. First, looking at the role of per capita income, all measures of private bank services reflect the aggregate finding of a positive association with per capita income. These results are highly significant except for bank branch existence, where the relation though also positive i s weaker. However, for public banks, there is no clear cut relation to per capita income. Three measures (existence of a branch, volumes of credit and volumes of deposits) indicate a negative relation, but this is significant only for the first of these measures. And for bank branchdensity, public banks also indicate a positive relation with per capita income, which i s significant. Second, in terms of associations with GDP the behavior of the two groups, public and private, is similar inmany respects; both have the same signs and reflect the generally positive association with GDP negative association of the aggregate group, for all measures except branch density (which are not very significant). Third, the negative association with population density i s similar for the first two measures, though surprisingly in terms of credit, private banks indicate a greater propensity to lendin more densely populated areas.35 35 As we shall see inPart 3, this is not a surprising finding; private banks have a large involvement in the small loan market, which could be associated with densely populated and poorer urban areas. Page 8 5raziL Access to Finaitciai Services Table 1.9 Brazil What Explains Publicversus Private Bank Branch Services across Municipalities? - Public Banks Income/cap. (R$ 000) -0.10 (0.00) 6.81 (0.00) -0.08 (0.64) -004 (0 72) -0.03 (0.00) Small firm share Existenceof public bank branch -0.40 (0.00) Note: Existenceof a branch is a dummy variable that takes on the value one if a municipality has a branch and zero otherwise. Branchdensity is the numberof branches per million persons, Brasilia is excluded from the regressions. However,the results do not change when including it. P-valuesare reported in parentheses Sources:Central Bank of Brazil, IPEAand IBGE. 49. The urban bias of both groups is similar, on all measures: the share of the population which is rural is negatively associated with bank access measures.The next finding, on the share of the population in agriculture, is surprising; for all measures public bank services are negatively associated with the agricultural share, while private banks services are positively associated with this. Measures are significant for branch density and for the share of credit. As discussed for the aggregate data, this may reflect policies requiring agricultural lending, at least in respect of credit. Results on smallfirm shares are positive for private banks, unlike public banks, for the first two measures of access, though not significant for measures related to credits and deposits. There are thus indications that municipalities with a larger share of agriculture and smallfirms are better served by private banks. 50. The likelihood that a municipality has a public bank branch i s negatively correlated with the existence of a private bank branch, and also the converse. This suggests some substitutability, and i s driven by the fact that nearly half of all municipalities with a bank branch have only public or only private bank branches. Public and private banks may compete for services as they operate broadly as substitutes in many areas. 51. As a final measure to better capture differences between private and public banks, the analysis defines two categories of municipalities, which are served with only private and municipalities with only public bank branches. Results reveal that municipalities with only public bank branches are signijkantly Page I!O Pari 1. Assessing Access poorer, in per capita income terms, but they also have a significantly lower share of agricultural work force and small firms.j6Average per capital income for those municipalities with only public branches was R$2,771, compared to R$4,060 for those with only private branches. But, the agricultural share of the group with only public branches was 3.8 percent compared to 5.1 percent for those with only private branches. 52. Summarizing, many characteristics of service provision by public and private banks are not very different, including the urban bias of both groups and also the broadly positive association with GDP. The finding of some degree of substitutability in their service provision is not surprising. But, there are also some interesting differences in the factors behind public and private provision of bank services: Public banks appear by some measuresto be more likely to serve the poor than private banks. However, they do not show the economic structure expected with a poverty focus - they are not more agriculturally oriented and are not associated with greater services to small firms. On some measures, private banks better serve municipalities with a larger share of agricultural and small firm activities. 3. DEMAND SIDEMEASURES OF ACCESS EXTRACT - FROMAN ENTERPRISE SURVEY 53. To what extent do Brazil's enterprises perceive the availability of finance to be a constraint? Data from the World Business Environment Survey (WBES) reports that Brazilian firms rate the high interest rates as the largest financing constraint, impacting upon operations and working capital needs, followed by collateral requirements and access to long-term loans (Table 1.10). Banks' access to loanable funds and access to lease finance are rated as relatively minor obstacles. While Brazilian enterprises perceive the financing of investment and access to long-term loans, to be no more constrained than entrepreneurs in other Latin American and upper-middle income countries, they rate high interest rates and collateral requirements as greater obstacles than enterprises in comparator countries. This might reflect the availability of long term investment funds from state sources such as the development bank BNDES, or deficiencies in the legal system and the short maturity of external debt financing in Table 1.10 Financing Constraints in Brazil in International Comparisons Brazil 2.69 3.71 2.91 2.90 1.81 1.71 2.35 2.65 2 25 1.80 1.59 1.68 Note:The ratings are as follows: 1 = no obstacle, 2 = minor obstacle, 3 = moderate obstacle, 4 = major obstacle, Numbers for individual countriesare averages and numbers for country groups are medians, due to the small numberof observations Source:World Business EnvironmentSurvey. World Bank (2001). 54. Brazilian firms of different size groups do not differ significantly in their assessment of financial constraints. Thus, unlike firms in other countries, small firms appear to face relatively lower financial constraints. There are however significant differences in overall financing patterns across firms of different sizes in Brazil. Small firms receive significantly less financing from equity or from development 36At the 1percent level, usingchi squared tests of differences. 37Describedby Beck (2000)"Impediments to the Developmentand Efficiency of Financial Intermediationin Brazil", Policy ResearchWorking Paper 2382, World Bank. Page 2i Brazil:Access to Financial Services institutions and more from retained eamings and informal sources than medium-sizedfirms. For firms of all sizes, highinterest rates are the largest constraint. 55. Comparing actual financing patterns across countries and firms, it i s found that Brazil on average relies to a lesser extent on retained earnings than other Latin American or upper-middle income countries, and especially relative to other middle income countries, more on supplier credit and bank finance. External equity financing i s also more important inBrazil. 4. DEMAND SIDEMEASURES OF ACCESS SURVEY OF URBANINDIVIDUALS - 56. Aggregate measures of the supply of financial services do not provide an answer to the question of who has access. Even if a financial institution exists in a given location, it is not clear without more disaggregated data whom the clients of such an institution may be. Therefore to complement the analysis of data on financial institutions, and on the financial constraints on enterprises, a survey was launched to examine the financial behavior of individual consumers.38 The aims of the survey were (i) to provide information on individual consumer's present patterns of access to financial services and potential additional demand; (ii) to explore perceptions of constraints to access; (iii)to examine the factors which are correlated with and could explain access, and in this context to try to identify policy directions which could be used to expand access. Broadly, the questions addressed are - what financial services do households want, what do they receive, how does this vary by income or poverty levels?What factors can predict the access to and use of financial services by individuals in different socio economic groups and how could this be useful in shaping public policy towards access? 57. The survey covered 2,000 urban residents in 9 metropolitan areas and 2 cities across 11 states. The unit interviewed was chosen to be any household member of 18or above.39 Around 55 percent of the surveyed population was in the two largest cities of SBo Paulo and Rio de Janeiro together; the remainder was distributed across small and medium cities in all geographical regions. In terms of regional distribution, the largest share is in the southeast - 61 percent (SI0 Paulo, Rio de Janeiro and Belo Horizonte). There is substantial coverage in Northeast (17 percent) and other regions (Center-West, North and South) account for 21 percent taken t~gether.~'The principal reason why the sample covers urban areas only is that an estimated 80 percent of Brazil's population lives in urban areas. Practical reasons were another consideration. 38This part of the study was undertakeninpartnership with a team of personsfrom IPEA and Universityof Niteroi in Brazil, and a group of Brazilian advisors.Data were collected and tabulated by the firm Sensusof Belo Horizonte. 39The individual rather than the household was chosen due to a combinationof reasons; financial characteristicsare not common to all householdmembers, there are variations across the age spectrum interms of financialaccess as well as financial needs. Some questionsespecially those relatingto credit depend on individual characteristics.See KinnonScott (Grosh, M.& Glewwe, P., 2000, chapter 21) on the reasons for selecting the individual as the point of data collection. 40The cities are: Goisnia, Brasilia,Porto Alegre, Curitiba, ,520 Paulo, Rio de Janeiro, Belo Horizonte,Salvador, Recife, FortalezaandBelCm. The sampleof cities, which includes locationsineach regionand cities of a rangeof sizes, follows the selection usedfor the periodic consumer expenditure surveys (POF) of IBGE, Brazil's statistical institute, and is basedon the IBGE Census 2000 results for the distributionof populationby city. The advantageof this procedurewas that a ready made existingsampling frame was available and known to be representative.The sample was constructedon a probabilisticbasisfor census sectors and for domiciles within the sector, with a requirementof a minimumof ten interviewsper city, and quotas for sex and age. In designingthe questionnaire, efforts were madeto build on financial data collectedin previous surveys in Brazil (in particular the PPV -pesquisa sobre padr6es da vida - or household survey;) and also to draw uponongoingparallelefforts underway in the Bank for Mexicoand Colombia. Additional details of sample construction, frequencies, or other sample resultsare availablefrom the task team at the World Bank. Page 22 Part 1. Assessinq Access 58. The questionnaire first identifiedthe key socio economic characteristics of each interviewee, and then attempted to map the interviewee's access to specific financial services, `unbundling' access in the following broad categories: (i) general access to a formal financial institution, (ii) to savings and access deposit services, (iii)access to credit, and finally, (iv) access to payments services.41Under eachcategory, access i s defined in terms of a series of alternative parameters. For example, access to a financial institution can refer to whether or not a person has an account with such an institution or has made use of it. It can also be described in terms of physical access; i.e., the distance or time taken to reach the financial institution used. The socio-economic characteristics of the surveyed population are summarized in the box below. The analysis of findings with regard to financial behavior is discussed below, inthree broad stages. First, we discuss the frequencies of responses to individual questions on financial access under the four broad groups of financial services discussed above. Second, responses are correlated with socio-economic characteristics of the respondents and we test the significance of apparent differences. Third (section 6), we examine the determinants of access to different forms of financial services and investigate the relative importance of the different factors at work. and around half were married by age 25. Forty month. Almost two-th respondents), and a th work certificate comp Household size householdsof 5 lived inrented homes. 41Other forms of `unbundling' could be used. For example, the typology usedinarecent study, conductedby the National Community InvestmentFund (NCIF) in the USA (2002; "Innovative Productsand Services for Low- Incomeand UnbankedCustomers") distinguishesbetween `jirst-order' or prior needs to help financial access (adequate income and minimal financial literacy); immediate financial needs, includingliquidity, money orders and bill payments, remittances(wire transfers),short-term credit during emergencies; and long-term financial needs, including savings, access to borrowing, homeownership, insuranceand investment. Page23 Brazil: Access to FinancialServices 4.1 Survey Results Access to Financial Institutions - 59. Forty three percent of the individuals interviewed had a bank account. This statistic is surprisingly close to estimates made by Unibanco, one of Brazil's prominent banks, which estimated the `banked' at 42.8 percent.42However, a much larger number of persons made use of financial institutions. Perhaps an unusual feature inBrazil, compared to elsewhere, is that for many if not most persons, contact with a financial institution consists of the use of payment outlets, whose establishment may be bank sponsored, but where it is not necessary to be an account holder to use select services offered. Such services have been dominated by the Casu Lote`rica or lottery shop chain of the Caixa EconGmia Federal, although other banks are now beginning to also offer such services. For almost half the respondents (47 percent), these correspondent outlets are deemed to be the most important financial institution. Services provided at the lote`ricas vary; most prominently, these have offered bill paying services for utilities and deposit services - and such services are offered without the need to open an account. Thus using this limited form of correspondent financial institution for making payments does not signify that an individual is a bank account holder. Not including these correspondents,, 46 percent of the sample claimed that banks were the most important financial institution for them.43Surprisingly, aside from the lote`ricas, the majority of respondents (58 percent) for whom banks were the primary financial institution use private banks. The use of non-bank financial institutions such as cooperatives / credit unions or microfinance was very limited (only 4 percent of responses). Combining these (and excluding Caixa's bill payment points) only around half the surveyed persons used any form of financial institution. 60. An attempt was made to see if physical distance has limited access; are banks far from their clients? But more than half of the individuals were able to reach the financial institution on foot (53 percent); a third uses public transport and only 12 percent use a private vehicle. More than a third of respondents (37 percent) took less than 10 minutes to arrive at their financial institution and almost three- quarters (73.5 percent) took less than 20 minutes. The relative unimportance of distance may be a reflection of the urban nature of the survey, or a confirmation that branch density at least in these urban areas, is not an important issue. Table 1.11 Indicator of Access :Access to FinancialInstitutions Use banks as their primary financial institution1 aa7 46 Use CEF lottery shops as their primary financial institution 909 47 Use mainly non-bankfinancial institutions2 67 4 Are able to reach the financial institution they use on foot 1014 53 Notes: This could include family members of account holderswho do not hold accountsthemselves. 1 2 Includes Credit cooperatives, micro finance institutions, and others. Source: World Bank, Survey of Access to FinancialServices in UrbanAreas of Brazil, 2002. ~~~ 42 Unibanco "Obstacles to IncreasingAccess" Presentation at World Bank seminar on Access to Financial Services. February 2003. However, the parameters of the surveyed population in the Unibanco sample are not known. The figure is also highly plausible in view of aggregate statistics on bank accounts in Brazil which amount to at least a third of the population. The present sample moreover only covers adults, in major cities; thus a higher mean ratio would be expected. 43 A second question was included as a cross-check. I t is however also possible that some persons who are intensive users of banks use the accounts of family members without setting up their own account. Part 1. Assessing Access 61. To what extent do the figures above indicate unmet demand for financial services? Around two thirds (64 percent) of those who did not have an account indicated an interest in having one. However, this does not indicate effective demand for bank services, as in many cases, there could be reasons why persons may be ineligible for an account, due to risk characteristics or due to bureaucratic requirements, bias or other factors. Possible reasons for not having an account were investigated grouped in two broad categories, based on voluntary factors (referred to as `difficulties due to banks') and involuntary factors (`difficulties due to respondent characteristics'). Among the voluntary factors, highfees clearly dominate -bank servicesareexpensivefor manypersons.Inparallel, individualscitedthattheirownlackoffunds was also an important consideration. References to factors such as lack of documentation and references and difficulties in opening an account suggest that bureaucratic requirements are also important. By contrast, factors such as hours of operation or distance, which suggest physical inconvenience, were not important. 62. Difficulties related to the costs of banking services or high threshold levels are also reflected in the responsesto a question addressedto those who had had an account and then closed it (629 cases). The main reason for cancellation was the difficulty of maintaining a minimumbalance (218 cases) combined with cancellation by the bank (also due largely to problems in maintaining the minimum balance) and high fees. These three factors combined amounted to 54 percent of responses.4 4.2 Survey Results Depositsand Savings Behavior - 63. Highlights of deposit behavior to emerge were that (i)sight deposits and savings accounts account for the bulk of most respondents' recent deposits; 43 percent of persons making a deposit in the past month used a sight deposit while 41 percent used savings deposits, as their primary vehicle. Less than 2 percent of respondents put money in term depositskertificates of deposit, less than 4 percent used rotating savings schemes (consortia), and less than 1 percent bought or sold shares. Around 10 percent made `other' investments. (ii)More than half the respondents (54 percent) have their largest deposits in private institutions, while 41 percent used public institutions. (iii) financial institutions, including Other cooperatives, accounted for less than 2 percent of responses, emphasizing the limited role of financial institutions other than banks for deposit taking, at least in urban areas. (iv) There is still client preference for branch bank use, despite technological change; 60 percent of respondents made their recent deposits at bank branches while only 30 percent used ATMs for deposits, while 10 percent used bank service posts(Figure 1.7). 64. Further, as discussed later in Part 3, (iv) the most important reason cited for making deposits was `security', especially for current accounts (215 out of 305 valid cases) and special savings accounts (259 out of 354 valid responses), but also, surprisingly, for term deposits. Respondents also acknowledged the lower transactions costs of current accounts (19 out of 305 cases) and special savings accounts (24 out of 354 cases), and for current accounts, the value of access to other financial services (30 out of 305 cases). Rates of return appeared third in order of importance, suggesting at an aggregate level, a weak association between use of deposit services and the interest rates paid for such services. It also appeared that when decisions were made to withdraw funds, it was not typically associated with diminishing returns or increasedrisk but for liquidity needs (92 percent) (v) Most account holders have limited accounts; only 61 percent of account holders had access to checking facilities. Finally, (vi) other forms of savings and investments, in non-bank assets, were low. Only 147 persons made voluntary contributions to pension funds (7.5 percent of respondents), and most were public (71 percent). 44Discussed in greater detail in Part 3 of this study. Around 40 percent of respondents cited `other' reasons for not having accounts. Bwzii: Access to FinancialServices Figure 1.7 Depositand Saving behavior-Survey of Urban Individuals Where do you havethe largestamount deposited? "Havevou depositedmonev in one of the followina inthe _ _ __ Bank branch ~~ ~ _.- Rivate Fublic Famlyi Coop/ Others institution institution Friends Credit Union Valid cases = 602 Source:World Bank, Surveyof Access to FinancialServicesin UrbanAreas of Brazil,2002. 4.3 Survey Results Loans and Credit Services - 65. Only 15 percent of respondentsclaimed to have applied for credit in the past 12 months, and only two-thirds of those respondents had their applications accepted. The use of guarantees or collateral in successful applications was low; only 12 percent (24 persons out of 194 successful loan applicant^).^' In terms of sources, a third of respondents usedprivate institutions as their primary source of credit, while a quarter (26 percent) used public institutions and a similar proportion (24 percent) used family and friends. Finance companies accounted for the next big group of creditors with nearly 10 percent, while credit unions and shops only provided for loans to for 1.5 percent of creditors respectively. The reported incidence of pawnbrokers was surprisingly low; only 1percent of accepted loans. 66. In terms of uses of credit, the majority of the respondentsapplied for a loan for personal purposes (160 persons or 83 percent of respondents).Only 14 borrowers (7.3 percent of respondents) cited business reasons, while the rest had 'other' reasons. The largest single use of loans (69 cases or 44 percent of respondents) was for family emergencies. Another 15 percent of respondents used the loan to purchase or renovate their homes, and a slightly smaller proportion, 13 percent used the loan for buying household goods.46Unexpectedly, loans for automobile purchase were very low (only 3 percent, perhaps because of the practice of leasing), and loans for education are also low, suggested a limited borrowing on a 'life cycle' hypothesis basis. However borrowing to smooth unexpected consumption needs appears to be common. Among business financing needs, working capital predominates (6 cases out of 12) although payments to creditors (2 cases) and routine maintenance as well as unforeseen business needs are also cited (1 case each). When not using formal credit, almost half of the respondents (49 percent) use loans from family and friends (35 percent from families and 14 percent from friends). One-fifth (20 percent) use private savings and around a third (31 percent) use other sources. And 15 percent of respondents had made a loan to their families inthe last 12months. 45Personal guarantees were the most common (6 cases), followed by business inventory/ bank accounts and institutional guarantees(3 cases each). 46At a different point in the questionnaire, it was asked how real estate had been financed. The response was that only 37 percent of respondents used a financial institution to obtain a loan for real estatepurchase and renovation Page ZE Pari 1. Assessina Access 67. Average loan amounts were low; 50 percent of respondents applied for R$500 or less; 74 percent applied for loan amounts of R$1,000 or less. Amounts approved very closely follow amounts requested, suggesting that in cases of doubtful applicants, credit is rationed by screening out the applicant, rather than by reducing the volume of credit extended. On the status of loans received in the last 12 months, a quarter were paid off, suggesting short maturitie~.~'Over 62 percent were up to date, while around 12 percent of respondents admitted to some late payments or to a renegotiation of the loan. Loan processing i s surprisingly rapid - more than half received a responseabout their loan application on the same day (58 percent), another 26 percent between 1-5 days. 68. Finally, apart from formal loans, around a quarter of persons had access to a debit card and a fifth had a credit card. (Table 1.12). Of these, more than half of respondents (58 percent) had credit card limits of R$600 or less (this is around twice the mean monthly income). Forty four percent of respondents had credit card debts of R$200 or less outstanding; another 42 percent had credit card debts between R$200- R$600. Table 1.12 Access to Loansand Credits R re Applied for a loan but didn't receiveit Did not apply for loans Received credit primarilyfrom: Private institutions 68 33 Public Received a businessIo Source:World Bank, Survey of Access to Financial Services in UrbanAreas of Brazil, 2002 69. What difficulties did the respondents face in obtaining credit? First, does the low proportion of applicants (only 15 percent) point towards low probabilities of obtaining credit or low credit needs? According to survey responses, most respondents who did not reply for a loan (70 percent) claimed that they did not need the money. However, 17 percent responded that they assumed the response would be negative. The balance did not know how to apply (4 percent), or had other reasons (13 percent). 70. Second, what reasons were offered to those who had applied for credit but had been refused? Respondents with loan application rejected cited not having a steady income, not having any guarantees and not having enough earnings as the primary reasons (Table 1.13); however the responses to this question were limited in number and these results can only be considered suggestive. 47Information on loan maturity suggests that 78 percent o f loans were for a period o f 12months or less and 20 percent were for less than one month. Interest rates on a monthly basis averaged 85 percent. Brazil: Access to Financial Services Table 1.13 Access to Loansand Credits Reasons for Loan Refusals - Didn't have a job Returns less than expenses Others Valid cases Source:World Bank, Survey of Access to Financial Services in Urban Areas of Brazil, 2002. .4 Survey Results Paymentsand MoneyTransmissionServices - 71. Cash was overwhelmingly the standard medium of payment for all types of transactions and respondents; 77 percent used cash for all transactions. There is surprisingly little variation by type of payment. Thus, 88 percent of respondents use cash for all daily purchases, but the proportions for education, health and house rent or mortgage payments are similar; 87 percent, 85 percent and 88 percent respectively. And surprisingly, 92 percent of utility bills are paid in cash. The only expenditure categories where cash use is lower is consumer durables; 77 percent for electrodomestic goods and 80 percent for furniture. This could reflect credit offered by consumer finance companies, especially, post- dated checks - in these two payment categories, check payment amounted to 12 percent and 13 percent respectively. Additionally credit cards are also used in these two categories, by 9 percent and 6 percent of respondents, respectively. Credit cards or debit cards are usedfor daily purchasesby only 7 percent of the persons surveyed, though debit cards are also usedfor utility payments by around 5 percent of persons. 72. Interms ofpoints of payment, a surprisingly highproportion (over 90 percent) make payments in person, at bank branches (54 percent) or other points (38 percent); only 2.5 percent use the mail (44 persons) andjust over 1percent use the internet or telephone. These numbers suggest that retail payments systems are still unsophisticated for the majority of persons. Are the present retail payments systems burdensome? Surprisingly, despite the high proportion of cash payments, between 30 and 40 percent of respondents claimed that payments for utilities, education and health could be made in less than ten minutes, and for 50-60 percent of respondents, less than 20 minutes. But, for 15-22 percent of the persons, these payments take from half an hour to an hour. Payments for furniture and appliances, surprisingly, seemed to take the longest time; over half an hour for three fifths of the persons surveyed. 73. Payments services were also investigated in terms of income receipt. More than two-thirds of respondents (68 percent) receive their primary income in cash, while 24 percent have direct deposits in their bank accounts. Surprisingly, only 4 percent receive checks. Cash preference i s high; the proportion of persons who would like to receive cash payments (72 percent ) i s higher than those who actually receive cash. Checks appear to have the greatest disutility as the proportion who would like to be paid by check (91.5 percent) is lower than the actual prop or ti or^.^^ The great majority (72 percent) would like to receive their incomes in cash, followed by deposits in bank accounts (24 percent). Check cashing is usually done at a public institution (50 percent); the proportion for private institutions is half this level at 26 percent. Eleven percent of respondents use family or friends and 9 percent use a commercial shop. It i s interesting that although there i s a preference for private institutions for overall banking services and also for deposit services, preferences are reversed for check cashing. This may indicated that private banks services for check cashing are more onerous or more expensive. Finally, remittances are rare -only 68 persons or 3.5 percent of respondents had sent or received remittances in the last year. Around two 48 Direct deposit proportions are unchanged. Page 28 Part 1. Assessing Access fifths of these were by interbank transfer and another ten percent through the postal system. Another two fifths were in cash or through personal transmission. 5. ANALYSIS OF SURVEY RESULTS-WHAT FACTORS ASSOCIATEDWITH ACCESS? ARE 74. The principal aim of this section is to investigate at the broadest level those factors which have been used, in a policy context, to address access issues, and to examine the scope for extendingthis menu to extend the options for policies favoring access. We analyze the associations or correlations between a series of measures of access and selected economic parameters important to policy, through the construction of cross-freq~encies~~and next through regression analysis. The measuresof financial access used are grouped in the same four broad categories as those described above; use of a formal financial institution, access to deposit and savings facilities, access to loans and credits, and finally, access to payments services. 75. Particular attention is thus paid to three specific issues. The first of these is location and the association between the location of the respondents and the availability of financial services. Locational measures of access have been the primary means through which the government has tried to track and to address shortcomings in access, as discussedin section 2 above. A central feature of government policies for the expansion of access has been the attempt to put some financial institution or point of service in each municipality. However, location can be interpreted at two levels; first the region of the country; and second, the microeconomic characteristics of the neighborhood and dwelling of the respondent. It i s proposed here that there may be significant micro differences in `location' within a region, state or even municipality, which could have pockets of poverty within broadly prosperous areas - and conversely, could have well-to-do persons even in municipalities which are less affluent. We explore the association between access and both definitions of location. 76. Second, we look at the role of different types offinancial institutions in the provision of financial services. There has been an assumption that due to market failures infinancial markets, there is a need for a public role in the provision of financial services, and that public banks have an important role to play as providers of such services. As shown in Section 3, public banks and private banks often behave in similar ways and are broadly substitutable in services to some segments of the population. We try to further investigate the role of public financial institutions, to see if they better serve the needs of the poorer segments of society. 77. Third, we pay special attention to the role of information inthe provision of financial services. It has been pointed out that one of the key reasons for the limited access to financial services to the poor is the huge difficulty in gathering information on such persons - the asymmetric information problem. We therefore investigate the extent to which proxies for information on creditworthiness, such as income, wealth or the possession of collateral, can influence decisions regarding the provision of financial services. If there are significant asymmetries in the availability of information, these would be illustrated by the importance of such proxy variables for creditworthiness. The policy implication would be to improve overall information on creditworthiness and also to address the specific biases noted. ~ 49Differences in the distributions of variables between the parameters examined (different regions, different institutions, different socio-economic categories, etc.) are tested for significance using chi-squared tests of independence. Significant statistics reported here, unless noted, are significant at a 1percent level. Detailed results are available in background papers. Results presented here are summaries only; more detailed background papers are available. Page 2Y Brarii: Access lo Financial Services 5.1 The Impact of Locationon FinancialAccess Location in terms of Regions of the Country 78. There i s a clear variation in a series of indicators of financial access across regions, which broadly indicate that the North, and sometimes the Northeast of the country are more limited in access than the central or southern regions (Figure 1.8 and Appendix Table A1.8).50In terms of use of a formal financial institution, for example, the majority of the respondents in the North (75 percent) and in the Northeast (71 percent) didn't have a bank account; in contrast around half the persons in the South (55 percent) and Southeast (45 percent) did have a bank account. Most persons in the Northeast (70 percent) used banking correspondents as their primary financial institution while in the Southeast one half of the persons used banks or their branches as their primary financial institution. Only 6 percent of the respondents who lived in the North had deposits in an account; in contrast to more than one-third of individuals who lived in South (37.5 percent). And 97 percent of persons in the North didn't apply for loans, compared to 83 percent in the Southeast. Regarding payments, only 18 percent of persons in the Northeast and 16 percent of persons in the North had a debit card, in contrast to 40 percent in the South and 44 percent in the center.51 Only 2 percent of respondents that lived in the North region used transfer or direct deposits to made payments while 15 percent of individuals in the Southeast used the same method for payments. Location in terms of Micro Regions Neighborhoodsand Areas within a city - 79. But differences in access can also be strong within a region, at the level of individual neighborhoods (Appendix Table A1.9). First, in terms of access to a financial institution, we distinguished for example between persons living in legalized areas and those in illegal areas. Forty five percent were bank account holders in the former in contrast to only 37 percent of persons in the latter. Similarly with regard to type of building, only 34 percent of people who lived in informally constructed shacks or in rooms had a bank account, in contrast to almost half (45 percent) the persons who lived in regularized houses or apartments. And only 13 percent of individuals with houses of less than 0.5 rooms per person had accounts, compared to more than half the individuals who lived in houses of more than 2 rooms per person. Interestingly, more than half the respondents (59 percent) who lived in shacks or rooms used mainly correspondents banks, 47 percent of persons who lived in house or apartment used mainly bank institutions. Again, we have the same pattern when we look for the number of rooms per person in a house. Almost two-thirds (56 percent) of individuals who lived in houses that had more than 2 rooms per person used mainly banks, while more than two-thirds of respondents who lived in houses with less than 0.5 rooms per person (64 percent) usedcorrespondents. 50Differences cited inthis section are all based on the results of chi-squared tests o f independence of distributions (see Appendix 1.2: Technical Note on Estimation) and are significant at the 1percent or 5 percent levels. Results with the corresponding test statistics are detailed in Appendix Table A1.%Appendix Table AI.14. 5'For credit cards, there are also differences in proportions ,which are significant but weaker. Page 30 Part 1. AssessingAccess Figure 1.8 Role of Location in FinancialAccess Yo Havinga bank account 60 - - ~ 50 - 40 - 30 - North Northeast Southeast South Central- North Northeast Southeast South Central- West West % Seeking credit Yo Payments 181 _ ~ _ ~ ___-. 70 151 - (%differerce) 12-1 I (0.0) North Northeast Southeast South Central- North Northeast Southeast South Central- 0 Credit sought inthe past Pmonths West West Credit sought and approved in the past 12months 0 Public fgPrivate flOther GDP per capita (2000)- Source: World Bank, Survey of Access to Financial Services in UrbanAreas of Brazil, 2002. GDP data providedby IBGE. 80. Results are similar when we analyze access to depositfacilities: 30 percent of individuals who lived in a unique permanent house had deposit balances compared to only 19 percent of individuals who lived incollective dwellings.52Forty one percent of the individuals who lived in a house with more than 2 rooms per person had deposit balances,in contrast to only 7 percent of persons who lived in a house with less than 0.5 rooms per person. Looking at credits, 83 percent of respondents living in homes with more than 2 rooms per persons didn't apply for loans, compared to 94 percent of people who lived in houses with less than 0.5 rooms per person. Thirty four percent of persons who lived in a house with more than 2 rooms per person had a debit card, while only 5 percent of persons who lived in homes with less than 0.5 rooms per person had a debit card. Differences in proportions for credit cards are also significant. Finally, persons who used cash for payments was bigger in illegal areas (84 percent) than in legalized areas (75 j2Defined in this case as money in either a special savings account, a current account or a term deposit account. A generic definition of deposits yielded similar results; 52 percent of persons living ina house with more than 2 rooms per person had no deposits; in contrast, 90 percent of people in homes with less than 0.5 rooms per person had no deposits. Brazik Access io Financial Services percent), while the number of persons who used debit or credit card and transfers or direct deposits was bigger in legalized areas. Observations are similar when we compare respondents who lived inhouses and apartments with respondents who lived in shacks or rooms. 81. There i s a significant suggested policy implication. Although there are indeed differences between the availability of financial services between regions of the country, placing financial institutions ineach region or municipality initself is not a sufJicientcondition for broadening access. There may well be important differences also within neighborhoods and districts of a municipality, region or state. Even if there is a financial institution present in a given location, its clients may be biased towards the better off persons. And conversely, areas benefiting from dense networks of financial institutions may still have pockets of the underserved. The implication is that targeted interventions looking at microregions, and at specific groups within regions, may be an important supplement to broad regional policies if access tofinancial services is to be promoted. e Role of Different Institutions and Financial Access 82. The next question examined concerns the role of public and private financial institutions as vehicles of access. Brazil's public banks have been considered important vehicles of outreach to underserved communities, and this role has been premised upon market failures or imperfections which make it too costly or too risky for private banks to lend to some communities. This section investigates the differences in outcomes for public and private banks in terms of the provision of various forms of financial services, and between groups of persons with different socio economic characteristics. The key findings are: 3 Individuals' useof public andprivate banks depends on the type of jinancial service sought. There are significant differences across different types of service. 3 Attempting to group individuals on the basis of thosewho may be more or less disadvantaged, (eg, those in poorer regions versus those in more afluent regions; those in poorer neighborhoods or homes compared to those in more afluent neighborhoods or better homes, those with higher income or education compared to those with lower income or education), there are some significant results where more disadvantaged groups make greater use of public bankfacilities. There are also significant cases of the reverse and many instances where there are no clear associations of the less well off with public banks. Next, looking at different measures of access and first considering overall access to financial institutions, disadvantaged groups do show some tendency to make greater proportional use of public financial institutions. However, the results are significantly injluenced by correspondent outlets. Both the disadvantaged and the well-to-do increase their use of public institutions once correspondent outlets are considered.53 3 Second,for the majority of persons seeking deposit or credit services, there is a general preference for private banks. 53 To analyze overall access toformalfinancial institutions, two separate measuresof public versus private banks were adopted. Inthe first, wide definition, we included correspondent banking outlets (lottery shops with banking services for public banks and post offices with banking services for private banks). Inthe alternative, narrow definition, correspondent outlets were not included. Results turned out to be very sensitive to these alternative definitions. Part 1. Assessing Access In the case of deposit services, test results find significant differences between public and private bank use, and location and personal characteristics, but the results show that the less well off groups tend to make lower use of both public andprivate depositfacilities. 3 In the case of credit facilities, tests of differences are mostly not significant. An inspection of frequencies again suggests that the association of disadvantaged individuals with public banks is unclear.. 3 Forpaymentsservices, in contrast, there is a clearpreferencefor public banksas these aredefined to include the correspondent outlets known as lottery shops. In all cases, the proportions of users are higherfor public banks thanprivate banks. 3 largepart of A thepreference for public banks may be traced to the servicesprovided by the loterica advantage in the provision of such services. If the correspondent outlets of private providers, especially correspondent outlets of the Caixa EconGmica Federal. At the time of the survey, the Caixa had a clear the postal outlets of Bradesco, are also able to compete in this market, this advantage may not remain. An implication is that policies adopted towards correspondent outlets will be very important in expanding access andpromoting competition in theprovision of services to the underserved. a Thereisalso aclearly greater useof public, comparedtoprivatepensionfunds, againby thepoorer and by better off social segments. =, Looking at creditfor housing,public banks dominate theprovision of services, though their services are used more extensively by the better off segments. 83. First,examining overall access to afinancial institution (our first measureof access) and looking across regions, including correspondents, 84 percent of individuals inthe Northused mainly public banks, compared to around two-thirds of individuals in the Southeast. Without correspondents, half the persons in the North continue to use public banks, but in the southeast users of public banks falls to around 38 percent. Both differences are significant, suggestingthat many persons used public bank services because of the lottery shops (Appendix Table A1.lO). Similarly, looking at "micro regions" and including correspondents, 81 percent of persons in homes of less than 0.5 rooms per person used mainly public banks compared to 61 percent of individuals in houses of over 2 rooms per person. But, with the narrow definition, only 48 percent of those with less than 0.5 rooms per person used public banks, and only 36 percent of those with more than two rooms per person. Both differences are significant. (see Appendix Table Al.10) 84. Looking at individual characteristics, and considering gender, women are less likely to use private banks than men. However, using the narrow definition for public and private banks, the majority of both men and women used mainly private banks. Including correspondent outlets, both women and men used mainly public banks. Interms of income, only 36 percent of persons with income in the highest quintile used public banks as opposed to 49 percent of individuals in the lowest quintile. Again, with the wider definition, use in both groups goes up dramatically, to almost half in the top quintile (51 percent) and 78 percent in the lowest quintile. Looking at wealth, half the persons without collateral used public banks compared to only 40 percent of those with collateral. On including banking correspondents, the majority of individuals with or without collateral usedpublic banks. Similar results are obtained when we look at education, and persons with or without jobs. 85. Our next series of measures of access describe access to deposit, payments and credit services (see Appendix Table Al.11). With regard to deposit and loan services, correspondent outlets, until Page 33 #mi/;Access to Financial Services recently, did not typically offer these services.54With regard to payments, public bank outlets and the lottery shops are grouped together. Payment services, as may be expected, are dominated by the public sector (defined to include the lottery shops), and this preference is true across social groups. Some indicators also suggest that public banks may be more oriented towards the underserved; (thus, 62 percent of persons in homes with less than 0.5 rooms per person preferred public banks for payments compared to 55 percent of those with more than 2 rooms per person; similarly 68 percent of persons in the North used public payment outlets, compared to 34 percent and 58 percent for the Center and Southeast, and perhaps most clearly, 58 percent of persons with less than primary education usedpublic payment outlets while 11 percent used private banks; compared to 41 percent with more than secondary education for public payment outlets and 38 percent for private). 86. Regarding deposits, the overall pattern in contrast to payments services is some preference for private banks for persons in each geographical or socioeconomic category. Looking at comparisons of the more and less disadvantaged groups of persons, according to a range of criteria, a first observation is that the better off persons have proportionally larger deposits in both public and private institutions than the less well off. For example, 18 percent of individuals who lived in houses that had more than 2 rooms per person had deposits in public sector and 28 percent had deposits in private sector, in contrast only 3 percent of persons who lived in houses that had less than 0.5 rooms per house had deposits in public sector and 6 percent had deposits in the private sector. Similarly, men had more deposits in both public and private sector banks than the women. Seventeen percent of the men had deposits in public sector and 22 percent in the private sector, while 11 percent of the women had deposits in public sector and 15 percent had deposits in private sector. The differences are pronounced with regard to income and education. Only 7 percent of persons with low income had deposits in public sector and 6 percent had deposits in private sector, in contrast 25 percent of individuals with high income had deposits in public sector and 42 percent had deposits in private sector. Similarly individuals with high levels of education had more deposits inpublic and inprivate sector than individuals with low education. 87. Do the more disadvantaged persons make proportionally greater use of public facilities for deposits? This hypothesis returns mixed results. Persons in the lowest income quintile had marginally more deposits in public banks (7 percent public compared to 6 percent private, while persons in the top quintile had 25 percent of their deposits in the public sector compared to 42 percent in the private sector) and persons in the northeast had proportionally more deposits in public banks (10 versus 7 percent, compared to 11percent public and 22 percent private for the southeast), but by contrast, persons living in illegal areas, or persons living in shacks or rooms instead of regular houses or apartments had proportionally more deposits in private banks. Similarly, those in the lower educational levels and those without collateral also made greater use of private compared to public facilities. These results, all of which are statistically significant differences, are at first surprising in view of the special savings accounts dominated by the Caixa, but can be explained by the high proportion of current account holders who are offered services at private banks.55 88. Next, looking at credit facilities, the case for public banks remains blurred (Appendix Table Al.11). In this case, the differences in proportions are usually not significant. They suggest again that the less privileged have weaker access to both public and private financial services, and there is no general tendency for the disadvantaged to make proportionally greater use of public banks. One significant result obtained is with regard to the micro-location variable on the nature of the dwelling: almost one third of individuals who lived in a house or in a apartment received credit from private banks and 26 percent received credit from public banks, while 48 percent of individuals who lived in a shack or in a room 54 A resolution of the National Monetary Council (Resolution 3110of July 31,2003) has expanded the scope of services offered by correspondents. j5 This is corroborated by data discussed in Part 3 of this study ' Downscaling Private Banks'. Page 34 Part 1. Assessing Access received credit from private banks and only 6 percent of those received credit from public institutions. The other significant difference is in the proportion of persons who possessed collateral, in the form of a car or house and those who didn't have collateral. Twenty-six percent of persons who had collateral received credit from public banks and 33 percent of those received credit from private banks, in contrast only 8 percent of individuals who didn't have collateral received credit from public banks and 28 percent received credit from private banks. In both cases, the less privileged groups had distinctly lower access to both public and private banks, and the frequencies suggest greater use even by these individuals, of private banks. 89. A fifth variable which was used to discriminate between public and private financial services was the voluntary contribution to public versus private pension funds. Results which were highly significant show clear differences between contributions in terms of location as well as individual characteristics, and here, as in the case of payment services, both the well off and more disadvantaged groups make greater use of public pension funds. Eleven percent of persons who had a high level of education had contributed to public funds and 10 percent contributed to private funds, while only 3 percent of individuals with low education contributed to public funds and only 1 percent contributed to private funds. The same relationship exists between level of income and contribution to public or private funds. In terms of employment, twenty two percent of employers had contributed to public funds and 11 percent had contributed to private funds, while only 3 percent of employees without a work certificate had contributed to public funds and 2 percent contributed to private funds. This may reflect that incentives to contribute to private pensions funds under a third pillar (previd2ncia complementar) need not have matched employer contributions and thus incentives for such savings are limited. 90. The final variable looked at was the use of credit from public versus private banks for home purchase. With regard to home purchase, the vast majority of the persons in all geographic regions used saving to finance home purchase. But, while in the North, 70 percent of respondents used savings, 10 percent used credit from public banks and another 10 percent used credit from private banks, in the Southeast 56 percent used saving, 22 percent used credit from public banks, and 11 percent used credit from private banks. Looking at micro definitions of location, 13 percent of persons with homes of over 2 rooms per person used public bank credit, while none of the respondents in homes of less than 0.5 rooms per person did so. Looking at income, almost one half of respondents with low income (49 percent) financed home purchase by savings, 5 percent used credit from public banks and 7 percent used private banks, while 39 percent of respondents who had high income financed home purchase by saving and 18 percent used credit from public institution. While 12 percent of respondents with a low level of education usedcredit from public banks to purchasea home, 16percent of respondentswith highlevels of education used such institutions. All these results, which are statistically significant, seem to perversely indicate easier access to public bank credit by the more well to do social groups, which may partly reflect superior economic capacity to repay. 5.3 The Role of Respondent Characteristicsand FinancialAccess 91. As discussed earlier, access can depend on the degree of client information available, because of the impact on risks and hence costs. Obtaining such information can assume special importance in environments where the capacity to enforce repayment may be difficult due to, first, the limitedresources of the client, and second, due to the nature of prevailing laws and practice with regard to debt recovery or bankruptcy. In the absence of full information, especially on poorer persons with limited credit histories, institutions tend to look for proxies to information on creditworthiness. Such proxy variables could include information on cash flows (income) which could service loans, or assets (wealth), which could aid loan recovery. It appears that banks also desire such information for opening an account, due to the need to service accounts, or even to accept a deposit. Other characteristics of this form could include position in household (heads versus dependents), gender (males versus females) or employment characteristics. Page 35 Brazii`:Access to Financial Services Individual Characteristics,AsymmetricInformationand Access to FinancialServices For most measures of financial access, virtually all the socio economic characteristicsincluded in the analysis turn out to be highly significant discriminators in determining access to financial services. Foremost among these were characteristics referring to income, wea employment characteristicswere also highly significant. Finally, for householdalso mattered (see Appendix Table Al.13). Income: In terms of access to bank acco nthe lowest quintile had a bank account, in contrast to 64 percentof deposits, only 9 percentof the persons that had income in the lowest account, a current account or a term deposit account, in contrast to 47 percent of persons in the top quintile. Using anoth deposits, 84 percentof persons in the bottom quintile didn't havedeposits, incontrast to only 32 perc quintile. Income i s also important in determining access to credit and loans. Ninety-two percent of respondentsin the bottom quintile didn't apply for credit, compared to 74 percent in the top quintile. Differences in sources of credit however are not very significant among the quintiles of income distribution. Only 5 percent of the persons with a low income had a debit card, in contrast to 40 percent of persons with a high income. Similarly only 7 percentof the persons with a low income had a credit card, comparedto 34 percentof persons with a high income. Education: is also a very important determinant in the access to financial institution. Among the significant differences, only 28 percent of persons with education less than primary level had a bank account, while 84 percentof the respondentsthat had more than secondary school had a bank account. When we comparethe access to public and private banks, the majority of respondents with low education used mainly public banks (including the correspondents), while 59 percent of person with high education used mainly private banks (including the correspondents).Only 19 percent of persons with low educationhad money in either a special savings current account or a term deposit account comparedto 64 percent of those with higher education Only of respondents with higher education didn't have deposits, in contrast to 78 percent of those with low Ten percentof personswith a low educationappliedfor and received loans comparedto 23 percentof re with a high education. Only 9 percent of persons with a low education had a credit card, while almost one half of respondents with a high education(48 percent) hada credit card. Wealth (collateral): Almost one half of respondents that had collateral (45 percent) had a bank account, while only 35 percent of the persons without collateral had a bank account. Collateral is also important for the pattern of deposits and savings, but not as important as income or education. However, possessing collateral i s the only characteristicwhich makes a significant difference to credit sources. Employment -havinga job: More than half the people who had a job in the last month (55 percent account, in contrast only 30 percent that people who didn't have a job in the last month had a bank account. 38 percent of individuals who had ajob in the last month had money in either a special savings account, a current account or a term deposit account, in contrast only 19 percent of persons who didn't. Thirteen percent of respondents that had a job in the last month applied for and received credit, while only 8 percent of persons who didn't have ajob were inthe same situation. Thirty-four percentof respondentsthat had ajob inthe last month had a debit card, in contrast only one half (17 percent) of persons who didn't have ajob in the last month had a debit card. The difference is very significant. The same patternoccurs with the possessingof a credit card. Employment having a work certificate: 68 percentof employees with a work cert$cate had a bank account, in - contrast to only 44 percent of employees without a work certi3cate. Forty-seven percent of employees with a work certlficate had money in either a special savings account, a current account or a term deposit account, while only 26 percentof employees without a work certi3cate had money inone of these accounts. Gender and Position in Household: Half the men had a bank account compared to a third of the women. One half of the heads of household had a bank account while only 37 percent of the dependents had a bank account. One half of men used banks as the primary financial institution, while 53 percent of the women used correspondents banks. Thirty-four percent of the men had money in either a special savings account, a current account or a term deposit account compared to 23 percent of the women. But gender is not a good determinant for credit and loan behavior. There are also significant differences betweenwomen and men inthe possessionof credit cards and debit cards. Page 36 Part 1. Assessing Access 92. How important are these factors in obtaining financial services in Brazil? The results cited in the box indicate that information on such characteristics is very important in determining access to financial services. Most important among these characteristics is the income of the respondent, which has a strong positive relation with access to a variety of financial services (Figure 1.9), including having a bank account, deposit and credit services. Indeed, in terms of credit, as income levels increase, the proportion of refusals of credit applications diminishes. Simple trend lines fitted to income and credit applications and acceptances tend to converge as incomes rise. Income levels appear to be particularly positively associated with the use of private banks, for deposit and payment services, where their use increases more than proportionately with increases of income, while the use of public banks diminishes. Figure 1.9 The Role of Income in Different Measuresof FinancialAccess Yo Havinga bank account Havingdeposlts oh Seekingcredit in last 2 months 90 - 70 ___- - - ___ ~~ . 30 Tp- ~~ ---- -- 60 45 - 30 - - ~ _ _ _ _ -- 20 15. ~ * - - 0 ~ .~ -~ 0 , , , 7 0 - , 0 1 2 3 4 5 6 7 8 9 0 0 1 2 3 4 5 6 7 8 9 0 0 1 2 3 4 5 6 7 8 9 0 Income decile *Credit sought inthelast Pmonths Institutional Choice Using PublidPrivate Institutions %share - 4 Credit sought and approved inthe last Pmonths - Payments Income decile l R b lnst x Pvt inst 4 Other lnst Source. World Bank, Survey of Access to Financial Services in Urban Areas of Brazil, 2002, 6. DETERMINANTS OF ACCESS AN ECONOMETRIC - INVESTIGATION 93. In the previous section, we examined the links between financial access and location (both in regions of the country and locations in the sense of neighborhoods or areas); institutions (primarily, public and private banks) and a series of individual characteristics. Following estimations of the cross- frequencies between these variables, we examined the extent to which differences in frequencies were significant. However, this could not permit us to evaluate the relative importance of the different factors 3r2zi! Access io Financial Services affecting access. The present section discusses the impact of these factors based on regression analysis.56 A series of regressions were undertaken, beginning first with locational explanatory variables and then adding variables on individual characteristics. 6.1 Determinantsof Access to Financial Services 94. Using a series of different measures of access and using a number of different explanatory variables related to location as well as individual characteristics, the most significant result to emerge was the importance of income, over and above most other parameters, as a determinant of access (Table 1.14 and Figure 1.9).57Closely parallel to this is the importance of education, and for access to credit, wealth or collateral. Although geographic region, used on its own, is a significant explanatory variable, its importance diminishes considerably after income is added to the equation. Thus for example, using the probability of having a bank account as our measure of access, geographic region, number of rooms per person, sex and age are all significant predictors until we included income, education level and the possession of collateral. Income and education are then very significant and in the first group of explanatory variables, only geographic location and number of rooms per persons continue to be significant. Similar results are obtained if we examine the probability of using banks as the primary financial institution. In this case, once socio-economic characteristics are included, the importance of geographic location and even dwelling characteristics disappears. 95. Broadly similar results are obtained when we look at access to specific services. Looking at deposit services, the probability of having money in either a special savings account, a current account or a term deposit account is significantly linked to income level, education, and two parameters of micro location. Geographic location i s important only until these socio-economic characteristics are included. In the case of credit services, income is the only independent variable that i s related in a significant way with the probability of applying for loans. Income and education are both very important for predicting the probability of having a debit card, and income, education and the possessing of collateral all explain the probability of having a credit card. Geographic region is not an important predictor. j6See Appendix Tables for a summary of the results. The estimation methods used are detailed in Appendix 1.2 - Technical Note on Estimation. Explanatory variables selected are those which were demonstrated to have significant differences across groups in the analysis of the preceding section (chi-squared tests of independence). Probit models were used for binary independent variables while multinomial logit models were used for multivariate independent variables. The usual care has to be applied in interpreting the results bearing in mind that although significant findings can help to predict some type o f access, these results do not imply causality. 57 A detailed analysis of parameters explaining access was also undertaken separately using data from a household survey undertaken in Brazil in 1997. This analysis, which definded a series of levels of access, depending on the range of services included, also had very similar overall results. "Access to Financial Services in Brazil - An empirical evaluation based on the Pesquisa Sobre Padr6es de Vida (PPV 1996/97)" Rio de Janeiro June 2002. Coordinated by RosaneMendonqa with a technical note by Daniel Santos. (Mimeo, World Bank). Page 38 Pari 1. Assessing Access Table 1.14 Econometric Results Determinantsof Access to FinancialServices- Select Results - Econometric Model Probit Multinomial Logit Probit Probit Multinomial Logit Probit Geographic region 0.04 (0.03) 0.26 (0 18) 0 16 (0 39) -0.02 (0.14) 0.02 (0.11) 0.3 (0.00) 0.2 (0.08) -0.6 (0.00 Sex -0.05 (0 19) -0.27 (052) -0.01 (0.97) 0.05 (0.13)-0.08 (0.01) -06 (0.01) -0.4 (0.03) -1.3 (0.02) Age 0.02 (0.35) 0.01 (0.46) (0.08) 0 02 (0.35) 0.02 (0.17) Position in the household -0.07 (0.10) 0 03 (0 36) 0 3 (021) 0 00 (0.99) 18 (0.01 Illegalvs legalized area -0.03 (0.57) -0.11 (0 38) 0 1 (075) 0.0 (094) 0.7 (0.49) Type of house' -0.09 (0 01) Type of buiiding -0 11 (0.04) 0.11 (0.66) 0.20 (0 74) 0.07 (0.09)-0.04 (0.06) -0 1(0.84) -0.20 (0.48) -35.5 (0.0 0.00 (0.93) No of rooms per person -005 (0.00) 0.04 (0.85) 0.15 (045) 0.02 (0.11) 0.18 (0.01)-0 1 (0.13) -0.2 (0.00) -0.5 (0.01 Having a job -0.04 (0.80)-1.12 (0 56) -0.86 (0.66) 0 12 (0.34) 0 18 (0.18) 2.1 (0.02) 0 00 (0.99) -0.5 (0.72 Role in Workplace -0.01 (0.62) 0.43 (0 18) 0.35 (0 27) -0.02 (0.46)-0.04(0.05) -0.3 (0.04) -0.1 (0.32) -0 1 (0.71 0.02 (0.38)-0.05 (0.39) Part time or full time empt. 0.01 (079) -1.06 (0 07) -1.25 (0.03)-0.04 (0.32) -0.01 (0.72) -0.3 (0.25) 0.00 (0.93) -02 (0.69 0.01 (0.64) Sector of activity -0.01 (0.62) 0 31 (031) 0.31 (0.30)-0.01 (0.56) -0.02 (0.29) -0 2 (0.10) 0 0 (0.94) 0 2 (0.27 Income 0.11 (0.00) 0.48 (0.00) 0.17 (0.27) -0.06 (0.00) 0 08 (0.00) 0.4 (0.00) 0.5 (0.00) -0.1 (0.44 Education 0.10 (000) 0.35 (0.02) 0.01 (094) -0.05 (0.00) 0 08 (0.00) 0.4 (0.00) 0.4 (0.00) 0.00 (0.86 0.05 (0.00) 0.06 (0.00) Possessionof collateral -0.03 (0.50)-0 12 (0.82) 0.05 (0.92) 0.01 (0.71) 0.01 (0.82) 0.1 (059) 0.0 (087) -0.2 (0.76 Observations 1077 942 1200 1085 1251 Note: Type of house refers to unique, improvisedor collective house. Type of building refersto (I) house or apartment or (ii) shack or room Source:World Bank, Survey of Accessto FinancialServices in UrbanAreas of Brazil,2002 Determinantsof the Volume of Credit Requestedand Approved 96, Attempts to evaluate factors determining the volumes of credit requested and approved, using ordinary linear regression show that income is the only significant explanatory variable. People who have more money request more credit and have received large loans. This supports the findings of Figure 1.9. Variables that are related to the hypothesis that individuals try to smooth consumption along their lives (like age, education and the number of dependents) are not significant. There i s thus no confirmation of the permanent income hypothesis in these results (Table 1.15), which also emphasize the importance of income in determining access to financial services. Table 1.15 Econometric Results: Determinantsof Volume of Credit Requested and Approved Possessing of collateral Observations Source, World Bank, Survey of Access to Financial Services in Urban Areas of Brazil, 2002. P values are in parentheses "3 ~eterminant~institutionalChoice of 97. The econometric analysis also suggests that income level and education are significant explanators of the probability of using public banks as the primary financial institution, relative to private banks. A high level of education and income decreases the probability of using mainly public banks. Page 34 Brari!:Access io Financial Services These results are true whether a wider or narrower definition of public banks is used (including or excluding correspondent services). However, the probability of having deposits in either the public sector or the private sector (in comparison to not having deposits) while positively and significantly determined by income, is also significantly determined by other socio economic and locational variables. When we analyzed credit, we saw that only possession of collateral has a significant relationship with the probability of received credit from public banks, but collateraldoesn't have a significant relationship with the probability of receiving credit from private banks (Table 1.16). Table 1.16 Econometric Results:The Probabilityof Using Public Banks for Access, Depositsand Credit Sex 0.05 (0.29) -0.6 (0.01) -0.4 (0.03) -1.3 (0.02) Type of building 2 0.08 (0.23) 0.07 (0.09) -0.1 (0.84) -0.2 (0.48) -35 5 (0.00) -1.02 (0.24) 0.73 (0 11) Number of rooms per person 0.02 (0.39) 0.02(0,11) -0.1 (0.13) -0.2(0.00) -0,5(0,01) Having a job 0.06 (0.75) 0.12 (0.34) 2.1 (0.02) 0.0 (0.99) -0.5 (0.72) Role in Workplace -0.03 (0.35) -0 02 (0.46) -0.3 (0.04) -0.1 (0.32) Part time or full time employed 0.00 (0.97) -0.04 (0.32) -0.3 (0.25) 0 0 (0.93) Sector of activity -001 (073) -001 (0.56) -0.2 (0.10) 0.0 (0.94) Income -0.05 (0 00) -0.06 (0.00) 0.4 (0.00) 0.5 (0.00) Education -0.01 (070) -0.05 (0.00) 0.4 (0.00) 0.4 (0.00) Possessing of collateral 0.01 (0.93) 0 01 (0.71) 0.1 (0.59) 0.0 (0.87) -0.81 (0.09) Age 0.54 (0.01) 0.26 (0.10) Position in the household 0.3 (0.21) 0.0 (0.99) 1.8 (0.01) Illegal area vs legalized 0.1 (0.75) 0.0 (0.94) 0.7 (0.49) Type of house3 450 1037 n of public banks included lottely shops with banking services. 2Type of building refersto (I) house or aparlment or (ii) shack or room. Type of house refers to unique, improvisedor 3 collective house. Source:World Bank, Survey of Access to Financial Services in UrbanAreas of Brazil, 2002. 98. Finally, income i s the only variable to have a significant (positive) relationship with probability of receiving credit from public banks. Both income and the respondents' sector of activity were significant explanatory variables for the probability of receiving credit from other government sources for home purchase. Table 1.17 EconometricResults: The Probabilityof Using Public Banks for Real Estate Purchase Position in the household Sector of activity Income Observations 130 Source World Bank, Survey of Access to Financial Services in Urban Areas of Brazil, 2002. * Significant statistic Page 40 Part 1. Assessing Access 7. SUMMARY OFFINDINGS POLICY IMPLICATIONS AND 7.1 Overall Findings - S y n t ~ e ~ i ~ ~ n g and DemandSide Findings Supply 99. The preceding sections have analyzed first, the factors associated with the supply of financial services, particularly in the form of institutional outlets for such services, and next, patterns of demand for such services and how these demands have actually been fulfilled, at the level of individuals in Brazil's urban areas. Combining and synthesizing the results, the key findings which emerge are: 3 Brazil'spopulation's access tofinancial serviceshas been dominated by banks and theprovision of banking services in the form of bank branches has remained broadly unchanged over the last decade. Brazil is not obviously underbanked in terms of international comparisons. 3 However, important new trends are emerging in theform of new outlets, such as correspondents, and as shown by the users survey, these have remarkably injluenced and increased overall use of financial services, both among the better off and less well to do sections of society. The effective use of banking services is considerably greater than the number of bank accounts alone would suggest. = While a first look at the supply of financial institutions points to wide regional differentials in the provision of services, these are considerably reduced when we correctfor differentials in regional GDP andfor geographical size. z Thesurveyof individualfinancial behavior alsosupports thefinding of the importance of locationin determining access, but demonstrates that location in terms of the micro characteristics of an area or neighborhood can be as important a discriminatorfor access. = The survey also strongly corroborates the findings of the supply side analysis regarding the importance of income, at a per capita level, and points also to the significance of a range of socio economic characteristics, such as education, and (for credit services), wealth, in determiningfinancial access. =? Gender bias appears to be present in some measures (access to a financial institution, and deposit services) but not in all measures (gender bias is not significant for credit). 3 Theimportance of socio economic characteristics in access tofinancial services suggests that there may be problems of information, or of enforcement, which cause lenders to emphasize such information. = Regarding the roles of public and private financial institutions, both the aggregated supply side analysis and the survey of users indicated that there is some association between lower income groups and public banks. z However the user survey also showed that the useof public banks varies by type of service. Thus, public banks are popular for payment services (due to the outlets of the Caixa Econdmica Federal), for all socio economic groups, and also dominate the provision of housing credit, but largely for the better off groups. For deposit taking and for credit, both the privileged and less privileged socioeconomic groups, broadly measured, have some preference for private banks, and with increases in income both increase their proportional use of private banks. =? Thus in many areas, the user survey would corroborate thefindings of the supply analysis, that in many respects,public and private banks are broadly substitutable. Brtizi,`:Access lo Financiai Services 7.2 Implications for Policy Directions 100. What are the implications of these findings, from the perspective of broad policy directions to adopt in addressing access related issues? The preceding analysis suggests a number of thematic directions which could help to guide policy choices. 101. A central and consistent theme to emerge from this analysis is the importance of income-related factors in determining access. A basic deduction is that access will be increased by overall growth oriented policies; thus improved financial distribution in addition to financial deepening is associated with growth. While clearly important, this finding does not suggest immediate measures for policy makers to address issues of access. Of far greater significance as an implication for policy makers is the corollary to this finding - that if access is so significantly influenced by income, then targeted policies, which particularly address lower income groups, are likely to be important in addressing access. What does this imply in practice? That policies such as for example, "lifeline" or basic accounts, affordability of minimumpackages of financial services, possibly special criteria related to documentation requirements for low income persons, or special financial products designed for low income persons, could be important for raising access. 102. Next in importance to income in terms of individual characteristics associated with access is education. One suggestion to follow up on this finding is that programs of financial education and awareness may be importantfor low income persons,even if overall educational policies, just as overall growth raising policies, may lie outside the scope of the financial sector policymaker. 103. Equally important are the implications of the analysis with regard to the importance of location. The analysis suggests that although location is important and there do seem to be some regions and locations which are underserved, muchof the difference with regard to locationcan be attributed to levels of economic activity and sparseness of population. Thus,policies which rely largely on the geographical equalization of services are not likely to be adequately effective in targeting access. This i s a significant comment in the context of Brazil where major efforts to expand access have focused on ensuring that there is at least one financial point of service (branch, service post or correspondent), in each municipality. Such policies cannot ensure that users of financial services in such locations are indeed the poorer segments, indeed the better off persons even in these neighborhoods are likely to have been more successful in achieving access. 104. Another important inference with regard to locational policies is that regions of the country or even municipalities, are perhaps not the best points of focus for the expansion of access. Rather, location defined at a micro level in terms of neighborhoods, with service expansion targetedat areas or parts of a city with specific microeconomic characteristics (e.g. high concentrations of low income housing), should be targeted. 105. Eventually, factors related to location, though initially significant in the analysis, lose significance once socio-economic factors such as income and education are included in the analysis. Additionally, other socio-economic characteristics, such as employment related variables, gender and role in household were also shown to be significant in the determination of many measures of access. One implication of this is that these factors have assumed importance in the absence of sufficient direct information on the financial behavior and creditworthiness of individuals. Wealth and collateral were also important in determining access to credit, as these may help with loan recovery. This finding could also suggest that in the absence of adequate enforcement mechanisms for financial claims, the factors noted above could serve as indicators of financial reliability. The implications are therefore that policies which can directly expand client information would expand access, such as the sharing of `positive' Page 42 Part 1. Assessing Access information in some forms of credit registries. Equally, procedures which streamline the use of guarantees,secured credit, bankruptcy or recovery could also help to expand access. 106. When we analyzed the type of financial institution, given that there are some specialized services where either public or private banks seem to dominate (e.g., public banks for payments or housing), a suggestion to emerge is that greater competition in these services would appear to be desirable, thus for example inviting in private services in housing or in payments (perhaps through an alteration in the present domination of housing finance by one large public institution, or through more competition in the allocation of correspondent franchises). 107. Finally, although the analysis did suggest that persons with lower levels of income tended to use public banks proportionately more, it also showed that along a broad spectrum of services, the roles of these institutions could be largely substitutable. Thus, for both public and private banks, deposit and credit services are available to lower income individuals, but as persons become better off, they tend to prefer private services for both types of transactions. This suggests that private banks could do more to refocus theirproducts, image and services in terms of outreach. Brazil:Access to Financial Services 108. Access to financial services should improve consumers' welfare and producers' productivity. The traditional neo-classical approach to demonstrate the advantagesof adding financial intermediation to the choices faced by economic agents has been to compare the "reachable (feasible) utility sets" under two, highly simplified, hypothetical scenarios: one without financial markets and another in which agents can lend (save) and borrow (spend beyond their initial endowment). In the first case, agents are able only to transfer physical goods from one period to the next (e.g., trading grain), and are not able to lend or borrow except also in physical goods (e.g., seeds). 109, The addition of financial assets enables agents to reach two objectives: (i) inter-temporal make transfers of income and consumption; and (ii)provide an objective reference to the cost of capital ("opportunity cost"), which permits benefits from allowing market determined interest rates to optimally allocate resources over time. To the extent that these objectives are reached, agents maximize their utility reaching higher utility functions. FigureAl.1 Income,Consumptionin Period2 Without any form of income transfers: Individuals optimize by consuming what they earn ineach period. Cln= YIaand C: = Y;. With physical income transfers: Individuals optimize by choosing consumption Clb 1....................... and Czb not equal to the income endowments o f Yla and Y;. 0 Clh Yj'l Income, Consumptionin Period 1 110. In Figure Al.1 we present a simplified two period model to illustrate the assertions (i) (ii) and made above. An agent earns income in the two periods, Yla and Y:, and derives utility U, by consuming C,, and C t in the same periods respectively. In the absence of any form of transfers, the individual 58Based on Robert Cobbaut, Thkorie FinanciBre, Collection Gestion, SCrie: Politique GCnCrale, Finance et Marketing, Economica, Paris, 1977, pages 37-40. Page 44 Part 1. Assessina Access must choose CIa= Y1,and C; = Y2, (and hence operate at point "a") to maximize utility. If however the person is permitted to save (but not borrow) from income in period 1for additional budget in period 2, the budget constraint for the individual would become the curve "ae". The concavity of the curve reflects the diminishingmarginal utility of greater amounts of income transfer (reflecting the shape of a standard Production Possibility Frontier). The same economic agent now has the option of choosing an optimal point "b" (tangency between the indifference curve and the budget constraint) that allows the person a utility level Ub> U, by selecting consumption levels of Clb and C: in periods 1 and 2 respectively. This new optimal choice between income flows and consumption flows imply a saving of Y1"- Clb in the first period that results in the financing of an additional consumption of C: - Y: in the secondperiod. Figure Al.2 Income,ConsumptioninPeriod2 With financial instruments N t allowing savings only: Individuals optimize by choosing consumption CIc and C i not equal to the income endowments of Y," and Y2a. U" U h u, 7 Income,Consumptionin Period 1 111. In Figure A1.2, there is a new scenario wherefinancial instruments that allow savings (though no borrowings) are available. In such a case, the new budget constraint will be governed by the market rate of interest on savings. The new budget constraint, the straight line aN allows the individual to optimize cumulative utility at point "c" where U, > Ub> U, by saving YIa - ClCin the first period resultingin an additional consumption of C; - Y2, inthe secondperiod. 112. In addition to the possibly of higher utility to the consumer, the matching of the individual's marginal rate of substitution (the slope of the indifference curve at "c") to the market rate of return (the slope of aN) provides an objective reference to the cost of capital - the "opportunity cost". It i s thereby a more efficient outcome than achieving U, at point "a". 113. Next, expanding the provisions of financial markets, if financial instruments allow both savings as well as borrowings, the budget constraint in Figure A1.2 would be extended to "MN' in Figure A1.3 below. Brazil: Access io Financial Services Figure A1.3 Income,Consumptionin Period2 t With financial instruments N I allowing savingsonly: Individualsoptimizeby choosingconsumptionClC and C; not equal to the incomeendowmentsof Y," and Y;. Cz' C2 0 I Income,Consumptionin Period 1 114. The implication of being able to borrow as well as save in both periods further expands the range of feasible options. For individuals with marked preference for saving in first period (as shown in the shape of indifference curves above) and choosing an optimal point in the range "a", this additional opportunity may not alter their decisions. But for individuals who prefer an optimal point in the range "aM", the potential to borrow in period 1 offers scope to both maximize utility and achieve optimal allocation between periods basedon opportunity cost consideration. 115. Aggregating this analysis for the entire market, the rate of interest in a competitive financial market can be expected to the equilibrating level that balances the net savers with the net borrowers in period 1, and hence make the mechanism stable and self-sustaining. 116. The financial instruments thus achieve the multiple objectives of (i) enabling intertemporal asset transfers (ii) ina manner that is economically efficient for the individuals and hence the collective society (iii) isrobust. and Page 46 Part f . Assessing Access -NOTE ON ESTI~ATION~URBAN FINANCIAL SURVEY 117. This technical note describes the statistical tests used to choose the independent variables of the models and then describes the formulation of the functional forms of models that were subsequently used in the analysis. The econometric models used to determine the measures of access are also described below. Chi-square tests of Independence 118. The chi-square "goodness-of-fit" test evaluates the hypothesis that the rows and columns in a two-way table are mutually statistically independent. With dichotomous variables, the chi-square test is similar to the z-test that tests the equality of proportions (or the frequency of incidence) between two variables using large-sample statistics. But unlike the z-test, the chi-square test is not restricted to comparisons of only two sample proportions. For variables with multiple categories the chi-square test is the most common technique to evaluate the degree of dependence between rows and columns of the categorized data. 119. Under the null hypothesis that the distribution of observations in the rows and columns is mutually independent (Le. follow a specified distribution), the chi-square statistic computed indicates the degree to which the distribution variables between the rows and columns differ significantly from the specified norm. 120. The chi-square statistic for data with (1-1)(J-1) degrees of freedom is computed as: x2 =xx (nu-mu)' i j mii where: nij= number of observations in i"row andj" column. Inour case, nij is just a count of frequency such that a,.= cJ n,, and n = cI n,, ,i= ..,Iandj=1,...,J 1,. n = 7, ai, (overall sum) I .I 121. Prior to the regression analysis, chi-square tests were conducted to confirm that the distribution pattern of the dependent variable was different from those that were to be used as the explanatory variables. For example, we checked whether the probability of a woman having a bank account was equal to the probability of a man having a bank account. Only in the cases where the null hypothesis could be rejected were the associated independent variables deemed to be good control or explanatory factors for inclusion inthe regression analysis., EconometricModels 122. Qualitative response models were estimated using models for binary response and for multinomial choices. Models of binary responses were designed to take the value 0 for persons who did Brazil:Access to FinancialServices not fulfill the given criterion (such as having access to financial institutions) and a value of 1for persons who did satisfy the specified criterion. Such variables were associated with additional information that captured personal characteristics, employment information and location information. Models of multinomial responses associated multivariate variables that followed a logical sequence of progression - such as don't save money, have money in a saving account, haye money in a time deposit account to some independent variables. The second sets of models were thus an expansion of the first set. 123. In the binary model, we attempted to estimate E[T/X] where X is the vector of independent variables that have some relation with I,and E i s the expectation operator. Assuming that vector "I" follows a Bernoulli distribution, we usedthe result: ELI XI =Pr (I=l X) =F(X:B I I 124. where F(.)denotes the cumulative distribution function of "I"conditional on the values of X, and i s an array of parameters which determines this function. Inthis analysis, we assumedthat X comprised of independent and identically distributed variables and that F(.) was linear in its arguments, resulting in the model: F ( X :p)=x'p+& 125. We also assumedthat 6 was independent of X. This allowed us to use a logit model for regression where the F(.)representeda cumulative distribution function of a logistic nature such that: 126. Equation (1) could be transformed into Log[P(I=l /X)/(l-P(I=l/X)]= x'p or simply h(I) =X'Q(2) 127. Equation (2) could then be estimated by Maximum Likelihood Estimation techniques. Unlike conventional linear regression models, the coefficients of the regressions cannot be interpreted as the partial derivatives of the dependent variable with respect to the respective explanatory. The partial effects were computed by: a) Estimating the probability of access to financial services of an individual with value J for a characteristic k, which is b) Estimating the average probability for every J. The difference between the averages calculated for J and J+l represented the marginal effect of increasing the value of the explanatory variable k of J to J+1. Page 48 Part 1. Assessing Access 128. Given the limitations of the statistical software used for the analysis which does not calculate the marginal effects of logit regressions automatically, the marginal effects from probit estimation was used as an approximation of the same results. The probit model i s similar to logit model, with F(xP)= @(xp),where CD is the standardcumulative normal distribution. 129. For analysis of multinomial models, multinomial logit estimation is the most widely used technique. In a choice set B={1,...,m}, where xB=(xl, ...xm)denotes the array of observed attributes of the available alternatives, vector yB= (yl, ..,y,) has a multivariate probability distribution: j E B with mean p'xBlike inthe binomial case. 130. Using Maximum Likelihood Estimation techniques, the sequential series of outcomes in y could be estimated in terms of the explanatory variables X. These results could then be used to estimate the coefficients corresponding to each outcome category such as: Pr(y =1)= expi + ,XP' + ,XP' Pr(y = 2) = expi + +ed3 eXP' Pr(y =3) = + +eXP3 131. The limitation of this technique is that the estimated solutions for coefficients (p', p2,p') neednot be unique. Hence by arbitrarily setting one of the 0's at the benchmark value of 0, the other coefficients can then be interpreted to measure the change relative to the control group. 132. In this study both types of models and estimation: binomial respondents models, logit and probit and multinomial response models, logit multinomial were used. Also used were linear models to predict the volume of credit requested by individuals and the volume of credit approved in each request. The logarithm of the amount of money request (or approved) for loans was regressed on identifiable personal characteristics such as age, education level, income, etc. The relation estimated was of the type: where g(X) = Po+ Xlpi + ...+ X E[ln(Y) XI =g(x) k P k+ E I 3 and Y representedthe volume of credit requested(approved) and X the characteristics variables Part 2. fxmndina Microfinance PART 2. EXPANDINGMICROFINANCE 1. MICROCREDIT, AND POVERTY:A NEWPARADIGM? ACCESS 1. As discussed inPart 1of this study, appropriate targeting of policies is neededto strengthen financial access, to support those groups, especially among the poor, where services are most needed. How this is to be accomplished has perplexed policymakers in Brazil, as elsewhere. Largely in the last two decades, excitement has grown worldwide in terms of a new institutional form of delivery of financial services to such communities; the `microfinance revolution'.' Loans are made available, in very small amounts, often under a new paradigm; unsubsidized lending at market rates, typically with low levels of formality and limited requirements of collateral, often to particularly vulnerable groups, such as village women. Repayment is undertaken frequently, and rates of repayment in many microfinance ventures are cited to be high. Many microfinance ventures also offer deposit taking services. Although many notions of microfinance exist, the small size of transactions involved, and delivery through channels other than formal banks are core themes. Additionally microfinance has sometimes been defined as the delivery of such services by financial institutions which are themselves small in size and informal in nature. Today, microfinance institutions (MFIs) are reaching out to over 9 million borrowers and 29 million savers worldwide, in over 53 countries.2 Microfinance is being considered as a preferred vehicle for extending access to the poorest in many countries. 2. The importance of microcredit for the self employed and for small entrepreneurs, and thus for jobs, is also increasingly recognized. Microentrepreneurs of Brazil account for the vast majority of all firms and contribute substantially to employment and GDP. According to recent estimates, 98 percent of Brazil's 4.1 million micro and small enterprises account for 45 percent of formal employment and more than 60 percent of urban jobs. Directly or indirectly, they provide the primary source of income to almost 60 million people and generate nearly 20 percent of GDP.3Firms with up to four employees account for 57 ' Global recognition was given to the potential role of microfinance in addressing poverty alleviation at the Microcredit Summit in Washington D C in 1997, when summit campaign members representing some 1,500 institution members prepared a Declaration undertaking to extend credit to 100million of the poorest by 2005. As of December 2001, some 2200 institutions had accessed 55 million clients, of whom 27 million were among the poorest. See www.microcreditsummit.org for details. 2 Microbanking Bulletin, November 2002, based on its 147 participating institutions, available through www.cgap.org. There are now several agencies and websites which provide rich information on microfinance, and the publications of the CGAP or Consultative Group to Assist the Poorest, including the above, the Focus Notes, and its Microfinance Gateway. Finance for the Poor is a quarterly periodical of the Asian Development Bank and the Microenterprise Americas magazine i s a publication o f the Interamerican Development Bank.PlanetFinance, Womens' World Banking and other agencies such as Accion International are rich private sources. The literature on the experience of microfinance is huge; surveys are available in Jonathan Morduch (1999), `The Microfinance Promise'; Marguerite Robinson (2001), `The Microfinance Revolution'; among others. Selected data are from the website of SEBRAE (January 2004). Brazil: Access to Financial Services percent of all firms in industry, 74 percent in services and 82 percent incomrner~e.~ addition, there are In an estimated 14 million informal micro and small enterprises in BraziL5Although the majority of such firms are located in the South and Southeast of Brazil, the impoverished North and Northeast have a higher share of new micro-enterprises, Beyond microenterprises, the need for loans for self-employment by the poorest is also widely acknowledged. Such ventures need working capital to survive, and investment funds to grow. 3. Related to the theme of provision of small loans, a second group of financial institutions catering to these needs is the community of credit cooperatives. Considerably older than microfinance, the concept of the credit cooperative dates from the 19thcentury, involving persons with a similar trade, employment or location to pool resources in a democratic economic organization with the objective of serving the common interests of the group. Credit cooperatives were first formed a century ago in the southern states of Brazil. Credit Unions, which have grown out from the cooperative system, are a mature alternative to banks in many developing countries worldwide today.` As in the case of microfinance, a key characteristic is the small size of transactions handled. Both deposit and loan services are offered, as loans are extended basedon the funds accumulated from members, inmost cases. 4. Both microfinance institutions and credit cooperatives have grown in Brazil over the past decade. Although the growth of the microfinance sector has been more rapid, the cooperative sector still dwarfs the microfinance sector in terms of size. The number of credit cooperatives increased from around 950 in 1994 to some 1,400 by end 2002, while their assets grew in parallel from around R$2.5 billion to R$11.5 billion, and loans grew from R$1.4 billion to R$4.6 billion. Members of the credit coops could be broadly estimated at around 1.5 million persons.' Many microfinance institutions are largely beyond the purview of formal financial system supervision, but estimates suggest that the total number of clients served increased from around 3,000 in 1995 to around 160,000 by end-2001, while the active loan portfolio grew to some R$140 million. This i s small compared to the credit cooperative system, and also small compared to other countries in the region, once adjusted for country size. Peru and Bolivia had estimated microfinance clienteles of around 185,000 and 380,000 at end 2001, respectively, while small countries such as Nicaraguaand El Salvador also had over 80,000 microfinance clients.8 5. There have been significant recent efforts to detail and analyze Brazil's microfinance sector in the last five years. The contribution of the present chapter lies in identifying and analyzing factors leading to the increase of both the microfinance and the credit cooperative sectors in Brazil and evaluating their future sustainability and potential for contributing to increased access. Alternative models under which there could be an increased potential for expanded access, and preconditions for such models, are evaluated, in 4 Sebrae (2002), basedon IBGEdata from 1994. These definitions are based on the number o f employees. The IBGE(Brazilian Institute for Geography and Statistics) defines micro-enterprises as those with upto 19 employees in industry and up to 9 employees inservices and commerce, and a small enterprise as those with up to 99 employees in industry and 49 employees in services and commerce. Alternative definitions, based on sales turnover are also used. 'Departamento Intersindical de Estatisticas e Estudos SocioeconBmicos DIEESE (2002). These numbers do not include an estimated 4 million small rural enterprises. Inall it is estimated that the majority of Brazil's 70 million `economically active persons work in or are linked to small enterprises. The National Credit Union Administration of the USA at www.ncua.org, and the World Council of Credit Unions, 'atThe Woccu.org provide detailed historic as well as comparative experience. 753 members of the SICOOB network, which accounted for over half Brazil's cooperatives in 2002, had 972,000 members. According to the Central Bank of Brazil, there were a total of 1.5 million total coop members at end 2001. World Bank estimates and BNDES / DAI `Understanding Microfinance in the Brazilian Context' Rio de Janeiro, 2002. Part 2. Expanding~ i c r ~ ~ n a n c ~ terms of options for policy makers and for the institutions themselves.' The present analysis is based largely on the period ending in 2002. New measures taken by the government in 2003 are briefly described, (Page 57), however, the outcome of such measures has yet to be assessedindetail. 1015 1 088 1183 1,235 1,333 1,374 621 1,366 1,728 2,203 2,816 3,736 4,562 1,730 6,881 3,339 Equity (R$million) 1.1 4.4 9.3 Note: 1 Data refers to microfinance institutionsSuDDortedbv BNDES. Numbers in parentheses show numbersnet of SCMs (Microcred, . . Socialcred and Rotula). Source:World Bank staff estimates basedon data providedby the Central Bank of Brazil, Bancodo Nordeste do Brasiland BNDES. 6. Key observations to emerge from the following sections, for Brazil's microfinance sector, are: 3 There was a remarkable recent acceleration in Brazil's microfinance sectorfrom the end of the 1990s, due to a series of factors: new political paradigms leading to a series of legal and regulatory changesfavoring the expansionof microfinance. Second, a major new large scale microcredit experiment the `CrediAmigo'program which incorporates many principles of good microfinance practice. And third, active governmentfinancial backing to emerging microfinance institutions, through credit offered by the BNDES bank. a Despite this rapid acceleration, microfinance penetration in Brazil today remains low, especially compared to neighboring countries. Yet paradoxically, the portfolio growth of small microfinance institutions is not high. a Microfinance expansionremainsgovernmentled,throughthesubstantialpresenceof thepublic banks BNB and BNDES, Microfinance institutions have also relied substantially on relatively low cost government lines of credit extendedat below comparable market rates. Recent changes introduced by the The teamof persons contributingto this chapter includes McdonaldBenjamin, Susana Sanchez and Sophie Sirtaine (Bank staff), and Alfred0 Ebentrich,Lydie Ehouman, RicardoGonGalves, Adam ParsonsandRobertVogel (consultants). Brazil: Access to Financial Services government include microfinance operations destinedfor low income eamers and small businesses, to be funded by a minimum 2 percent of thefinancial institution's sight deposits. The new measures create the possibility of additional reserve requirementsfor banks which do not participate in certain programs of access. While such additional reserve requirements would be small, as implemented in tandem with other programs of directed credit, they would add to the already high implicit tax burden of banks and could work against the desired broader development of the microfinance sector. 3 Conversion to the new institutional forms, although increasing access to wholesale finding, is rendered less attractive by the implications of new regulatory requirements, both non-prudential reporting as well as prudential requirements. There are also tax implications which would subject the new institutions to thefill range offinancial sector taxation. 3 Growth indicators for the portfolios of MFIs supported by BNDES suggest nominal annual growth of 22percent a year; much lowerfor larger MFls. Theslow growth of Brazil's independent MFIs (other than CrediAmigo) despite apparent large unmet demand is paradoxical. Somefactors responsible could include the dificulties encountered by independent MFIs in expanding client outreach without a branch network or savings product on offer to support their efforts. Partnerships with large banks can help build a client culture, provide a branch network and also, reduce competitionfrom similar products offered by the banks, such as consumer loans. 3 In terms of performance, indicators of loans overdue are average (possibly with some trend increase for BNDES supported MFls) and loan provisioning may be somewhat low. But financial sustainability has been good, perhaps due to low costfinding. Eficiency indicators in terms of loans per loan officer though low have been rising. lmpact indicators in terms of loan size suggest that outreach is reaching small client segments. 3 Some so-called microfinance institutions in Brazil, run mainly by municipal governments to achieve social objectives such as employment, provide highly subsidized credit for this purpose. Such price-distorting subsidies could affect the expansion of market oriented microfinance. Other transfer mechanisms (eg lump-sum transfers) could be investigated instead, and a clearer distinction could be drawn between these activities and microfinance. Such distinctions between income transfers and the market based extension of credit may need to be made more extensively in the context of the new microfinance measures of 2003. 3 Brazil'sformal microfinance instiutuions, known as its microfinance credit societies (Sociedades de Crkdito ao Microempreendedor, or SCMs) are subject to regulatory and reporting requirements which may be more rigorous than comparator models, in view of their non-deposit taking scope of activities. However should Brazil contemplate the introduction of savings products or deposit taking facilities, further regulatoryfine tuning may be needed. 7. Suggestions for the future expansion of the MFIsector, basedon these findings, follow below: =, An impact evaluation of the new microfinance measures should be undertaken, to assess their performance in terms of outreach and sustainability as well as their impact on participating microfinance institutions and banks. 3 Brazil's MFls could make more extensive use of recognized best practice methods for microfinance; although extensively used at CrediAmigo, small MFIs do not appear to havefollowed these principles. Part 2. Expandinq Microfinance 3 Both to attract more wholesale bank funding, and to better incorporate good microfinance practice in its MFIs, Brazil could consider more emphasis on organizational forms which allow partnerships with large banks, which could help with training loan ofSicers, tracking quality indicators and incorporating good practice. In this context, the BNDES bank could consider a more indirect support to MFIs than its present model, through theprovision of wholesale credit to commercial banks which then liaise with self help groups either directly or through NGOs. Recent moves to establish microfinance operations in commercial banks, such as the Banco Popular, are a step in this direction. In addition to the Banco Popular, new programs are also being introduced by public banks such as the Caixa and BNB. Care should be taken to ensure that such operations are undertaken on a sound basis. 3 In the future, ifsuflcient maturity and sustainability is achieved, Brazil's microfinance sector could also contemplate deposit taking not only to expand its funding base if large scale roll-out is to be contemplated, but also to offer needed savingsproducts and expand client inte$ace. Brazil does not allow any form of deposit takingfor its MFIs, and wholesalefunding from the banking system has been limited. However, at present, the MFI sector may not be adequately prepared for such a major step, which inter alia would also require regulatory reform. 3 Microfinance in Brazil may need a separate regulatory window, but the reasonsfor the current multiple regulatory windows of ONGs, OSCIPs and SCMs may be driven partially by the need to counter other regulations which need overhaul, such as the Usury law. In the future, a simplification of this framework may be desirable. Harmonizing (and perhaps eventually integrating) the framework for regulated MFIs with the banking system may also be considered. 3 Within current regulatory norms, SCMs have very low entry capital requirements both relative to other countries and relative to Brazil's own banking system. This couldpose a problemfor supervision if numbers of SCMs expand significantly. If separate regulatory windows are maintained, models of shared, delegated or independent supervision could be considered. =? Conversely leverage ratios, atfive times net worth, though not strictly comparable, appear much higher than bank capital adequacy ratios, at eleven percent of risk weighted assets, and higher than in many other countries, whichfrequently use the same capital adequacy ratios asfor banks. Especially in view of thepresent non-deposit taking nature of Brazil's MFIs, this could be reviewed. 8. With regard to credit cooperatives, principalfindingsand suggestions to emerge are: 3 The Brazilian credit cooperative system has made important advances, particularly since the mid-1990s, with the permission to establish cooperative banks. These have allowedfor rapid expansion of credit cooperatives combined with growing professionalism in cooperative management, information, accounting, staff training and incentives, and internally administered prudential standards. New measures introduced in 2003 havefurther favored the expansion of cooperatives, expanding membership in remote areas by allowing `open admission' credit cooperatives to be established, and by creating a more level playing field with banks through the harmonization of capital requirements. 3 Permission to establish central credit cooperatives has existed since 1971.'' Incentives to individual cooperatives tofederate are high as leverage ratiosfor federated cooperatives are set at twice the levels of individual cooperatives. This may mitigate the extremely low start-up capital requirements, which are even lower than for microfinance institutions. This may provide a safe way to encourage loArticle 8 of Law 5764 of 1971.However, the first Central cooperative to commenceits operations was not establisheduntil October 1980. Brazil: Access to FinancialServices cooperatives to combine the benefits of local knowledge in mitigating moral hazard and adverse selection problems with the advantages of risk reduction and diversification through networking. s In turn,federated or central cooperative members benefit through an expanded range offinancial services (such as internet banking, credit cards, CPRs, insurance and custodial services) in addition to more accessible loan and savings products, and in most cases they have been able to count on more reliable prudential oversight of the resources they have invested in cooperatives. 3 Nevertheless, the legal and regulatory impediments of earlier decades (notably in the 1960s and 1970s) resulted in a system that still lags behind that in other countries. Penetration ratios remain extremely low by intemational comparisons. One constraint on expansion has been the high liquidity requirement, required by apex cooperatives, considerably higher than prudential norms advocated by international standards. However, new legislations introduced in July, 2003 have harmonized capital requirementsfor credit cooperatives and banks. s Other prudential regulations recently introduced could improve the soundness of the sector but will raise supervision resource requirements, notably the expansion to open forms of organization in remote regions. Non-prudential disclosure-based advances on the regulatory side comprise the inclusion of cooperative clients with loans above R$5,000 in the Central Bank's credit information system (Central do Risco), and reviews of cooperatives Boards of Directors. 3 Cooperatives, especially rural cooperatives, have also served to channel directed credit to the rural sector, often at below market rates, sometimes at unattractive spreads and with regulatory dispensations in terms of leverage ratios. Although the dispensations in terms of higher leverage ended in 2003, there is still the risk that the cooperative sector could be viewed as key to the implementation of government credit programs. 9. In this light, this study proposes the following recommendations for consideration by Brazilian policymakers and practitioners. 3 Provide incentives for the establishment of a liquidity facility for credit cooperatives to reduce opportunity costs relative to commercial banks backed by the FGC. One option to be studied would be to encourage the establishment of a privately managed central liquidity facility for credit cooperatives, run by representatives of cooperatives under Central Bank supervision and standards for liquidity support, capitalized via a percentage of deposits. In turn, liquidity requirements on sight and time deposits could be eased, as the degree of capitalization of the facility increases. After evaluating pe$ormance, consideration could be given to public backing for thefacility. 3 Apply similar financial regulation for credit cooperatives as for other financial institutions. As financial institutions, they should asfar as possible be under the same standards of regulation applied to other financial institutions, to promote competition and reduce perceived higher risk associated with the lack to access to a liquidity facility and deposit insurance. However some special features which would digerentiate their regulatory framework, need to be taken into consideration. 3 Review norms, regulations and incentives governing public rural credit programs channeled through credit cooperatives. Policies such as those that doubled the leverage permitted in the event of financing using public rural credit programs such as PRONAF reinforced incentives which undermined the cooperative spirit and the sustainability of cooperatives. These regulations ran counter to what is in general a prudent regulatory framework and the lifting of thisfacility with new regulations introduced in June 2003 was a welcome move. Part 2. Expanding Microfinance In addition, consideration should be given to reviewing the extent to which the relative market positions of Banco do Brasil versus smaller rural credit cooperatives explains their relative shares of the spread on PRONAF loans, and what measures can be taken to better relate spread to credit risk and administrative expenditures on suchpublic transfers. This undermines members ' sense of ownership and reduces incentives to repay. 3 Review present capital standards, which are much lower than for microfinance institutions (SCMs)and also compared to microfinance institutions in other countries. Although thesefacilitate broad access to cooperative financial services they can also imply a proliferation of small formal financial institutions, with high costs to the supervisory entity, particularly in a country the size of Brazil. Maintain a prudent balance between direct and delegated oversight of credit cooperatives. The present system of delegation appears to be beneficial though if the sector grows further formalized delegation may be needed and a choice of appropriateframework for this will need to be selected. 10. Finally, for both the microfinance and credit cooperative sectors, it i s suggestedthat: 3 A number of technological innovations are available which can greatly enhanceproductivity and delivery capacity and reduce risk in microfinance transactions. An exploration of thesefor adaptations in the Brazilian context is suggested. s The collection and maintenance offurther information on a regular basis should be strengthened. Easing the present shortage of reliable, consolidated data on these sectors would be a first step to strengthening oversight. More detailed information could usefully be obtained for both supervisory and market development purposes on the performance of microfinance institutions and credit cooperatives and determinants of such performance (for example, data on administrative cost structures in different geographic contexts, risk profiles of clients by location and type of enterprises, etc.). 2. MICROFINANCE INBRAZIL: EVOLUTION AND STATUS1' 11. Despite rapid growth in the microfinance sector in neighboring countries such as Bolivia and Peru in the 1980s and 1990s, Brazil's microfinance sector remained dormant through most of this period, for a combination of reasons. First, a strong belief was held that the public sector was key in terms of ensuring financial access to the poor and a series of specially designed programs proliferated, particularly for rural access. The proportion of credit which was `directed' towards specific sectors was estimated to be as high as 54 percent, as recently as mid-2000, and was still as high as 40 percent in March 2003. Second, the regulatory environment for microfinance was limiting; as a result microfinance was mostly offered through non governmental organizations, which relied largely upon access to donor funds for onlending; thus Brazil's microfinance sector was largely driven by philanthrophic partnerships with foreign donor organizations. Interest rate caps based on the Usury law violated the principles of microfinance lending. Third, there was little `demonstration effect', in terms of successful large scale in country programs of 11Considerable recent literature on Brazil's microfinance sector is now available, beginning from Schonberger (2001), "Microfinance Prospects inBrazil", Goldmark and Vechina (2000),`A Situa@o no Brasil, BNDES, OrganizaGdoInternacional do Trabalo Goldmark,, Pockross and Vechina (2000),`The Status of Microfinance in I, Brazil', Goldmark, Nichter and Fiori (2002), `Understanding Microfinance inthe Brazilian Context', Central Bank TCcnica, Central Bank o f Brazil (2002),`As Cooperativasde Crkdito e o Banco Central do Brasil' - Nota TCcnica, of Brazil (2002), `Democratizapio do Crkdito no Brad - Principais Desafios - AtuaqEo do Banco Central' - Nota Mezzera (2002), `Microcredit inBrazil: The Gap Between Supply and Demand', and UNDP (2001), paper for Caixa EconGmica Federal on `Status o f Microfinance inBrazil', mimeo, Brasilia. Brazil: Access to Financial Services microfinance. Experience with microfinance was largely anecdotal and limited. The acceleration in microfinance in Brazil inrecent years can be attributed to a change in these circumstances, as discussed in the following sections. Microfinance:New Measuresintroducedin 2003 Recent Changes: Microfinance Regulations Law No. 10735of September 11,2003 (formerly Provisional Measure fJune25,2003) andCM Resolutions 3109 and 3128of July 24,2003 and October 30,2003 res te new vehicles microcredit available to low income earners and small businesses. The measuresdesigned to easeconditions for establishingbank accounts and settingup discussed further inPart 3 of this study. Under thenew regulations, microcredit operations will befundedby a deposits, or alternatively such deposits will be held as unremuneratedreserves. Interestrates of cre funds will be cappedat 2 percentper month. Loans will be limitedto R$600 for individuals and$1 businessmen. Loan terms will be no less than 120days (terms canbe smaller as long as rates are adjusted accordingly). Loanoriginating fees mustbe less than 2 percentfor individuals and4 percentfor small businesses. Individuals and enterpriseswith simplifiedaccounts (with maximum deposits of R$l,OOO for individuals and maximumdepositsof R$10,000 for enterprises), and low income individuals will be eligible for such loans I can be extendeddirectly by the banks or onlent through microfinance institutions. Recent Changes: Credit CooperativeRegulations CMNResolution 3106 of June 25,2003 eases membershipcriteria for credit cooperatives, hitherto only opento members of the same activity or profession, by allowing new credit cooperativesto form as `open admission' credit cooperatives inmunicipalities with up to 100,000 inhabitants.Existingcredit cooperatives, operating for more than three years, can transform into `open admission' credit cooperatives, but only in municipalities (or contiguous municipalities) with up to 750,000 inhabitants.Minimumcapital requirementsfor the million for entities locatedinmunicipalities inmetropolitan regionswith more than 1 million for the rest. Inthe North and Northeast, this requirement expandthe provisionof financial services in thinly populated ar reluctant to venture. The sameresolution removedthe additional leveragepermitted to public rural credit programs such as PRON ircular 3196of 2003) harmonizedcapital requirementsfor credit cooperatives of 11percent for cooperatives affiliatedto a `central banks) and 15 percentfor cooperatives not 2.1 Evolution of the Political Debate and the Legal Framework 12. Parallel to the accelerating intemational dialogue on the role and contributions of microfinace, as witnessed in the international Microcredit Summit o f early 1997, internal debate begain in Brazil on the potential contribution o f microcredit, through the ComunidudeSolidciriu,a public body created in 1995 to strengthen civil society initiatives in Brazil. The Communidade Solidhria adopted micro-credit its principal theme o f political debate in 1997-98, and resuming this theme in 2001. Largely through the efforts o f working groups organized in this context, a number o f legal impediments to the expansion o f micro-credit were identified. Based on the recommendations o f these groups, dramatic improvements in the legal framework for microenterprises have been introduced over the last five years. Parf 2. Expanding Microfinance 13. New forms of micro-credit institutions, excluded from the scope of the usury law were established by regulation.l2Prior to the changes, microfinance was undertaken largely by non-profit organizations (ONG, or Organizqfio Niio Governamental),and these were constrained by the one percent per month ceiling on interest rates imposed by Brazil's usury law, which has been meaningless in Brazil's high inflation periods, and is also limiting in the case of microfinanceloans, which traditionally are offered for short periods of weeks or months, at higher interest rates than loans in the formal and large scale sectors. NGOs were also constrained in terms of access to capital, to grants or credit lines from government or foreign donors. Two new legal entities were created in 1999, with considerably more flexibility in terms of both funding and lending, and outside the purview of the Usury law.13 These were (i)the OSCPs (Organizqfio da Sociedade Civil de Interesse Pu'blico) or Public Interest Civil Societies, and (ii) SCMs (Sociedadesde Crkdito ao Microempreendedor)or Microenterprise Credit Societies.I4 14. The first of these organizational forms, the OSCIPs, can have a range of objectives, which are not limited to microfinance. OSCIPs exist in many fields such as science, education, research, etc. Like NGOs, they are deemed to be not-for-profit organizations. Importantly, OSCIPs can `sign Cooperation Agreements with the public sector to implement activities and projects of public interest', which implies that they could have access to public sector funds, in addition to donor funds and owner/sponsor funds. OSCIPs are not able to mobilize deposits from the public. OSCIPs are subject to a low level of non- prudential regulatory requirements, such as some reporting requirements to the Ministry of Justice, but are not required to comply with any prudentialregulations. 15. The second institutional form created, the SCM, i s much closer to a microcredit society, as they exist in many other c~untries.'~SCMs are formal for-profit financial entities regulated by the Central Bank of Brazil. SCMs can take loans or lines of credit from foreign or domestic financial institutions, due to regulations issued in July 2001, where their access to funding was further widened, together with permission to use instruments such as fiduciary alienation for extending credit.l6 However, issuingdebt or mobilizing deposits remain restricted. Subsequent regulations issued in July 2001 maintained various prudential norms; including operational limits on capital, leverage and risk. SCMs are requiredto comply with minimumstart-up capital, ongoing capital adequacy and limits on risk concentration. The regulations issued in July 2001 also permit SCMs, like banks, to operate through Microcredit Service Points (Pusto de Atendimento de Microcr&dito),and require them to submit information on their lending operations to the Central Banks credit risk center - the Central de Risco de Crkdito - which in principle could permit them inthe future to obtain greatly enhancedcredit information on prospectiveclients.l7 16. These new institutional forms are clearly a major step forward for microcredit in Brazil. Some existing NGOs, which are microfinance institutions in practice, are considering modifying their legal l2A more extensivediscussionof the regulatory framework is in Part 4 of this chapter. l3Medida Provisdria 1914-4of June 28, 1999excluded both OSCIPsand SCMs from the Usury law. 14Law 9790 of March 23, 1999, which came to be popularly known as the New Law of the Tertiary Sector, establishedthe OSCIP. DecreeNo 3100of June 30, 1999provided details on its authorization, functioning and supervision. Reporting requirementsare minimal, but for lines of credit through public entities, external audits are required.For details see especially `OSCIP - Organiza@o da SociedadeCivil de InteressePdblico - A Lei 9.790/99 conlo Alternativa para o Terceiro Setor. Comunidade Solidhria'. Brasilia, July 2001. l5CentralBank Resolution 2627 of August 2, 1999establishedthe SCM. Guidelines for its authorization and functioning were issuedrapidly after (Central Bank circular 2915 of August 5, 1999) and further circulars establishedreporting requirementsto the Central Bank as well as public disclosure requirements for financial statements (Circular No. 2964 of 3 February 2000) and a standardizedchart of accounts in COSIF (Carta-Circular No. 2898 of the Central Bank of 29 February 2000). 16Law 10194of February2001 extendedfunding sources to includethe SEBRAE, and also, companiesregistered with the CVM, and investmentfunds. OSCIPs were permitted to own SCMs. l7Resolution 2874 of the National Monetary Council of 26 July 2001. Brazil:Access to Financial Services framework in order to benefit from the ability to tap external sources of finance, including in some cases splitting up existing organizations into two entities, an NGO rendering non-financial services (training and consulting) and an SCM for all financial operations. Ownership of an SCM by an OSCIP is also permitted. SCMs have grown rapidly in numbers, from around 6 in October 2001, to 36 by October 2002 (with 25 functioning and 11in a startup phase). 17. However as the preceding paragraph also shows, this arrangement, with multiple regulatory windows for different forms of microcredit institutions, is complex. Arguably some of these forms have been established to circumvent obstacles elsewhere in the regulatory environment (such as restraints on interest rates). In the future, the need for a range of different regulatory windows for microfinance can be investigated, and possibilities of simplifying the present framework can be explored (see Section 88 below). 18. SCMs have pointed to their heavy tax burden, in common with other regulated financial institutions. This is a deterrent to NGOs or OSCIPs considering converting to an SCM to expand funding sources. Documentary and reporting requirements have also been considered burdensome, although the reporting of credit operations to the Central de Risco could in future be beneficial to SCMs in terms of enhanced client screening.'* Third, an extension of the product range of SCMs to include individual or consumer credit, i s also debated. Most controversial, in terms of further measures to aid SCMs, is the question of whether they should be allowed to mobilize deposits and offer savings products. This issue, and experience in this regard inother countries, is discussedfurther inSection 2.3 below. 2.2 Establishmentof CrediAmigo A Major Player in Microfinance - 19. In 1997, enthused by its new management and by the new attention to microfinance in political spheres within Brazil, the Banco do Nordeste do Brasil, (BNB), a state-owned development bank with a mandate to promote economic development in the northeastern states, launched a large scale microfinance program, which has remained unrivalled in scale in Brazil and comprises a significant part of its microfinance expansion of the last decade. At end 2001, it served nearly 60 percent of MFIclient micro- entrepreneurs and held about 45 percent of their outstanding loans (Figure 2.1). 20. The program, known as CrediAmigo, benefited from the technical advice of ACCION International (a group with strong experience in solidarity group lending) as well as from the CGAP (Consultative Group to Assist the Poorest), and was financially supported by the World Bank.'' CrediAmigo was committed to incorporate best-practice principles emerging from successful microfinance institutions in the world (see Section 3.4 for a further discussion). These included (i) solidarity group lending; (ii)targeting the informal sector; (iii) charging interest rates highenough to provide a return on assets sufficient to permit financial sustainability; (iv) starting with small loan amounts and gradually escalating loan size with repeat loans; (v) amortizing loans regularly; (vi) offering incentives for regular repayment through discounts on the last installment, and penalizing borrowers if repayment falls behind schedule. The program also adopted the principles of product differentiation (discussed in greater detail in Part 3), separating its identity from BNB through a separate entrance or premises, for each branch office. Unlike some successful microfinance programs in other countries, there are no obligatory savings requirements (microfinance entities inBrazil cannot accept deposits). As discussed in Part of this chapter MFIs under separate and formal regulatory windows would havereporting requirements, which are harder to elaborateand require better informationsystems. There is a trade-off betweenthe supervisoryvalue of such reporting requirementsand the point where they becomeexcessiveand become a burden rather than a tool to improve the managingof their operations. I9Consultative Group to Assist the Poor Pari 2. ExDandina Microfinance Key LegalCharacteristicsof OSClPs and SCMs OSCIPS OSCIPs are non-profit organizations whose social objectives must fall within a specified list; They are exempt from Usury Law restrictions; They are subject to reporting requirements (including audited accounts of activities carried out under a Cooperation Agreement with a public entity) to the Ministry of Justice; Sources of financing are limited to donor funding, owners' investments and Government funding (including lines of credits from the Brazilian state-owned development bank BNDES); OSCIPs can own an SCM as a subsidiary SCMs SCMs are for-profit financial institutions deemed to be a part of the national financial system; They are thus exempt from Usury Law restrictions; They are also thus subject to the tax regime for financial institutions including income tax, CPMF, etc. As financial institutions, they are subject to reporting (non-audited accounts) and regulatory requirements of thc Central Bank of Brazil, which supervises them; They are subject to minimumcapital requirements (R$lOO,OOO); liquidity requirements; and restrictions on leverage (five times liquidassets); Sources of financing include all those allowed to OSCIPS, plus loanor credit lines from national and foreign financial institutions and from OSCIPs; They can extend loans and guaranteesto individuals and micro-enterprises, up to R$10,000 per client, for professional, commercial or industrial use only; they are prohibited from extending consumer loans; They can operate Micro-credit Servicing Points (Post0 de Atendimento de Microcrkdito) with full flexibility in terms o f location, opening hours, etc, and no additional capital requirement; They cannot collect deposits from the public, participate in the interbank deposit market, or issue securities for public offering. Figure2.1 Brazil:Growth of Loan Portfolioand Client Baseof Main MicrofinanceProviders _____ 101.8 -~ 80.3 7 2 . 0 ~,jI b i n m'crofinanceproviders -a- Credi Note: In 2002, June 2001 data was usedfor 3 MFls (Bancodo Povode Si0 Paulo,RioCredand SindCred),as 2002data are not available. Source: Bank staff estimatesbasedon data from BNDES.BNBandindividualinstitutionsdata. 21. CrediAmigo adopts additional good practices to safeguard its portfolio. A potential client microenterprise must be at least a year old with demonstrated cash flow potential. Its solidarity group members must know each other well but relatives are excluded. Each group elects a representative and adopts a name. Newly formed solidarity group undergo training by loan officers on group liability and loan characteristics. Initially, loans were exclusively for working capital purposes. Later, individual loans were introduced based on client records of at least two solidarity group loans. And by offering life Brazil:Access to Financial Services insurance policies to its borrowers during the term of its loans, CrediAmigo protects itself from eventual death of borrowers. Productsand Growth 22. CrediAmigo's minimum loan sizes, for first time loans, have ranged from around R$200 to R$700 and each subsequent loan may be 50 percent larger than the previous loan. Its average loan size in December 2002 was only R$605, confirming a focus on the poor. Female participation however corresponds to the population average, at 48 percent. Loan terms range from three to six months for solidarity group loans. Interest rates are higher than popular public bank enterprise loans but lower than consumer credit or even rates charged by some non-banks such as factoring companies. The program began with a 5 percent flat monthly rate, which has since been reduced to 3.5 percent. Initially, there was a high loan renewal rates, of about 85 percent, and an increasing number of new clients per loan officer." Low transaction costs and quick loan decisions contribute to this. Time between application and disbursement for first time borrowers is seven days, while for repeatedborrowers it is 24 hours. Links to BNB 23. CrediAmigo represents less than one percent of loan assets of BNB, and is being managed as an independent profit center with the goal of monitoring progress towards self-sustainability and eventual separation from BNB. As part of BNB, CrediAmigo is supervised by the Central Bank, and the question of how it could in future be separated is not clear. Should CrediAmigo opt for SCM status, it will be required to maintain appropriate capital adequacy. So far, it has received its funds on a 100 percent basis from BNB, which are indexed to the CDI rate, which closely tracks the Selic or interbank rate. However, if CrediAmigo had to raise funds from deposits (and comply with reserve requirements) or borrow at market rates from commercial banks or through the issue of its own paper, its funding costs may rise. In terms of operating costs, the program has made every effort to separate its costs from those of BNB and also to prorate the use of BNB resources. Such a separation i s however partially dependent on the capacity of BNB for monitoring cost centers. 24. Critically, CrediAmigo has kept costs down and enhanced client focus by separating its labor and staff from the BNB. Not only has this permitted the adoption of specialized training and offering of bonuses and incentives, but also, it has been relieved of the necessity of paying the wages which public sector workers and formal bank employees earn. However, such arrangements could face increased scrutiny. Its image differentiation in terms of its branches has also relieved it from obligations of complying with costly bank branch opening requirements. Portfolio Quality and Financial Performance 25. CrediAmigo initiated operations in 5 BNB branches inNovember 1997, and expanded in five months to 50 additional branches, but with ensuing poor portfolio quality. With a renewed commitment to focus on portfolio quality and productivity, CrediAmigo was able to achieve a more stable and sustainable rate of growth. Today, CrediAmigo distributes its products through 164 of Banco do Nordeste's 174 branches. By May 2003, CrediAmigo was among the largest microfinance institutions in Latin America, with 123,000 clients and an active portfolio of R$72 million. 2o However the retention rate has dropped to47 percent, partly due to the economic slowdowns and perhaps also due to less understood demand factors. Part 2. Exoandina Microfinance 26. Delinquency rates at CrediAmigo were initially uneven. Delinquency rose sharply in the first year following the surge of expansion, but managementreacted promptly by writing off bad loans and in early 1999, by modifying its performance-basedincentive scheme for staff. It also installed a detailed portfolio monitoring system for delinquency at the loan officer level. Since its inception, CrediAmigo has been aware of the needs to monitor and control costs and to operate on the basis of profit centers. Today, each branch is a profit center. For example, initially CrediAmigo offered enterprise development services. But when it faced challenges in terms of providing these on a sustainable basis, and found also that demand was low, it discontinued such services. CrediAmigo was also aware of its high assets utilization and administrative cost ratios compared to its peers, and launched programs to increase loan officer productivity. Today its portfolio quality and staff productivity compare favorably with international good practice. Only 4 percent of its loans were overdue, using a strict 30-day portfolio-at-risk measure, in accordance with Central Bank requirements. Its annualized loan loss rate i s 2.7 percent, after fully provisioning all loans with payment 360 days or more overdue. As to productivity, loan officers with nine months or more of experience are each handling an average of 313 clients. Salary expenses as percentage of loan portfolio decreasedfrom 139 percent inDecember 1998 to 27 percent inMay 2001. 27. In terms of profitability, CrediAmigo has progressed positively. About 145 percent of CrediAmigo's 164 branches are operationally sustainable. Since June 2000, CrediAmigo has presented positive returns on average assets but in 2002, profits dropped. Learning to maintain good portfolio quality is one of the key challenges which CrediAmigo faces in controlling operating expenses, which remain high at 37 percent of total assets, as of December 2002. Implicationsfor Brazil's Microfinance Sector 28. The story of CrediAmigo clearly dominates Brazil's microfinance sector. It has clearly demonstrated the existence of a market niche for microfinance and has also demonstrated the means for its achievement. It incorporates many examples of recognized good practice in microfinance lending techniques. It also shows that rapid growth must be tempered with an eye on quality and that learning the microfinance culture takes time. 29. To what extent does its story provide commercial banks with a model for `downscaling'? CrediAmigo demonstrates that, under appropriate conditions, and following specific practices, a "downscaling" strategy for commercial banks (i.e., targeting lower income individuals or smaller businesses) could be viable. CrediAmigo also suggests that an existing branch network can greatly help the rollout of microfinance products. Alternatively, partnerships allowing microfinance specialists to distribute their products via bank networks could also be successful strategies to develop large scale microfinance services. 30. With the introduction of new measures for microfinance nationwide in 2003, CrediAmigo has also decided to offer microfinance loans under the new terms, accepting the interest rate caps for small size loans, which other microfinance agencies will now offer. Although CrediAmigo maintains that it will nevertheless be able to sustain its performance, clearly this will be an additional strain and the extent to which it is possible inpractice will need to be monitored. 2.3 Evolution of the Government's Role- BNDES and SEBRAE 31. The Communidade Solidaria and the government also recognized through their working groups the need for hands-on support to microfinance, both through the provision of wholesale finance, and through programs of institution building. From the late 1990s,a range of new forms of support for microfinance Brazii: Access to Financial Services has been provided in particular by the BNDES bank, and also by the agency SEBRAE.'l The most significant of these was the launching of a new program by the BNDES in 1997, for the provision of credit to microfinance entities; the Programa de Crkdito Produtivo Popular - PCPP. This program remained largely unchanged until new microfinance measures were introduced from mid-2003. The analysis below refers to the program from 1997 untilmid-2003. 32. Under this program, BNDES provided revolving wholesale lines of credit to microfinance institutions. Institutions supported by BNDES represented a high proportion of Brazil's MFIs, including most of the larger and more established institutions." From the time of initiation of the program in 1997, until mid 2001, BNDES had contracted a total value of around R$44 million with 31 microcredit institutions. Of this, around R$25 million represented first time contracts, and R$18 million represented repeat contracts with around eleven MFIs. Total credits extended by BNDES' client MFIs by end 2000 amounted to R$84 million, based on an active portfolio of around R$31.7 million. The active portfolio of these institutions grew to R$53 million by end 2002 (Table 2.2), representing an estimated 40 percent of total microfinance loans (Table 2.1). Thus,the role of the BNDES bank in Brazil's microfinance industry, albeit indirect, has been large, and its microfinance support program, the PCPP, contributed significantly to the expansion of microfinance. The role of the BNDES is particularly significant in view of the contribution to microfinance of another public sector bank, the Banco do Nordeste do Brasil, though the CrediAmigo program, discussed further below. Table 2.2 BNDES Program of Support to Microentrerprises(1997-2002). Notes:`This may include some contracts other than the 31 MFls referredto in the other columns. Source: BNDESArea Social (See Appendix Tables A2,la and A2.1b) 33. The share of BNDES in the total resources of the microfinance institutions it works with has been constrained by its own regulations; initial leverage ratios have been 1:l for NGOs and OSCPs, which could be expanded to 2:l over time. Leverage ratios of 3:l (BNDES capital to own capital) were permitted for SCMs. According to BNDES, it provided for up to 50-80 percent of the funding of these institutions. Loan value averaged R$1,155 inJune 2001, which had changed little since 1998 when loans averagedR$ 1,091 (Appendix Table A2.la). The average number of clients per institution rose 50 percent from 1998 to 1999, in the early years of the program, from 969 to 1,489, but declined again thereafter to around 1,100 by mid-2001. Average portfolio size showed a similar initial increase, from R$0.8m to R$1.3m (1998 to 1999) before diminishing to R$l.lm in 2000/2001. This suggests that there was some initial overexpansion, and the early surge then diminished. In terms of loan composition, the majority of loans went to the informal sector (76 percent); mainly for working capital needs (91 percent); with some 21Banco Nacional do Desenvolvimiento Economico e Social, and the ServiGoBrasileiro de Apoio cis Micro e Pequenas Empresas. SEBRAE, established in 1990, focuses largely on technical assistance for micro and small enterprisesin Brazil, as well as the maintenanceof information, advice on credit and support to businessincubators 22Portosoi, Vivacred and Blusol, all of which havepioneer status, and large clusters of MFIs belonging to groups, including nine in the CEAPE network, three in the network of `Banco da Mulher' and five belonging to `Vis20 Mundial'. Part 2. Expandincr Microfinance preference for women (53 percent of all borrowers). These characteristics conformed to those observed in most international microfinance industries, although the proportion of women i s often much higher elsewhere. 34. The program was very popular. Seven out of ten applicants were rejected at the first visit and BNDES worked with one or two o f the remaining for a year or two before a loan approval. It maintained that findingquality MFIs was its biggest constraint. BNDES had an institutional development programfor its clients, under which it prepared manuals for managers, educational material on financial management, regulation, information system, accounting, markets, et^.^^ Average loan value: R$1,155 Average term: 5 years LoanDistributionby Sector and Client Informal sector: 76% Fixedcapital: 7.1% Commerce: 65.9% Male: 46.5% Formal sector: 24% Working capital: 90.8% Service: 18.7% Female:53.5% Industry: 15.4% BNDESLoanTerms to MFI: Maximumamount per loan: R$3 million. Interestratechargedto MFI: BNDES financial costs, TJLP - "Taxa de Juros de Longo Prazo" Term: 8 years, grace period of 9 months; quarterly amortization. Leverage permitted: (Ratio of BNDES funds to the capital provided by the MFI) NGOs (ONG) and OSCIPs: 1:1in a first operationand 2:1ina secondoperation SCMs: 2:1inafirst operationand 3: 1ina second operation Other Conditions Total leverage shouldremain below 5:l; inc Ifnon-performing loansarebetween6% and quarter, then an amount equivalent to their value must b Ifnon-performing loansexceed 15%of total loansina 15% of funds from BND ould beput aside for loan p Thepublic sector is allowed to representonly up to 113of Onlendingterms to final client Interest rates: Variable by client MFI; maximum spread allowed Average loan term: 4.4 months Source: World Bank, basedon data providedby BNDESArea Social. 35. Although the client MFIs onlending rates, limited to 6 percent per month under these programs, were usually close to market rates, interest rates charged by BNDES to its clients, as with other programs o f BNDES lending, were linked to the official long term interest rate (TJLP), which has generally been well below the Selic (interbank) rate, and considerably below commercial lending rates. BNDES was able to sustain this due to its own funding being substantially at below market rates, based on the FAT 23 Supported by a technical assistance loan from IADB. It launchedthe development of rating systems for MFIs (in collaboration with Planet Finance) and auditing systems (based on CGAP audit principles) (www.portaldomicrocredito.org.br). Tools were provided on the website for performance evaluation including a set of performanceindicators and methodology for self-assessment. Brazil: Access to Financial Services employment tax.24In the case of MFIs, BNDES did not charge a spread over the TJLP, as these loans were deemedto be socially ~trategic.'~ 36. In terms of regional distribution, although initially concentrated in the northeast, which had 68 percent of client borrowers and 56 percent of total loans in 1998, these shares diminished to 51 and 43 percent respectively by 2001 (Appendix Table A2.4). Over time the more affluent South and Southeast regions taken together have increased their participation, from roughly 26 percent of clients in 1998 to 42 percent in 2001 and from 38 percent in 1998 to 51 percent in 2001 in terms of value. The North, Northeast and Center West regions decreased their number of clients' participation from around 74 percent to 58 percent and from 62 percent to 49 percent in terms of value in the same period. In September 2001, the two regional groups each had around half total value of loans, although the North and Northeast had a much higher share of clients (58 percent versus 42 percent) implying a much smaller average loan size Implications of the PCPPfor Microfinance Development 37. Clearly the PCPPprogrampushed forward significantly the frontiers of microfinance inBrazil, and in scale was surpassed only by the efforts of the Banco do Nordeste and the CrediAmigo program. Many sound practices were incorporated- such as onlending at market rates, working with client M F I s to impart good business practices, portfolio monitoring and evaluation systems, etc. Yet other features associated with sound principles of microfinance were not incorporated. Notable among these was the `solidarity group' principle of group lending and coinsurance, which is not a requirement of BNDES for its client MFIs.Although some, suchas the CEAPEMFIs, follow this practice, it is relatively rare inBrazil. 38. Moreover, an analysis of the BNDES portfolio and its evolution over this period suggests that its client enterprises did not grow particularly rapidly, despite the rapid growth in the BNDES portfolio. Rapid growth episodes, in terms of increases in clients and portfolios, may have been correlated with higher default rates when they occurred. There was no trend decline in bad loans; if anything, there may even have been some increase. 39. Other good practices as discussed in the context of CrediAmigo and in Section 3.4 below, could also have been followed more consistently.26The below market funding of the client M F I s of BNDES, even if justifiable in a startup phase, lacked an articulated `exit' strategy. Ideally, BNDES should have been able to gradually disengage itself from its client M F I s and possibly transfer its role in credit provision for such MFIsto entities suchas commercial banks. 24FAT, Fundo de Amparo ao Trabalhador (Worker`s Assistance Fund) 25Note that all rates would have changed with the new measures introduced in 2003. 26These `good principles' of microfinance have now been accepted by a number of authors; see Section 3.4 for a fuller discussion. The erratic adhesion o f BNDES supported entities to these principles has also been pointed out elsewhere, notably DAI (2002), `Understanding Microfinance in the Brazilian Context'. Part 2. Expanding ~icro~inance dependence on unrealistic Other BNDES Programs - `Downscaling through Affiliated Financial Znstitutions ' 40. BNDES also extends credit to small and mediumenterprises directly, though its subsidiary FINAME, and indirectly, through special lines of credit to private financial institutions. Disbursements to micro and Brazil: Access to Financial Services small scale enterprises amounted to over R$3 billion in 2000.27Adding medium enterprises raises these totals to R$4.4 and R$3.9 billion respectively. These figures dwatjf the R$43 million of BNDES' contracted lines of credit to microfnance institutions under the PCPP program and demonstrate that the overall importance of the BNDES bank in programs of financial access for micro, small and medium enterprises, is huge. BNDES has a series of different products for funding MPME' s permitted activities (Appendix Tables A2.6 to A2.7). Prominent among these are the FINAME lines of credit for loans to final borrowers up to R$7m. Although not restricted to small or medium enterprises alone, the criteriafor awarding these lines of credit provide incentives for loans to SMEseZ8Funds are made available at the TJLP rate, plus a markup, which is still somewhat below market sources of funding. The markup has two elements; one based on assessedrisk of the agent, which varies by institution, and the other i s the spread of the agent, which varies by activity. Spreads for loans to SMEs are 1percent only, while spreads for other activities have been 2.5 percent. Institutions which are able to demonstrate higher proportions of loans to the SME sector obtain more funds. Banks' eligibility limits are re-evaluated every six months. The top twenty five financial agents used by BNDES are indicated in Appendix Table A2.6. 41. BNDES also participates, together with otherfederal financial institutions, in other special programs targeted at support to microenterprise for purposes of employment generation - the so-called "Programa B r a d Empeendedor - PBE', specially designed to stimulate the development of micro, small and medium size companies. BNDES also participates in a guarantee fund, the "Fundo de Garantia para a Promoglio da Competitividade", created with resources from the National Treasury, managedby BNDES, which gives collateral to guarantee credits from financial institutions to micro and S M E exporters." Start- up micro-entrepreneurs in Brazil can also apply for funding from a dedicated fund of R$200 million created to finance 5,000 new micro-c~mpanies.~~BNDES received 60,000 applications and approved 4,000. 42. SEBRAE ("Sewigo Brasileiro de Apoio as Micro e Pequenas Empresas"), a special agency for the support of small and medium enterprises, runs programs of technical and financial support to microenterprises, through a Loan GuaranteeFundfor long term investments, FAMPE("Fundo de Aval cis Microempresas e Empresas de Pequeno Porte"), only however available for loans issued by a handful of public banks.3' From2000, SEBRAE began a Venture Capital Program(Programa de Capital de Risco - PCR), for participation in funds investing in emerging micro and small companies. And in October 2001, SEBRAE launched a new scheme to provide both technical assistance and credit to microfinance institutions too small to qualify for BNDES assistance, across 300 municipalities with low HDI(Human Development Indicator) indices. Loans will be available for up to R$300,000 for 8 years with a 3 year grace period. Interest rates on SEBRAE lines of credit to MFIs are based on the TJLP with no spread. 43. Thus the government, and especially BNDES, has partnered a wide segment of the nation's financial institutions to improve access, through financial incentives and preferential credit rates. But as shown in Appendix Table A2.6, despite its efforts, it is not clear that BNDES i s meeting its objectives. Overall ratios of small and medium enterprises in the BNDES lines of credit through its MPME programs have 27TheFederalgovernment, the BNDES and SEBRAE, each offer different definitionsof micro, small and medium enterprises.BNDES defines microenterprisesas businesses with gross annualrevenues of up to R$900,000, while small businesses are classifiedby revenues betweenR$900,000 andR$7,875,000 and mediumbusinesses are classifiedby revenuesbetweenR$7,875,000 and R$45,000,000. This includesbusinesses from industry,commerce and services. '*Regionaldevelopment policy and export promotionare also goals which are explicitly supported and given additional weight, together with energy conservation, environmentalsafeguards, etc. 29Created by Law 9531(10/12/97) and Decree 3113 (6/07/99). 30"Cre`ditoOrientadopara Novos Empreendedores" -CreditDirectedto New Entrepreneurs. 31Working capital loans are acceptedifthey complement a long term financing loanand representless than 50 percentof the total. Part 2. Expanding Microfinance been only around half, suggesting that more appropriate targeting mechanisms could be explored (see Section 3.2 below). In its PCPP program, BNDES has shown a tendency to go towards some of the best performing institutions in microfinance, and also, to the more well to do regions, over time. 2.4 Microfinance in Brazil Present - Size and Growth 44. Given that the incentives to expand microfinance began in the late 1990s, it i s not unexpected to find first that although a handful of the older microfinance institutions date from the 1980, most (over 80 percent) have been in existence only from 1998/99 onwards. Apart from the state-owned bank initiative, Crediamigo, microfinance institutions in Brazil are mostly small, private, and independent of formal financial institutions. Private banks in Brazil have not been active players in mi~rofinance.~~ Numbers vary; but estimates suggest that by mid-2002, there were around 120 microfinance players, of which only a dozen catered to more than 2000 clients.34 45. Among the 40 to 50 larger private institutions, the CEAPE network of Non Governmental Organizations (NGOs) constitutes the second largest microfinance provider in Brazil, with over 26,000 clients in 220 municipalities across the country, served via a network of 13 CEAPE entities. Banco da Mulher, a group of NGOs, some affiliated to Women's World Banking, had more than 2,300 clients and a loan portfolio close to R$2.4 million (US$l.O million) in June 2001. Several small institutions are also active in the Rio area. They include VivaCred and RioCred, with a loan portfolio of R$2.0 and 1.2 million and about 2000 and 1000 clients respectively in June 2001.35In terms of nature of services provided, commerce and services in the informal sector dominate (Appendix Table A2.5). And in terms of regional distribution, most microfinance is urban in character (especially in the Northeast), and thus well equipped to target the urban poor, who account for the bulk of the poor in Brazil.36 32 There is now a sizable literature on microfinance inBrazil, which beganfrom Schonberger(2001) "Microfinance Prospectsin Brazil", and has been added to notably by the DAY BNDES group publications; notably Goldmark (2000) `A SituapTo das Microfinancas no Brad';Nichter, Goldmark and Fiori (2002) `Understanding Microfinance inthe BrazilianContext'; Brusky and Fortuna (2002), `Understandingthe demandfor MicrofinanceinBrazil: A Qualitiative study of Two Cities; Martins, Winograd and Salles (2002), `RegulamentapTodas Microfinangas' and Mezzera(2002), `Microcredit in Brazil: The Gap Between Supply and Demand', MicroBanking Bulletin, November 2002, Washington, D.C. 33 The foreign bank ABN AMRO is developing a pilot project, in the context of aUSAIDcredit line, to start providing loans to micro-entrepreneurs. The project, locatedin the Northeast of Brazil and in Sao Paulo, would use distinct facilities and specially trained credit officers, but ABN AMRO's branch network would be used to channel funds. However, this activity is not intendedto becomea real businessline of the bank. 34 Estimatesbasedon DAI (July 2002) (Nichter, Goldmark and Fiori); basedon an aggregationof data on banks, SCMs, credit OSCIPs, NGOs and government initiatives. 35 RioCred, ajoint ventureof Fininvest and the International FinanceCorporation, has since undergonea transformation of its regulatory status as well as name, and is now operating as MicroInvest. 36 See World Bank (2001), `Attacking Brazil's Poverty : A Poverty Report With a Focuson UrbanPoverty Reduction Policies'. Brazil:Access to FinancialServices Bancodo Povo de Juiz de Fora (FAEP3') RioCred Vitoria Credisol Rotula SindCredI Credito Cidadao Note:a End2002 figures suggest considerablyhigher figures Averages between December 2000 and 2001 data; number of active clientsproxiedby number of active contracts Includesthree affiliatesonly (Parana, Santa Catarina and Bahia) Estimate Source: BNDES; institubons'brochuresand websites; data are for June2001 46. There are also at least 20 municipal or state-supported `microfinance' NGOs inBrazil, many of which are expanded versions of public work and employment subsidies. Banco do Povo de Sao Paulo in Sao Paulo is now the largest one of these, with a loan portfolio of R$13.8 million (and an estimated 16,000 clients. Although these too are dubbed microfinance institutions, as they make small loans, they do not attempt to follow the sound financial principles adhered to in microfinance, and their onlending practices are heavily subsidized, on the grounds of adhesion to the Usury law. 47. As discussed earlier (Figure 2.1), microfinance in Brazil only began to grow significantly from the end of the 1990s. There was a rapid burst of growth for the next three to four years, with total clients tripling between 1998 and 2001 (from about 46,000 clients in December 1998 to more than 140,000 in June 2001, an average annual growth rate of 49 percent). Total loan portfolio nearly quadrupled (from R$25 million in December 1998 to R$95 million in June 2001, an average annual growth of 61 percent). But a part of this success is the result of low initial levels. Growth has been unequal across institutions. Crediamigo has managed to maintain rapid growth since its inception but some other large microfinance institutions such as the CEAPE network have been largely stagnant. PortoSol, the fourth largest microfinanceprovider, has also grown slowly.38 37 Banco do Povo de Juiz de Fora's official name is Fundo de Apoio ao Empreendimento Popular FAEP (Support Fundfor Popular Employment) 38 Some sources detect a recent slowdown, as witnessed by the slow loan renewal rate even inCrediAmigo (see Nichter, et. al). However, against this, the number o f SCMs has clearly increased rapidly. Part 2. ExpandingMicrofinance Supply and Demand Compared; PenetrationRatios 48. Available estimates suggest that demandfor microfinance in Brazil far outstrips supply. The total loan portfolio for microfinance was some R$135 million by end 2002 (Table 2.1), for a total clientele of some 140 thousand persons (Table 2.3). ILO estimates contrast this supply with an estimated 6.4 million potential microfinance clients in Brazil, and a potential loan demand of R$11 billion.39 This implies an average loan size of around R$1,700 per borrower. Even if this is adjusted for our present estimated average loan balance of only R$1,100 per borrower, potential loan demand i s around R$7 billion. Thus according to these estimates only 2.2 percent of potential microfinance clients, and 1.4 percent of their potential credit needs, are served by microfinance institutions today. Other estimates are similar; 8.2 million potential client enterprises and a penetration rate of 2 percent.40If this i s compared with other countries in the region, using comparable techniques of estimation, Brazil i s revealed to have a penetration ratio much lower than that of its neighboringcountries (Table 2.4). 49. However these estimates follow a standardized assumption of a 50 percent eligibility rate for all microenterprises. It can be argued that effective eligibility rates may very likely be lower, due to such factors as the availability of special government support especially inthe rural sector, or due to the use of consumer or installment credit in lieu of microenterprise demand, and also that all microenterprises may not wish to borrow. However, the 50 percent ratio was maintained for purposes of comparability with other countries' data. As shown in Table 2.4, microfinance penetration in Brazil i s much lower than in other Latin countries, even if the two percent estimate is low in absolute terms. Table 2.4 MicrofinancePenetration Braziland Other LatinAmerican Countries (2001) - Bolivia 232.3 379.1 163 Nicaragua 116.4 84.3 72 ElSalvador 136.3 93.8 69 83.0 30.2 Peru 618.3 185.4 30 Chile 307.8 82.8 27 Brazil 7,875.6* 158.7 2 Note: Brazilmicrofinance penetration calculationsuse 2001 potentialdemand data, since current market data are also from 2001. Source: BNDES, 2002, `Understanding Microfinance in the BrazilianContext', Rio de Janeiro, Brazil, Microfinance Growth ZnternationalComparisons - 50. The apparently highdemand estimates for microfinance seem paradoxical compared to microfinance growth rates in Brazil, which outside of CrediAmigo, have not been high. As the analysis of the BNDES clients showed, growth rates of the overall microfinance clientele have been high, but this i s more due to new MFIs than to the growth inthe portfolio of existing MFIs. 51. True, internationalexperience suggests that developing an extensive microfinance market i s generally a long process. After three years of operations, the Grameen Bank of Bangladesh had 377 client solidarity groups, reaching a total of 2,200 poor people. These numbers are close to those reached by recently created Brazilian institutions. Growth at the Grameen Bank started to pick up after about ten years of operations. The experience of India and Bolivia also confirm that, independent of the structure of the microfinance market, building significant outreach is a lengthy process. India's main microfinance 39See also Mezzera (2002, op. cit.). 40Nichter, Goldmark and Fiori (2002), `Understanding Microfinance inthe Brazilian Context'. Pnge 71 Brazil: Access to Financial Services model, the Self-Help Group program, was initiated at the beginning of the 1990s. Yet it was only after eight years that it began to witness acceleratedgrowth. 52. The Self Help Group (SHG), model was launched by the National Bank for Agriculture and Rural Development (NABARD) in 1991 as a pilot project to link small solidarity groups of poor women, the Self-Help Groups, with financial institutions for funding. The success of the pilot led NABARD to formalize the Self-help Group-Bank Linkage Program in 1996 as a mainstream activity for commercial banks under their priority sector lending obligations. Under strong impetus from NABARD, the model seems to have taken off, with almost 265,000 self-help groups linked to 318 banks by March 2001 through the participation of more than 750 NGOs. Total loans outstanding amounted to Rs.1,930 million (US$39 million) in March 2000. The Government of India is now focusing on making the model sustainable without subsidies, and on eliminating barriers to the development of private microfinance institutions. Bolivia, where private microfinance institutions dominate, has also seen its industry grow slowly. Growth started to pick up in 1993 after Prodem, a microfinance NGO transformed into a licensed bank (PortoSol), which prompted similar changes by other NGOs and the creation of a fund to promote private regulated microfinance institutions. 53. These comparisons also suggest that the models of large-scale expansion of MFIportfolios, intandem with a sponsoring institution such as a bank, may lead to more rapid growth, than through small individual MFIs.This is discussed further in Section 3.2 below. Figure2.2 Grameen Bank, Bangladesh;Self-Help Groups: India; and Top 10 MFls: Bolivia Growth - Growth in annual disbursements and outreach # of groups /members (millions) Disbursements ($USm) # of SHGs Yearlydisb linked('000) (Rs Cr) 2.5 300 350 First soft loan 450 400 250 300 350 250 300 200 1.5 250 200 150 1 0 200 150 150 100 100 0.5 100 50 50 50 0 0 0 0 0 - - - -Number of members -Numberof groups -Total numberof SHGs linked ~ *- SHG linkedduringthe year .Yearlyloan disbursements(US$") - - - -Yearlydisbursements (Rs Cr) Source: Grameen Bank web site; NABARD, India, and lnidividualInstitutions' data, Bolivia. Part 2. Expanding Microfinance 2.5 Institutional Capacity of Brazil'sMicrofinance-Performance and Impact Operationaland Financial Performance 54. A large proportion of Brazilian microfinance institutions claim that financial sustainability is a core objective. This section looks at a series of indicators of operational and financial performance, which contribute to the overall assessment of sustainability and performance. First, indicators of the quality of the loan portfolio suggest that Brazil is an average performer relative to comparators, and by some estimates,possibly below average. Analysis of the client institutions of BNDESshowed that average rates of loans in arrears has fluctuated considerably over the period from 1998 to 2001, from around 3.5 percent and 8.4 percent. In June 2000, loans in arrears averaged 6.12 percent for all 31 clients.41 Looking at the largest institutions only, the average appears lower; at 4.7 percent, with a range from 1.0 to 10.5 percent (Table 2.5). Variations between institutions are large. Among the BNDES group, eliminating the two worst performers reduced loans in arrears from 6.1 percent to 4 percent. Variations over time are also sharp, even within the same institution; at some points bad loans have risen to a quarter or third of total portfolio; only to apparently recover six months to a year later. This may reflect the short maturity of loans and periodic write-off^.^' 55. A benchmarking table, prepared from a sample of 30 leading microfinance institutions in Latin America shows that these institutions had an average portfolio at risk (overdue by over 30 days) of 6.5 percent in June 2001 (Table 2.6).43 This suggests that Brazil i s around the reference rate used for similar institutions. Moreover, default rates are dependenton macroeconomic conditions. For instance, BancoSol (Bolivia), which i s often cited among best practice examples in microfinance, had a percentage of overdue loans (by one day and more) of 4.5 percent in 1998, but it increased to 16.9 percent inJune 2001 as macroeconomic conditions deteriorated. Table 2.5 Brazil: Selected Performance Indicators of Main Microfinance Institutions 3.8 * n.a. ma. 8.6 ** n.a. n.a. POrtOSOl 20.0 4.6 Bancodo Povo de Sao P n.a. Notes: August 2001; ** June 2001. 41 Development Alternatives Inc., estimated the average percentage o f gross loans overdue by more than 30 days for a representative sample of 12 BrazilianMFIs at 7.5 percent (with a range from 2.3 to 12.0 percent) in 1999, and found that this compared poorly with a reference rate of 4.6 percent (with a range from 0.6 to 13.0 percent) for a sample of 17 microfinance institutions in Latin America in 1998. Differences in results are likely attributable to fluctuations over time and sample selection, both of which have high variation, as discussed above. 42 I t mustbe remembered that guidelines adopted by BNDES require that if bad loans exceed 8 percent in two consecutive quarters, loans must be returned. 43 MicroRate is a private rating agency dedicated to the evaluation o f microfinance institutions in Latin America. I t has been operating since 1997 and has analyzed a total of about 70 institutions. The extent to which loan classification and recognition of bad loans differs materially across countries i s not taken into account. Brazil: Access to Financial Services Source: individual institutions, BNDES. Brazil Table 2.3 sample 4.6 169.7 Colombia I, WWB Popayan,WWB Bucaramanga, El Salvador Calpia, Enlace Mexico Comparlamos Edyficar,CMAC Ica,CMAC Tacna, Proempresa,Crear Tacna, Crear Arequipa, Confianza Uruguay Fucac 6.5 76.3 6.5 118.9 Source: MicroRate, website; ratios are for June 2001. 56. Next, provisioning for past due loans may also be low. Most Brazilian microfinance institutions do not disclose their levels of provisioning. Those which do show a wide versatility of provisioning levels, from 19 percent for Banco da Mulher's three largest affiliates to 420 percent for Banco do Povo de Juis de Fora. DAIestimates an average provisioning level of loans overdue over 30 days of their sample of 12 microfinance institutions in 1999 to be 37.5 percent. This compares with 79 percent for BancoSol in 2000, and 68 percent in June 2001. MicroRate's sample of microfinance institutions had a median provisioning rate of 81 percent in June 2001. Applying recommended rates of provisioning for various degrees of past due loans, DAI estimated that the average provisioning level of Brazilian microfinance institutions should be 64 percent.44 It has been suggested that current low level of provisioning may be influenced by loan classification and provisioning norms of the Central Bank', which require full provisioning for a loan with payments overdue by 180 days or more, although these are not applicable to microfinance institutions other than SCMs. Given the short maturity of most micro-loans, microfinance institutions should be encouraged to provision bad loans earlier. By contrast, capitalization appears generally strong. DAIcalculated the average ratio of capital to gross loans of its sample of institutions at 119 percent. This high average level may be in part due to the young age of most microfinance institutions in the sample, where the loan portfolio is expected to grow substantially from its current level. 57. Some evidence also suggests that the operating efficiency of Brazilian microfinance institutions appears lower than international best practices. DAI's data suggest that the average number of clients per loan officer was only 191compared to an average of 326 for a sample of 30 Latin American microfinance institutions used by MicroRate. Similarly operating costs amounted on average to 49 percent of the loan portfolio for Brazilian institutions compared to 23 percent for MicroRate's sample of Latin American institutions. However, these indicators may be biased upward for Brazilian institutions by the fact that they have not yet reacheda size sufficient to benefit from any significant economies of scale. 58. Yet many microfinance institutions in Brazil seem close to being operationally and financially sew- sufficient. Operational self-suficiency was estimated at 91 percent in the DAI sample of microfinance 4410%on loans overdue by 1to 30 days, 50% on loans overdue by 31to 60 days, 75% on loans overdue by 61to 90 days and 100%on loansoverdue by more than 90 days. Part 2. Expandina Microfinance institutions, adjusted for subsidies but excluding loan loss provisions. Adjustedfinancial self-suficiency was estimated at 92 percent.45 These results compare very favorably with international standards. However, differences across institutions seem to be large, and only four of the 12 institutions analyzed reachedfinancial sustainability. Another study of four microfinance institutions conducted by BNDES in May 2000 reached similar conclusions. 59. These favorable results may be explained in part by high spreads combined with subsidized funding, which compensates for their high operating expenses. Most of Brazil's microfinance providers charge market-based interest rates (Table 2.9). Rates are in line with those charged by microfinance institutions elsewhere in Latin America.46 A comparison conducted by BNDES between four Brazilian microfinance institutions and comparable samples of peers shows that the rates charged by the Brazilian institutions were in line with their peers in 1999, whose average interest rate reached 53 percent p.a. for medium size institutions and 63 percent for smaller institutions (Figure 2.3). Figure 2.3 Comparisonof InterestRates Chargedon Loanswith other Institutionsin LatinAmerica Interest Rate Policy, BrazilianMicrofinance Institutions vs. Peer Groups CEAE Portosol Peer Group 1 BancodoPovo Vwacred Peer Group 2 Pernarrbuco de Juiz de Fora -19 Source: Estimatesbasedon Calmeadow (2000). 60. However, some institutions in Brazil charge heavily subsidized interest rates, violating a generally accepted basic principle of sustainable microfinance.This is for instance the case for municipal programs such as Banco do Povo de Siio Paulo or de Juiz de Fora. These programs are essentially social services, financing the creation of work for employment or other welfare purposes, which justify their low interest rates on social grounds. Other microfinance providers cite this unfair competition as a key obstacle to their growth. There is an ongoing debate in Brazil between advocates that microfinance should be sustainable and therefore market-driven, and advocates that microfinance i s a social policy instrument to distribute interest rate subsidies to the poorest. New measures introduced in mid 2003 clearly attempt to support the expansion of credit to the poorest through measures which include lending quotas and interest rate ceilings. It remains to be seen whether these will be effective in increasing sustainable long term microfinance. as desired. 45Operational self-sufficiency is defined as the ratio of Total Loan Income / Total Operational Expenses. Financial self-sufficiency is defined as the ratio of (Total Loan Income Operational Expenses + Total + Total Income from Financial Investments) / (Total Financial Expenses). 46These comparisons are only approximate however given differences between base rates and other market conditions across countries. lJcl82 Part 2. Expanding Microfinance advantageof coinsurance and social collat the country. In 1983 the project was tran Bank decided not Source Bank Indonesiaand GTZ (2000), www.grameen-info.org, www,baac.org.th Mibanco 15.1 68.6 27.2 CMACs Cajas Municipales de Ahorro y Credito 33.4 52.2 9.2 CRACs Cajas Rural edrto 27.6 78.2 32.4 EDPYMEs 11 8 34.8 31.5 Source: Banking Superintendency, Peru. 3.3 Public Banks and CompetitionIssues 80. The presence of institutions (mostly those established by municipalities), which charge highly subsidized interest rates, creates a distortion in the market likely to be a barrier to entry for new private players. Banco do Povo of Sao Paulo is an example of such an institution, which although apparently successful, is extremely costly to the government and where financial self-reliance i s clearly a subsidiary objective. In the long run, such a programcould perversely choke off total credit to the system, as private providers are not able to compete in its market. The problem of such crowding out, even in microfinance, i s accentuatedby the existence of numerous `low cost' programs and credit lines, implemented mostly by Brazil's development, federal- or state-owned banks. The danger of such crowding out has increased with the recently introducedmeasures of 2003, which include large scale subsidies to small borrowers. Brazil: Access to Financial Services Banco do Povo, an Exampleof an Institutionwith Subsidized InterestRates Banco do Povo is a microfinanceinvestment fund createdin 1998by the State of Sao Paulo. Banco do Povo receivesan annual donationfrom the Governmentof Sao Paulo's budget. The funds are administeredby the State bank NossaCaixa, which makes disbursementsand collects paymentson loans made by Bancodo Povo. Banco do Povo's trainedcredit agents distributeloans mostlyaimed at financingfixed ca investmentsby small businesses and morerecently, start-up companies. The bank's primeobjectiveis to stimulateemployment inthe State, and it is thereforelocated in the officesof the City Hall and the Labor Secretary. Banco do Povohas beenquite successfulat increasingits loanportfolio while maintainingapparently good quality of loans. Loans disbursedhave increasedfromR$.3 millioninthe first year of operation (Sept. 98 to Sept. 99), to R$.13 million last fiscal year (Sept. 2000 to Sept. 2001). Non-performingloans (overdueby 30 days or more) reach2 to 3 percent only. However, Bancodo Povo's loans carry a subsidized interest rateof one percentper monthinaccordancewith the limits set by the UsuryLaw for non-regulatedentities. Source: Banco do Povo 3.4 Microfinance Methodology Principles of Good Practice - 81. As alluded to earlier in the present chapter, economic theory, as well as supporting evidence from a number of countries, points to a number of principles which can mitigate the risks and reduce the costs associated with microfinance lending. Many of these principles have been adopted by CrediAmigo, but could be extended especially to newer startup microfinance initiatives. These principles point particularly to the value of group contracts and dynamic incentives for loan repayment. The Grameen group lending model, where the default of one member implies that all group members are subsequently denied loans, has been successfully replicated in a number of countries. Credit cooperatives and rotating savings schemes also make use of the principle of `social insurance' which makes these approaches successful. Such models reduce the risks of `asymmetric information', or limited knowledge of the client by the lender. Selection by peers of group members also mitigates `adverse selection', where potential borrowers may include an unusually highproportion of high-risk persons, who know that their default is too small to be followed up by the 1endera5' 82. Not all successful MFIs strictly follow the group approach alone, however; BancoSol in Bolivia also required some form of guarantee on around a quarter of its portfolio, and BRI Indonesia does not use a group lending mechanism and does require some form of collateral (see below). However, as a consequence the poorest borrowers are excluded in the BRI model, which targets the group of the `better o f f poor. Interestingly, some evidence suggests that overall repayment rates are higher in models with greater focus on the poor. Such models tend to rely more on solidarity mechanism^.^^ 83. Other good practices identified include `dynamic incentives', where loan amounts have progressive or `step lending' features, increasing over time; regular repayment schedules which begin immediately after the loan (and help capture a cash-flow before it is dissipated), and lending to women or particularly disadvantaged groups with few other options. Finally, although few programs require formal collateral, a j7 See in particularMorduch(1999) `The MicrofinancePromise'; Besley (1994) `How do MarketFailures Justify InterventionsinRuralCredit Markets'; Besley and Coate (1991); "Group lendingrepayment Incentives and Social '*MicrobankingBulletin collateral" and Udry (1990) `Credit Markets inNorthernNigeria'. November 2002, notes in its highlights that MFIS that target the low endclientscover almost all of their costs and have an average financialself-sufficiencyratio of 92 percent, comparedto 42 percent for all its 147 participatinginstitutions.There is other evidence to support this. Wenner (1995) finds ina survey of village banksinCosta Rica supported by FINCA that delinquencyrates are higher inbetter off towns, Sharma and Zeller (1996) also found this to the case invillageprograms inBangladesh.However, other studies are less conclusive and more empiricalinvestigationof this issue is needed. Pad 2. Expanding Microfinance collateral substitute is not uncommon, for example, the Grameen bank requires borrowers to contribute to an emergency fund, in the amount of 0.5 percent of every unit borrowed. An additional 5 percent of the loan is taken out as a `group tax' which goes into a group fund account, thus constituting a form of `forced savings'. 3.5 Other Constraintson Expansion 84. Other factors at play include insufficient information to small entrepreneurs on microfinance a~ailability.~~ This emphasizes the importance of financial education and training efforts for government entities such as BNDES and SEBRAE. The creation of the "Portal do MicrocrCdito" (www,portaldomicrocredito.org.br) sponsored by Comunidade Solidaria is a step in the right direction, although the low level of internet connectivity among small entrepreneurs will limit its short-term use. 85. Parallel to this is the strengthening of information disclosure at the M F I s themselves. Only some of the largest programs disclose financial information on their performance publicly. This contributes to the perception that microfinance providers do not devote sufficient attention to profitability and sustainability. Adopting standard charts of accounts and the publication of results are potential solutions. The microfinance internet portal could also be used for disclosure purposes. An alternative i s to encourage the rating of microfinance providers and publish these ratings. Increased communication on the benefits of microfinance compared to other more traditional finance sources would also help increase the demand for microfinance. 86. Second, venture capital funds inBrazil have had a limited interest inmicrofinance. Although there are several venture capital funds active in Brazil, most of them invest in middle market firms. HSBC, with the InterAmericanInvestment Corporation (IIC), has recently established a US$105 million private equity fund to invest in small and medium size enterprises in Latin America with a special focus on Brazil. It remains very limited however compared to the potential demand. The International Finance Corporation (IFC) of the World Bank Group is also a small actor, and the other player i s the government owned SEBRAE. as described above. 87. Third, the scarcity of trained loan officers willing to work for small microfinance institutions i s also mentioned as a constraint to growth. The difficulties of prevailing labor laws also make it financially difficult to hire good loan officers on a permanent basis, as returns to these activities would not support the maintenance of costly staff with high overheads. Providing additional training to candidate or existing credit officers as proposed by the Comunidade Solidaria i s therefore certainly needed, at least as a prelude to the investigation of means to flexibilize labor laws. 4. MICROFINANCE REGULATIONAND SUPERVISION: FUTUREPOLICY OPTIONS 88. The present section does not evaluate in any detail the recent changes introduced in microfinance regulation in 2003. These measuresand their potential impact alre alluded to elsewhere in this chapter. 4.1 Should Microfinancebe Regulatedor Supervised? 89. A discussion of this issue i s facilitated if a basic conceptual distinction i s pointed out between prudential (or protective) and non-prudential (enabling or preventive) regulation. Prudential regulation is 59Interviews of micro-entrepreneurs in Sao Paulo and Recife confirmed that most of them had never heard of MFIs (Brusky,2002). Brazil: Access io Financial Services concerned with the safety of the financial system, and of deposits in institutions which intermediate funds. Such regulation which comprises for example, capital adequacy, reserve or liquidity norms, i s usually complex and would typically require a financial regulator's supervision for implementation. Non- prudential regulation on the other hand, (e.g., disclosure of ownership and control, fees and rates, consumer protection or financial performance), could be largely self-executing. Until 1999, Brazil's microfinance institutions, as NGOs or non-profit entities were not subject to financial regulations and were entitled to special tax treatment. But they were constrained by the Usury law and limited in their capacity to mobilize funds. Today, the majority of Brazil's microfinance institutions operate under special regulatory windows, as civil society organizations (OSCIPs), or microcredit societies (SCMs). However one major MFI, CrediAmigo, operates as a department of a bank, and i s therefore regulated under this umbrella. Conceivably, other banks, public or private, could also have microfinance operations among their activities. SCMs, as financial institutions, are also subject to regulatory requirements which encompass both enabling regulations (reporting requirements) and prudential regulations (minimum capital, leverage, and liquidity); OSCIPs have reporting (non-prudential) obligations to the Ministry of Justice but are not subject to prudential regulation or supervision as a financial entity. To what extent is the present framework in line with international practice and what modifications may be needed to support the future evolution of this sector? There i s an increasing international consensus on a large range of issues concerning the regulation and supervision of microfinance institutions, and the present section discusses broad guidelines which may be helpfulgoing forward.60 The Structure of Liabilities and the Needfor Prudential Regulation 90. A first consensus area in the microfinance regulation and supervision debate is that it is the structure of the liabilities side of a microfinance institution that determines whether it should or should not be regulated, and the extent to which supervision and oversight is required. The emerging consensus points out that credit-only MFIs need not in general be subject to prudential regulation, because the limited resources of a bank authority would be distracted regulating and supervising an institution where public interest is not at stake. However this does not preclude non-prudential or enabling regulations, which is aimed at preventing abuses in the industry, and also at promotion, through greater transparency. Indeed protection against abusive lending and collection practices, or truth-in-lending, can be particularly relevant for institutions such as MFIs.Ownership and governance criteria (e.g. limitations against public sector participation on boards of SCMs belong to the same category). 91. Brazil's MFIs today are subject to both non-prudential and, in the case of SCMs, prudential regulations, although its SCMs are not deposit taking. Prima facie the present level of prudential regulation of the SCMs appears to be greater than warranted by theirfunding or scope of activities; it is possible however that regulations were imposed with a view to thepossible future expansion of the scope of activities ofSCMs. Waiting until the MFIsector reveals itself to be sufficiently mature to take deposits can be a prudent strategy. If Brazil believes that the MFI sector i s today sufficiently mature for it to contemplate this major step, the structure of the prudential regulation of this sector going forwards requires planning. 6oThe growing literature in this area includes, inparticular,Christenand Rosenberg(2000) `The Rushto Regulate'; Van Greuninget. al, (1999), `A Frameworkfor RegulatingMicrofinance Institutions;Staschen(1999), `Regulation and Supervisionof MicrofinanceInstitutions:Stateof Knowledge'; andCGAP (2002)`ConsensusMicrofinance Policy Guidance: Regulationand Supervision'.See also a discussionof issues inthe Sa-DhanNewsletter of October 2002 (wwwsa-dhan.org); an associationof microfinanceinstitutionsinIndia. Part 2. ExpandingMicrofinance Are Special Regulatory WindowsNeeded? 92. Regarding regulation, the debate focuses on the question of whether there should be a special regulatory window for microfinance activities, or for microfinance institutions, or whether this should come under the code applied to other financial institutions. As discussed above, their risk profile i s unique, and regulations may do well to require distinct treatment for microloan or institutions specialized in microfinance. Such regulatory windows may for example lower entry requirements or establish a higher capital ratio. But against this, the proliferation of different, specialized, regulatory windows can rapidly increase the number of regulated microfinance institutions, because the entry requirements can be so low that many microfinance institutions can apply for it. This situation reduces the effectiveness of the prudential supervision, due to both resource constraints and need for appropriate supervisory experience. Today Brazil has a specialized framework for SCMs; and their numbers have increasedfrom none to 36 in 3 years. This could swiftly become a supervisory burden. There i s also a separate framework for credit OSCIPs; the need for this third window, in addition to the NGO and the SCM, could be reevaluated over time. At present its major purpose may be an instrument to avoid the interest rate restrictions on the NGO. 93. Opinion today i s increasingly that integrating microfinance institutions into formal financial systems i s key to expanding their outreach, and could also facilitate the commercialization, or downscaling of banks into microfinance activities. Banks, with their large networks of offices and information systems, have some advantages with respect to smaller microfinance institutions, but in general, they still lack the lending methodologies to enter aggressively into the new field of microfinance. The higher minimum capital requirement of a bank is often cited as a barrier for microfinance institutions, but in some countries, this capital can be so low that sustainable microfinance institutions can afford to comply with it.G'In Brazil, the new regulatory framework for MFIs have much lower minimum capital requirements than banks, but the largest and arguably most successfil of Brazil's MFIs is sofar constituted as an arm of a bank, with no special regulatory window. 4.2 Specific Prudential Norms: Minimum Capital, Capital Ratiosand Credit Risk 94. If a special regulatory window is to be established for MFIs, it i s argued that the smaller balance sheets of MFIs, and, in many cases, the lower capacity of their shareholders to make additional contributions to the equity capital, imply a lower minimumcapital, both for entry to the market and for ongoing business operations. Establishing such different capital requirementfor microfinance institutions can be very dijjicult and appropriate levels can varyfrom country to country, depending on the size ofthe existing microfinance providers, the competition in thefield of microfinance, and required investments to start the licensed microfinance company, which depends on other requirements made by the Bank authority. 95. The size of existing microfinance providers matters, because if too high, applicants will be few, but too low a level will invite an excessive number of applicants, regardless of their sustainability, also posing supervision problems. The degree of competition matters because the higher the competition, the lower the spread and thus the larger the loan portfolio must be to cover administrative expenses. A larger loan portfolio can be financed either by larger equity or by higher financial leverage; but the latter (as discussed below) i s not desirable. Finally, higher regulatory requirements require better information 61Christen and Rosenberg mention the following cases: BASIC(India), K-REP (Kenya), BancoSol (Bolivia), Kafo Jiginew {Mali), BRAC Bank (Bangladesh), FinAmerica(Colombia), Compartamos (Mexico), FINSOL (Honduras), CARD Bank {Philippines), Financiera Calpia (El Salvador), ACEP (Senegal),Genesis (Guatemala), Banco ADEMI and Banco de la Pequefia Empresa (Dominican Republic), CERUBEB (Uganda), Banco Solidario (Ecuador), and also MIBANCO {Peru). Brazil:Access io FinancialServices systems, which also argue for higher equity capital. These factors vary by considerably by country, and there is no golden rule for establishing a minimum capital requirement, but many factors need to guide the decision. As Table 2.11 indicates, Brazil's minimumcapital for SCMs at present is low. Whether this is justified in view of thefactors discussedabove should be debated. Table 2.11 Capital Requirementsfor RegulatedMicrofinanceInstitutions Bolivia Private Financial Funds (FFP) 820 El Salvador Savings and Loans Societies (SAC) 2 900 Panama Microfinance Banks (BMF) 3 000 Peru MunicipalOffices, Rural Offices and EDPYMEs 233 Ghana Savings and Loans Companies/ Rural Banks 50/ 20 (*) For rural and cooperative banks, the paid-upcapital varies from US$50 000to US$260 000, depending on the class of municipality or metropolitanarea the bank is located in. Inthe case of thrift banks,those operating outside metropolitan Manila are requiredto have US$ 1 millionof paid-upcapital, and those inside US$6.5 millions. Sources: Gallardo (2001), Rosales (2002). 96. Differences in the minimumcapital requirements in countries that have opened regulatory windows for microfinance institutions can also be viewed in terms of the ratio of minimum capital for banks compared to the minimumcapital for the licensed MFI. This ratio is 3.3 in Panama, 3.9 in El Salvador, 9.1 in Bolivia, and 19.3 in Peru, to mention some countries. In Brazil by contrast the diference in minimum requirements (and thus this ratio) is very high; 175 timesfor commercial banks and 125 times for development or investment banks.62 CapitalRatios and Leverage 97. Capital ratios establish limits to financial intermediation, reducing the risk from opportunistic behavior from the shareholders, who want to maximize their profits, but at the same time reducing their risk. Capital ratios also induce asset diversification through risk weighting. Currently, the capital ratio does not discriminate among types of loans in terms of risk weighting. However, micro loans are relatively risky for the reasons discussed before; lack of physical collateral, higher volatility and nature of clients; as well as more costly to administer. Especially in countries with low financial intermediation, and a large informal sector, these pressures can be compounded by excess demand for microloans which raises liquidity risk as well as leverage. A higher capital ratio would be required to limit the exposure of microfinance institution to these risks. 98. However, conversely, there are two arguments against setting higher capital ratios for microfinance institutions: it results in a less efficient use of capital, and therefore lower profits, and also it restricts microfinance institutions access to borrowing at market rates, limiting their growth. But cross-country data suggest that sustainable microfinance institutions have achieved operating self-sufficiency with fairly highcapital ratios. One option to reconcile these issuesis to have aphased approach to capital ratios, with higherinitial requirements for some period of time, or until some conditions have been achieved, usually financial sustainability, a large equity capital, and better internal and external controls. 62Entry capital for different types of banks is given in a Central Bank notificationof 29 August 2001 (Acesso ao Sistema Financeiro Nacional);for SCMs the most recentrequirementsare in a circular of 27 March 2003. Limits are currently R$100,000 for SCMs, and R$17.5 million for commercial banks and R$12.5 million for multiple banks (higher if foreign exchangeoperationsare sought). Part 2. Exmndina Microfinance 99. An alternative to reducing capital ratios, institutions may be allowed a wider range of products as their maturity increases. Peruvian regulation adopts such a scheme; more equity capital and better audit systems allow microfinance institutionto perform a wider variety of operations. However the capital ratio remains at the 9.1 percent minimum. And in many countries the same capital ratios are required for licensed microfinance institutions as for banks, as it is the case of Peru, Bolivia, El Salvador, and Philippines. Panama and Mexico are still determining the capital ratio for these institutions. In Brazil, a credit-only SCM is constrainedto lowerfinancial leverage, i.e., a capital ratio, of$ve times liquid assets almost twice as high as the capital ratio of 11percent of Tier 1and Tier 2 capitalfor the banking system. Credit Risk Regulation 100. As discussed, the characteristics of microfinance clients increases their credit risk. This may suggest that more conservative limits on credit risk are needed, in terms of provisioning practices. But it i s difficult to determine a cutoff limit for microcredit, microenterprises, or microfinance and any limits set are to some degree arbitrary. Looking at regulatory experiences elsewhere, it i s common for the risk of microcredit to be categorized in terms of past-due days, and for each category a rate of provisioning is determined. Credit risk regulations should also specify how to categorize refinanced or rolled over loans, loans to related parties, and loanconcentration ratios. 4.3 Supervision Issues Who Should Supervise? - 101. Related to this i s the issue of who can provide oversight to MFIs; the banking or financial sector supervisor, a separate or delegated supervisor, or self-regulation. These choices are based partially on the changes that regulation imposes on microfinance institutions in the form of more reporting requirements and the corresponding need for more oversight, and also on the numbers of such institutions, compared to banks and other entities under formal supervision. It can be argued that the supervision of deposit-taking microfinance institutions should ideally be carried out by a bank supervisor. However, the large number of regulated financial institutions has induced supervisors in many countries to adopt other types of supervision, as auxiliary supervision, delegated or hybrid supervision. Inthe case of auxiliary supervision, the bank supervisor i s supported in its task by a third party; in the case of delegated supervision, the supervision of microfinance institutions is transferred to a new supervisor; finally, in the case of hybrid supervision, some criteria are establishedto separatethose institutions which are supervised by the bank regulator, and the remainder, who are supervised by a delegated supervisor. Usual criteria used in this type of supervision are size of assets, degree of financial intermediation (deposit base), and integration with the payment system.63 102. In Indonesia, indirect supervision is used to supervise credit-only microfinance institutions, the Badan Kredit Desas - BKDs. According to law, all BKDs, that are a special type of Bank Perkreditan Rakyat, or People's Credit Banks, are to be supervised by Bank Indonesia. However, considering their large number (5,345) and the fact the BKDs are not mobilizingdeposits from the public, Bank Indonesia decided to entrust Bank Rakyat Indonesia with the supervision of BKDs, performing the duty of a delegated supervisor. 103. Some countries, such as South Africa, have a special supervisor for microfinance. Apart from numbers, there is a case for a specialized supervision approach, for microfinance operations, on account of special characteristics: small size and multiple small transactions, inherently greater risk and volatility 63 These types of indirect supervision are commonly usedin the supervision of closedcredit cooperatives, because the risks faced by the credit cooperativeare self-contained in it, and would affect only its shareholder (discussed further inpara. 149below). Brazii: Access io Financial Services due to the short maturity of their assets, a market concentrated in low segments of population with a lack of physical collateral, a lending methodology based on information rather than guarantees, a loan portfolio composed of a large number of micro loans with a short term maturity, highly volatile, and an institutional structure where shareholders have no "deep pockets". These features change the risk profile of these institutions, which must be adapted for appropriate supervision methods.64These argue for some differences in the approach to supervision, compared to a formal financial institution, as formal supervision could rapidly become very costly. There is thus a tradeof between the introduction of prudential regulation and the costs of supervising compliance with this regulation. This will be a key issuefor Brazil going forwards if SCMs increase in numbers. Brazil will have to consider thefeasibility of maintaining high levels of supervision by the banking superintendent of these entities. 104. Finally, self-regulation could be appropriate for a infant industry providing limited services, and for institutions that do not take deposits from the public. However, self-regulation i s rare and to be effective, should be combined with public ratings of such entities. But there are few rating agencies in developing countries able to rate microfinance institutions. Moreover, rating systems lack sanction mechanisms. There is a general consensus today that although desirable in some circumstances, successfil self regulation is hard to achieve. 5. THECREDIT COOPERATIVEMOVEMENTITSCONTRIBUTIONSTOACCESS AND 5.1 An Overview of the Credit CooperativeSector in Brazil 105. Credit cooperatives were established inBraziljust after the turn of the century, in the state of Rio Grande do Sul, and more than 60 were established, in the same region, until the mid-1960s. At this time, the `open' form of cooperative, known as the Luzzatti cooperative prevailed, with no membership restrictions and thus no restrictions against who could obtain credits. Financial scandals coinciding with reforms in the banking laws in the mid-1960s revealed their risks to regulators and made them uncompetitive with banks. In 1967 there was an outright ban on the establishment of new credit cooperatives. A new cooperative law in 1971 sharply restricted the activities of existing cooperatives and set tight membership criteria. Over the following 20 years more than 50 cooperatives closed down, with losses for their members. Credit unions are legally constituted not-for-profit cooperative financial institutions most part, under national cooperative law and created to meet the basic financial se middle-income citizens. They provide a means to learn the value of regular savings which provides savings, credit and financial management services a credit union only accepts Source:World Councilof Credit Unions: www.woccu.org 64Inthe caseof Peru, it was estimated that if all the costs of supervising microfinance institutions were charged to these institutions, interest rates would increase by an estimated 3 percent to meet these additional costs. Part 2. ExpandingMjcrofinance 106. The cooperative credit movement began to witness a renaissance inthe early 1980s, when central credit cooperatives were permitted which began to provide apex services to member single credit cooperatives. Thus the central cooperative Cocecrer-RS was formed for nine member cooperatives inRio Grande do Sul, and similar central credit cooperatives were established in most Brazilian states over the following two decades. These central cooperatives provided a way for liquidity surpluses and needs to be addressed within a network. In 1995, the National Monetary Council authorized the establishment of cooperative banks, which are regular commercial banks that are owned and controlled by cooperatives. This has enabled cooperatives to begin to gain access to several of the benefits of the regular banking system. Cooperatives continue to count for a very small share of the total financial system, but have grown significantly inrecent years (Table 2.1 and Figure 2.4). Figure2.4 Capital, Depositsand Loans of the Sicoob and Sicredi CooperativeSystems, 1998-2001 R$ millions 2500 2000 1500 1000 500 0 1998 1999 2000 2001 ns Sources: Sicoob: www.bancoob.com.br andSicredi:www.sicredi.com.br 5.2 The Principal Players in Cooperative Finance 107. Cooperatives are formed for a wide range of activities inBrazil and credit cooperatives rank third in importance in terms of number of cooperatives and of employees, but second interms of numbers of members (Table 2.12). Non-credit cooperatives can act as channels for banking loans to their members, but only credit cooperatives are authorized to mobilize deposits and offer other financial services to their members. Credit cooperatives can take three forms: mutual credit, rural credit or Luzzatti. Mutual credit cooperatives can be formed by a minimum of twenty persons who share a common workplace or professional association. Rural credit cooperatives can be formed by persons who work predominantly in agriculture or related activities in rural areas. Up untilJuly, 2003 only ten Luzzatti or open cooperatives were operating in the country and no new ones were being authorized. New regulatory measures introduced in July, 2003 allow for the creation of open admission cooperatives again, under certain condition^.'^ 65Resolution no. 3106 of June 25,2003. The resolution allows new `open admission' credit cooperatives (in municipalities with up to 100,000 inhabitants), subject to the approval of the Central Bank, thus allowing them to be created by location, as well as by activity or profession. Existing credit cooperatives, operating for more than three years, can transform into `open admission' credit cooperatives, in municipalities (or contiguous municipalities) with up to 750,000 inhabitants. Minimumcapital requirements for the transformation is R$6 millionfor entities located in municipalities in metropolitan regions with more than 100,000 inhabitants and R$3 million for the rest. Inthe North and Northeast, this requirement i s reduced by 50 percent. I'rrge 91 Brazil: Access to Financial Services Table 2.12 Number of Coonerativesin Brazil. bv Tvne of CooDerative Agroindustry 1,587 822.3 108.3 Credit 1,038 1,059.4 20.7 Education 278 73.3 2.7 Special 7 2.1 n.a. Health 2 21.4 Housing 297 69.7 1.4 Infrastructure 187 Mining 37 Production 147 Tourism 5 Workers 2,391 322.7 7.4 Total 7,026 4,779.1 175.4 Source: ABRACOOP, www.abracoop.com.br 108. Central cooperatives are apex institutions for member credit cooperative which operate at a state level, and are owned by at least three member cooperatives. They serve to review standards applied by member cooperatives in such areas as accounting, loan origination, staffing and facilities, facilitate flows of funds from cooperatives with surplus liquidity to those requiring additional resources, and assume some auxiliary responsibility from the Central Bank for supervision. 109. Since central credit cooperatives still lacked access to the payments system, cooperative bunks were authorized in 1995.66Two such banks have emerged since then (Bancoob and Bansicredi), owned respectively by the central credit cooperatives in their networks, which act as regular commercial banks. This allows them to operate as correspondent banks for their respective networks and to offer a broader range of financial services that can be undertaken by the cooperatives alone. 66Resolution2193 of the National Monetary Council. Part 2. Expanding Microfinance 5.3 Size, Composition and ServicesOffered 110. In 2001, the Central Bank reported 1,306 credit cooperatives in Brazil. The Brazil Cooperative Organization (OCB) provided data on 1,035 cooperatives, of which 670 are mutual credit cooperatives, and therefore predominantly urban in nature, 355 are rural credit cooperatives and the remaining ten are Luzzattis. An alternative disaggregation from the three major credit cooperative systems in early 2002 indicates 769 in Sicoob, 107 in Sicredi and 33 in Cresol, plus 10 remaining Luzzatti cooperatives or 919 credit cooperatives. In addition, Unicred operates a network of 400 credit cooperatives for medical doctors in 14 states, which has a working agreement with Bansicredi for access to central clearing facilities. (Figure 2.5). 111. Despite numerical dominance by mutual credit cooperatives, at least in the case of Sicoob, the largest cooperative network, rural credit cooperatives have a considerably higher proportion of cooperative financial resources (around 70 percent). Taken together, credit cooperatives as a group account for one out of every seven cooperatives in Brazil, ranking third behind agroindustry and worker cooperatives in number (see Figure 2). They also account for almost one-quarter of all affiliates of cooperatives. Nonetheless, membership in credit cooperatives, at slightly over one million, remains low at less than one percent of Brazil's total population and less than 3 percent of the workingpopulation. Figure2.5 Cooperativesin Brazil,by Type of Cooperative,as of 2001 4 en7 IPICredit 402 1,038 0Education MHousing Health OWork Mother Source: OCB - Organiza@io das Cooperativas Brasileiras: www.ocb.org.br 112. Consolidated assets for Sicoob and Sicredi cooperative networks have risen substantially from 1998 to 2001 to over R$4.5 billion as of 2001.67These assets arefinanced mainly through deposits,which rose from around R$900 million in 1998 to Rs2.5 billion in 2001 and liquid capital, which rose from some R$920 million to R$1.4 billion over the same period, with the difference funded by borrowing. The bulk of the resources went to finance loans to members, which rose from R$1.24 billionto R$2.68 billion. 113. Within the two networks, the respective apex cooperative banks (Bancoob and Sicoob) account for 15-30 percent of all loans and 22-40 percent of all deposits. Deposits and loans grew remarkably between 1998 and 2001 (around 50-150 percent for deposits and 135-200 percent annually for loans). The two banks invest liquidity mobilized by member cooperatives, largely in liquid instruments issued by the National Treasuty and by commercial banks. They also act as channelsfor public resources to the cooperatives,notablyfrom BNDES, to which the cooperativesdo not have direct access. 67 Sincethese two cover at least three-quarters of all credit cooperatives and havereliablemember information,data here are based principally on their information. Brazii: Access to Financial Services The first Cresolc governments. The system grew Sicredi and Sic Mimeo, 2001. 114. In the area of rural credit, the presence of cooperatives is more important, reaching 7 percent of formal rural credit in 2000 (Table 2.13). The sharp difference between their share of the rural market and of the overall market attests to the potential of cooperatives to reach more remote rural clients relatively more easily than banks. This can also be seen in Appendix Table A2.3 below, which clearly shows the presence of a few cooperatives in states in which there i s virtually no presence for commercial and other banks, or even for microfinance companies or finance companies. Their greater rural role probably also reflects the disbursement of government lines of credit at concessional terms to rural entities. Rural CreditCoops 710 5.2 95 7.0 Working Capital 72 5.3 Investment Capital 12 0.9 1,051 77.9 270 29 Source: Banco Centraldo Brasil,2000, "Anuariofstatist/codo Cred/foRural`, Brazil. Outreach Geographicand By Income Level - 115. Between them, Sicredi and Sicoob cover 18 states, with neither cooperative organisation covering the two Amazon states of Amazonas, Roraima and Amaph, or the three North-eastern states of MaranhZo, Piaui, Cearh, Alagoas or Sergipe. With the exception of one state, Mato Grosso, there is no overlap between the states covered by Sicredi and Sicoob, so that member cooperatives are obliged to work under one or the other network, depending on the state, (or establish a separate, smaller and costlier, network or work under the more restrictive regulations applicable to unaffiliated credit cooperatives). 116. In terms of outreach by income level, Sicredi estimates that three to four percent of credits are large, less than 20 percent are medium and the vast majority are small, where small refers to eligibility criteria used for access to PRONAF D loans (namely under R$30,000). A 1999 study of the Cresol rural cooperative network offers more detailed data on the membership by farm size and by income level Part 2. Expanding Microfinance (Table 2.14). In particular, as of that year, 95 percent of Cresol members earned incomes of under R$12,000. This clearly is a lower income strata than the larger cooperative networks. Three-quarters of members have average areas under 20 hectares. 6.6 56.4 10,001 to 12,000 5.3 Over 12,000 5.1 Total 100.0 18.3 Total 100.0 Source. Gilson Bittencourt,1999, "Cooperabvasda Credto",CONTAGKUT. Serie Experibncias09. SBo Paulo/SP. Services Offered 117. The financial services that credit cooperatives can offer depend on whether or not they are affiliated to a cooperative network with central credit cooperatives and a cooperative bank.68 Unaffiliated cooperatives are allowed to mobilize sight and time deposits from members, obtain national and international loans and grants, offer credit to members, place liquid assets in financial institutions, offer collection and custody services, and enter into agreements with other financial institutions to gain access to the check clearing and settlement system and to correspondent banking services. 118. By linking with a cooperative bank, credit cooperatives are able to offer a broader range of services. For example, members of Sicoob and Sicredi issue credit cards, offer internet banking, issue trade credits including letters of credit, provide insurance (life, non-life and rural), as well as extend working capital and investment loans funded by public programs, such as PRONAF and lines of credit under BNDES (e.g. FINAME, Propasto, Prosolo, etc.). They are also able to facilitate forward sales, notably by coffee growers, through the Ckdulade Produto Rural (CPR) in~trument.~~Credit cooperatives cannot offer their members foreign exchange services, leasing or housing finance. 5.4 Penetrationof Brazil'sCredit Cooperatives:InternationalComparisons 119. A comparison of the outreach of credit cooperatives inBrazil with that in other countries suggests that membership in Brazil is very low (Table 2.15)." With around one million members in a population of 170 million, the penetration of cooperative finance remains below one percent of the total population, and around 3 percent of the working population. Deposits in Brazilian credit cooperatives are under 1percent of total deposits in the banking system; the share of loans by credit cooperatives relative to total credit to the private sector is similar. This is very low compared to many countries in North and Latin America. Eligible services were specified under Resolution 2771 of the National Monetary Council, and later by Resolution No. 3106 of June 2003. 69See Part 5 of this study on Expanding Rural Finance. 70Based on data from the World Council o f Credit Unions (WOCCU), which reports more than one hundred million members in almost 37,000 credit unions in 91 countries. However, WOCCU does not report members insuch important countries for credit unions as Italy and Germany, and in other countries, only some of the national credit union networks are members of WOCCU. For example, inBrazil, WOCCU lists 423 credit unions, or less than half of all credit unions in the country. Nonetheless, even allowing for understatement, the conclusions would remain correct. I'crge 9.5 Brazil: Access to Financia/Services Also despite recent overall growth in credit union numbers, the performance of WOCCU-affiliated credit cooperatives in Brazil has been virtually stagnant in the last five to six years (Table 2.16). The low penetration rates of Brazilian cooperatives (when compared with other countries at similar levels of economic development), notwithstanding the long history of cooperative credit in Brazil and the relative depth of Brazil's overall financial system, suggest that regulatory factors played a significant role in retardingthe development of cooperative finance inthe country. Table 2.15 Braziland Other Countries:Credit Cooperatives(2001) (US$ m and percent) Note I/Credit union members over economically active population, percentage.21 Include credit unions and other financial cooperatives. Source:World Council of Credit UnionsStatisticalRepolt 2001 5.5 FinancialStructure and Performance of BrazilianCredit Cooperatives 120. Some widely acknowledged common failings of credit unions in various countries have been avoided to a great extent by the large Brazilian cooperative networks. The common pitfall of low deposit as well as low lending rates does not apply; both deposit and lending interest rates in Sicredi and Sicoob are competitive with the banking system. Second, an effort has been made to maintain skilled staff. In Sicredi, around 60 percent of staff have tertiary level education inrelevant skills. Sicredi management has indicated that it intends to try to maintain salaries at levels that are competitive with private banks so as to retain skills. Third, considerable emphasis has been placed on systems, accounting, training and standards. Part 2. ExDandin0Microfinance Note:1 Credit union membersover working population, percentage. Source: World Council of Credit Unions Statistical Reports riskier institutions, credit union deposi when credit unions eposit rates to this Sources and Uses of Funds 121. Based on data from a sample rural credit cooperatives in the Sicredi network, it appears that (i) such cooperatives arefairly highly leveraged, at ratios at or just below the 10:1 limit. Deposits account for the bulk ofjnancing and borrowingsfrom banks (interbank lines of credit) account for another sizable share (16 percent). Due to high liquidity requirements, 40-50 percent of resources are held as cash and bank deposits (with Bansicredi), and around the same is on-lent in credit operations to members, with rural credit accounting for slightly over one-third of total credit. Fixed assets are kept to a minimum, at 3 percent of total assets. Within the capital account, reserves are more significant than member contributions or retained earnings. Around 40 percent of loans are personal loans, another 40 percent are working capital loans and 20 percent are investment loans, with respective maturities averaging 90 days, 8 months and 3 years. Off-balance sheet items include custodial services, which can be popular. 122. At the central credit cooperatives close to half of funds (48 percent at Sicredi) are kept liquid, around 40 percent is loaned, divided evenly between rural and non-rural credit and the remainder is held Brazil:Access to Financial Services as fixed and other assets. The discounting of public funds accounts for almost three-quarters of rural credit). On the financing side, deposits account for around 60 percent of resources, with most of these (80 percent) held as time deposits and the remainder as sight deposits. Loans and discounts account for just under 20 percent of funding, and the rest is capital. 123. The two cooperative banks, Bancoob and Bansicredi have some structural differences. The Sicoob system is larger than Sicredi in terms of members, but Bansicredi has substantially larger assets than Bancoob (their respective financial statements are in Appendix Tables A2.13 and A2.14). Bansicredi, with slightly over R $ l billion as of December 2001, essentially mobilizes resources through market operations (54 percent) and in turn places over 60 percent of its resources in the market. With a highshare of non-risk assets, its shareholders' equity is equivalent to only 4 percent of total assets. Inthe case of Bancoob, net loans are the most important use of funds, reaching 37 percent compared to Bansicredi's 26 percent of assets. On the liabilities side, deposits account for 37 percent of resources, compared to 17 percent for Bansicredi. Bancoob has a somewhat larger share of its liabilities in the form of long-term liabilities. Leverage i s high, with Bancoob's capital equivalent to just 6 percent of total resources. 124. Inthe case of Cresol, lending was around R$35 million for the 2000/2001 cropping season, with Banco do Brasil financing R$20 million in PRONAF C and D loans, BNDES financing R$9 million in PRONAF C, PROSOLO, PROLEITE and other lines of credit, and own funds supporting close to R$6 million in loans, divided into R$4 million for personal loans and R$2 million. On-lending and deposit interest rates 125. Onlending by cooperatives appears to be at competitive rates for loans from free resources. Sicredi reports that loans to cooperative members using its free resources are at the CDI reference rate plus 4 percent to 12 percent, i.e. about 22 percent to 30 percent, based on a CDI of 18 percent in June 2001. These can be personal loans, working capital or investment loans. Lines of credit and rediscounts by the cooperative banks and central cooperatives to their member cooperatives are at interest rates that adjusted daily according to liquidity and market conditions. For example, Bancoob issues a daily rate sheet in its daily release called "Bom Dia Bancoob". Regarding deposits, cooperatives offer both sight and time deposits which are largely channeled to the cooperative banks. Bancoob offers an ovemight deposit rate to member credit unions of 90 percent of the CDI rate, an indexed deposit instrument for deposits offering a rate that starts at 90 percent of CDI for deposits under 30 days, risingto 100percent of CDI for deposits over 120 days, and time deposits (certificates of deposit) with pre- or post-fixed interest rates and which are in some cases negotiable prior to their expiration. These are subject to withholding tax. Withdrawals prior to 30 days lead to imposition of the IOF financial transactions tax. Cooperative banks place their liquidity largely in Treasury instruments (LTNs) at the SELIC rate, or in commercial bank CDsjust above the SELIC rate 126. Interest rates on publicly sourced funds for rural credit are consistent with their requirements typically for below market interest rates. Most lines of credit stipulate 8.75 percent (e.g. PROSOLO, PROLEITE, PROPASTO), 11.95 percent for tractor modernization, or 1-4 percent on PRONAF loans.'l To fund these, cooperative banks issue intrabank rural CDs to capture the directed rural credit of commercial banks, offering them between 6 percent and 8.5 percent. Where the rate to borrowers is below the cost of funds, as in PRONAFlines of credit, cooperative banks are the only private institutions that receive support through the "equilization" scheme of the National Treasury. 7`See the related paper on ruralfinance by Jacob Yaron and McDonald Benjamin, opcit. I'fr`qe Y8 Patt 2. Expanding Microfinance 127. In the case of PRONAF loans, the Cresol system acts as a channel for Banco do Brasil (BB) to Cresol members. Banco do Brasil receives a spread of 8.5 percent per year, together with a fee of R$13 per month per contract, as `equalization compensation' from the Treasury for handling PRONAF loans. However, only 1.5percent of this spread is passed on by Banco do B r a d to Cresol, which is clearly insuficient when the latter must assume the credit risk in addition to administrative costs. Cresol contends that were it to administer the PRONAFfunds without BB as an intermediary, it could include technical assistance toproducers within the spread that BB receives.12 128. Loans funded by BNDES through its BNDES Automhtico or FINAME lines of credit (the latter are for private companies and rural producers who are purchasing domestically produced machinery and equipment) are at TJLP plus from 4 percent to 5.5 percent. BNDES charges the cooperative banks TJLP plus 1percent, leaving the cooperatives with a 3 percent spread. With TJLP currently at 10percent, the rate to ultimate borrowers is 14-15.5 percent. Portfolio Quality 129. Overall, the delinquency rates in the vast majority of Brazilian credit cooperatives compares favorably with the private banking system, as well as with the prudential standards advocated by the World Council of Credit Unions, namely of total delinquencies to total loan portfolio under 5 percent. Whereas loan delinquencies proved to be a problem in the 1960s and 1970s, and combined with other internal management and external (macroeconomic and regulatory) factors to cause the collapse of a number of cooperatives, loan delinquencies are apparently not a significant issueamong cooperatives that are members of the two major networks-Sicoob and Sicredi. Atlantic Rating's review of Bancoob in June 2001 notes that "Bancoob has never experienced a delinquency since it opened for business.73The bank provisioned R$322,000 under loan loss reserves in the first half of 2001. Write-offs totaling R$567 thousand refer to accounting movements while the effective loss within Sicoob is very low, around 1.5 percent." The cooperative banks are insulated by guarantees from the central cooperatives. Sound management information systems, incentives for cooperative directors and managers and prudent loan origination and recovery procedures (including screening and pressure for recoveries that avail of the local cooperative spirit) help to reduce credit risksat the local level. Note Bankingsystem as of Apr.02; Bancoob as of Jun.01, Sicredi-RSand Encantado as of Dec.01 Sources:Banco Central do Brasil:Nota Para a Imprensa, May 2002, www.bcb.gov.br; Atlantic Rating: Bancoob- Banco Cooperativo do Brasil S.A., June 2001; Sicredi: Communications to Bank Staff; April 2002. 130. In the case of Cresol too, delinquencies average of about 5 percent; ranging from 3.5 percent for rural credit with its own loans and rising to 8 percent for personal loans, with a reduction to 2.6 percent to 72See GilsonBittencourt (2000): "Cooperativismo de CrCdito Solidario: Constituigdo e Funcionamento" ADSKUT, Sao Paulo. 73Atlantic Rating: a localrating companyinBrazil. r v ~ 99~ e ~ Brazil: Access fo Financial Sewices 6 percent after recovery effort^.'^ The delinquency rate is also around 5 percent for loans funded with public resources (such as PRONAF), although a portion are renegotiated to bring the delinquency rate down to 1 percent to 2 percent. Loan losses are substantially higher in a small share of Cresol cooperatives where delinquencies are concentrated in a small number of borrowers, usually cooperative managers who diverted loan funds. Other early problems were excessive loan amounts per member or financing of high-riskactivities, poor managementinformation systems and accounting. Cresol's liquidity was compromised by the need for member cooperatives to assume the credit risk and repay Banco do Brasil on behalf of delinquent members. Consideration should be given to reviewing the extent to which the relative market positions of Banco do B r a d versus smaller rural credit cooperatives explain their relative shares of the spread on PRONAF loans, and what measures can be taken to better relate spread to credit risk and administrative expenditures on suchpublic transfers. Profitability 131. Theprofitability of the two cooperative banks in Brazil is low. The return on assets for Bancoob was below 0.6 percent in 2000 and 2001, and in spite of high leverage, this converted into a return on equity of just 7-9 percent, Le. less than the rate of inflation. The returns for Bansicredi were lower still, with a return on average assets of 0.3 percent and on equity of close to 6 percent. These rates compare to returns on assets at the same time of about 1.5percent in the private banking system and on equity of around 20percent. 132. On the other hand, these two banks are not designed to generate profits at the expense of their cooperative owners, but rather to assist these to generate market retums and increase their coverage. While Bancoob generateda slightly higher profit than Bansicredi, at the systemic level profitability in the Sicredi system seems slightly higher than that in Sicoob (Table 2.18). In particular, returns on equity averaged 22 percent for the Sicredi system as of December 2001, compared to under 10 percent for the Sicoob system, due in large measure to Sicredi's concentration inthe profitable southern and center-west states which have well-established rural producers. Table 2.18 Kev FinancialIndicatorsfor the Sicredi CooDerativeSvstem Rio Grande do SUI 211.7 1,061.4 747.7 1,317.4 40.6 23.7 133. The Cresol system reported a loss in the year 2000 of R$20,000on revenues of R$3.5 million and loans of R$35 million. The loss was attributed to their thin spreads to benefit members, subsidize the establishment of new cooperatives, and cover the credit risks of members on public sector loans such as those of the PRONAFprogram. The system had produced small profits in 1996(R$4,000 on revenues of R$189,000) and in 1997(R$38,000 onrevenues of R$526,000).75 74See Gilson Bittencourt (2000),Central Bank norms require provisioning of 0.5 per cent for loans that are 15 days overdue, 3per cent after 30 days, rising to 100%after six months. Inthe case o f CRESOL, about 20 per cent to 25per cent of the provisioning for loan losses is eventually reversed into revenues due to loan recoveries. 75Source: Gilson Bittencourt and Ricardo Abramovay, 2001, "1novaGo"eslnstitucionais no Financiamento a Agricultura Familiar: 0 Sistema CRESOL" Mirneo. Part 2. Expanding Microfinance Figure 2.6 Evolutionof Profitsand Equityfor the Sicoob CooperativeSystem F@rrillions Source: Sicoob: www.bancoob.com,br 134. A commonfactor depressing profits for all credit cooperatives is the absence of a central lender of last resort. This is because, in the absence of a Govemment-backed liquidity fund, or indeed of any central liquidity funds at the national levels, cooperatives are obliged to auto-insure by maintaining a high ratio of liquidity to deposits, especially to sight deposit^.'^ The 90percent liquidity ratio on sight deposits results in a ratio of liquid to total assets well above the maximum of 20 percent advocated by the World Council of Credit Unions and a low loaddeposit ratio, impeding an effective financial structure. While exemption from income tax provides some relief for cooperatives, one way to compensatefor relatively unremunerative returns on this large amount of liquidity could be to charge correspondingly higher interest rates on loans. This is rendered less feasible through subsidized public programs in rural areas and through competitionfrom banks in urban areas. This intum places a greater onus on cooperatives to cut administrative costs relative to banks, which can affect their public image as well as their competitiveness in terms of staff salaries. Already, at slightly under 4 percent, personnel and administrative expenses for the Sicredi system in Rio Grande do Sul are indeed low by Brazilian banking system standards, and also at the low end of the 3-10 percent range advocated by WOCCU. 135. An additional factor affecting not only profitability but the viability of cooperatives in certain areas i s membership restrictions. Clearly banks are concerned that the exemption that cooperatives enjoy from IOF and income taxes would lead to "unfair" competition if membership restrictions were to be lifted; on the other hand there are areas of the country in which a restriction by location or by profession results in insufficient members to make cooperative banking viable. There i s no clear answer to this. The temptation to adjust membership restrictions with a view to expanding access is high, but these may result inadvantagesto cooperatives purely from regulatory arbitrage, and would foster the formation of separate and fragmented regulatory windows. It is not obvious that the members of such cooperatives would be the poor of these areas. 76 This appearsto be a frequently encountereddifficulty for credit unions.Recently, ABCUL, the associationof UK credit unions, has recommendedthat a Central Services Organization be established in the UK to provide credit unions with central treasury and liquidity management functions, inthe hopethat this will assist UK credit unions to grow. (ABCUL ReportonFinancial Exclusionand Credit Unions, August 2001). Brazii: Access io Financial Services 6. THELEGAL, REGULATORYAND SUPERVISORYFRAMEWORK CREDIT COOPERATIVES FOR 6.1 Legal Framework and Prudential Regulation 136. UntilJune 2003, credit cooperatives fell under the general law on cooperatives, Law No. 5764/71 of 1971, though as financial institutions, they are also subject to the general Financial Institutions Law No. 4549/64 of 1964, and the Resolutions of the National Monetary Council.77Just prior to the specification of prudential norms for SCMs, prudential guidelines and membership criteria for credit cooperatives were ~pecified.~'Together with other new measures for microfinance, new regulations for credit cooperatives were also introduced inJune, 2003.79 137. Eligibility criteria for membership of credit unions in urban areas, related to work or place of employment have been similar to those in closed cooperatives elsewhere. For example, in the United States, just as in Brazil, credit unions enjoy tax exemptions that commercial banks do not have, and therefore eligibility restrictions have been placed on membership of credit unions in order to avoid "unfair" competition with banks." Although the new regulations of June, 2003 have somewhat reduced constraints against open admission cooperatives, these are only allowed in thinly populated areas, where other financial services may be absent Prudential regulations against credit risk were also specified, with limits on risk concentration through single party exposure limits (credit to any member cannot exceed 10 percent of liquid capital for affiliated cooperatives or 5 percent for unaffiliated cooperatives); and limits on exposure to a member cooperative and at a central cooperative (less than 20 percent of liquidcapital).*' 138. The same resolution established minimum entry capital requirements for central cooperatives (R$60,000 for new institutions, rising to a minimum capital of R$300,000 by their fifth year of operations) and member cooperatives (R$3,000 at the outset, rising to R$60,000 over five years). In line with sound prudential principles, more stringent minimum capital requirements apply for non-federated cooperatives (R$4,300 at the outset, rising to R$86,000 over four years). Nevertheless these minimum capital requirements are very low even by comparison to Brazil's microfinance societies, and more so relative to other countries. Like SCMs, there are no ongoing capital adequacy ratios, but instead limits to leverage (ten times liquid capital for single credit cooperatives that are members of a cooperative system and five times liquid capital for unaffiliated credit cooperatives). However leverage ceilings are relaxed substantially for rural credit cooperatives that participate in the PRONAFprogram. The decision to relax leverage ceilings in the event of PRONAF clients is not prudentially sound in light of the additional covariant risk exposure it createsfor cooperatives among a clientele whose capacity to repay loans at market rates remains to be proven. 77 The 1988 Constitution in its Article 192 provides for distinct treatment for cooperatives relative to other financial ''institutions. Resolution No. 2771 o f August 2000. 79 Resolution No. 3106 o f June 25'h, 2003 which superseded the rules of August 2000. These were followed by Circular 3.196/03 of July 17* 2003, the Central Bank, which specified new capital adequacy rules for credit cooperatives, of 11percent for affiliated coops (as in banks) and 15 percent for cooperatives not affiliated with a central cooperative. 8o See the National Credit Union Administration: www.ncua.org for the US case. Inthe case of Brazil, credit cooperatives are exempted from the 1.5 percent IOFtax on financial transactions, the income tax on profits, and reserve requirements although the cooperative banks that they own need to meet these requirements in the same way as regular commercial banks. ''Resolutionclearlyno No. 3106of June 2003. There are exposure limits upwards regarding the placement o f deposits by member cooperatives in their central cooperative or cooperative bank. Part 2. xpanding Microfinance 139. The most stringent requirements for Brazil's cooperatives are related to reserve requirements on deposits. Whereas cooperatives are not required to place such reserves with the central bank, they are obliged by central cooperatives to place 90 percent of sight deposits and 25 percent of time deposits in liquid instruments. Thus loanable resources are substantially lower than for banks. This prudential requirement reflects the lack of access to deposit guaranteesfor cooperatives. Commercial bank deposits are insured by the Fundo Garantidor de Crkdito (FGC), but the FGC does not cover deposits in cooperatives. With the introduction of new regulations in mid-2003, formal requirements for the extension of credit, based on deposits, havejust been introduced for open cooperatives, which require a minimumof 50 percent of deposits to be usedfor credit.83 140. In the United States, there has been a separate deposit insurance fund for credit unions since 1970, known as the National Credit Union Share Insurance Fund (NCUSIF), which i s built up with premia from credit unions to an amount close to 1.3 percent of deposits, and backed by the full faith and credit of the United States Government. In addition, a lender of last resort was established for US credit unions in 1979, known as the Central Liquidity Facility. Inthe case of Brazil, apex commercial banks for each coop network have been partially able to offer such services for cooperatives, through their own access to lender of last resort funds. In addition, Sicoob i s considering establishing a network-wide deposit insurance system, most likely with a R$20,000 per account ceiling at a cost of R$0.03 per Real of insurance coverage. Four of the fifteen central credit cooperatives in Sicoob already operate state-level self-insurance funds. Unlike the FGC, or the NCUSIF in the case of the United States, these insurance funds do not have public backing, therefore the risk will inevitably continue to be higher even ifthe funds are eventually consolidated at the system level, but this may prove to be a valuable first step to assess the viability of the fund before extending sovereign backing. 141. There is a clear casefor providing incentivesfor the establishment of a liquidity facility for credit cooperatives. Credit cooperatives currently face high opportunity costs related to liquidity requirements, relative to commercial banks that are backed by the FGC. One option to be studied would be to encourage the establishment of a central liquidity facility for credit cooperatives, that i s inthe first instance privately managed by representatives of cooperatives, under BCB supervision, and that enjoys clear, BCB- approved standards for liquidity support and i s capitalized via a percentage of deposits. Inturn, liquidity requirements on sight and time deposits could be eased, with the reduction in liquidity requirements related to the degree of capitalization of the facility. After evaluating the performance of the facility, consideration could be given to public backing for the facility. 6.2 Supervision 142. Credit cooperatives are supervised by the Banco Central do Brasil, which approves applications for their establishment and appointments of boards of directors, and establishes reporting requirements. Routine oversight, as a form of auxiliary supervision, is offered by all cooperative networks and has increased at the two main cooperative credit networks, Sicoob and Sicredi. These networks apply standards to varying degrees through their central credit cooperatives, which use moral suasion to enforce prudential standards. This moral suasion is of course backed by legal remedies in extreme cases, but more typically standards are enforced through performance-related pay for cooperative managers, network-applied sanctions against the single cooperatives' directors in the event of non-compliance with guidelines, restrictions on credit to weaker cooperatives (e.g. Bancoob does not lend to member cooperatives with C ratings in its internal scoring system), and the ultimate sanction of expulsion of the cooperative from the network (while the ultimate sanction of the Central Bank i s closure and liquidation). 83 Resolution 3.106 of June 25ith2003, of the National Monetary Council, which requires a minimumof 50 percent of the mean value of daily balances of monthly deposits to be used for credit. Page IO3 Brazil:Access to Financial Services 143. Over the last five years, the Central Bank`s confidence inthe ability of all apex cooperatives, and notably, based on information collected by this study, Sicoob and, Sicredi, to supervise their member credit cooperatives has clearly grown, due to the implementation of effective information systems and reliable reporting by the two networks, which allow for integrated supervision and benchmarking of financial performance. The two networks clearly differ substantially in terms of their philosophies, with Sicredi applying more rigorous harmonization of standards from loan origination guidelines down to the physical appearance of Sicredi affiliated cooperatives' buildings. Moreover, while Sicoob affiliates have often conserved their earlier depository relationships with various commercial banks, Sicredi affiliates deposit their surplus funds exclusively in Bansicredi. The differences in philosophies led member cooperatives in the state of Santa Catarina to disaffiliate as a group from Sicredi andjoin Sicoob. 144. As with banks, all loans of R$5,000 or more must be reported to the "Central de Risco de CrCdito" or central credit bureau run by the Central Bank.*4 Thus clients who fail to repay one institution should be unable to obtain credit at others. Cooperatives can apply the same sets of policies regarding loan origination and enforcement as applied to other financial institutions. As in microfinance, cooperatives take advantage of the `reputational knowledge and collateral' in terms of community spirit and knowledge of clients to ensure more effective screening of clients, as well informal sanctions to promote timely repayment of loans. While Sicredi has enough information in its system to be able to design a credit scoring system, it does not want to substitutefor the personal knowledge that comesfrom the local operation of a cooperative venture. 6.3 Conceptual Issues in the Regulation and Supervisionof Credit Cooperatives Regulation 145. The distinctive characteristics of credit unions have induced regulators to introduce unique elements in the regulatory framework for closed or open-common bond credit cooperative^.^^ First, unlike banks, who are open to provide financial services to any person or enterprise, credit cooperatives restrict their operations to their members (closed credit cooperative) or those eligible for membership (common-bond credit cooperative). Therefore it is argued that the public interest motive for financial regulation and supervision is reduced to only those who are members of the cooperative. Nevertheless, a closed credit cooperative can be larger in asset size than a bank, as it i s the case in Peru, where the three largest credit cooperatives are larger in size than the smallest bank in the system. Therefore, selecting a criteria of openness or closeness to regulate and supervise a credit cooperative can be misleading. 146. It is not uncommon to encounter, as in Brazil, more relaxed regulation in terms of entry capital, and in some cases more relaxed capital ratios (in the case of the US). Reserve requirements at the Central Bank are not usually required although credit cooperatives take deposits from their members. This is the case inBrazil and Peru. There is also a lack of access to deposit insurance. This increases the public perception that credit cooperatives are more risky institutions although these institutions may manage more money from the public than open credit unions or even banks.86There i s also no access to a 84Resolution 2798 of November 30,2000 (See Chapter 6 of this study for a further discussion). Hennie Van Greuning, et. al., 1998. (op. cit). 86InLatin America, Peru, Ecuador and Argentina have formal deposit insurance for credit cooperatives, but in reality they do not work. Inthe case of more developed countries, like Germany and the US, credit cooperatives have their own deposit insurance fund. Brazil and Colombia are moving towards establishing their own deposit insurance fund. See Matthias Arzbach and Alvaro Duran "Proteccidn de Depdsitos para Buncos y Cooperativas de Ahorro y Crtdito en Amtrica Latina" DGRV (2002) for a discusi6n. Part 2. Expanding Microfinance liquidity facility.87Instead, central cooperatives or federations perform that duty, to a limited extent since they are also financial institutions. 147. Credit cooperatives also commonly face a number of privileges and constraints with respect to other financial institutions. There are usually tax advantages, not only in Brazil, but also in Germany, before cooperative banks became open; geographical restrictions on operations to level the competition between cooperatives and banks." Commonly, credit cooperatives also suffer from instability of equity as exit of a member from a cooperative can not only damage liquidity (withdrawing their deposits) but also solvency (by withdrawing contributions, the essential part in a cooperative's eq~ity).'~Problems of governance are much higher in credit cooperatives, both due to incapable management or abuse of position. ConceptualIssues in the Supervisionof Credit Cooperatives 148. There are two central and interrelated issues in this context: which credit cooperatives should be supervised, and who should undertake this. Regarding the former question, it is argued that open cooperatives should clearly be subject to the supervision of either a Superintendency, the Central Bank or other state agency. The question concerns closed or common-bond cooperatives. But criteria such as juridical form, opennesskloseness or common-bond have generated inconsistencies since some large credit cooperatives have been left out. A combination of size and openness, as used in Bolivia, Chile, Colombia and Ecuador, is suggested. In Brazil, all cooperatives are supervised by the Central Bank. This may a large burden on supervision capacity goingforward. 149. To reduce supervisory costs, a number of countries have adopted methods of indirect supervision. These include, first, auxiliary supervision.The public agency in charge of the supervision of the financial system remains with the responsibility of the supervision of credit cooperatives, but it i s supported by private agent(s) (federation(s) or association(s) of credit cooperatives, bank(s) or central(s) cooperatives) basically in the process of on-site supervision. The authority to intervene or sanction a credit cooperative remains in the public agency, which conducts regular reviews on the performance of the private agent. This is the case of Germany and the United States. Delegated supervision is an alternative. All supervisory functions are carried out by the private agent(s), even interventions and sanctions. However, the design of the regulatory framework remains in the public agency, that also conducts reviews on the performance of the delegated supervision. This i s the case of Peru and Costa Rica. Lessonsfrom Regulation and Supervisionof Credit Cooperativesin Developed Countries 150. In Spain, the United States, Canada and Germany credit cooperatives are much more developed than in developing countries, not only by their outreach (measured as penetration ratios) but also by size in the financial system. A stable macroeconomic and political environment may have contributed, as well as the supervision and regulation applied to credit cooperatives by these countries. Potential factors could include the following: = The United States and Germany have very strong and organized indirect supervisors of the cooperative sector, The National Credit Union Administration in the US and the Regional 87In the US, the revisedFederalCredit UnionAct of 1998 includes inthe Title I11the NationalCreditUnionCentral LiquidityFacility to improvegeneral financial stability by meeting the liquidity needs of credit unions. The idea is that cooperativescan hardlyexpandtheir services to a large geographical area since their qualifying members are usually concentratedinone area. But limiting geographical spread may increaseriskconcentration. 89Regulationon this matter has already beenimplementedinBolivia, Peruand Germany. Brazil:Access to Financial Services Federations of local cooperatives in Germany are highly trained in the supervision of credit cooperatives, and both are under permanent inspection from the banking supervis~rs.~~ 3 There is a clear delimitation of supervisory roles from representative roles in these indirect supervisors. In Germany, auditing and auxiliary supervision is conducted by regional Federations, while their representation is done by two other federations, DGRV (German Confederation of Credit Cooperatives) and BVR (Association of German Popular Banks and Raiffeisen Banks). This reduces the conflict of interests inside the supervisor when it both has to supervise and represent the credit cooperatives. + The regulationapplied tocreditcooperativesdoesnotdifferfrom theoneapplied tootherfinancial institutions. This reduces regulatory arbitrage and provides incentives fair competition, reducing also the public perception that credit cooperatives are riskier than banks. 3 Credit cooperatives have also had access to liquidity facilities, deposit insurance and credit bureaus, either establishing their own (deposit insurance and liquidity requirements) or using the same available for other financial institutions (credit bureau). 6.4 Suggestions for Brazil 151. Brazil should try asfar as possible to apply similar financial regulations for credit cooperatives asfor otherfinancial institutions. Credit cooperatives, as financial institutions, whether are closed, open, or common-bond, should be under the same standards of regulation applied to other financial institutions, to motivate competition and reduce their perceived higher risk associated with the lack to access to a liquidity facility, deposit insurance and credit bureau. Certainly some special features or credit cooperatives may differentiate their regulatory framework, as the composition of their capital and their loan portfolios, but in general, regulations on solvency, credit risk, liquidity risk, market risks and other aspects of financial regulation could be the same for credit cooperatives as for other financial institutions. Recent measures taken to ensure common capital ratios for cooperatives and banks are therefore a step in the right direction. 152. This also suggests a need to review regulations and incentives governing public rural credit programs channeled through credit cooperatives. Credit cooperatives are at their most effective when members feel considerable ownership of their cooperatives. When resources are perceived as coming from the State, particularly for disadvantaged populations, there are strong incentives not to repay. Policies such as those that doubled the leverage permitted in the event of financing using public rural credit programs such as PRONAF, simply reinforced incentives that can undermine the cooperative spirit and the sustainability of the cooperatives. These regulations run counter to what i s in general a prudent regulatory framework and the removal of such policies under a new regulation introduced in2003 is most welcome.g1 153. Direct supervision of the entire cooperative network would strain supervisory capacity and the present delegation of supervision appears to have many advantages. In thefiture, a clearer delineation of duties between the Central Bank and the central cooperatives may be desirable. A prudent balance must be maintained between direct and delegated oversight of credit cooperatives. The relatively low capital standards that facilitate broad access to cooperative financial services can also imply a proliferation of small formal financial institutions, with high costs to the supervisory entity, particularly in a country the size of Brazil. For example, in one Asian country, rural banks account for only 2 percent of banking system assets, yet they outnumber commercial banks by more than 20 to 1innumber and as a group they absorb one-third of banking regulators. Recent regulatory changes permitting open membership 90The NationalCreditUnion Administrationis the independentfederal agency that supervisesand insures the federal credit unionsand state-charteredcredit unions. 9'ResolutionNo. 3106 of June 25,2003. Part 2. Exmndina Microfinance cooperatives, and conversions to such structures by existing cooperatives, could lead to a potential expansion in numbers of cooperative members. In order to avoid high costs of supervision and the attention of a large share of Central Bank supervisors to a small share of systemic assets and systemic risks, it is important to continue to strengthen capacity for prudential oversight among cooperative networks, particularly through strengthening systems, accounting standards, and transparency in reporting, and to continue gradual extension of delegated supervision to cooperative networks that meet Central Bank standards of oversight, while trying to avoid the conflict of interest of these networks who at the same time supervise and representcredit cooperatives. 154. Another way to supervise the large number of credit cooperatives in Brazil is to adopt hybrid supervision, combining criteria of size and openness, in which a few credit cooperatives would be under the direct control of the Central Bank, while the rest, a large number of small credit cooperatives, would be under delegated supervision of the cooperative networks or other third party. Brazil: Access to Financial Sewices ANNEX 2.1 MICROFINANCE AND TECHNOLOGY 155. All forms of microfinance institutions have greatly benefited in recent years from a series of technological innovations which have expanded the outreach, enhanced the productivity, reduced costs and expanded the product range of microlending and lending to the poor. Yet with each, there are also limits to their usefulness in terms of substituting for conventional lending techniques. This annex briefly describes some such technological innovations, discusses their advantages and limitation and presents examples of their recent applications in mi~rofinance.~' 1. AUTOMATION AND COMPUTERIZATIONINMICROFINANCE Handheld computers/PersonalDigital Assistants (PDAs) 156. Handheld computers and PDAs which can be connected by modem to central management information systems, allow an MFIto deal with the geographic dispersal of clients by taking the office or evaluation techniques to the field. They allow loan officers to record client data, upload applications from the centralized database, and conduct quick and automated credit analysis. The accuracy and efficiency of lending decisions is enhanced while processing time and costs are greatly reduced. The time saving increases loan officer capacity, which is a leading indicator of efficiency, and hence returns. 157. Unlike traditional banks, microfinance institutions do not make loans based upon revenue or collateral alone, but also look at references from customers and neighbors, in addition to the loan officer's own per~eption.~~Although handheld computers are a powerful tool for making micro-loans, they will never replace loan officers. The personal knowledge or client relationships of loan officers or of loan groups remains integral to the lending process. ATMsand Debit Cards 94 158. ATMs are cost effective means of providing financial services, as they increase geographic outreach, maximize hours of operation, and reduce employee costs, as human tellers do not have to conduct the transactions. The ATM option is especially cost effective for geographic targeting within poor pockets in city locations; maintaining ATMs in geographically remote areas has limits on cost effectiveness just as bank branches have, due to needs to service machines and install infrastructure. The use of conventional ATMs may also be limited in some low-income contexts due to illiteracy, or where persons may be intimated by technology. To promote the uses of ATMs among the poor, they have been customized in many countries. For instance, in Bolivia, special ATMs capable of overcoming language barriers and illiteracy, and with functional simplicity have been developed. Instead of entering an identification number to access their account, customers places a finger in the given location to enable the machine to detect the fingerprint and identify the client. After the identification stage, a friendly voice 92See especially Campion and Halpern, 2001, `Automating Microfinance-Experience inLatin America, Asia and Africa', Microfinance Network, Occasional Paper No.5 and Marker, P, Kerry McNamara and Lindsay Wallace, (2002), `The Significanceof information and communication technologies for reducing poverty', Department for International Development, U.K.; for recent surveys. 93Beebe, Christianna, 2000, `Palm Pilots: Lending Efficiency to Microcredit', Interaction - Monday Developments, Washington DC, USA. Available at http://www.microfinancegateway.org/cache/palm.pdf 94For more information on ATMs, see Chapter 3 -Downscaling Private Banks Part 2. ExpandinaMicrofinance welcomes the customer in their native language. and directs users through the desired transaction, through touch-activated screen icons based on pictures and symbols.g5ATMs have been particularly popular for transmission of funds to rural pensioners who otherwise paid check-cashing fees to merchants. It reduces theft of checks, fraudulent claims and transaction costs for the recipient. Provincial governments in some countries award contracts to cross-subsidize this marginally profitable service. Smart Cards 159. Smart cards are ATM cards with an embedded microprocessor chip which can be used to complete financial transactions. A smart card resembles a credit card in size and shape, but inside it i s completely different. Unlike a normal credit card which is a simple piece of plastic with a magnetic stripe, the smart card contains an embedded microprocessor chip that can hold up to 800 times the information that a magnetic stripe can, including personal information (e.g. identification via biometrics) and consumer information. It can function as a debit card and/or credit card, store account information for several different financial products, and be programmed to hold and transfer money.96 The user can preload a certain amount of money into the card. Vendors that accept smart cards can pass the card through a reader that immediately transfers the appropriate amount in their accounts. Where infrastructure exists, smart cards can pay for public transportation, goods, services, items from vending machines, loans, pay phone calls and other consumers goods. Because they hold customer account information and personal details, smart cards are more secure than ATM or credit cards. Smart cards if programmed can offer an unprecedented level of security with fingerprint imaging. Unlike debit cards and ATMs which require expensive terminals and reliable telecommunications, smart cards can function off- line with battery-powered readers without a permanent network connection. 160. For microfinance, smartcards improve efficiency through card security and bulk electronic settlement routines that produce accurate accounts, provide a fully transparent and tamper-proof electronic audit trail, and reduce central accounting time. Similarly, this encourages merchants participation becausethey benefit from reduced cash handling. The improved efficiency and the low price suggest that smart card technology will become more important in the transactions of poverty-oriented programming, but it is premature to predict the types of products that will be developed inthe future. Credit Scoringand risk metric^^^ 161. Credit scoring i s used by lenders to help them calibrate the credit risk of loan applications. Information such as income, bill-paying history, late payments, past records on loans, age are collected, aggregated using a statistical program and then compared to the credit performance of consumers with similar profiles. Credit scoring models are being usedextensively in microfinance. A typical microfinance model may take into account past repayment records, levels of indebtedness, the length of time a customer has been borrowing, net family income, net business income, whether the applicants rents or owns his house and place of business, among others. However, to be able to make accurate decisions, credit scoring models require information which may not be easy to come by with microclients, especially in countries where comprehensive credit histories from credit bureaus and collection agencies are scarce, as in most developing countries, that possess adequate records of microfinance loan clients. 98 Even in advanced countries, such as the UK and USA, credit scoring methodologies are often not applied to new 95For more information see http://news.bbc.co.uW2/lowlbusiness/l3863l0.stm 96Campion and Halpern, (2001op. cit.). 97See also Chapter 6 - Installing Financial Infrastructure 98Schreiner, Mark. "Credit Scoring for Microfinance: Can it work?" Published by the Center for Social Development at Washington University in Saint Louis. Available at http://econwpa. wustl.edu/eps/dev/papers/O 108/0108003.pdf Brazii: Access to Financial Services small business entities, who have little prior credit history. Beyond credit scoring, there are comprehensive risk managementtools for extending microcredit, which attempt to minimize unnecessary exposure to underperforming loans and capital losses. Two Mexican financial institutions currently using such risk management tools, perfected by a Spanish company, AIS, are Nacional Financiera and Financiera Rural (formerly Banr~ral).'~ Telecommunications and the 162. Telecommunications, particularly the Internet have a huge poverty alleviation potential. The internet promises to make virtually unlimited amounts of information accessible from anywhere in the world, and to defy traditional borders and boundaries by allowing people from various countries, communities, religions and ethnicities to exchange ideas openly and instantaneously. Making true the potential of the Internet, however, poses enormous challenges. The greatest challenge by far i s overcoming the so-called digital divide. The digital divide i s the disparity between those with access to information and communication technology, versus those without. Inthe developing world, a computer is a luxury, and even phone lines and electricity are not always available. Although the average OECD country has approximately 11times the per capita income of a South Asian country, it has 1,036 times as many Internet hosts, 40 times as many computers, and 146 times as many mobile phones."' 163. The most promising developments seem to be happening in finding alternatives to landline dial- up access. Wireless networks can be set up at a much lower cost than landline networks. Inimpoverished areas where isolation and poor infrastructure are often the norm, cell phone use can play a crucial in enhancing the social and economic development. Cell phones have been particularly helpful for enhancing markets for mkroentities; women in rural areas can use cell phones to make sell calls, to order supplies, or to check commodity prices inthe city to negotiate better prices for their crops. InKenya and in India, fishermen have used cell phones to determine catch prices at different ports so that they can direct their returning trawlers back to the most profitable locations. In Bangladesh, thanks to the Village Phone Program, several thousands villages have access to cell phones through Grameen Telecom, a subsidiary of Grameen Bank. Loans repayments for phones are in accordance with Grameen Bank procedures. 2. EXAMPLES OF APPLICATIONS OF NEWTECHNOLOGIES INMICROFINANCE 2.1 Debit Cards Banco ADEMI - 164. Banco ADEMI is a large MFIand one of the most profitable banks in the Dominican Republic. Banco ADEMI offers its own plastic payment system, the ADEMI+ debit card"."' Under this scheme,the debit card i s linked to a new loan product, the "Prestamo con ahorro" (loan with savings). When a customer applies for a loan up to RD$200,000 (or $12,500), she agrees to deposit 10 percent of the loan value into a Banco ADEMI savings account and to allow the bank to automatically withdraw future loan payments from that account. Inexchange, Banco ADEMI authorizes loan amounts 10percent higher than otherwise. The additional loan funds are then placed in savings at the time of the loan disbursement. The funds are not specified as a loan guaranty because the customer can withdraw them at any time. However, the customer must keep the savings account open throughout the loan cycle, which implies O9AIS hasdevelopedtools which can measureand managecredit, liquidity and market risks (see www.ais-int.net). looSee especially Part 3 of this study - DownscalingPrivateBanks. Io'The World Bank,2002, `Information and CommunicationTechnologies', WashingtonDC, USA. Io*Campionand Halpern, (2001op. cit.). Pad 2. fxmndina Microfinance maintaining sufficient funds to cover loan payments and a minimumbalance of RD$100 (or $6.25). Upon opening the savings account, Banco ADEMI issues a debit card to the customer to make deposits or withdrawals from the account from any automatic teller machine (ATM) in its network.Io3 2.2 Credit Cards for Farmers The Experience of Kisan - 165. The Kisan Credit Card Scheme was introduced in India in 1998 so that farmers could gain easy access to credit for working needs in their agricultural operations. It as been opened to all farmers prepared to borrow above a threshold of Rs 5,000 (around US$lOO), who have not defaultedin repayment of loans in the past three consecutive years, with a minimumof 5 acres of irrigated land. The credit limit is set on the basis of land holding, the value of the crop cultivated, and income. Farmers can get loans up to five times the annual farm income or 50 per cent of the value of the land as collateral, whichever is lower.'04 Kisan Credit Card Scheme participants are given a credit card and a check book and are issued a pass book incorporating identity and transaction details.Io5 Revolving credit i s extended with unlimited withdrawals and repayments within the overall ceiling - no loans should remain outstanding for more than 12 months. These repayment conditions provide much-needed flexibility to farmers. Kisan cards are valid for three years subject to an annual review basedon performance. 166. Participating commercial banks, regional rural banks and cooperative banks provides the Kisan cards to the farmers. "The Scheme has gained popularity and its implementation has been taken up by 27 Commercial banks, 373 District Central Co-operative Banks/State Cooperative Banks (two-tier structure) and 196 Regional Rural Banks throughout the country. As on November 30, 2001, 20.4 million Kisan Credit Cards involving credit sanctions of around US$ 7.5 million had been issued. Cooperative Banks accounted for 66 percent of Kisan Credit Cards followed by commercial Banks (27 percent) and Regional Rural Banks (7 per cent). Personal insurance cover for accidental death or permanent disability for KCC holders has also been finalized, which offers protection to creditors.Io6 167. However according to the Reserve Bank of India (RBI), the success of the KCC scheme is debatable from some points of view. On the whole the scheme has been well received by farmers in terms of timeliness, hassle-free operations and adequacy of credit. But on the other hand relatively few farmers get Kisan Credit Cards."' Many who could benefit from the scheme do not possess land, or title, or are simply illiterate and unaware that such a scheme exists. There i s an unfortunate absence of any village survey by banks to identify and target eligible farmers. Another key drawback of this scheme i s its fungibility. Indeed, the State Bank of India, has recently found that most farmers are using the Kisan cards for purchase of consumer durables instead of farm inputs.'08This practice could lead to more unproductive indebtedness. IO3 Campion and Halpern, (2001 op. cit.), p.19 104 The Hindu Business Line. "SBI Kisancards turn FMCG golden goose". Available at http://www.blonnet.com/b1ine/2002/03/09/stories/2002030902100100.htm. IO5 National Bank for Agriculture and rural development. "Kisan Credit Card Scheme". Available at http://www.nabard.org/roles/kcc.htm lo` The Governmentof India, Ministry of Finance. Available at http://indiabudget.nic.in/es2OOl- 02/chapt2002/chap813.pdf. lo'Menon, Rajalakshmi from The Hindu Business Line. "Farmers don't fancy credit card facility". Available at http://www.blonnet.com/bline/2002/06/25/stories/2002062501571000.htm The Hindu Business Line. "SBI Kisancards turn FMCG golden goose". Available at http://www.blonnet.com/bline/2002/03/09/stories/2002030902100100.htm Brarii: Access to Financial Services 2.3 HandheldslPDAsand Smart Cards Swayam Krishi Sangam - 168. Swayam Krishi Sangam (SKS) i s a relatively successful MFIin one of the poorest drought-prone regions of India. It was founded in 1997 and works with solidarity groups (Sangams) of very poor women, with a mission of sustainability. It provides loans, savings, insurance, and other financial services to its customers. Service delivery, including loans disbursement, savings withdrawals, collecting due payments and making proposals, would occur during solidarity group meetings, and used to be very time-consuming as all transactions were recorded manually. Moreover, meetings could only took place within a small window of time early in the morning since the rest of day i s spent working inthe field. To increase the effectiveness of micro-loan delivery, SKS has developed a technology based using Smart Cards and hand-held computers (HHCs) to automate approval, disbursement, and collection. SKS provides each group member a smart card which i s read by a card reader on the loan officer's hand held computer, which displays the information contained, and allows a synchronization of information between the HHC and smart card once transactions are completed. After meeting with all the groups, the loan officer returns to the SKS office to synchronize and update the branch computer with the new data on the HHC. SKS affirms that the HandheldSmart Card technology has significantly lowered SKS's cost of operation, even beyond anticipation. Annex Table A2.1 SKS Time Savings and Cost Savings: Anticipated and Real (US$ and minutes) - Pledgeand Attendance 2 2 2,3 2, 3 8 4, 11 4,11 ' RecordingCollection Sheet 10 0 Printed Material 4,143' 1,3162 1,316 Total 76,143 56,229 51,515 Total Savings 19,906 24,628 Note: lPassbooks=$3,556 and Collection Sheets= $587. zReceipts= $889 and Monthly Statements= $427 Source:Campionand Halpern, (2001 op. cit.), 169. The financial sustainability of microfinance institutions (MFIs) like SKS has continuously been under strain-not only due to their clients' generally small loan portfolios, but also due to the hightravel costs to reach clients remote villages. With the need technology, the need for sangam managers to manually enter transactions into passbook is completely eliminated. Thus, staff can cut down sangam meetings time and service more groups while reducing the risk of fraud and error. The time and cost savings can be productively used for scaling up the SKS microfinance activity. 2.4 BASIX and the Sudama Project 170. BASIX, a leading MFIin India, focuses its operations on addressing the tiny credit need of low- income borrowers, often located in remote villages and ensuring commercial viability of their ventures. Part 2. ExDandinaMicrofinance Its most popular products among the poor involve tiny transactions of say R s l O ($0.20) in regular intervals. These could be weekly savings and or weekly repayments of small amounts. Small daily savings are also popular. Borrowers are issued smart card technology based Sudama cards, which act as identification cards and electronic loan books. The card itself is issued for a fee on registration and repaid in small installments. Once the payments for the card fee have been completed, the borrower becomes eligible for a loan. Cards are read at the `BASIXPOT', or Basix Point of Transaction, by a service agent with a computer equipped with a smart card reader and modem. Information is securely transferred to the BASIX Unit server each evening with a call which takes about a minute and costs less than 5 US cents. The agent uses the local bank for collections and disbursement of loans. He works on a commission basis linked to the nature and number of transactions, number of new borrowers registered and the quality of the portfolio. Once the new product was introduced, the agents in one location (Pala Sammudra) registered 70 new borrowers in one month and cut costs to a third and it has enabled its micro-credit operations to work with poor borrowers spreadover a large rural area. This project was partly funded by a Technical Assistance Facility of the InternationalFinance Corporation (IFC). Annex Table A2.2 BASIX No. of branches 10 Loan portfolio outstan Financialself-suficie Source.www.basixindia.com 2.5 Postal Services and the Internet-South Africa's PiTsiog 171. PiTs or Public Information Terminals (Internet kiosks), ajoint project between the South African Department of Communications and the South African post office, was introduced in 2001 as a means of utilizing the vast rural reach of the South African post office network to address the financial needs of low-income and rural citizens. In 2001, South Africa had a population of 40 million that were serviced by South Africa Post Office in over 2700 outlets and 5500 service points. More than 100rural post offices had PiTs installed, providing internet access to over 20 million people living in rural areas. The post office plans to install an additional 600 PiTs in post offices in 2002-2003, with the eventual goal of installing about 5000 PiTs. PiT kiosks are installed with an easy-to-use touch screen interface, speakers, microphones and web-cam. It is extremely user-friendly, with universally recognized icons to overcome the constraint of illiteracy. Post office staff have undergone extensive training to assist customers on the usage of PiTs. 172. Electonic banking services offered at the PiTs will provide medium, small, and micro enterprises greater access to financial services. Encryption technologies - such as biometric thumbprint and retina recognition - will be employed to ensure banking security. Within the next two years, the post office also plans to offer e-finance services to small and medium sized enterprises (SMEs) in the underserved parts of South Africa. The PiT will offer SMEs access to electronic financing in their own local communities by providing an infrastructure for firms to apply for loans from all 26 competing banks in South Africa. Annamalai, Claessens, Furst, Glaessner, Klapper and Klingebiel, `Excerpts from South Africa - Electronic Finance and the Delivery of Financial Services to Lower Income Groups,' The World Bank, Washington D.C. Brazil: Access to Financial Services ANNEX 2.2 BNDES MFI PORTFOLIO ANALYSIS 1 Correlation Analyses 173. Attempts to separate the sample between institutions which received repeated lines of support from BNDES and those which are still first time clients do not suggest a significant difference inbehavior between these groups. 174. There is however some positive correlation between percentages in arrears and numbers of clients, as well as numbers of active credits, suggesting that as some institutions grow, control over portfolio quality may suffer. Correlation between arrears and numbers of credits: 0.37; correlation between arrears and active clients: 0.33, and correlation between arrears and loan size: -0.21. There is also some negative correlation with loan size, suggesting that those microcredit institutions which concentrate on somewhat larger borrowers have a better portfolio quality. One of the best performers, with loans in arrears from 1%-2.4%, had loans around three times the average for all MFIs in this sample. 2. PortfolioGrowth Estimation 175. Over time, the total active portfolio of microcredit institutions that received funds from BNDES has grown from R$16.4m in 1998 to R$31.7min 2000 and R$33.9m inJune 2001. But also the number of institutions that received funding from BNDES has grown from 16 in 1998 to 31 in 2001. Average portfolio size, as discussed above, rose erratically over this period, from R$0.8m to R$l.lm per institution per year. 176. Attempts to find out whether the growth behavior of BNDES client institutions has been dynamic over time was analyzed by fitting a statistical growth trend to this array, as follows: 1ert +b(AXit) Pit =p,10 ] where Pi, i s the Active Portfolio for Institution ``2' inPeriod"t" and X,, is a vector of other control variables as described below, to include: s The institution's default rate s Thenumber of credits oflered by the institution 3 The average loan size oflered by the institution 3 The number of active clients catered to by the institution The linear transformation of the model resulted ina GLS regression lnapch = BAtime) +Bdcrgrch) +Biavlnch) +B+( wdfch) where the variables were defined as: lnapch ln(Pit-Pio),the dependent variable time number of periods over which the panel of observations was based crgrch change in the number of credits being granted duringthis period (current to base) avlnch change in the average loan size of the institution duringthis period (current to base) wdfch change in the weighted default rate of the institution during this period (current to base) - the weights were the contributing proportion of each institution in each period to the overall Active Portfolio Part 2. Expanding Microfinance 177. The results of the pooled cross-sectional time-series GLS regression reveal a statistically significant coefficient of time of (0.116) for the half-yearly continuous compounding rate. Hence the annual growth rate will be around 0.22 {or 22%) and is statistically significant. Change in average loan size and change in number of credits granted both have positive and significant coefficients. The effects of growth of default rate, number of credits and average loan size, and the number of active clients on the dependent variable (the growth of active portfolio) are controlled for at the level of individual institutions. The results denote an average annual nominal rate of growth of around 22 percent, for any given institution, suggesting that even apart from the growth in the number of institutions, individual institutions do also show some tendency to grow over time"'. Discounting for inflation, real rates of growth would be in the range of 16-18 percent, (based on a coefficient of +0.116 for every six months, which is statistically significant after controlling for other influences captured through Xi(). Average annual consumer inflation rates over the period 1996 to September 2001 ranged between 9.5 percent and 1.7percent; and were largely in the range of 5-7 percent over most of the period inquestion. 178. The other statistically significant coefficients are those for a change in number of credits granted (coefficient = 0.0001) and for change in average loan size (coefficient = 0.0003) and i s also statistically significant. The variable Weighted Default Rate within Xi, i s however not significant. 3. Default Rate Trend Measures 179. Average rates of loans in arrears over thirty days for all BNDES clients, according to BNDES' data, has fluctuated over time, between around 3.5% and 8.4%. In June 2000, loans in arrears averaged 6.12% for all 31 clients. Again, variations between institutions are large and eliminating the two worst performers reduced this dramatically to 4%. Variations over time, even within the same institution, are sharp; at some points in time bad loans have risen to a quarter or third of total portfolio, only to apparently largely recover six months to a year later. To test for any improvements in performance over time, the model below investigates if there is any trend over time in default rates, through the specification: WDF,, is the Weighted Default Rate for Institution "i"inperiod "t" and Xi, is a vector of other control variables that include: 3 The number of credits offered by the institution 3 The average loan size offered by the institution lnwdfch =BAtime) +Bd crgrch) +BJ avlnch) The linear transformation of the model resulted ina GLS regression where the variables were defined as: lnwdfch In(WDFit-WDFia),the dependent variable time number of periods over which the panelof observations was based crgrch change in the number of credits being granted duringthis period (current to base) avlnch change in the average loan size of the institution during this period (current to base) "'In terms of the active portfolio levels. Brazil: Access io Financial Services 180. The results suggest a coefficient of time of (0.059) as the half-yearly continuous compounding rate. Hence the annual (positive) growth rate will be around 0.12 (or 12%)and is statistically signi$cant. Change in average loan size and change in number of credits have non-significant coefficients. Including the "Change in Active Portfolio" as an explanatory variable did not add any explanatory power to the model. Part 3. Downscalina Private Banks PART3. DOWNSCALINGPRIVATEBANKS 1. INTRODUCTIONAND SUMMARY 1. Brazil's 175 private banks represent overwhelmingly the majority of its 191 commercial banking institutions in numbers, and also today in the provision of broad aggregates of financial services. Their role in deposit and lendingactivity has grown to a dominant position over the last decade, from 42 percent of total deposits in 1994 to 52 percent in 2001, and from 44 percent of total loans to 56 percent over the same period. Four of these private banks are substantially larger than the others: Bradesco with 9 percent of total banking assets in June 2001, Itaii with 7 percent, Santander with 6 percent (via their acquisition of Banespa) and Unibanco with 5 percent. Nevertheless, Brazil's two large public retail banks - Banco do Brasil and Caixa EconBmica Federal still account for close to 40 percent of deposits (20 and 19 percent respectively in December 2001) and three federal banks (including the wholesale bank BNDES) account for around 31 percent of total lending (13, 6 and 11 percent of total bank loans respectively). 2. To what extent do these private banks `downscale' their services; that is, contribute to the provision of financial services for the poorer segments of Brazil's society? This issue i s investigated in the following section in a number of dimensions. First,the role of private banks in the provision of small deposit and small loan services i s examined, and conditions for opening of bank accounts in terms of initial documentation, minimum deposits as well as ongoing fees for services, are evaluated. Next, delivery channels presently used are discussed, from branches to ATM services and correspondent banking. Third, the role of new technologies in the provision of financial services to smaller and less affluent clients, is discussed, assessing the present use of such technologies in Brazil and offering examples from other countries. Finally, policies affecting the banking system as a whole which could differentially impact the willingness to engage with smaller clients are acknowledged, although the detailed discussion of these factors i s left to Part 7 of this study. It i s true that several factors in the macroeconomic and regulatory environment reduce the incentives for outreach to small borrowers, and these are compounded by the shortcomings of the legal framework and infrastructure for the provision of financial services. Nevertheless, there are measures which banks can adopt to increase their outreach to the poor, through partnerships with other institutions, through special services and windows, and through the use of more appropriate technology. Key findings to emerge are: a First,commercialbanksinBrazilalreadyplay asignificant role intheprovisionoffinancial services to small clients. As such, the scopefor further expansion is promising. Hence, small transaction size per se does not appear to be an absolute barrierfor private banks in entering this segment. a Second, there is however a high degree of segmentation in the types of banking services which private banks wish to provide. On the deposit side, they are more likely to enter the possibly more lucrative segments of the market, where remuneration is minimal (sight deposits} or much lower than the prevailing interbank rates (special savings deposits). This appears to be the case despite the directed credit requirements placed on these categories of deposits. Private banks are not keen to serve small Page 117 Brazil: Access to Financial Services clients' term deposit needs, which are remunerated at rates close to the interbank rate, presumably because its advantages compared to safe assets such as government securities declines. Thus, transactions-based needs and needs for safe stores of value are met, though needs for returns and investments are served less. There is considerable variation between private banks' individual business plans in terms of outreach, whether they are wholly national or partly foreign-owned. 3 Third, the important corollary is that public banks are not the principal providers of services to small account holders; however they do provide services in this segment to those account holders who are more interested in investments, and returns through term deposits, compared to those whose needs are based on transactions or safe stores of value. And low income clients do not show a strong preference for public banks. 3 Fourth, looking at the costs and prices of banking transactions, althoughfees for a basic package of banking services do not appear high, they could nevertheless constitute a significant proportion (one to twopercent per month) of the income of a less well off individual. In terms of institutional variation of transactions costs, although average fees for private and especially foreign banks may be higher than federal banks, the difference is not significant, looking at comparable public and private banks. Some large private banks are as competitive in terms of prices and service range asfederal banks, and in other instances, somefederal bank servicefees may be high. Transactionfees appear to be a greater obstacle for personal banking services thanfor the provision of credit to enterprises; high interest rates are the most importantfactor for enterprises, especially small enterprises. a Fifh, some specific banking services appear to be more uniformly available across different banks (such as ATM use and debit card facilities) while others are less likely to appear in basic packages of services without extra charges, and may be more competitively priced across institutions (checking facilities, overdraftfacilities, provision of printed statements, and debit orders). 3 Sixth, the choice of delivery channelsappears to be affected by regulatory requirements which raise the costs of establishing banking services, especially bank branches. The growth of correspondent banking has been partly in response to these, but correspondents also help to spread fixed costs. In future, competition in correspondentfranchises andfurther broad basing of their services (e.g. including government bond distribution) could help to ensure their competitive edge is maintained. But also, prices of other types of services such as bank card networks may also be affected by limited competition in their provision. Differentiated solutions are required to address these issues. Seventh, new technologiesbased on phone connectivity (telephone and internet based) are widely used in Brazil but notfor services to the lower income segments. Teleconnectivity in Brazil is relatively low in average terms, although there are many well-to-do who use internet banking. Many countries including Brazil are making a conscious effort to encourage more universal access to and use of telecommunication services, through nondistortive targeted subsidies, and several examples of good practice are available. a Eighth, there are a number of good practices which have been identified by banks in other countries topromote the downscaling of their business, which could be considered by banks in Brazil. A first group of such practices depends on the development of appropriate products: a second approach rests upon the development of partnerships, with microfinance institutions, with community organizations, etc., and a third rests upon image differentiation. 3 Nevertheless,financial institutions willfind success dificult in the absence of both staff commitment at the seniormost levels, as well as the appropriate staff trainingfor these segments. Page 118 Pari 3. Downscalina Private Banks 3 Finally, financial training of the target population would be greatly beneficial and can be delivered in combination with outreach programs. 3. The analysis of this chapter was undertaken largely prior to the introduction of new measures affecting the role of the banking system in access to financial services. These measures would not affect the analysis o f the situation prevailing in Brazil, although there are implications for policy going forwards. The new measures are summarized below and references to these policies are included in subsequent text. New Basic Bank Accounts CMNResolution 3104of June 25,2003 and CMNResolution 3113 of Ju to certain transaction limits. No charges are made unle month; gets more than 4 statements a monthor makes Conditions for opening abank accountare account will however still requirea taxpay other basic personalinformati offering such accounts will in Low Cost Microcredit Loans Law No. 10735 of September 11,2003 ( Resolutions 3109 and 3128of July 24,20 expandthe opportunities for microfinance loansdestinedfor low income earners and small businesses.Interestrates on such loans will be capped at 2 percentper month. Loans will be limitedto R$600 for individuals and$1,000 for Feesfor accessing credit are not to exceed2 percentof the loan amount for in business. Credit will be offered to individuals with simplifiedaccounts (with max income individuals, and enterprises(with maximum Obligatory lending and Bank Reserve Requirements According to the same regulations,such microfinance loans will be funde depositsof banks.Microlending may be u are not made, bankshaveto increasetheir Payroll Loans Automatic Dedu - automatic paymentof lo limited to thirty percentof Expanded Correspondent CMNResolution 3110of July 31,2003 expands th many banking services such as opening accounts (sight deposits, savings deposits, term deposits and investment funds), paymentandcredit facilities. Central Bank authorization will be required for entering into suchcontracts. Page 119 Brazii; Access to Financial Services 2. DEPOSITSERVICES-SMALL CLIENTSAND SPECIALSAVINGS 2.1 Deposit Structure -An Overview of Aggregate Deposits 4. For most small clients, especially for individuals or households who are not engaged in entrepreneurial activities, the most important financial service is the deposit of liquid assets and cash wealth. The extent to which private financial institutions are engaged inthe provision of deposit services i s thus the first issue investigated here. Looking at the composition of aggregate deposits, for all banks in Brazil, a first finding is that in terms of deposit type, the overwhelming majority of clients in Brazil's banking system (totaling almost 97 percent of all clients) hold their money in the form of either (unremunerated) sight deposits, or special savings deposits (caderneta de poupaqa). Special savings deposits are remunerated at rates which are below the prevailing overnight rate and sometimes negative in real terms (Figure 3.1 below).' Only a relatively small number of clients (3.3 million persons or 3.1 percent of all depositors) hold term deposits, which however amount to a huge R$128.9 million (41.2 percent) of all deposits by value. Thus the role of banks in the provision of services to depositors in the form of sight or special savings deposits accounts is more important, in terms of client access, compared to aggregatedeposits,or afortiori, term deposits (Figure 3.2). Figure 3.1 Brazil InterestRatesfor Special Savings Depositsand Term Deposits (1999-2002) - - Selic _ _ _ _Special . _ . Savings Deposits ---- T e r m Deposits Source:World Bank staff calculationsbasedon data from Central Bank of Brazil, 5. The second observation is that private banks are already very active in terms of provision of services to large numbers of clients, in the form of sight deposits in particular, and also special savings deposits. However, with regard to the more highly remunerated term deposits, private banks deal with a relatively small number of large depositors. There i s a striking difference in the composition of term deposit services between public and private institutions; 75 percent of term deposits, interms of value, are held by private institutions, but these account for only 25 percent of term deposits clients -andvice versa ' Note that banks are obliged to direct 25 percentof the total of their unremuneratedsight deposits towards the rural sector (in addition to reserverequirements, applicable also to other deposits). Rules regarding specialsavings depositsare complex.Remunerationsare indexed to the TR (Taxa Referential) plus 6 percent, and are tax exempted.However at least 65 percentof these deposits must be directed towards housing finance, four fifths of which must be loanedat controlled terms of credit, (TR+12 percentto individuals andTR+13 percent to builders). Inpractice, such deposits can still be very profitable. Page 120 Pari 3. Downscaling Private Banks for public banks. By contrast, for sight deposits, private banks serve a larger number of clients than public banks (59.5 percent), and hold over half the total value of deposits (54.5 percent). For special savings deposits, these ratios are lower, but private banks still hold a larger proportion of total clients (44.7 percent) than total deposit value (42.6 percent). The inference i s that private banks do not hesitate to provide services to large numbers of small account holders, provided the deposits are lucrative (as in the case of the unremunerated sight deposits). For deposits where reimbursements are much higher, such as term deposits, private banks prefer to focus on larger clients (Figure 3.2). Figure3.2 Brazil Structureof Aggregate Deposits December2001 - - ai Rbllc - Value Compositionof Deposits -Clients (%) Riate 350 7 . Total Deposits 0Value (R$b) and Clients (m) 80 ,% 94.0 - 250 -O: =I i 150 ' 3 100 50 0 Term Deposds Slght Deposits SpecialSavings All Institutions All Rblic All Riate Deposits Compositionof Deposits Compositionof Deposits -Value (R$b)and Clients (m) -Value (??) 350 - 94.0 80 300 -- 70 -P.-z250 -- 60 200 -- 50 H 5 150 -- 40 30 20 10 0 All Deposits Term DepositsSqht Deposits Special 0 Savings Term Deposits Sight Deposits SpecialSavings Deposits Deposits )me:World Bank calculations based on data from the Central Bank of Brazil- Fundo de Garantidor de Creditos. 6. Third, the average size of client accounts inpublic banks is around the same as incomparable private banks (Table 3.1). Public banks taken together have a larger share of total clients (5 1.4 percent) than total deposits (43.1 percent); a difference of around 8 percent, and hence, somewhat smaller average deposits per client (R$2,800, compared to R$3,900 for all private banks). But the larger average client size of private banks i s considerably influenced by some foreign controlled institutions which appear to focus on high net worth clients. The apparent public bank advantage with small size clients is reversed if we compare similar large retail banks, public or private. Thus the average deposit size inBanco do Brasil, at R$2,400, is not very different from the average size in the private banks Bradesco and Itali (R$2,500), and average deposits in the Caixa in fact exceed this, at R$3,200. Looking at sight deposits, the average size in the largest private banks i s below that of the public banks (R$1,100 and R$900 respectively in contrast to R$1,300 in Banco do Brasil and R$1,700 in the Caixa. Bradesco retains the lowest size Page 121 Brazil: Access fo FinancialServices ranking with regard to special savings deposits as well (R$1,800, compared to R$2,700 and R$2,400 at Banco do Brasil and the Caixa respectively). Table 3.1 Brazil Structure of Depositsby Type and Institution Average DepositI Client (R$ 000) (December - - 2001\ All Institutions 3.3 1.2 2.3 Foreign Participation Foreign Citib Note:Caixa has a negligible role in Term Deposits Source:World Bank calculations based on data from the Central Bank of Brazil 2.2 Deposits Below R$S,OOO 7. The findings above suggest that private banks play an active role in some segments of the deposit market which serve the largest numbers of clients, and also that their intake of deposits in these segments is similar to that of the public banks. Does this remain true also for smaller sized deposits? In the second stage of our analysis of deposit behavior for different institutional categories, we look specifically at deposits below R$5,000. The broad message to emerge is that the signijicance of private banks for deposits below R$S,OOO remains largely unchanged compared to their role in total deposits (Figure 3.2 and Appendix Table A3.2). In terms of numbers ofclients, the share of private banks in this segment remains at 49 percent; in terms of value there is a small decline (57 percent to 53 percent), suggesting, in combination with the stable share of numbers of clients, a somewhat greater concentration on smaller depositors. Private banks' share in clients, for sight deposits below R$5,000 is close to its share in total sight deposits (around 60 percent), (with a marginal decline invalue) and the share of clients in special savings deposits i s also stable (at 45 percent) for deposits of all sizes compared to those below R$5,000 (with a small increase in value from 43 percent to 47 percent) (Appendix Table A3.2 and Figure 3.3). Data on individual institutions also reveals largely stable behavior, for sight and savings deposits, for the larger public and private banks. 8. Finally, we examine the behavior of banks with regard to deposits in the smallest size categories. Included in the four size categories defined below are deposits below R$100 and three additional size categories below R$5,000. These data are only available for sight deposits and special savings accounts. Page 122 Pafl3. Downscaling Private Banks Figure3.3 Brazil Structure of Depositsbelow R$5,000 by Type, Institution, Nos. of Clients and Value - (December 2001) Ail Deposlts Clients - All Deposits - Value 0Rivate Special Savings Special Savings Deposlts Deposlts Sight Deposits Sight Deposlts TermDeposits 5 TermDeposlts O h % 0 20 40 60 80 Deposits below ~$5,000-clients Deposits below FS5,OOO Value - Special Savings SpecialSavings Deposits 55 2 Deposlts Sight Deposes 52.q Sight Deposits TermDeposns F , , TermDeoosits I I I , , 2 0 10 20 30 40 50 60 70 80 0 20 40 60 EO wce:World Bank calculations basedon data from the Central Bank of Brazil - Fundode Garantidor de Creditos. 9. These data reveal surprisingly that for unremunerated sight deposits, private institutions dominate the provision of services in the smallest size categories in terms of numbers of clients (63 percent), and also value of deposits (87 percent). However, in terms of (remunerated) special savings accounts, the private share declines, to 43 percent of clients and 46 percent of value, in the smallest size categories. As deposit size categories increase, ratios of public and private provision of services stabilizes, converging to the totals for all deposits below R$5,000 (Figure 3.4 and Appendix Table A3.3). There are also strong differences between institutions, revealing that these are market segments which certain institutions choose to specialize in . Among private players, Unibanco accounts for 70 percent by value of the deposits below R$100, although only 21 percent of clients. Itaa and Bradesco take smaller deposits, as both have client shares of around 17 percent and value shares in the rante of 5-6 percent. For special savings deposits, the dominance of Caixa is clear; 30 percent of clients and value. However, the roles of the private bank Bradesco and the public Banco do B r a d are similar with 18 and 17 percent respectively of clients; though Bradesco has a somewhat higher share of savings accounts by value; 21 percent versus 13 percent for Banco do Brasil (suggesting somewhat larger account sizes within this size category). 10. In sum, private banks are very significant players today in the provision of deposit services to even the smallest clients, and they are aggressive in terms of presence in the more lucrative segments of these markets, with lower remuneration, notably sight deposits. Some private banks (Bradesco and Unibanco) have a major presence in special savings deposits for the smallest segments although Caixa remains the dominant player. Provided returns are sufficient, therefore, barriers in terms of establishing and maintaining small accounts do not appear to be an obstacle. Page 123 Brazil: Access to Financial Services Fi, re 3.4 Brazil - Structure of Sight and SpecialSavings Depositsfrom R$1,000to R$5,000 (December 001) Sight Deposits below R$lOO to R$5,000 Clients - 00RNate Sight Deposits 0hblic below R$100 to R$5,000 Value - ~1 100% FromI31,OOl to 90 Rs5,OOO 80 70 FromI3501to 60 F31,ooo 50 40 FromI3101to 30 w o o 20 10 0 Below I3100 Below F$100 FromR$101to FromW01 to FromW1,OOl % I3500 ~ 1 , 0 0 0 to R$5,000 0 10 20 30 40 50 60 70 80 90 100 SpecialSavings Deposits SpecialSavings Deposits Rrvate below R$100to R$5,000 Clients - below R$lOO to R$5,000 Value - % FromF31,OOl to R%5,000 FromF3501to I31,ooo FromI3101to F3500 Below %loo Below I3100 FromRSlOlto FromRf601to FromF31,OOl % I3500 I31,ooo to Rf6,ooo 0 10 20 30 40 50 60 so e:World Bankcalculationsbasedon datafromthe Central Bankof Brazil-Fundode Garantidorde Creditos. 11. As a caveat however, small deposit size does not necessarily imply low income clients, although a correlation would be expected.To the limited extent that data are available on client income, it suggests that the proportion of low income clients would be smaller than the proportions of small depositors. One large bank for instance has 5 percent of its deposits below R$10,000, but only about 1 and 2 percent respectively with individuals in the lowest two income classes E and D, compared to 7 percent with middle income individuals inclass C and 11percent with middle market companies. 12. Only limited data are available in the present study on the income levels of small depositors, based on information provided by banks. These data suggest that the very poor have a higher deposit to loan ratio than richer individuals. The average deposit balance held at banks by their Class E customers oscillate between 4 and 6 times the amount of outstanding loans they have. This compares to an average ratio of 2 to 3 for all other customer classes. Most of this balance i s in special savings accounts. In some banks, the average deposit of a Class E individual can be as high as about R$2,000 (US$900), a comparable level to the average for Class B individuals. Class C and D tend to have lower deposit balances, varying between R$400 and R$1,000 on average.2 2.3 The Client's Perspective Demand for Bank Accounts and DepositServices - 13. The extent to which services are actually provided i s not a measure of the difference between desired and actual levels of services. In order to gauge the gap between supply and demand of banking Banks in Brazil divide their clients into five notional income groups; Classes A (the wealthiest) to E (the least well-off). Limits for each class at the beginning o f 2002 were approximately: A (above R$ 3,000); B (above Rs 1,000); C (above R$500); D (above R$250); E(below this). Page 124 Part 3. Downscaling Private Banks services in general and deposit services in particular, and also to assess the extent to which the services actually provided conform to clients' needs, the information above has been supplemented by the evidence from the survey of access to financial services inurban areas, discussed in Part 1of this study. 883 65.9 113 18.3 52 12.9 services Transactionswl lower costs Attractive rates of return Others Valid Cases Curr Security 215 70.5 27 18.8 16 15.5 Other financial services 30 80.3 35 7 Transactionswl low 51 3 Attractive rates of return 3 87.5 6 82.6 13 70.9 Others 38 100.0 25 100.0 30 100.0 Transactionsw/ lowercosts 24 79.9 39 43.9 11 35.2 Transactionswl lowercosts 21 44.4 n 29 73.7 Others Valid Cases (347) (133) (99) Source:World Bank, Survey of Access to Financial Services in UrbanAreas 14. Results confirm that a gap between demand and supply does exist.3Although a large proportion of the 2,000 persons surveyed in urban areas of Brazil already had some form of bank account (43 percent or around 850 persons), of the approximately 1,150 persons who did not have an account, two thirds stated that they would like to have an account (Table 3.2). It also appears that demand is primarily for `store of value' and transactions purposes, and not for returns, as the primary reason cited for wanting a bank account was `security'. Additional reasons suggest a dominance of needs for `other financial services' (presumably services such as transactions or payments), particularly in the cases of current or special savings accounts. Rates of return were not considered an important reason, except for those with term deposits. In the case of current accounts and special savings accounts, lower transactions costs were also cited. These responses would suggest that the dominance of current and special savings accounts in small deposits reflects bothclient needs and the banking sector's interests. Note that this does not establish effective demand at prevailing prices. Table 3.8 below suggests that a large number of persons believe that their lack o f "money" (which could be either income or wealth) is an obstacle to access to bank services, and also, prices of services are deemed "high". Page 125 Brazil: Access fo Financial Services 2.4 New Possibilitiesfor Small Savings Other than Deposits Retail Bonds - 15. To the extent that savers or account holders desire higher returns, new channels for small savings instruments through the direct purchase of retail government bonds are worthy of a mention. Sales of such bonds can be made through the internet, under the Treasury Direct program. Closely modeled upon the program of the same name in the USA, and widely used in a number of OECD countries, such retail debt sales programs have the double advantage of allowing a safe savings vehicle for small savers, remunerated at market rates, while at the same time, allowing the government more efficient mechanisms for tapping local debt markets4These schemes permit small investors with as little as R$200, or less than US$lOO, to invest in a low risk, high return investment with relatively high liquidity. B y investing via the internet, costs involved are lower. The maximum amount allowed is R$200,000 per month. Prior to this, only banks, bond dealers and other financial institutions registered in the Sistema Especial de LiquidacrZo e Custddia, (SELIC) with a minimum amount of R$50,000 could purchase government bonds. November 9.9 8.7 5.3 Decembe Source:National Treasury - Brazil 16. In its first two months of operation (January 2002 to March 2002) more than 31 percent of investors involved invested less than R$l,OOO, and almost two thirds of all investors invested sums of below R$5,000. More recent data offer the average transaction size rather than the detailedbreakdown by size category. These suggest that average transaction size has increased progressively over time (Table 3.3). Thus, although in principle, very small deposits can be accepted, it appears that in practice, average deposit size is several multiples of the minimumsize. This suggests that efforts could be made to raise the publicity of the program to the attention of the smallest savers or increase the involvement of distributing financial institutions to use branch outlets or bank correspondent outlets. 3. CREDIT SERVICES PROVISIONOFSMALL LOANS - 17. Banks in Brazil offer a wide variety of loan products to businesses, including, in principle, the full range of products which could be of interest to lower income groups. The issuesare to what extent are bank services targeted towards these groups, under what conditions and at what price? The following 4The BrazilianTreasury (Tesouro Nacional), http://www.tesouro.fazenda.gov.br/tesouro-nacional, on December the internet and on offers beganonJanuary 7tR. 12Ih,2001 issuedOrdinance no554, which ap roved the regulation of the public offering of bondsto individuals via Page 126 Parl3. Downscalina Private Banks sections investigate the extent to which bank credit i s made available to small borrowers and lower income groups. 3.1 The Provision of Small Loans 18. The choice of products offered to each segment also depends on the probable value addedeach customer segment would generate with each product. Evidence suggests that consumer loans are the most profitable products among those of interest to low income segments. By contrast, banks find that carrying out payment transactions with lower segments is not profitable because the high administrative costs compared to the margin. However, the cost of carrying transaction services could be lowered by the use of new delivery channels, as discussed later in this section. 19. Data on small loans cover loan value and loan size, but not the number of clients served by an institution or market segment. In terms of the aggregate value of funds, it is not surprisingthat Brazil's largest commercial banks, both public and private are the main providers of small size loans, followed by non bank finance institutions. InNovember 2001, banks provided a total of R$39billion (US$17 billion) of loans below R$5,000 (US$2,150), with an average size as low as R$400for some banks (US$180) at then prevailing exchange rates. Non-bank finance institutions provided another R$5 billion (US$2 billion), while microfinance institutions provided only R$O.1 billion (US$43 million) of loans, with an average size of R$1,087. 20. One public bank, Banco do Brasil, provides 14 percent of all loans below R$5,000. The bank estimates that the average size of its consumer loans was R$1,100 in 2001 (less than US$500), placing it indirect competition with microfinance institutitons. Banco do Brad is followed by Brazil's other large banks: Bradesco, Caixa Econbmica Federal, Itad and ABN AMRO, each with 6 to 8 percent in each market.5 Unlike deposits, foreign banks play a significant role in the small loan market, with a total market share of 26 percent in the market for loans below R$5,000 and 36 percent for loans below R$10,000(Table 3.4). 21. Since Caixa's restructuring in 2000, private banks dominate the small loan market in terms of volume of loans. The share of public banks in the small loan market was substantially higher in 2000 before Caixa's balance sheet was restructured and a large part of its loan portfolio sold or written off. As a result, Caixa's share of the loan market segment below R$25,000 (US$10,700) dropped from a dominant 38 percent to 11 percent. Caixa i s now in second place after Banco do Brasil which accounts for 15 percent of loans in this segment. The small loan market became thus more competitive, and private commercial banks became the largest players with 63 percent of all loans in this segment while public banks dropped from 61to 37 percent. 22. However, follow up discussions with the banks in question, and an analysis of their credit data suggest that a large part of small size loans i s not provided to lower segments. Detailed data available for some banks suggest that the majority of their small loans are in fact borrowed by middle income individuals and middle market businesses. In the case of one large bank for instance, 11 percent of its loans are below R$5,000. However, only 1percent of its outstanding loan portfolio i s to individuals in class E, and 2 percent in class D, while 10 percent goes to middle class individuals in class C and 9 percent to middle market enterprises. Data on the number of clients served for different types of loans and within lower size groups would be desirable to supplement this conclusion. Note that Unibanco appears to have a smaller share in the small loan market than its natural market share would suggest, but this is in part due to its concentrating small loans in its subsidiary Fininvest. Page 127 Brazil: Access to Financial Services Table 3.4 Share of Brazil'sCommercialBanks inthe Small Loan Market in terms of Loan Value (R$000, Private Brazilianbanks 40 31 24 26 31 Public banks 34 33 46 52 40 Unibanco Private mixed 4 4 3 2 5 HSBC Brasil Totalten largest players 58 54 57 70 34 Source: Central Bank of Brazil, Centra/ de RISCO Table 3.5 Recent Evolutionof Bank Shares in the Small Loan Market (% shares, December 2000 - November 2001\ December 2000 % - November2001 -Yo Private Brazilianbanks 22 20 33 30 Public banks 49 Foreign banks 17 21 29 21 Bradesco ltalj ABN AMRO Foreign 3 3 7 2 Santander (Banespa) Source: Central Bank of Brazil. Centralde Risco 23. For most large Brazilian banks the small loan segment is important. For the ten Brazilian banks with the largest volume of loans below R$5,000 (see Table 3.4 and Table 3.5 above), these loans represented 24 percent on average of their total loan portfolio. By comparison, all other loan size categories represented less than 10 percent each of their total loan portfolio. Clients with loans below 6With foreigncontrol. Page I28 Part 3. Downscalina Private Banks R$5,000 accounted for a very significant part of their clientele, and those with loans below R$10,000 for another very large tranche. There are also some smaller banks with a focus on low income segments, such as Banco BBM, which provides working capital financing to small businesses, backed by their receivables. 24. Some banks channel their loans to lower segments through specialtyfinance subsidiaries. For several private banks, activities with low income clients are concentrated in dedicated subsidiaries with a focus on consumer lending. This enables the banks to maintain their image vis-&vis their middle and high income clients, and to present themselves with a friendlier face to lower income clients. For instance, while Unibanco S.A. has only 1 percent of its loan portfolio with individuals in class E and a similar percentage in class D, Fininvest, its consumer finance subsidiary, has 11 and 28 percent respectively in those categories. In terms of their client base, individuals in classes D and E constitute only 10percent of Unibanco S.A.'s clients but 54 percent of the clients of Fininvest. Table 3.6 The Importanceof Small Loansfor the 10 Largest BrazilianBanks (Value and O share, November h 2001\ Bancodo Brasrl 6,095 13 3,368 7 1209 4 Itair 3 260 16 1164 Unibanco 1864 11 797 5 504 LloydsTSB 1,337 73 neg. 0 H NossaCaixa 997 35 298 10 TotalIaverage 26,179 24 11,654 7 Source.Central Bank of Brazil, Central de R m o 25. Many specialtyfinance companies dedicate their activities to lower segments. Many large non bank finance institutions in Brazil, such as Fininvest and Panamericano, are consumer finance specialists, which dedicate their activities to low income segments. These institutions play a much more limited role in terms of volume than large banks, but are significant players in the small loan market compared to microfinance institutions. 3.2 Small Size Transactions versus Transactionswith Low Income Segments 26. The above analysis suggests obstacles to lending to lower segments or to collecting their deposits, do not necessarily lie in the small size of their transactions per se. In the case of deposits, it was demonstrated that initial and ongoing transactions costs associated with small deposit intake and monitoring are less significant than interest rate differentials on different types of deposits. In the case of loans, profitability depends not only on initial and ongoing transaction processing costs but also on the default risk associated with each class of borrower, the cost of assessing such default risk and the potential to set lending rates which adequately reflect these costs and risks.Key elements of initial (entry) Page 129 Brazii: Access to Financial Services and ongoing (per transaction) costs and prices are investigated in Section 4 below. Since these costs vary significantly by the delivery channel chosen, Section 5 examines alternative delivery channels. Table 3.7 Non Banks with a Significant MarketShare in the Small Loan Market (R$'000 and percent, November mni\ Fininvest Consumerfinance 1,274 94 Source:Central Bank of Brazil, Centralde Risco 4. ENTRY REOUIREMENTS.PRICESAND TRANSACTION COSTS 4.1 Entry Requirements -the Client's Perspective 27. If banks are prepared to offer services to small scale clients, what are the reasons which hold back such clients from opening bank accounts? A variety of factors could be possible. From the point of view of the individual, requirements for opening an account may have been too stringent. Until recent changes introduced in 2003, opening an account required proofs of identity, income, clean credit records, references, and in the case of prospective borrowers, collateral. Additionally, the prices and conditions at which services were offered by banks could have been unattractive for less well-to-do clients. These could include for example high initial deposits or high minimumbalances, high ongoing maintenance fees, or high fees per transaction. Unattractive conditions could include inconvenient locations or hours of service, or even behavior towards unaccustomed clients which could be overwhelming or ~nwelcoming.~ Finally, it i s also possible that there has been limited interest in having a bank account among some persons who are very poor, for whom the need to store value i s not important.8 A partial investigation of the importance of these issues was investigated in the World Bank 2002 survey of access to financial services in urban areas of Brazil. Second, information was also obtained from individual banks on entry requirements for bank services and the pricing of such services, and third, Central Bank data on the pricing of services was evaluated. 'In mid 2003, new regulations were introduced (CMN Resolution 3104 o f June 25,2003 and C M N Resolution 3113 of July 31, 2003) which permit the opening o f basic bank accounts for low income clients, with eased requirements. Such accounts, which offer limited services, are now available free o f charge. Further details are on page 119. Inthe World Bank survey, of the around 1,150 persons who didnot have bank accounts, only around 700 persons or 64 percent wanted a bank account. Ina parallel survey by Unibanco of Brazil, a very similar 60 percent of the unbanked wanted an account. The Unibanco survey also ranked the reasons why persons did not have accounts, and financial reasons (insufficient income, insufficient proof of income or insufficient money for opening or maintaining balances), together with costs, were the first four reasons cited, accounting for 80 percent o f responses. Next came bureaucracy and documentation, and finally, adverse credit reports. Page 130 Pari 3, DownscalinaPrivate Banks 28. First of all, results from the World Bank survey suggest that both sets of reasons were valid; that is, reasons pertaining to entry requirements in the formof perceived levels o f minimumfunds needed to use banking services, as well as documentation and references, and in addition, the costs and fees o f service provision (Table 3.8). Official regulations pertaining to requirements for opening an account, until recently, have been onerous and diffuse, and added to these were further requirements which most banks required in practice.' Official requirements include an identity card, proof o f address, and a taxpayerhndividual identification number (CPF).loSuch requirements still prevail for accounts other than the new basic accounts for low income persons. Table 3.8 Brazil Survev on Access to FinancialServices. Reasonsfor Not Havina a BankAccount - Group 1 Client Characteristics - Don't have enough money 817 93.3 45 21.3 31 25.4 Don'thave the right documents 22 95.8 83 60.7 12 35.2 Don't know how to open an account 13 ferences 12 tory 8 Hostile bank environment 4 100.0 14 100.0 4 100.0 High fees 125 42.2 40 31.5 9 11.0 Transactions take too long Badservice 23 96.0 24 87.0 24 84.0 B nt Source: World Bank Survey on Access to Financial Services in Urban Areas of Brazil (2002) 29. On standard accounts, the branch manager's signature is required to attest to the validity of the documents, and not only does this make a branch visit necessary but the implied personal liability o f the manager in case o f misinformation has made managers reluctant to vouch for individuals perceived to be risky. According to standard requirements, financial institutions could open accounts for persons with Official regulations on minimum information and other data required to open any type of deposit account can be found in Resolution 2025 (article l), of the National Monetary Council, as amendedespecially by Resolution 2747 (article 1) and also by 2078,2303,2817, as well as Circulars 2452,2520,2556, 2677 and 3006. Requirementsare basically the same for demand depositsand savings deposits, with some additional regulations for term deposits. An individual must furnish name, parents' names, nationality, date and place of birth, gender, civil status, name of wife or husband, profession, and an official document of identification. Legal entities must presentdocumentsinrelation to the entity (statutes, principal activity, form and dateof constitution), as well as to the individuals authorizedto operate the accounts, including their residential and commercial addresses, phone numbers, referencesand signature. Financial institutions must maintain copiesof these documents.Banks are also obligated to provide some information to their clients. For demand deposits,information on minimum balancesto maintain the accountcurrent, conditions for furnishing acheckbook, fees chargedfor inactive accounts, etc., mustbe included inthe contract. Checkbooks may not be furnished untilclient data has beenverified. loThe CPF (Cadastrode Pessoa Fisica) i s the tax registration number for Brazilian individuals at the Secretariada ReceitaFederal(the BrazilianInternal RevenueServices). While all citizens are supposed to have aCPF, only 11 millionof the estimated 170million population paid income taxes in 1999. Page 131 Brazil: Access to FinancialSetvices negative information in credit bureaus, but may not furnish checkbooks if they have issued checks drawn with insufficient funds. However, beyond these, many banks have required clean credit records, proof of income and minimum income levels, and minimum initial deposits. These have included the publicly owned federal banks, where these minima have been no lower than comparable private banks (Table 3.9). 30. Brazil's banks have usually applied a credit scoring methodology to all account applicants and defined ex-ante what products the account holder could have access to. Applications for these products could then be sent on-line or via an ATM. The only account which could be opened without credit scoring were the "Caderneta de Poupanga" (passbook savings deposits). Persons unable to provide all the above information were sometimes deemed to be ineligible by these methods. Additionally, for loans to individuals, institutions have required a minimum salary level, or a co-signer or guarantor. Minimum salary levels for loans have been higher for banks than for consumer loan agencies or non-banks. There has therefore certainly been scopefor simplijkation of conditions requiredfor the opening of an account. Recent measures introduced by the government establishing simplified accounts for individuals are in broad terms a step inthe right direction (see page 119: ExpandingAccess to Banks). 31. Non-banks or consumer credit agencies (such as merchant or store credit) have apparently shared most of these requirements, with one key difference; lower importance for the credit record, which they usually do not require or check. However supermarket chains and large stores which issue credit cards in their name do require credit checks as well as the minimum salary, although minimum salary levels can be much lower than at banks, at levels (such as R$150 to 270) which are below the prevailing minimumsalary in2001, of around R$180. 4.2 Pricing of Banking Services for Individuals and Households 32. A first measure of the extent to which the prices or costs of bank services may have been considered high by some client groups can be gauged from Table 3.9, which looks at the costs of provision of basic bank services associated with a sight deposit (current account). Fees detailed in this table are on a per service basis. The data suggest that monthly fees although low in absolute terms may nevertheless have been expensivefor a low income client. Fees which range from R$3.00 to R$5.00 for monthly services, could constitute 1.5 percent to 2.5 percent of an individual earning R$200 per month, which was the cut off level for the first quintile of our sample, and exceeds a monthly income at the minimumwage, of R$180 per month. 33. A second and possibly more realistic measure of bank fees is the cost of provision of a basket of services. Most banks provide their services inthis manner, offering two or three standard combinations of services to clients in broad categories of service needs. Thus, Itali offers three broad packages; an electronic package, a basic package, and an all-inclusive package. ABN-Amro also offers three broad standard baskets and the public banks Banco do Brasil and Caixa each offer a basic as well as a `special' package (Annex Table A3.4 ). Basic baskets of services have similar monthly charges; Caixa and Banco do Brasil, the two public banks, offer service packages of R$3.00 and R$3.50 respectively, while services from private banks begin at the same level (R$3.50 in Banco do Brasil, and somewhat higher at ABN Amro (R$4.00) and Itali (R$6.00). The diference between basic and more expensive service packages appear to lie in services such as the number of withdrawals, the provision of monthly statements, and signijkantly, checking services and overdrap (cheque especial)facilities. Page 132 Part 3. Downscaling Private Banks Table 3.9 Fees and Requirementsfor Sight Deposit Services at Key Brazilian Banks (2002) Caixa Econdmica Bancodo Brasil(ee) (BANERJ). ita" Banco Real ABN AMRO BANK Bradesco Identitycard Yes Yes Yes Yes Yes Notes:Option 1: Specific fees are charged for each transaction.; Option 2: Packages of servicesare offered, for a single monthly fee, for a package of pre-determinedservices (See Annex Table A3.4). Refersto Credit Order Documents (DOC) or Credit transfers which are usedto make transfer of funds between differentaccounts in the same or different banks Source: Bank informationthroughwebsites and site visits 34. Packages offered by some private banks such as Bradesco appear very competitive compared to the public banks, and there i s little to choose between them. However, other private banks in the table above prefer to scale prices of some services based on the volume of transactions. Additionally (and not cited in the table), most banks offer volume discounts on service fees, based on overall deposit size. As deposit size increases, monthly maintenance costs reduce and finally disappear. Although this i s logical from the point of view of scale economies for the bank, it i s regressive from the point of view of small consumers. 35. Per transaction prices too can be expensive,for the poor, especially for some transactions. In order to open an account, registration of client particulars (the client `cadastre') is required, and is regularly updated by some banks. Many institutions charge an initiation fee for this service and some require fees for the ongoing updating of client particulars. The average cost of this service was R$31 in February 2002 (with a range varying from 0 to 270), and the average cost of a credit history check was R$21(with a range from 0 to 300). The average cost of opening a current account was R$14 (with a range from 0 to 30). This means that it is possible to open an account at no cost in some banks, but that the average cost is R$66, nearly 37 percent of the minimumwage. The average cost of a credit assessment is R$105 (with a range from 0 to 800), or nearly 60 percent of the minimumsalary. Charges for checking and overdraft facilities ("cheques especiais") are particularly high, especially in terms of interest rates charged for the latter. Overdraft facilities may be accessed by debit cards, with pre-approved overdraft limits, which i s a key factor in their popularity from the clients' perspective. Another reason on the suppliers side may be that problems associated with the recovery of formal loans do not apply to overdrafts. 36. Interest rates are very high for those categories of lending which are important to small consumers and to small entrepreneurs, especially overdrafts. Loans on overdraft represent the overwhelming proportion of bank credit; 72 percent of new lending to individuals in December 2002 and over 45 percent of new loans to enterprises. Moreover, the percentage of new lending which consists of overdrafts is much higher than the stock of all loans (13 percent of outstanding loans to individuals and Page 133 Brazil: Access to FinancialServices around 29 percent of loans to enterprises) and the interest rate on this category of lending is very high, up to as much as 160 percent on an annualized basis (Table 3.10). 37. Moreover, overdrafts and other expensive categories of loans form a large proportion of new loans. Overdrafts as a part of outstanding loan stock are growing particularly fast for enterprise overdrafts. Vehicle and personal loans are the most rapidly growing categories for personal loans (see Figure 3.5). Table 3.10 Interest Rates,OutstandingStock and New Loansto Consumersby LoanType (Free Credit) (Dec 02) (R$ million and percent) Loans to Individuals Overdrafts Personal Total "Average interest rates weighted by outstanding loans until May 2000 and by daily average of new loanssince May 2000 Source: Central Bank of Brazil 38. Apart from overdrafts, debit card services are cheap to provide and are the most common product used by lower segments.All the 'baskets' of services offered (see Annex Table A3.4) include one debit card for an account holder. Brazilian banks interviewed indicated that up to 60 to 85 percent of their clients in classes D and E have debit cards with pre-approved overdraft limits; against only 20 to 40 percent of clients in classes A and B. By contrast, about 20 percent of clients in classes A and B have a credit card, compared to less than 5 percent of class D and Eclients. Nearly 95 percent of the outstanding loan balance of class E individuals on average comes from an overdraft, rather than a formal loan. For classes C and D, the percentageremains highat around 80 percent. 39. Under new arrangements, a basic basket of services is now offered free of charge to low income consumers. Such baskets offer limited services and do not include check writing facilities. These resemble, in many respects, basic accounts offered in many other countries, and as such are a welcome development. However in Brazil such accounts are offered entirely free of charge, which may make their sustainability difficult, as well as expensive and unattractive for sponsoring financial institutions. A second unusual feature i s the offering of low cost loans to the holders of such accounts (see page 119 for details). Page 134 Part 3. Downscalina Private Banks Figure3.5 Credit Operationswith NonearmarkedFunds (R$ billionJuly 00 to Dec 02) Outstanding Loans (end of period) 30 I 8 16 25 14 20 12 c 10 15 -gB - 8 ~ 10 6 4 5 2 0 4 0 4 , I I I , , , , , , , , , , , -Enterprises Overdraft x IndividualsOverdraft -Enterprises Fixed Assets Individuals Fixed Assets _ _ _ _ Enterprises Working Capital IndividualsVehicles -Enterprises Vendor Individuals Personal Enterprises Discounting of Duplicates Source:World Bank staff based on Central Bank of Brazil data. 4.3 Pricing of BankingServicesfor Small Enterprises 40. In contrast to individuals, banking services sought by small entrepreneurs are primarily credit, and interest rates are perceived as the key constraint to access. A study conducted by the World Business Environment Survey (WBES) on a sample of 201 Brazilian firms of various size showed that the high level of interest rates was perceived as the main operational and growth constraint by companies in Brazil, followed by collateral requirements. These two factors have been more important in Brazil than in other Latin American countries or elsewhere in the world. And for small firms, interest rates are an even stronger constraint than for larger firms (Table 3.1l)." Table 3.11 Constraintsto Enterwise ODerationsand Growth, an InternationalComDarison All firms Brazil-all firms 2.69 3.71 2.91 2.90 1.81 1.71 Latin America 2.86 3.57 2.79 2.91 2.22 2.08 Upper-middleincomecountries 2.69 3.41 2.37 2.54 1.83 1.71 Small firms 2.47 3.79 2 76 2.54 1.73 1.88 Mediu Large firms 2.33 Note Ratingsare as follows: 1= no obstacle, 2 = minor obstacle, 3 = moderate obstacle, 4 = major obstacle Source:Beck, Thorsten (2001) "Accessto Financial Services in Brazil -An Overview 41. By contrast, bank transaction fees appear reasonable to micro and small companies. Most banks charge lower service fees for small enterprises. Bank registration costs average R$52 for I 1New regulationswere introduced in 2003 which offered subsidizedlow interest loansto small entrepreneurs(see page 119). Page 135 Brazil:Access io Financial Services enterprises, account opening fees are R$15, credit checks average R$29 and loan processing fees average R$140. Obtaining a loan backed by receivables is muchmore expensive (about R$600). 42. An additional source of information on the links between bank fees and type of institution, provided by the Central Bank, suggests in contrast to the observations above that on average, federal banks may be significantly cheaper than private and foreign banks (Table 3.12). The difference varies from 20 percent to open a current account to nearly 200 percent for a credit file analysis. But these averages for all banks within each category may be influenced by those private or foreign banks targeted at high net worth individuals. Comparisons of the largest private and public banks, as undertaken above (Table 3.13), reveal outcomes which are more similar. And the data below support the finding that the highest prices charged by some federal banks are in line with average private bank prices. Foreign banks tend to be more expensive than private national banks. 43. Finally, it is worth noting that non-bankfinancial institutions are generally more expensive than banks. On average, it costs R$180 to become a client of an NBFIand obtain a debit card equivalent to one minimum salary. A credit history averages R$319 on average, or almost 80 percent above the minimumsalary. Table 3.12 Average Transaction Costs by Type of Banks in Brazil Data on all Banks - Client cadastre 0 15 20 0 31 270 0 34 250 15 0 0 9 10 0 0 0 2p.m. 0 Overdraft facilities 0 15 30 0 Credit granting 15 0 Notes. pm: per month, p.a.: per annum Source: Central Bank of Brazil,February 2002 Table 3.13 Transaction Costs Comparison, Banks and Non Banks in Brazil Client cadastre 0 32 270 0 45 130 Credit history check 0 21 300 0 Current account opening 0 14 30 0 6 0 26 105 800 0 139 800 Note: p.m: per month;p.a: per annum Source:Central Bank of Brazil, February 2002 5. DELIVERYCHANNELS-BRANCHES, ATMs AND CORRESPONDENTBANKING 44. Bank fees and transaction prices discussed above are affected by the real costs of delivery of services, by the risks associated with the provision of specific types of services to different categories of consumers, and by the contestability of entry of other banks into each market segment. The present section evaluates the role of different modes of delivery of services in terms of their possible contribution Page 136 Pad 3. 5ownsca/inoPrivate Banks to the expansion of outreach. More widely used traditional channels are discussed inthe present section - bank branches, ATM networks, and correspondent banking services. New techniques for delivery of services, based on internet and phone technology, are discussed in the following section. The relative provision of these services is described inTable 3.14 below. Table 3.14 Types of DistributionChannels Utilizedby Different Bank Groups Source:Central Bank of Brazil - "Evo/u@odo Slstema FinanceiroNational'- Annual Report from December 2000 5.1 Brick and Mortar Institutions -Bank Branches 45. Bank branches are the most traditional and still most important channel for the provision of financial services in Brazil. Bill payment is still mostly conducted through branches as the postal system i s perceived as less reliable. As discussed in the first chapter of this study, Brazil's banks serve a much higher number of inhabitants per branch than in developed countries, but service levels are above average compared to other developing countries. Brazil has more bank branches relative to population than other Latin American countries or other large developing countries. Measured relative to GDP, its bank branch intensity is lower than Colombia, India and Indonesia, which are all poorer than Brazil, but the two latter are much more densely populated. Branch intensity relative to GDP inBrazil i s higher than in Mexico or Chile, although both are richer than Brazil. It is true that regional distribution of bank branches differs widely, with lower services in the north and north east, but these are in large measure explained by differences in population density and regional GDP, and thus, by differentials inpersonal income. 46. In any event, the provision of bank branch services is expensive, and expanding branch networks in underserved regions i s not a profitable strategy. Banks agree that the high labor, security and capital costs associated with branch opening make it unprofitable to attend to poor people in remote areas through new branch services. 5.2 BankingCorrespondents 47. The high fixed costs of establishing a bank branch are raised disproportionately by a series of security regulations, labor laws, restrictive hours, etc. In response, from the late 1990s, Brazil's banks began to develop physical outlets for their services which were not deemed to be bank branches. Since March 2000 such `correspondent banking' agreements have been officially recognized, and there has been a dramatic increase in the numbers of financial institutions entering into correspondent arrangements, as well as the numbers of correspondent outlets, in the last two years (Table 3.15)." Correspondents include non-financial legal entities such as convenience shops, supermarket chains, etc, and more recently, post offices; largely for the provision of basic payment, deposit and simple cash withdrawal transactions to a wider public and particularly to low income segments. " Official recognitionwas first givenby ResolutionNo. 2707 of March 30,2000, and extendedby CircularNo. 2978 of April 19,2000, Circular No. 2983 of June 7,2000 andResolutionNo.2953 of April 25,2002. Most recentlythe scope of activities was extended by Resolutionof the NationalMonetaryCouncil, No. 3110of July 31, 2003, whichexpandedthe outreachof correspondent banks, authorizingthem to offer manybankingservices such as opening accounts (sightdeposits, savings deposits, term deposits and investment funds), payment and credit facilities. Page 137 Brazil:Access to FinancialServices Costs of Maintaining Bank Branches in Bra Security Costs Law 7102 of June 1983 of the Ministry of Justicerequiresall financial outlets public andprivatebanks, savings banks, credit societiesand savings associations(defined as branches where movementsof funds occur) to implement acertain number of security measures, includingthe provision of (i) trained armedguards; (ii) alarmenabling an communication with another branch of the institution, a securitycompany or the police; (iii)a camera, or ablinded cabin with apermanentarmed guard, or other equipmentsdelaying the actions of criminals. Transport Costs The law also requires that transportationof funds should be undertakenby professionalso from the financial institutionapprovedby the Ministry of Justice. For sums aboveR$7, used but with two armed guards. Beyond R$20,000, a special vehicle is required. Labor Costs and Restrictions The Bankers Union Agreement imposesa maximumof six hours of work employees (but this also tends to serve as aminimum, requiring branches little use). Overtime must be paid for extra hours. Labor costs are higha bank employees. Inaddition, social contributions and taxes allegedly amount to ar these, 59 percentare taxes, 11percentsocial security contributions and 12 etc). Capital Requirements An additional BIS ratio capital requirementof two percentis required perbranchestablishedinSI0 PauloandRio de Janeiro andone percent per branch located elsewhereinBrazil. Other Costs Brazilianlaw also requiresdisabled access in all branches. Table 3.15 Correspondentsin Brazil(2001and 2002) Number of Contracting Institutions 52 68 Banks 53 Finance Companies 10 15 2.0 5.4 Source:Central Bank of Brazil 48. The best established of these chains i s run by Caixa, the giant public sector bank, which has a monopoly franchise on the network of about 8,000 lotto game outlets spread out across the country. In addition, Caixa is establishing new relationships with 2,400 correspondents, with the objective of being present in all 2,000 municipalities in which it i s currently absent. As a result, by June 2002, Caixa expected to be present in all municipalities of Brazil with at least one correspondent banker offering payment and deposit services. Payment transactions worth R$65 million are made each year through Caixa's lotto game outlets, of which 55 percent are worth less than R$40 each (US$16). Caixa Econbmica Federal, which runs the Casa LotCrica chain, estimates that the average cost of a bank transaction at a correspondent bank is only a fourth as much as at a branch. 49. The most significant expansion of correspondent banking services prior to the year 2003 was the creation of postal service correspondents, through the auction by the government, in August 2001 of the postal franchise. It was awarded to Brazil's largest private bank, Bradesco, for a price of Page 138 Pari 3. DownscalinaPrivate Banks R$201 million (US$8l million), which won the exclusive right to distribute financial services through the 5,532 agencies o f the Brazilian post office, ECT (Empresu Brusileiru de Correios e Telkgsrufos). The contract has a maturity of five years, which begins when each post office agency will be connected to Bradesco. All agencies are expected to be operational by 2004. The concession will therefore last at least until 2009. It can be renewed without a fresh biddingprocess. Under the terms of its contract, Bradesco has undertaken to connect agencies located in municipalities without banking service on a priority basis. ECT has a greater relative presence in the poor Northeast area of Brazil than the banks. Bradesco will be allowed to distribute basic financial services through the ECT such as payment and account transaction services, in exclusivity. The post office is present in 1,738 municipalities without a bank.13 Table 3.16 Geographic Coverage of the Banking System and the Postal Network ib Source:Lehman Brothers 50. Bradesco hopes to gain some 3.5 million new current account holders in two years from the post office client base. These clients would bring Bradesco's current client base of 11.8 million current account holders to 15.2 million. Through the post office, Bradesco intends to target low income individuals, especially in remote areas, which otherwise could not be targeted profitably even using ATMs. Bradesco may also transfer some of its existing clientele to the post office in order to reduce the cost of serving them. The current agreement covers the following services: demand deposits, savings deposits, check cashing, bill payment, remittance operations and collections. Bradesco undertook to start working with about 1,000 post office agencies established in the Northeast in order to fulfill its contractual obligation to start with the poorest regions of Brazil. On April 5, 2002, Bradesco opened its first twenty seven Postal Bank branches, with one for every state inBrazil, including the Federal District. According to the original timetable, 3,000 branches were to be opened by the end of 2002, and inpractice, 4,000 branches had been reported open by end-January 2004. 51. With the ECT network of Bradesco, and the Casa LotCrica franchise, it is aimed that all municipalities in Brazil will have access to at least one bank, and sometimes two, for basic financial transactions (payment, and account deposits and transfers), when these two networks are fully operational. Both institutions plan to extend the services offered by these networks to opening of accounts and credit granting, but several identification and security issues still need to be solved. 52. Other large banks in Brazil are increasingly establishing associations with local distributors to reach out to the masses. The Central Bank estimates that there are about 1,500 correspondent outlets in use in Brazil, excluding the LotCricas and the Post Office. Banco do Brasil for instance recently signed a correspondent agreement with a supermarket chain in Rio de Janeiro and ABN AMRO entered into an agreement with a 24 hour drugstore chain in Fortaleza. Many poor beach vendors inFortaleza use these facilities to make deposits or payments at night. Unibanco's subsidiary finance company Fininvest has signed agreements with the supermarket chains Magazine Luisa and Globex's Ponto Frio through which l3An estimated 25 percent of the total populationdo not or are not able to use bank services. Of these, about 4 million live in municipalitieswithout a bank, but with a post office. Page 139 Brazil; Access lo Financial Services it can distributeconsumer loans to customers in the stores. Through these three outlets, Unibanco reaches some seven million people of the lower middle and low income groups, mostly in the South and Southeast of Brazil. 53. However, the Fininvest agreements (and some other retail financial arrangements) are not considered correspondent banking agreements as Fininvest deploys its own employees in the store, instead of using the store's employees. This is more costly than a correspondent agreement given the relatively higher labor costs of Fininvest's own employees (though employees of Fininvest (which is a non-bank finance company) are likely to be less costly than its parent Unibanco, as they are not subject to the same labor laws. Nevertheless, there are still savings infixed costs relative to establishing a branch. 54. One reason cited for banks' reluctance to further expand their correspondent networks is the fear of labor union complaints, as union leaders consider that correspondents should be subject to the same labor laws as bank employees. Threats of court action have been made and banks believe that the court system in Brazil would tend to favor labor over employers, and potential indemnities could be costly. 55. In addition, correspondent agreements are profitable in small cities and remote areas, given the high cost of using supermarket space and installing the required technological connections, if the number of daily transactions is small. In addition, negotiating separate agreements with small chains with only a local presence in smaller cities could be difficult and costly. The extent to which the dramatic growth in correspondent agreements witnessed over the last two years can be sustained remains to be seem. Future agreements could remain limitedin size and present mostly inlarge cities. 56. To conclude, banking correspondent arrangements have a significant niche role in Brazil, but this is at least partly the result of regulatory costs imposed on bank branches. If the legal footing of correspondents is not clearly established to be independent of banks, this advantage could be eroded. Non-bank finance companies' arrangements with retailers represent the next level of outreach. The expansion of the major distribution networks of the Caixa and Bradesco represent the most important correspondent chains to date, and highlight the importance of outlets such as post offices and lottery shops. 57. The possibility of permitting some competition in the award or at least renewal of the franchise of such chains could help to ensure their future competitiveness. The attribution to Bradesco of the right to distribute banking products through the branches of the post office resulted from a competitive process. However, the decision to grant it with an exclusive right over the entire network ended up limiting potential competition in some municipalities with no other financial service provider. Post office franchises awarded in many other countries are subject to periodic contest, at the time of both contract award and renewal. 5.3 Bank Automatic Teller Machines 58. Private Brazilian banks attempt to use segmented delivery channel strategies, matching the cost of reaching certain segments of customers with the value added they generate. Lower cost technology, particular ATMs, are popular for reaching poorer (and potentially lower value added) individuals. Access to automatic teller machines has increased substantially over the last five years (from about 5,000 in 1996 to more than 16,000 in 2001) in all regions of Brazil, though service may again be biased in favor of the Page 140 Part 3. Downscaling Private Banks southern regions (Appendix Table Al.2).I4Fifty-nine percent are inthe Southeast, compared to only four percent in the North. Interms of ATMs per inhabitant, the Southeast, Center-West and South of Brazil are clearly the best served (Figure 3.6). But adjusting for differences in levels of economic activity, regional differentials are less distinct. The Northeast of Brazil actually has a fairly high density of ATMs relative to GDP per capita, while ratios for the South and Center-West are low. However, the Southeast is still clearly the best off region, even adjusting for income differentials, while the Northi s clearly the worst off. Figure3.6 Brazil Regional Distributionof ATMs Relativeto Populationand Income - Number of A'IMs per million inhabitants 1600 6 0 !40% 140 nor/, PO K ) O 1 l30 80v 80 60% 60 40% 40 20v 20 0% rthedst Soutli Southeast Center-west North Northeast South Southeast Cenler-west 59. Overall ATM coverage at four per 10,000 inhabitants is high in Brazil, even compared to more mature markets such as the USA (which has two ATMs per 10,000 inhabitants), based on its ten largest networks. Moreover, ATM growth has stabilized in the USA but is still growing inBrazil. Table 3.17 ATM Fleet in the USA and in Brazil American Express 8,700 8,600 Bradesco 18,940 7,708 6,500 Caixa EconBmicaFederal' 12,052 ClBC 4,314 4,324 Santander (Banespa) 985 Totaltoo ten 57.395 58.215 Toti these banks 81A34 Note:1This figure totals the number of Agencies and Service Outposts (PAB, PAE, lottery houses and bankingcorrespondents etc.) and they serve as a proxy for the total numberof its individual ATMs. Source:Card Industry Directory 2002 for the USA; individualbank datafor Brazil 60. Apart from correspondents, automatic teller machines have been the main means for banks to reach lower segments. I t a estimates for instance that 68 percent of its consumer loan applications are ~ made at ATMs. The cost of processing such applications is less than one percent of the same transaction made at a branch. Banco do Brasil estimates that 78 percent of all its transactions are made at ATMs, with a daily value of R$40 million (US$18 million). Fininvest, whose client base includes mostly people in classes C, D and E, calculates that clients in classes D and E use ATMs twice as often as the telephone, while customers in class C are indifferent. 14These figures include the number of Electronic Service Points o f each bank ("Postos de Atendimento Bancdrio Eletrhico" as defined by the Central Bank of Brazil), plus the Banco24Horas ATM network shared by 61 financial institutions in Brazil. Page 141 Brazil: Access to Financial Services 61. However, despite efforts to continue expanding networks, coverage remains partial. Further expansion of ATM networks to increase the outreach of private banks in lower segments may not be profitable. Expanding networks to small cities or rural areas would create large money transportation costs and pose severe security issues.Banks estimate that 4,000 to 5,000 daily transactions are requiredto make an ATM profitable. Such levels would be near impossible to reach in small cities, and even less so inremote areas. Competition Zssues in ATM Provision 62. Today most banks in Brazil have their own ATM networks. Sharing ATMs could facilitate the profitable expansion of ATM networks in remote areas. Large infrastructure sunk costs create barriers to entry and reduce competition. Some large banks are reluctant to share their ATM networks, due to their own sunk costs. Banks argue that ATMs are sales points and therefore are proprietary and should not be shared. In most other countries however, ATMs are shared. This creates an enormous barrier to entry for other banks who would want to penetrate lower segments but would not have an extensive ATM networks. Increasing competition would improve efficiency. It i s not clear that replication of the same infrastructure by competing financial service providers is really in the interest of final consumers, particularly in sparsely populated areas of very low income. Costs would likely be lower and entry easier with more sharing of existing infrastructure. In addition, reduced sunk costs would likely reduce bank spread^.'^ Banks with large existing networks could consider charging fees to other banks for widening the use of their networks. An investigation of the competitive compatibility of the present system could also be considered. As inthe case of the exclusive award of the postalfranchise, this could limit long term competitiveness in service provision. 63. An inadequate retail payment infrastructure also contributes to reducing competition and increasing the costs of small value transactions. Today in Brazil small value transfers are dealt with in a highly decentralized manner. Banco do Brasil clears checks (Compe); the Banco24Horas network clears electronic checks; credit cards are cleared and settled over two separate switching systems (Redeshop held by Credicard, and Visa Net). Several other small systems compete to clear electronic fund transfers for retail payments. Commercial banks in Brazil are currently establishing the SIP system for clearing and settling payments, in parallel with the new RTGS system for large value payments. These arrangements and their links to the new RTGS system need to be evaluated from the vantage point of how they are consistent with proper competition policy. Such reviews have been made in other countries such as Canada, South Africa, and the UK. 6. NEWTECHNOLOGIES -USEOF THE INTERNETAND PHONEBANKING 64. Developments in new technologies for information collection, storage, processing and transmission could potentially strongly influence banks' costs, in at least three important ways. First, they can lower costs by substituting computers for paper-based and labor-intensive methods of transactions accounting and intemal operations. Second, it could reduce personnel costs at the teller, for example through automated teller machines, telephone banking and PC banking, enabling institutions to reach a wider geographic base. Third, on the credit side, institutions can greatly increase the efficiency and l5Brazil's two largest public banks, Banco do B r a d and Caixa Econ8mica Federal, have indicated a willingness to begin sharing their A T M networks, and there are plans to launch a pilot program inmajor cities from early 2004, with a subsequent country wide rollout later in the year. Banco do Brasil and Caixa EconBmica Federal alone make up 40 percent of all A T M transactions in Brazil. Page 14.2 Part 3. Downscaling Private Banks decrease the risks of providing credit services by using credit scoring models.16The possibility of remote access of these technologies could eventually reduce the strategic importance of branches and ATMs. 65. High inflation and high labor costs provided incentives for banks in Brazil to develop such technologies. Surveys by private banks and the Central Bank suggest that costs of delivery are lowest for internet transactions, followed by ATMs, then branches. Transaction costs too usually are commensurately lower. Although internet would be the most profitable channel to reach out to lower segments, connectivity and fixed costs are obstacles. Brazilian banks find that nearly all of their internet clients belong to the richest social classes. Efforts to provide cheap financing for the purchase of personal computers and offer free internet access have not met with large success, or have not been targetedat the appropriate income segments." 66. Surveys indicate that a much wider range of persons, in terms of per-capita income, own cell- phones than computers, and a greater proportion of low income bank clients use phone-banking rather than internet banking. Phone users are forecast to grow much faster than internet users. Moreover, the voice activated methodology of phone based systems may be easier for less sophisticated clients. But again, both the cost of the instrument and connectivity remain large obstacles. 6.1 Telephone Banking 67. The telephone may allow banks to better penetrate lower segments. Some surveys indicate that the income distribution of cell phone owners is much wider than for computer owners, and the cost of a cellular phone call is muchcheaper than in many other countries in the world. Figure3.7 Braziland Other Countries BankTransactions and Cellular Phone Costs - Estimated cost per banking transaction Analog cellular 3 minute call, average price ($) 140 1.2 120 1 100 0.8 0 80 0.6 060 040 0.4 020 0.2 000 0 Bnnch Onhne 1998 1999 2000 0US Latin America Brazil Brazil 0 Canada HongKong South Africa Source World Brok.Tower Group Research Source:World Bank, Tower Group Research 68. Brazil's phone industry appears quite competitive, with at least two operators for fixed and cellular lines in each region, and two national providers for long distance and international calls. Cellular phone charges have fallen significantly, from US0.77 per three minutes in 1998 to US0.24 in 2000. An equivalent fixed line call i s muchlower, at around US$0.04 16See Part 6 of this study for details. Credit scoring i s an automated statistical method used to assess the risk of default of a credit applicant. This substantially reduces costs of underwriting a credit since it reduces the role of human evaluation, thus allowing financial institutions to profitably underwrite credits of smaller sizes. For example, Bradesco financed about 20,000 computers, but these were directed first to its employees, many of whom were well-to-do. Page 133 Brazil:Access to Financial Services 69. Although these fees are in line with, and even slightly lower than in many other countries, they represent a higher proportion of Brazil's GDP per capita. Connections charges represent about 0.75 percent of the GDP per capita of the poorest Northeast region (compared to 0.05 percent in most of Europe) and monthly subscription fees about 0.5 percent (compared to 0.05 percent in most of Europe). They represent a very high9 and 6 percent respectively of the minimumsalary. Inaddition, the subscriber must pay the cost of the phone itself, an activation fee of about US$6 and a monthly subscription fee of about US$11. Most banks charge a monthly fee ranging from US$O.XO to US$2.00 for phone banking. The relatively highcost of a telephone line is also a factor. Brazil, like many developing countries, has a Universal Access Plan for telecommunication services and, the Government is investing about $600 million a year in building out telecommunication infrastructure in underdeveloped areas. Effective penetration will depend as much on the nature of the regulatory regime adopted, including tariff structures, as on the installation of physical infrastructure. Figure3.8 Teledensity and GDP per capita and Telecommunications Brazil and Other Countries - 8 0 , Teledensity and GDP per capita Cellular phone charges (?hGDP per capita) -- I 1.oo 0.10 0.01 0.00 , % Brazil Mexico Hangkong United Finiand 0, Kingdom 0 5 a 15 20 GDP per Capita US$ -25 30 35 40 j lCelluiarcallcost IJMontlysubscr charge IJConnectioncharge Note:Teledensity = Total telephone subscribers per 100 inhabitants (2001). GDP per capita (US$) (2000). Source:International Communication UnionData Bas Figure3.9 Main Telephone line, Cellularand Internet penetration Brazil and Other Countries - 1 100 o M a i n telephone lines RaCellular mobile s ubs c ribers =Internet Source: InternationalTelecommunications Union, 2002. Page 144 Part 3. DownscalinaPrivate Banks Page 145 Brazil:Access io FinancialSewices 70. Overall teledensity in Brazil is however low by international standards, and in particular for lower segments (Figure 3.8). At the end of 2001, there were about 45 million fixed telephone unitsand 30 million cell phones in use in Brazil, for a population of about 170 million. But, a large number of these telephones are used by offices and many rich families have more than one phone. A study inSlo Paulo in 1999 showed that 98 percent of the highest income bracket individuals owned a phone, compared to only 8 percent of the lowest bracket. And on a per unit of population basis, international comparisons confirm that overall connectivity is low in Brazil compared to developed or emerging economies, and is also lower than in large neighboring countries. Therefore, as for the internet, Brazilian banks have not yet used phone-banking to reach out to lower segments, although the potential may be larger (Figure 3.9). 6.2 Internet Banking 71. The Central Bank of Brazil estimates that the average banking transaction using the internet costs R$0.24 (US$lO cents), compared to R$0.33 (US$13 cents) for dialed-line banking services and R$2 (US$80 cents) for transactions at a branch. Estimates by TowerGroup Research confirm these wide savings." Brazil's largest banks have all developed extensive internet-based activities, and comparisons with Asia's emerging economies based on survey evidence suggest that a higher percentage of users in Brazil make extensive use of new technologies (Figure 3.10). Brazil's banks are even succeeding in closing the gap with the USA in terms of internet penetration among clients (Figure 3.11). Whereas the use of computers and the internet inBrazil for recreational and leisure activities i s lower than in the USA, for purposes of work and services such as banking, Brazil scores better than the USA, at least among the connected segment of the population. Figure 3.10 Braziland Asia On-Line Internet Banking - Activities of regular internet users Online Banking Millio ns o f o o-line us e rs Customer base Cliatted (iii-liiie Scl,oaIIC*C.IIL/l CScd 5CdILh cllglll~ Work rc\urLIi 0 I O 20 30 40 50 60 Source:World Bank staff estimates l8 Estimates all confirm huge cost differentials. Based on US data for 1996, Booz Allen and Hamilton estimate that the average transaction cost through the internet is less than 1percent of that of a transaction at a branch, and about 4 percent of A T M transactions. Institutional Investor Magazine estimates that transactions at a teller cost US$1.53 in the USA, versus US$0.6 at an ATM and US$O. 10 via internet. Page 146 Pari 3. DownscalinaPrivate Banks Figure 3.11 Braziland the USA Internet Use and Online Banking - Telephone Teller I I I I tise for i o m e o r t e w 30 60 72 brnkmg iidnsactm ns Use torrllor most 11 bdnkmg i 5 9 p q : : trrnsactmns 72. However, huge disparities in access to internet services limit the use of this technology to extend financial services to the poorer segments of the population. The Central Bank reveals that only 10 million people (Le. 6 percent of the population) own a personal computer and 8 million use internet. An estimated 90 percent of the 8 million intemet banking users belong to highincome classes.ig Only about 4 percent of total individual bank clients used internet banking in 1999 and only 10 percent of corporate users. Access is in part limited by supply, as the smaller Brazilian banks do not have websites, but also due to low connectivity, especially inrural areas, although growing rapidly. And buying a computer i s out of reach for most of Brazil's poor. Unless products offered and internet access channels are modified, it will be difficult to use this technology for large scale popular outreach.20And growing internet security concerns would also have to be addressed." 7. DOWNSCALING OFBANKS ELSEWHEREWHAT CANBANKSDO? - 73. What have banks in other countries done to achieve `downscaling'? In some countries the government has instituted special schemes to promote access through the banking community. These forms of government assistance are discussed in Part 7 of this study. There are however also measures which banks can take themselves, and typically, banks are more motivated to enter the markets of the "banked' in environments which are more competitive, and where the growth of credit services is the l9Bradesco reveals for instance that 83 percent of their internet clients have an income above R$15,000 (US$6,700) and 65 percent have a university degree. Unibanco estimates that more than 75 percent o f them are in the richest income class and less than 5 percent in the two lowest groups. They are highest in a few o f Brazil's largest cities, Brasilia coming first, followed by SZo Paulo and Rio de Janeiro. Brazil is not alone in it limited success in using the Internet for outreach of financial services. Citibank in India had initiated a small-savings based internet scheme, Citibank Suvidha, which accepted initial deposits of as little as Rs 500 (US$lO). But low income depositors have been few and the initial deposit doubled within a short while of the opening of the scheme. See www.citibank.com/india/suvidha/ Comit&Gestor da Internet - Brazil. In2000 alone, there were 5,997 reported successful system intrusi0ns.Arecent report by the Gartner group indicated that while the largest 500 companies inBrazil spent an average of 16 percent of their budgets on informationtechnology in 2001, very little o f this was spent on security. Gartner's survey stated that Brazilian security investments were just under 2 percent of their technology budget against a worldwide trend of around 4 to 6 percent. Only 13 percent of the largest 500 Brazilian firms are preparedfor,large losses; 28 percent have some kind of crisis management and business continuity plans and an estimated 36 percent have insurance plans against sizable losses. Page 147 BraZik Access to FinancialServices main vehicle of growth for the financial system. These approaches have several identifiable elements which are sometimes presentexclusively and sometimes incombination. These include, first, the adoption of new technological platforms which are more suited to microlending, or serving a community with the profile of the unbanked. A second common element i s the building of partnerships with microfinance institutions, non-banks, correspondents or non-government organizations. Third, the use of image differentiation and the adoption of special programs of outreach. The following sections describe some examples of these practices. 7.1 DownscalingThrough Partnerships Alliance of Standard Bank and Afn'ca Bank Partnerships and New Technology - 74. In South Africa, outreach by the formal banking community has been encouraged by the government, although there are no explicit legal requirements. Moreover, competition in the banking industry is immense and domestic growth opportunities are scarce.22Commercial banks have chosen different ways to grow into this market; through joint ventures, acquisitions of existing microlending institutions, correspondent type low-cost retail chain branches and also through the introduction of new products such as smart cards (Table 3.18). The most successful of these have been the alliance of Standard Bank with African Bank, which illustrates the advantages of a regular bank teaming up with a microlendinggroup to extend its outreach to small clients.23 75. Standard Bank's E-Plan takes advantage of the relative strengths of the two institutions - the strong technology platform and credit scoring techniques of Standard Bank and the familiarity and outreach to the black community of African Bank. The revenue is split 60:40 between Standard Bank and African Bank. Potential E-Loan customers are required to have a paycheck and a direct-deposit account with Standard Bank for at least three months (African Bank does not take deposits). The success of the product may be traced to a series of factors. First, the automation of all transactions. After an initial application and interview with a branch employee, customers undertake all financial transactions (including rolling-over their loan) at the ATM machines in Standard Bank "E-Branches", which cuts the cost of providing numerous, low-value transactions. Second, the Bank has real-time access to the borrower's account information, and can withdraw loan payments electronically when paychecks are deposited. Thus loan defaults are very low (about 2 percent). Third, interest rates are very high at about 75 percent per annum, as this category of borrowers has limited financing options. Additionally, automated credit scoring techniques are used, using a Standard Bank proprietary credit model and the customer's deposit-account information to "score" the customer and give immediate loan approval or rejection. Standard Bank also checks the MicroLenders Credit Bureau (MLCB) for any unreported outstanding loans. 22Moody's South Africa Banking Sector Outlook, December 2001 23Some of these initiatives admittedly had disappointing results, though these could have been the result of unsound banking practices. According to the Economist Intelligence Unit, Unifer, now a subsidiary o f Absa, admitted that i t had overstated profits by some 10percent, and that poor lending had resulted in a shortfall inprovisions. Saambou, one of the smaller institutions active in micro-lending, issued a profit warning and said that it was likely to record a half-year loss o f R5m (US$436,5 lo). Unifer and Saambou financed personal capital purchases with aggressive sales targets. Page 148 Part 3. Downscalina Pn'vafe Banks Table 3.18 South Africa DownscalingThrough Partnershipsby Commercial Banks - Standard Entered into co-operation with African Bank,leading microlenderin South Africa, to provide African Bank loans through Bank AutoBank E centers. This electronic banking service targets low end customers and does not have a minimum income requirement, but customers need to have regular incomestreams. African Bank offers StandardBanks microlending productsand carries out credit approval and collection functions. Both banks increasedtheir customer basefor micro loans ABSA Aquired FBC Fidelity Bank and merqed it into its own Peoples Bank to offer bankinqservicesto the underbanked mass Nedcor redit card and transactionalbusiness. servicesto those who have been excluded BoE Bought Credcor, a retailfinance company using proprietary credit assessment techniquesto providefinance mainly in the retailconsumer credit arena, through a network of retailers but also directly throughits outlets. take in Saambou bank, another pioneer in the industry, through the acquisitionof Fedsure. Sa increasedits exposure in the market significantly(upto R2 billionfrom R80 milliontwo years ago), using its network of agents lnvestec In additionto the cooperationwith Standard Bank,African Bank formed an alliancewith Edcon Group to expand its distributionnetwork. Sources:Moody's, Mediaarticles,Bank web pages. 76. Overall, the E-plan is an important source of consumer microcredit to the black community in South Africa and i s a first step for offering formal banking services to an underbanked population. The provision of microcredit through banks i s an improvement over microlenders as interest rates, though high, are lower than microlender rates (which generally exceed 200 percent annually). Although the current product offering i s small and limited to consumer loans, there i s the potential for customers to buildcredit histories that could lead to great consumer and commercial credit. 77. The South African examples illustrate the importance of the enabling environment, in two specific areas. First, the usefulness of widely available and easily accessible credit information, available for microlenders, retailers and bothsmall and large banks. The success of this i s due to the presence of the leading international bureaus such as Equifax and Transunion - who compete on price and services provide various end-user products such as credit-scoring models (to banks and retailers) and customer reports (to microlenders). The role of credit bureaus inBrazil i s discussed ingreater detail inPart 6 of this study. 78. A secondfactor is the increase incompetition which has created pressuresfor the largest banks to move downscale. Nevertheless, there i s a perception that the costs of some financial services in South Africa are highdue to insufficient competition. South African authorities have conducted a major analysis on access to financing for small and medium enterprises which takes a systemic look at competition issues across payment, savings, and credit-related services in South Africa. One of the conclusions of the report i s "the problem i s not so much the availability of (debt) finance as such for small enterprises, but rather inefficiencies in the provision of financial services to such users due to competitive factors such as barriers of entry of potential new providers. Many of its conclusions are along the lines of a similar study on competition in financial services undertaken two years earlier in the UK (the Cruickshank rep~rt).'~ This illustrates again the desirability of looking in greater detail at the role of competition issues in the downscaling of financial services inBrazil. 24"UKBanking: A Report to the Chancellor of the Exchequer" HMTreasury UK;Pressrelease 12810020 March 2000; and from the Reserve Bank of South Africa, 2002, SMEi's Access to Finance in South Africa -A Supply- " side Regulatory Review" Hans Falkena et. al. available at http://www.finance.gov.za/documents/sme/. Page 149 Brazil: Access to Financial Services Peru And Bolivia 79. In Peru, three of the largest banks inthe financial sector have substantially higher participation in the financing of microenterprises than the overall banking system.25Banco Wiese Sudameris was one of the first commercial banks to enter microfinance, and has been recently followed by Banco Continental BBV and Banco de Credito del Peru. Smaller banks and finance companies like Banco de Trabajo, Financiera Solucion and Financiera Daewoo have changed their focus from consumer loans to microloans.26In Bolivia, Banco de Credito de Bolivia increased its presence in microfinance, and Banco Economico has also a much higher share of its portfolio in microloans than the rest of the banks in Bolivia. The Bolivian banking system is not as concentrated as in Peru, but Banco de Credito i s one of the largest Bolivian banks, and BancoEconomico i s a medium size bank. Table 3.19 Peruand Bolivia: Banksand FinanceCompanies in Microfinance(US$ millionand%) Banco Continental BBV Peru 9 0 4 114 6 0 109 5.6 5 6 BancoWiese Sudameris Peru 45 1.6 33 1.2 28 0.8 Bancode Creditodel Peru Peru i a 0.6 23 0.7 27 1.o 40 34.3 51 37.0 47 75.8 51 Bancode Creditode Bolivia Bolivia 43 Banco Economico Other Banks Banking System Bolivia 123 3.7 ia6 6.2 122 4.3 I/ForBolivia,theinformationisfromJune2001.Fortheothercases,theinformationISfromtheendoftheyear 21Share in total loan portfolio of each financial institutionlisted Source: SBS and SBEF 80. Commercial banks are the second largest investors in regulated microfinance institutions in Peru, predominantly through their financing of CMACs (municipal savings and loan offices), and to a lesser extent, their participation in EDPYMEs (regulated microfinance institutions). The high number of closed CRACs (rural savings and loan associations) precludes banks of financing these type of institutions. 25Banco de Credito del Peru i s the largest bank inPeru measured by the size o f is assets at June 2002. It i s followed by Banco Wiese Sudameris and Banco Continental BBV. 26Consumer loans are a close cousin of microfinance, and is quickly moving closer. According to Christen and Drake (2001), small commercial banks and finance companies are more flexible and adaptable than the large banks, although the later have a large network o f branches and information systems that are very important for the provision of microfinance services. Page 150 Pari 3. Downscaling Private Banks Table 3.20 Other LatinAmerican Banksand FinanceCompaniesin Microfinance o Agricola (L), Banco Salv ode Pichincha(L), Banco Credito Familiar Note :S: Small, L: Large Source: Valenzuela (2001),`Gettingthe Recipe Right: The Experiencesand Challengesof Commercial Bank Downscalers',Working Draft, MicroenterpriseBest Practice Project, USAID. Table 3.21 Sources of External Financinafor Peruvian MicrofinanceInstitutions (Dec 2000 -nuevos soles ml CRACs 106.1 3.9 13.7 123.7 85.5 89.1 0 Total 99.4 Source: SBS 81. Besides funding microfinance institutions, commercial banks have established partnerships by investing in the capital of microfinance institutions. Banco Wiese Sudameris owns 6 percent of Mibanco. Some banks have established or reorganized existing financial institutions following acquisition or change of control. This is the case of Financiera de Credito in Peru, which is fully owned by the same holding company that owns Banco de Credito. Originally focused inthe market of consumer loans, inrecent years has become one of the largest providers of microloans in Peru. In Chile, Banco Santander bought a finance company that has already established a strong microenterprise credit program within its consumer loan portfolio. This finance company, BANAFE, i s the leading supplier of microcredits inChile." 7.2 Downscaling - A New Image for a Different Community? 82. Successful banks which have extended their outreach to the underserved emphasize the importance of image differentiation and the use of different channels of service delivery for different client groups. One recently cited example in the USA i s Mitchell Bank, a small bank which has focused on the underbanked community of immigrant teenagers, in the suburbs of Milwaukee. Its service branch, located in a low income district, i s runby students who earn credit towards their graduation requirements, while working under adult supervision. At least 90 percent of its customers have never had a bank account before. One indicator of its successful outreach i s that many students' parents now deposit their pay checks there. Another key element in their strategy i s the establishment of ties with community organizations, such as schools and clubs. Another U S bank, First Bank of the Americas, has likedup with church networks serving immigrants. Such channels could also be used for financial education ''Christen andDrake (2001). Page 151 Brazii:Access to Financial Services programs - First Bank has hired a Spanish speaking marketing consultant who addresses churchgoers about the dangers of money lenders and the benefits of established banking organizations supported by the church or community leaders.28 83. Some financial institutions in Brazil are very aware of the role of image differentiation; Unibanco points to its consumer finance company, Fininvest, a 100 percent owned subsidiary, as an example of its product differentiation strategy. Fininvest focuses on consumer credits and personal loans, and had over R$8 million in current operations in 2002 and 109 branches. Unibanco also offers these financial products through its 50:50 joint venture companies, Pontofrio and Luizacred, which operate through store chains (of 358 and 126 stores respectively). Among the benefits of using such companies, it cites reduced client acquisition costs, cross-selling opportunities, specialized credit expertise for low income segments and appropriately tailored financial products.29 84. Products successfully used for outreach include `payroll card' programs, given to employers, who then put funds directly into the accounts of employees, typically teenagers or immigrants. The employees are then able to use debit cards to operate these accounts, cobranded by popular names such as Visa. Check writing or credit facilities are not included, thus limiting the banks' risk, and helping to reduce the conflicts between the marketing and risk-control departments of banks interms of accessibility versus fiduciary responsibility. For immigrant communities, another popular device has been dual ATM cards, which allow a party in the USA to deposit funds which could be withdrawn by a recipient abroad, as an alternative to expensive wire transfer services. Designated community banks such as the First Bank of America, in Chicago's Hispanic district, have the support of the US Treasury in their basic loan programs aimed at the previously ~nderserved.~' 85. There are also examples of more relaxed entry requirements. Thus some banks in the USA tacitly acknowledge the cards issued by Mexican consulates, the `matricula C O I ~ S U Z U Y ~ ,regardless of legal status, as proof of identity. 86. Finally, government programs to boost access in the USA have gained ground. Legislation was recently passed requiring all federal payments to be made electronically, which exposed the large numbers of unbanked. The Treasury Department has set up a new Office of Financial Education, which has launched a pilot program called First accounts, operating in 26 states, designed to bring 35,000 persons into the banking system for the first time (Part 7 has details). 7.3 Downscaiing -Suggested Good Practice 87. Largely due to the Community Reinvestment Act and its various modifications, there has been a keen interest in the USA among analysts and academics inexamining `what works' in terms of effective 28One of the four leading UK banks has recently set up a separate community banking unit focusing on providing savings and loan services for housing-associationtenants, banking social enterprises, childcare organizations, and community developmenttrusts. This bank has also createdan innovative financial inclusion product. I t announced a partnership with ProspectCommunity Housing Association whereby Prospect's tenants would be able to open savings accounts (paying interestof 2.75 percent) and apply for loans of up to 22,000 (up to four times savings). Bank personnelwould be trained to accept rent booksandpower bills as forms of ID.Prospect staff would assist tenants to take advantage of these products, reducing their reliance on sub-prime lenders. 29Presentationof Unibanco at the World Bank seminar on Access to Financial Services, February 2003. Washington DC. 30Main banks in the UK have also startedproviding basic bank accountsto meet the special needs of low-income people.Typically, these accounts do not include an overdraft option or checkbook; the Bank of Scotlandhas gone further, offering a bill-payment account linked to the basic account. Page 152 Part 3. Downscalina Private Banks outreach. One practiced observer has proposed a menu of five key elements for any strategy for bringing the unbanked into the financial main~tream.~'These include: 88. First, open specialized bank branches that provide commercial check-cashing outlet (CCO) services. These "outlets," should be conveniently located for lower-income households and should provide fee-based check-cashing services in addition to traditional consumer banking services. Opening such outlets serves three purposes. First, by offering check cashing services in a bank branch, the bank builds a relationship with its check cashing customers. Second, the establishment of the outlets recognizes that many check cashing customers are likely to be slow to open deposit accounts. Third, check-cashing fees provide new sources of revenues for banks with branches in lower-income areas which often report that it is difficult to cover the costs of these branches with traditional services. In Brazil, the opening of bill paying outlets such as the Caixa's LotCricas may serve a similar purpose - a proportionally smaller part of payments are received in the form of checks. Additional emphasis on check cashing could however be added. 89. Second, offer "starter" bank accounts that have low minimum-balance requirements, cannot be overdrawn, and include low-cost money orders for making long-distance payments. In addition to check- cashing services, the outlets should offer traditional deposit accounts and low-cost, low-minimum-balance savings account that gives account holders the option to buy money orders for long-distance payments. Brazil's banks have offered such minimumpackages of services; and recently new regulations have led to a reduction to zero, for the client, of a basic account, which has also been combined with a simplification of eligibility and entry documentation. 90. Third, offer accounts specifically designed to help people accumulate savings. Savings building accounts should entice customers to systematically save. They should include features such as financial penalties for closing savings accounts, and automatic but low periodic transfers to savings accounts. In Brazil, these would be different from the current passbook savings accounts, which have the benefits of flexibility and low entry requirements but which do not remunerate the depositor at market rates, or help to develop good savings practices. 91. Fourth, offer deposit-secured emergency loans to individuals whose credit histories make them ineligiblefor traditional mainstream credit. In this scheme, collateral could be the balance that a member has accumulated in a savings-building account. For individuals without savings, the outlets might consider working with a third party, such as a not-for-profit community-based organization to raise funds. The new provision for simplified micro loans are a step in this direction. However, it i s proposed that the present recommendation would expand such loans to include collateral defined more broadly, and also include persons with limitedor impaired credit histories.32 92. Finally, seek community-based organizations and offer financial literacy programs. The value of community partnerships has been widely acknowledged in microlending; it could also be incorporated into the downscaling of mainstream banking. Partnerships with such groups help promote outlets to the community. Such outlets can also leverage the community organization's partnership to conduct financial counseling programs. Such an outreach based strategy i s preferable to the traditional approach for at least four reasons (i) it will attract the unbanked into bank branches; (ii)will offer the unbanked services that it are specially designed to meet their needs; (iii)it will help the unbanked to become traditional bank 3'The term "unbanked" hererefers to individuals who do not have deposit accounts of any type. See JohnP. Caskey (2001), 'Reaching Out to the Unbanked', http://www.chicagofed.org/cedric/2001/session1paper3.pdf 32Law No. 10735 of September 11,2003 (formerly ProvisionalMeasureNo 122of June 25,2003), CMN Resolution3109 of July 24, 2003 and CMNResolution3128 of October 30,2003, createdadditionalmeasures which include microfinanceoperations destined for low incomeearners and small businesses. Page 153 Brazil: Access to Financial Services customers, and (iv) in the USA, it is appealing to banks because delivering financial services to lower- income community organizations counts towards banks' Community Reinvestment Act obligations while offering them some additional credit quality protection and hence a less risky rate of return on their investment. Page 154 Part 3. Downscalins Private Banks ANNEX 3.1 BANK DOWNSCALING - ONE BANK'S VIEWPOINT 93. This annex provides an overview of successful practices in commercial bank downscaling to low-access groups, as perceived by one commercial bank in the USA with considerable practical experience in this area - ShoreBank of Chicago, which has been noted for its successful outreach in the poor and minority neighborhoods of urban Chicago.33ShoreBank now has partnerships with institutions in other countries, especially EastemEurope, and also Latin America, aimed at promoting access for the poor, for persons in remote locations or in minority communities. It has developed programs of overseas outreach, which are also touched upon in this annex. It seeks to identify (i) appropriate products, both retail (for individuals, in Section 1below) and (ii) products for small and microbusiness (Section 2) and (iii)successful market outreach strategies (Section 3). Within these contexts, it discusses sound principles of client selection and eval~ation.~~ 1. Retail Products Get Checking 94. Get Checking is aimed at individuals who have never had an account or those who have mishandled a checking account in the past. It began in 1998 in Wisconsin as a partnership among financial institutions, community-based financial educators, and an eFunds Corporation (the owner of ChexSystems). Participants must undergo six hours of instruction and training in checking and savings accounts at a cost of $35 for course materials. Upon completion and after passing a test, participants receive a certificate that guarantees the right to open an account, provided previous credit or account problems have been resolved. Participating banks waive the ChexSystems report requirement when opening an account. This is an important feature of the program, since most banks block the opening of accounts for a period of five years to those who have had a negative ChexSystems report. As of December 2001, 1,012 individuals participated in the classes, and 914 received certificates. Nearly 75 percent of participants were below 35 years of age, and 79 percent listed their race/ethnicity as African- American, Latino, Asian, or other. Eighty-eight percent of participants had incomes 80 percent or less of the Milwaukee County median. Today, Get Checking i s offered in 37 locations across the country. For example, as of June 2002, Wells Fargo Bank Wisconsin, which played a major role in developing the program, had opened 85 checking accounts and nine savings accounts for program graduates. U S Bank was the first bank to offer the programnationally in 2002.35 95. In terms of applicability in Brazil, such a program could help those who have been on a `negative list' at SERASA or other credit reporting systems. The participation of the credit reporting 33 This annex is based on a background paper prepared by ShoreBank Advisory Services for the World Bank. 34 ShoreBank points out that there are significant differences in cross-border concepts of low financial access. For example, in the United States ShoreBank's target markets in the South Side of Chicago suffered exclusion 30 or 40 years ago largely due to minority status and location in less desirable neighborhoods. As commercial banks have discovered the business opportunities offered in these communities, access to financial services has become more a function of income level, business size, credit history, and immigrant status. Inthe transitional economiesof Eastern Europe, access is low due to political instability (or continued military conflict), nascent financial systems and short operational histories of small or micro enterprises. InLatinAmerica, income level, geography, race, and business size all limit access to financial services, as do popular distrust of financial institutions following currency crises and other shocks to the financial system. 35Williams, Marva and Kimbra Nieman, January 2003, "The Foundation of Asset Building, Financial Services for Lower Income Consumers". Woodstock Institute. Page 155 Brazil:Access to Financial Services agency itself together with the banks i s noteworthy, as i s the incentive compatible structure, which requires applicants to purchase course materials, thus reducing moral hazard. Payday AlternativeLoan (PAL) 96. Payday loans are a high-cost source of emergency cash for low-income individuals. Payday lenders (often check-cashing companies) provide loans to individuals based on future paychecks. These loans are also known as "cash advance loans" or "post-dated-check loans." The fees and interest charged during the 14-day lending period between paychecks translate into APRs as high as 300 percent or more.36This expensive payday debt can accumulate quickly. The borrower profile would be someone with a poor credit history, or outside of the formal financial system. 97. North Side Community Federal Credit Union in Chicago has developed a new product within the last year-the Payday Alternative Loan (PAL) -to help those who have accumulated payday debt get out of the trap. It is a $500, six-month loan made at 16.5 percent APR. PAL i s approved automatically and funded immediately for any community resident who has an income of over $1,000 per month, regardless of credit history. The credit union seeks repayment from payroll deductions to reduce delinquency, and charges an application fee. Researchconcerning the clients who have used the product indicates that 16 percent had filed for bankruptcy in the last 3 years, and that they had an average credit score of 548 (only 14 percent had a credit score above 600). The credit union i s reporting all repayments to the credit bureau to help members build positive credit histories, and i s partnering with a financial literacy provider. The product i s currently inthe pilothesting stage, but to date 523 loans have been made; while delinquencies are common, write-offs are minimal.37The product will be break-even under the best of circumstances, but it provides an important entry to financial services to low- and moderate-income individuals. Note that the use of positive credit histories is especially noteworthy in the Brazilian context, where positive credit history i s limited. Extra Credit SavingsProgram 98. In 2000, ShoreBank launched an initiative at two of its branches in Chicago aimed at bringing low-income new depositors into the financial system, through partnerships with the Center for Law and Human Services (CLHS).38ShoreBank offered its space on nights and weekends for CLHS to process tax returns of low-income individuals for free. Most of these low-income working households were eligible for the Earned Income Tax Credit, a refund worth as much as $4,000. The Program arranged for the opening of accounts at ShoreBank for the direct deposit of tax refunds to the new accounts. Over two years, the Programattracted 202 new accounts, and over $200,000 in deposits. Twenty of the new account holders also opened savings accounts, checking accounts, and Certificates of Deposit. While 63 percent of the accounts were spent with two to three months of the program, 14 percent continued to use their accounts for savings, and the remaining 22 percent for transactions. More than 70 percent of respondents to a follow-up interview said the account changed the way they used their tax refund and helped them save more than they would have without the account.39 36 State of California, Department o f Corporations website, http://www.corp.ca.gov/pub/payday.htm,"Payday LendingFact Sheet", September 2001. 37EdJacob, Manager, North Side Community FederalCredit Union, Periodic Report, March 31,2003. 38CLHS: a non-profit agency that seeks to increase the resources o f low-income households through improved access to public entitlement programs 39ShoreBank and the Center for Law and Human Services, 2001, "Money in the Bank: The Extra Credit Savings Program". Page 156 Part 3. Downscaling Private Banks 99. While this Program was institutedon a small scale and centers around a specific aspect of the U.S. tax system, it points to the potential for building access to financial services through establishing accounts for electronic payments of any kind, including paychecks or remittances. It also provides an example of a successful partnership between a community financial institution and a non-profit community group. Remittances 100. Remittances have received much attention in recent years as a profitable new product for commercial banks, such as Citigroup, Wells Fargo, and Bank of America. This profitable niche for transfers and foreign-exchange fees, estimated at between $900 million and $1.8 billion annually, has ' been dominated by Western Union.40An example of such a new remittance product i s Bank of America's SafeSend, a wire transfer service that can be accessed from 20,000 ATMs in Mexico using a SafeSend card sent to the user via courier with a transaction cost of $10. Bank of America views the remittance product as a means of capturing new Hispanic clients (39 million in the U.S.). Currently, 37 percent of those that have signed up for SafeSend have opened other accounts. The bank expects to receive no less than 80 percent of future growth inretail banking from the Hispanic market. Only 58 percent of Hispanics have checking accounts, versus 93 percent of non-Hispanic whites.41Mexicans in particular are gaining greater access to the financial system in the U.S. with the Matricula Card, the Certificate of Consular Registration offered to Mexican Nationals by the Consulate of Mexico.42 The Matricula provides an acceptable form of identification for opening bank accounts, and thus moves savings from under mattresses into bank accounts. As an outreach strategy, Bank of America has been actively marketing to Mexicans as they wait in line at the Consulate for their Matricula document.43 The growth of the remittance business can also be used to attract more clients to the financial services sector inthe countries of the recipients, as they utilize ATMs to access their funds. Remittances form a relatively low part of financial receipts in Brazil, incontrast to other countries in the Latin America region. 2. Small and MicrobusinessProducts Term Loans -Upto One Year 101. Commercial banks typically have entered low-access markets with pilot products offered through test branches, later expanding their product offerings as profit and performance indicators demonstrate market potential. Demonstrationeffects can be powerful inthis context. Loans with a term of up to a year are the most standard product for entering this market niche. Long-term loans to small enterprises in certain neighborhoods were rare in the late 1960s when one of the founders of ShoreBank first made a loan to a minority-owned McDonald's franchise in the urban core of South Chicago. At the time, this was considered a highly risky transaction by mainstream banks, given the location of the business and the structure of the McDonald's franchise agreement. Today, however, these loans are readily underwritten and later securitized by mainstream banks. 102. This has beenfacilitated also by readily available credit bureau information, although borrower credit histories in the low-access market may be unfavorable. Income tax returns are also widely available, and many customers have established account relationships in the bank where they apply for a ~~ ~ ~~ 40 Krebsbach, Karen, September 2,2002, "Following the Money", U.S. Banker. 41 Tully, Shawn, April 14,2003, "Bank of the Americas; BofA is betting its future on the Hispanic market", Fortune. 42 Williams, Marva and KimbraNieman, January 2003, "The Foundation of Asset Building, Financial Servicesfor Lower Income Consumers", Woodstock Institute. 43 Tully, Shawn, April 14,2003, "Bank of the Americas; BofA is betting its future on the Hispanic market", Fortune. Page 157 Brazii: Access to Financial Services loan. While some banks, such as Wells Fargo and Bank of America, have developed credit-scoring models for term loans to small businesses, traditional credit analysis i s still employed by many banks. Credit-scoring models may improve efficiency and allow banks to reach a greater number of clients, but they also may exclude potentially viable borrowers, or fail to identify a future default. 103. ShoreBank has explored developing a scorecard for loans below $150,000, but has chosen to rely on its traditional analysis process, to avoid disqualify potential borrowers outside the realm of a traditional scorecard. ShoreBank's lending methodology i s a rigorous underwriting process, utilizing all available information on the company, its owners, and affiliate companies. Efficiency i s reached by streamlining the application processes so that documentation requirements do not overwhelm clients; by combining these with seasoned lenders with the ability to quickly assess a potential deal (based primarily on cash flow projections, market potential and skill level and character of the management team; and by adjusting the depth of analysis for new versus repeat borrowers, and for loan type - trade, production or start-up lending. 104. Consistent monitoring of the outstanding small business portfolio i s a significant factor in success. While ShoreBank may have a higher level of delinquencies than its peers (banks with similar asset size), suggesting higher risk-taking, it also has higher write-offs. Monitoring includes quarterly collection of updated financial information, site visits and tracking of performance against projections. ShoreBank has incorporated a risk rating system to inform management of asset quality. In its partnerships overseas, ShoreBank encourages similar quarterly monitoring to look for changes in management, observe the flow of business, and check the presenceand condition of collateral. Structuring frequent (even monthly) payments prevents the entrepreneur from spending available cash for loan repayments on other businessneeds and these payments serve as an opportunity to visit the client. 105. ShoreBank contrasts the effectiveness of its continued personal involvement as an early warning system, with the lower monitoring of most mainstream banks that rely more heavily on credit scoring methodologies. In developing markets, where credit bureau information i s scant and small businesses lack substantial track records, credit analysis is more labor-intensive in terms of data collection. ShoreBank Advisory Services has been working to adapt a partial scorecard on a pilot basis with some of its partner banks in Eastern Europe. Banorte in Mexico has adopted a "mini-scoring" model for its small business portfolio. While Banorte can take a credit decision in a couple of hours based on the scorecard output, it spends days collecting the necessary information for the scorecard. The scorecard automatically excludes companies with less than three years' experience. 106. Beyond intensive due-diligence, ShoreBank has always sought to use government or other guarantees to "limit the risk," via guarantee programs and cooperation with state economic development agencies. This demonstrates the importance of third-party guarantees and guarantee programs such as those of the Small Business Association (SBA) in small-business loans in the U.S. When ShoreBank first began lending to small businesses in the early 1970s, nearly 100 percent of their small-enterprise loans were term loans guaranteed by the SBA. ShoreBank actually limited its offering of other loan products, such as credit lines, to small enterprises until the mid 1990s, because of the importance of the SBA guarantee as the secondary source of repayment. ShoreBank leveraged these guarantee programs to further its outreach, and was a substantial user of the state-run Illinois Capital Access Program. Guarantee programs can thus be a useful tool inbroadening access to credit. 107. ShoreBank has also leveraged third-party guarantees from individuals and enterprises to make deals happen. The bank has also worked with state and local grant programs, or mezzanine funds, to structure multi-layered financing packages for its clients. This pulling together of multiple sources of financing helps to mitigate risk.However, these deals involve considerable coordination on the part of the lender. Also, guarantee programs, though very important in small business lending in the U.S. through the Page 158 Part 3. Downscaling Private Banks 199Os, may not work in the same manner internationally. Some of the concerns voiced by donors are that guarantees would take the place of rigorous credit analysis. Banks have expressedconcern that guarantee programs would add additional layers of bureaucracy and in the end are not worth the effort. It would be interesting to review the performance of guarantee programs, such as the USAID Development Credit Authority, outside of the U.S. Loans to Start-up Companies 108. Loans to newly established companies, or start-ups, are in a higher risk category. While the entrepreneurs may have a track record, their businesses do not. Few banks are willing to take on this risk without third-party guarantees. ShoreBank makes loans to start-ups if owners contribute a signijkant amount (30percent) to the total project, and assuming the business intrinsics make sense. Theseloans are almost always backed by SBA or other guarantees.Typically, start-ups are candidatesfor non-profit loan finds, or for venture-capital type financing through the SBA's SBIC program or other community developmentventure capitalfinds. 109. In Eastern Europe, SAS has partnered with local banks to structure financing packages to start- up businesses without guarantee programs. Erste Bank in Austria, owner of Ceska Sporitelna in Czech Republic, has established specialized branches in Austria to process loan applications to start-up companies, where skilled loan oficers with small business consulting experience connect these borrowers to training and small business support. Rejectionrates are high(80 percent or higher), but this provides an opportunity to work with promising growing companies. Erste i s planning to start such a program in the Czech Republic. Start-ups also require heightened vigilance in monitoring. In many cases, start-ups involve construction and equipment installation, and lenders should perform on-site visits at milestones identified during the analysis process, such as the delivery of equipment. Sales targets, cost projections, and margins should be monitored against projections, with financial updates on a monthly basis. Credit Lines 110. Credit lines are typically extended to small enterprises to increase working capital through their primary accounts. While they can be unsecured, they are more often secured by accounts receivable or inventory, rather than fixed assets. ShoreBank did not initially offer credit lines because of the weak collateral position and the costs of linking outstandings with specific inventories or receivables. Now, a greater percentage of ShoreBank borrowers receive credit lines. Many banks provide credit lines as part of a suite of productsfor small and micro clients. For example, Ceska Sporitelna in the Czech Republic has developed the "Profit Program," a bundle of products targeting micro and small businesses that incorporates a special direct deposit account, a credit card, and a credit line of $3,000 to $6,000. Credit lines are typically higher priced than term loans and thus are a more expensive but more convenient option for the borrower. Fees for credit lines can be higher due to increased maintenance costs for the bank. However in some environments, taking receivables or inventories as collateral may be difficult to document or enforce. This partly explains the huge differentials in interest rates of lines of credit and term loans, in Brazil as well as elsewhere. Approval procedures may be relatively easy; commercial banks often base their approvals on account balances and collateral appraisal alone.44Servicing for credit lines frequently involves additional monitoring of collateral if linked to accounts receivables or inventory levels. 44At Ceska Sporitelna, to qualify for acredit line, the customer needs to have beenabank customer for a minimum of two years (either through an account or lendingrelationshipwith the bank), andhave documentedannual sales of at least $30,000 for the last two years. ShoreBank usuallyincludes acredit line as partof a larger financingpackage, and traditional cash-flowanalysis is performedon the entire package. Page 159 Brazil: Access to Financial Services 111. Credit may also be extended through corporate or personal credit cards. Corporate credit cards are a now a basic product offered to small businesses at banks of all sizes in the U.S. In Eastern Europe and Latin America, they are being offered increasingly to small business clients as well. Personal credit cards are available for smaller loans. Finally, an overdraft entitles account holders (retail or commercial) to exceed account balances for fixed periods at a rate of interest comparable to a credit card. These loans serve as emergency funding, like credit cards, for companies experiencing temporary cash shortages. They are typically smaller in size than a credit line, and are not secured. Often, the primary consideration in qualifying for an overdraft is the average account balance. Amounts granted are linked to income or revenue levels. Again, like a credit line, this product is more costly to the borrower than traditional debt, but the processing is instantaneousif the customer qualifies for the product. 112. As in Brazil, where the `cheque especial' overdraft is popular, consumer overdrafts have been controversial as it is sometimes not immediately apparent to account holders that they have slipped into an overdraft situation and are faced with mounting fees or interest charges. As with credit lines, overdrafts are frequently approved based on a credit-scoring type of approach, or simply account balances. For example, TBC Bank in Georgia has been offering overdrafts to individuals and businesses based on individual income levels (limited by two times monthly salary), or up to $50,000 per company. Monitoring for overdrafts i s typically a matter of tracking account balances and receipt of periodic financial reporting. Because these loans are unsecured, rates are highto compensatefor this higher risk. Gold CollateralizedLending Pawn Lending Through Banks - 113. Product Features Some commercial banks in developing economies have entered into gold pawn lending to micro enterprises, traditionally a microfinance institution product. TBC Bank inGeorgia has recently launched this product to microenterprises. Servicing associated with these loans is related to the storage and security of the collateral. Loans are provided in the range of $25-$250, based on the appraisal of the collateral alone at a rate of 36 percent per annum. The benefit for the borrower is rapid disbursal. The benefit for the bank i s its ability to attract new customers to the bank that might not qualify for consumer or business loans based on income-generation capacity. The downside is the high administrative costs for the pawn lending in appraising and storing the collateral. In developed economies, this niche is occupied by pawnshops, check-cashing services, and other issuers of urgent cash. These products would be utilized by consumers with low or no credit scores, or by the undocumented. In the absence of positive credit history, these loans offer low-access individuals an alternative to traditional bank financing and an entryway into the financial system. They can be offered along with other microfinance or consumer-lending products to complete market coverage. However, this is an expensive option for the borrower, and is costly for the bank to administer. Such lending lies largely outside the banking sector in Brazil, although interest rate spreads are high. Other Productsfor Businesses 114. ShoreBank and other banks with a focus on the micro and small business segment have broadened their product offerings to better serve their customers and generate fee income as they face increased competition. B y offering bundled products and services, the bank reaches new customers and provides convenience and savings to its clients. In recent years, ShoreBank has developed several products targeted at its core small business customers. These include Small Business Checking (for businesses with a low monthly volume of deposits and checks), Sweep Accounts, and Zero Balance Checking (which enables businesses to track expenses and monitor costs by department, location, and type of expense). Page 160 Part 3. Downscaling Private Banks 115. Alpha Bank in Romania has a goal of providing at leastfour products for any type of customer, including small businesses. They believe this solidifies the relationship with the client, making it more difficult for customers to switch to another bank. One of the products Alpha Bank has marketed to employees of small businesses is the Employee Credit Card, which is essentially an overdraft credit limit for employees with direct deposit services. The use of funds i s flexible, and the limit generally increases over time basedon the employee's salary. 3. Outreach Strategies and Marketing 116. Banks embarking on a new product or entering downscaling for the first time need an outreach strategy. One approach i s to leverage partnerships with community organizations, for-profit businesses, state, national, and multilateral programs to generate wider outreach. Another is to establish collaboration with microfinance. Banks can leam the culture of the microfinance borrower from the MFI, while the MFI could receive financing from the bank and benefit from the bank's branch networks and infra~tructure.~~Fundamental to both is the recognition that success depends on attracting and retaining a solid customer base within the target market that generates profits for the bank. Those banks with a winning customer outreach strategy inthis segment will ultimately shape the market. 117. A first hurdle sometimes faced by commercialbanks is their negative market perception among potential clients. Banks, especially older, well-established banks, may be viewed as unsympathetic to the needs of small businesses,requiring excessive documentation and collateral coverage, and lacking interest in small customers. Outreach requires that the bank dismantle this image, and approaches and principles towards this are discussed below. 118. Selecting the appropriate staff to market to low-access customers i s the first critical element of outreach. Such staff working should have the desire to service this market segment, and an ability to work with clients having low levels of education and financial sophistication. Prior work experience in a small business, or through small business support services, i s advantageous.Extensive work experience in traditional commercial lending can be detrimental, given that micro and small businessfinance may strike these lenders as too risky or too small for the trouble. Specialized training in analyzing micro and small businesses can be beneficial, not only to improve the quality of credit analysis, but also to appreciate the particular characteristics of this market segment. 119. Hiring staff from the target communities helps to break down cultural barriers and ease the interaction with potential clients. For example, ShoreBank hired many African-American lenders to work with businessesin its urban "footprint" in Chicago, Cleveland, and Detroit. When traditional banks were not welcoming to minority entrepreneurs, Shorebank projected an image of commitment to the local community. In Brazil, this could mean hiring persons from those urban areas and disadvantaged communities wherefinancial access is to be expanded. 120. A high level of customer orientation among all levels of bank staff is another essential tool in serving low-access communities. This i s frequently an area of weakness at traditional commercial banks. If service issues are not addressedprior to a sales effort, new customers will be repelledby their banking experience. Customer service among low-access clients entails basic concepts such as courtesy, client- needs assessment, and service recovery in the case of complaints. In addition, customer service entails an ability to work with clients who have an innate distrust of banks, or a low level of education. Therefore, specializedtraining in both customer service and cultural sensitivity is recommended. 45As discussedinPart 2 of this study. Page 161 Brazil:Access to Financial Services 121. A fundamental structural change implemented by banks active in downscaling to low-access segments i s the creation of specialized divisions to target these segments. Many banks working with small businesses in the USA have established separate micro and small business lending departments with dedicated loan officers and resources. This allows for recruitment of appropriate staff and application of appropriate loan methodology. Large banks often choose to deliver micro and small business products throughout the bank. While this gives a national bank the ability to reach all areas where it may have branches, not all lenders may have the skills or desire to work with this segment, and some may not actively promote the products. Having a department head or internal champion responsible for micro and small businessproducts is essential to facilitate growth of these products. 122. Another example of an innovative approach to organizational structure is the establishment of specialized start-up branches by Erste Bank of Austria. B y focusing lenders with higher-level skills on analyzing and servicing start-up businesses in one location, the bank will serve this subsegment more efficiently and mitigate risk. Yet another innovation is the creation of mobile sales forces, and separating lenders from the sales forces. The dedicated mobile sales force travels to potential commercial and retail clients, including outlying rural areas. This leaves lenders more time to concentrate on loan appraisals, thus addressing the issue that good lenders are not necessarily good sales people, although many lenders have volume targets and are expected to engage in direct marketing. Similarly, client relations are better maintained by separating the collection departmentfrom lending. Banorte in Mexico has established special call centers to deal with delinquencies beyond 30 days on their small- business loan portfolio. This provides lenders with another resource should their customers have delayed repayments. The phone call from a different department allows the lender to continue to play the role of supportive partner, while collections can place additional pressure on the client if needed. 123. Moving downscaling to the retail lending department of a bank is another possibility. At Ceska Sporitelna, businesses with over $800,000 in annual sales are handled by the corporate division, and anything below that amount is retail. Staff is differentiated along these lines as well. Alpha Bank in Romania and TBC Bank in Georgia have actively been shifting their focus to retail products as competition in the banking sector has increased. Both banks have developed new retail products, such as "gift cards", special accounts for children, offering of free safe deposit box services with new accounts, etc. This retail approach helps these banks inEastern Europe reach previously unbanked populations. 124. Another outreach strategy has been the use of specializedproducts or programs to attract new customers. For example, the Profit Program developed by Ceska Sporitelna, in which a credit line of $3,000 - $6,000 i s offered to existing clients of the bank, has attracted new lending business for the bank. Many of these customers previously had a non-borrowing relationship with the bank, and did not apply for a loan in the past because the process was too onerous and bureaucratic. TBC Bank uses the gold- collateralized loan product to reach new retail customers in outlying areas and compete with informal lenders. They are also currently developing a loan product for farmers of vineyards in rural areas, which they hope will attract a new subset of customers to the bank. 125. Banks can further enhance their outreach potential by buildingpartnerships and alliances with grass roots organizations, for-profit businesses, non-profits and NGOs. ShoreBank staff has traditionally participated on boards of local non-profit organizations and church groups. ShoreBank's holding company structure lends itself to these partnerships as well. In each community in which it operates, ShoreBank has a commercial bank, non-bank loan funds, non-profit business development services, and real estate development corporations. The group of ShoreBank banks partner the non-banks, typically in cases where collateral coverage is insufficient to provide the total financing package needed by the business. These alliances allow ShoreBank to offer greater access to financing to viable businesses. ShoreBank and other commercial banks in the U.S. have also partnered with providers of financial literacy trainingandfree tax-return compilation to bring low-income account holders into the bank. Page 162 Pari 3. Downscalina Private Banks 126. Similarly TBC Bank in Georgia has provided financing to FINCA International's micro- lending activities, and has held a stake in the EBRD and IFC sponsored Microfinance Bank (which it recently sold). It started partnering with FINCA by virtue of a long-term account relationship FINCA had with TBC. TBC also saw this as an opportunity to place funds profitably. Ownership in the Microfinance Bank allowed TBC to learn more about the sector. Banorte in Mexico and Alpha Bank in Romania have partnered with car dealerships and other providers of retail goods and services to access new clients and markets. 127. Ina highly competitive banking environment, most banks find it necessary to engage indirect marketing and "cold calling" to potential customers. Direct marketingefforts may involve hosting events for entrepreneurs in partnership with the local mayor's office, sponsorship of small business training seminars, visits to trade shows and open-air markets, and other approaches. The Bank of America example above, in which employees handed out brochures and coffee to Mexicans waiting in line at the Consular section to receive their Matricula identity card, i s another type of direct marketing. Most banks also engage in some form of mass-media advertising, whether through radio spots, newspaper articles and advertisements, or sponsorship of local community activities. Advertising and marketing efforts are most successful when the product is well definedfor the target group. 128. Some banks have chosen to target low-income or micro customers through special purpose entities. If initial market research has shown that the bank image is negative, rather than expending resources to improve the image, the bank creates a new arm. For example, Banorte inMexico has chosen to launch its micro lending below $6,000 through a special purpose entity. Some of the bank infrastructure will be used, but in most cases the client will not be aware that Banorte owns the entity. The primary reason is the negative image of bank in this market as a result of the peso crisis inthe 1990s. 129. While the financial institutions may differ in location, asset size, and motivation for downscaling, all agree that commitment at the highest levels of the organization to reach these markets i s critical. Efforts needed are not sustainable to scale unless top management makes it a priority. 130. Finally, and to summarize, other key factors of success in bank downscaling include, the extent to which the product addresses an unmet demand; the rigor of the lending methodology;the quality of the staff;pricing; and the impact of the outreachstrategy. Page 163 Brazil:Access to Financial Services Page 164 Part 4. Partnering "?-Banks PART4. PARTNERINGNON-BANKS 1. INTRODUCTION 1. To what extent can financial institutions other than banks contribute to financial access for the underserved? An examination of the institutional distribution of the assets of the financial system indicates a heavy concentration in the banking system. Between 90 and 95 percent of total financial assets in the last five years are in banks, in contrast to only 3 to 4 percent in finance companies and leasing companies taken together and around 2 to3 percent in all other financial intermediaries (Table 4.1 and Appendix Table A1.3). But, these data may overemphasize the importance of the banking system. Banking system assets as representedbelow consolidate data for Brazil's multiple banks, and to the extent that departments of these banks also engage in non-bank activities, these would be reflected under bank assets.I Banks (1) 91.40 90.25 91.42 90.67 91.58 93.08 93.64 Multiple Banks Cooperatives 0.20 0.25 0.31 0.41 0.54 0.62 0.75 Finance Companies Leasing Companies Other FinancialIntermediaries (2) Notes: I / Includes multiplebanks, commercial banks, caixa economica, development banksand investment banks 21Includes housingfinance entities, securitiesmarkets brokers&dealersand foreign exchange dealers. Source: Central Bank of Brazil 2. As discussed in the previous chapter, a number of credit instruments other than `plain vanilla' bank loans are used increasingly by both individuals and enterprises.' These include particularly, overdrafts, credit cards and vehicle loans for individuals (these accounted for 11percent, 6 percent and 35 percent of outstanding loans to individuals (December 2002), and 55 percent, 10percent and 7 percent of new loans contracted respectively). Enterprise financing also makes considerable use of overdrafts (15 percent of outstanding loans and 30 percent of new loans), and though smaller in volume, the discounting of commercial invoices or `duplicatas' (4 percent of outstanding loans and 8 percent of new loans) as well as the provision of vendor credit (around 6 percent of new or outstanding credit)(Figure 4.l).3 `For example, multiple banks had 34 leasing portfolios at the end of 2001. However, according to persons from the Central Bank, if these activities are undertaken through separate subsidiaries such assets would not be included in these data. See Part 3 Downscaling Private Banks, Table 3.10. Overdrafts for individuals are referred to as `cheque especial' lending; overdrafts for enterprises are included in the category referred to as the `contagarantida'. Export notes, letters o f credit, lines o f credit and import finance are important, accounting for a significant 42 percent of outstanding enterprise financing (December 2002). Page 165 Brazil: Access to Financial Services Figure 4.1 Brazil Loansto Companiesand Individualsby Instrument (%) - - - 8. Workmg Capital Cverdrafts 6Cverdraft c Wrsonal Loans @VendorCredit hvoice & R.Notes !%count ExportCredit .Eternal Credit ~ Source:Central Bank of Brazil,appendix Tables A4.1 and A4.2 3. Some of these instruments are closely associated with and move in parallel to other forms of financing which lie outside the banking system. For example, the rapid growth of the discount of `duplicatas' (commercial invoices) by commercial banks i s significant. However the activities of factoring companies, which are the main providers of credit through the purchase of receivables, are not included among data on financial institutions as factoring is deemed to be a purely commercial activity. In addition to the discount of invoices, included in credit data, which amounted to around R$6 billion in 2001, the estimated factoring portfolio at the end of 2001 was R$27billion, and factoring representedan estimated 2.3 percent of Brazil's GDP. These figures indicate that factoring companies may have a significant role in providing access to financial services, indeed particularly for smaller enterprises which may face difficulties in obtaining conventional bank finance. The purchase price of receivables through factoring companies is higher than the interest rates applied by banks for the discounting of invoices. Thus, at the end of 2001 the nominal interest rate applied was 50.1 percent for invoices discounted at banks while the implicit average "purchase factor", real and effective, was 58.2 per~ent.~ 4. On the consumer credit side, vehicles credit and credit card loans have both shown an upward trend. These activities are also associated with leasing companies and consumer finance companies. Indeed, the clear upward trend in vehicles credit is due in part to the decrease of leasing activity over the past few years, as discussed further below. Credit cards receivables are only included in the statistics above to the extent that these may be refinanced by banks.5Finance companies' portfolio was 0.4 percent of GDP, amounting to R$4.4 billion in value, compared to an aggregate consumer credit portfolio of R$70 billion for the entire financial system.` And Brazil's outstanding leasing portfolio represented 1 The yearly factor has been calculated by annualizing the monthly factor for December 2001 provided by ANFAC (Associa@o Nacional das Sociedades de Foment0 Mercantil - Brazilian Factoring Association), of 3.9 percent, i.e., (1.039)"12=1.582. Credit card receivables are issued by special credit card companies, which are non-banking commercial activities, which finance themselves either through the securitization of their receivables or through conventional bank loans, for which receivables provide the collateral. Credit card receivables are due at the end o f 30 days and, if not paid off by the expiry date, loans are originated by the sponsoring financial institutions. `December 2001. Figures for February 2003 are R$79 billion for total consumer credit. Page 166 Part 4. Partnering Non-Banks percent of GDP at end 2001, with a portfolio of R$11.6 billi~n.~ Information about trade finance is dispersed, making it very difficult to estimate a global figure for this form of financing. Table 4.2 Brazil Finance, Leasingand FactoringCompanies(December2001) (R$ billion) - 27.4 Source:Central Bank of Brazil 5. Given the significance of these activities, the present chapter examines their contribution to financial access especially for small scale enterprises and for low income consumers, and assesses the potential for an expanded contribution.' This section suggests that the weak bankruptcy and secured transactions laws which complicate the use of collateral in Brazil create incentive for forms of financing (factoring, leasing, trade finance) other than straightforward credit, which reduce risk by securing repayments against accounts receivable, or by maintaining title, and using self-executable legal instruments. Additionally, these institutions do not face many of the high costs of entry of banks, and some non banks such as factoring and finance companies are better equipped to handle the high operating costs for small scale clients. As such, they could have an important role, although factoring and trade financing only provide working capital and cannot meet investment needs. Leasing could potentially contributeto investment finance but has been little usedfor this purpose. 6. The contributions of these sectors have been below their potential, inpart becausetheir own legal underpinnings are fragmented and sometimes challenged as to their legitimacy. Also from a practical point of view, their distinguishing characteristics are not always evident in the services they provide. Factoring does not have a law governing its services and the treatment it has received in civil litigation has sometimes been perceived as arbitrary. In both leasing and factoring, there remains confusion between these specialities and normal bank credit. Thus, in the case of leasing it is sometimes concluded that leases are essentially similar to loans; also distinctions between financial and operational leases are sometimes unclear; in the case of factoring, the discounting of invoice or pre-dated checks is considered at par with the purchase of receivables. Until a mature distinction regarding these novel contracts is established, it will be hard to expand the growth of these industries inBrazil. 7. Nevertheless there i s scope for an increase in the services these industries offer to access to financial services. Key findings are summarized below. There is a sizable real contribution of thefactoring industry to the provision of services to small enterprisestoday which could be strengthenedand legitimized to extend its outreach. 3 At present, the distinction between genuinefactoring and the discounting of invoices or checks is not suficiently clear in practice. Consolidation of the dispersed legal framework of the factoring * The principal correspondingfigures are 0.7 percentand R$8.9b for February 2003. The contributor to this chapter is Mario Guadamillas, World Bank, with inputs also provided by Professor Ricardo Leal and a team from CoppeadBusiness School, Rio de Janeiro. Page 167 Brazil: Access to Financial Services industry could contribute towards this. A proposed draft law on factoring has been pending in Congressfor years. Itsfinal passage would be beneficial to the industry. The legal framework should also clarify distinctions between recourse financing and non-recourse financing byfactors. Financing with recourse though notforbidden has not been recognized in legal jurisprudence. Lack of recourse in cases of fraudulent transactions by clients is costly for factoring companies. The tax treatment of factoring needs investigation. At present, they are subject to all financial transaction taxes as well as to commercial taxes. A case has been pending in the Federal Supreme Court to resolve this. Permittingfactoring companies wider access tofinance for example through the issue of commercial bills or debentures, as permitted elsewhere, could be explored. This could be eased if the proposed draft law is passed. Finally, intemational factoring could be eased by the relaxation of certain restrictions against holdingforeign currency accounts. Leasing today is an activity mainly for large firms, where the obsolescence of equipment and / or contracting of maintenance for operational leasing makes this an attractive option for investment financing. For individuals, leasing has been used largely by the better ofl primarily to acquire vehicles. Leasing is largely beyond the domain of the less well off segments of consumers, or small firms, and thus today of limited immediate importance in terms of expansion of access. The legal framework for leasing is fragmented andfor many purposes leasing falls under the Civil Code. As of mid 2003, a dedicated leasing law was under preparation and was due to be submitted to Congress, This will be important in terms of clarifying theframework for leasing operations. The leasing industry has fluctuated markedly in Brazil in the past decade. Regulatory arbitrage in the tax system has considerably influenced the development of the leasing industry and contributed to its boom, especially in consumer vehicle leasing, in the latter 1990s. Further changes in the tax system together with the impact of devaluation on dollar indexed loans, and also, judicial activism which tended to shelter the lessee led to its subsequent contraction and to the move away from individual leasing by lessors. The relatively slow growth in leasing of machinery and equipment has also been ascribed to the availability of alternative sources offinance (such as the FINAME credit lines of the BNDES bank), which are longer in term (3 to 5 years). Leasing has tax advantagesfor the lessor due to provisions for accelerated depreciation of leased assets at around 30percent per year. Leasing also has potential tax advantagesfor the lessee due to the write off of periodic payments against income. But, it may be difficult for smallerfirms to benefit from this without using more complex tax forms (`lucro real') which permit itemized deductions, instead of the simplified taxforms (based on `lucropresumido') The tax regime for leasing was also made more complicated by the treatment of residual value. If low, leases were sometimes not recognized but treated as loans. The industry sought legal clarification, and in 2003 the Superior Court of Justice (Superior Tribunal de JustiCa) clarified the use of residual value in leasing contracts, to include all payments, irrespective of the timing of such Page 168 Part 4. Partnerha "?-Banks payment during the life of the contract. It will considerably influence the extent to which the industry regains its attractiveness. Leasing has an accounting advantage in Brazil, asfinancial leases are booked as operational leases, thus allowing capital goods acquisition without raising the gearing (indebtedness) ratio. However, this accounting treatment is unorthodox. Overall, a review of the accounting framework of the leasing industry should be undertaken to (i) clearly distinguish between loans and lease as well as betweenfinancial and operating leases; (ii) examine the possibility of permitting write offs against income for small firms and (iii) harmonize accounting methodology with international practice. Partnerships betweenfinancial intermediaries and commercial companies could constitute afeasible alternative to promote operational leasingfor smaller enterprises. This would be promoted by more competition in the industrial segments which could constitute the lessees. Finance companies are an important alternativefor consumer credit at low and medium income levels, which have the advantagefor the consumer of easier access than bank accounts. Forfinancial institutions, they offer an avenuefor product diflerentiation as well as lower costfinds (as there are no reserve requirements). Access tofunding is tightfor thosefinance companies which operate independently, as they are not deposit taking institutions. However, most of their services could today be provided by multiple banks. 2. FACTORING 2.1 Factoring in Brazil - Evolutionand Development 8. Factoring, known in Brazil as "jomento mercantil", is a service activity which includes ongoing advisory work on credit, risk, accounting, inventory and working capital management, in tandem with the irrevocable purchase of credit rights, in the form of receivables which arise from the sale of goods or services, with a typical maturity of 30 to 60 days. The factoring company assumes the insolvency risk associated with these credit rights.' Factoring must be based on commercial sales, and is governed by Brazil's Civil Code." It can only be conducted with `legal persons' or enterprises, and not with individuals. Real estate, as in international practice, is also excluded. The most common accounts receivable are `duplicatas', which account almost for 90 percent of all factoring receivables (see below). Others are bills of exchange, bills of lading, warrants, promissory notes, checks and pre-dated checks (which can only be used if they originate from mercantile sales). Factoringas defined inBrazilconforms to internationalpractice.Brazil together with 53 other countriesaccepted the definition adoptedat the OttawaDiplomatic Conventionon InternationalFactoring(May 1988),organizedby the Institut International pour L'Unification du Droit Prive` (UNIDROIT). The definition was formally adoptedin Brazil inArticle 28 of Law 8981195, ratifiedby Resolution2144195 of the NationalMonetaryCouncil(CMN) and by article 58 of Law 9532197. loArticles 191to 220 of Brazil's CommercialCode and Articles 1,065 to 1,078 until January 11,2003, and since then by Brazil's new Civil Code, which unified civil and commercialobligations. (Articles 286 to 298. Law No. 10406, of January 10,2002). The provisionof services attachedto a factoringcontractis regulatedby Articles 594 to 609 of the New Civil Code). Page 169 Brazil: Access to Financial Services 9. There are an estimated 720 known factoring companies in Brazil, which provide services to more than 65,000 small and medium enterprises (end 2001)." Eighty percent of these enterprises belong to the industrial sector with a monthly turnover of around US$l billion. Brazilian enterprises obtained about US$10 billion (R$27 billion) as creditor rights, resulting from mercantile sales, representing 6 percent of all domestic sources of financing in 2001. This figure constitutes an impressive increase of around 50 percent over 1998 (US$7 billion; R$19 billion). Factoring companies had over 6,000 employees and were estimated to provide another 710,000 indirect employment opportunities. Thus, factoring is a sizable industry in Brazil and an important source offinance for a number offirms. CeardPiaui /Maranh8o 38 23 577 4.1 310 4,750 8 125 15.2 0.61 1.86 15.3 Total all states 717 1406 27,397 Source:Estimates based on data from ANFAC BrazilianFactoring Association; Appendix Table A4.5 - 3.8 6,186 66,070 9 92 38.2 1.96 4.43 10.7 10. Brazil's factoring companies are located mainly in the Southeast (5 1percent), South (21percent) and Northeast (17 percent), with over 250 in the state of SHo Paulo alone (Appendix Table A4.5). As shown in Table 4.3, factoring firms tend to have no more than 8 to 10 employees, in any part of the country, but the average portfolio handled per employee tends to be much larger in SHo Paulo (around R$8.5 million) than the national average (R$4.4 million) while the northeast states of Piaui, Maranhao and CearA have ratios of around R$1.9 million per employee. In terms of average numbers of firms per employees, however, the reverse applies; factoring companies in SHo Paulo handle around 8 firms per employee while in the northeast each employee deals with an average of 15 firms (or 71 firms per factorinng company in Sa0 Paulo compared to 125 in the northeast). These figures illustrate the much smaller clients that such firms have in the northeast. Innational terms, following some consolidation, two broad groups of factoring companies have emerged; large, conglomerate-affiliated companies, and smaller, independent companies which specialize in funding smaller accounts and servicing smaller clients. 11. Factoring companies serve a broad spectrum of sectors. The industrial sector accounts for 65 percent of overall activity, the commercial sector another 20 percent, and the services sector for 15 percent. The overall delinquency ratio for receivables is estimated to be around 3.8 percent although this figure can be an underestimate becausefirms use credit recovery techniques, such as discounts, to receive, at least in part, past due credit which may not be declared delinquent. 11These are the factoring companies affiliated with the National Association of Factoring (ANFAC). Since factoring i s deemed to be a non-financial activity, data on factoring companies are not readily available through institutions such as the Central Bank, which maintains data on all financial institutions. ANFAC provides a valuable service in terms of providing a focal point for the industry and its causes. '* ANFAC distinguishes three groups of members of factoring companies according to size of turnover: below R $ l m (30 percent); between R$1 and R$5 million (60 percent) and above R$5 million (10percent of all members). Page 170 Part 4. Parhering Non-Banks Figure4.2 Evolutionof the Portfolioof FactoringCompaniesAssociatedwith ANFAC Portfolio No. of Clients Thousands 8o __ 60 2o 172- 199319941995 199619971998 199920002001* 1993 1994 Source: ANFAC - BrazilianFactoring Association Figure4.3 FactoringCompanies Portfolio by Sector (2000-2001) - 2000 3 n% 15% OMetallurgicindustries WCommerce HSewlces BChemicalIndustries HPrlntlng Industry OTextile Industry .Sugar-Alco hol industries OTransportation OOther industries `Preliminary Source:ANFAC - Brazilian FactoringAssociation Pricing Interest Rates and the Purchase Factor - 12. The cost of a factoring operation has two components, a fee for the provision of services and the price, known as the "purchase factor", for the purchase of receivables. The fee for the provision of services is fixed "ad valorem". There i s an important conceptual difference between interest rates and purchase factors although they are closely associated. Interest is the remuneration of capital for the time duringwhich it is used in a credit operation. The purchase factor is the price paid for the sale of a credit right. Factoring companies use as a base the "factor ANFAC" to quote the price of factoring services in Brazil. The factor i s published daily and varies with the money market or CDB (Cert$cado de Depdsito Banca'rio).The components of the "factor ANFAC" include: the opportunity costs of resources, assessed Page 171 Brazil:Access io Financial Services in terms of the CDB rate, operational costs, taxes and expected profits. The purchase factor varies with industry conditions, and the creditworthiness of the customer. As an illustration, the purchase factor was more than twice the inter-bank lending rate in February 2003; at 4.4 percent per month (67.7 percent per year equivalent), compared to the prevailing CDB interest rate of 1.83 percent (25.7 percent per year equivalent).13 As expected, this was also higher than the prevailing average bank discount rate for `duplicatas' which was 58.5 percent per year at the same time. Factoring companies point out however that this rate i s effective and final, while banks' effective interest rates are higher as other bank charges are added to the rate, such as a credit issuance fee, a price for receivables, overheads, retention of a percentageof the loan's value, product sales etc. However, many of the factor's clients may not be able to obtain lower rates through bank discounts or any other formal financing. 13. As shown in Figure 4.4, the spread between the purchase factor and the bank discount rate for `duplicatas' showed some downward tendency over time, from mid-2000 to mid-2002, although there was some fluctuation in this trend over the past year. The spread relative to the CDB has declined more consistently. Both suggest improved competitiveness inthe factoring business. Figure4.4 Brazil Evolutionof "Purchase Factor", Interest Ratesand Spreads (%p.m.) (Jul 1994-Feb2003) - 30% spread over Duplicata ,--- % spread over CCB - ~ ~ -1I ^ " ^__I---_" 7300 250 E sa 200 150 100 -Factor -- -Duplicata CCB -I CDB [Xlplcaia Source:ANFAC - Brazilian FactoringAssociation Factoring versus Bank Services Contrastand Complementarity - 14. There are a number of respects in which factoring activities differ from the provision of bank credit. Typically, factors provide a bundle of services to their clients, and forms of factoring can vary from the most basic to more complex bundles of services. Under conventional factoring, the merchant client receives a cash payment at a discounted price for the sale of credit rights to the factoring company. Under trustee factoring, the factoring company provides the additional service of managing the merchant's cash flow. More sophisticated arrangements also exist, such as raw materiavinput factoring, where the factoring company guarantees the purchase of the product to the supplier, with first claims for its merchant. Factoring companies manage their credit risk through traditional credit information services available in Brazil, complementing traditional credit analysis with personal contact with the sellers to gain knowledge of their business practices, such as the timeliness and product quality, and their market. l4 The l3The CDB (Certifcado de Depdsito Bancario) is a security with a fixed incomewith an average termof 30 days. 14DiscussedinPart6 of this study. Page 172 Part 4. Paftnerinu "?-Banks factor must also verify the authenticity of documentation directly with the buyer to ensure that sales are legitimate protect itself against false trade documentation. 15. Thus, factors can play an important role for SMEs, not only as buyers of receivables but more importantly as supporters in strategic decision making, providers of accounting and marketingknowledge, and also, knowledge of cash flow and inventory management. The factoring company and the factor have a tight interdependence of interests - thus there are incentive compatible objectives that help overcome problems of information asymmetries which SMEs encounter with banks. Factors operate very close to their clients as they are assuming important credit risk and need effective credit information about the merchants' clients. is assumed to be a cash operation (Article 137)'' are the only credit instrument that can b They are therefore deemed to be causal sale of products and services. 16. The purchase of receivables i s normally on a non-recourse basis and without any pledge or collateral. The factor buying the merchant's credit receivables makes a credit evaluation in advance and assumes all risk of non-payment. Factoring companies purchasing receivables do not have the same rights as banks. Sellers transfer their credit risk tofactors as they buy the receivables. Banks,on the other hand, may charge the seller when the buyer does not pay as they lend to sellers and use the receivables as collateral. The only exception is when the buyer does not pay becausethe contract was not fulfilled or the receivables are fa1~e.l~ A factoring operation is thus different from a credit operation and, by law, factoring companies cannot carry out operations exclusive to financial intermediaries." One consequence is that factors typically have a close acquaintance with their clients' business and also make efforts to undertake detailed credit risk assessments than would be typical for a bank. 17. Factoring companies cannot normally receive resources from third parties but have to operate with their own resources. Thus, for example, factoring companies cannot take deposits or issue bills of exchange. Factoring companies point out that in other countries the issue of commercial paper or of debentures has beenmore readily admitted. lSWhen the buyer does not pay, the factor may file a delinquency declaration with a notary public, called a protesto, which serves as a proof of the existence o f the debt. Thus a cognizance action i s not necessary and the debt is self- executable (see Part 6 of this study). l6Laws 4595164, the basic financial institutions' law inBrazil, (Articles 17 and 18)and 7492186 (Articles 1and 16) define operations of financial intermediaries which cannot be carried out by factoring companies, including bank discounts or payment I receipt of interest. Page 173 BrazitAccess io Financial Services 18. Factoring is an important player in both the provision and use of credit information systems, as both user and provider of information. Factoring normally uses the negative credit information systems of SCI and SERASA. Factoring companies are both users and producers of information. As of May 2003, ANFAC, together with SERASA and Equifax, has had an ambitious project for the creationof a dedicated information data base on members' risks, to cover the 70,000 of factoring enterprises. Credit operations Nocredit operations but purchase of credit rights (receivables) Obtains funds from the public to invest in credit operations by Remunerationconsistsof interest rates for the use of capital Remunerationdoes not have an interest rateor discount rate but the during a periodof time. payment of a price for the sale of servicesand the purchaseof credit rights Needs a operate operates with legalpersons.-Doesnot requireauthorization from the Central Bank. Operateswith legal persons an Only operates with legal personsor professionalsthat are considered 19. As a result of these differencesfactoring has several benefits overfundingfrom bank loans. First, funding can be obtained more rapidly. Second,for the provider of credit, the exposure is limited to the amount of invoices issued to the seller and the creditworthiness of the seller is less important. Third, this allows for companies in legal diflculties to borrow from factors because factoring depends on the creditworthiness of the seller's customers. On the contrary, banks do not have an incentive to lend to firms in financial distress as this could imply a higher risk classifcation of the loan and would mean higher capital requirements. Under the prevailing bankrupcy law, unlike factors, bankshave not been permitted to operate with companies in receivership. Fourth, it is clear that factoring will have considerably higher implied rates of interest than bank loans but this can be largely explained by the composite nature of services provided, the needfor much closer client scrutiny and knowledge, and also the limited sources offinance available to thefactor, who must rely largely on internal resources. 20. Because of high implicit interest rates, factoring companies in Brazil are vulnerable to legal challenges, due to the provisions of the Usury law. This challenge to their legitimacy serves to further raise their costs of service provision and the purchase of receivables. Technically, until January 2002, interest rates were capped at 12 percent per annum under Brazil's Usury Law of 1933. Although it does not remove a cap on interest rates, the new Civil Code caps these at the rate of servicing charges on overdue Treasury paper." Although an exemption from the restrictions of this law was largely obtained for all entities which are part of the national financial system, factoring being deemed a commercial activity, falls outside the safe territory of other providers of finance. Factoring companies point out that they do not charge interest as their acquisition of receivables constitutes an outright sale. However the implicit `factor' or interest rate equivalent can be easily calculated and thus factors have often been the subject of disputes in the court system, due to their charging of `usurious' interest rates. This may also explain the small size of Brazil's factoring firms in an international comparison; Brazil's over 700 factoring companies are ahead of even the USA in numbers, which has less than 300 such entities, but the US firms have a total portfolio value around ten times as large as inBrazil (Appendix Table A4.7). 17Law No. 10406of January 10,2002. Page 174 Pari 4. Partnerina Non-Banks 21. These challenges to the legitimacy of factoring are compounded by suspected illicit activities of some firms, which have been known in particular to offer services outside their legal scope, for example the taking of deposits or provision of loans. One of the main objectives of ANFAC, the factoring company trade association, i s to delimit factoring from other activities, sometimes illegal, of financial intermediation, without the required approval of the central bank, or grant loans at usurious interest rates under the name of "factoring". ANFAC has agreed on a model of factoring contract (contrato de foment0 mercantil) to be used by its associates, has issued a Code of Ethics (Cbdigo de Etica e Auto-Regulaqlo), has approved a set of General Rules for Factoring Operations18(Regulamento Geral das Operaq6es de Foment0 Mercantil) and in 1999, reached a Technical Cooperation Agreement (Acordo de Cooperaqgo TCcnica) with the Economic Law Secretariat of the Ministry of Justice (Secretaria de Direito Econamico do MinistCrio da Justiqa). The agreement has three basic objectives: to recognize and preserve factoring as a means of supporting productive activities; to protect affiliated companies by means of a `qualification and quality' certification and; to establish a system of technical cooperation with the Federal public administration to prevent the emergence of illegal activities under the guise of factoring. ANFAC has been petitioning strongly for a specific factoring law, to help eliminate illegitimate activities and protect its' members. International Factoring and Znternational Comparisons 22. International factoring in B r a d is undeveloped, especially relative to the large size of the domestic factoring sector. The most signijkant obstacle towards the development of intemational factoring is thatfactoring companies in Brazil are not allowed to holdforeign currency accounts. Thus, they are not able to make advances to the exporter in the currency of their invoices, which introduces exchange rate risk. It would thus be difficult to serve the export community without involving commercial banks. If conditions for permitting intemational factoring could be eased, export companies could reduce their risks. Italy and the United Kingdom have extensively used factoring to involve SMEs in export activity. InLatin America, countries such as Mexico have also begun to engage in international factoring. Incontrast, inBrazilexports are carriedout by bigcompanies with more access to financial services from convential financial intermediaries. 23. Today, factoring operations are carried out in more than 50 countries with an estimated total worldwide factoring value exceeding Page 175 Brazii: Access to Financial Services France 52,450 Germany 5 1 700 15 1715 Total World Source:Factors Chain International-Annual Repott 2001 (AppendixTable A4.7). 2.2 Brazil's Factoring Regime: Regulation, Supervision and Tax Treatment 24. Brazil laws provide a framework for factoring although i t i s fragmented and sometimes lacks clarity. The question o f whether factoring is a financial or commercial activity was unclear prior to 1988. This was clarified by Circular 1359/88, which clearly recognized factoring as a commercial activity, to be regulated by commercial law.'9 The specific characteristics o f factoring call for a clear legal framework but today this is dispersed and incomplete. A draji law (230/95) which seeks to clarify the operational and legalframework for factoring has beenpending in the Congressfor approvalfor a long period.20The law was approved by a commission o f the Senate in December 2002. However, a number o f amendments were proposed after the new Congress was reconstituted in 2003 and these are still under discussion. Prospects for the supplementary vote needed in Senate to clear this are indefinite. l9A previous regulation, under Circular 703/82 had createdsome confusion stating that factoring activity had similar characteristics to financial intermediaries subject to central bank authorization. This was however clarified in 1988.Subsequent Laws and regulations (8981/95, ratified by Laws 9249/95,9430/96 9532/97; Resolution CMN 2144/95 and Declaratory Act (At0 Declaratdrio) 5 1/94 of the Secretciriada Receita Federal (SRF), endorse the osition that factoring activities are regulatedby commercial law. Presentedby Senator Jose FogaGa. InJuly 1996it was approved by the Comissdode Assuntos EconBmicosdo Senado and passedto the Comisscio de Constitucdo e Justica (CCJ) of the Senate, where it remains.Inparallel, the Congress has discussedeight draft laws directly or indirectly dealing with factoring. InNovember 2001 the Cowiissdode Economia da Ccimarados Deputados approved anotherdraft law pertaining to factoring, presentedby Deputy EmersonKapaz. This is currently seeking approval at the CCJ da Ccimara.Both projects, inSenate and Congress, are very similar although the project presentedto Congresshas beenmore recently updated. Page 176 Part 4. PartneringNon-Banks A Framework of Control for Factoring Activities? One frequent allegation against factoring companies is that they undertake functions o f financial intermediation without authorization. The law governing financial crimes (Law 7492186; "Crimes contra o Sistema Financeiro Nacional") considers criminal those actions where an institution acts as a financial intermediary without Central Bank authorization. There are also protections against this in Law 4595/64; the basic Financial Institutions Law. Despite this, "false" factoring companies are encountered, which illegally offer remuneration by means of deposit taking and bond issuance. Some factoring companies offer a menu of services which include both lending and factoring, and thus it is difficult to separate legitimate and less legitimate activities. Another allegation faced by factoring companies is that some lend at `abusive' interest rates or extort unfair fees from clients. These allegations are largely based on the old Usury law of Brazil, which capped real interest rates at 12 percent per year, and which financial institutions are not bound by. As non-financial enterprises however, the protection against such allegations for factoring companies, which lies in the purchase o f receivables rather than on the charging of interest, is sometimes questioned. Medidn Provisbria (MP) 1820/99 (reissued as MP 2172/2001) echoed the principle o f avoiding excessive interest charges. I t is not clear to what extent such activities are really illicit or extortionate. Some factoring companies are now seeking conversion to SCMs - for-profit microcredit firms (see Part 2 of this study), to avoid allegations of illicit lending activities and to obtain protections against allegations of usury. 25. The framework for oversight for factoring companies, as commercial companies, i s limited. Unfortunately, this situation has been used by some companies to undertake financial intermediation activities under the name of factoring. Draft law 230/95 besides trying to establish a specific legal framework for factoring supports the creation of a regulatory agency with limited powers. ANFAC has also agreed to enter into a Technical Cooperation Agreement (Acordo de CooperapTo Tkcnica) with the Secretaria de Direito EconGmico do Ministbrio da JustiGa, which would provide a certificate of qualification and quality to those factoring companies, affiliated with ANFAC, that comply with the Agreement. 26. Inpractice, there is no clear distinction in factoring laws in Brazil between financing without recourse and financing undertaken on a recourse basis (accounts receivable financing). In Brazil, financing on a recourse basis is undertaken through a "pro clause in factoring contracts, which allows for the repurchase of receivables by the merchant in case of debtor insolvency. In 1989, after seven years of factoring activity in Brazil, the factoring contract "contrato defoment0 mercantil" evolved from a cession of credit to a commercial contract which however included some restrictive clauses affecting recourse financing. This contract can be voluntarily altered by the parties if all legal pre- requisites are fulfilled. It was deemed in an appeal to the Conselho de Recursos do Sistema Financeiro Nacional (CRSFN, a high level appeal committee for all financial sector issues) that there is no regulation in the Brazilian legal framework which prohibits the "pro solvendo" clause.22 Tax Regime 27. To what extent does the tax regime for factoring companies differ from the framework for banks? Does this constitute any advantage or disadvantage for such ~ompanies?'~ First, factoring companies, unlike banks, do not pay interest as there are restrictions against their access to funds from third parties in the market. Thus, they cannot deductfrom their income any interest expense and taxes afsects them more 21When the operation is with recourse from the merchant it is said to be undertaken on a `prosoluto" basis. 22Some countries such as Italy and Spain have introduced the concept o f "passive solidarity" which allows for recourse financing thus diminishing the risks faced by factoring companies. 23A discussion of the various taxes on financial intermediation is available inthe documentation for Brazil's Second Financial Sector Programmatic Loan. Page 177 Brazil: Access to Financial Seruices heavily. This tax burden is passed to clients as the additional costs of factoring services. Second, as they undertake commercial transactions to purchase receivables, the tax note, invoice and `duplicata' (credit right) associated with such sales must be issued. The commercial transaction is subject to the ICMS (Impost0 de Circulagdo de Mercadorias) tax. Law 5474168 (article 20) allows service providers to issue invoices and `duplicatas' representative of their rights. Next, the service provider i s subject to the ZSS tax which is locally collected with a tax rate varying from 0.7 percent to 5 percent depending on location. Sometimes, factoring companies are asked to collect the ISS tax on overall revenue (service provision plus revenues from purchase of receivables). The Cofns tax does not take into account the dual nature of factoring activities and the tax base includes both the revenues for service provision and the revenue for receivables purchase.24Until recently, PIS and COFINS were also cascading taxes, with implied double taxation, however recent legislative modifications in 2002 and 2003 have alleviated this. 25 28. Third,factors are taxed in exactly the same way as banks in terms offinancial transaction taxes. The IOF tax on financial transactions, which was established by Law 9532197 to credit operations of financial intermediaries, includes factoring companies in its scope (Article 58), although factoring companies are not financial intermediaries. Moreover, factors complain that they should be exempted from the IOF tax. On the other hand, credit transfers ("cessdo de crkditos") between financial intermediaries or between financial intermediaries and leasing companies, are not subject to IOF (Decree 2219/97, article 8, VII), and factors do not benefit from this. These appear to be contrary to principles of equality of tax treatment. 2.3 Factoring Summary and Key Issues - 29, Factoring is an important alternative to the banking sectorfor thefinancing of small and medium enterprises. Given difficulties of entry into the formal banking system, factoring companies have found an important market niche. It is currently a key, if not the main, source of finance for SMEs. Factoring companies work at a much smaller scale than banks, inclose association with the creditor. Thus, factoring companies are able to perform a credit risk assessment based on a day to day working relationship rather than on financial statements. 30. Weak bankruptcy and secured transactions laws generate dificulties in the use of collateral in Brazil create special incentives for forms of financing such as factoring. In general, firms receive financing contingent on their future expected cash flows. Accounts receivables are already expected cash flows and, thus, a company eliminates its clients' business risks through factoring. However, factoring can thus only provide financing for working capital needs and not for long term investments. 31. Factoring enables quick but costly access tofinancing. But, alternatives such as bank credit based on the discount of invoices i s not much cheaper, and bank credit in the guise of personal or consumer overdraft facilities are more expensive. Although there are some provisions for credit through public banks such as the FINAME programs of the Banco Nacional de Desenvolvimento EconBmico e Social (BNDES) for small and mediumenterprises, and considerably smaller volumes of credit offered by public bodies such as Sewigo Brasileiro de Apoio as Micro e Pequenas Empresas (SEBRAE), these funds can only be obtained after a bureaucratic process that takes much longer than financing provided by factoring companies. If forthcoming, these other sources of finance can constitute a complement to factoring in the sense that they can be used for long-term investment purposes while factoring remains a flexible alternative for working capital financing. _ _ _ _ _ _ ~ ~ 24However, most localentities accept the principle that tax must bechargedonly on services provisionrevenuesand not on revenues from the purchaseof receivables. 25FromDecember 1, 2002, PISceased to be calculated on a cascadingbasis.COFINSceased to becalculated on a cascading basis with the passingof the new Tax Reform, ProvisionalMeasureNo. 135 of October 30,2003 (later convertedinto Law No. 10833,of December29,2003). Page 178 Part 4. Parlnering Non-Banks 32. Factoring in Brazil has to trade off adequate demand for sewices it provides against the mitigation of risks by offering recourse financing. One of the key elements of the demand of factoring services is the decrease of non-payment risk for merchants. Factoring companies are in a privileged position regardingmerchants' client credit information. But, with non-recourse financing (i.e. pro soluto), factoring companies bear all the non-payment risks.In addition, they can only finance their activities with their own resources, funds from partners or shareholders, or bank lines of credit. The combination of those elements could require some recourse financing in order to alleviate factoring companies' exposure. However, recourse financing requires a very clear and comprehensive legal and supervisory framework that is currently missing in Brazil. 33. A consolidation of the dispersed legal framework and clarification of factoring with and without recourse is needed. As mentioned before the legal framework for factoring is dispersed and consolidation efforts underway are welcome. The legal framework is now unclear with regard to distinguishingbetween factoring with or without recourse, although there are no impediments to this in the new Civil Code, and jurisprudence has been established that receivables purchase with recourse i s a legal activity. This is also normal in other countries. As indicated in the paragraph above a combination of both could be important for balancing risks and a clarification could lower costs to the client. 34. Factoring companies would also benefit greatly if it could be clarified that the Usury Code would not apply to their legitimate factoring activities. Some allegations of high interest rates charged are clearly due to the continued role played by this now outdated regulation. 35. Some structure of oversight may also help to achieve this. Currently, there i s no framework of oversight for factoring other than the self-regulatory adoption of the Code of Ethics of ANFAC. Factoring still has a low public image due to its association sometimes with illegal activity and usury (agiotagem). Some closer scrutiny by such an entity of factoring companies which undertaking illicit financial intermediation may be useful. However it must be recognized that in most countries, factoring would not be subject to the formal supervision needed for deposit taking financial sector entities, and thus care should be taken to structure oversight bearing in mindthe recognition of the contractual and commercial status of this industry. The difficult decision following this i s what form of agency could be involved. Supervision of the form undertaken by the Central Bank for financial institutions is not appropriate, nor would it be normally permitted, yet a new agency established for this purpose, even if minimal, could be very costly. 36. In order to take advantage of synergies many factoring companies are associating with microfinance institutions, or even converting to such institutions. There are several advantages to being associated with an MFI; achieving a mantle of respectability by signaling its regulated nature, access to financing, reduction of administrative costs, and use of the client's databases. Sociedades de Cre`dito ao Microempreendedor (SCMs) are supervised by the central bank as they are considered financial intermediaries, which would confer upon factoring companies the status of being `regulated'. Moreover, SCMs have access to BNDES and multilateral bank financing inaddition to their own resources. Owners of factoring companies may also be owners of SCMs and although this does not permit a direct contribution to capital, it can still help to lower shared overheads and enables the sharing of credit information and the establishment of common collection structure. As of mid 2003, a number of factoring companies were knownto have the same owners as SCMs. 37. Finally, a simplification of the tax treatmentfor factoring would contribute to diminishfactoring costs. Almost 25 percent of the "purchase factor" is due to taxes. The other components are estimated to be opportunity cost (25 percent), risk and profit (25 percent), and operational expenses including collection expenses for the remaining 25 percent. Page 179 Brazil: Access to Financial Services 3. LEASING 3.1 Origin, and Concept and Evolution 38. Leasing, known as "arrendamento mercanti?', accelerated in Brazil from the 1970s when its core legal framework was also established.26Initially, consisting largely of government leasing to finance huge civil construction projects, with a limited number of large domestic and foreign banks as lessors, it expanded and diversified in the 1980s when additional banks entered the leasing business. Private companies began to consider leasing as an alternative for long term investment financing. Leasing to individuals accelerated only at the end of the 1980sand 199Os, with the boom of vehicles leasing.27 39. Leasing i s a contract giving the right of possession and use of an asset for an specific period of time in exchange for a flow of payments that cover the asset redemption, financial costs and taxes plus the lessor return.28When a lease i s terminated, the leased asset reverts to the lessor, but the lease agreement can give the lessee the option to purchase the asset or take on a new lease based on the residual value of the asset. Leased assets can include real estate, vehicles, machinery, equipment and generally any asset able to generate returns. Leasing differs from renting because the asset is gradually redeemed, and because there is an option to buy the asset for its residual value. It is different from other forms of financing as the client receives the asset and not the funds and because duringthe duration of the lease the ownership of the asset remains with the lessor. 40. Both financial and operational leases are recognized in Brazil. Afinancial lease is designed for the use of a lessee-specifiedassetfor a long period of time while an operational lease is designed for the use of an existing asset for a short period of time.29Financial leasing transfers to the lessee all risks associated with the use of the leased asset. Most financial leases are arranged for new assets. The lessee identifies the asset, arranges for the leasing company to buy it from the manufacturer and signs the contract with the leasing company. Operational leasing has two components: the financing of the asset and the provision of services (maintenance of the equipment, technical assistance, insurance, etc). Inboth type of leases, the lessee could be a legal person or an individual. 41. The tremendous growth of leasing after 1994followed the macroeconomic success in bringing down inflation, which made it possible to enter into longer term financial contracts. The boom was sectorally reflected in increased leasing to individuals, mostly for vehicles, despite a parallel decline in leasing to the government, possibly a reflection of client privatization. Leasing volume peaked in 1998, just before the devaluation of the real, and declined thereafter backto the levelof 1995by the year 2000. 26Leasing is relatively new even inthe US. I t followed the Roosevelt's wartime Lend and Lease Act. Inthe1950s this activity expanded to the business sector, initially largely for equipment leasing. Brazil's legal framework for leasing was introduced with Law 6099174 and modified by Law 7132183. However, these define only the fiscal treatment of leasing and its oversight by the Monetary Council, respectively. Thus leasing activities are governed by the Civil Code for issues concerning property, rent, penalty or repossession. A comprehensive law on leasing i s under preparation and is to be submitted to Congress in mid 2003. 27ABEL (AssociaG6oBrasileira das Empresas de Leasing, at http://www.leasingabel.com.br/), the leasing association of Brazil, is very active in advancing the interests o f its associates, and the industry. 28Leasing payment flows inBrazil are defined to be of not less than 30 days and not more than 180 days, except in the case of leasing activity for the rural sector for which payment flows up to one year are permitted (Central Bank Resolution 2309196, Article 7 IV). 29Defined respectively in Resolutions 2309196 and 2465198 o f the National Monetary Council. Page 180 Part 4. PadnnerinaNon-Banks Payment Payments and other charges are enough for the lessor to: I) Paymentsand other leasingcharges cover the full recover the cost of the leased asset and obtain a retumon the cost of making the asset available to the lessee. funds invested; ii) freely negotiate the option price of the asset The presentvalue of payments must not be ter nt of the cost of the asset. MinimumTerm The term of the contract is calculated basedon the economic life Ninety days or lessthan 75 percent of the of the asset. The minimumterm is 24 monthsfor assets with an economic life of the asset. whichever is shorter economic lifeequal or less than 5 years and 36 months for assets with an economic lifeof over 5 years. If these terms are not respected the transactions is considered a commercial sale. Optionto Buy Forthe lessee the interest in acquiring the asset can only be Normally,the lessee does not intendto buythe executedat the end of the lease contract. The price maybe asset and there is no obligationof a residual negotiated at contract signature (ValorResidual Garantido,VRG, payment. However, the lessee may acquire the price limit. 1 to 95 percent of the asset cost); or market value at the asset at its market value (Resolutions 2309/96 end of the contract. In the latter, the residual value of the asset is and 2465/98). There is no VRG. excludedfrom the calculation base for periodic payment. Services Maintenance, technical and operating services are not provided by Maintenance, technicaland operatingservices the lessor may be providedby the lessor Option to The lessee does not have an option to cancel the contract The lessee has an option to cancel the leasinq Cancel under the conditions'establishedinthe contr&t Asset The lessor must acquir the lessee, exclusiv 42. Regulatory arbitrage in the tax system has considerably injluenced the rises and declines of the leasing industry. The boom in leasing in the late 1990s was significantly also a reflection of the increase of the financial transactions tax (IOF) inMay 1997, which rendered leasing more attractive to individuals compared to consumer credit (Crkdito Direto ao Consumidor, CDC). Consumer credit is subject to the financial transactions tax (IOF) while leasing is subject to the services provision tax (Impost0 Sobre Servicos de Qualquer Natureza, ISS). The IOFrate rose from 6 percent to 15 percent of sales value, while the ISS on leasing transactions varied between 0.25 percent to 0.5 prcent of all transactions. This unequal treatment provoked a tax-incentive expansion of the leasing activity inthe vehicles sector which however declined again due to the reestablished greater equality of tax treatment. 43. By 1999, the IOF rate declined to 1.5 percent. Meanwhile, new regulations introduced in June 2002 have tried to establish uniform rates for ISS for leasing across different areas, with a 2 percent minimum to 5 percent maximum rate.30Preference has swung once again in favor of consumer credit. Moreover, the minimum term of a leasing contract is 24 months compared to 6 months for consumer loans, which also reduces the attractiveness of leasing in periods of macroeconomic uncertainty. Besides, with direct consumer credit, the ownership and title of the vehicle remains with the consumer while in leases the title stays with the lessor. 44. The steep decline of leasing in 1999 was partly due to macroeconomic events; some leases were dollar denominated or indexed, due to the nominally lower interest rates. With the devaluation of 1999, many lessees suffered difficulties and leasing portfolios shrunk.31B y 2001, leasing in domestic currency (prC-fixados) represented almost 85 percent of total contracts. Less than 5 percent were indexed to the 30Constitutional Amendment No. 37 of June 12, 2002. 3'In terms of contract type andpricing, leasingcontracts may have fixed interestrates @r&jixados), US dollar indexedrates, or rates indexed to a reference, or benchmark domestic rate (e.g. the Taxa Referencial, TR, or Taxa Basica Financeira, TBF.The law does not include indexation to the Consumer Price Index (fndice de PreGos ao Consumidor, IPC), althoughthis too has been used in some contracts. Page 181 Brazil: Access lo Financial Services dollar (4.4 percent) while contracts indexed to domestic rates represented another 11.5 percent (Figure 4.6) ." Figure4.5 Brazil Evolutionof the Leasing Portfolioand Operations by Type of Asset (1990-2003*) - i 0 7* , 8 s % a# s' ? ? ? ? z z%,z 8z z z a3, a , * < - I , , ~,6 8, a8, a6,g2 j$ i T m~"w?n~~~aaag 0 % - N a 2 4 3 X 6 8 3 8 s 8 ' No of Contracts W Portfolio ...... rTEquipment Vehicles Others Machines and Equipment Note: *February, 2003 Source.ABEL - Brazilian LeasingAssociation Figure4.6 Brazil Types of Indexationin LeasingContracts (1998-2003*) - ---I-I Note:'Februaly, 2003 Source:ABEL - Brazilian LeasingAssociation 45. Currently, there are around 65 leasing companies or institutions with leasing portfolios in Brazil. Brazilian enterprises obtained about US$5 billion in the year 2000 from leasing companies, representing around 3 percent of domestic enterprise financing. A large proportion of the top ten leasing companies include subsidiaries or affiliates of Fiat, Ford, and Daimler Chrysler, though some computer equipment leasing companies (IBM) are also present. The ten largest leasing companies accounted for 70.6 percent 32This indexation to the long term interest rate (Taxa de Juros a Longo Prazo, TJLP)is normally used when the leasing operation is financed with BNDES resources (Finame Leasing). Finame leasing credit line is devoted to legal persons (if national shareholders do have the control of the firm) and individuals (transport and rural sector professionals) for the leasing of machines, vehicles and new equipments registered at Finame and financed by BNDES resources. Page 182 Part 4. Paltnering Non-Banks of the present value of leasing assets in February, 2003 (Appendix Table A4.6). In terms of asset type, vehicles leasing represented 57.5 percent of total leasing, machines and equipment 25.5 percent, technology and equipment 12.3 percent and others 4.7 percent by February, 2003 (Figure 4.5). As observed, there have been dramatic shifts in the asset composition of the leasing industry in the last decade, with the boomand subsequentdecline inconsumer vehicle leasing. Figure4.7 Evolutionof LeasingOperations by Sector (1993-2003*) 100% 80% 60% 40% 20% 0% 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 200Y Note: *February, 2003. Source:ABEL - Brazil LeasingAssociation 46. In terms of lessee sectors of activity, the service sector showed a big increase over individual leasing from the year 2001, increasing from 31 percent to 39 percent of activity, while individual leasing declined from 29 percent to 19 percent as vehicle purchasers began to move towards direct consumer credit (Crkdito Direto ao Consumidor, CDC) and away from leasing (Figure 4.7). 47. By international comparisons, Brazil's leasing industry had attained sizable proportions by 1999, prior to devaluation (Table 4.7). The experience of developed countries suggests a potential for the growth of leasing activity in Brazil, especially, operational leasing, or the provision of a service together with adequate maintenance for its efficient use. Effective demand for operational leasing appears to lie mainly in the chemicals, pharmaceutical, and consumer goods sectors. Table 4.7 billion, 1999) United States 183,000 Japan 63,000 Germany 37,000 United Kingdom 20,000 France 18,000 Italy 13,600 13,300 Source:Dataprovided by ABEL. Page 183 Brazii: Access to Financial Services International Leasing 48. International leasing (Le., with a foreign lessor) in Brazil, limited to airplanes in the 1980s, grew in importance in the 1 9 9 0 ~The~lessee is considered the importer of the assets if acquired from abroad, . ~ and thus is responsible for the payment of tariffs, freight and insurance and taxes. Lower nominal interest rates for the lessee have been the attraction, together with longer terms. The foreign lessor i s also exposed to `country risk'. International leasing grew in importance from 1996, facilitated by a faster approval process at the Central Bank34and a more open Balance of Payments capital account. International operational leasing has been extensively used for airplanes. It has also been used for medical equipment. Other sector that recently have been active in international leasing in Brazil are power, telecommunications and infrastructure. 49. It therefore appears that leasing is an activity mainly for largefirms, where the obsolescence of equipmentand/or contracting of maintenancefor operational leasing makes this an attractive optionfor investmentfinancing. For individuals too, leasing appears to be used largely by the better 08primarily to acquire vehicles. Thus, leasing today is therefore largely beyond the domain of the less well off segments of consumers,or smallfirms. 3.2 Legal and Supervisory Frameworkand Tax Treatment of Leasing 50. Leasing regulation is fragmented. Law 6099/74 defines the tax treatment of commercial leasing and Law 7132/83, which altered Law 6099/74, empowered the Central Bank to supervise this activity. As the industry evolved, new Resolutions in the 1990s allowed for operational leasing (Resolution 2309/96) and introduced leasing export (Leasing ExportapTo) with lessees abroad, as a form of `export' of Brazilian assets.35A recent Circular (3036/01) introduced more flexibility in leasing contracts to permit seasonal variations in periodic payments and the lessee's payment capacity. Additional regulations are detailed in Appendix Table A4.8 and the ABEL ~ e b s i t eThe . ~ ~leasing industry has prepared a draft law on leasing and expects its submission to Congress by the summer of 2003. However, this will take an estimated couple of years to be voted upon. 51. Leasing is a financial activity and is thus subject to Central Bank supervision. Lessors must be leasing companies organized as corporations or multiple banks (with a leasingportfolio) or other financial institutions authorized to operate a leasing portfolio. Leasing companies are therefore subject to entry conditions, including minimum capital requirements as established for financial institutions, as well as prudentialregulations. 52. Until recently, prudential regulations on concentration limits which imposed a 25 percent exposure limit to a specific client did not aid the commercial flexibility of this industry. If the primary client of a company is the government (as was historically true for a large part of leasing activity) this rule could be very limiting. Recent modifications to the regulations however have eased this.37 33 International leasing is regulated by Resolutions 1969192 and 2302196 of the Monetary Council; Circulars 2249192,2397193 and 2731196;and Law 9959100. 34 In 1997 the Central Bank implemented an electronic system for approval, called "Siscomex", which reduced the average term for approval from 120 days to 15/20days. 35 The contract is signed by a Brazilian lessor and a foreign lessee, just the opposite o f international leasing, in which the lessor is from abroad and the lesseeBrazilian. 36 At http:llwww.leasingabel.com.br 37 Under Resolutions 2921 and 2923 o f 2002 o f the Monetary Council, which permitted `derivative' credit operations, and related operations (Operacties vinculados), more flexible limits many be established within diversification limits permitted for each client. Page 184 Part 4. Parlnering Non-Banks Tax Regime and Accounting Treatment 53. Leasing operations may provide tax shields for enterprises because payments are treated as expenses and may be deductedfrom taxable income. However, individuals cannot deduct lease payments as operational expenses. Also, a 30 percent accelerated redemption of the leased asset to the lessee is permitted, which could aid the distribution of profit flows. Leasing activity is subject to the expected taxes; the service tax (ISS or Impost0 Sobre SewiGos),which must be collected in the location where the services were provided. This creates difficulties due to the diversity in ISS rates in different locations and for different economic activities, varying from 0.25 to 5 percent. Leasing i s also subject to the federal profit and contribution tax (ContribupTo Social Sobre o Lucro, CSSL) and to the social security tax (ContribuGcZo Financiamento da Seguridade Social, COFINS) as well as the PIS (Programa de IntegraGcZoSocial). 54. Under conventional accounting practices, leasing also has an accounting advantage in Brazil, as it does not increase afirm's indebtedness ratio. Most leases are recorded as operating leases even though most contracts are capital leases. Thus, lease payments are expensed as incurred and the lease obligation i s not recorded as a liability against capital. Goods leased are accounted for as assets of the lessor, similar to financial leasing. The lessee i s therefore able to finance the purchase of the asset without increasing its indebtedness ratio. This accounting treatment i s not in line with international standards, where leases analogous to a financed purchase of an asset are classified as capital leases, and the leased asset and the lease obligation are recognized on the balance sheeta3* 3.3 The Scope for Extensionof Leasing 55. For many products, especially those with high obsolescence, operational leasing may be more appropriate than financial leasing. Moreover, the tax treatment is more favorable, for legal persons. This i s because in financial leases the residual value i s not deductible while in operational leases all payments by the lessee are ded~ctible.~'Lesseesalso have the flexibility to terminate the contract with two months' notice. Yet, true operational leasing in Brazil is limited and currently almost all leasing contracts signed in Brazil are defactofinancial leases. This can be explained by a combination of several factors. First, most of thefinance providers in leasing contracts arefinancial intermediaries who lack the interest and expertise in the commercial aspects of leasing, in particular, pricing the residual value of leased assets.4a Financial intermediaries thus try to avoid the market risk of retaining the leasedasset. Second, the lessee's rights of termination in operational leases increases the lessor's exposure to market risk related to the residual value of the leasedasset in addition to the normal credit risk relatedto the creditworthiness of the lessee. Third, the lack of secondary markets for leased assets also impedes the development of operational leasing, and makes the pricing of assets even more difficult. Fourth, this may reflect limited competition in the manufacturing industries - in highly competitive environments commercial companies use operational leasing to help the sale of their products even if they have to undertake more risks.41Fifth, financial intermediaries can access financial resources more cheaply than commercial companies, which tends to further limit the role of real commercial companies in the business.Finally, on the demand side, a 38 International Accounting Standard 17 issued by the International Standards Accounting Committee and Financial Accounting Standards Board Statement 13 inthe USA. 39 Inoperational leasesthe present value of payments must not be greater than 90 percent of the cost of the asset. 40 Even some companies from the real sector entered in the competition for providing leasing from a financial perspective. For example, IBMstrategy inthe early 90s was to provide finance through leasing not only for IBM products but in general for IT solutions. This strategy changed in the late 90s and their leasing operations were only concentrated to facilitating the acquisition o f IBMproducts. 4' In those cases the higher risk is worth as operational leasing helps to strengthen the sales, keeps clients captive, facilitates access to secondary markets and normally comes with services provision increasing the value added of the lessor. Page 185 BraZii: Access to Financial Services cultural preference in favor of ownership is sometimes alluded to, which may also create a preference for financial leasing. 56. Partnerships between financial intermediaries and commercial companiers could constitute a feasible alternative to promote operational leasing. On the one hand, financial intermediaries have cheaper access to financial resources than commercial companies and better access to potential clients' credit information. On the other hand, commercial companies know how to price used assets, have better access to secondary markets, if they exist, or if not, they have the capacity to create them through remarketing activity and better access to potential clients.42 Increasing competition between supplying enterprises could foster the emergence of these partnerships, and promote the emergence of operational leasing, which provides comprehensive services of maintenance and technical assistance in addition to financing. In Brazil, sectors that offer financial leasing sometimes accompany these contracts with a service provision contract, if necessary. Broadbasing this partnership to include a wide spectrum of financial institutions could create the basisfor offerin operational leasing to small clients. 57. The sometimes arbitrary treatment accorded to lessors at the time of the depreciation of the real, and the court judgements in favor of lessees, also led to a loss of faith in the leasing industry and its consequent contraction. The deleterious effects of this form of `judiciary activism' are discussed further in Chapter 6 below. As mentionedabove, the depreciation of the real in 1999 led to difficulties for many lessees, as many vehicle leasing contracts were indexed to the dollar.43 Court rulings determined that these contracts were no longer pegged to the US dollar and lessees paid the rest of the lease according to arbitrarily defined dollar values, except in cases when the lessor obtained foreign resources to finance the acquisition of the asset. Some rulings of the Superior Tribunal de JustiGa overruled contractual clauses stating that leasing payments indexed to the dollar constituted an `excessive' burden for the lessee (based on the Cddigo de Defesa do Consumidor) and substituted the indexation with the Consumer Price Index (indice Nacional de Precos ao Consumidor,INPC) even though this i s not a financial index and does not interest rate evolution into account. In most cases renegotiation of the contracts with better conditions for the lessee were agreed. This inability to adequately manage risks for the lessor has been a principal cause for the current drop in individual leasing inparticular. 58. The distinction between leasing contracts and normal commercial sales is sometimes blurred. The Treasury has deemed some leasing operations to be normal commercial sales due to low residual values for the leased asset indicating that the real purpose of the operation could be tax evasion. But, this conclusion has been contested and overturned in some cases in the courts of the Primeira Turma do Superior Tribunal de Justiga and Ciimara Superior de Recursos Fiscais. Lack of clarity in treatment is another factor impeding the development of this industry. 59. Prepayment of the guaranteed residual value (VRG) sometimes has created dificulties for the repossession of the leased asset in cases where the lessees failed to pay. This has been an important problem in the vehicles sector. In Brazil, lessors in the vehicles sector have accepted traded in vehicles from the client as a down payment of the new purchase. However, until very recently leasing contracts could not include a down payment so it was considered a prepayment of the residual value and thus 42There are some examples at the international level of how important secondary market considerations are for operational leasing. Inone case, one important car manufacturer launched a new model before selling in the secondary market an important stock of the old model car creating huge losses. Inother occasions some car manufactures had a very aggressive leasing policy concentrating the maturity o f the lease contracts insuch a way that a huge car supply went at the same time to the secondary market creating an important price drop. So even experts in the markets have experienced important losses when for several reasons they have been unable to adequately price used assets coming from operational leases. 43The real went from R$1.12 for 1US$ to more than R$2 after the depreciation o f 1999,stabilizing afterwards in a R$1.70-R$1.80range. It depreciated again until R$2.55 inSeptember 2001. Page 186 Part 4. Partnering Non-Banks legally regarded as point of view as an anticipated purchase option. In these cases, the operation was not considered to be leasing but a normal financial operation for the purchase of an asset and, thus, subject to the "aZienq6o fiduciaria" process in case of repossession. This process usually takes longer that the repossessionin case of a leasing operation as the ownership of the asset remains in the lessor.44However a recent court ruling recognized the claim that leasing laws were to apply in such cases. The leasing industry is challenging this in the courts and is awaiting the decision of a pending case before the Justip which could resolve this. 4. CONSUMER FINANCEFOR INDIVIDUALS AND TRADEFINANCEFOR ENTERPRISES 4.1 Consumer Finance Companies 60. Finance companies (`tfinanceiras") are special purpose non-bank financial institutions, referredto in the regulation as Sociedades de Crkdito, Financiamento e Investimento, SCFI. Their activities are subject to regulation to direct consumer credit (Crkdito Direto ao Consumidor, CDC), most frequently used for installment purchase of consumer goods, but also used for vehicles, letters of credit, refinancing of credit and real estate. Finance companies were originally permitted a wider range of activities, including the financing of trade and to some extent construction financing. Their activities were progressively limited in scope during the 1960s and 1970s, to restrict them from taking on the more broad-based intermediation permitted to banks.45It i s not therefore surprisingthat finance companies are a relatively small segment of the financial system, and their operations have not grown significantly despite the recent boom of consumer lending (Figure 4.8). At end of 2001, finance companies had a portfolio of R$4.4 billion compared to an overall R$70 billion consumer credit portfolio for the entire financial sector. The number of finance companies in Brazil i s small, at around 40 over the past two to three years. Like factoring companies, they are mainly concentrated in the state of Siio Paulo, where over half the country's finance companies have been located. The others are distributed among nine other states, leaving seventeenstates without any finance companies (Appendix Tables A4.3). 61. The access of such companies to finance i s limited; finance companies may not accept deposits or issues `duplicatas' or promissory notes. Finance companies may however raise financing through assignment of credits (cess60 de crkdito) to banks and increasingly have obtained capital from their associated banks. Since 1988, when Brazil introduced universal banking, its bank conglomerates (buncos mu'ltiplos) have also engaged in consumer credit and many finance companies were absorbed into such banks.46This allowed diversification of their credit portfolio and permitted them to obtain financing through term deposits (which can only be issued by banks). But, many multiple banks are now directly engaged in the consumer credit portfolio, which imposes a higher regulatory burden for finance companies without the benefits of diversifying their portfolios. With the launch of the new Payments System in April 2002, finance companies claim that costs have increased further as all banks are direct participants while finance companies must use their services. 44 Inthe cases of normal financial operations subject to "alienagciofiduciaria" the repossession of the collateral is done by means of a process called "busca e apreenscio" that takes longer than the "reintegrapTo deposse" process used in the case of leasing. 45The concept of direct consumer credit was originally introduced inDecree Laws 7583145 and 9603146.A later regulation or Portaria 309159 defined finance companies' activities to include the financing of purchases and sales; issue o f `duplicatas', promissory notes and bills of exchange, working capital finance and export and import finance. But later, Central Bank Resolution 45/66 banned their issue of `duplicatas' and promissory notes. With subsequent Resolutions (103168)working capital finance was capped at 60 percent of a finance company's operations. Export and import finance was also forbidden (by Circular 29/66). 46Resolution 1524of September 21, 1988. Page 187 Brazil:Access to Financial Services Figure4.8 Brazil: FinanceCompanies Role in ConsumerCredit - 60 50 .Ea - 40 te 30 LT 20 10 0 Jan-OO Jun-00 Dec-OO Jun-01 Nav-01 Con Source: Central Bank of Brazil 62. For banks, however, it can still be an advantage to use their finance company subsidiaries for a portion of their consumer lending. First, since finance companies are not deposit taking institutions, they are not subject to the reserve requirements of banks and can thus lend more cheaply, without the burden of this implicit tax. Second, they can access clientele whom banks wouldfind it difficult to accept, given the stringent entry requirements for the opening of bank accounts. Thus, finance companies can also serve the objective of providing a diversified image and product for lower income consumers, with a dedicated brand name and marketing techniques, and diferent entry requirements. Finally, a large proportion of finance companies' CDC assets are secured by liens, under the "aliena@o fiduciaria em garantia system, and creditors may take possession of the asset and sell it in case of non payment. This has been particularly successful for automobiles although more difficult to impose on other consumer durables due to the absence of clear title and ownership registration. Such assets can therefore be sold by their owners when the execution of a guaranteei s required. 63. Finance companies have therefore lost some of their original rationale, which was clearer in the years prior to multiple banking. Their continued role is largely due to regulatory distincitions in reserve requirements (which are not required due to their non-deposit taking nature) and in entry requirements, both of which can lower lending costs, but this has to be counterbalanced by possibly increased service costs, recently, for the use of the payments system. Another role played by finance companies i s in aiding product and market differentiation to serve different market segments. The principle difficulty impeding the growth of the finance companies business i s the high costs of funding, given their lack of access to deposits, which has also acted as a disincentive to the continued survival of independent small scale finance companies. Additionally, they face highcosts originating in a highproportions of non performing which require provisioning. 64. Finally, finance companies also have to face the challenge of the persistently high informal economy. Post-dated checks are a better accepted instrument in the Brazilian market than bills of exchange, which can also be discounted at banks, which is an advantage in terms of capturing the consumer credit business. 65. Brazil's finance companies enjoy the benefit of a trade association, ACREFI, AssociapZo Nacional das Instituipjes de CrLdito, Financiamento e Investimento, established in 1958, to promote the their objectives4' ACREFI has 61 associated members including `pure' finance companies as well as those which have been transformed into or absorbed by multiple banks. 4'For more informationon ACREFI visit http://www,acrefi.com.br/. Page 188 Part 4. Parinering "?-Banks 4.2 Trade Financeand Vendor Credit 66. Vendor or supplier credit, used to finances sales, i s the least expensive form of credit with one of the lowest default rates. As indicated in Figure 4.1, vendor credit has represented a stable proportion of finance to enterprises, averaging around 5 to 6 percent. Products are sold on credit by the seller who receives cash back from the bank. The seller transfers its credit to the bank, which charges an intermediation fee, pays the seller and finances the buyer. Apart from the elimination of credit risk, the seller is able to offer a more competitive price based on reduced working capital costs and a smaller receivables account. As financing costs are excluded from the invoice price, the seller also has the advantage of lower value-added taxes. Vendor credit averaged 72 days of financing in December 2000, a term longer than the usual 30 days granted by suppliers. Transactions may be with fixed interest rates (30 to 90 days), floating interest rates (120 to 360 days) or exchange rate indexed (90 to 180 days). Vendor credit is executed through three instruments; a contract to open a line of credit between the financial institution and the seller; ii)a contract between the financial institution, the seller and the buyer; and iii) docments corresponding to each credit transaction. Collateral, such as promissory notes, are generally required. Although some companies use supplier credit as their favored way of financing working capital, this is not usual. The availability of such credit would depend on the creditworthiness of the potential creditor. 67. New forms of finance though the discounting of credit card receivables are also emerging. Although genuine credit card use i s still limited in Brazil (as most such cards are in fact debit cards or payable in full every month), retailers can discount credit card receivables in banks or at credit card companies. Receivables in this case may not be represented by certificates but by a book entry note, which may make this less secure from the point of view of legal recovery.48There is however growth potential for this form of financing. Another emerging alternative form of financing i s through franchises. In this case, the franchiser provides not only the brand name and products as common in other countries but also, finance for the retailer for terms that vary between 45 and 90 days. This is a relatively new form of financing in Brazil which began inthe last decade. 68. Finally, there are specialized channels for trade finance for export promotion incorporated into the formal credit market. Although not explored in detail here, they are briefly described in the present section. As shown in Table 3.10 and Figure 4.1, trade finance forms a relatively large part of total enterprise finance in Brazil, amounting to around a third of outstanding loans and ten to fifteen percent of new loans. In Brazil, there are two formal instruments to promote export activities; the ACC (Adiantamento de Contrato de Cdmbio) and ACE (Adiantamentode Contrato de Exporta@o). Programs associatedwith these are based on receivables expected from exports and are thus similar to export letters of credit. ACCs are extended for terms of up to 6 months while ACEs are for up to two years. These credit facilities are only open to well established businessesand especially inthe case of ACEs require the exporter to have good credits records. Thus, present forms of export financing favor established businesses, and suggest however that there may be good opportunities for international factoring to promote export activities for start-up or small firms which do not have easy access to financial services. 48See Part 6. A self-executable contract cannot be structured unless both parties sign. Page 189 Brazil: Access to Financial Services Page 190 Part 5. Channeiing Rurai Finance PART5. CHANNELINGRURALFINANCE 1. INTRODUCTION: SPECIALIZED FINANCE AND DIRECTED CREDIT INBRAZIL 1. A large part of the issue of financial access which has occupied the attention of policymakers in Brazil has been the problems of access to specialized financial services, in sectors where it has been particularly difficult to attract private finance. Two areas of the financial system which stand out in this context are rural financial services, and access to housing finance. A substantial part of the system of `targeted' or `earmarked' credit which has existed for decades in Brazil arose due to the need to address these specialized credit needs. The present chapter examines the evolution of one of these areas; rural finance, evaluates the rationale for and cost of the present system, and discusses other options for meeting these needs.' 2. It has been pointed by many that rural finance faces special difficulties and thus merits special solutions. There are first geographical issues of access in remote rural areas. These are combined with typically high rates of poverty, lower population densities, isolated markets, highly covariant risks, limited opportunities for risk diversification, sharp seasonal variations in income and in the demand for and supply of financial resources, and lack of traditional collateral. All these factors result in high transaction costs and higher perceived risk in providing financial services in rural areas2It i s argued that as a result, credit to rural markets is `rationed', in the sense that lenders tend to limit quantity, even if prices are flexible. 3. Traditionally governments have tended to address the challenges of rural financial intermediation via sectorally directed credit (quantity controls) combined with lower than market interest rates (price controls), or budgetary or off-budget support, often funneled through publicly owned banks, with the volume of agricultural credit disbursed as the key performance indi~ator.~While they have succeeded in thus channeling substantial resources to rural areas, there has been less emphasis on loan recovery and institutional sustainability, or on serving non-agricultural rural activities that can help to spread risks, or on strengthening intermediation by mobilizing rural savings. Such interventions have often resulted in ` Although this study intended at the outset to also include housing finance issues inits study o f specialized financial segments, this was omitted due to the ongoing parallel advisory work in this area, initially launched jointly with the present study. See, `Brazil Progressive and Low IncomeHousing: Alternatives for the Poor', Report No. 22032-BR, 2002, and `Housing Policy inBrazil: Maintaining Momentum for Reform', Report No. 24335-BR, 2002. Washington D.C., World Bank. I s there a case for interventions in rural finance markets due to the existence o f real market failures in such markets?This is a widely debated issue. Besley (1994) points also to imperfect information, high transaction costs, contract enforcement problems, and the need to safeguard against the monopoly powers o f some lenders, as factors which reinforce the case for intervention insuch credit markets. Even if market interventions are considered necessary (discussed further in Section 4 of this chapter, page 212 below), there is a case for increasingtheir efficiency and effectiveness according to clearly defined criteria. InMexico and India, for example, nationalized financial institutions bore a large part of the responsibility for rural lending. Inthe Philippines as well as inIndia, rural lending was also an obligation for private banks. Page 191 Ei-azil: Access to FinancialServices poor loan portfolios and institutionally unsustainable rural financial institutions leading to costly failures arising from incentive problems and poor governance. 4. Brazil's response to the challenge of rural finance has been similar to that of a number of countries; direction of credit, with a combination of quantitative controls (in the form of mandatory lending to agriculture) and price controls (lending at below market rates of interest). InBrazil as inmany other countries, selective or directed credit has been channeled largely through public financial entities, either directly or through onlending arrangements with private financial entities. Brazil's largest banks, the Banco do Brasil and the Caixa Econ8mica Federal, have served as the primary vehicles for channelling credit to the rural and housing sectors respectively. There are also mandated private sector programs for directed credit: 25 percent of unremunerated demand deposits must be devoted to rural finance, and 65 percent of passbook savings deposits are earmarked for housing loans. Directed credit shrank from mid-2000, but following the government-financed restructuring of the federal banks that removed a substantial volume of directed credit from their books in exchange for government debt. But despite the decline from about 54 percent of total credit in June 2000, to about 38 percent inMarch 2002, directed credit remains high, and the proportions going towards housing and rural credit also remain high (Figure 5.1). Both banks suffered in differing degrees from problems with their loan portfolios and less than adequate attention to operational efficiency, resulting in the portfolio quality difficulties which eventually led to their recapitalizationinJune 2001. Figure5.1 Trendsand Compositionof DirectedCredit (2000-02) R$ billion Free & Drected Credit Composition of Directed Credit 250 T 60% 160 7 0FreeCredit Drected Credit +%Drected Housing rural OBNDES gothers Source:World Bank Staff based on data from the Central Bank of Brazil. 5. In Brazil, the provision of large volumes of directed credit has been sustained over time partly due to the financing of such credit by off-budget taxes, such as the FGTS and the FAT and earmarked constitutional funds, thus representing an implicit tax on the beneficiaries of these funds.4As discussed further below, a large part of this directed credit fails to meet intended targets, however, with better-off farmers for example, capturing much of the subsidies, rather than the poorer groups for whom the loans, while the smallest 75percent of borrowers receive only 6percent of red it.^ subsidies are intended. I n agriculture the largest 2 percent of the borrowers receive 57percent of the 6. Apart from such questions regarding the appropriateness of the targeting of such directed credit, the present system of directed credit is expensive to sustain. In 2002, volumes of free credit and directed 4FGTS, or Fundo de Garantia do Tempode Servifo, is the Employees Severance and IndemnityFund.The FAT,or Fundo de Amparo ao Trabalhador, is a workers' support fund. Similarly in the case of directed credit for housing it can be shown that the principal beneficiaries are middle class homebuyers. These findings of the `capture' of selective credit by the better off is common; it has been pointed out also by other authors (e.g. Adams and Vogel, 1986). Page 192 Part 5. ChannelinaRural Finance credit amounted to around R$212 billion and R$143 billion respectively. According to one set of estimates based on simplified assumptions, assuming all directed credit i s at the TJLP rate, and assuming all free credit is at the SELIC rate (which is a lower bound), the government financial outlay involved would amount to around R$18.6 billion or around 5 percent of total credit. Even if we refine this estimate, separating the volumes of rural, housing and BNDEScredit, adjusting interest rates and adding a spread for BNDES lending, broad parameters of the subsidy element involved remain unchanged.` These are conservative estimates; an attempt to calculate the full subsidies in the PRONAF program alone estimates these at R$l.l billion (page 212 below). 7. Sections 2 to 4 describe the present system in Brazil and evaluate its efficiency. This is followed inSection 5 by adiscussion of broad approachesto widen accessbasedonnewer internationalparadigms. 8. The principal findings and conclusions of this chapter are first of all, that 3 The Brazilian rural financial system today, contrary to best practice, has limited outreach at high cost instead of mass outreach at low cost. Most rural households still have no access to formal financial services. Credit remains highly concentrated, with three-quarters of all clients receiving less than 6percent of credit while the top 2 percent of contracts accountedfor 57percent of all credit 3 Second,public sector programs and institutions play a predominant role in the national rural credit system, and these have `crowded out' the private provision of credit at market rates. Freely provided private sector resourceshave declined to as little as 5percent of formal rural credit. 3 Third, to move awayfrom thepresent situation, Brazil needs to implement a gradual but credible transition program to market determined interest rates on all rural credit and savings. While overnight removal may create at least a transitory decline in the allocation of resources to the rural sector and may not be politically acceptable, (i) there could be increases in subsidized rates on the most heavily subsidized programs (notably PRONAF, whose enormous timely repayment rebates result in highly negative nominal interest rates). (ii)Another area wherephased rates could be applied is with regard to the 25 percent directed lending requirement. (iii) While the special privileges of the Banco do Brasil, in the form of the `equalization' payments for administering subsidized funds have been ended, other financial institutions could be encouraged to compete activelyfor this role. 3 Fourth, in parallel, remaining subsidies can be explicitly monitored and budgeted and refocused awayfrom interest rates to transaction costs. Subsidieshave largely been captured by well-to- dofarmers and nonfarming corporations,for whom there is nojustification for a credit subsidy on either equity or ejiciency grounds. The net impact has been to inflate rural land prices as subsidies are capitalized into the value of land, and to promote a more regressive rural income distribution. In view of thefact that ultimately publicfundsfinance programs such as both CrediAmigo and PRONAF, there is an urgent need to apply a more uniform approach in applying lending interest rates in suchprograms. 3 Fifth, Brazil can try to address information and insurance constraints to reduce risks in rural lending, There is good potential of advancing a range of interrelated financial instruments for risk management; price risk management for agricultural commodities, area based index insurance and `Specifically,we assumethat rural credit (R$35billion) i s at an average rate o f TJLP-3 percent (since many rural credit programs are at rates below TJLP); housing credit (R$22billion) is at TR+5 percent (since rates vary from TR+3 to TR+6 percent, depending on income) and BNDES(R$85 billion) credit is at TJLP+2 percent (since BNDESchargesa spreadon most of its lending). Under this scenario, and assuming that the opportunity cost remains the SELIC rate, total government financial outlay declines little; from R$18.6billion to R$18.1billion. Page 193 Brazil: Access lo FinancialServices warehouse receipts. These instruments allow farmers to increase their specialization and adopt higher risk, higher return production strategies. 3 Sixth, in a wider context, strengthening rural finance requires attention to the same issues as those affecting overall financial intermediation: addressing issues of high spreads, high administrative costs, and relatively highprofits for banks operating in Brazil versus in comparableeconomies,combined with highfinancial sector taxation. Finally, Brazil could benefitfrom greater support to innovation and research in ruralfinance. 2. AN OVERVIEW OFRURALFINANCEINBRAZIL 2.1 The Basic Model and Volumes of Overall Credit 9. The Brazilian agricultural finance system has witnessed some significant changes in the last decade. Much of the budgetary support that was given to the sector has been removed in recent years and price supports to agricultural produce (and the resulting storage costs borne by the government) have either been eliminated or reduced substantially. However, agriculture still benefits significantly from concessional credit, through both budgetary and earmarked off-budget sources, which is substantially channeled through three federal banks, especially Banco do Brasil, as well as directed or mandated credit subject to both quantitative and price controls. This i s implemented via a 25 percent mandatory lendingrequirement on all demand deposits in the banking system, which can either be deposited with the Central Bank without remuneration or be used for lending to agriculture at controlled interest rates.7 Ceiling lending interest rates set by the Government were highly negative inreal terms prior to the Real Plan in 1995 as inflation soared above 1000percent. However recently, with an exception in 1998,rates turned positive for a brief period in the late 1990s, although they remain substantially below reference market rates such as the SELIC rate (Figure 5.2). Thus the present agricultural credit system broadly follows the `traditional' model outlined above. Figure5.2 Brazil: Nominaland Real Interest Rates in Rural Credit (%) (1995-2001) YO 45 , -15 1995 1996 1997 1998 1999 2000 2001 x CR-M- Selic * RuralCredit Rate RealRuralRate Sourcedata:Central Bank of Brazil. 'And at a minimum, a fifth of such resourcesare to be usedfor livestock and up to R$60,000 per borrower. The government points out however that limits per borrower are lower for crops generally cultivated by large producers. Page 194 Parf 5. Channeling Rural Finance 10. In terms of total volume, over time there have been sharp variations in the amount of credit provided to agriculture. Inreal terms, volumes of credit were as high as R$51.7 billion in 1975, declining to a low of below R$5 billion in 1993. Real credit rose again thereafter, to R$13.8 billion by the end of the decade for the 2000 crop season, and further to R$21 billion in 2002. Although still substantially below their peak, (Figure 5.3 and Table 5.1) a new upwardtrend may be emerging. Table 5.1 Brazil Aggregate Credit to the Rural Sector (1969-2002)(Constant2001 R$ millions) - 1970 14,796 1979 50,835 1988 11,198 1997 13,809 1971 17,206 1980 46,289 1989 1972 21,562 1981 43,170 1990 1973 30,364 1982 40,910 1 1974 35,827 1983 25,275 2 1975 51,667 1984 15,286 1993 1976 51,042 1985 21,174 1994 1977 Note: In millions of constant 2001 Reais, deflated by IGP-DIindex. Source Central Bank of Brazil: Annual Statistical Report on RuralCredit, 2001. Figure5.3 Credit Flowsto Agriculture-Reductionswith Fluctuation(1969-2002) 50 40 30 20 10 Note: In millionsof constant 2001 Reais, deflated by IGP-DIindex. Source:Central Bank of Brazil: Annual StatisticalReporton Rural Credit, 2001. 2.2 PrincipalPlayers: Demandfor Rural Finance 11. On the demand side, close to 20 percent of Brazil's population (approximately 35 million people and an estimated 4.8 million rural establishments, or farm units) constitute the client base for rural financial services. Brazil has an extremely skewed land distribution, with the wealthiest 1 percent of farmers accounting for 45 percent of landholdings and the poorest 50 percent of farmers holding just 2 percent of agricultural land (Table 5.2). Page 195 Brazil: Access to financial Services Table 5.2 Brazil: Agricultural Landholdingsare Highly Concentrated && F `1. 0 -10 ha 10- 50 ha 50 - 100- 500 ha 8.47 23.57 5 Above 1,000 ha 1.43 45.11 100.00 essel, 2001, "0Credit0 Rural no Bras//',Nota Jecnica,BCB-DEPEC. 12. There are also significant regional distinctions inthe rural clientele. The northern Amazon region has limited agricultural potential and small, scattered populations; the semi-arid north-east region has the highest concentrations of poverty, particularly among inland smallholdings producingnon-tradable crops; the central western region is characterized by large landholdings with a heavy emphasis on production of Brazil's main crop, soy for export; the south-eastem region, which includes the wealthier state of SZo Paulo with an agricultural hinterland for Latin America's largest city, as well as less wealthy states north of SZo Paulo; and the south, which includes three states that account for close to 50 percent of Brazil's agricultural GDP, and which have the longest traditions of cooperative organization inBrazil. 2.3 PrincipalSources: Credit Supply 13. The principal sources of funding for rural credit include: (a) mandatory lending of 25 percent of all demand deposits; (b) private bank resources (deposits and borrowings) freely lent to the agricultural sector; (c) the rural savings (poupanga) program of federal banks, whereby rural savings are channeled to rural credit; (d) the Workers' Support Fund (Fundo de Amparo ao Trabalhador-FAT); (e) constitutionally mandated allocations of taxes to rural credit in the northern, north-eastern and central- westem regions (Fundos Constitucionais-FCO); and (f) resources from the National Treasury.8 As of November 2001, the total outstanding loan portfolio for agriculture amounted to about R$26.7 billion, equivalent to 8 percent of the loan portfolio of the banking system, and to 33 percent of agricultural GDP, of close to R$80billion for 2000. Fiaure 5.4 Brazil: Flow of Funds underthe National Rural Credit Svstem I 1. Mandatorv bank funds 1, Bancoao Brasil 1. Non-sDecificIhes of credit 1. Individualaariclrltural 2, Non-cont;olled bank 2. Bancodo Nordeste 2. PRONAF producersoperating resources 3. Bancoda AmazBnia 3. Crop-specific or input-specific family farm 3. Constitutionalfunds 4. State Banks lines of credit 2. Large-scale farm 4 Workers' Support Fund - FAT 5. PrivateBanks enterprises 6. Credit Cooperatives 3. Membersof credit 5. Rural savings cooperatives (Poupanqa Rural) 4. Agro-enterprises 6. Treasury R Source: Bank staff. 14. There have been marked changes in the relative share of resources providedby different sources of finds (Table 5.3). In 1995, Treasury resources accounted for more than 20 percent of total credit extended but by 2000 the Treasury's share of lending had declined to zero (there was a small reemergence These funds mentioned here including the FAT and FCO, are not exclusively used for rural credit, which are a part of a range of purposes they are used for, such as rural infrastructure etc. Page 196 Pari 5. ChannelinqRural Finance in 2001 and 2002). Similarly, free resources of banks accounted for close to 20 percent of total financing of the sector in 1995 but have remained at around 5 percent between 1998 and 2002. On the other hand, the share of the banks' obligatory resources, (Le. the 25 percent of the value of demand deposit accounts that is earmarked to finance agriculture), rose from around 20 percent in 1995 to over 50 percent in 2000 and 59 percent in 2001, declining somewhat to 53 percent by 2002. This suggests that the growth of lending to the sector has been related to the growth of demand accounts in the banking sector. This is noteworthy since much of the subsidy embedded in this arrangement i s left with the financial institutions, who are able to choose the best and most risk free borrower. Thus subsidies don't necessarily reach target small producers. Fundingfrom rural saving has fluctuated between 10 and 19 percent between 1998 and 2002. The share of BNDES / Finame has increased somewhat (4.1 percent in 1998 to 7.2 percent in2002, and there has been some increasein the share of the FCO ConstitutionalFund(from a 2.7 percent share to 4.7 percent). FreelyAllocatedResources 1,710.4 591.8 522.4 581.3 651.8 722.4 1,159.4 Obligatory Resources 1,389.7 4,417.4 4,538.5 4,887.7 7,137.6 10,577.9 11,832.6 Rural Savi FAT -Workers` Fund 0.0 1,813.2 1,946.7 1,931.7 1,751.1 1,789.4 2,283.5 ConstitutionalFunds 962.1 555.0 1,039.1 914.5 806.4 1,084.4 1,544.6 NationalTreasury 2,138.0 148.1 185.4 13.7 2.6 289.1 376 6 Other Sources 641.4 752.6 239.6 373.6 77.0 78.9 Total 10,689.9 9,839.5 11,443.8 11,992.1 13,779.5 17,942.1 22,443.3 Sources:Benedito Rosado E. Santo. 2001. `Os Caminhos da AariculturaBrasileira',BM&Fand EditoraEvoluir.2001, andCentral Bankof Brazil: "Anuario statistic0do Credit0 Rural. (SeeAppendixTa6es A5.1, andA5.la to A5,le and A5.3 for furtherdetails) Figure5.5 Sources of Fundsfor AgriculturalCredit- Increased ObligatoryLending 100% Freely Allocated Resources 80% Obligatory Resources NationalTreasury 60% RuralSavings 40% FAT- Workers' Fund 20% ConstitutionalFunds 0% 13Other Sources Source:MoysesKessel: "0Credit0 Rural no Brasil', NotaTecnica,BCB-DEPEC.2001 Sources of Credit: Public Programs and Federal Banks 15. The principal financial institutions involved in rural agriculture have been the three federal banks: the Banco do Brasil, which operates nationally, and the Banco do Nordeste and Banco da Page 197 Brazil:Access to Financial Services AmazGnia. Private banks play an important role in channeling public funds and mandatory allocations of their demand deposits to agriculture, and in certain rural areas, particularly in the south, they mobilize deposits for lending in urban areas.' Credit cooperatives, which were discredited and almost disappeared entirely during the 1980s, are making a gradual come-back." Production cooperatives act as channels for loans to their members, but they are not allowed to mobilize deposits. Microfinance i s insignificant inthe rural sector, and traditional moneylenders play a minor role, although they can be important for poorer rural households particularly before the harvest. Another growing source of finance is traders' and suppliers' credit (e.g. from local merchants, fertilizer companies, wholesalers, etc.), which frequently takes the form of credit for pre-arranged sales, especially through the Bills of Rural Production (CPRs, described below). 16. Banco do Brasil has maintained its predominant role with over 4,200 branches nationwide, accounting for 61 percent of loans granted and 43 percent of the credit extended, with an average loan size of R$6,700 in 2000. Other federal and state banks together accounted for around 17 percent of total loan contracts but only 9 percent of total credit. In contrast, private banks accounted for 15 percent of contracts but 43 percent of credit extended in 2000, with an average loan size of R$28,600, about 4 times larger than the average BB loan amount (Appendix Table A5.6). Credit cooperatives, with average loans around R$7,500 each, accounted for 5 percent of loans and 6 percent of credit extended in 2000. Bancodo Brasiland Rural Credit Banco do Brasil is the largest provider of rural credit in Brazil, particularly of small farmer credit through the funded by earmarked taxes. Banco do Brasil is therefore a ke total assets of R$157 billion, a domestic distribution network ATMs and almost 90,000 staff as of September 2001, the s via open market funding. Personnel and administrative expens credit operations as of September2001, while provisionsamoun There has been a proposals to increaseprivatee Novo Mercado may not substantially impro on the board of directors, and international e include deposits from special funds. A substantial share of its depo including R$13 billion from judicial deposits. About 7 percent of Ba onlending from BNDES, FINAME,CaixaEconBmicaFederal, and other financial institutions. wrce: Central Bankof Brazil"Useful Information, 3rd Quarter, 2001",2001; staff discussions. 17. A large proportion of funding channeled by public and private banks to the rural sector is earmarked through a series of special programs (Figure 5.4 and Appendix Table A5.2). The key programs are: (i)PRONAF, the National Program for Family Agriculture, launched in 1994; (ii) a series of crop- Inrecent years privateparticipationhas beenconcededfor some earmarkedfunds, for example the Fundo de Defesa da Economia Cufeeira - FuNCAF6 (NationalMonetaryCouncilResolution2779 of October 18,2000). loSee Part2 of this study: ExpandingMicrofinance. Page 198 Part 5. Channeiina Rural Finance specific programs channeled through the second-tier National Development Bank (BNDES), for products such as coffee, milk, cashew, wine, ornamental horticulture, aquaculture, etc., and for modemization of the tractor fleet. The underlyingpurpose of some of these is to provide income support. PRONAFi s more a social than commercial program. Inconsistencyof PublicSupport: PRONAFversus CrediAmigo The PRONAF program (National Program of Strengthening Family Farming) started operations in 1995. It benefits agricultural producers who rely on family labor by allowing them to borrow cheap funds at an interest rate from 1 to 4 percent (negative in real terms in 1999).The maximum loan amounts under this program are R$5,000 for working capital and R$lj,000 for investments in fixed assets. The program grew in its volume of lendingand accounted for about 16.3 percent of total agricultural lendingin 1999, comparedto about 13.1 percent in 1998. Banco do Brasil i s its main implementing agency, extendingR$0.87 billion of PRONAF loans in 1999, benefiting about 350,000 families. Loans averaged about R$2,000 each. The plans for 2000, indicate similar amounts. The resources financing PRONAF, come from the Fuiido de Ainparo ao Trabalhador (FAT) program and the Ministry of Finance covers the difference betweenthe actual cost and the low lending interest paid by the borrowers. According to Banco do Brasil its PRONAF loan collection performance is about 97-98%. With these high repayment rates, PRONAF i s considered successful in an environment that has become used to poor financial discipline and low loanrecovery. But the low lending interestrate should be a matter of concern as nothing seems to justify lending at negative real interest rates (in 1999) when the lion's share of the farming households have neither access to formal credit nor a chance to obtain it in the foreseeable future. Setting the lendingrate higher would allow - other things remaining equal - the expansion of lending to this clientele as the accumulation of interest revenue will make the program more financially independent and less susceptible to reductions in budgetary allocationsneeded to continue financing the program- contrast to CrediAmigo, discussed in Part 2 of this study, which charges much higher interest rates. In 2002, PRONAF realized965 million contracts for a total of R$2.3 billion. PRONAF also extends loan guarantees. It i s now pioneering the introduction of rural credit through magnetic cards (for proven good borrowers). PRONAFhas now begunto accept certificates of borrowing capacity extended by third parties, such as technical assistance organizations or labor unions. Although it is true that microfinance is used with non-agricultural activities and personal loans, while PRONAF is also usedfor agriculture, current expenditures and investments,since ultimately public funds finance both these programs, there is a need to apply a more uniform approach in applying lending interest rates in such programs. Many clienrs of CrediAmigo are uoorer than PRONAF's clients. 18. Loan terms to agriculture are highly preferential with a standard rate of 8.75 percent for most lines of credit using mandatory funds (Appendix Table A5.7). Interest rates reach 10.75 percent for tractor loans over R$250,000 under the MODERFROTA program and 11.95 percent for equipment purchase or maintenance loans under CMN Resolution 2662. On the other hand, family farm loans under PRONAF range from 1to 4 percent (Table 5.4), with substantial timely payment incentives for some categories of clients, although mandatory life insurance and compensatory balances can raise the costs to farmers to 9 percent. Finally, agricultural loans in arrears prior to 1995 have typically been renegotiated at rates of 3 percent over 25 years. Page 199 Erazj? Access to FinancialServices Table 5.4 PRONAFProvidesLow-Cost Loansto FamilvFarms Investment:Group A Agrarian reform One-time only; beneficiaries with no 40% rebate for timely earlierinvestment loans R$4,000-R$12,000 1% repayment; < I O years with <5 years grace Investment: Group B Farmersand rural 40% rebate for timely workers in North-east repayment; Regionwith income