Report No. 2604-BR Brazil Human Resources Special Report Annex IV: Housing, Water Supply, and Sewerage July 13, 1979 LAC II FOR OFFICIAL USE ONLY Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CURRENCY EQUIVALENTS Currency Unit: Cruzeiro Exchange Rates Effective June 12, 1979 Selling Rate US$1.00 = Cr$25.655 US$1 million Cr$25,655,000 Cr$1 million = US$38,978 Buying Rate: US$1.00 = Cr$25.615 Average Exchange Rates 1970 1971 1972 US$1.00 Cr$ 4.593 Cr$ 5.228 Cr$ 5.934 US$1 million Cr$ 4,593,000 CR$ 5,228,000 Cr$ 5,934,000 Cr$1 million US$ 217,723 US$ 191,278 US$ 168,520 1973 1974 1975 US$1.00 Cr$ 6.126 Cr$ 6.790 Cr$ 8.129 US$1 million Cr$ 6,126,000 CR$ 6,790,000 Cr$ 8,129,000 Cr$1 million US$ 163,239 US$ 147,275 US$ 123,016 1976 1977 1978 US$1.00 Cr$ 10.675 Cr$ 14.144 Cr$ 18.047 US$1 million Cr$ 10,675,000 CR$ 14,144,000 Cr$ 18,047,000 Cr$1 million US$ 93,677 US$ 70,701 US$ 55,410,000 FOR OFFICIAL USE ONLY PREFACE This report is based on findings of a mission which was in Brazil in October-November 1977 composed of: Peter T. Knight (Chief of Mission) Ricardo Moran (Deputy Chief of Mission) Constantino Lluch (Senior Economist) Dennis Mahar (Health and Nutrition Consultant) Francisco Swett (Education Consultant) The report consists of a summary volume and four annexes. The title and principal authors are as follows: Summary Report Peter T. Knight and Ricardo Moran Annex I: Population Thomas W. Merrick and Ricardo Moran Annex II: Employment, Constantino Lluch Earnings, and Income Distribution Annex III: Health, Nutrition Peter T. Knight, and Education Dennis Mahar, and Ricardo Moran Annex IV: Housing, Water Peter T. Knight and Supply, and Sewerage Ricardo Moran Substantive contributions to this report were also made by health consultant Ernesto Calderon (Annex III); education consultants Claudio de Moura Castro, Ernesto Schiefelbein, and Francisco Swett (Annex Ill); and nutrition statistics consultant Joseph Quinn (Annex III). Joseph Quinn and Roger Bove of the U.S. Bureau of the Census contributed to the development of the Long Run Planning Model used for the demographic and other simulations and assisted Thomas W. Merrick with the demographic work. Research assistance was provided by Julie Otterbein. This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. TABLE OF CONTENTS Page No. MAP GLOSSARY OF ACRONYMS SUMMARY AND CONCLUSIONS ............................ i - viii I. LONG TERM TRENDS IN BASIC INDICATORS ...............1 A. Housing ........................................1 B. Water Supply ................................... 6 C. Sewerage ....................................... 9 II. EVOLUTION OF PUBLIC POLICIES AND PROGRAMS 1964-1977 11 A. The Situation in 1964 ...... .................... 12 B. The New System for Financing Housing and Related Urban Services: Key Institutions and Funding Sources ......... ............................... 13 1. The National Housing Bank and the Housing Finance Systems ............................ 13 2. The National Housing Bank: Basic Modes of Operation - Housing ..... ................... 20 3. The Financing System for Water and Sewerage 23 C. The Housing Finance System: Achievements, Problems and Policy Issues .......... ........... 24 1. Measures of Performance ..... ............... 24 2. Major Factors Affecting Popular Housing Finance Performance ..... ................... 33 (a) Poverty ............................... 33 (b) The Internal Financial Viability of the National Housing Bank .... ............. 36 (c) The Rising Relative Price of Housing since 1972 ...... ...................... 37 (d) Deficient Local Level Implementation Capacity .............................. 38 3. Policy Response to Factors Affecting Popular Housing Finance Performance ........ 39 (a) Changing Terms for Popular Housing Loans 40 (b) Efforts to Control Construction and Land Costs ...... ...................... 44 (c) Strengthening the Popular Housing Companies (COHABs) ..... ............... 44 (d) New Housing Subsidies ..... ............ 47 (e) The New Approach to Low Income Housing: Loans for Sites, Services and Construction Materials ....... ...................... 49 TABLE OF CONTENTS (Continued) Page No. 4. Current Housing-Specific Policy Issues ..... 55 (a) The Fiscal Benefit System .... ......... 57 (b) Increasing Incentives to COHABs to Build More Sites and Services Projects ...... 58 (c) Mixed Housing Projects .... ............ 58 (d) Improving Land Use Legislation ........ 59 (e) Reducing Construction Costs ........... 59 (f) Improving Rural Housing .... ........... 60 D. The Sanitation Finance System (SFS): Achievements, Problems, and Policy Issues .... ................ 60 1. The National Sanitation Plan (PLANASA): Background and Objectives .... .............. 60 2. Measures of PLANASA/SFS Performance ........ 62 (a) Water Supply .......................... 66 (b) Sanitary Sewerage .... ................. 72 3. Major Factors Affecting PLANASA/SFS Performance ................................ 80 (a) Possibilities for Subsidization of Low Income Consumers ..... ................. 80 (b) The Weak Financial Situation of States and Municipalities .... ................ 81 (c) Resistance by Operating Sanitation Companies ............................. 8l 4. Policy Response to Constraints Affecting PLANASA/SFS Performance .... ................ 81 5. SFS-Specific Policy Issues .... ............. 83 (a) Subsidies Benefitting the Poor ........ 83 (b) The Need for Comparative Cost Studies by Location and Technology .......... 84 (c) Rural Sanitation ...................... 86 TABLE OF CONTENTS (Continued) Page No. 111. SIKULATION EXERCISES ......................... 86 A. Type I Simulations .88 1. Method and Assumptions .89 2. Results. 91 B. Type II Simulations .98 1. Households ... 98 2. Adequate Dwelting.- Urban .99 3. Adequate Dwelffii,ns: Rural .102 4. Access to Water: UJrban .103 5. Urban Sewerage ............................. 104 6. Rural Water nld ;,ewerage .105 7. Summary .109 APPENDIX A. Housing Servi(e,-. and Per Capita Income: Historical Reu,,'arities . .111 APPENDIX B. Statistical Appitndix .118 LIST OF TABLES Table Number Page Number I Dwellings by Type of Construction and Location: 1960, 1970, and 1972. 2 2 Percent of Dwellings Classified as "Durable" by Region and Urban-Rural Location, 1972 ...... 3 3 Percentage of Dwellings with Water Supply Services Available by Region and Urban-Rural Location, Selected Years, 1950-1973 .... ....... 7 4 Percentage of Dwellings with Sanitation Facilities by Region and Urban-Rural Location, Selected Years, 1950-1973 ..... ................ 10 5 Average Annual Value of the UPC and the Rio de Janeiro Minimum Wage, 1965-1977 15 6 Liabilities of the BNH, 1966-1977 .17 7 BNH Summary Flow of Funds Statement, Selected Years, 1971-1976 ....... ....................... 18 8 BNH Summary Balance Sheets, Selected Years, 1971-1976 .......... ........................... 18 9 BNH Applications of Funds, Actual 1964-1976 and Budgeted 1977-1979 ..... ................... 21 10 Performance of the Housing Finance System by Periods in the Development of the BNH, 1964- 1977 . .......................................... 26 11 Value of BNH Housing Loans by Periods, 1964- 1977 ..........................................27 12 Average Value of BNH Housing Loans by Type of Loan and Periods, 1964-1977 .... ............ 28 13 Performance of the Housing Finance System by Regions and Periods in the Development of the BNH, 1964-1977 .30 14 Value of BNH Housing Loans by Regions, 1964-1977 .......... ........................... 31 15 Average Value of BNH Housing Loans by Type of Loan and Region, 1964-1977 ..... ............... 32 Table Number ___ ___________ Page Number 16 Monetary Income of Urban Families by Size and Region, 1972 ........ .......................... 34 17 Comparison of BNH Construction Cost Index, General Price Index, Value of the UPC, and Minimum Salary, 1970-1976 ..... ................ 38 18 Loan Terms for Final Borrowers of Housing Programs, 1969-1974 ...... ..................... 41 19 Distribution of BNH Housing Loans by Market, 1969-1974 ......... ............................ 42 20 Interest Rates Charged by BNH to its Financial Agents Under PLANHAP by Size of Loan .... ...... 45 21 Distribution of Fiscal Benefits under D.L. 1358/74 to Housing Finance System Beneficiaries, 1976 ...... ..................... 48 22 Comparison of COHAB Income from a 120 UPC Urbanized Lot and a 300 UPC House with Lot Prior to October 31, 1978 ..... ................ 56 23 PLANASA: Water Supply Objectives and Implementation as of January 1978 .... ......... 65 24 PLANASA: Number of Water Connections of the State Sanitation Companies, 1972-1977 ......... 67 25 PLANASA: Total Water Supply Investments Realized by Source and Region 1968-1976 ....... 69 26 PLANASA: Water Supply Investments Realized by Source and Year 1968-1976 ..................... 70 27 PLANASA: Estimated Investment Required to Achieve 1980 Water Supply Goals, 1977-1980, by Region and Source .71 28 PLANASA: Sanitary Sewerage Objectives and Implementation as of January 1978 .73 29 PLANASA: Number of Sewer Connections of the State Sanitation Companies, 1972-1977 75 30 PLANASA: Sanitary Sewer Investments Realized by Sources and Region, 1968-1976 .76 31 PLANASA: Sanitary Sewer Investments Realized by Source and Year, 1968-1976 .77 Table Number ___ ___ ___ ___ Page Number 32 PLANASA: Summary of Investments Realized 1968-76, 1977 and Required 1978-1980 to Achieve Goals ...................... ................... 78 33 Adequate Dwelling Coverage Ratios: Estimates, 1970; Simulation Results, 1980-2000 ........... 92 34 Access to Water Coverage Ratios: Estimates, 1970 and 1973; Simulation Results, 1980-2000 .. 94 35 Sewerage Coverage Ratios: Estimates, 1970; Simulation Results, 1980-2000 .... ............. 96 36 Households by Urban and Rural Location: Estimates, 1970; Assumptions, 1980-2000 ....... 97 37 Adequate Urban Dwellings, Type II Simulations; Average Annual Public Sector Outlays by Decade, 1970-2000 ..................................... 101 38 Adequate Rural Dwellings, Type II Simulations; Average Annual Public Sector Outlays by Decade, 1970-2000 ..................................... 102 39 Urban Access to Water, Type II Simulations; Average Annual Public Sector Outlays by Decade, 1970-2000 ..................................... 104 40 Urban Sewerage Type II Simulations; Average Annual Public Sector Outlays by Decade, 1970-2000 .... 105 41 Rural Water Type II Simulations; Average Annual Public Sector Outlays by Decade, 1970-2000 .... 107 42 Rural Sanitation Type II Simulations; Average Annual Public Sector Outlays by Decade, 1970-2000 ..................................... 108 43 Type II Simulations: Summary of Public Sector Outlays as Percent of GDP by Decades, 1970-2000 109 Table Number ____ Page Number APPENDIX A A.1 Cross Section Estimates of the Relationship Between Income per Capita and Housing Services Indicators at the State Level, 1970 .... ....... 113 A.2 Observed and Predicted Values of Certain Housing Indicators by Region ..... ............. 115 A.3 Population, Income and per capita Income by Region, 1960 and 1970 ...... ................... 116 A.4 Permanent Dwellings and Dwellings with Certain Housing Services, by Region and Location, 1970 117 APPENDIX B B.1 All Non-Institutional Dwellings and Those Classified as "Durable" by Region and Urban- Rural Location, 1972 .119 B.2 Dwellings with Water Supply Services by Region and Urban-Rural Location, Selected Years, 1950- 1973 .120 B.3 Permanent Dwellings by Region and Urban-Rural Location, Selected Years: 1950-1973 .121 B.4 Dwellings with Sanitation Services by Region and Urban-Rural Location, Selected Years, 1950-1973 .122 B.5 Housing Services Unit Cost Assumptions for Simulations .123 B.6 Type II Simulations: Summary of Average Annual Public Sector Outlays by Service and Decade, 1971-2000 .124 .BRD 13912 GUYANA BRAZ I L i wEsEZ;EtA rj 5 s S POPULATION DENSITY VENEZUELA R / PRNCOl. o'~ \ / SUR INAM ' \GU0ANA 1,INHABITANTS /KM' CL sfi~ B.. RoV, iN M.lh,le bFfEh-N _ o 100 COLOM5IA 50/ 1oo \. L,' A M AP \. 1 ; ~ '/ ~ 'i 03\n;'2 5 - 5 0 R 0 ,,, \RORAIMA 5-10 ~~~K~~~-.- ~~~~~~ 1 ) AOooei2~~~~~~~~~L-- y. a o - k V I A R M A z r C ..~~~~~7 , - 00< '0 t 0s J t3 7 2 GLOSSARY OF ACRONYMS AID - United States Agency for International Development BNH - Banco Nacional de Habitacao (National Housing Bank) CEF - Caixa Economica Federal (Federal Savings Bank) COHAB - Companhia de Habitacao Popular (Popular Housing Company) COS - Carteira de Operacoes Sociais (Social Operations Portfolio) CPH - Carteira de Programas Habitacionais (Housing Programs Portfolio) FAE - Fundo de Financiamento para Agua e Esgotos (Water and Sewerage Fund) FICAM - Financiamento para Construcao, Ampliacao, ou Melhoria de Habitacao de Interese Social (Financing for Construction, Expansion or Improvement of Social Interest Housing) FISANE - Fundo de Financiamento para Saneamento (Sanitation Finance Fund) FGTS - Fundo de Garantia do Tempo de Servico (Time-on-Job-Guarantee Fund) FUNDHAP - Fundo Estadual de Habitacao Popular (State Popular Housing Fund) IBGE - Fundacao Instituto Brasileiro de Geographia e Estatistica. (Brazilian Institute of Geography and Statistics Foundation) IDB - Inter-American Development Bank OPEC - Organization of Petroleum Exporting Countries ORTN - Obriagacao Reajustavel do Tesouro Nacional (Indexed Treasury Bond) PLANASA - Plano Nacional de Saneamento (National Water Supply and Sewerage Plan) GLOSSARY OF ACRONYMS (Continued) PLANHAP - Plano Nacional de Habitacao Popular (National Popular Housing Plan) PNAD - Pesquisa Nacional por Amostra de Domicilios (National Household Sample Survey) PRODEPO - Programa de Apoio ao Desenvolvimento de Polos Economicos (Program of Support to Development of Economic Poles) PROFICO - Programa de Financiamento e Complementacao da Habitacao (Housing Complement Finance Program) PROFILURB - Programa de Financiamento de Lotes Urbanizados (Sites and Services Finance Program) RECON - Refinanciamento ou Financiamento ao Consumidor de Materiais de Construcao (Program for Financing Construction Materials by Consumers) SABESP - Companhia de Saneamento Basico do Estado de Sao Paulo (Sao Paulo State Sanitation Company) SAC - Sistema de Amortizacoes Constantes (Constant Amortization System) SBPE - Sistema Brasileiro de Poupanca e Emprestimo (Brazilian Savings and Loan System) SFH - Sistema Financeiro da Habitacao (Housing Finance System) SFS - Sistema Financeiro do Saneamento (Sanitation Financing System) SIC - Service - Income Coefficient TAC - Taxa de Apoio Comunitario (Community Support Rate) TCA - Taxa de Cobranca e Administracao (Administration and Collection Rate) UPC - Unidade Padrao de Capital (Standard Capital Unit) SUMMARY AND CONCLUSIONS Historical Evolution of Basic Indicators i. Brazil's very rapid population growth, particularly since the 1950s, has led to mounting demands for housing and for essential residential services such as potable water and sanitary waste disposal. This has been especially acute in urban areas, where massive in-migration has added to a high rate of natural increase of the urban population. Between 1950 and 1970 the number of urban households increased 2.8 times, and the current annual growth rate is estimated at over 4%, which means that each year over half a million new urban families add to the demand for housing and essential residential services. Supply responses for the provision of these services have been, by and large, quite strong. Measured in terms of proportional coverage of these services, the situation has definitely improved since 1950, yet serious qualitative and quantitative deficiencies remain. Squatter settlements and illegal subdivisions, where self-help housing is the rule, are common phenomena, especially on the peripheries of Brazilian cities. ii. Deficiencies in essential residential services, mainly water and sewerage, as well as inadequate location and overcrowding rather than lack of shelter as such are the key housing problems in Brazil. Available data for recent years show that a large majority of families live in residential structures characterized in census and survey data as "durable". However, this minimum shelter standard (of durability) fails to reflect the precarious location (on steep hillsides, in areas subject to periodic flooding, or subject to hazardous environmental pollution) and overcrowding of many dwell- ings as well as their lack of essential services. Given the modesty of this standard, those families living in structures which fail to classify as "durable" (15% of urban and 42% of rural dwellings in 1970) must make do with very inadequate living conditions. iii. The lack of readily accessible water and adequate sanitation facil- ities affects very substantial proportions of the population. These inade- quacies not only have strong negative effects on the quality of life of the affected people, but more importantly threaten their health and survival. As in the case of most socioeconomic indicators, pronounced regional and rural-urban differentials in the availability of water supply and sanitation services in Brazilian homes have prevailed throughout past decades and remain to the present. But progress has been quite rapid since 1950, and generally most rapid in those regions (the Northeast and the Frontier) and locations (rural as opposed to urban) that were lagging farther behind in terms of these services at that time. The percentage of urban dwellings with piped water rose from almost 40% in 1950 to 53% in 1970; and in rural areas, from less than 2% to 6% respectively. However, the absolute number of dwellings without this service also rose during this period, more than doubling in urban areas and increasing by 11% in rural locations. Coverage ratios for all dwellings with sewerage network or septic tank connections increased from 24% in 1960 to 27% in 1970. The absolute number of dwellings without such services increased - ii - by almost 90% during this decade. Compared to other countries in the same per capita income class, Brazil's standing with respect to the water supply indicator was below the average in 1970, reflecting its very low level around 1950. Since 1970 Brazil's progress in water supply has been quite rapid as the National Sanitation Plan has been implemented, but comparable statift.fc are not available, pending the publication of the 1980 census. Establishment and Expansion of the National Housing Bank (BNH) iv. The rapid progress of urbanization and industrialization experienced by Brazil in the post World War II period, coupled with an inflationary spiral in the early 1960s, brought Brazil's housing finance institutions as well as those serving public utilities such as water supply and sewerage to a state of virtual financial collapse by 1964. In August of that year the National Housing Bank (Banco Nacional de Habitacao - BNH) was established to orient and control the Housing Finance System (Sistema Financeiro da Habitacao - SFH), which was created at the same time. The BNH is a public enterprise under the Ministry of the Interior, but with financial and administrative autonomy, subject to the control of the Central Bank of Brazil. The SFH's principal objective was defined to be "to promote the construction and acquisition of owner-occupied homes, especially for the lower income classes." v. Since 1964 the SFH has grown rapidly in complexity as well as scope and volume of operations, to the point that, in addition to the BNH and a system of 22 Popular Housing Companies (COHABs), in 1976 it also included the world's fourth largest savings and loan system (Sistema Brasileiro de Poupanca e Emprestimo - SBPE) composed of 81 savings and real estate credit agencies. The BNH has expanded its activities to include all kinds of urban infrastruc- ture, but especially water supply and sewerage after the establishment of the Sanitation Finance System (Sistema Financiero do Saneamento - SFS) in 1968. Today the BNH is better characterized as an urban development bank than a housing bank, though its activities in urban infrastructure, including mass transportation, can be related to housing services, broadly conceived. The BNH, in addition to being an investment bank, operating exclusively through financial agents and promoters, serves as a central financial institution controlling the diverse entities making up the SBPE. By 1976, 14.2% of the BNH's US$2.5 billion in commitments were for water supply, sewerage, and drainage, and the percentage budgeted for these services in 1977-79 was 18.5%. In 1976 "housing and its complements" (including urbanized lots and construc- tion materials loans) accounted for 48.5% of total BNH commitments; this proportion was budgeted to rise to 61.9% in 1977-79 with a greater emphasis on popular housing, including an expanded sites and services program, and less investment in mass transportation and other urban infrastructure with the exception of water supply and sewerage. Total BNH commitments in 1976 were the equivalent of about 1.7% of GDP. The increase in the value of outstanding loans of the SFH in the same year was the equivalent of 3.1% of GDP or 11.4% of the country's gross fixed capital formation. - iii - vi. The institution of monetary correction (designed to maintain the real value of BNH, SFH, and SFS assets and liabilities in an inflationary environment) has been a key factor in establishing viable long-term capital markets serving urban development activities. The other major factor under- lying the expansion of the BNH's activities was the establishment of the Time- on-Job Guarantee Fund (Fundo de Garantia do Tempo de Servico - FGTS) in September 1967. The FGTS, a form of forced savings financed by an 8% tax on payrolls, is deposited in the BNH (generally in the form of individual accounts bearing 3% interest and monetary correction) and can be withdrawn by workers in the formal sector of the economy only under specified conditions, including involuntary unemployment, purchase of housing, sickness in a worker's family, retirement, etc. At the end of 1977 FGTS deposits were 66.9% of total BNH liabilities. The Housing Finance System (SFH) vii. Over the period 1964-77 the SFH made a total of 1.8 million loans for finished housing, urbanized lots, and construction materials. Of these, 30% were for "popular" housing. The definition of popular housing loans has been broadened over time, and since 1975 has included loans up to about US$6,650 (average 1977 values and exchange rates) for families with incomes up to five minimum wages. Assuming all housing loans resulted in a new housing unit, the total of all such units associated with total housing loans through the SFH would have amounted to 25% of the increase in the number of urban households over the period 1964-77. But an undetermined number of loans were for upgrading existing housing, and the increase in urban families understates potential demand by failing to include substandard units which should be replaced. Therefore, the measure given overstates the extent to which the SFH satisfied the potential demand for housing. The performance of the SFH has varied both as regards the percentage of popular housing loans and the total number of loans per year, with 1973-74 clearly being a low period for popular housing, and the years 1975-77 showing a marked improvement both for total housing loans and especially for pbpular housing, though the expanded definition of the latter coincides with the beginning of this period. Throughout the period 1964-77 there was a marked tendency for both total and popular housing goals to exceed achievements. viii. Four basic institutional and economic factors underlie the relatively poor performance of the BNH in financing popular housing. They are (a) Brazil's high proportion of poor families, which means that a majority of urban families have not been able to afford the kind of popular housing financed by the BNH, at least until the advent in 1975 of a sites-and-services program; (b) the fact that the BNH does not receive any major funds for which it does not have to pay monetary correction plus at least 3% interest means that there are only limited possibilities for direct BNH subsidization of low income housing by charging lower than average interest rates to low income borrowers; (c) land and construction costs have risen considerably faster than the general price index or the minimum wage, thus raising the real cost of housing; and (d) at least until 1973 many COHABs were not financially viable and did not have the technical capacity to carry out large scale operations. ix. Government housing policy, as set forth in planning documents and executed largely through the BNH, has not been unresponsive to the problems mentioned. In particular, major evaluations of popular housing programs were undertaken in 1972 and again in 1974 and a wide range of corrective measures - iv - were taken. In general the policy responses of the Brazilian housing autho- rities designed to bring some form of officially financed housing within financial reach of the low income population fall into five basic categories: (a) changing the terms of loans (maximum loan size, interest rates, maximum family income, repayment systems, and system of monetary correction); (b) attempts to monitor and influence construction and land costs; (c) intro- duction of subsidies from outside the BNH-FGTS system financed by general government revenues; (d) introduction of new programs to support self-help housing construction; and (e) strengthening of the COHABs, both financially and technically. In 1975 these measures combined to produce a rapid expan- sion in popular housing loans by the BNH and the expansion has continued since that year. x. While some measures, such as increasing the maximum size of loan and family income qualifying for "popular" housing loans, did not actually make housing finance more accessible to the poor, others--especially a new system of Treasury-financed subsidies, the sites-and-services program, and a new program for financing construction materials acquisition by low income borrowers--have had this effect. Together they represent a new approach to low income housing emphasizing the provision of basic water, sewerage, and electricity services to prepared housing sites on which self-help housing can be constructed. This new approach recognizes that traditional finished "popular" housing is, in the vast majority of cases, priced out of reach of families with incomes of two minimum wages or less, 1/ that the provision of basic water and sanitation services should have first priority on available resources, that the majority of the urban poor are already involved in self- help housing construction without any assistance from official sources, and that it is better to upgrade existing settlements through loans for construc- tion materials than to demolish them. x. Despite these improvements, and without considering broader questions such as general urban development policy, employment generation, and income distribution, there are a number of measures which could be taken to further strengthen the popular housing program. They include increasing the redistri- butive character of the fiscal support system for housing, strengthening incentives to the COHABs to produce sites and services projects, developing a broader variety of housing options within housing projects at the same time opening up greater possibilities for direct cross-subsidization within such projects, improving land use and taxation legislation to reduce urban land prices and speculation, reducing construction costs, and expanding the popular housing loan system toward smaller cities and rural areas. The Sanitation Finance System (SFS) xii. The National Sanitation Plan (Plano Nacional de Saneamento - PLANASA), which sets national objectives for water supply and sanitary sewer- age, was formulated and execution begun by the BNH in 1971. Already in 1968 BNH began to finance water supply systems through the SFS and the first BNH investments in sanitary sewer systems took place in 1970. PLANASA grew out of an analysis of the deficiencies of the previous arrangements 1/ The equivalent of about US$1700 per year in the Rio and Sao Paulo metro- politan areas in calendar 1977 at the average exchange rate for that year, assuming 24 monthly minimum wages. - v - for planning, financing and executing water supply and sanitary sewerage systems in Brazil. Among the problems which PLANASA sought to attack were a chronic lack of financial resources; application of resources without recovery; uncoordinated action and multiplicity of municipal, state, and federal agencies involved in these activities; absence of overall planning; and lack of an adequate and realistic tariff system. xiii. The SFS is oriented and controlled by the BNH, which has financed just under half of its investment program over the years 1968-76. The remainder comes from state and municipal governments either directly as grants (a fundo perdido) or from the state governments through the State Water and Sewerage Funds (Fundos de Agua e Esgotos - FAEs) which together with the BNH make loans for water supply and sanitary sewerage investments to state sanitation companies (one per state) that provide water and sewerage services at the metropolitan and municipal levels. xiv. PLANASA's long-term objectives include the elimination of the deficit in the basic sanitation sector and creation of new capacity to meet growing demand once the deficit is met. Its financial structure is designed so as to eventually become self-sustaining. PLANASA seeks to realize economies of scale through stimulating the expansion of state-wide sanitation companies. Cross-subsidization takes place between states only through the interest rates charged on SFS loans. Within states, it may occur through the tariff structure. Current short-term goals include providing safe drinking water to at least 80% of the urban population in at least 80% of Brazilian muncipalities and all nine metropolitan regions by 1980 and the provision of adequate sanitary sewerage systems by 1980 to the metropolitan areas, state capitals, and cities over 50,000. While statistics on PLANASA implementation present serious problems of interpretation, the water supply goal may be within reach if a vigorous effort is maintained through 1980. This would be an impressive achievement. Sewerage investments are more complex and costly and have not received the same priority as water supply in implementation. xv. As of January 1978, BNH statistics show that 1842 municipalities (58.2% of the 1980 goal) have joined PLANASA, which generally means transfer- ring decisions over operations, investment, and rate setting to the state sanitation companies. Investments already financed would provide enough capacity to connect 71.4% of the 1980 target population to general network water systems. In contrast, only 129 municipalities had received the benefits of PLANASA's sewerage program by January 1978 and financing approved as of that date would allow construction of enough capacity to connect 55.9% of the 1980 target population, assuming no major cost overruns. BNH officials warn that translating capacity into effective connections is more problematic in the case of sewerage than for water supply. In neither case is the process automatic, and therefore, both felt needs of the population and ability to pay the rates required may limit effective capacity utilization by the poor. The total investments required during the period 1978-80 to mee.t the 1980 goals for both water supply and sewerage according to SFS officials may not seem high as a percent of GDP--less than 0.4% assuming 6% GDP growth over this period (historically, investment in water and sewerage peaked at 0.4% of GDP during 1977). However, the amount is equal to 73% of total PLANASA investments for water supply and sewerage over the period 1968-77. - vi - xvi. All the major factors which affect the performance of the popular housing finance system also affect PLANASA/SFS performance. However, there are some differences inherent in the collective nature of water and sewer systems and the financing mechanisms of PLANASA which merit comment. Under PLANASA guidelines, the combined minimum water and sewerage tariffs for residential connections cannot exceed 7% of the regional minimum wage. For water alone the minimum tariff cannot exceed 5% of the regional minimum wage. Tariff setting may pose a dilemma to the state sanitation companies. On one hand they are supposed to recover their costs and exercise strict finan- cial discipline. On the other, social criteria, including public health considerations, suggest they expand their service to the poorest strata of the urban population. xvii. Assuming fixed terms for BNH and FAE loans to sanitation companies, there are in essence two ways in which the rate constraints may be met if a system appears financially unviable. The first is cross-subsidization through the tariff structure at any level of the system (state, metropolitan, or municipal) controlled by the state sanitation companies. In general, this method works in the richer, more industrialized states where there are enough industrial and high income residential customers to allow very low minimum tariffs for the poor. But in poorer states, particularly those with large numbers of relatively small cities and towns, tariffs that are affordable by many of the poorer communities may be incompatible with an overall tariff structure that would allow the companies' financial viability. The other option is the application of budgetary resources on a grant basis to subsidize the sanitation companies at any level. Both are possible within the PLANASA framework, and greater use of these mechanisms is likely to be necessary, especially if rates are to be affordable for families with incomes below two minimum wages. A method similar to that used in the electric power sector for transferring financial resources between rich and poor states might allow at least partial resolution of this problem. There, a special fund is used to subsidize the poorer state power companies and a federal regulatory agency carefully monitors physical performance indicators to assure efficient use of the funds supplied. If no transfer mechanism within the sanitation sector can be developed, the financial situation of states and municipal governments, which deteriorated over the 1970s, suggests that the poorer states and municipalities are likely to require more assistance than they are already getting through Brazil's revenue sharing system and the SFS if PLANASA's ambitious goals are to be achieved. xviii. Government's basic sanitation policy has been expressed in the original PLANASA documents, in important revisions made in PLANASA in April 1975, and in the quantity of resources made available for basic sanitation investments--largely through the SFS. The principal changes in response to the constraints referred to include changing BNH loan terms through the SFS so as to provide greater subsidization of poorer states, and establishing in 1977 a system to implement water supply systems on a large scale in small communi- ties. PLANASA does not seek to provide water supply and sanitary waste disposal in rural areas. In the case of water supply to small communities, some progress is evident which may pave the way for expansion in these areas in the 1980s. - vii - Simulation Exercises xix. In spite of deficient historical data on basic residential services and the key rdle of non-economic factors in _etemining their supply, it is o inLteresL to .^xplore questions concerning the future evolution of thes, v rvices. Simple (and in some ways simplistic) arithmetical exercises, or simulations, are a convenient way of dealing with some such questions. xx. The specific objectives of the simulations reported here are to derive broad boundaries on the range of plausible values for: (i) time paths involved in expanding coverage of basic residential services in the remainder of this century; and (ii) maximum amounts of public sector outlays that might be required to meet coverage targets for specified years in the future. More- over, since it is of interest to explore the likely demographic impact of alternative future paths in the coverage of these services, the exercises are useful in providing some rough notions on what might be plausible limits within which such paths can be expected to fall. The simulatioris deal with the three basic residential services of central concern to the report: (i) adequate dwellings, (ii) access to water and (iii) sewerage. The defini- tion of each differs according to the urban or rural io.-ation of the dwelling. xxi. Depending on whether "optimistic" or "pessimist4`:" assumptions are made concerning the (long run) rate of Brazilian economic growth during the last two decades of this century (7% or 4% per annum, respectively), and concerning certain quantitative historical relation.hbips between per capita income and coverage *.f these residential services, simulation results suggest that by century's end probably: 10% to 30% of all Brazilian households would be living in dwellings that are considered "sub-standard" by most Brazilian poiicy-makers today -- for urban households, these proportions would be some-Ab -; higher; the corresponding percentage for 1970 is about 45%; between less than 10% and 30% of urban dwellings would lack connections to a general water network, in 1970 46% lacked such connections; one half to two thirds of rural dwellings would lack piped water, compared to 94% in 1970; 40% to 60% of urban dwellings would lack connections to a general sewerage network, compared to 78% in 1970; 20% to 50% of all rural dwellings would lack any sanitary device, compared to 75% in 1970. xxii. To some observers of Brazilian social develup,ent, even the optimistic simulation results might appear disappointing. For these, the question of what would be the rough dimensions of the financial outlays that would be required by the public sector to attain certain coverage targets that are more ambitious than the optimistic simulation results would be of some interest. A related question is whether such financial - viii - outlays would imply an unrealistic fiscal stress. These sorts of questions beg the crucial issue of the role of non-financial constraints -- such as organizational, staffing and logistical bottlenecks -- in drastically acceler- ating the pace at which coverage for the services is extended. Yet it may still be of interest to explore the extent to which "money is a problem". A second set of simulations address such preoccupations. xxiii. For the purpose of investigating whether the magnitude of financial requirements may present a binding constraint on attaining coverage targets that are ambitious by historical standards, these simulations are based on assumptions tending to overstate the required amounts. These simulations indicate that only a small proportion of GDP would be required as financial outlays by the public sector to substantially accelerate the expansion of basic housing services in Brazil during the next couple of decades and approach full coverage by century's end. Based on assumptions that systematically sought to produce upwardly biased projections of such public sector outlays, the simulations suggest that less than one and a half percent of GDP would be required during each of the two decades between 1980 and 2000, assuming GDP growth of 4% per year. Under 7% annual growth, GDP share would be about 40% lower by the decade of the 1990s. (In recent years some 1.7% of GDP were devoted to such expenditures.) There is no a priori reason to believe that this share of GDP would have to be substantially exceeded in any particular year during the entire simulation period. xxiv. These simulations also indicate that of the three housing sub-sectors included in the exercises (dwellings, water and sewerage), dwelling would take the lion's share in attempting to approach full coverage in the three sub- sectors by century's end. This share (for dwellings) would probably exceed 50% of public sector outlays for these housing related services over the remaining period to the year 2000. In turn, urban dwellings would probably absorb over 80% of the total dwelling's share. xxv. In conclusion, the simulations indicate that, compared to historical trends, a substantial acceleration in coverage of basic housing services might not involve a serious strain in the Brazilian economy. However, institutionlal and other factors pose obstacles to rapid future progress in these sectors. Important among these are legal and financial constraints on the capacity of public sector agencies to provide these basic services to families that cannot afford them under the current rules. Other potential constraints relate to the supply of scarce non-financial resources (e.g., top- and middle-level managerial talent) which may prove insufficient under current incentive patterns. Consequently, an effort of the magnitude implied by these ambitious simulations would probably require increasing financial subsidies to these sectors from outside the BNH system and finding ways to overcome institutional and other impediments, especially in rural areas. Analysis of the policy, program and administrative measures that could be taken to deal with such other constraints seem to warrant top priority for further work in this area. I. LONG TERM TRENDS IN BASIC INDICATORS This section reviews long-term trends in basic indicators showing the extent to which housing, water supply, and sewerage services have spread to the urban and rural populations of the three major regions used in this report (Northeast, Southeast, and Frontier). 1/ The basic sources used for this purpose are the 1950, 1960, and 1970 demographic censuses and the 1972 and 1973 National Household Sample Surveys (Pesquisa Nacional por Amostra de Domicilios - PNAD) all carried out and published by the Brazilian Institute of Geography and Statistics (Fundacao Instituto Brasileiro de Geografia e Estatistica - IBGE). This statistical review serves as the backdrop against which the evolution of public policy and programs during the period 1964-1977 are examined in part II of this Annex as well as the starting point for the projections contained in part III. A. Housing Considering Brazil's generally mild weather, 2/ the need for shelter as such (as opposed to safe water, sewerage, and other basic services) has probably never been particularly critical. 3/ By now, it appears to be fairly well satisfied for the vast majority of its population, though the quality and location of structures available to the poor leave much to be desired, parti- cularly in the Northeast and Frontier regions. According to Brazilian official statistics, only an insignificant proportion (less than 0.2%) of the population has lacked a "residential structure" 4/ in which to dwell during recent times--at least since the census year of 1960. Moreover, by 1970, a sizeable majority of families were living 1/ Throughout this report the Northeast means the 9 states classified by IBGE as the Northeast; the Southeast includes Minas Gerais, Espirito Santo, Rio de Janeiro (and the former Federal District of Guanabara), Sao Paulo, Parana, Santa Catarina and Rio Grande do Sul; and the Frontier is the rest of Brazil, the regions called Central West and North by IBGE. 2/ Monthly data collected between 1931 and 1960 by 60 meteorological stations scattered throughout Brazil show that the lowest average minimum tempera- ture for any of the 12 calendar months during the 30-year period was 6.50C and the absolute lowest temperature recorded was -8.40C. Throughout the post-colonial period, the vast majority of the population has lived within a fairly narrow coastal strip where temperatures rarely fall below 20 0C. (IBGE, Anuario Estatistico do Brasil, 1976, Table 1.2.3, pp. 28-24). 3/ It should be recognized, however, that protection from the elements is only one of housing's functions. Others include the provision of facilities for raising children, resting, preparing food and protecting personal property. 4/ I.e., one which was originally built for residential purposes; as opposed to "improvised" (e.g., a church, an abandoned streetcar, a cave). -2 - Table 1: DWELLINGSAa' BY TYPE OF CONSTRUCTION AND LOCATION. 1960. 1970, AND 1972 (in thousands of dwellings and percent distribution) Location and Type 1960 1970 19724/ Construction #% % All Brazil Total Houses 13504.6 100.0 17643.4 100.0 18233.0 100.0 - Permanent 13497.8 17628.7 18218.1 Durable N.A. 13008.0 73.7 15993.1 87.7 . Rustic N.A. 4620.7 26.2 2225.0 12.2 - Improvised 6.8 c/ 14.7 c/ 14.9 cl Urban Areas Total Houses 6355.3 100.0 10283.4 100.0 11510.0 100.0 - Permanent 6350.1 10276.3 11499.7 Durable N.A. 8779.6 85.4 10776.4 93.6 Rustic N.A. 1496.7 14.6 723.3 6.3 - Improvised 5.2 c/ 7.1 c/ 10.3 cl Rural Areas Total Houses 7149.4 100.0 7360.0 100.0 6723.0 100.0 - Permanent 7147.7 7352.4 6718.4 Durable N.A. 4228.4 57.4 5256.7 78.2 Rustic N.A. 3124.0 42.4 1501.7 22.3 - Improvised 1.6 c/ 7.6 c/ 4.6 c/ Notes a/ Non-institutional dwelling units. b/ Definitions: - Permanent are those built for residential purposes. Durable are those that are permanent, and in addition, are located in buildings with: walls predominantly made out of bricks, stone, clay or lumber; roof predominantly made out of shingles, zinc or concrete; and floors predominantly made out of wood, cement, stone slabs or tile. Rustic are those that are permanent but don't qualify as durable. - Improvised are those that are not "permanent" (i.e., under bridges, church steps, stores, etc.) c/ Less than 0.2%. d/ Frontier region is excluded with the exception of the Federal District. N.A. Not available. Sources; For 1960 and 1970: Demographic Censuses; For 1972; IBGE, Indicadores Sociais (Rio de Janeiro: 1977). Table 4, p. 165, - 3 - in residential structures classified as "durable" 1/ (Table 1). For the country as a whole, almost 75%. of all non-institutional dwelling units (or simply "dwellings" or "houses" hereafter) were classified as durable in 1970. For urban areas this coverage ratio was almost 1.5 times higher than in rural locations. Data from the 1972 PNAD indicate that for all of Brazil the coverage of durable dwellings had increased to 88% and the urban-rural relative differential dropped to 1.2 times. The rate of progress in extending durable shelter to the Brazilian population entailed in these data, however, appears too great to be either plausible or sustainable. They imply a rate of net in- crease in durable dwellings of over 9% per year in that (2.2 year) interval which, if maintained, would have resulted in enough durable dwellings for all Brazilian families by 1975. 2/ It is probable, however, that progress in this area was considerable during that period. 3/ Table 2: PERCENT OF DWELLINGS CLASSIFIED AS "DURABLE" BY REGION AND URBAN-RURAL LOCATION, 1972 /a (percentages) Region All locations Urban Rural Northeast 73.6 86.9 62.8 Southeast 94.4 95.9 90.7 Federal District N.A. 75.1 N.A. /a Each cell is the percent of dwellings classified as "durable" in each regional/locational category to the total number of dwellings in the same category. Source: IBGE, Indicadores Sociais (Rio de Janeiro; 1977). Tables 2 and 2A, p. 163. Absolute numbers are shown in Table B.1 of Appendix B. 1/ That is: with walls made predominantly of bricks, stone, clay or lumber; roofs of single, concrete or zinc; and hardfloors (wood, cement, stone slabs or tile). It should be stressed that this criterion of durability, which is the only one available from the IBGE data sources with respect to the quality of the dwelling, is a rather broad one, even by devel- oping country standards. Many of the "durable dwellings" are in rather bad shape. In addition, this dimension of dwelling adequacy says nothing about the degree of crowding existing in these premises which in many cases is great. Indeed, durability cannot be said to be the official indicator of minimum shelter standard. 2/ The most common and plausible explanation for the figures, in addition to a strong actual improvement, is the presence of sampling related errors in the PNAD survey. 3/ Recently available data from the 1976 PNAD show coverage of durable dwellings at 77%. These data imply a rate of net increase in durable dwellings of 4.4A per year for the period 1970-76 (IBGE, PNAD 1976, Table 1, p. 89). These numbers lend support to the suspicions concerning the 1972 PNAD figures. - 4 - 4. Compared to other housing related indicators to be examined later, the spatial breakdown of this ("durable dwelling") indicator does not reveal particularly great divergences across regional/locational categories. Durable dwelling coverage was only 1.2 times greater in urban areas than in rural loca- tions 1/ in 1972 (Table 2). The maximum gap was between urban Southeast and rural Northeast (1.5 times). The quality of shelter in the large areas of Brazil's interior where Chagas'disease is endemic, however, presents a serious health problem, since the vector of this disease inhabits the palm frond roofs and wattle and daub walls commonly found in the homes of the poor (see Annex III of this report). 5. Over half of all non-durable dwellings in Brazil in 1972 were located in the rural Northeast and 75% of them were occupied by households reporting a total monthly monetary income below the minimum wage. 2/ However, 75% of all such households were living in durable structures. 6. Thus, it would seem safe to assume that households residing in non-durable dwellings are by and large among the poorest of all. Using such a non-durable residence as a criterion to map the locational distribution of the very poor in Brazil in the year 1972 (but excluding from consideration, owing to lack of data, approximately 1.6 million households--with or without durable dwellings--located in the Frontier outside the Federal District). 3/ yields the following array: Non-Durable Dwellings, 1972 /a (Thousands of Dwellings and Percentages of Total) All Locations Urban Rural Dwellings (%) Dwellings (%) Dwellings (%) Northeast 1,308.6 (64) 137.9 (7) 1,170.7 (57) Southeast 696.6 (34) 362.7 (18) 333.9 (16) Federal District 32.4 (2) 32.4 (1) - - Total 2,037.6 (100) 533.0 (26) 1,504.6 (74) /a Source: Appendix Table B.1. 1/ I.e., the ratio of durable dwellings to total dwellings in urban areas was 1.2 times the corresponding coverage ratio in rural locations. 2/ IBGE, Indicadores Sociais (Rio: 1977), page 162, Tables 162, 163. Households reporting a total monthly monetary income below the minimum wage comprised the bottom of 37% of the household income distribution in 1972 (ibid.). 3/ This assumes only one household per non-durable dwelling. - 5 - 7. Further, the minimum shelter standard (of durability) being used here fails to reflect the precarious location (on steep hillsides, in areas subject to periodic flooding, or subject to hazardous environmental pollution) of many dwellings as well as their lack of essential services such as safe water and sanitary waste disposal. Overcrowding is also often cited as a serious problem by students of housing in Brazil. According to the 1970 census, the average number of persons per bedroom was 2.5 in urban areas and 2.6 in rural areas. While tabulations of bedrooms per inhabitant in relation to income per inhabitant cannot be derived from the published census data, no doubt the ratios are higher for the poor. Given the modesty of this "durable" standard, those families living in structures which fail to classify as "durable" must make do with very inadequate living conditions. 8. Between 1950 and 1978 the urban population of Brazil has almost quadrupled, and it is still increasing at over 4% per year as rapid rural- urban migration continues to augment the high natural rate of increase in the cities (see Annex I of this report). This means that each year 500,000 new urban households must find a place to live. Squatter settlements and illegal subdivisions, where self-help housing construction is the rule, are common phenomena, especially on the peripheries of Brazilian cities. Lack of clear title to the land they occupy often deters feasible investments in upgrading, they are often located far from sources of employment, and essential urban services are often absent. These problems are discussed further in part II of this annex. Summing up: (a) Although the housing problem, broadly defined, is a serious one, the lack of shelter from inclement weather, that is structures as such, would seem to rank fairly low in terms of either its serious- ness or priority among basic needs in Brazil. Virtually all Brazilians live in residential structures, and the majority in structures that meet the durability standards implicit in the census definitions. (b) Inferior shelter conditions, gauged by durable dwellings, are concentrated in the rural Northeast both in terms of coverage ratios and absolute numbers; (c) Durable shelter appears to have been reasonably affordable in that 75% of families reporting average earnings below the minimum salary in 1972, were then living in a durable dwelling. 1/ Thus, residence in non-durable dwellings would seem to be a robust index of extreme poverty; and 1/ The minimum salary is equivalent to about US$775 per year in the Rio and Sao Paulo metropolitan areas. - 6 - (d) The key housing problems of the urban poor are access to urban services, overcrowding, location, and legal status, not the lack of structures as such. In rural areas the prime shelter problem is upgrading walls and roofing to reduce exposure to Chagas disease. B. Water Supply 9, There are at least two aspects of water supply availability that seem important from this report's perspective. One is the presumably strong effect on family welfare of water accessibility for household purposes (particularly for washing, cooking and, to a lesser extent, drinking--since not as much of it is consumed for this latter purpose). The second aspect is the suit- ability of the available water for drinking purposes. 10. Brazilian data allow the estimation of two indicators relating to household water supply, encompassing reasonably long historical periods and susceptible to spatial disaggregation. One such indicator is the proportion of dwellings with piped water ("com canalizacao interna"). This appears to be as good as indicator of the "welfare aspect" of water availability as can be reasonably hoped for. (Ideally, one would like to have a breakdown among the dwellings that do not have piped water by, say, distance to the water source.) Considering the relevance of this indicator to the potability aspect of water supply, it would seem that while piped water is hardly a necessary condition for potability, it seems reasonable to suppose that piped water is more likely to be safe for drinking than water brought into the house by more rudimentary methods. The second indicator, the proportion of dwellings connected to a general water network (as opposed to a well or spring), is perhaps an even better indicator of potability, 1/ though in some of the older water systems contamination of water supplies arising from lower water pressures and obsolete, unmaintained piping systems presents serious problems. An additional limita- tion is that up to 1973, at least, it was close to zero in rural areas and it can only be traced back to 1960. On the other hand, it sheds considerable light on the performance of the public sector across regions and over time in providing this vital service to urban households. 11. Historical patterns and trends in these two indicators can be sum- marized as follows (refer to Table 3). 1/ Strong statistical association is found between a composite "safe water" indicator covering both urban and rural areas and infant mortality at the state level in 1970. (See Annex I). Table 3: PERCENTAGE OF DWELLINGS WITH WATER SUPPIY SERVICES AVAILABLE BY REGION AND URBAN-RURAL LOCATION,ai SELECTED YEARS, 1950 TO 1973 Average annual rate of Service, region Percentage of all dwellings in each category chance (in percent pyer-ear) and location 1950 1960 1970 1973 1950-1970 1950-1973 Piped waterh/ All regions 15.6 n.a. 33.3 41.1-/ 3.9 4.3 . urban 39.5 n.a. 52.6 58.1 1.4 1.7 . rural 1.4 n.a. 6.2 9.6c/ 7.7 8.7 Northeast 3.8 n.a. 11.1 14.6 5.5 6.0 . urban 13.0 n.a. 25.3 31.2 3.4 3.9 . rural .2 n.a. .8 1.1 7.2 7.7 Southeast 23.9 n.a. 45.7 53.7 3.4 3.6 . urban 54.2 n.a. 62.4 68.2 0.7 1.0 . rural 2.5 n.a. 11.2 17.3 7.8 8.8 Frontier 6.4 n.a. 17.8 n.a. 5.2 n.a. . urban 22.3 n.a. 34.0 36.9 2.1 2.2 rural .4 n.a. 3.3 n.a. 11.1 n.a. General network 1960-1970 1960-1973 All regions n.a. 21.1 32.8 40.2c/ 4.5 5.1 urban n.a. n.a. 54.4 60.8 . rural n.a. n.a. 2.6 2.4L/ Northeast n.a. 5.2 12.4 17.5 9.1 9.8 urban n.a. n.a. 28.7 38.1 rural n.a. n.a. .5 .6 Southeast n.a. 30.8 44.2 50.3 3.7 3.8 . urban n.a. n.a. 63.3 68.8 rural n.a. n.a. 4.5 4.0 Frontier n.a. 7.5 19.6 n.a. 10.1 urban n.a. n.a. 39.9 47.5 rural n.a. n.a. 1.6 n.a. a/ Each figure is the percentage of all houses in each region/location/year category having the indicated service. b/ Includes water from general network, wells or springs, delivered through internal plumbing ("Con canalizacao interna"). c/ Excludes rural Frontier. Source: Apnendix A, Tahles A2 and A3. - 8 - (a) The gaps 1/ between leading and lagging spatial categories 2/ were very large in 1950 (for piped water) and 1960 (for general network). By and large, these gaps narrowed considerably over time, but they were still rather wide by the year of the most recent available observation (1970 or 1973). By 1973, water coverage for urban dwellings was 6 times greater than for rural ones (down from 28 times in 1950); for dwellings in urban Southeast, coverage was about 2 times as great as in urban Northeast (down from 4 times in 1950). In 1973, dwellings in the Southeast were almost 3 times as likely as those in the Northeast to be connected to a general network (down from almost 6 times in 1960). (b) For the periods as a whole (i.e., 1950-73 for piped water; 1960-73 for general network) the rates of progress in both indicators seem generally rapid and were fastest for those spatial categories that had the lowest levels in the earliest years. (c) For all spatial categories for which data is available on both water supply indicators, their rate of progress 3/ accelerated between the periods 1950-70 and 1970-73, in the case of piped water, and between 1960-70 and 1970-73 in the case of general network connection. (d) For the years 1970 and 1973, for which data on both indicators are available, the proportions of urban dwellings with each type of service available in each year are fairly close between most corresponding spatial categories. The Northeast and the Frontier in 1973 are outliers in these comparisons. In both cases, the proportion of dwellings served by general network are appreciably greater than those with piped water. 4/ 12. Brazil's performance since 1950 in terms of water supply indicators appears impressive. However, when considering its countrywide status in 1970 in the context of international comparisons, Brazil does not seem to have been particularly well off then. On or about that year, countries in Brazil's per 1/ Although absolute gaps (defined as percentage point difference between two values of the same indicator) are also meaningful, we refer here (and hereafter unless otherwise noted) to relative gaps (defined as the ratio between two values of the same indicator). 2/ The term "spatial categories" is used here to include both regional categories (i.e., Northeast, Southeast and Frontier) and rural-urban categories within and across regions. 3/ Here and hereafter, "rate of progress" of housing related indicators is measured in terms of average annually compounded rate of change (in percent per year). 4/ Recently available data from the 1976 PNAD show the piped water coverage in that year at 50% of all dwellings (urban, 68%; rural, 14%). Corre- sponding figures for general network coverage are 43% for all dwellings (urban, 62%; rural, 4%). PNAD 1976, Table 1, p. 89. - 9 - capita income group (i.e., $370 to $1,000--in 1973 US$) averaged 42% of all dwellings with piped water, versus 33% for Brazil in 1970. 1/ Doubtlessly, this is largely attributable to the very low levels of this indicator in 1950 in Brazil rather than to its rate of progress since that time. Considering the rapid progress in Brazilian water supply indicators between 1970 and 1973 implied by the data, it is likely that by the latter year the gap in such an international comparison would have narrowed somewhat, but international comparison data is not available beyond 1970. C. Sewerage 13. Together with accessibility to safe drinking water, availability of sanitary sewerage facilities has important favorable effects on health. 2/ Two indicators of sanitation facilities available to Brazilian dwellings are examined here. One is the percentage of dwellings that is either connected to a general sewerage network or has access to a septic tank (the "strong" sanitation indicator). The other is the percent of houses having any sort of sanitary device (this "weak'1 sanitation indicator includes as a sanitary device anything ranging from indoor plumbing to "...a rudimentary ditch or some other drain, when used directly as such, or a river, a lake, etc."). 3/ The two indicators differ in the degree of comfort and expected health hazard implied for the household in question; they differ also in a qualitative sense. 14. The weak indicator corresponds to a situation which threatens the health of neighboring households, particularly through water contamination. This risk can be expected to grow with the density of residential agglomera- tion. 4/ Still, it is probably true that health consequences of sanitation conditions are even more serious when not even the weak indicator conditions exist. 15. These are the salient features of the recent historical evolution of the two sanitation indicators in Brazil (Table 4). (a) As in the case of water supply accessibility, gaps between leading and lagging spatial categories were very wide in the initial periods (1950 for the weak indicator and 1960 for the other one) for each of the two indicators. (b) Although for Brazil as a whole these indicators did not improve as rapidly as in the case of those for water supply between 1/ World Bank, World Bank Tables 1976, Table 6, p. 526. 2/ For a discussion of this issue in the Brazilian context, see Annex IV. 3/ IBGE, Censo Demografico 1970, p. XXXVI. 4/ In cases of large communities where network sewerage goes untreated-- as is typically the case in Brazil to this day--a similar problem may also arise even if the strong indicator criteria are met. - 10 - Table 4: PERCENTAGE OF DWELLINGS WITH SANITATION FACILITIES BY REGION AND URBAN-RURAL LOCATION, SELECTED YEARS. 1950 TO 1973 Percentage of all dwellings in Service, Region each catego gy / Average Annual Rate of Change and Location 1950 1960 1970 1973a' (in percent per year) A. Any Sanitary Device 1950-70 1960-70 1950-73 All Regions 33.0 49.0 60.6 67.5b/ 3.4 2.0 3.2 Urban 71.3 n.a. 85.6 87.3 0.9 0.9 Rural 10.4 n.a 25.6 31.0 4.6 4.9 Northeast 12.6 24.1 29.1 31.3 4.3 1.9 4.0 Urban 39.3 n.a 60.8 62.0 2.2 2.0 Rural 2.4 n.a. 6.1 6.1 4.8 4.1 Southeast 46.6 63.8 76.1 82.6 2.5 1.8 2.5 Urban 84.1 n.a 92.7 94.4 0.5 0.5 Rural 17.1 n.a. 41.7 53.2 4.6 5.0 Frontier 24.3 43.6 55.0 n.a. 4.2 2.3 Urban 66.3 n.a. 87.2 90.7 1.4 1.4 Rural 8.6 n.a. 26.4 n.a. 5.8 B. Sewerage Network or Sceptic Tank 1960-70 1960-73 All Regions n.a. 23.8 26.6 39.1-/ 1.1 3.9 Urban n.a. n.a. 44.2 57.6 Rural n.a. n.a. 2.0 5.0-/ Northeast n.a. 6.1 8.0 15.2 2.7 7.3 Urban n.a. n.a. -18.6 31.7 Rural n.a. n.a. 0.3 1.7 Southeast n.a. 34.6 37.2 50.5 0.7 3.0 Urban n.a. n.a. 53.4 67.5 Rural n.a. n.a. 3.6 7.9 Frontier n.a. 8.9 12.6 n.a. 3.5 Urban n.a. n.a. 25.7 35.3 Rural n.a. n.a. 0.9 n.a. a/ Definitions applicable to 1973 figures are not strictly comparable with figure for other years. See footnote to item (c) on page 11. b/ Excludes rural Frontier Source: Appendix A, Tables A3 and A4. - 11 - 1950/1960 and 1973, the spatial patterns in their evolution were more markedly convergent--regions and locations that were lagging in the earlier year improved even more rapidly relative to the leading categories. This was partly due to the high level already attained by 1950 by the urban Southeast in the weak sanitation indicator. By 1973, urban dwellings were almost 3 times as likely to have some sanitary device as rural ones (down from close to 7 times in 1950); dwellings in urban Southeast were 1.5 times as likely to have such facilities as in urban Northeast in 1973 (down from 2.1 times in 1950). In 1973, dwell- ings in the Southeast were 3.3 times as likely to be connected to a general sewerage network or to a septic tank as in the Northeast (down from 5.7 times in 1960). The virtual totality of dwellings having either of these utilities must have been in urban areas, since by 1970 only 2% of houses in rural areas were so equipped. (c) The fastest overall rate of improvement in the weak sanitation indicator took place between 1950 and 1960. It slowed consider- ably during the following decade and accelerated again between 1970 and 1973. The 1960-70 period was of even slower progress in terms of the strong indicator and the data imply that its improvement between 1970 and 1973 proceeded at an extra- ordinarily high pace (almost 14% per year) but the improvement in the latter year appears implausibly great. 1/ Judged by these indicators, Brazil's progress in the availability of sanitation facilities to private households since 1950 has been very con- siderable, particularly in rural areas generally and in the Northeast. Room for further improvement, however, is still quite ample, considering for example that by 1973 only about 40% of all dwellings were connected to a general sewerage network or to a septic tank. II. EVOLUTION OF PUBLIC POLICIES AND PROGRAMS 1964-1977 This section reviews and analyzes institutional development, concrete achievements, contraints, and public policy evolution in the areas of housing, water supply and sanitary sewerage in Brazil since 1964. Emphasis is placed on the provision of these basic services to the low income urban population. To date there has been only a minimum institutionalized effort to upgrade rural housing and related services. As for the period before 1964, only a very limited amount of information is presented here. The National Housing 1/ There are two possible reasons for this rapid rise of the "strong" indi- cator: differing definitions and sampling error. Households sharing bathroom facilities in the general network, septic tank or other catego- ries were included in the PNAD of 1973. In the Census definition, these cases were classified as being without facilities. In other words, the Census definition of units connected to the general system, a septic tank or with some other more primitive system does not include dwellings where the bathroom is shared among two or more households, while the 1973 PNAD does. This source of error is probably more important than sampling error. The 1976 PNAD definition, however, is the same as the Census definition. Recently available data from the 1976 PNAD show the sewerage or septic tank coverage at 37% of all dwellings (urban, 54Z; rural, 5%). Corresponding figures for any sewerage device are 71% for all dwellings (urban, 88%; rural, 31%). - 12 - Bank (Banco Nacional de Habitacao - BNH), created in August 1964, is by far the most important institution operating in the housing sector. Its activities have been progressively expanded to embrace water supply, sanitary sewerage, municipal drainage, and other forms of urban infrastructure, so that it has become a general urban development bank. Considerable attention is therefore devoted to the BNH's own development, accomplishments, and the factors which have constrained this institution in pursuing public policy objectives. A. The Situation in 1964 18. The rapid process of urbanization and industrialization experienced by Brazil in the post-World War II period coupled with an accelerating infla- tionary spiral in the early 1960s left Brazil with a housing deficit estimated by the Government at 7 million units as of 1964. 1/ The housing finance system as well as that for public utilities such as water supply and sewerage were in a state of virtual financial collapse. The privileged few who could obtain long-term housing loans through the Savings Banks (Caixas Economicas), the Retirement and Pensions Institutes (Institutos de Aposentadoria y Pensoes) and the Popular Housing Foundation (Fundacao da Casa Popular) at fixed interest rates in effect received a real transfer of income, often at the expense of low and middle income depositors. As a result, the Retirement and Pension Institutes had by 1964 virtually ceased financing housing. In the case of water supply and sewerage, provision of these services was the responsibility of the municipios. State or federal funds were allocated sporadically on a grant basis to municipal sanitation companies which often operated with unreal- istically low tariff structures. There were almost never enough grant funds to finance integrated systems on a regular basis permitting systematic expansion, and what funds were available could provide at best partial short-term solu- tions to the most pressing problems. 19. The 1964-66 Program of Economic Action of the Government noted that the serious housing shortage was in part due to inadequate urban planning in the fact of the cities' explosive growth. But the principal cause was seen as inflation which had three negative effects: (a) Principally in the first years after the war, inflation gave rise to speculation in unused urban land (and housing units which could be sold for cash to relatively rich individuals) as inflation-protected assets. The value of these lands increased faster than inflation, thus rendering them inaccess- ible to the poorest strata of the population; (b) Uncertainty regarding the future rate of inflation dried up private sources of long-term housing finance needed by poor and middle class families to purchase their own dwellings. Only a small volume of such finance remained available through official institutions; (c) The freezing of rents by legislation discouraged investment in housing units for rent. 1/ Ministerio do Planejamento e Coordinacao Economica, Programa de Acao Economica do Governo 1964-1966, Sintese (Rio de Janeiro: EPEA, 1964), p. 87. - 13 - 20. According to the BNH, the old official housing finance system financed the construction of only 120,000 units between 1938 and 1964. 1/ The Popular Housing Foundation, created in 1946 to intensify the production of housing for the lowest income strata, had financed only 17,000 units in 18 years. In these conditions, the only kinds of commercially constructed housing available to the vast majority of the urban poor were slum tenements formed by subdividing old single family dwellings (corticos), attached one- room dwellings sharing a common bathroom built specifically for the low income housing market (casas de comodos and cabecas de porco), and similar rows of one-room dwellings constructed on back alleys (avenidas), etc., in decaying parts of the cities. The principal alternative was self-constructed dwellings in semi-legal private subdivisions (loteamentos populares) or on illegally invaded lands (favelas, mocambos, etc.) These improvised dwellings formed the mushrooming shantytowns to be found in and around almost every Brazilian city. In many cases the residents could not obtain title to their land (often because the land was occupied illegally), many of the sites occupied were on steep hillsides or in areas subject to periodic flooding, and thus not considered safe for permanent construction, and water and sewerage services connected directly to the dwellings were for the most part non-existent. But given the tropical and semi-tropical climates prevailing in most of the country, this "informal" housing sector served to provide a modicum of afford- able and sometimes relatively well located (in terms of access to employment) housing for 15-45% of the population of major Brazilian cities in the early 1960s. 2/ B. The New System for Financing Housing and Related Urban Services: Key Institutions and Funding Sources 1. The National Housing Bank and the Housing Finance Systems 21. This, then, was the situation in 1964 when the BNH was created and the Housing Finance System (Sistema Financeiro da Habitacao - SFH) instituted. Measures were also taken to permit more realistic rents to be charged. Law 4380 of August 21, 1964 defined the principal objective of the SFH to be "to promote the construction and acquisition of owner-occupied homes (casa propria), especially for the lower income classes" and gave the BNH the task of "orienting, disciplining, and controlling" the newly created system. The most important financial innovation introduced by the SFH was the institution of "monetary correction" in housing loan contracts and deposits in the SFH's lending institutions, thus maintaining the real value of assets and liabili- ties of the new financial system in an inflationary environment. The insti- tution of monetary correction was a major reform designed to make long-term 1/ Olavo G. Otero and Luis C. Gurgel do Amaral, 0 Plano Nacional de Habitacao, (Sao Paulo: Telepress, no date, probably 1971), p. 35. 2/ Janice Perlman, The Myth of Marginality: Urban Poverty and Politics in Rio de Janeiro, (Berkeley, Los Angeles and London: University of California Press, 1976), p. 16. - 14 - capital markets serving the housing sector viable, and on the whole has proven highly successful. 22. The basic monetary unit of the SFH is the Standard Capital Unit (Unidade Padrao de Capital - UPC) equal to the nominal value of one Indexed Treasury Bond (Obrigacao Reajustavel do Tesouro Nacional - ORTN) in the first month of each quarter of the year. The value of the ORTN is closely related to the wholesale price index. The UPC is thus revalued each three months and that value prevails for the entire quarter. Table 5 shows that the average annual value in 1976 cruzeiros of the UPC (using the general price index for domestic supply 1/ as the deflator) remained remarkably constant from 1965 through 1973, never falling below its 1965 value and at its highest point (1972) being only 4.5% above its 1965 value. A change in the indexing formula in December 1972 introduced what turned out to be an underestimate of future inflation into the indexing formula. The direct link to the wholesale price index was reestablished in March 1974, but since September 1975 a new adjust- ment has been made in the wholesale price index for the purpose of calculating the monetary correction coefficient. The adjusted index excludes the impact of those price changes which are considered the result of "fortuitous" factors or factors beyond the control of the Government. These may include changes due to extreme weather conditions or changes decreed by outside forces such as OPEC. Furthermore, in June 1976 the correction formula was again changed to reincorporate an estimate of future inflation (weight 20% as against 50% in the period December 1972-March 1974) which has consistently underestimated the realized rate of inflation. As a result of these changes, the real value of a UPC in 1977 stood 14% below its 1965 value and 17.7% below its peak 1972 value. If continued, this trend could erode savers' confidence in the SFH. 23. The value of the UPC as a percentage of the Rio de Janeiro minimum wage rose during the period 1965-69 and then stayed relatively constant through 1974 before declining to slightly below its 1965 value in 1977. While it is difficult to draw definitive conclusions from minimum wage data, when coupled with increasing construction and land costs, reflected in rising average loan values (see below Tables 12 and 15), these movements suggest that families whose incomes were the same multiple of the minimum wage would have faced rising real debt burdens over time for the purchase of the same quality of housing. As will be seen below, attempts were made to reduce the impact of these relative wage and price movements by altering interest rates and amor- tization periods. 24. The BNH began operations in 1965 with an initial capital of Cr$1 million (US$910,000 at the exchange rates then prevailing). In addition it was to receive a payroll tax of 1% on all wages and salaries subject to labor legislation and 20 year loans from owners of buildings equivalent to 4% of the value of rents received each year. These two sources soon proved insuffi- cient to finance the large scale programs envisioned by the BNH. The problem 1/ This broad inflation index is a weighted average of the wholesale price index (weight .6), the cost of living index in Rio de Janeiro (weight .3) and the construction cost index (weight .1). Table 5: AVERAGE ANNUAL VALUE OF ThE UPC AND THE RIO DE JANEIRO MINIMUM WACE, 1965-1977 Year Average Value of Average Value of Average Value ° Average Monthly Average Monthly Average Monthly Average Value of UPC UPC (Current Cr$) UPC (1976 Cr$) a/ IlPC (1976 US$) - Minimum hage, Minimum Wage, Minimum Wage, as 7. of Average Rio de Janeiro Rio de Janeiro Rio de Janeirg Rio de Janeiro (Current Cr$) (1976 Cr$) a/ (1976 US$) Monthly Minimum Wage 1965 13.95 167.09 15.49 62.31 746.34 69.20 21.0 1966 18.92 164.34 15.24 81.0C 703.57 65.23 23.4 1967 25.36 171.56 15.91 102.35 692.46 64.20 24.8 1968 31.06 169.16 15.68 122.23 665.73 61.72 25.4 1969 37.99 171.09 15.86 147.20 663.93 61.55 25.8 1970 45.21 170.22 15.78 176.8 665.69 61.72 25.6 1971 54.21 169.48 15.71 212.8( 665.29 61.68 25.5 1972 65.30 174.54 16.18 254.4Q 679.97 64.71 25.7 1973 74.43 172.81 16.02 297.bC 690.94 64.06 25.0 1974 89.01 160.59 14.89 355.2( 640.84 59.41 25.1 1975 116.25 164.22 15.23 480.8' 679.23 62.97 24.2 1976 149.63 149.63 13.87 689.60 689.60 63.93 21.7 1977 204.86 143.61 13.31 993.6(' 696.29 64.55 20.6 al The General Price Index (Domestic Supply) was used to convert current to 1976 prices. b/ Converted at average exchange rate for 1976, Cri10.786 per US$. Source: Banco Central do Brasil and Calculations. - 16 - was resolved by Law 5107 of September 1967, which created the Time-on-Job Guarantee Fund (Fundo de Garantia do Tempo de Servico - FGTS), a form of forced savings obtained by an 8% tax on all payrolls subject to labor legis- lation. The establishment of FGTS and its link to the BNH was the second major innovation (after monetary correction) which has permitted the growth of a major system for financing urban development in Brazil outside the normal budgetary process. FGTS funds are accumulated it individual accounts in the BNH subject to quarterly monetary correction and bearing an additional minimum 3% rate of interest. 1/ They are available to workers and thleir heirs under specified circumstances including involuntary unemployment, serious sickness of the worker or a member of his family, purchase a dwelling, retirement, and death. Table 6 shows BNH liabilities as of December 31 for each year from 1967-1977, broken down into FGTS deposits and other liabilities (including capital, reserves, and various others). Despite the fact that since 1971 FGTS deposits have gradually decreased as a percentage of total liabilities, their real value was still increasing at over 20% per year in 1977, when they accounted for 66.9% of total BNH liabilities. 2/ Over time FGTS disbursements have risen as a percentage of gross receipts from 3.1% in 1966 to 51% in 1977. Total FGTS accounts in 1976 totaled 21.4 million, or 60.5% of the employed population. 3/ 25. Tables 7 and 8 provide summary financial statements indicating the overall development of BNH's financial activities from 1971 through 1976 expressed in current U.S. dollars. BNH's growth during this period is evident from this data. The total loan portfolio expanded by 369% and total assets by 337%. By 1976 BNH was making new loans at the rate of US$2.6 billion per year. 1/ The rate was initially set at a minimum of 3% for worker with less than three years service with the same firm. If the worker chose to forego the benefits of "labor stability" (which virtually assures job tenure after 10 years of service) thereby obtaining direct ownership of his FGTS account, the rate rose progressively to a maximum of 6% after 10 years of service. In September 1971 Decree 69,265 set the rate at 3% for all new accounts, but retained the previous rules for those who had entered the system under the original rules. The quarterly monetary correction and capitalization was abandoned in favor of annual monetary correction and capitalization in December 1972 (Decree Law 71,636), but the original quarterly system (more beneficial to the depositor] was reestablished in January 1976. 2/ In 1977 other BNH liabilities included own resources--capital, reserves, etc. (15.1% of the total), deposits of SFH institutions (3.2%), a liquid- ity fund to assist SFH financial institutions (7.2%), foreign loans (1.0%) and all others (6.6%). 3/ Roberto Cavalcanti, "O Governo Geisel e o Desenvolvimento Social," p. 4. - 17 - TABLE 6 LIABILITIES OF THE BNH, 1966 - 1977 (As of December 31 each year, current Cr$ millions and % of total) All Other Total Year FGTS Deposits Liabilitiesa/ Liabilities 1966 155 155 (0.0) (100.0) (100.0) 1967 629 305 934 (67.3) (32.7) (100.0) 1968 1,902 469 2,371 (80.2) (19.8) (100.0) 1969 3,611 778 4,389 (82.3) (17.7) (100.0) 1970 6,040 1,391 7,431 (81.3) (18.7 (100.0) 1971 9,813 2,073 11,886 (82.6) (17.4) (100.0) 1972 14,788 3,609 18,397 (80.4) (19.6) (100.0) 1973 20,982 5,402 26,384 (79.5) (20.5) (100.0) 1974 32,897 13,283 46,180 (71.2) (28.8) (100.0) 1975 48,413 21,981 70,394 (68.8) (31.2) (100.0) 1976 79,011 38,646 117,657 (67.2) (32.8) (100.0) 19771 128,749 63,644 192,363 (66.9) (33.1) (100.0) Source: Boletim do Banco Central do Brasil, August 1978. Note: a/ Includes capital, reserves, SFH deposits, foreign loans, and others. - 18 - Table 7 BNH SUMMARY FLOW OF FUNDS STATEMET, SELECTED YEARS, 1971-1976 (Millions of U.S. Dollars) 1971 1973 1975 1976 Operating Income 61.0 144.6 220.6 256.0 Indemiity Deposits (FGTS) 424.4 727.1 1,135.1 1,199.6 All Other Sources 249.0 361.8 904.7 1,132.5 Total Sources 734.4 1,233.5 2,260.4 2,588.1 New Loans Granted 618.4 1,021.9 1,939.4 2,569.7 All Other Applications 116.0 211.6 321.0 18.4 Total Applications 734.4 1,233.5 2,260.4 2,588.1 Source: BNH and IBRD/LCP Table 8 BNH SUMMARS BALANCE SIMETS, SELECTED YEARS. 1971-1976 (Millions of U.S. Dollars) December 31 1971 1973 1975 1976 Loan Receivables 1,545.3 3,131.7 5,461.4 7.246.1 Negotiable Bonds 203.8 761.1 1,120.0 1,048.8 All Other Assets 378.1 345.4 893.2 995.8 Tctal Assets 2,127.2 4,238.2 7,474.6 9,290.7 Indemnity Fund Deposits 1,741.5 3,373.3 5,337.7 6,400.3 Capital and Reserves 283.8 696.4 1,349.8 1,495.2 Other Liabilities 101.9 168.5 787.1 1,395.2 Total Liabilities 2,127.2 4,238.2 7.474.6 9,290.7 Source: N and I=/LCI. - 19 - Beyond administering the FGTS, the BNE serve %. L central bank for the Brazilian Savings and Loan System (Sistema Brasileiro de Poupanca e Emprestimo - SBPE) established in 1966. As of Sentember 1977 the SBPE con- . sted of a large variety of savings and real e-tate credit agencies. These included the Federal Savings Bank (Caixa Economic Federal - CEF) and 5 S-late Savings Banks--these 6 institutions being public enterprises already in existence at the time the SFH was set up and which adapted to its rules. in additio, there weri- "4) Real Estate Credii Associations (Bra ' companies) and 36 Savings and Loan Associations. In 1976 the SBPE was the fourth largest system of savings and loan associations in the world as measured by the value of accounts, smaller than those of the USA, Great Britail: and Federal Republic of Germany, but larger than those of Australia, France, and Venezuela. In terms of numbers of accounts, only the USA, Great Britain and Germany led Brazil. 1/ The SBPE institutions obtain their re-;urces directly from tbe public through two mechanisms: passbook savings accounts (cader.etas de poupanca) and housing bonds (letras imobiliarias), both of which are guaranteed by the BNH up to the value of 1,000 UPCs (approximately US$14,800 in 1977). Thus the BNH, in addition to being an investment bank, serves as a central bank controlling the entities forming the SBPE. As of October 1977, the total resources captured by the SBPE were Cr$182.9 billion (US$12.1 billion) compared to Cr$125.6 billion (US$8.3 billion) in FGTS deposits in the BNH. The number of savings passbooks in the SBPE has grown extremely rapidly in recent years from 1.3 million at the end of 1970 to over 17 million by the end of 1977. The total value of passbook savings accounts in nominal terms fell for the first time in November 1977, however, which suggests that falLing real yields on this savings instrument have begun to affect flows of volu, ary savings to the SFH. In fact the real value of all passbook savings accounts remained below the October 1977 level through February 1978. The interest rate on passbook savings accounts is monetary correction plus 6%, but in larc!h 1978 number of new options for contractual savings bearing interest rates up to 7.2% were introduced. At the same time income tax deductions equal Lo 2% of average balance were granted for accounts over 400 UPC. 2/ These moves were designed to reverse the outflow of funds in larger accounts in response to falling real yields. The BNH's own resources and those of the SBPE together were about Cr$40 billion in October 1977, so that the total domestic resources mobilized by the SFH came to Cr$348.5 billion (US$23 billion). As of July 1977 an addi- tional Cr$1.6 billion (US$114.7 million) had been obtained by the BNH from foreign sources (AID, IDB, and World Bank). The increase in the value of loans outstanding by the entire SFH during 1976 war: the equivalent of 3.1% of GDP and 11.4% of gross fixed capital formation. 1/ BNH em Resumo, Setembro/Outubro 1977, p. 27. 2/ A deduction of 47 of the average balance on accounts up to 400 UPC continues. - 20 - 29. Table 9 shows the breakdown of all BNH applications, measured in UPC, over the years 1964-76 and those planned for 1977-79 according to the pluriannual budget covering that period. The prospective acceleration in the growth of the system is evident from the fact that planned applications in 1977-79 almost equal those of the preceding 13 years. All BNH applications in 1976 equaled 1.7% of GDP. During the period 1964-76, loans for "housing and its complements" (which include loans to users of construction materials) amounted to 66.7% of all applications. Largely due to the relatively bigger share planned for urban development loans (especially water supply, sanitary sewerage, and drainage) during the 1977-79 period, the share for housing and its complements is expected to fall to 61.9%. It was much higher in the 1960s and early 1970s. During the period 1964-76 only 7.8% of all loans were used for low income housing despite the mandate of the BNH to give special attention to this field. 30. Four major factors which conditioned this performance in the area of low-income housing are examined in Section IIC2 of this annex: Brazil's large proportion of poor families (a problem which has deep historical roots and cannot be solved by purely sectoral economic policies), the need to maintain BNH's internal financial viability, the relative rise in land and construction costs since 1972, and deficient local level implementation capacity. In general, BNH has proved more successful at gathering financial resources than in financing low-income housing, though designing and carrying out successfully low-cost housing programs has proven a difficult task in virtually all countries where it has been attempted. One reason for the diversification in the direc- tion of urban development has been the need for construction of urban infra- structure and the development of the building materials industry, themselves critical for tackling the housing problem. Another is that the BNH has had to find ways to invest the huge influx of funds from the forced and voluntary savings systems at adequate rates of return in the face of costs for producing finished low-income housing which effectively price most of the poor out of the official housing market. What low- and middle-income housing as has been produced has been absorbed by the market and indeed there are long waiting lists for completed BNH financed housing units. 2. The National Housing Bank: Basic Modes of Operation - Housing 31. When acting as an investment bank, the BNH always acts through intermediaries known as financial agents and promoters. From its inception, the BNH has actively encouraged the formation of such intermediaries. In the area of low-income housing, these are known as Popular Housing Company (Companhias de Habitacao Popular - COHABs). The COHABs are controlled by state and/or municipal governments and perform the functions of both finan- cial agents and promoters. As of October 1977 there were 22 COHABs in all the states and the Federal District and an additional 10 controlled by municipal- ities or groups of municipalities. In addition, under the National Popular Housing Plan (Plano Nacional da Habitacao Popular - PLANHAP) initiated in January 1973, each state was supposed to establish a State Popular Housing Fund (Fundo Estadual de Habitacao Popular - FUNDRAP) which would finance a part of the cost of popular housing from state resources, supplemented by BNH - 21 - TABLE 9 BNH APPLICATIONS OF FUNDS, ACTUAL 1964-1976 AND BUDGETED 1977-1979 1964-76 1977-79 Programs million UPC % million UPC % Housing and Its Complements 488.0 66.7 439.5 61.9 Low Income Housing 57.2 7.8 86.1 12.1 Middle Income Housing 143.5 19.6 122.8 17.3 High Income Housing b/ 170.7 23.3 110.0 15.5 Urbanized Lots 0.1 0.0 11.7 1.6 Construction Materials Loans to Consumers 93.8 12.8 65.5 9.2 Other 22.7 3.1 43.4 6.1 Urban Development 154.8 21.2 221.2 31.1 Water Supply, Sewerage and Drainage 78.9 10.8 131.5 18.5 Urbanization 22.0 3.0 14.5 2.0 Community Equipment 12.8 1.7 8.2 1.2 Urban Transport 24.1 3.3 12.5 1.8 Other 17.0 2.3 54.5 7.7 Technical and Financial Support Operations 46.0 6.3 49.8 7.0 Research, Training and Technical Assistance 1.0 0.1 7.8 1.1 Loans to Construction Materials Industry 45.0 6.1 42.0 5.9 Complementary Operations c/ 43.2 5.9 - - TOTAL 731.9 100.0 710.5 100.0 a/ Totals and subtotals may not equal sums of components due to rounding. b/ BNH support for SBPE. c/ Includes purchase of Indexed Treasury Bonds, but excludes acquisitions of real estate Sources: Calculated from BNH, Relatorio de Atividades em 1976 and BNH, Orcamento Plurianual 77/79. - 22 - loans. It took almost three years until all states had officially adhered to PLANHAP and established their FUNDHAPs, in part due to the changeovers in state and federal administrations which took place during this period. In 1975 the regulations on FUNDRAP contributions were changed, drastically reducing their required contributions, and the FUNDHAPs have not played a significant role in popular housing finance. 32. Initially "popular housing" constructed by the COHABs with loans to purchasers of up to 200 UPC was available only to families with incomes of up to three minimum wages, but over the years the maximum size of loan increased first to 320 UPC and then to 500 UPC. These increases were motivated by rapidly rising construction costs and land prices, which effectively priced most "popular" housing out of reach of even those families earning three minimum wages. In 1975, the maximum family income for popular housing loans was raised to five minimum wages or 22 UPC, whichever is lower. The COHABs as well as other financial agents including savings and loan associations, real estate credit societies, savings banks, and state and regional development banks, also operated in the "economy" housing market (originally for families with 3-6 minimum wages income and with loan values of 200-400 UPC) through promoters, principally housing cooperatives. Since 1975, the "economy" and "medium" (initially loans between 400-900 UPC) classifications have been dropped, and loans through cooperatives and other promoters, such as mortgage companies, are now available to families with incomes between five and ten minimum wages. The average cooperative loan was budgeted at 1,100 UPC in the 1977-79 BNH budget. Higher income groups have been served through loans for construction materials (the RECON program of BNH) made to individuals through a wide variety of financial agents. Finally, the BNH guarantees loans of up to 3,500 UPC made through the SBPE which are available to families with more than six minimum wages monthly income. The average loan through the SBPE was budgeted at 2,200 UPC in the 1977-79 BNH budget. 33. The institution of monetary correction has been vital in channeling both forced (FGTS) and vqluntary (passbook savings accounts, housing bonds, etc.) savings into the SFH. But the fact that the BNH must pay an average of somewhat over 3% interest on FGTS deposits, and the SBPE 6%, means that, including administrative costs, the cost of funds to BNH is almost 5% and on the order of 8% for the SBPE I1 (though at least one SBPE institution, the Caixa Economica Federal, obtains some non-interest bearing government deposits). In addition, all assets and liabilities of the SFH are denominated in UPC and thus subject to monetary correction. Thus the principal means by which loans to low-income borrowers can be subsidized within the SFH is to charge rates of interest over the average cost of funds to higher income borrowers, and then use the profits to subsidize loans at the low end of the income spectrum. The system of cross subsidization is widely used by the BNH, as is explained in greater detail below. 1/ Mauricio Schulman, "The Housing Challenge in Brasil," paper presented at the Fourteenth World Congress of Real Estate Savings and Loan Credit Companies, September 1977, San Francisco, California, USA. - 23 - 3. The Financing System for Water and Sewerage 34. In 1968, after the establishment of the FGTS, the BNH began financing urban water and sewerage systems. The Sanitation 1/ Financing System (Sistema Financeiro do Saneamento - SFS) is controlled by a superintendency of the BNH. Resources are mobilized at the federal, state, and municipal levels. At the federal level sources of funds include BNH's own resources, budgetary alloca- tions to the Fund for Sanitation Finance (Fundo de Financiamcnto para Saneamento - FISANE) and foreign loans to the BNH. On the state level BNH has stimulated the formation of Funds for Finance of Water and Sewerage (Fundos de Financiamento para Agua e Esgosos - FAEs) which obtain their funds principally from state taxes and federal revenue sharing, the returns of loans made to state sanitation companies, and supplementary credits obtained by the state governments from the BNH. Finally, at the municipal level both municipal taxes and state revenue sharing funds may be used to supplement resources obtained through the direct operation of water and sewer systems by state sanitation companies. Beginning in 1971, these companies have been taking over the concessions of existing municipal companies which have joined the National Sanitation Plan (Plano Nacional de Saneamento - PLANASA) as well as establishing wholly new municipal systems, again under PLANASA. The state sanitation companies are the final recipients of SFS loans. They must show the financial viability of their respective systems at the state level, but may carry out cross-subsidization of weaker municipal systems by stronger ones through centralizing control over cash flow and investment programs at the state level. The use of budgetary resources at each level of the federal system to supplement loan funds (overwhelmingly from BNH) means that the average cost of funds through the SFS is considerably less than it is through the SFH. 35. In general, when a state sanitation company is financially strong enough to support an investment in the form of a loan, the loan will b,e com- posed 50% of BNH funds and 50% of FAE resources. FAE loans to state sanita- tion companies bear interest rates equal to the rate of urban growth oi the state, currently ranging from 3 to 6%, and are subject to monetary correction. The long-term objective of the SFS is to make the state systems self-sustaining on the basis of FAE resources and operating for revenues. In states where an investment exceeds the maximum recommendable for the state sanitation company, the difference is covered with non-reimbursable resources (a fundo perdido) which may have a federal, state, or municipal origin. The FAEs obtain their funds from state revenues which may not exceed 5% of their total tax revenues, as well as returns on their loans to state sanitation companies. 36. When the state government participation, together with returns on earlier FAE investments, is insufficient, the BNH may make a supplementary loan to the FAE so that the 50% FAE participation in new loans to state water companies will be met. In sanitation loans to its financial agents the BNH 1/ The term sanitation which is used in Brazil and in this report includes water supply as well as sanitary sewerage. - 24 - uses a system of cross-subsidization of interest rates (2 to 7%) to lower the cost of its loans destined to benefit the poorer states and municipios as measured by tax collections. The financial agents are normally allowed a spread of 1 percentage point, so that the interest rate on loans to state sanitation companies and FAEs is 3 to 8%. At the state level the factor of population density also is taken into consideration, with lower density states obtaining more favorable interest rates. Most BNH sanitation loans have terms of up to 18 years plus up to 3 years grace. As in the case of housing, the BNH uses a system of financial agents and promoters to carry out its water and sewerage programs. At the level of individual consumers the tariff structures of operating sanitation companies may also incorporate the principle of cross-subsidization. C. The Housing Finance System: Achievements, Problems, and Policy Issues 1. Measures of Performance 37. The establishment and growth on a sound financial and technical basis of the BNH, the COHABs, the housing cooperatives, and the many financial institutions composing the SBPE, as outlined above, are in themselves major achievements of Brazilian housing policy. In this section, however, the emphasis is on performance measured in terms of housing units financed within broad market categories. Particular attention is focused on popular housing. 38. To facilitate the analysis, the period 1964-77 has been broken down into four subperiods, each representing a different phase in the development of the BNH and national housing policy. During the first subperiod, 1964-67, the institutional infrastructure was being established and the FGTS had not become operative as a major source of finance. In the subperiod 1968-72 FGTS was fully operative, but the National Popular Housing Plan (Plano Nacional de Habitacao Popular - PLANHAP), a major effort to reorient popular housing policy, had not yet been instituted. The years 1973 and 1974 began within the PLANHAP framework, but it rapidly became evident that PLANHAP's ambitious goals were not being achieved. These are years of crisis in the popular housing market. The underlying problems and the measures taken to deal with them are discussed in parts 2 and 3 of this section. Finally, the period 1975-77 began with another attempt to reorient national housing policy, particularly in the area of popular housing, in a way consistent with the Government's stated concern for implementation of a social policy favoring the lower income groups. 39. The overall operations of the SFH are broken down into loans for popular housing (including serviced lots and housing improvements), middle- income housing, SBPE-financed housing (which has covered essentially the upper middle- and higher-income brackets), loans to final users of construction materials (Programa de Financiamento ou Refinanciamento do Consumidor de Materias de Construcao - RECON - which has served essentially the same income classes as the SBPE), and the Program of Support to Development of Economic - 25 - Poles (Programa de Apoio ao Desenvolvimento de Polos Economicos - PRODEPO), a relatively small program to provide a variety of housing in new economic growth poles such as the petrochemical complex outside Salvador, Bahia. 40. The definition of popular housing here is all housing financed through the COHABs (by far the largest number of loans), the program for Financing to Complement Housing (Programa de Financiamento a Complementacao da Habitacao - PROFICO) and predecessor programs which were designed to upgrade existing low-income housing, and the Program for Financing of Urbanized Lots (Programa de Financiamento de Lotes Urbanizados - PROFILURB), a sites and services program initiated in 1975. All popular housing so defined is financed by BNH's Social Operations Portfolio (Carteira de Operacoes Sociais - COS). As will be explained in greater detail below, the reader should bear in mind that over time the income and loan limits for what is considered popular housing have risen, most notably in 1975, therefore a substantial portion of the "popular" housing loans in the 1975-77 period would have been in the medium-income bracket had they been made before 1975. Medium- income housing is considered to be all housing financed through all programs of BNH's Housing Program Portfolio (Carteira de Programas Habitacionais - CPH), the most important of which has been the housing cooperatives, but this category also includes loans through the Social Welfare Institutes (Institutos de Previdencia Social) and the private mortgage market. Some of these loans are probably in the upper-income class, but on the other hand some SBPE and RECON loans were in the middle-income bracket. 41. Table 10 shows the total number of loans made in each of the cate- gories mentioned in each of the four subperiods as well as for the entire period 1964-77. There may be some double counting of units in the case of RECON and SBPE loans, since RECON loans for construction materials made to builders sometimes have not been netted out of the total of SPBE loans made to final owners, so if anything the total of non-popular housing units financed may be overstated. 42. Table 10 also compares the number of housing loans made with goals for housing units to be financed as stated in national economic plans and BNH planning documents (always using the most recent goal when more than one source was available for a given year) as well as with the estimated increase in the number of urban households as calculated from census data and the "baseline" population projections described in Annex I of this report. It should be noted that each housing loan does not necessarily correspond to an addition to the stock of housing, since some loans may be for upgrading existing housing. Furthermore, since there is generally about a two-year delay between the financing and completion of a housing unit, the comparison between housing loans made and increases in the number of urban households is a less than perfect indicator of the relation between housing supply financed via the SFH and potential demand for short periods. Finally, this measure takes no account of any demand from rural areas, the need to overcome the urban housing deficit existing in 1964, nor the need to replace some dilapidated units. For these reasons the denominator of the ratio of housing loans to the increase in the number of urban households underestimates the potential demand for urban housing in Brazil while the numerator overestimates the supply of such housing effectively financed by the SFH. As a result, this measure if anything overstates SFH impact on the urban housing problem in Brazil. Table 10: PERFORM!ANCE OF THE HOUSING FINANCE SYSTEM BY PERIODS IN THE DEVELOEMENT OF THE BNH, 1964-1977 1964 - 67 1968 - 72 1973 - 74 1975 - 77 1964 - 77 Total Avearg Per Total Average Per Tota1 Average Per otal Average Per Total _' Average Per Year Year Year Year Year Thousands 3f HouSing Loans Popular Uousing, BHH '1 76.6 19.2 155.9 31.2 26.5 13.3 262.7 87.6 521.6 37.3 Middle Incoee Housing. BhH - 31.4 7.9 212.4 42.5 40.0 20.0 90.2 30.1 373.9 26.7 Brazilian Savings and Loan System (SBPE) Housing 30.4 7.9 278.3 55.7 136.9 68.5 229.4 76.5 676.0 48.3 Construction Materials Lnane (RECON), BNH .4 .1 49.3 9.9 41.5 20.8 49.4 16.5 140.7 10.1 Development Pole Housing (PRODEPO) - - - - - - 25.9 8.6 25.9 1.9 Total -1 139.9 35.0 695.9 139.2 244.9 122.5 657.4 219.1 1738.1 124.2 Thousand. of Housing Units Popular Housing Units Goal 150.0 37.5 334.0 66.8 92.0 46.0 839.0 279.7 1415.0 101.1 Total Housing Units Goal 250.0 62.5 805.0 161.0 414.0 207.0 2261.0 753.7 3730.0 267.9 Thousands of Households Increase in Nunber of Urban Households 1744.4 436.1 . 2503.5 500.7 1016.6 508.3 1699.2 566.4 6963.7 497.4 Petcentages Popular Housing Loans as 7. of all Hoosing Loans 54.8 22.4 10.8 40.0 30.0 Popular Housing Loans as % of Popular Housing Coal 51.1 46.7 28.6 3103 36.9 Total Housing Loane as % of Goal 56.0 86.4 59.2 29.1 46.6 Total Housing Loans as 7. of Increase in Urban Households 8.0 27.8 24.1 38.7 25.0 a/ Totals may not equal sum because of rounding. b/ Includes COHABs, PROFICO, and PROFILURB (Social Operations Portfolio). The definitions of popular housing loans vas broadened in 1975 to include sany loans voich -ould previously have fallen in the middle income category. See tent. c/ Includes Housing Cooperative., Institutes, 'Mortgages on beusing by private builders, and housing constructed by economic enterprises for their -orkers (Housing Progra Portfolio). Sources: Number of Housing Loana; BNH. Heusing Coals: Programa de Acan Econorica do Govermo 1964-66, Piano Entrategico 1968-70, BNH Orcamento Pl-rianual 1970-73, BNH Ore-mento Plorianual 1973-75, BNH Orcamento Plurianual 1976-78, BHH Oreanat Pluriannal 1977-79. When amer than one goal was available for a given year, the s.aller goal -as chosen. For the years 1964 and 1967 no goals vera found. Increase in nuaber of Urban House- holds: Calculated from Baseline: desographie projections (sea Annex I). Table 11: VALUE OF BNH HOUSING LOANS 3Y PERIODS, 1964-1977 1964 - 67 1968 - 72 1973 - 74 1975 - 77 1964 - 77 Total AeaePgTo tal Aotal Average Per Total Average Per Totala/ Average Per Year Year Year Year Year Thousainds of UPC Popular Housing-/ 10,460 2,615 30,195 6,039 5,645 2,823 81,337 27,112 127,637 9,117 Middle Level Housing 13,076 3.269 85,898 17,180 21,951 10,976 93,254 31,085 214,180 15,299 Support tor the Brazilian Savings and Loan System (SBPE) 6,799 1,700 54,494 10,899 38,717 19,359 94,890 31,360 194,900 13,921 Construction Macerials (RECON) 419 105 20,878 4,176 34,810 17,405 52,981 17,660 109,088 7,792 Development Pole Housing (PRODEPO) d/ - - - - - - 30,417 10,139 3u,417 2,173 Total a/ 30,754 7,689 191,465 38,293 101,124 50,562 352.879 117,626 616,222 48,302 Percent Popular Housing as % of Total 34.0 15.8 5.6 23.0 18.9 a/ Totals may not equal sum of components due to rounding. b/ For detinition see note b, Table 10. C/ For definition see note c, Table 10. d/ Includes infrastructure. Table 12: AVERAGE VALUE OF BNH HOUSING LOANS BY TYPE OF LOAN AND PERIODS, 1964-1977 (In UPC) 1964-67 1968-72 1973-74 1975-77 1964-77 Popular Housing a/ 137 194 213 310 245 Middle Level Housing 416 404 549 1034 573 Support for the Brazilian / Savings and Loan System (SBPE) - 217 196 283 414 287 Construction Materials Loans (RECON) 1048 423 839 1072 775 Development Pole Housing (PRODEPO) d/ _ - - 1174 1174 a/ For definition see note b, Table 10. b/ For definition see note c, Table 10. c/ BNH support per SBPE loan. Does not include resources loaned directly by SBPE institutions. d/ Includes infrastructure. Source: Calculated from BNH data. - 29 - 43. Table 11 presents the value of housing loans of each category, measured in UPC for the same periods, and Table 12 shows the average value per loan as derived from Tables 10 and 11. Regional breakdowns of loans made, total value, and average loan value, using the three major regions defined for other parts of this report (Northeast, Southeast, and Frontier), are presented in Tables 13, 14 and 15 respectively. These last three tables give figures for the entire 14-year period. 44. A number of conclusions can be drawn from an examination of Tables 10-15. The most salient are the following: (a) The number of popular units financed as a percentage of all units financed fell from 54.8% in 1964-67 to 10.8% in 1973-74, rising to only 40.0% in the period 1975-77 despite the increase in the qualifying maximum family income from 3 to 5 minimum salaries which occurred at the beginning of this last period. For the entire period 1964-77 only 30.0% of SFH financed housing units fell in the "popular" category despite the original mandate of the BNH to give special emphasis to popular housing. (b) BNH-SFH goals have consistently been well in excess of achieve- ments, by a ratio of over 2.1 to 1 throughout the entire period for all housing loans, and 2.7 to 1 for popular housing loans. In the case of popular housing the goals became increasingly unrealistic over time except for a slight improvement in 1975-77, the period during which the definition of "popular" housing was relaxed. (c) If the total number of SFH housing loans as a percentage of the increase in urban households is taken as a measure of the extent to which the SFH has satisfied potential demand (and here the caveats mentioned above must be kept in mind), there has been steady progress over the period 1964-77 except during the crisis years of 1973-74. Over the whole period the number of loans came to only 25.0% of the increase in households, reaching a maximum of 38.7% in the years 1975-77. Assuming that all urban households had some form of permanent shelter, during the period 1964-77 3 out of 4 homes have been financed outside the SFH if at all. 1/ (d) Roughly the same pattern emerges from Table 11. The value of "popular" housing loans as a percentage of all loans is less, given that the average value of loans, as shown in Table 12, is lower than for other categories. The rise in the average value of popular housing loans from 137 UPC in 1964-67 to 310 UPC in 1977 is note- worthy. There was also a sharp rise in the average value of middle income housing loans in the period 1975-77. The low average value 1/ These homes include favela dwellings; self-constructed dwellings on lots in private subdivisions often purchased on an installment basis; and dwellings financed by the sale of assets, informal loans within extended families, and/or constructed little by little from current earnings of their owners, to mention only some of the most common modes of obtaining housing outside the SFH. No reliable statistics are available on a national basis regarding the distribution of modes of finance and con- struction outside the SFH. Table 13: PERFORMANCE OF THE HOUSING FINANCE SYSTEM BY REGIONS AND PERIODS IN THE DEVELOPMENT OF THE BNH, 1964-1977 NORTHEAST SOUTHEAST FRONTIER UNALLOCATED BRAZIL - Thousands of Loans Popular Housing, BNH b/ 158.9 281.1 81.6 - 521.6 Middle Income Housing, BNH - 58.1 260.2 41.9 13.8 373.9 Brazilian Savings and Loan System (SBPE) Housing 69.0 566.3 40.7 - 676.0 Construction Material Loans (RECON), BNH 19.6 115.6 5.5 140.7 Development Pole Housing (PRODEPO), BNH .2 13.9 11.8 - 25.9 Total a/ 305.9 1237.1 181.4 13.8 1739.1 Thousands of Households o Increase in Number of Urban Households 1022.3 5337.8 603.6 - 6963.7 Percentages Popular Housing Loans as % of Total 51.9 22.7 45.0 30.0 Housing Loans Total Housing Loans as % of Increase in Number of Urban Households 29.9 23.2 30.1 25.0 a/ Totals may not equal sum of components due to rounding. b/ For definition, see note b, Table 10. c/ For definition, see note c, Table 10. Source: Number of housing loans; BNH Increase in number of Urban Households, calculated from "Baseline" demographic projections (See Annex I). Table 14: VALUE OF BNH HOUSING LOANS BY REGIONS, 1964-1977 NORTHEAST SOUTHEAST FRONTIER UNALLOCATED BRAZIL Thousands of UPC bI Popular Housing7 34,669 76,046 16,922 127,637 Middle Level Housingc 30,745 152,222 25,821 5,392 214,180 Support for the Brazilian Savings and Loan System (SBPE) 22,657 150,189 21,870 184 194,900 Construction Materials (RECON) 18,373 85,948 4,766 1 109,088 Development Pole Housing (PRODEPO 4/ 1,950 20,364 8,103 30,417 Total 108,395 484,769 77,482 5,576 676,222 Percent Popular Housing as % of Total 32.0 15.7 21.8 0.0 18.9 a/ Totals may not equal sum of components due to rounding. b/ For definition see note b, Table 10. c/ For definition see note c, Table 10. d/ Includes infrastructure Source: BNH and mission calculations. Table 15: AVERAGE VALUE OF BNH HOUSING LOANS BY TYPE OF LOAN AND REGION, 1964-1977 (In UPC) NORTHEAST SOUTHEAST FRONTIER UNALLOCATED BRAZIL Popular Housing -/ 218 271 207 - 245 Middle Level Housing-/ 529 585 616 391 573 Support for the Brazilian Savings and Loan System (SBPE) 328 265 537 - 287 Construction Materials (RECON) 937 743 867 248 775 Development Pole Housing (PRODEPO)A/ 9750 1465 687 - 1174 a/ For definition see note b, Table10. 'h/ For definition see note c, Table 10. c/ BNH support per SBPE loan. Does not include resources loaned directly by SBPE institutions. d/ Includes infrastructure. Source: Calculated from BNH data. - 33 - of loans shown for the SBPE in Table 12 reflects the fact that these figures include only BNH funds relent through the SBPE and thus exclude the value of resources mobilized and lent by the SBPE institutions themselves. (e) When the measures of the degree to which SFH-financed housing has supplied the increase in housing demand are broken down by major regions (see Table 13), it appears that the Northeast has done relatively better than the Frontier and Southeast in terms of popular housing (31.3% v. 28.2 and 24.2%, respectively). The Northeast also led in terms of popular housing loans as a percentage of all housing loans (51.9% v. 45.0% in the Frontier and 22.7% in the Southeast). The same pattern emerges in the total value of loans (Table 14). As for the average value of popular housing loans the Frontier had a slightly lower value (207 UPC) than the Northeast (218 UPC) and as might be expected the Southeast had a higher average (271 UPC). On the whole the regional analysis suggests that despite the fact that over 70% of both the number and the value of housing loans was con- centrated in the Southeast, the Northeast and Frontier had a greater portion of the increase in housing needs satisfied through the SFH during the period 1964-77. 2. Major Factors Affecting Popular Housing Finance Performance Four basic institutional and economic constraints underlie the relatively poor performance of the BNH in financing popular housing despite the rapid expansion of the SFH. These are (a) Brazil's large proportion of poor families, which means that a majority of urban families have not been able to afford the kind of "popular" housing financed by the BNH, at least until the advent of the PROFILURB sites and services scheme; 1/ (b) the fact that BNH does not receive any major funds for which it does not have to pay monetary correction and at least 3% interest (even though this is somewhat below the 6% paid on SBPE passbook savings accounts) means that there are only limited possibilities for direct BNH subsidization of low-income housing via charging lower than average interest rates to low-income borrowers; (c) the rise in land and construction costs considerably faster than the general price index, the average value of the UPC or the minimum wage, which has increased the real cost of housing; and (d) deficiencies in local level implementation capacity. In the following paragraphs each of these constraints is analyzed. In the next section government policy responses to the con- straints, largely implemented through the BNH, are reviewed. (a) Poverty Table 16 shows the percentage of families residing in urban areas receiving monetary incomes (measured in terms of Rio de Janeiro monthly minimum wages) within certain ranges as determined by the 1972 PNAD, conducted during the fourth quarter of that year. The sample covered all of Brazil except the Frontier but including the Federal District. A simplified regional breakdown 1/ For a description of the PROFILURB program, see paragraphs 67 - 83 of this annex. TABLE 16 MONETARY INCOME OF URBAN FAMILIES ' BY SIZE AND REGION, 1972 Monthly Monetary Income Northeast Southeast & Federal District Totall,c (Minimum Wages) Families x Families Z Families % (thousands) (thousands) (thousands) 1 or less] 1,208 48.5 1460 16.2 2668 23.2 2 or less Cumulative 1,833 73.6 3748 41.3 5548 48.3 3 or lessj 2,077. 83.3 5358 59.6 7435 64.7 3 - 5 167 6.7 1508 16.8 1675 14.6 5 - 7 69 2.8 715 7.9 784 6.8 7 - 10 63 2.5 557 6.2 620 5.4 10 - 15 56 2.3 425 4.7 482 4.2 Z 15 - 30 40 1.6 302 3.4 342 3.0 More than 30 10 0.4 102 1.1 112 1.0 Without Income 3 0.1 6 0.1 8 0.1 No Answer 8 0.3 23 0.3 32 0.3 Total- 2,492 100.0 8997 100.0 11489 100!0 Source: Calculated from IBGE, PNAD - 2. 4 Trimestre de 1972, Volume III, page 40 and Volume IV, page 42. Notes: a/ The technical definitions of the term "Family" is a "consumption unit", that is "the group of members of a Family who effectively make up and spend the same family budget, that is one in- cluded in the same domestic budget." The minimum wage is that prevailing in Rio de Janeiro. b/ Total of Northeast, Southeast, and Federal District. c/ Total may not equal sum of component elements due to rounding. - 35 - is therefore presented which divides the area covered by the survey into onily two regions, the Northeast and the augmented Southeast (including the Federal District). The global figures show that in 1972, 64.7% of urban families had monetary incomes totaling three Rio de Janeiro minimum wages or less. There were substantial regional differences as might be expected, the corresponding statistics being 83.3% in the Northeast and 59.6% in the augmented Southeast. Returning to the global figures, 22.3% of urban families had incomes of at most one minimum wage and 48.3% earned at most two minimum wages. Just under 80% of the urban families had incomes of five minimum wages or less. But since these data are in terms of the highest minimum wage prevailing in the country, they overstate the number of families who had any given multiple of the regional minimum wage outside of Rio de Janeiro, Sao Paulo, and Brasilia. The lowest minimum wage, that in the smaller Northeastern capitals, was 68% of the Rio de Janeiro minimum wage. The experience in the first phase of PLANHAP, when a total of only 27,000 popular housing units (for families with up to three minimum wages) were financed in all Brazil despite the announced goal of two million in a ten-year period, suggests that BNH financed housing was simply priced out of reach of roughly half of all urban families, the very population which had been singled out for priority attention under PLANHAP. Two examples will suffice to indicate that this statement probably remains true even after the revisions in PLANHAP terms in 1975. The cheapest standard housing unit for which detailed construction costs are available from BNH is a roughly finished two bedroom single family house with a constructed area of 25.65 square meters. 1/ In November 1976 direct construction costs alone (not including the costs of supervising construction, interest during construction and profits for the builder) for such a house in Recife were 191.85 UPC. 2/ Even assuming a loan of only 200 UPC (probably low for the Northeast as well as Recife since even a minimum size serviced lot would probably cost about 30 UPC), 3/ the total first monthly payment on this loan, which would bear an 1/ The Pernambuco COHAB is constructing a 35 m2 one bedroom house estimated to cost 276 UPC in mid 1978. A similar 33 m2 house costing 261 UPC is being constructed by the Bahia COHAB. A 27 m2 "embryo" house constructed by the Pernambuco COHAB cost 212 UPC and a 23 m2 embryo constructed by the Bahia COHAB cost 176 UPC, all in mid 1978 prices. 2/ This cost does not include the investment costs of providing water supply, sanitary sewerage, and electrical generation and distribution capacity sufficient to connect these serviced lots to these utilities. Such investment costs are normally recovered through the utilities rate structures. 3/ Interest and supervision charges on a sample of recent COHAB projects totaled about 8% of construction costs. Assuming a 10% profit for the builder, direct construction costs should be increased by 19%. A recent dissertation suggests that these indirect costs may be as much as 44% of direct construction costs. See David Michael Vetter, "Low Income Housing Policy for Development: An Evaluation of the Brazilian Experience," - 36 - interest rate of 1% for a 25 year mortgage would be 1.11335 UPC including the required insurance, Administration and Collection Fee, and Community Support Fee. A family of five members or less with an income of two Recife minimum wages would not qualify for this loan--this size family would have to have a minimum income over 2.23 Recife minimum wages under the rules governing PLANHAP loans. 73.8% of urban families in the Northeast reported having incomes of two minimum wages or less in 1972. 1/ The same house in a working class area of Sao Paulo, also in November 1976, cost 229 UPC to build. Adding another 71 UPC as an estimate of the cost of a serviced lot, this house could be purchased for about 300 UPC, excluding indrect construction costs. 2/ A 25 year mortgage for this value carries an interest rate of 2.6%. The total first monthly payment would be 2.158 UPC. Under the PLANHAP rules, a family income of 2.92 Sao Paulo minimum wages would be required to assume this mortgage. In Sao Paulo 31.8% of urban families in 1972 had incomes of two Sao Paulo minimum wages or less and 51.6% had incomes of three Sao Paulo minimum wages or less. 3/ 48. Given these rough estimates, which are deliberately on the low side because indirect construction costs are excluded, and since according to the 1970 census 30% of urban families have more than five members (under PLANHAP rules larger families must have larger incomes to qualify for the same loan), probably about two-thirds of Brazilian urban families could not afford to purchase this minimum commercially constructed house under the present set of rules. (b) The Internal Financial Viability of the National Housing Bank 49. The two principal purposes of the FGTS are effectively in conflict. The FGTS is to serve as an inflation-proof forced savings instrument to accumulate reserves on which Braziian workers can draw in times of need. But it is also to provide the principal source of funds for the BNH to be used to finance popular housing and other forms of urban development benefitting the low-income population. In the first place even in the absence of with- drawals, FGTS accounts have lost real value as measured against the general price index in every year since 1972 except 1975, as is shown in Table 5. Secondly, the basic 3% rate of interest is already below the rate offered on voluntary savings accounts by a full 3 percentage points. An attempt was made during the period 1973-75 to reduce the realized rate of return on FGTS deposits further in order to increase the financial viability of subsidizing popular housing loans. During this period a system of annual rather than 1/ Recently available data from the 1976 PNAD show that in that year 57.4% of urban families in the Northeast had incomes equal to twice the highest minimum wage prevailing in the country or less. Since the minimum wage prevailing in Northeastern cities was between 71% and 78% of the highest minimum wage, the number of urban families in the Northeast below two regional minimum wages was probably less than 50% in 1976. 2 2/ In mid 1978 a 28 m one bedroom house being produced by the Sao Paulo COHAB cost 404 UPC. 3/ These data are from the 1972 PNAD. The 1976 PNAD data indicate that 17.4% of urban families in Sao Paulo state had incomes of two Sao Paulo minimum wages or less in that year. The published data does not allow a direct calculation of the percentage of families with three minimum wages or less. - 37 - quarterly monetary correction and interest capitalization prevailed. This arrangement probably had negative distributional effects on FGTS depositors, since the available data show that employees with less than two minimum wages were, directly or indirectly, responsible for over half the value of with- drawals in May 1973. 1/ In any case the vast majority of housing loans during this period went to middle- or upper-income earners. Employees making with- drawals after January 1 forfeited monetary correction and interest for up to 12 months. If these withdrawals had been for the purpose of acquiring a hous- ing unit financed through the BNH, the probable negative distributional effects might be neutralized. But in fact during the years 1973-75 only an average of 0.6% of the value of all FGTS withdrawals was used to finance purchase of housing through the entire SFH. 2/ 50. Thirdly, assuming subsidies are necessary in order to bring popular housing within financial reach of the low-income population, if the system is to be internally viable in the long term and maintain the value of FGTS deposits, the only way to subsidize low-income housing without introducing subsidies from outside the FGTS-BNH system is to charge interest rates higher than the total cost of funds and administration to the BNH (estimated at 5%) on loans for purposes other than the construction of popular housing. This permits a system of cross-subsidization of interest rates, which is what the BNH has in fact done both within the housing program and by moving in the direction of urban development. Reliance on cross-subsidization within the FGTS-BNH system rather than obtaining subsidies from outside the system was a second best solution for a politically weak sector which had trouble obtaining grant funds on a regular basis. But the higher the subsidies for popular housing financed in this fashion, the greater must be the volume of lending at higher rates outside the popular housing sector. Thus, ironically, this means of subsidizing popular housing encourages BNH diversification away from popular housing finance in order to expand popular housing finance. (c) The Rising Relative Price of Housing Since 1972 51. Relative to the general price index, the index of construction costs prepared by the BNH rose sharply over the period 1973-75. As shown in Table 17, taking the value of each index and their ratio in 1970 as 100, after fall- ing in 1971 and 1972, the ratio of the construction cost index went from 94.0 in 1972 to 121.3 in 1975, and increase of 29%, then fell to 119.5 in 1976. During the same period, the ratio of the housing cost index to the average value of the UPC rose even faster, from 91.7 in 1972 to 125.7 in 1975 or 31%, and the rise continued to 136.0 in 1976. The ratio of the construction cost index to an index of the minimum wage rose 29.2% between 1972 and 1975. According to BNH sources, urban land prices rose even faster than construction costs, though no good index of urban land prices is available. 52. BNH officials assert that the rapid rise in construction costs beginning in 1973 was caused by the large increase in public works during the last year of the Medici administration which created excess demand for con- struction materials and the services of construction companies, thereby pushing up production costs for popular housing. The rise continued in 1974, however, 1/ Wanderly J.M. de Almeida and Jose Luiz Chautard, FGTS: Uma Politica de Bem-Estar Social, IPEA/INPES Colecao Relatorios de Pesquisa N 30 (Rio de Janeiro: IPEA/INPES, 1976), p. 97. 2/ Ibid, p. 92, Table I11.5. TABLE 17 COMPARISON OF BNH CONSTRUCTION COST INDEX, GENERAL PRICE INDEX, VALUE OF THE UPC, AND MINIMUM SALARY, 1970-76 (1) (2) (3) (4) (5) (6) (7) BNH Construction General Price Index of Index of Cost Index Index UPC Value Minimum Wage 1 -;- 2 1 - 3 1 4 Year (1973 100) (1973 = 100) (1973 = 100) (1973 = 100) x 100 x 100 x 100 1970 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1971 115.0 120.4 119.9 120.4 95.5 95.9 95.5 1972 132.4 140.9 144.4 143.9 94.0 91.7 92.0 1.973 161.9 162.2 164.6 168.3 99.8 98.4 96.2 1974 246.3 208.7 196.9 200.9 118.0 125.1 122.6 1975 323.2 266.5 257.1 271.9 121.3 125.7 118.9 1976 450.1 376.5 331.0 390.0 119.5 136.0 115.4 Sources: BNH Construction Cost Index - BNH/APC "Indicadores Economicos e de Custos da Construcao Habitacional" 3:2 (Abril-Junho 1977), Table 1.1. General Price Index, Domestic Supply - Conjuntura Economica, September 1977. Index of LTC Value - Calcualted from date in Table 1. Minimum Salary Index - Calculated from date in IBGE, Anuario Estatistico do Brazil. 1976, p. 476. The average minimum salary for each year is an average of the minimum salary prevailing each month in Sao Paulo. - 39 - perhaps owing to the boom in speculative luxury housing which occurred about this time. In any case, concerning the fact of the relative price increase for housing (as opposed to its explanation), there is little doubt. (d) Deficient Local Level Implementation Capacity 53. Within the SFH, the BNH acts as a financial wholesaler, while the responsibility for design and execution of popular housing projects lies with state and local governments or agencies under their control (the COHABs). Financial and technical weaknesses of the COHABs as well as lukewarm support from and the financial difficulties of state and local governments have been a key bottleneck impeding better popular housing performance. Before the 1973-75 reforms (to be described in the next section) many COHABs were not financially viable and did not have technical capacity to carry out large- scale operations. 3. Policy Response to Factors Affecting Popular Housing Finance Performance 54. Government housing policy, as set forth in planning documents and executed largely through the BNH, has not been unresponsive to the problems discussed in the previous section. In the following paragraphs the principal measures which have been taken to bring "popular" housing within financial reach of the low income population and improve the implementation capacity of the COHABs are reviewed. Two points should be emphasized at the outset. First, this summary cannot do justice to the extreme complexity and frequent changes in BNH programs and the terms under which their "products" are sold in the low-income housing market. Second, while the SFH performance fell short of the need and effective demand for shelter, it compares favorably with the performance of other countries at a similar level of development in Latin America. To develop a large-scale., practically self-financing housing delivery system is a long-term task, and what has been achieved so far, especially in terms of institutional development, is impressive. 55. Since 1975 there have been encouraging new developments which hold great promise for bringing BNH-financed housing to the urban poor. In the PROFILURB sites and services program and the new program to finance the purchase of construction materials (Financiamento para Construcao, Ampliacao, ou Melhoria de Habitacao de Interese Social - FICAM), BNH now has available two significant instruments which together represent a new and potentially viable approach to low-income housing emphasizing the provision of basic water, sewerage and electricity services to prepared housing sites on which self-help housing can be constructed. This new approach recognizes that traditional "popular" housing is priced out of reach of families with incomes of two minimum wages or less, that the provision of basic sanitation should have first priority on available resources, and that the majority of the urban poor are already involved in self-help housing construction without any assistance from the official sources. Another major breakthrough has been the granting of housing subsidies from general federal tax revenues on a progressive basis. Considerable attention, therefore, will be given to these new program in this section. - 40 - 56. In general, the policy responses of the Brazilian housing authori- ties to the problems analyzed in the previous section may be divided into five basic categories: (a) changing the terms of loans (maximum loan size, interest rates, maximum family income, repayment period, and system of monetary correction), (b) attempting to monitor and control construction and land costs,(c) improving the financial viability and technical capability of the COHABs, (d) introducing subsidies from outside the BNH-FGTS system, (e) introducing new programs to support self-help housing construction. (a) Changing Terms for Popular Housing Loans 57. Table 18 summarizes the terms for housing loans to final consumers applying in the four market brackets recognized by BNH during the period 1969-74, namely "popular, economy, middle, and upper." The changing defini- tions of these categories are also summarized in Table 18. In general over this period, the maximum income to qualify for "popular" market loans fell when measured in UPC while the maximum loans remained constant at 200 UPC. The "popular market" does not coincide with the range of housing financed through the COHABs, which in 1972 were able to build and finance housing units worth up to 320 UPC and therefore by BNH definition penetrated well into the "economy" market. The rate of interest charged by the BNH to its financial agents (including the COHABs) in the popular market fell, however, from 5-7% before October 1969 to 1-3% in the period October 1971-December 1974. The maximum loan period also rose from 20 to 25 years. In 1970, the system of monetary correction applying to popular housing loans was modified to permit an annual correction 60 days after the announcement of the new minimum wage rather than quarterly adjustments. 58. All of these changes had the effect of reducing the monthly install- ments for any given size of loan. However, as can be seen from Table 19, which also covers the period 1969-74, the total value of loans made in the "popular" market fell steadily after 1970 both absolutely and relatively. This was also true in the "economy" market after 1971. The opposite trend, namely increasing absolute and relative value of loans, prevailed in the "middle" and "upper" income markets. Thus, the changes introduced were insufficient to overcome the negative impact of the severe constraints dis- cussed earlier. The lower interest rates and longer payback periods intro- duced in 1971 were in part made possible by the less favorable conditions applying to FGTS deposits as explained in a footnote on page 16. Despite this, a major government study of the FGTS concluded that the need to maintain the rate of return on FGTS deposits was a major motive in leading the BNH toward applying a greater proportion of its resources in programs with higher interest rates and/or shorter repayment periods such as the RECON construction materials loans, loans in the medium- and high-income housing markets, and programs outside the housing sector. 1/ 1/ Almeida and Chautard, op. cit., pp. 197-219. - 41 - Table 18: LOAN TERMS FOR FINAL BORROWERS OF HOUSING PROGRAMS. 1969-1974 Markets Terms Popular Economy Middle Upper 1. Limits on BNH Loans, in UPC a. Up to October 1969 Up to 200 200 -1400 400 - f720 720-j 2250 b. October 1969-October 1971 Up to 200 200 -4400 400 -- 720 720 -1 2250 c. October 1971-December 1974 Up to 200 200 - 1400 400 -1900 900 --42250 2. Classes of Minimum Necessary Family Income a/ in UPC a. Up to October 1969 Up to 6.20 6.20--417.24 17.24 -143.92 Above 43.92 b. October 1969-October 1971 Up to 5.76 5.76 -17.24 17.24-1443.92 Above 43.92 c. October 1971-December 1974 Up to 4.68 4.68-116.00 16.00-f50.00 Above 50.00 3. Maximum Percentage of Family Income Permitted for Loan Payments a. Up to October 1969 b/ 27 25 25 b. October 1969-October 1971 b/ 25 25 25 --- c. October 1971-December 1974 c/ 20 20-25 25 Up to 35 4. Annual Interest Rate in Addition to Monetary Correction (Z) a. Up to October 1969 5r-4 7 8 -H o 10 10 d/ b. October 1969-October 1971 4-H 6 7F-110 10 l1l-413 e/ c. October 1971-December 1974 1 - 13 4 -i8 9 -l10 10 1-12 f/ 5. Maximum Loan Period (years) a. Up to October 1969 20 20P-415 121-4 8 10 b. October 1969-October 1971 20 20t--115 12 - 8 10 c. October 1974-December 1974 25 25 251-115 15 Notes: a/ Calculated based on maximum commitment of family income permitted by BNH. b/ Payments according to the "Tabela Price" Amortization System. c/ Payments according to the Constant Amortization System (SAC). In this case, maximum commitment is determined by the first payment as a percentage of family income. d/ For loans in the 720 - 1020 UPC range. e/ Effective interest rate. f/ Effective interest rate for loans in the range 900 - 1800 UPC. For loans above 1800 UPC, the effective interest rate was freely fixed by the SBPE. Source: Wanderly J.M. de Almeida and Jose Luiz Chautard, FGTS: Uma Politica de Bem-Estar Social, Colecao Relatorios de Pesquisa No. 30 (Rio de Janeiro: IPEA/INPES, 1976) Table IV-7, pp. 120,121, Original Source: BNH regulations. Taible 19: DISTRIBUTION OF BNII HOUSING LOANS BY ArLTKET, 1969-19741-'/ b/_ Loans Popular Economy Middle Upper Total 1. Value, in 1000 UPC 1969 7,323 16,283 9,409 6,863 39,879 1970 9,835 17,898 8,243 6,066 42,043 1971 2,618 19,299 10,238 7,872 40,027 1972 1,857 13,898 13,143 7,782 36,680 1973 1,924 9,492 18,342 12,876 42,634 1974 679 5,178 17,877 14,774 38,508 2. Percentage Distribution 1969 18.4 40.8 23.6 17.2 100.0 1970 23.4 42.6 19.6 14.4 100.0 1971 6.5 48.2 25.6 19.7 100.0 1972 5.1 39.7 35.8 21.2 100.0 1973 4.5 22.3 43.0 30.2 100.0 1974 1.8 13.4 46.4 38.4 100.0 Source: Wanderly J. M. de Almeida and Luiz Chautard, FGTS: Uma Politica de Bem-Estar Social, Colecao Relatorios de Pesquisa No. 30 (Rio de Janeiro: IPEA/INPES, 1976) Table IV-9, p. 126. Original source: BNH/APC. Note: a/ The distribution of loans under the Program of Stimulus to the SBPE and the RECON program were estimated on the basis of the distribution of all loans by the SBPE. b/ All loans made through the Programs of Popular, Economy, Middle and Upper Urban Markets; Program of Stimulus to the SBPE and the Program of Loans for Consumers of Building Materials (RECON). - 43 - 59. The crisis in popular housing so manifest in 1973 and 1974 resulted in major modifications in PLANHAP in January 1975. These modifications were the fruit of a thorough review of housing policy conducted by the Geisel administration after it took office in 1974. One of the conclusions of this analysis was that the increase in land and construction costs faster than the increase in the general price level had made the previous PLANHAP maximum loan (320 UPC) and maximum family income (three minimum wages) limits unrealistically low. Beginning in 1975 the new limits for PLANHAP "popular" housing loans were 500 UPC maximum loan value and five minimum wages or 22 UPC (whichever is lower) maximum family income. 1/ In this way the problems of rising housing costs and poverty were "overcome" simply by redefining "popular." At the same time the maximum percentage of family income constituting the first total monthly payment 2/ (previously 25% independent of family size) was made a decreasing function of family size and an increasing function of family income. The percentage now varies from 5% (family size ten or more, monthly income 8 UPC or less) to 25% (family size five or less, monthly income 20-22 UPC). These measures simply recognized the fact that larger and poorer families can afford to pay smaller shares of their income on housing, despite their equal or greater needs for this service. Thus, in addition to increasing the probability that families qualifying for loans would be able to keep their payments up-to-date, the new regulations achieved this result by effectively downgrading the quality of housing for which larger and poorer families could qualify and in some cases excluding them entirely from this particular program. 60. Interest rates (i) charged by the BNH to the COHABs were made a continuous function of the loan value according to the following rules: Loans of 200 UPC and less 0.0% per year Loans over 200 UPC up to i = (loan value - 2.03 and including 500 UPC ( 100 UPC ) % per year Under these rules the maximum interest rate for popular housing to the COHABs under PLANHAP was 3%. The interest rate on loans from the COHABs to final borrowers is given by the rules: 1/ The ratio of the minimum wage to the value of the UPC varies over the course of the year since the UPC is revalued each quarter and the minimum wage is increased only once a year on May 1. 2/ The total monthly payment includes the following components: (a) amor- tization of principal, (b) interest, (c) comprehensive insurance, (d) Collection and Administrative Fee, and (e) Fee for Community Support. The latter two fees go to the COHABs. - 44 - Loans of 200 UPC and less 1% Loans of over 200 UPC up to i = 1.66 loan value - 2.3 % thus making the maximum interest rate to final borrowers 6%. This spread between borrowing and lending rates for COHABs increases with the size of the loan in the range 200-500 UPC from 1% to 3%. This provides an incentive to the COHABs to build more expensive housing units. But these changes represented a significant downward shift in the interest rate structure to the COHABs as indicated in Table 20. (b) Efforts to Control Construction and Land Costs 61. The BNH has sought to obtain more precise information to monitor construction costs. Beginning in December 1973, through an arrangement with the IBGE, the BNH has prepared monthly construction cost estimates for 17 stan- dardized types of completed housing units in each of 62 Brazilian cities through its National System of Research on Construction Costs and Indexes (Sistema Nacional de Pesquisa de Custos e Indices da Construcao - SINAPI). In December 1967 BNH established special loan subprograms to expand production of construction materials as part of a broader construction materials program including RECON, the subprogram providing loans to users of these materials. BNH loans for fixed assets and working capital to construction materials enterprises under these subprograms expanded the supply of construction materials and thus may have kept prices under what they otherwise would have been. But the criteria for making such loans have been essentially financial and little if any effort has been made by BNH to achieve standardization of materials, to promote the production of relatively cheaper materials, to fully define and exploit potential returns to scale, or to orient the location of construction materials plants so as to decrease marketing costs. In March 1978, BNH sponsored a symposium on reduction of construction costs which may foreshadow new moves in this area. As for land costs, no indexes exist, but since 1976 the BNH has made loans to COHABs to purchase blocks of urban land with the purpose of anticipating speculative price rises and building land stocks for future popular housing projects. In the absence of any detailed cost data it is impossible to verify whether this program has been effective in keeping the price of land to COHABs lower than it otherwise would be. (c) Strengthening the Popular Housing Companies (COHABs) 62. Recognizing that the financial and technical status of many COHABs was less than solid, within the PLANHAP framework a number of important steps were taken in 1973 to strengthen the COHABS. These included: (i) global renegotiation of the COHABs' debts to the BNH in January 1973, with a reduction in interest rates and increase in amortization periods for projects with marketing problems or large numbers of beneficiaries badly behind in their payments; - 45 - Table 20: INTEREST RATES CHARGED BY BNH TO ITS FINANCIAL AGENTS UNDER PLANHAP BY SIZE OF LOAN Annual Interest Rate BNH/COHABs Loan Size PLANHAP I a/ PLANHAP II b/ (UPG) Up to 120 1.0 0.0 121 - 160 2.0 0.0 161 - 200 3.0 0.0 201 - 240 4.0 0.0 - 0.4 241 - 280 5.0 0.4 - 0.8 281 - 320 6.0 0.8 - 1.2 321 - 400 --- 1.2 - 2.0 401 - 500 2.0 - 3.0 Source: BNH Notes: af 1973 - 1974. Id Since 1975. - 46 - (ii) loans to the COHABS, at subsidized interest rates and 18 year amortization periods, for construction of shops within the housing projects (conjuntos habitacionais) destined for sale to small businesses, and which can then be sold at market prices, interest rates and amortization periods; (iii) inclusion, within BNH, housing loans to the COHABs, with 25 year amortization periods, of an amount equal to 6% of COHAB housing construction costs to be used to defray COHAB expenses for planning, design, and supervision of construction of the housing projects; (iv) inclusion, in the monthly payments for housing units financed, of administration and collection rates (Taxas de Cobranca e Administracao - TCA) and community support rates (Taxas de Apoio Comunitario - TAC) fixed at 5% and 3% respectively of the initial payment by final beneficiaries in UPCs; (v) loans to the COHABs for working capital at 2% interest and a maximum value equal to 10% of the repayment of the COHABs to the BNH over the next 12 months and amortizable in 36 months; and (vi) loans for improving the liquidity of the COHABs through opening special lines of credit. In addition, in 1975 the changes in loan terms to both COHABs and final borrowers increased the spread to the COHABs on most loans, especially those over 200 UPC. 63. BNH and COHAB officials claim that these measures have been quite effective in strengthening the COHABs and accelerating the production of popular housing. The acceleration is clearly visible over the years 1975-77 as was shown above. The 6% included in BNH loans to COHABs for planning and administration of projects has been particularly helpful, since these substan- tial funds are received prior to construction, permitting the COHABs to prepare projects more adequately before implementation. The expenses incurred are then passed on to the final beneficiaries in their mortgages. The increase in spreads has also strengthened the COHABs, but, as will be discussed below, the interest rate structure in force since 1975 has provided greater incentives to the COHABs to produce more expensive housing than PROFILURB serviced lots. There is also some question as to whether the 6% planning and administration component in BNH loans is sufficient in the case of PROFILURB projects, and this issue is currently under study. - 47 - (d) New Housing Subsidies 64. A major breakthrough occurred in 1975 with the institution of "fiscal benefits" through the tax system which result in substantial subsidies to SFH mortgage holders, subsidies which are larger in proportion to loan payments the lower the borrowing family's income. These new subsidies, which were made retroactive for 1974, are entirely outside the BNH-FGTS financial system. Beginning in 1971, the federal government authorized deduction from taxable income (federal income tax) of both interest and monetary correction on SFH housing loans. This system was regressive since it benefitted only those with incomes high enough to be subject to the income tax, thereby excluding the majority of families supposed to benefit from PLANHAP. 65. To partially correct this distortion, Decree Law 1358 of November 1974 specified that all persons with SFH loans, whether or not they were subject to the federal income tax, would receive a rebate, beginning in 1975, equal to 10% of their actual payments on housing loans the previous year, with a minimum of Cr$240 and a maximum of Cr$3,000 (respectively US$29.57 and US$369.59 in July 1975). Deduction of SFH interest payments (but not monetary correction) was still allowed for calculating taxable personal income. Unlike the rebate, the interest deduction effectively increases the rate of subsidy the higher the taxpayer's income. Decree Law 1431 of December 1975 raised the fiscal benefit to 12% of payments in the previous year, with a minimum of Cr$480 and a maximum of Cr$3,960 (respectively US$44.26 and US$365.15 in July 1976). The limits were again raised in 1976 to a minimum of Cr$648 and a maximum of Cr$4,554 (respecitvely US$44.29 and US$315.70 in July 1977). 66. Table 21 shows how the fiscal benefit system (exclusive of the interest deduction) worked for the year 1976. This table is particularly interesting since it breaks down the total of 892,893 borrowers benefitting from the subsidies in that year by size of loan and minimum family income required to obtained a loan of that size. Minimum family income required is not necessarily equal to family income in 1976 since loans can be for less than the maximum borrowing capacity of a family at the time they are made, and most of the loans were made before 1976 so that even if its maximum loan capacity was effectively used at the time the loan was made, the family might qualify for a larger or a smaller loan in 1976 depending on how its income changed. Minimum family income required therefore probably underestimates actual family income in 1976. Table 21 nevertheless gives a rough estimate of the distribution of SFH loans outstanding in 1976 by income class. The original data on which Table 21 was based had eight loan size and income classes. The lowest two were for minimum incomes up to and including 6.3 UPC or 1.4 minimum wages 1/ and from above 6.3 up to and including 9.2 UPC or above 1.4 up to and including 2.0 minimum wages and are shown separately as 1/ UPCs have been converted to minimum wages on a fixed basis of 4.6087 UPCs per monthly minimum wage, the average ratio in 1976 in Rio de Janeiro and Sao Paulo. Table 21: DISTRIBUTION OF FISCAL BENEFITS UNDER D.L. 1358/74 TO HOUJSING FINANCE SYSTEM BENEFICIARIES, 1976 Loan Required Minimum Borrowers Total Total Fiscal Fiscal Benefit as % of Class Amount Family IncomeP Benefited Loan Payments Benefits Granted Payments (UPC) (UPC) (Minimum (Number and (r 106 and Wages)k/ x of total) % of total) I -qI 200 .- 6.32 -4 1.37 133632 155.6 86.6 (55.7) (15.0) (1.4) (6.4) 2 200 -4 300 6.32 -j 9.21 1.37 -i 2.00 126994 315.1 82.3 (26.1) (14.2) (2.9) (6.1) 1+2 (Low) -4 300 -1 9.21 -4 2.00 260626 470.7 158.9 (35.9) (29.2) (4.3) (12.5) 3+4+5 (Middle) 300 -j 1100 9.21 - 46.47 2.00 - 10.08 399976 3674.7 464.4 (12.6) (44.8) (33.5) (34.4) 6+7-18 (High) 1100 -3500 46.47 - 80.07 10.08 -4 17.37 232291 6833. 718.6 (10.5) (26.0) (62.2) (53.2) Total- - - - 892893 10978.5 1351.8 (12.3) Sources: Calculated from BNH,"Concessao do Beneficio Fisial do D.L. 1358/74 aos Mutuarios do Sistemo Financeiro da Habitac-ao," Sept. 77. Notes: a/ The interpretation of these two columns is that as the loan amount increases over the range indicated in the first column, the required minimum family income increases over the ranges indicated. b/ On basis of 4.6087 UPC's per minimum wage, the average value of theminimum salary divided by the average value of the UPC in 1976. c/ Totals may not equal sum' because of rounding. - 49 - as well as in aggregate form, labeled as the "low income" category, which roughly corresponds to what was called "popular" housing prior to 1975. The next six classes are grouped into two categories called "middle" and "high" income (above 2.0 up to 10.1 and above 10.1 up to 17.4 minimum wages, respec- tively). 67. The following points can be made based on a perusal of Table 21: (i) 29% of the borrowers fell in the "low," 45% in the "middle" and 26% in the "high" income classes. (ii) The fiscal benefits granted in these same three classes as a percentage of loan payments were 35.9%, 12.6%, and 10.5% respectively. For the poorest subgroup (class 1) the benefit was 55.7% of loan payments. The new subsidy is therefore notably progressive in its distributional effect and repre- sents a major improvement over the system prevailing during the period 1971-73. (iii) Of the total benefits paid, the low income class received only 12.5%, compared with 34.4% for the middle income class and 53.2% for the high income class. Thus over half the total value of subsidies paid went to borrowers whose minimum required family income was over 10 minimum wages. 1/ (e) The New Approach to Low Income Housing: Loans for Sites, Services, and Construction Materials 68. Another major conclusion of the 1974 housing policy review was that it was unrealistic to expect low income families (particularly those with incomes below 2 minimum wages) to be able to afford commercially constructed housing units of even the minimum standard. It was therefore necessary to create another alternative to meet the housing needs of this very large group which constituted 48.3% of all urban families in 1972 (see Table 16). The PROFILURB sites and services program was the first major step toward imple- menting a policy of encouraging self-help housing construction. Through this new program the COHABs (and other agents such as state and municipal govern- ments) are given an instrument for financing urban lots complete with water supply, a sanitary sewer system or septic tank, and electricity at prices within reach of the urban poor. The average PROFILURB loan for the period 1977-79 was budgeted by BNH at 80 UPC (US$1,110 at the average 1976 UPC value and exchange rate), equivalent to about 18 times the average minimum monthly wage in Rio de Janeiro or Sao Paulo in 1976. 69. The PROFILURB program represents a major move toward realism in meeting the housing and sanitation needs of the urban poor. It implicitly recognizes that families with incomes of two minimum wages or less are already for the most part forced to construct their own housing, often on land illegally occupied and without any connection to water or safe sanitary waste disposal 1/ Furthermore, the percentage of total benefits paid to the high income class rose. It was 38.6% for 1974 and 45.3% for 1975. - 50 - system. Another common phenomenon, which affects somewhat less poor families, is the purchase of lots from private dealers, often in unauthorized subdivi- sions in which the lots are again often lacking basic urban services. Bad sanitary conditions rather than the quality of self-constructed dwellings as such are a major cause of the alarmingly high infant mortality rates and other indicators of poor health in Brazil's cities as is pointed out in Annex III of this report. Furthermore, the lack of clear title to land in illegally occupied squatter settlements and unauthorized subdivisions is a disincentive to investment in quality, permanent housing by the families concerned. The PROFILURB program gives priority to the provision of adequate water, sanitary waste disposal, electric power connections, and drainage on lots to which clear title can be obtained, leaving the owners (or leasers) of the lots to build their own permanent structures or have them built when they can afford to do so. 70. In principle the PROFILURB program offers an alternative to the favela removal philosophy which has prevailed in Brazil's official housing institutions. This philosophy sees the favela as a social cancer to be surgically removed and replaced by neat commercially constructed housing developments often located far from the original favela sites which are then devoted to uses such as the construction of luxury apartments and hotels. The alternative philosophy asserts that the favela and the unauthorized subdivision with self-constructed housing represent legitimate adjustments of urban poor to the realities of the labor market and the budgetary constraints they face. According to this line of thought, destruction of such housing together with forced removal to relatively high-priced (even though highly subsidized) official housing developments, often distant from sources of employment, not only reduces the welfare of those relocated, but also squanders scarce resources which would have a far higher social rate of return if devoted to: (a) upgrading existing settlements through the provision of basic sanitation and other urban infrastructure; (b) obtaining clear title to such sites for their occupants, thus encouraging the new owners to gradually invest their available labor and financial resources in more permanent forms of shelter; (c) developing new serviced sites on which self-help housing can be constructed, thus allowing greater planning of the urbanization process; (d) providing financial and technical assistance for self-help housing construction and other community improvements accomplished largely through the residents' own labor; (e) in general stimulating mutual aid and community involvement in the urban development process, making use of the naturally arising forms of social organization which tend to develop in such urban settlements. - 51 - This alternative has been officially espoused in other Latin American countries, most notably Peru, apparently with considerable success. Recent studies by social scientists have pointed out the social and economic benefits to be derived from such an approach in Brazil, basing their findings on empirical research in Rio de Janeiro. 1/ 71. PROFILURB projects may be established on previously unbuilt land, in existing subdivisions, or in previously existing squatter settlements (favelas) which are to be upgraded. In all cases the areas and lots to be financed under PROFILURB must respect existing urban plans (planos diretores), land use laws, subdivision laws, building codes, and other legal requirements. PROFILURB will finance all costs involved in acquisition and legalization of the land involved; planning, administration, and supervision of works up to 6% of their total direct costs; land leveling and drainage; individual sanitary sewers (on the lot); and connections to water sewer, and electric power systems. In the case of favela upgrading programs, the expenses involved in compensating families removed from the project area or forced to relocate within it for any improvements they have made are also covered. BNH may finance 100% of all the above costs as well as infrastructure and community equipment not covered in the loans for lots themselves. 72. PROFILURB lots may be sold outright or leased with an option to buy within a maximum period of 5 years. The latter method is obligatory in the case of favela upgrading and may be made so by the financial agent in other cases. The option to buy may be exercised when the beneficiary has regularly met his payments for the previous 6 months (a) after the contract has been in effect 2 years or (b) at any time, within the contract period, if a durable house has been built on the lot or another loan received to build a house. 73. Since the acquisition of an urbanized lot represents only the begin- ning of the process by which a family's housing needs will be met, PROFILURB establishes limits to the value of each lot and the percentage of family income which can be committed to loan payments so as not to exhaust the bene- ficiary's debt servicing capacity, thereby permitting him to obtain a second loan for the purchase of construction materials or for the construction of a house at a later stage. Such loans are now available under the FICAM pro- gram described below. The limits for a PROFILURB loan are 120 UPC for the total loan and total monthly payments not more than 9% of family income. The maximum interest rate to the final borrower for PROFILURB lots is 1% and the maximum loan repayment period is 15 years. The total first monthly payment on a 15 year PROFILURB loan for 80 UPC including insurance, Administration and 1/ See Janice Perlman, The Myth of Marginality: Urban Poverty and Politics in Rio de Janeiro (Berkeley, Los Angeles, and London: University of California Press, 1976) and David Michael Vetter, "Low Income Housing Policy for Development: An Evaluation of the Brazilian Experience," Ph.D. dissertation, Department of Urban and Regional Planning, University of California at Los Angeles, 1977. - 52 - Collection Fee, and Community Support Fee comes to .72866 UPC (Cr$109, equiv- alent to 16% of Sao Paulo and Rio de Janeiro monthly minimum wages or US$10.11 at the average values and exchange rate prevailing during the year 1976). Under the Constant Amortization System (Sistema de Amortizacoes Constantes - SAC) used by the BNH, the real value of the monthly payment on such a loan would fall over time even if the UPC maintained its real value because, with amortization constant, the interest on the outstanding balance falls. In addition, such a loan, if the first payment were in January 1976, would have been subject to the minimum fiscal benefit of Cr$648 prevailing in 1976 equal to about 50% of loan payments, albeit this benefit would only be received in 1977 and only if the borrower was up to date in his payments. These numbers suggest that the PROFILURB program does in fact bring serviced lots within the financial grasp of a much larger group of the urban poor than would be able to afford completed housing. However families having incomes of less than one minimum wage (about one in every eight urban families in 1976 according to the 1976 PNAD)l/ are still likely to find most PROFILURB lots priced beyond their means unless PROFILURB loan terms are made still more favorable. 74. A number of safeguards have been established in an effort to assure that the PROFILURB program will in fact achieve its social objectives. These regulations are intended to (a) direct PROFILURB loans preferentially to lower income families, (b) increase the amount of assistance which the COHABs and other promoting agents give to the purchasers of lots, and (c) avoid transfer of PROFILURB benefits to real estate speculators. 75. With the first objective in mind, a minimum of 50% of the total value of PROFILURB loans must be applied in loans with a maximum amount of 60 UPC and be to families with incomes less than or equal to two minimum wages or 8 UPC whichever is lower. The regulations to attain the second objective include the following: (a) all projects must include community development programs to promote the social development of the beneficiaries; (b) all projects must provide essential urban services (potable water, sanitary waste disposal system, electric power, drainage, easy access to public transportation, schools, etc.); (c) the promoting agent must provide technical assistance in house construction (even in the "rustic" stage) to the beneficiary families; and (d) there must be a possibility that those acquiring lots can receive another loan for construction materials. 76. Given the high degree of mobility of the low income population and high rate of job turnover, there is a danger that purchasers of PROFILURB lots might abandon them or sell them at low prices to speculators. It is 1/ IBGE, PNAD 1976, Vol 1, Timo 8, p.74. The data in this source are given in Rio de Janeiro minimum wages hence the proportion of urban families with less than one regional minimum wage was certainly lower. - 53 - (a) the borrower must reside on his lot for 12 months after the signing of the contract in the case of unbuilt areas and 4 months in the case of favela upgrading projects; (b) the explicit consent of the promoting agent (normally a COHAB) is required when a beneficiary cedes his rights to another person; (c) the lot must be returned to the promoting agent in case of a friendly recision of the contract, with compensation for improve- ments made and return of that part of payments representing amortization of principal (with monetary correction) after deduc- tions for any damage (other than normal wear and tear) to the basic installations on the lot (sanitary facilities, electric installation); (d) the market value of the lot will be its price on resale in the case of early liquidation of the loan contract, therefore eliminating subsidies which may have been implicit in the initial sale by the promoting agent. These anti-speculation measures are good in theory, but as yet untried. 77. A program to finance the cost of building materials to low-income borrowers (the RECON program did not reach the urban poor) was necessary to complement PROFILURB as well as to meet the needs of many potential borrowers who have already acquired suitable lots outside the PROFILURB program. The FICAM program, established in October 1977 within the PLANHAP framework, pro- vides this missing link in the institutional structure now potentially capable of meeting the housing and sanitation needs of most of the urban poor. The final beneficiaries of FICAM may obtain loans to construct a complete house, finish a house construction of which has already been initiated but not yet completed, or to improve or expand a dwelling. They may contract the full job (materials, labor, and administration) with a COHAB, with a construction firm, or with an autonomous and legally qualified professional. They may also assume direct responsibility for the execution of the job and purchase only construction materials and/or contract only specific professional services. It is also possible to include in the loan up to 80% of the value of the lot where the house will be constructed (even if the lot has been bought on an installment basis and not fully paid off). FICAM borrowers must prove that they meet the legal requirements for using land for housing purposes and have an income meeting housing finance requirements under PLANHAP. 78. FICAM loans can be made to families who have already obtained a first loan from a COHAB for acquisition of an urbanized lot or complete dwell- ing. In this case, the sum of the two loans cannot exceed 500 UPC and the value of payments of the two loans together must be compatible with the income and size of the family under the PLANHAP regulations. The land on which the works financed with FICAM resources are to be executed must, in addition to - 54 - meeting legal and technical requirements for residential use; have dimensions compatible with local customs; have or permit the installation of potable water, basic sanitation, and electric power; and be in the uncontested possess- ion of the borrower as proved through a legal title. These rather stringent provisions rule out loans to many existing favela dwellers. At present there is no program specifically oriented toward obtaining titles, though, as men- tioned earlier, PROFILURB will finance the costs of acquisition and legaliza- tion of land in squatter settlements. 79. In addition to the normal responsibilities of recipients of BNH loans, Promoting Agents of FICAM must (a) give technical assistance to bene- ficiary families in planning and executing the work financed, (b) counsel these families regarding compliance with legal formalities relating to the selection and acquisition of construction materials when the execution of the work is being done directly by the borrower, and (c) supervise the util- ization of the materials financed to assure that they are used for the purposes originally proposed. 80. While there is not yet any experience with the FICAM program to review, the PROFILURB program has a longer history and some observations regarding achievements to date can be made. Some issues related to the future development of these two programs and housing policy in general are set forth in the next section. Through October 1977 BNH had approved PROFILURB projects for the production of 18,340 lots, of which 2,128 (11.6%) were in the Northeast, 12,889 (70.3%) in the Southeast, and 3,323 (18.1%) in the Frontier region. Only one PROFILURB project with a total of 393 residential lots had been completed. It is located in Vitoria, the capital of Espirito Santo State. PROFILURB has gotten off to a very slow start and has lagged far behind the goals set for it in BNH budgets. 1/ 81. A number of factors in addition to the fact that the program is a new one seem to be responsible for this disappointing performance. First of all, in large metropolitan areas, cheap land is available only at peripheral locations involving long commute times and expensive transportation. The single family dwelling mode, which is the only one viable for PROFILURB, requires more land per housing unit than apartment buildings, townhouses, or semi-detached houses. Yet the need to keep total costs down means that the land costs must be relatively low. This means that it is often hard to find sites meeting PROFILURB standards. There appears to be little or no coordina- tion between implementation of water and sanitary sewerage systems under PLANASA (see Section IID of this Annex) and that of PROFILURB projects. In fact, BNH officials indicate that many COHAB projects require sanitation investments which end up being financed by BNH's Urban Development Portfolio rather than through the SFS. Favela upgrading, which is envisioned in both 1/ The 1976-78 budget planned for 50,000 PROFILURB lots in 1976 and 62,500 in 1977. The 1977-79 budget planned for 35,000 lots in 1977. In 1976 5,733 lots were financed and in 1977, through October, 9,014 lots had been financed. - 55 - the PROFILURB and FICAM programs, might appear to be a way around the problem of finding relatively convenient sites for PROFILURB projects. Yet favelas are often located on land which has a high potential value to real estate speculators if the favelados (favela residents) can be removed. This results in pressures from powerful interest groups in favor of removal rather than upgrading. Given that favelas are by definition illegal occupations of public or private land, and in the case of private land, the legal situation is often rather confused (the title may be in litigation), there may be substantial costs involved in regularizing land ownership which would have to be capital- ized in the price of a PROFILURB lot. Legal problems can also result in long delays in project implementation. 82. Interesting the COHABs in the PROFILURB program has been a problem. First, the "image" presented by a PROFILURB project, particularly in its early stages when unsightly temporary dwellings are highly visible, may seem less attractive than that of neat rows of commercially constructed housing. More important may be the problem of financial incentives. In addition to the potentially greater problems in administering projects with lower-income beneficiaries (given the instability often characterizing their income and employment as well as cultural traits), prior to November 1978 the COHAB simply obtained a greater income per unit on a fully constructed house than an urbanized lot, both during production and after sale of the units. Table 22 compares what a COHAB could earn on the maximum value PROFILURB lot (120 UPC) with a house costing 300 UPC. During the construction phase, the COHAB's income from a house would be 3.75 times that of an urbanized lot, while after completion the annual income ratio would be 1.84, again in favor of the house. These ratios would increase for a cheaper lot. Finally, given that 25 year loans were available on completed houses and only 15 year loans for urbanized lots, if some kind of house (perhaps an "embryo" with a partially finished core) could be constructed for 200 UPC, the first monthly payment on the house have been only 0.02 UPC higher for the house. Twenty-five year loans for PROFILURB lots were not originally contemplated since the fixed minimum fiscal benefit could exceed the loan payments. 83. On October 31, 1978 a new BNH regulation (RC 11/78) went into effect which may eliminate many of these problems. Among its provisions are allowing the financing of a sanitary unit or embryo house along with the serviced lot, the extension of the amortization period for PROFILURB loans to 25 years for that portion of the loan exceeding 50 UPC, and eliminating the 6% ceiling on the cost of planning, administration, and supervision of construction which can be passed along to the final borrower and financed. A subsequent regulation increased the upper limit for PROFILURB loans to final borrowers from 120 to 200 UPC, with as much 250 UPC loans being permitted for embryo houses in certain high areas such as Greater Sao Paulo. 4. Current Housing-Specific Policy Issues 84. A number of policy issues related to housing are best viewed in the context of overall urban development since they also affect the provision of other housing related services. Among these general issues are (a) how to assure the financial viability of the BNH, the SFH, and the FGTS--that is the efficiency of financial intermediation and the average real rate of return on - 56 - Table 22: CONPARISON OF COHAB INCOME FROM A 120 UPC URBANIZED LOT AND A 300 UPC HOUSE WITH LOT, PRIOR TO OCTOBER 31, 1978 Lot House (UPC) (UPC) Production Phase Planning administration, and supervision of works (6% of direct costs, assuming these are 50% of loan value for the lot and 75% for the house) 3.60 13.50 After Sale (annual receipts) Interest rate differential (1% per annum for PROFILURB and 1.6% for house) a/ 1.20 4.80 Collection and Administration Fee (10% of value of first payment for lot, 5% for house) 1.07 .97 Community Support Fee (10% of value of first payment for lot, 2% for house) 1.07 .39 TOTAL 3.34 6.16 Source: BNH Note: a! Under the constant amortization scheme used in all PLPNHAP loans, the interest charges will fall over time as the principal is amortized. - 57 - assets and liabilities within these systems; (b) the weak financial status of many state and municipal governments, which under the existing system are expected both to incur BNH debt and to provide direct budgetary support for housing infrastructure (roads, public transportation systems, and public utilities including water supply and sewerage); and (c) the need for an over- all urban development strategy at the federal, state, metropolitan and muni- cipal levels within which industrial location, urban land use, public trans- portation, and public utilities policies are coordinated and harmonized. To put matters in perspective, these issues or very similar ones are common to many countries and (c) in particular has not been fully resolved even in advanced industrial countries. These matters are beyond the scope of the present report and therefore will not be discussed further here. In the following paragraphs attention will be focused on less comprehensive issues more specific to housing policy: (a) the efficiency and redistributive character of the fiscal benefit system; (b) the incentives to COHABs to plan and implement PROFILURB-FICAM sites and services projects; (c) the development of "mixed" housing projects including a variety of housing options for differ- ent income groups in the same location; (e) land use and taxation legislation to reduce land price; (f) reduction of construction costs; and (g) the question of how to improve housing conditions in rural areas. (a) The Fiscal Benefit System 85. The introduction in 1975 of housing subsidies financed from general federal tax revenues was a major step forward in financing adequate housing and related services for the urban poor. Nevertheless, the present system of minimum and maximum payments, which assures that all SFH borrowers receive a rebate which is a higher proportion of loan payments the lower the borrower's income, has a major shortcoming. As shown in Table 21, in 1976 it resulted in over half the total benefits being paid to borrowers whose incomes were over ten minimum wages (measured in terms of the highest regional minimum wage) when they obtained their loans. Paying such subsidies to the relatively well-off, who also can deduct their interest payments for income tax purposes, certainly made no contribution to solving the housing problems of the urban poor. In principle these resources could be used in a variety of ways to reduce the cost of housing and related services to those families in greatest need rather than to subsidize the well-to-do. 86. One way to solve this problem would be to relate the subsidy directly to loan payments and family income, with the maximum subsidy rate (allowing token loan payments) going to families with the lowest incomes (say less than one minimum wage). The subsidy would taper to zero above an upper limit, say ten minimum wages. Subsidies could be deducted directly from payments for the same month if loan payments were up to date, reducing the out-of-pocket expenses of low-income borrowers from the date of their first loan payments (rather than after up to a year as is now the case) while maintaining the incentive to keep payments up-to-date. The main problem would be obtaining relatively accurate estimates of total family income on a regular basis. This administrative problem would be greatly simplified if, as is already done, the determination of income is done when a loan is made. Each time an additional loan is made (for example a FICAM loan for housing - 58 - materials or construction) the debts could be consolidated and a new determin- ation of family income made. Even so, administrative problems and possible disincentives to obtain new loans if income rises may rule out a direct link between the rate of fiscal benefit and income. If these considerations pre- vail, then there are still two modifications which would improve the equity and efficiency of the current system. First, the inherently regressive interest deduction for tax purposes could be abolished. If the interest deduction is maintained, a provision could be added that no fiscal benefit be paid to those taking the interest deduction, thus eliminating double benefits for those taking the interest deduction. Secondly, as suggested above, the fiscal benefit, which is considerably more progressive than the interest deduction, could be deducted directly from payments if the latter are up to date. (b) Increasing Incentives to COHABs to Build More Sites and Services Projects 87. As pointed out above, the interest rate spread allowed COHABs increases from 1% to 3% over the range 200-500 UPC. This results in a per- verse (from the viewpoint of solving the low -ncome housing problem) incen- tive to build more expensive units which are beyond the means of the poor. Recognizing that there is excess demand for completed housing units as well as urbanized lots does not rule out a change in the method of financing the COHABs which would eliminate this distortion. The spread could be made uniform. If the COHAB's must be subsidized, permitting higher spreads for larger loans appears to run directly counter to the objective of keeping housing costs to a minimum. A direct subsidy proportional to the number of lower value loans made would work in the proper direction. It is likely that community development and plot settlement assistance costs will be much higher for PROFILURB-FICAM projects than for completed housing units, and if the COHAB's are to be induced to move massively into sites and services projects some means of meeting these costs which does not price the intended benefi- ciaries out of the market must be found. The whole question of financial incentives within the PLANHAP framework was under study by BNH in February 1979. (c) Mixed Housing Projects 88. Housing projects containing only PROFILURB lots or fully constructed popular housing units tend to segregate housing by income class, even within the "popular" market, while restricting the range of solutions to the housing problems of the urban poor. It may be argued that this is socially undersire- able. A broader range of housing options, including some outside the popular range (such as that provided by the housing cooperatives for the five to ten minimum wage income class) would decrease this segregation by income strata, increase the range of solutions to the housing problem facing the urban poor, open up nearby sources of employment as domestics and providers of other services to higher income classes, and create a possibility of cross- subsidization of infrastructure costs within a large-scale housing project. Some COHABs are already experimenting with this concept for finished, semi- finished, and "embryo" housing units as well as urbanized lots, and others have established joint projects for finished popular housing and housing - 59 - cooperatives in the middle-income range. Combining these approaches anid al . allowing at least partial self-help construction of semi-detached or townhouse styles, which allow greater economy in the use of expensive urban land, might - permit closer-in construction sites, saving on transportation costs in the larger cities and metropolitan regions. Without mixed projects, high urban land costs are likely to force PROFILURB projects into peripheral loca- tions in major metropolitan areas such as Sao Paulo. (d) Improving Land Use Legislation 89. The cost of urban land is heavily influenced by the existing land tax structure and land use laws. Without entering into details here, it is sufficient to note that speculative holding of undeveloped urban land is encouraged by the combination of tax and land use legislation now in effect in most Brazilian cities. For example, in Sao Paulo, according to the muni- cipal planning authorities, 42% of the land suitable for building was being held in an undeveloped state in 1976. The existence of large areas of unde- veloped land raises the cost of providing essential urban services to other areas, while too much vertical development may put excessive strain on urban infrastructure in other parts of a city. Progressive taxation of undeveloped urban land, a capital gains tax on gains from holding urban real estate, and implementation of the system of air rights (solo criado) whereby the right to build over a certain height must be purchased from public authorities are policy measures which might help alleviate this problem at the same time creating public revenues which could be used to subsidize popular housing and related urban services. All these measures would work best if there is a well- defined urban planning system. In any case, the distortions in urban struc- ture occasioned by the present tax and land use laws are being intensely and widely discussed at both the federal and local levels. (e) Reducing Construction Costs 90. Reducing constrUction costs offers another promising line of attack in bringing housing costs within reach of the poor. As pointed out above, BNH has made loans for the production of building materials since 1967. But little effort has been made to: (a) bring down production costs for these materials through a program of research and development; (b) standardize low-cost materials and designs for use in popular housing to the extent feasible given the diversity of climates and available raw materials over Brazil's immense territory; and (c) use the building materials production loan program to achieve maximum scale economies while assuring lively competition among producers to the extent feasible and controlling prices when it is not. A serious effort along these general lines could result in important cost reduc- tions for housing, whether commercially finished or self-constructed. In addition, bulk purchases of building materials by the COHABs may help keep down costs for all PLANHAP programs. In the case of commercially finished popular housing, experimentation with new construction technologies including fully or partially prefabricated housing is under way, but costs tend to be as high or higher than for conventional housing. - 60 - (f) Improving Rural Housing 91. To date, the BNH has made no significant effort to improve housing in rural areas. Its progressive definition as an urban development bank does not suggest that it is moving in this direction. This means that the insti- tutional infrastructure for operating in rural areas is not in place, that is unless the BNH were willing to use Bank of Brazil offices as financial and promotional agents. The Bank of Brazil is the Brazilian public institution with the broadest institutional presence in rural areas, since it is the principal source of agricultural credit. The quality of rural housing units in terms of shelter often leaves much to be desired, even without considering their water supply and sanitary waste disposal facilities. This is particu- larly important in the extensive areas where Chagas' disease is endemic and indeed an agency of the Health Ministry has initiated a small program of rural housing improvements as part of the attack on this disease (see Annex III of this report). A program like FICAM could be adapted to rural areas, with an emphasis on upgrading houses so that they no longer provide hiding places for the vector of this debilitating disease. But in the near- to medium-term future the expansion of the PROFILURB and FICAM programs in small- and medium- size cities not presently served by the COHABs appears to offer the most promising line of movement toward rural areas. It will be important to assure careful coordination of these programs with the establishment and expansion of basic sanitation systems under PLANASA. D. The Sanitation Finance System (SFS): Achievements, Problems, and Policy Issues 1. The National Sanitation Plan (PLANASA): Background and Objectives 92. The National Sanitation Plan (Plano Nacional de Saneamento - PLANASA), which sets national objectives for water supply and sanitary sewerage, was formulated and formal execution begun by BNH and other parts of the SFS in 1971. However, BNH finance of water supply systems began with the creation of the SFS in 1968 and the first BNH investments in sanitary sewer systems took place in 1970. These earlier loans have been incorporated into the PLANASA framework. PLANASA grew out of an analysis of the deficiencies of the previously existing non-system of planning, financing, and executing water supply and sanitary sewerage systems in Brazil. Among the deficiencies which PLANASA sought to correct were: a. lack of financial resources; b. application of what resources were available on a grant basis; c. uncoordinated action and multiplicity of municipal, state, and federal agencies; - 61 - d. absence of overall planning; e. absence of adequate and realistic systems of rates; and f. excessive construction time. 1/ 93. The Sanitation Financing System (SFS) was established by the BNH, which plays a central role in it. But it also includes the State Water and Sewerage Funds (FAEs); state banks as financial intermediary agencies; and state sanitation companies (one per state), which are the promoting and executing agencies for the provision of water and sewerage services at the metropolitan and municipal levels, as described earlier. The emphasis here will be placed on the objectives set by PLANASA. The following objectives are declared to be "permanent": a. elimination of the deficit in the basic sanitation sector through an adequate program which permits achieving equilibrium between demand and supply of these services, in the shortest time, at minimum cost; b. maintenance, on a permanent basis, of the equilibrium attained between supply and demand of goods and services in the field of basic sanitation; c. service for all Brazilian cities, even the poorest urban centers; d. institution of a rate policy in accord with the possibilities of consumers and the demand for resources and services so as to attain a permanent equilibrium between receipts and expenditures; e. institution of policies to reduce operational costs through achievement of scale economies with direct impact on the rate scheme; and f. development of research, training, and technical assistance programs. 2/ Even in these "permanent" objectives, PLANASA is explicitly confined to urban areas. Implicitly it is postulated that under Brazilian conditions, solving basic sanitation problems in rural and semi-rural areas requires individual 1/ Alberto Klumb, PLANASA: National Sanitation Plan (Rio de Janeiro: BNH, 1976). 2/ BNH, Modificacoes No Sistema Financeiro do Saneamento (Rio de Janeiro: BNH, 1975) 39-40. - 62 - rather than collective solutions. This was explicitly stated in a diagnosis of basic sanitation problems prepared for the 10 Year Economic and Social Development Plan in 1967. 1/ 94. The original PLANASA set specific goals to supply drinking water to at least 80% and sanitary sewerage to at least 50% of the urban population by 1980. 2/ These objectives were revised in 1975, resulting in what may be interpreted as a more ambitious goal for water supply, at least in terms of increasing the geographical distribution of beneficiaries, and a lower, less precisely defined goal for sewerage systems, namely; a. drinking water service by 1980 to at least 80% of the urban population in at least 80% of Brazilian municipios and all the metropolitan regions; b. adequate sanitary sewerage service by 1980 to the metropolitan regions, state capitals, and larger size cities; and c. service with simpler sewer systems, to the extent possible, for cities and towns of smaller size. 3/ BNH has defined "larger size" cities as those with more than 50,000 population. This would bring the number of municipios to be served by adequate sanitary sewers in 1980 to 160. 4/ The new goals reflected the priority placed on water supply and the relative success in this area as opposed to sanitary sewerage, where investments and achievements have lagged as will be seen below. 2. Measures of PLANASA/SFS Performance 95. In this section. the achievements of PLANASA in bringing local water and sewerage systems into state level systems coordinated at the federal level and in financing new basic sanitation investments through the SFS are evaluated. 1/ Ministerio de Planejamento e Coordenacao Geral, Plano Decenal de Desenvolvimento Economico e Social, Tomo VI, Vol. 3, Saude e Saneamento (Versao preliminar) p. 45. 2/ Presidencia da Republica, Secretaria Geral do Conselho de Desenvolvimento Social, "Plano Nacional de Saneamento - Nova Sistematica" in CDS II (Janeiro/Junho de 1975) p. 75. 3/ Ibid, p. 75. 4/ Ivandro Mendonca Pires, "O Atual Estagio do PLANASA", paper presented at the 9th Brazilian Congress of Sanitary Engineering, Belo Horizonte, 3-8 July 1977, p. 8. - 63 - A strong word of caution is in order from the outset regarding interpretatiun of PLANASA statistics regarding (a) municipios "integrated" in PLANASA's water supply program, (b) the 1980 population with water or sanitary sewer service already in place or financed, and (c) the total number of water or sewer connections 1/ in the state sanitation companies. These statistics cannot be linked with any precision to those of the censuses or PNADs on the number of urban houses connected to general water and sewer systems (rede geral) as provided in the first part of this annex, nor can thi average cost of new connections be calculated by coupling the population or connections data with investment data. 96. When a municipio is integrated into the PLANASA water or sewerage program, if it has any water or sewer systems, the population served by them (usually estimated as a multiple of the number of connections) will increase the statistics of total population served under PLANASA without any investment being made. The quality of service of those actually connected to these systems may be highly deficient and therefore require substantial investments to bring it up to standards regarding volume, continuity and water potability or sewage treatment. The fact that a housing unit has a connection to either a water or a sewer system does not mean that these connections provide adequate service, and BNH officials state that such sub-standard connections have been common in pre-existing systems entering PLANASA. Furthermore, at the time of the 1970 census and the 1973 PNAD, there were many existing water and sewer systems which were not integrated into PLANASA. Integration into PLANASA has meant that a local system loses autonomy over its finances, rate structure, and investment program. This is one of the reasons that some of the more financially viable systems have been slow to join PLANASA. 2/ While the number of systems outside PLANASA had decreased substantially by 1977, there were still a significant number, especially in the Southeast (the most advanced region of the country, but also one which appears to lag most in PLANASA implementation). Some federal programs also finance basic sanitation outside 1/ The term "connection" is used here despite the fact that in Brazil the word refers to billing units rather than final users. For example, a condominium apartment building with multiple housing units is generally considered as one "connection" with multiple "water economies" and/or "sewer economies." In this annex the Portuguese word economia is trans- lated as connection. 2/ In the case of Sao Paulo, an arrangement has been worked out whereby a municipal water company may (1) join PLANASA in the normal way, turning its concession over to the state sanitation company (SABESP), (2) agree to turn over the concession to SABESP by a given future year, but continue to operate the municipal system until then, and (3) agree for SABESP to supervise planning and execution of new investments, stipulate the accounts to be used by the municipality, and prepare tariff increases at SABESP's request--but the municipality may continue to operate its municipal system. Extension of this system or some variant to other states might speed integration of the systems still remaining outside PLANASA. - 64 - the PLANASA framework. Among them are the Foundation for Public Health Services (Fundacao Servicos de Saude Publica - FSESP), attached to the Ministry of Health and working mainly in rural areas of the Northeast and Amazonia; DNOS and SUDENE in the Northeast; and SUDAM in Amazonia. The magnitude of resources devoted to basic sanitation under these organizations is small compared with PLANASA, but still significant. 97. Of the federal programs outside the PLANASA framework, that of FSESP is probably the most important. FSESP completed 306 water supply systems, expanded 112, and began work on an additional 248 new systems and 123 expansions during the period 1974-November 1978. FSESP estimates that approximately 4 million Brazilians in 21 states and territories will benefit from these systems. Through 1973 an additional 467 water supply systems were constructed by FSESP. As of 1978, 493 localities in 280 municipios in 17 states were receiving technical and administrative assistance from FSESP. 1/ Detailed investment and cost data for the FSESP program were not obtained, and FSESP statistics are not easily comparable with PLANASA statistics. Some systems constructed by FSESP have been integrated into the state sanitation companies and thus included in PLANASA coverage statistics. 98. The number of water or sewer connections includes industrial and commercial ones as well as residences, even though residences are a high proportion of the total. Furthermore, in addition to population already served, BNH/SFS statistics on 1980 population served in relation to PLANASA water supply and sanitary sewerage goals include an estimate of the population which should be served if all loan contracts signed through a given date are executed on schedule, if there are no cost overruns, and if the potential link between installed capacity of a system and actual connections to dwelling units is realized. There are obviously a number of ways in which achievements could fall short of these projections of 1980 population with service already in place or financed (populacao ja equacionada) as shown in Table 23. Finally, though the official PLANASA water supply goal is to serve at least 80% of the urban population in 80% of the municipios in each state, 2/ the BNH estimate of the 1980 target population for Brazil (67.3 million) exceeds 80% of IBGE's total urban population (62.5 million). With these serious caveats, the avail- able measures of PLANASA/SFS achievements are reviewed below, first for water supply, and then for sanitary sewerage. 1/ "Sintese de Atividades da Fundacao SESP, 1974-1978" (mimeo, n.d.). 2/ The goal is itself ambiguous, since it is not clear whether the objective is to serve at least 80% of the entire urban population and reach 80% of the municipios, or to serve at least 80% of the urban population in each of the municipios to be served (80% of all municipios). The latter inter- pretation would allow meeting the goal without reaching 80% of the total urban population. Table 23: PLANASA: W4ATER SUPPLY OBJECTIVES AND I1PLEMENTATION AS OF JANUARY, 1978 Municipios Population (millions) in 19SO-/ To be Integrated Goal to be Service Already in Service not Regions Total Goal Integrated % by 1980 served Place or Financed % yet Financed (A) (B) (C) (C/B) (B-C) (D) (E) (E/D) (D-E) Northeast 1375 1100 772 70.2 328 15.0 11.46 76.3 3.55 Southeast 2128 1703 886 50.3 817 46.1 32.02 69.4 14.15 Frontier 451 361 184 51.0 177 6.1 4.55 74.6 1.55 Brazil 3954 3164 1842 58.2 1322 67.2 48.03 71.4 19.25 1 0%' Source: Calculated from BNH /SFS data. Note: a/ The population figures are BNH projections. - 66 - (a) Water Supply Table 23 compares PLANASA water supply objectives with implementation as of January 1978, both in terms of municipios integrated into PLANASA and 1980 population with service already in place or financed. The data are broken down by the three regions used elsewhere in this report. With just under three years remaining to reach the goal of 80% of all municipios, only 1,842 or 58.2% of the goal had been integrated. The highest level of goal achievement is to be found in the Northeast with 70.2% of the target municipios signed up and 76.3% of the estimated 1980 target population with service already in place or financed. The relatively favorable performance of the Northeast may be explained by the attractiveness of the PLANASA program in a region where pre-existing systems were few, costs of systems low compared with major cities in the Southeast, and interest rates on BNH loans relatively favorable. The Southeast region lagged with only 50.3% of the target municipios and 69.4% of the 1980 target population "served". The explanation for this lag seems to be twofold. First, the Southeast includes the states of Minas Gerais and Sao Paulo which between them account for 31% of all Brazilian municipios. In Minas Gerais, many of the municipios are relatively small and poor, and only 175 or 30.3% of the 570 target municipios in this state had joined PLANASA, a lower ratio than any other state except Mato Grosso, in the frontier region. For a period in 1976 and 1977 the Minas Gerais state sanitation company simply refused to accept applications of additional municipios wishing to join PLANASA, since the company did not have the financial resources necessary to absorb them. The other explanation is that a number of relatively important existing municipal water systems (including Porto Alegre in Rio Grande do Sul, Juiz de Fora in Minas Gerais, and several medium-sized cities in Sao Paulo state) have been reluctant to enter PLANASA because they have been able to finance their own systems without recourse to PLANASA and fear that integration will mean loss of autonomy with their cash flow being used to finance systems elsewhere in their states. The Frontier, with 51.0% of target municipios signed up, but 74.6% of the target 1980 population "servel" occupied an intermediate position. The Frontier's performance in terms of population would be much less impressive if the Federal District (which includes Brasilia) were left out. The overall figures for Brazil are 58.2% of the target in terms of murlicipios and 71.4% of the 1980 target population. As mentioned above, the population "served" statistics are subject to question. Furthermore, the 1322 municipios to be integrated by 1980 are for the most part relatively small and poor. A major government study points out that PLANASA by 1976 had already completed its most viable phase. From 1976 onward, what remained were "projects of smaller size, dispersed in a greater number of urban centers which present greater deficiencies in the effective supply of a population with smaller purchasing power." 1/ 1/ Wanderly J. Manso de Almeida, Abastecimento de Agua A Populacao Urbana Uma Avaliacao do Planasa, Colecao Relatorios de Pesquisa No. 37 (Rio de Janeiro: IPEA/INPES, 1977) p. 114. - 67 - Table 24: PLANASA: NUMBER OF WATER CONNECTIONS OF THE STATE SANITATION COMPANIES, 1972-1977 Absolute Percentage Increase Increase Region 1972 1973 1974 1975 1976 1977a/ 1972-77 1972-77 (Thousands) Northeast 761 882 1002 136 1284 1157 696 (91) Southeast 2203 2530 2779 3057 415h 4807 2603 (118) .rontier 171 20' 362 406 51h 6r7 136 ( 255) Brazil !/ 3135 3616 hlh49 4599 5951 6860 3734 ( 119) a/ 1977 Data are BNH estimates based on information available in January 1978. bf Brazil total may differ from region sums due to rounding. Source: Calculated from BNH data. - 68 - Table 24 shows the number of water connections incorporated in the 21 state-level sanitation companies for the years 1962 through 1977 and by major regions. In all Brazil in 1977 there were 6.9 million connections, considerably less than the 7.6 million urban dwellings estimated as connected to general water supply networks in 1973 according to the PNAD of that year even without discounting for the number of industrial and commercial connec- tions included in the PLANASA statistics. Assuming 5% of the PTkNASA connec- tions were industrial and commercial, and an average of six persons per residen- tial connection, water supply systems incorporated in PLANASA in 1977 served approximately 39.2 million Brazilians, or only 58% of the target 1980 popiL- lation. This is a more sober measure of the gap which PLANASA must fill by 1980 by incorporation of more existing systems and expansion of effective capacity. Of course SFS loans already contracted will permit doing part of the job, but the task remaining to be accomplished in only three years is substantial. Nevertheless, the number of water connections in the PLANASA system has expanded by 119% over a five year period, the most rapid expansion being in the Frontier region (255%), and the slowest in the Northeast (91%), which seems inconsistent with the high degree of coverage for the Northest indicated in Table 23. Furthermore, assuming 95% of the water connections are to housing units and six persons per connection, the Northeast by 1977 had reached only 55% of the goal for 1980. This suggests that despite having ambitious plans, the Northeast sanitation companies may lack sufficient revenues and grant funds to expand water supply systems as rapidly as they plan and/or that the low income levels of the urban populationi in the Northeast has resulted in fewer connections to expanded systems than expected. The comparable figures for the Southeast and Frontier regions are 59% and 57%. Table 25 presents data on total water supply investments realized under PLANASA during the period 1968-76 by source and region. Over two thirds of total investaents (68.7%) were in the Southeast, 23.0% in the Northeast, and 8% in the Frontier. These proportions are very close to the proportions of urban households in the three regions during this period. Of the totals for Brazil, direct BNH loans accounted for 37.6%; supplementary loans to the FAEs 10.4%; another 26.7% from amortizations, budgetary allo- cations, and other sources channeled through the FAEs; and finally 25.3% from grant funds (fundo perdido) and others. BNH supplementary loans to FAEs were highest as a perceatage of the total BNH contribution (43%) and total investment value (31%) in the Northeast, as might be expected since the purpose of these loans is to supplement insufficient local resources. Table 26 breaks down total investments in water supply systems under PLANASA over the same period by source and year. While there have been fluctuations in the level of effort over time, there has been a marked acceleration from 1974 onwards from all sources except fundo perdido and others. This is consistent with the present administration's allocation of high priority to PLANASA's water supply program. However, the total invest- ments for 1976 came to only 0.25% of GDP. Finally, Table 27 presents estimates of the investment required to achieve the 1980 water supply targets during the years 1977-80, broken down by source and region, as estimated by BNH. Measured in UPC, the investment required exceeds the total for the period 1968-76 by 4.4%, which gives an Table 25: PLANASA: TOTAL WATER SUPPLY INVESTIENTS REALIZED BY SOTTRCE ANM RuGIOM 1968-1976 a/ (Millions of 1976 Cr$ and Percentages) % of "Fundo Total Total Perdido" Investment Investment BNH c/ and Others Region Value Value b/ Direct Supplementary Total FAE- (A+C+D)- (A) (B) (A+B) (C) (D) Northeast (23.0) 3946.3 1585.9 1207.3 2793.2 1929.9 430.5 Southeast (68.7) 11799.1 4210.3 347.2 4557.5 3981.7 3607.1- Frontier (8.3) 11421.9 654.8 228.7 883.5 458.2 308.8 Brazil (100.0) 17167.3 6451.0 1783.2 8234.2 6369.9 4346.4 (%) - (100.0) (37.6) (10.4) (48.0) (37.1) (25.3) a/ At average value of UPC for 1976, Cr$149.63. b/ Numbers in this row or column may differ from sum of components due to rounding. c/ Includes Supplementary BNH loans (B). d/ Includes resources from World Bank loans to the State of Sao Paulo Source: Calculated from Ivandro Mendonca Pires, "O Atual Estagio do PLANASA" paper presented at the 9th Brazilian Congress of Sanitary Engineering, Belo Horizonte, 3-8 July, 1977, Annex 8. Table 26: PLANASA: WATER SUPPLY INVESTMENTS REALIZED BY SOURCE AND YEAR 1968-1976 (Millions of 1976 Cr$- and Percentages) Total "Fundo Investment BNic Peridido" d/ Year Value b/ Direct Supplementary Totalb, FAE- and Others- (A+C+D)- (A) (B) (A+B) (C) (D) 1968 22.6 11.0 - 11.0 7.1 4.5 1969 1108.4 166.5 30.9 197.4 157.6 784.3 1970 2020.6 339.4 104.0 443.4 350.2; 1331.0 1971 745.2 295.9 61.1 357.0 279.7 169.7 1972 1217.5 454.9 129.3 584.2 390.1 372.6 1973 2822.5 1183.6 119.3 1302.8 1092.5 546.4 1974 1965.8 768.0 336.7 1104.6 869.3 328.5 1975 3363.5 1395.6 238.3 1633.9 1304.2 663.6 1976 3901.2 1836.1 763.7 2599.8 1919.2 145.9 1968-197611a 17167.3 6451.0 1783.2 8234.2 6369.9 4346.4 (%) (100.0) (27.6) (10.3) (4b.0) (37.1) (25.3) a/ At average value of UPC for 1976, Cr$149.63. b/ Numbers in this row or column may differ from sum of components due to fotnding. c/ Includes Supplementary BNH loans (B). Source: Calculated from Ivandro Mendonca Pires, "O Atual Estagio do PLANASA", paper presented at the 9th Brazilian Congress of Sanitary Engineering, Belo Horizonte, 3-8 July, 1977, Annex 9. Table 27: PLANASA: ESTIMATED INVESTMENT REQUIRED TO ACHIEVE 1980 WATER SUPPLY GOALS, 1977-1980, BY REGION AND SOURCE (Millions of UPC) % of Total Total "Fundo Investment Investment BNH Perdido" Region Value Va lue)Direct Supplementary Total FAE and Others (A-fC+D)2J (A) D Northeast (17.6) 21.1 6.6 3.6 10.2 8.6 5.8 Southeast (71.1) 85.2 42.1 10.0 52.1 33.8 9.3 Frontier (11.3) 13.5 6.0 1.7 7.8 4.7 2.8 1 BrazilA' (100.0) 119.8 54.7 15.3 70.0 47.1 17.9 (%) - (100.0) (45.7) (12.8) (58.4) (39.3) (14.9) Notes: a/ Numbers in this row of colum may differ from sum of components due to rounding. b/ Includes supplementary BNH loans. Source: Calculated from Ivandro Mendonca Pires "O Atual Estagio do PLANASA", paper presented at the 9th Brazilian Congress of Sanitary Engineering, Belo Horizonte, 3-8 July 1977, Annex 15. - 72 - idea of the magnitude of the effort required. In terms of 1976 GDP projected over the period 1977-80, the total investment required over the same period comes to 0.25%, or the same effort relative to GDP as actually took place under PLANASA in 1976. 71.1% of the total would be invested in the Southeast, 17.6% in the Northeast, and 11.3% in the Frontier. TIaplicit in these figures is an average investment cost of roughly 6 UPC (US$83.24 at average 1976 UPC value and exchange rate) per person served. A study prepared in the state of Sao Paulo estimates water supply investment costs at 8 UPC per person served for cities larger than 5,000 and 6 UPC for cities under 5,000, 1/ but SFS officials indicate that costs per person served may be as low as 3 or 4 UPC for simple systems in small towns where construction costs are low, increasing to over 10 UPC per person for completely new capacity in large aetropolitan areas such as Sao Paulo and Rio de Janeiro. Of course, costs are less for partial expansion of distribution systems if there is excess capacity in existing treatment and/or collection systems. This suggests that the BNH estimate may be on the low side in the case of medium-sized and larger urban areas. Given that nlich of the investment will take place in relatively small cities where costs are probably below those in Sao Paulo state, the BNH estimate of 6 UPC (1976 prices) per person served appears reasonable for projection purposes. (b) Sanitary Sewerage Information on implementation of the sanitary sewerage program in relation to PLANASA goals is less complete than in the case of water supply. Table 28 shows the situation as of January 1978 in a manner analogous to Table 23 in the case of water sutpply, except that the target number of muni- cipios is not broken down by region. The number of municipios which had joined PLANASA's sewerage program as cf January 1978 was 129 or 81% of the 1980 goal. In terms of 1980 target population with service already in place or financed, the Southeast was fac ahead of other regions, with 66.3% of the goal "achieved." The Northeast and Frontier had only 34.2% and 33.3% respectively. SFS officials suspect, however, that translating 1980 expected capacity into effective connections may prove more difficult than in the case of water. They suggest that for residents of housing units with some rudimentary (even though unsafe) form of sanitary disposal, such as a simple drain buried in a "septic trench", the felt need for a connection to the general sewer network is often less than that for safe drinking water from the general system, despite the social desirability of the general network solution, especially in densely populated areas. Sanitary sewer systems have per capita investment costs well above water systems, and establishing a connection may require substantial additions to plumbing inside the residence being served. Hence "marketing" may prove to be more a problem. Experience through 1978 has shown that connections to the systems which have been con- structed under PLANASA are being made at a rate well below that foreseen in the original projects. As a result, the rate of return to the state sanitation 1/ SABESP, "Estudo de Viabilidade Globe" (Sao Paulo: August 1977). Table 28: PLANASA - SANITARY SEWERAGE OBJECTIVES AND IMPLEtENTATION AS OF JANUARY 1978 a/ Municipios Population (millions) in 1980 Goal to be Service Already in Service not Regions Total Integrated % served Place or Financed yet financed (A) (B) (C/B) (C) (D) (C/D) (C-D) Northeast 1375 25 18 9.38 3.21 34.2 6.17 Southeast 2128 100 4.7 27.92 18.50 66.3 9.42 Frontier 451 4 0.9 3.81 1.27 33.3 2.54 Brazil 3954 129 3.3 41.11 22.97 55.9 18.14 Notes: a/ The population figures are BNH projections. b/ Brazil may not equal sum of regions due to rounding. Source: Calculated from BNH/SFS data. - 74 - companies has been lower than expected. 1/ In any case, the expected popula- tion served in 1980 under PLANASA, 23 million, is exactly the number of persons living in houses with connections to general sewer systems according to the 1973 PNAD, assuming 5 people per housing unit. 107. Table 29 provides evidence of an even more sobering measure of the gap between PLANASA sanitary sewerage goals and the situation in 1977. The total number of sewer connections in the state sanitation companies at the end of 1977 was 2.6 million, or only about 56% of the total number of urban dwell- ings indicated as served by sewer systems in the 1973 PNAD. Of the existing sewer connections, 85% were located in the Southeast. Whatever the nature of the service provided in 1973, it seems that many of the existing sewerage systems in that year still had not been integrated into the state sanitation companies by the end of 1977. Compared with 1980 PLANASA goals, and assuming 95% of the sewer connections were residential and six persons served per connection, 2/ only 36% of the total PLANASA target population for 1980 was actually served by the state sanitation companies in 1977. 108. Technical difficulties in establishing broader sanitary sewerage programs (basic surveys of each river basin and coastal area have had to be completed in order to indicate measures necessary to keep pollution within acceptable bounds), as well as the priority given to water supply for use of limited financial resources are the reasons given by SFS for the relatively low investments under PLANASA in Sanitary Sewer Systems to date. Table 30 shows the financial effort which has been made under PLANASA in sanitary sewerage programs over the period 1968-76 by sources and region, as in the case of water. Of the total investment, which began in 1970, 27.5% has been in the Northeast, 69.7% in the Southeast, and only 2.8% in the Frontier region. 36.6% of total investments were direct loans by the BNH, 12.0% supplementary BNH loans to the FAEs, 20.6% FAE resources other than BNH supplementary loans, and 30.8% fundo perdido and others. 109. Table 31 shows that total investments became substantial beginning in 1972 and rose to a new plateau of about Cr$1.2 billion (US$115 million) per year in the period 1974-76 (the Geisel administration), measured at 1976 average value of the UPC and exchange rate. More detailed breakdowns of the investments made through 1976 indicate that 79% of the total value was made in four states (Bahia, Pernambuco, Rio de Janeiro, and Sao Paulo). It is likely that in reality these investments were concentrated in four metropolitan areas: Salvador, Recife, Rio de Janeiro, and by far the most important, Sao Paulo. Of the sewerage investment programmed for the period 1977-1980, about 50% will take place in Sao Paulo alone. Table 32 shows that while investments in water 1/ Ivandro Mendonca Pires, "PLANASA: Avaliacao dos Resultados e Perspectivas," Paper presented at the 10th Brazilian Congress of Sanitary and Environmental Engineering, Manaus, January 21-26, 1979. 2/ Six persons per connection are assumed since some connections serve multiple dwelling units. In Sao Paulo 6.21 persons per connection is the figure used by the state sanitation company (SABESP). Table 29: PLANASA: NUMBER OF SEWER CONNECTIONS OF THE STATE SANITATION COMPANIES, 1972-77 Absolute Percentage Increase Increase Region 1972 1973 1974 1975 1976 1977 1972-77 1972-77 Northeast 89,385 112,094 127,367 145,856 164,900 207,651 118,266 (132.3) Southeast 1,401,871 1,566,648 1,631,558 1,885,663 2,042,790 2,193,806 791,935 (56.5) Frontier 60,276 64,750 98,807 105,999 147,804 178,214 117,938 (195.7) Brazil 1,551,532 1,743,492 1,857,732 2,137,518 2,355,494 2,579,671 1,028,139 (66.3) Source: BNH/SFS Table 30: PLANASA: SANITARY SEWER INVESTMENTS REALIZED BY SOURCES AND REGION, 1968-1976 (Millions of 1976 Cr$A' and percentages) % of Total Total '"Fundo Investment Investment BNH b, Perdido" Region Value Direct Supplementary Total FAE- and Others (A+C+D)h! (A) (B) (A+B)- (C) (D) Northeast (27.5) 1418.0 650.8 550.2 1201.0 721.4 45.9 Southeast (69.7) 3597.4 1185.4 22.6 1208.0 923.5 1488.5 Frontier (2.8) 144.4 54.5 47.4 101.8 36.3 53.6 b/ Brazil- (100.0) 5159.8 1890.6 620.1 2510.8 1681.2 1588.0 (%) (100.0) (36.6) (12.0) (48.7) (32.6) (30.8) Source: Calculated from Ivandro Mendonca Pires, "O Atual Estagio do PLANASA", paper presented at the 9th Brazilian Congress of Sanitary Engineering, Belo Horizonte, 3-8 July 1977, Annex 9. Notes: a/ At average value of UPC for 1976, Cr$ 149.63. b/ Numbers in this row or column may differ from sum of components due to rounding. c/ Includes Supplementary BNH loans. d/ Includes resources from World Bank loan to State of Sao Paulo. Table 31: PLANASA: SANITARY SEWER INVESTNENTS REALIZED BY SOURCE AND YEAR, 1968-1976 (Millions of 1Q76 Cr8 and Dnrcentages) Total "Fundo Investment BNH Perdido" Value Direct Supplementary Total FAEc/ and OthersA/ (A+C+D)b/ (A) (B) (A+B)W/ (C) (D) 1968 1969 1970 86.9 32.6 32.6 - 54.2 197-1 44.3 20.5 - 20.5 3.6 20.2 1972 548.5 109.7 26.3 135.9 123.6 315.3 1973 753.1 227.3 39.6 266.9 201.9 323.9 1974 1283.2 422.2 204.7 626.9 386.5 474.6 1975 1255.2 521.8 102.3 624.2 353.5 379.9 1976 1188.6 556.5 247.3 803.8 612.1 20.0 Total 1968-76k/ 5159.8 1890.6 620.1 2510.8 1681.2 1588.0 (M) (100.0) (36.6) (12.0) (48.7) (32.6) (30.8) Notes: a/ At average value of UPC for 1976, Dr$149.63. b/ N'umbers in this row or columan may differ from sum of components due to rounding. c/ Includes supplmentary BNII loans d/ lnclzides resources from World Blank loan to State of Sao Paulo. Source: Calculated from Ivandro Mendonca Pires, "O Atual Estagio do PLANASA" paper presented at the 9th Brazilian Congress of Sanitary Engineering, Belo Horizonte, 3-8 July 1977, Annex 9. - 78 - Table 32: PLANASA: SUMMARY OF INVESTMENTS REALIZED 1968-76, 1977 AND REQUIRED 1978-1980 TO ACHIEVE GOALS (Millions of UPC) 1968-76 1977 l978-80 Total Water Supply 114.7 31.2 88.6 234.5 Sanitary Sewers 34.5 9.9 50.1 94.5 Total 149.2 41.1 138.7 329.0 Source: Calculated from Ivandro Mendonca Pires, "0 Atual Estagio do PLANASA" paper presented at the 9th Brazilian Congress of Sanitary Engineering Belo Horizonte, 3-8 July 1977, Annex 14 and BNH supplied data for 1977. - 79 - supply during 1977 appear to have been at a level consistent with meeting the 1980 targets, this was not the case for sewerage, where 13% more must be invested in the three remaining years than has been invested in the entire pe-iod 1968-77. It is unlikely that the objective of providing adequate service in 160 municipios can be met by 1980. Estimated investment costs for new sanitary sewerage systems vary widely. A BNH/SFS document based on 48 contracts approved or pre-ratified prior to August 1975 indicated a linear cost function for a sewerage system with pumping stations and treatment (unspecified level) for cities up to 40,000 inhabitants of 4.5 UPC per person. 1/ Correcting for cost increases and depreciation of the UPC, an estimate of 5 to 6 UPC per person in 1976 prices could be derived from this. A study by the Sao Paulo State Sanitation Company (SABESP) completed in 1976 estimates investment costs for medium- and large-size cities of 10 UPC per person. Finally, the Greater Sao Paulo Sewage and Collection and Treatment Project which has been appraised by the World Bank would expand the number of connections in Sao Paulo by 550,000 and in general upgrade the entire sewage system. Total costs (including secondary treatment) for this project are very high if divided by the number of connec- tions added (120.37 UPC per connection). The project estimates 6.29 persons per connection, which gives 19.14 UPC per person. If the cost of the treatment plants are excluded (they would serve existing as well as new connections), the cost per connection falls to 92.16 UPC or 14.65 UPC per person. These diverse estimates suggest that sanitary sewerage costs may be considerably higher in large metropolitan areas than for medium-sized cities, but that smaller cities can satisfy their sanitary sewerage requirements still more cheaply, at about the same per capita cost as for water connections. Theoretical and empirical analyses of costs for water and sewerage systems are few. A recent review of the available literature coupled with an empirical analysis of Colombian data reached the following conclusions: "In summary, one may conclude that while declining unit costs may be associated with increased system capacity for water supply an sewerage disposal, offsetting factors cannot be neglected. First, cities do not grow in discrete jumps but continously over time. Thus, capacity must be added sequentially involving frequently units of constant, rather than increasing size. Second, as cities grow they tend to expand into areas which are more difficult and costly to service. Third, increased congestion and pollution, and reduced absorption capacity of the environment associated with increased city size result in the need for more extensive or intensive treat- ment at the source of water, or at the disposal of effluents. In concentrating on technological considerations and neglecting entirely what might broadly be called "environmental" factors associated with growing city size, the literature ... has postulated 1/ BNH/SFS, Assessoria de Analise e Programacao, "Sistema de Acompanhamento de Custos para Saneamento, Esgotos Sanitarios, 1a Parte" (Agosto 1975). - 80 - the hypothesis of declining unit costs for such urban services as water supply and sewerage. Consideration of the offsetting environmental factors leads us to a more cautious approach a priori to the question of economies to scale. Indeed, the empirical evidence in this paper appears to indicate that at least in the case of Colombia the assumption of diseconomies of scale is on balance more justified." 1/ For the purpose of making projections, a cost of about 9 UPC per capita is probably justified if a single cost is to be used. For a more sophisticated analysis, a step function with cost related to city size (say 5 UPC for cities up to 50,000, 8 UPC for cities in the 50,000 to 500,000 range, 10 UPC in the 500,000 to 1,500,000 range, and 15 UPC for large metropolitan areas might be used. 3. Major Factors Affecting PLANASA/SFS Performance 112. All the major factors which affect the performance of the popular housing finance system -- namely Brazil's large proportion of poor families, the need to maintain the long-term financial viability of the BNH, and the increase in real construction costs after 1972 -- also affect PLANASA/SFS performance. There are some differences, however, inherent in the collective nature of water and sewer systems and the financing mechanisms of PLANASA which merit comment. (a) Possibilities for Subsidization of Low-Income Consumers 113. The financial viability of a sanitation company requires that its receipts from operations exceed its operational plus its financial expenses. Given PLANASA objectives and the loan terms prevailing from the BNH and the FAEs, it is possible to project financial costs at any point in time for any sanitation company at the state, metropolitan, or municipal level. By law, the combined minimum water and sewerage tariffs for residential consumers cannot exceed 7% of the regional minimum wage and the minimum water supply tariffs cannot exceed 5% of the regional minimum wage. Tariff setting may pose a dilemma to the state sanitation companies. On one hand, they are to recover their costs and exercise strict financial discipline. On the other, social criteria, including public health considerations, suggest they expand their service to the poorest strata of the urban population. 114. Assuming fixed terms for BNH and FAE loans to operating sanitation companies, there are in essence two ways in which the rate constraints may be met if a system appears financially unviable. The first is by cross- subsidization through the rate structure within the system at any level (state, metropolitan, or municipal). This may be feasible if the capacity to I/ Johannes F. Linn and Nelson A. Valverde, "Public Service Costs and City Size: The Case of Urban Water Supply and Sewerage Cost Functions in Colombia" (Draft, IBRD Development Economics Department, June 1976). - 81 - pay off industrial, commercial, and more well-to-do domestic consumers is large enough and they are sufficient in number and demands on the system to support subsidization of low income consumers. In general this method works in the richer, more industrialized states. But in poorer states, particularly those with large numbers of relatively small cities and towns to be served, tariffs must either be set too high for the poor and/or even charging the maximum minimum tariff may not make it financially viable to extend service to many communities. The other option is the application of budgetary resources on a grant basis (a fundo perdido) to subsidize the sanitation companies at any level. The grant funds could come from the federal as well as state or municipal sources (to date the metropolitan regions have no independent tax base). Both options are possible within the PLANASA framework, and both can be used to reduce water and sewerage rates to very low income customers below the limits set under PLANASA. This may well prove necessary, especially for families with incomes below two minimum wages, given the other costs associated with housing which they must also bear. (b) The Weak financial Situation of States and Municipalities 115. In general, the financial condition of state and municipal governments has deteriorated over the 1970s, and some projections show that even Sao Paulo may be hard pressed to meet its current operating budget, much less finance new investments, in the remaining years of the decade. The situation is worse in the poorer states and municipios, despite the existing arrangements for revenue sharing which bring a fraction federal revenues to the states and state revenues to municipios. This suggests that the poorer states and municipios are likely to require even more assistance than they are already getting through the revenue sharing system and the SFS if PLANASA's ambitious goals are to be achieved. (c) Resistance by Operating Sanitation Companies 116, As pointed out above, some sanitation companies have resisted incor- poration into PLANASA. They have feared losing control over their rate struc- tures, expansion plans, and finances to state level sanitation companies which may have priorities which differ substantially from their own. The poorer companies, and municipios where there is no water or sewerage company, have felt less threatened by PLANASA than the richer, more established compa- nies which might lose control of their cash flow which could be used to expand systems other than their own. But since there is no other major source of funds, other than self-finance through the tariff structure, to finance new investments in water and sewer systems, it is likely that even the more prosperous municipal systems will be induced to join PLANASA to finance their expansion. 4. Policy Response to Constraints Affecting PLANASA/SFS Performance 117. Government basic sanitation policy has been set forth in the original PLANASA, the revisions made in PLANASA in April 1975, SFS regulations, a law governing tariff structlres and rates of return promulgated in May 1978 and the resources made available for basic sanitation investments, pri*cipally - 82 - through the SFS. The principal changes in response to the CoTIStratats discussed above, include (a) revision of water supply and sewerage targets for 1980 as discussed above, (b) changing BNH loan terms through the SFS, and (c) estab- lishing in 1977 a system to implement water supply systems in small communities on a large scale. 118. In April 1971, with the initiation of PLANASA, interest rates ([il addition to monetary correction) charged on BNH sanitation loans to financial agents were lowered from 10% to a range of 4 to 8%. In May 1975 the interest rate structure was again lowered to a range of 2 to 7%. The rate actually charged a financial agent is an increasing function of the state's tax receipts and its population density. These regulations provide for interest rate subsidies to the poorer and more sparsely populated states. The maximum spread between a financial agent's borrowing rate and the rate for lend Ll1, i:,) operating sanitation companies is 1%. The normal maximum amortization period is 18 years plus up to 3 years grace. But beginning in 1976 "special cases" were allowed 30 years for amortization plus 54 months grace. 119. Recognizing that a special effort would be necessary if a large number of small communities (defined as those with an urban population of 5,000 or less according to the latest census) are to obtain water supply systems under PLANASA, in 1976 BNR began establishing a special program (a) to help finance such systems, (b) to reduce the amount of red tape involved in incor- porating them into PLANASA by requiring less sophisticated feasibility studies for such systems, and (c) to reduce their cost by encouraging the use of simple, relatively cheap systems. This program became effective in 1977. On the financial side, a new sub-program was established under the Program for Sanitation Finance called the Supplementary Loan for Forming State Financial Participation in PLANASA Execution (Emprestimo Suplementar Para Composicao da Participacao Financeira dos Estados na Execucao do PLANASA - FINEST). FINEST is designed to encourage the states to increase their financial support for PLANASA in general and water supply systems in small communities in particular by making loans to state governments. 120. FINEST has two new subprograms. Under one subprogram, BNH provides financial assistance to the states for carrying out water supply projects in small communities. BNH makes loans to its financial agents for relending to their respective states. The amount of the BNH loans is equivalent to the amount that the states make available to their water companies, substantially as a capital contribution, for projects in small communities. Under the other, BNH provides financial assistance to the state water companies. BNH makes loans to its financial agents for relending to their respective states. The states make available the proceeds of BNH loans, together with an equivalent amount from their own budgetary resources, to their water companies, substan- tially as a capital contrbution, for projects in small communities. Interest rates (2% below normal BNH supplementary loans to FAEs), amortization periods, and loan limits for both subprograms are structured so as to provide an incen- tive for the states to increase their support for PLANASA implementation in small communities. This new program is expected to allow the state water companies to step up the level of their investments in such communities with- out unduly impairing their financial position. - 83 - 5. SFS-Specific Policy Issues (a) Subsidies Benefiting the Poor Assuming the interest rate structure for basic sanitation loans is already as low as is consistent with long-run financial viability of the BNH/ SFS/FGTS system, a key financial problem of the SFS is likely to be )btaining more funds to permit subsidization of those sanitation companies serving the poor, thus permitting lower minimum rates. The most serious problem exists in the poorest states, particularly in the Northeast. In such states the present system of cross-subsidization at the state ilevel through the tariff structure and at the federal level through the interest rate structure on BNH/SFS loans is least likely to be adequate if water supply and sewerage service to the poor is to be expanded significaatly. Of course, state sanitation companies should take full advantage of the National Tariff Law for the Water Supply and Sewerage Sector approved by the Brazilian Congress in May of 1978 by establishing progressing tariff structures and adjusting them so as to keep up with inflation. The stated objectives of the law are (a) to relate tariffs to service costs; (b) to ensure the financial viability of the state sanitation companies; (c) to put a ceiling on the cost of obtaining basic water supply and sewerage services for the low-income population by establishing a minimum consumption tariff related to income level; and (d) to set tariff levels which would produce the necessary revenue to pay for all costs and allow a rate of return of up to 12% on capital investel. The concept of the minimum-consumption tariff related to income level can be used to subsidize consumption deemed adequate to meet basic needs for those poor households with individual service. Beyond the minimum consumption level tariffs would reflect costs. But for water supplies available collectively through public standpipes or pumps rather than piped to individual residences (a system appropriate for small towns and villages and the poorest urban areas) relating charges to consumption levels and income is very difficult. If at the state level a tariff structure consistent with the National Tariff Law will not allow increased service to the poor, even after full use of the potential for cross subsidization at the state and federal level inherent in the present system, a method similar to that used in the el.tow-ric power sector for transferring financial resources between rich and poor states might allow at least partial resolution of this problem. There a special fund is used to subsidize the poorer state power companies and a federal regulatory agency carefully monitors physical performance tndicators to assure efficient use of the funds supplied. In the case of the water supply and sewerage sector, the first approach might be to subsidize capital costs by using the fund to make grants of equity capital to the state sanitation companies in the poorer states. In establishing sitch subsidies to state sanitation companies, criteria should be set to assure that the companies receiving the subsidies are well-managed and have set their tariffs according to nationafly' accepted standards. Then if the rate of return (revenues less operating coYsi-s as a percentage of the current value of capital including equity) is less than some minimum (say 5%), operating costs might be subsidized as well at a level suffi- cient to establish the minimum rate of return. - 84 - 124. If no transfer mechanism within the sanitation sector can be devel- oped, and perhaps even if it can, more federal budgetary resources to be applied on a grant basis are likely to be necessary. If the government wishes to achieve even the present 1980 goals, much less advance beyond them, more budgetary resources applied on a grant basis are likely to be necessary. Loans to state governments, such as the new FINEST sub-program discussed above, are in the last analysis limited by the debt service capabilites of the states, which are already strained especially in the poorer states. Given that the total non-reimbursable funds for both water supply and sewerage investments under PLANASA for all Brazil amounted to only Cr$166 million (US$15.4 million) in 1976, 1/ there would appear to be room for a federally financed effort of this kind. Of course, the FAEs may receive state (and federal, via revenue sharing) tax revenues, but these resources are eventually returned to the FAEs with monetary correction and interest and are thus the basis for the future autonomy of state level sanitation companies. It would not be realistic to expect a greater financial effort by the states unless new sources of state level taxation are obtained. (b) The Need for Comparative Cost Studies by Location and Technology 125. It is likely that the per capita costs of expanding water and sewage systems varies substantially between different Brazilian cities. More precise data on such urban infrastructure costs would be a helpful guide to elaborating a national urban policy which embraces industrial location, migration, transportation, housing and urban infrastructure. But beyond the comparative costs of the standard urban solutions of general network water and sewer systems with individual house connections, there is a need to explore alternative technologies much more thoroughly than has been done in Brazil to date. In the case of water supply, a mix of house connections and standpipe service for the poorer urban areas deserves consideration. But it is in examining carefully the alternatives to waterborne sewerage that the greatest effort should be made. The potential for large savings in terms of economic and environmental costs is high given the present low coverage ratios. Accord- ing to a recent World Bank research report, 2/ it is becoming very clear that this conventional developed country solution has proven technically, economic- ally, and culturally unsuitable for many countries in the developing world where it has been tried. Even in industrial countries, the enormous water requirements and very high environmental cost of the point discharge of treated waste necessitated by sewerage are leading to reexamination of its desireabil- ity. There are many technologies for exereta disposal between the unimproved pit privy and conventional sewerage which can be recommended for adoption subject to the physical conditions of the site and the social preferences and economic resources of the beneficiaries. They include improved pit latrines, pour-flush toilets, composting toilets, modified septic tanks, 1/ The PLANASA statistics for investments do not include state grants to state sanitation companies for general budgetary support. 2/ "Appropriate Sanitation Alternatives: A Technical and Economic Appraisal: Summary Report," The World Bank Energy, Water and Telecommunications Department, Public Utilities Notes (P.U. Report No. RES 20), February 1979. - 85 - vault and cartage, and smallbore sewerage. Except in unusual circumstances, scattered, rural and densely populated, and urban communities should find themselves with two or more technically feasible options, each with a range of design alternatives. The principal exception is very high density urban areas with high rise construction. One of the most important technical contributions of this research done for the World Bank is the design of "sanitation sequences": step-by-step improvements leading from one option to another and designed from the outset to minimize costs over the long run. These sequences also permit a decentral- ized approach to excreta disposal, at least in the early phases. A community may initially select one of the low-cost technologies in the knowledge that, as its socioeconomic status improves, the technology can be upgraded in a known series of improvements to a sophisticated "final" solution. This is something that is not possible with conventional sewerage. For it to function properly, large investments and large waterflows are needed from the outset. It is noteworthy that none of the sanitation sequences developed to date results in conventional sewerage. In urban areas the final upgrading is generally to a low volume cistern flush toilet connected to a vault which overflows into a small diameter sewer. From the user's viewpoint the only difference between this system and conventional sewerage is the size of the toilet tank in his bathroom. But the cost savings, particularly if it is installed by utilizing an existing facility, are significant. The installa- tion and use of such a system over a 30-year period costs about one half the present value cost of conventional sewerage. Its environmental cost is also significantly lower than that of conventional sewerage since its eventual waste discharge is less. As this example makes clear, sewers are required to dispose of sullage water (laundry and kitchen wastes plus flushing water), not excreta; and therefore the key element in an economic solution to sanitation is the reduction of non-essential water use. To this end, work is underway to adapt water-saving showerheads and other appliances for use in developing countries. This also leads to the conclusion that water supply schemes, even when the sanitation component is not included, should consider in their design and in the establishment of service standards the requirement to dispose of the water being supplied to the community and the costs (even in the future) associated with its disposal. It is not possible to recommend any given alternative to the general network sewer systems being implemented under PLANASA without careful study of the alternatives in each case. But, as noted earlier, even when sewers are constructed, economic constraints appear to limit the extent to which they are used by the intended beneficiaries. Therefore, what can be recommended with- out reservation is that the alternative technologies be seriously considered, since the solution which has been chosen to date under PLANASA is the most expensive one. - 86 - (c) Rural Sanitation 129. Finally, it is clear that the PLANASA/SFS framework has not made inroads on the serious problems of rural basic sanitation. The Ministry of Health, through FSESP, has been given the responsibility of "expanding water supply and sewerage services in rural areas" 1/ but the resources devoted to this task appear to be very small relative to needs. Provision of deep wells or other sources of safe drinking water and establishment of adequate latrines or septic tanks in small towns and villages are objectives which are critical to improving the quality of life in rural areas and could have a major positive effect on rural mortality and morbidity rates (see Annex III of this report). Setting such objectives would represent a natural expansion of PLANASA in the 1980s. In the case of water supply to do so would imply a major expansion of the state sanitation companies into a "market" where collective rather than individual service is likely to be the norm. The building of latrines might best be pursued through an expanded rural health service. Installation of septic tanks or deep wells serving isolated dwellings could be accomplished with the help of individual loans to those who can afford them, again making use of the Bank of Brazil's extensive branch offices, as in the case of rural housing loans. III. SIMULATION EXERCISES 130. Previous discussions indicates that available estimates of key quantitative variables associated with the provision of housing related services (e.g., the number of dwellings connected to a general water or sewerage network) since the 1970 demographic census -- or at best, since the 1972/73 PNADs -- are not altogether reliable. Moreover, even when estimates of the number of facilities in place appear reasonably reliable, there is little to go on in assessing the extent to which such facilities are operational (for example, the extent to which the general network systems deliver potable water consistently). These and other considerations such as non-economic factors affecting the supply of these services ought to discourage attempts to forecast the future evolution in the provision of housing related services. Yet, there may be some interest in performing certain arithmetical exercises to explore conditional implications concerning plausible alternatives for the future. 131. Various approaches can be taken to the stipulation of assumptions, depending on the purpose of the simulations. In this case there are three objectives: 1/ BNH, Modificacoes No Sistema Financeiro do Saneamento (Rio de Janeiro: BNH, 1975) p. 19. - 87 - .To derive some broad orders of magnitude concerning the time schedule involved in expansion of coverage of housing related services in the future, assuming a continuation of certain historical trends. Type I simulations address this objective. . To derive some broad orders of magnitude on the amounts of public sector outlays that might be required to meet date specific targets, (e.g., to achieve a coverage ratio of 100% by year 2000) in the provision of housing related service; Type II simulations serve this purpose. . To set ranges of plausible value to quantify relations between the availability of safe water and sanitary sewerage, on one hand, and mortality, a key variable in the demographic projections that are part of this report (See Annex I). This is of interest in exploring the likely demographic impact of alternative coverages of these services. 132. The sub-sectors involved in these simulations are: (i) adequate dwellings, (ii) access to water and (iii) sewerage. Concepts of service differ depending on the urban or rural location of the dwelling. Adequate Dwellings 133. Although census and 1972 PNAD data exist for "durable dwellings", it has already been noted that this criterion is an extremely modest one for classifying dwellings as inadequate -- particularly, for dwellings in urban areas. It would become increasingly so as Brazil becomes a wealthier nation in the years ahead. A more reasonable standard of basic adequacy for dwellings in the future would seem to be one derived from the current PROFILURB sites and services projects. Consequently, for the purpose of the simulations, we define: . Adequate urban dwelling, as a residential structure located on an urbanized lot served by roads and drainage, and equipped with connections to water and sewerage networks. For the purpose at hand, this definition has the advantage that reasonably solid unit cost estimates can be obtained from the ongoing PROFILURB program and World Bank appraisals of specific projects within the program. Its disadvantage is that there is no correspondingly defined category in Brazilian nationwide data sources. To operationalize this definition in the historical context, a one-to-one correspondence is assumed to exist between adequate urban dwellings and dwellings enumerated in the demographic census as having "piped water", based on the conjecture that by and large, residences with this service will also meet the other criteria for adequacy. - 88 - Adequate rural dwelling, as a residential structure meeting the IBGE's criteria of durability (see footnote to Table 1). The regional distribution of adequate rural dwelling coverage is assumed to be proportional to the distribution of coverage of rural dwellings with any sanitary device in the 1970 base year. Access to Water 134. In urban areas, a dwelling is defined to have access to water if it is connected to a general water network by internal plumbing, and for rural areas, if it is either connected to a general water network or to a well. Sewerage 135. In urban areas, a dwelling is defined as having sewerage if it is connected to a general sewerage network; and for rural areas, if it has any sanitary device. For the purpose of Type II simulations, rural water and sewerage projections are treated as a package, it being assumed that future installations to any particular dwelling will provide both these services at once. 136. All simulations stipulate that GDP will grow at 6% per annum in 1978 and 1979 and use either a 4% or a 7% per annum assumption concerning GDP growth during the period 1980-2000. 137. The rate for 1978-79 is the "consensus" number being used by Brazilian official agencies and by the Bank. The 4% and 7% rates limit a plausible range for GDP growth over the long term. The average annual rate of growth of GDP for the period 1921-1947 was 4.4% p.a.; 1/ for the period 1948-1976, it was 7.4% p.a. 2/ A. Type I Simulations 138. The outcome variables in these simulations are future dates in which specified coverage ratios for the different services are attained under alter- native assumptions concerning trends. These simulations come in pairs; each outcome variable is derived twice. Once, on the basis of assumptions that are considered "pessimistic"; and the other, based on "optimistic" assumptions -- in relation to historical trends. The guiding principle for deciding how pessimistic or optimistic the assumptions would be was that they should approximate boundaries outside of which actual future values for the corres- ponding variables seem very unlikely to fall -- although these boundaries may conceivably be approached. 1/ M. H. Simonsen, Brazil 2002 (Rio de Janeiro: Ap,-,j 1977). 2/ Annex II, Table 9, p. 17. - 89 - 139. These assumptions are stipulated without attempting to spell out what sorts of policy and implementation strategies would be required to achieve a given coverage ratio. An alternative approach was explored and abandoned on the grounds that the detailed sectoral planning and programming required to attain any measure of "operational validation" to the choice of prescribed numerical values is beyond the scope of this report. Consequently, the assumptions are offered as the mission's best guesses of plausible bounda- ries for the variable in question. To the extent possible, these conjectures draw on certain empirical associations between per capita income and coverage ratios in Brazil and on the findings concerning public sector involvement in the housing, water supply and sewerage sectors as reported in Section II of this annex. 1. Method and Assumptions 140. Type I simulations rely heavily on empirical relations between per capita income and coverage ratios for the different services. The fundamental macro-economic rationale for a positive relation to exist is that these services are superior goods. Whether they are transacted as private goods (i.e., dwellings) or as public goods (e.g., a general water or sewerage net- work), the ability and willingness of households to pay (effective demand) for them increases as households, communities and the nation become wealthier. 141. Using 1960 and 1970 state level census data, various statistical exercises were performed to identify robust, yet simple functional specifica- tions relating per capita income to coverage ratios for various housing related services in Brazil. Some of these exercises are reported in Appendix A. The most consistently satisfactory specification turned out to be a semi-log form relating first differences in coverage ratios to first differences in the logs of per capita income through a single parameter, the "service-income coefficient" (SIC). 1/ This parameter i.s estimated as the "regression coefficient" for the equations specified in Appendix A, Table A.1. 1/ In symbols, let: R ; R = the coverage ratio for service(s) at time o and t, o,s t,s respectively; Y ; Yt = per capita income at time o and t, respectively, then, R - R = b(ln Y - ln Y ), where b, the "service-income s,t s,o coefficient" is a parameter that generally differs for each service. - 90 - 142. While coverage trends since 1950 indicate that progress in the Northeast has been generally faster than in the Southeast when it is measured in proportional changes in curresponding coverage ratios (Section I), this ts not the case when progress is measured in percentage point terms. In the absence of a plausible basis on which to postulate different rates of progress (by the latter standard) in the simulation, it is assumed in the exercise that such a rate of progress will be i.d i;.al for all regions; except, of course, when aity o ae region reaches the 100% upper coverage limit before the end of the simulation period. Consequently, to simplify the exposition, reference to regional values is generally omitted from the discussion. 143. With the exception of adequate urban dwellings, empirical data is available for at least two point-. itl i:t-ne to permit direct estimation of "historical trends" in terms of the SIC. In the case of adequate urban dwellings, the SIC is derived under the assumption of a one-to-one corres- pondence between those and dwellings with piped water, and using historical data for the latter. 144. In general, the pessimistic simulations use the historically derived SIC, the low GDP growth assumption, and the Baseline population projections (refer to Annex I) to derive coverage ratio values at future dates from the 'service-income relationship" described above. The optimistic simulation, on the other hand, adopts a SIC that is higher than the empirically derived one for the same service, the high GDP growth rate assumption; and the same Baseline population projection 1/ to derive the corresponding coverage ratios using the same formula. 145. The decision not to scale down the historically derived SIC for the pessimistic simulations is based on two considerations. First, improving living conditions of the lower income population has increasingly become an explicit objective of Brazilian policy makers and the importance attached to this objective is unlikely to decline in the future. Second, cross section statistical estimates of SICs for certain housing related service using first 1960 data and then 1970 data indicate a slight increase in the value of this coefficient. Therefore, it seems implausible to assume that for given increments in per capita income in the future, coverage improvement in those services will be less than in the past -- at least until the coverage ratios approach the (100%) upper bound. 146. The extent to which the historically derived SIC is scaled up for the optimistic projections is methodologically arbitrary, but influenced by factors such as titi- flli'1' S assessment of trends and prospects in the priority that the government has been giving to the different services, institutional factors tending to influence the success of ambitious implementation programs, logistical cornplexities, and the like, as they differ acro- . ;hes. I/ For this purpose, using an alternative population projection would make no significant difference. - 91 - 147. As a general rule, simulations for the period 1970-1980 use pessimistic SICs derived from 1970 census data and 1972 or 1973 PNAD data to reflect shorter term trends in the respective service. This seems particular7y important in the case of urban services. Ambitious new programs to accelerate expansion of these services at the national level got underway in the 1960s: the Housing Finance System, in 1964; and the National Sanitation Plan, in 1971. Since the full impact of these programs would not yet have been reflected in the 1972/73 data and since additional programs to further strengthen those efforts were launched after 1972 1/, this approach seems appropriately pessi- mistic. For the period beyond 1980 pessimistic simulations generally use SICs derived from longer term intervals: either 1950 to 1970 or 1960 to 1970, depending on data availability. 2. Results Adequate Dwellings 148. Recalling that adequate urban dwellings are assumed to have a one-to- one correspondence to urban dwellings with piped water, the coverage ratio for this service starts out at 53% in the 1970 base year (Table 33). For the pessimistic case, the SIC derived from urban piped water data, takes values of 0.19 for the 1970-80 simulations and 0.20 for 1980-2000. 149. The resulting coverage ratio, rises to 63% in 1980, and then more slowly to 65% in 1990 and 69% in the year 2000. The faster rise in the 1970s reflects primarily the effect of the more rapid GDP growth estimated and assumed for this decade, compared to the 4% p.a. "pessimistic" rate postulated for the remainder of the century. The somewhat faster progress in the 1990s relative to the 1980s is the result of the slower projected population growth in the final decade. 150. For the optimistic case, the SIC is augmented by 30% of its 1970-73 derived value, to 0.25. Together with the 7% optimistic GDP growth rate beginning in 1980, their combined effect gives a coverage ratio that by the year 2000 is almost 30% greater than in the pessimistic case. 151. Since the operational definition of adequate rural dwellings is the more lenient one of "durability" of the structure, 1970 coverage for thlis service is somewhat higher than for urban areis (Table 3.1). The pessimistic simulation uses a SIC of 0.19 for the 1970-80 period and 0.21 for 1980-2000. These coefficients are derived from data on rural dwellings with any sanitary device (refer to Table 4) because the PNAD 1972 data for rural durable dwellings yield are implausibly large increment when compared to 1970 census data (refer to section I). The procedure is likely to give a downward bias to the estimate of the "true" SIC, since the coverage ratio of 1/ Notably, the National Popular Hou-sing Plan, initiated in 1973; the Program for' -iLtling of Urbanized Lots and the Program to Finance the Construction, Expansion or Improvement of Popular Housing, both initiated in 1975. - 92 - Tabl1e 33 ADEQUATE DWELLING COVERAGE RATIOS: ESTIMATES, 1970; SIMULATION RESULTS, 1980 - 2000 (percentage of households) 1970 1980 1990 2000 Base Pes. Opt. Pes. Opt. Pes. Opt. A. Urbana! Northeast 25 35 39 37 49 41 60 Southeast 62 72 76 74 86 78 97 Frontier 34 44 48 46 58 50 69 Brazil 53 63 67 65 77 69 88 B. Ruralb/ Brazil 57 67 69 70 80 73 91 a/ Urban adequate housing in 1970 is equated to urban dwellings with piped water. Changes in the regional share of urban households throughout the simulation period are not taken into account in deriving the results. b/ Rural adequate housing in 1970 is equated to "durable dwellings". No plausible empirical basis was found to allow regional disaggregation of rural adequate housing coverage in base year. Abbreviations: Pes. = Pessimistic simulations Opt. = Optimistic simulations - 93 - "any sanitary device" for rural dwellings in 1970 is over twice as great as that of "durable dwellings". Therefore, the derived SIC is probably appropriately pessimistic. In this simulation, the coverage ratio reaches 73% by the year 2000. 152. For the optimistic case, the SIC is scaled up by 15% of its 1970-73 value for 1970-80 simulations and by 30% of its pessimristic 1980-2000 value, to 0.25, for later simulations. A rationale for the different scalo- up factors is that forceful national programs to improve housing conditions in rural areas are not yet underway -- although, "optimistically," there is some indication such programs may be launched within the next few years. In this simulation, the coverage raLto in the year 2000 is almost 25% higher than in the pessimistic case. Access to Water 153. The procedure used in these simulations is similar to the one used for adequate housing. It is also somewhat less speculative because the available empirical data is congruent with the definitions used f)r- this service. 154. In 1970, nearly 54% of all urban dwellings were connected to a general water network, which is the definition of urban access to water used in the simulations (Table 34). The SICs are derived from 1970-73 data for each of the three regions, yielding values of 0.32 for tH Nurtheast, 0.21 for the Southeast and 0.26 for the Frontier. These are used in the pessimistic case simulation for the period 1973-80 1/. As noted earlier (Section II), the National Sanitation Plan came into being in 1971, and water supply investments under this program in 1971 and 1972 were small compared to levels attained thereafter (Table 26). For the period 1980-2000, the SIC was derived statistically from 1970 state-level data, yielding a value of 0.26 (Appendix A). According to these (pessimistic) simulation results, the urban water coverage rises from 54% in 1970 to 67% in 1980 and then, at slower pace, to 75% in the year 2000. 155. For the optimistic simulations of urban water, the National Sanita- tion Plan target of covering 80% of the population in 80% of the munici- palities and all metropolitan areas by the year 1980 is equalled to an 80% coverage ratio for all urban locations in all regions in that year, and is accepted as such as a plausible, albeit ambitious, upper bound for the eventual outcome. This target implies a SIC of 0.46 for the 1970-80 period. Since it appears unlikely that this level of effort would be maintained over the longer term, the SIC is scaled down to 0.36 for the period beyond 1980. (This value is the mean of the SIC used for the pessimistic case (0.26) and the 0.46 coefficient). This would yield coverage ratios of 95% in 1990 and 100% by the early 1990s. 1/ In these simulations, the 1973 data is deemed sufficiently reliable to be used as the base. - 94 - Table 34: ACCESS TO WATER COVERAGE RATIOS: ESTI4AiTES, 1970 AND 1973; SIMULATION RESULTS, 1980 - 2000 (percentage of dwellings) 1970 1973 1980 1990 2000 Pes. Opt. Pes. Opt. Pes. Opt. a! A. Urban- Northeast 29 38 46 80 50 95 54 100 Southeast 63 69 74 80 78 95 82 100 Frontier 40 47 54 80 59 95 63 100 Brazil 54 61 67 80 71 95 75 100 B. Rural- Northeast 1 1 10 13 15 33 21 54 Southeast 11 17 26 29 31 49 37 70 Frontier 3 n.a. 15 18 20 38 26 59 Brazil 6 n.a. 18 21 23 41 29 50 Notes: Changes in regional shares among urban dwellings and among rural dwellings are disregarded here. a/ Urban access to water is equated to urban dwellings with general network connection. b/ Rural access to water is equated to rural dwellings with piped water. Abbreviations: Pes. = Pessimistic simulations Opt. = Optimistic simulations. - 95 - 156. The pessimistic simulati-ons for rural water (defined as rural dwellings with piped water) use a single SIC of 0.36 for the entire period 1973-2000. This is the statistically derived value from 1970 state-level data. In the simulation results, the coverage ratio increases from 6% in 1970 to 18% in 1980 and to 29% in the year 2000. 157. Optimistic simulations for rural water use a SIC value of 0.41 for the 1973-80, which is about 15% higher than the corresponding value used in the pessimistic case. For the period beyond 1980, the SIC is scaled-up by 30%, to a value of 0.47. The simulated results are a 21% coverage ratio by 1980 and 50% by the year 2000. Sewerage 158. Similarly to the case of water, urban sewerage disposal service is defined in the simulations as urban dwellings connected to a general sewerage network. In 1970, the coverage ratio for this service was 22% and data from the 1973 PNAD implies that in that year the corresonding ratio had increased to 37%. This seems highly implausible. It implies a SIC of 0.53, which is substantially beyond the range of values for the same period derived for each region with respect to general network water (the maximum value being 0.32, for the Northeast). Furthermore, PLANASA investment during the 1969-73 period in urban water supply exceeded its corresponding one in sanitary sewerage by more than five times (Tables 26 and 31). Consequently, other things being comparable one would expect lower SIC values for sewerage than for water. 159. Thus, instead of using the 1970-73 coverage data as a basis for computing the SIC, the statistically derived coefficient from 1970 state- level dat-i n) 0.24 is used in the pessimistic simulations for the entire period. For the optimistic simulations, this coefficient is scaled-up by 15% to 0.28 for the 1970-80 period and by 30% to 0.31 for the 1980-2000 period. The simulations yield coverage ratios of 35% and 38% by 1980, and of 42% and 65% by the year 2000 under the pessimistic and optimistic cases, respectively (Table 35). 160. Pessimistic simulations of rural sanitation (rural dwellings with any sanitary device) use a SIC of 0.26 -- statistically derived from 1970 cross section data -- for the entire simulation period. Optimistic simulations scale-up this value by 15% (to 0.39) for the 1970-80 period and by 30% for the 1980-2000 period. Simulations yield 40% and 48% by 1980 and 48% and 81% by the year 2000 under tii- pessimistic and optimistic simulations, respectively. Summary of Results 161. The simulations suggest that: if Brazil should follow a path of moderately low GDP growth -- averaging 4% per annum during the two final decades of the current century; if population growth slows down somewhat -- from the current annual rate of about 2.8% to 2.2% by the year 2000; and if historical quantitative relationships between per capita income and coverage of certain housing related servii(4', ;'emain roughly stable in the future; - 96 - Table 35: SEWERAGE COVERAGE RATIOS: ESTIMATES, 1970; SIMULATION RESULTS, 1980 - 2000 (percentage of dwellings) 1970 1980 1990 2000 Pes.- Opt. Pes. Opt. Pes. Opt. A. Urbana/ Northeast 5 18 21 21 34 25 48 Southeast 28 41 44 44 57 48 71 Frontier 11 24 27 27 40 31 54 Brazil 22 35 38 38 51 42 65 B. Rural-b Northeast 6 20 28 24 44 28 61 Southeast 42 56 64 60 80 64 97 Frontier 26 40 48 44 64 48 81 Brazil 26 40 48 44 64 48 81 Notes: Changes in regional shares among urban dwellings and among rural dwellings are disregarded here. a/ Urban sewerage is equated to urban dwellings with general network connection. b/ Rural sewerage is equated to rural dwellings with any sanitary device. Abbreviations: Pes. = Pessimistic simulations; Opt. = Optimistic simulations. - 97 - then progress in the proportional expansion of the services between now and the year 2000 will continue to occur, but at a pace that could be considered slow. By century's end: . roughly a quarter of Brazilian households would be living in dwellings that are considered "sub-standard" by most Brazilian policy-makers and World Bank specialists today (compared to about one half in 1970) -- for urban households, this proportion would probably be somewhat higher; . roughly a quarter of urban dwellings would lack connections to a general water network (compared to about one half in 1970); . roughly two thirds of rural dwellings would lack piped water; (compared to more than 90% in 1970); . roughly six out of ten urban dwellings would lack connections to a general sewerage network (compared to more than three quarters in 1970); and . roughly half of all rural dwellings would lack any sanitary device (compared to about three quarters in 1970). 162. On the other hand, the simulations suggest that: if Brazil could sustain an average GDP growth rate of 7% per annum between 1980 and 2000; if population growth slows to 2.2% per annum by the end of the century; and if forceful and effective policies, such as those being currently pursued in the area urban water supply are extended to the other housing related sectors (e.g., policies such as those suggested in Section II above) within the next two years; then coverage progress in the remainder of this century would proceed at a pace that could be considered reasonably fast. In this case, by century's end: only about one out of ten urban households would be living in dwellings that are currently considered sub-standard -- and this proportion would probably be slightly lower in rural areas; virtually all urban dwellings would be served by a general water network; roughly half of all rural dwellings would have piped water; roughly two thirds of all urban dwellings would be served by a general sewerage network; and roughly 2 out of 10 rural dwellings would lack any sort of sanitary device. While this progress may appear satisfactory from the point of view of the urban population, the significant proportion of rural households without these basic services by the end of the century seems disappointing. - 98 - B. TYPE II SIMULATIONS 163. The question of what would be the rough dimensions of the financial outlays that would be required by the public sector to attain ceriain coverage targets that are more ambitious than the optimistic simulation results is of some interest. A related question is whether such financial outlays would imply an unrealistic fiscal stress. These questions beg the crucial issue of the role of non-financial constraints--such as organizational, staffing and logistical bottlenecks in drastically accelerating the pace at which coverage for the services is extended. Yet it may still be of interest to explore the extent to which "money is a problem". Type II simulations address such preoccupations. 164. Deriving magnitudes of financial outlays that would be required by the public sector to attain certain coverage targets by future dates is an inherently questionable exercise. It involves making numerical assumptions that are quite arbitrary within a broad range of plausibility and then perform- ing one or two simple arithmetic operations to arrive at results. Consequently, any such results are vulnerable to the criticism of being predetermined by questionable assumptions; it is too easy to construct desired results by ,inple manipulation of stipulated values. Since the criticism is fundamentally valid, the magnitudes of required public sector outlays derived from such exercises are interesting only to the extent that they turn out to be "large" ("small") in spite of underlying assumptions at, or beyond, the boundaries of plausibility leading to downward (upward) biased esitmates of the required outlays. 165. For the present purpose of investigating whether "money is a problem" in attaining coverage targets that are ambitious by historical standards, it is therefore appropriate to choose numerical assumptions at boundaries of plausibility tending to overstate the required amounts. These assumptions bear on: (a)-the number of households requiring each of the services of interest by specified future dates; (b) the extent to which public sector outlays would be involved in meeting such requirements; and (c) unit costs in terms of public sector outlays. 1. Households 166. Numbers of urban and rural households at future dates are derived from the Baseline population pr-,,'ections presented in Annex I assuming average household size of 4.8 and 5.3 for urban and rural families respectively. 1/ The resulting projections of urban and rural households, and 1/ Data from the four decennial demograplhic censuses from 1940 to 1970 yield average urban household sizes ranging from 5.2 in 1940 to 4.9 in 1970; for rural households, 5.3 in 1940 and 1950, and 5.4 in 1960 and 1970. - 99 - their 1970 base, appears in Table 36. 1/ Since differences betwee[ cnle numbers of permanent dwellings and households were insignificantly small in all regions and locations in the year 1970, the projections assume that by i9,0 and beyond, the number of dwellings equals the number of households. Table 36: HOUSEHOLDS BY URBAN AND RURAL LOCATION: ESTIMATES, 1970; ASSUMPTIONS, 1980-2000 (Thousands of households) 1970 1980 1990 2000 A. Stocks Urban 10,899 16,518 22,763 29,813 Rural 7,703 8,488 9,751 11,343 Total 18,602 25,006 32,514 41,156 B. Decennial 1970-80 1980-90 1990-00 Net change Urban 5,619 6,245 7,050 Rural 785 1,263 1,592 Total 6,404 7,508 8,642 Source: Baseline population projections (Annex I). 2. Adequate Dwellings: Urban 167. About 53% of all urban households were living in an adequate dwelling in 1970 accordLnig to previously reported estimates; this coverage ratio would reach 88% by the year 2000 under the optimistic case of Type I simulations (Table 33). For the purpose of Type II simulations, suppose instead a more optimistic targ.1-i oF lt)0% by century's end and a linear path in the coverage ratio during the period 1980-2000, assuming that the Type I pessimistic coverage ratio of 63% were to be attached in 1980. 1/ Although use of an alternative population projection presented in Annex I would have produced slightly higher numbers of households in the future (by a maximum of less than 3% by the year 2000), this would not have made a material difference in the qualitative implications to be derived from the exercise at hand and would have introduced additional complications. - 100 - Previously discussed estimates indicate that during the period 1964- 1972, the BNH-centered Housing Finance System participated in the financing of dwellings for considerably less than 30% of net additional households established in urban areas during the same period (refer to Table 10 and corresponding text). Yet, the available data indicate a marked improvement in the coverage ratio for adequate urban dwellings in the interval 1960-72. To err on the high side, the calculations assume that public sector outlays are used in financing 40% of net additional urban adequate dwellings built from 1970 to 2000. The remainder, plus all replacements of existing adequate dwellings are assumed to be financed through the private sector. Unit cost to the public sector for the adequate dwellings it finances is set at US$3,800 1/ for the entire simulation period. 2/ This amount includes: roads and drainage (US$600) + site and services (US$1,100) + construction materials (US$2,100). Site and services costs covAprise land purchase and terracing for an "average" individual lot of abou-t 10 i2, and on-lot connections to water and sewerage networks. Construction materials correspond to those necessary to build a 25 m2, two-bedroom "embryo" house. Labor costs are assumed to be met by the beneficiary -- e.g., by using family labor and that of friends (the multirao systemi of mutual assistance). These unit costs correspond to maximum estimates derived from current PROFILURB and other low- cost housing projects being designed and implemented by BNiI (the World Bank is also involved in some of these projects). They are slightly higher than the typical cost estimate being used by BNH and the Bank in its projections, which range up to US$3,680 for the same set of items. Simulation result derived from the foregoing assumptions are reported in Table 37. The computed average annual public sector outlays for the 1970s is approximately 51 million UPC. While this amount is large compared to the value of BNH popular housing loans during the period 1973-77, it is considerably less than the value of all BNH housing loans in the same plrtod (refer to Table 11, above). Average annual popular housing loans in that interval amounted to 17 million UPC; average annual total loans amounted to over 90 million UPC. In comparison, the average annual value of BNH housing loans during the period 1975-7? came to about 1.15% of GDP; the corresponding share of GDP accounted for by "popular housing" loans was 0.27%. The derived public sector outlays accumulated over each of the current and future decades in this century turn out to be less than 0.6% of the corresponding "pessimistic" GDP projecVon. Under the pessimistic GDP 1/ This, and all other US$ amount presented here, are in 1976 constant dollars. 2/ One justification for deciding against having this unit cost rise with either time or per capita i.n'i'ne is that the cost of renting a dwelling in the U.S. fell by over 13% (relative to the all-items consumer price index) between 1950 and 1977. (Source: U.S. Department of Commerce, Statistical Abstract of the United States, 1977; Table No. 770, p. 478). - 101 - growth assumption of 4% annual increase in the period 1980-2000, the percentage share of GDP accumulated over 10 year periods would rise slightly from 0.5% in 1971-80 to a maximum of 0.57% in 1981-90 and then decline in 1991-2000 (Table 43). Thus, the simulations indicate that under the stated assumptions, meeting these seemingly very optimistic targets on urban housing would not involve any additional strain on the Brazilian economy. However, given the financial constraints imposed on the BNH at present, (specifically that it must loan money at above the average cost of funds and administration in order to be able to subsidize popular housing loans) an effort of the magnitude outlined here would probably require subsidies from outside the BNH/FGTS system. The subsidy system which exists now (see paragraphs 64-67 above) provides a basis for such transfers. Very rough calculations indicate that the 0.27% of GDP that the BNH was devoting to popular housing around 1975-77, "required" about 0.07 percentage points worth of extra-BNH subsidies. Extend- ing this logic the achievement of the 0.6% GDP share for popular housing indicated in the simulations and noted above would require increasing the subsidy rate to about 0.4% of GDP. Table 37: ADEQUATE URBAN DWELLINGS, TYPE II SIMULATIONS; AVERAGE ANNUAL PUBLIC SECTOR OUTLAYS BY DECADE, 1970-2000 1970 1980 1990 2000 A. Stocks (in year) Households (in thous.) 10,899 16,518 22,763 29,813 Coverage ratio 0.53 0.63 0.82 1.00 Adequate dwellings (in thous.) 5,776 10,406 18,666 29,813 1971-80 1981-90 1991-00 B. Flows (in decade) Av. annual net additions (in thous.) 463 826 1,115 Av. annual public sector outlays (in mills US$'s) a/ 703 1,255 1,695 a/ In constant 1976 dollars, at US$3,800 per dwelling, assuming the public sector finances 40% of all net additions. - 102 - 3. Adequate Dwellings: Rural 172. Around 57% of all rural households were living in an adequate dwelling in 1970 according to previously reported estimates; this coverage ratio would reach 91% by the year 2000 under the optimistic case of Type I simulations. For the purpose of Type II simulatiorm,, suppose instead a more optimistic targel (>F 100% by century's end and a linear path in the coverage ratio during the period 1980-2000. 173. In the past, the public sector has had no significant involvement in financing rural dwelling construction. Yet the evolution of housing sel-ri-ce indicators in rural areas since 1950 (Table 3 and 4) suggest that the share of adequate housing has increased substantially since that year. Type I simulations therefore assume no public sector financing of rural dwellings. Public sector outlays required to meet lil11e Type II targets are derived by calculating the difference between the net additions to adequate rural dwell- ings under the Type I pessimistic simulation coverage ratios and the Type II targets (Table 38). Such differences are then multiplied by US$2,100 per dwelling -- wlilh i s the same unit cost for construction materials uNi>d in the case of urban dwellings, assuming that a comparable dwelling unit would be built. Table 38: ADEQUATE RURAL DWELLINGS TYPE II SIMULATIONS; AVERAGE ANNUAL PUBLIC SECTOR OUTLAYS i3Y 0)0CADE, 1970-2000 1970 1980 1990 2000 A. Stocks (in year) Households (in thous.) 7,703 8,488 9,751 11,343 Coverage ratio 0.57 0.67 0.84 1.Q() hdequiate dwellings (in thous.) 4,390 5,687 8,191 11.343 1971-80 1981-90 1991-00 B. Flows (in decade) Av. annual net additions (in thous.) a/ - 136 170 Av. annual public sector outlays (in mills US$'s) b/ - 286 357 a/ Additions under Type II coverage target in excess of those that would be forthcoming under Type I pessimistic coverage ratios. b/ In constant 1976 dillars; at US$2,100 per public sector finance(d ilet - 1 #l.ion. - 103 - 174. The public sector outlays derived from the assumptions turn out to be much smaller than for urban dwellings: Public sector outlays for this purpose remain below 0.2% of GDP for the two decades between 1981 and 2000 on the same basis of comparison used before (Table 43). 4. Access to Water: Urban 175. Under the optimuistic case in Type I simulations for urban access to water, the coverage ratio increased from its 54% level in 1970 to 100%, attained during the last decade of the century. This path in the coverage ratio seems sufficiently optimlsiei- to be adopted as the target schedule for Type II simulations. By definition, access to urban water in the simulations entails connections to a general water network. Current and expected policy trends, as well as the stated objective of biasing simulation results towards iv.eh public sector outlays, are reflected in the Fact that all construction or extension of general water networks are entirely financed through the public sector. Unit cost, reflecting average investment cost per dwelling connection, is set at US$500 throughout the simularLon period. 1/ This amount is higher than the typical estimates being used by BNH and the Bank in their projections, which range up to US$420. To the extent that user charges allow for amortizing invested capital, as Bank financed projects are intended to, investment costs assumed in the project-imis will tend to overstate net public sector outlays, increasingly as the projection period proceeds. 176. Simulation results derived from these assumptions show average annual public sector outlays for this service during the 1971-80 period at a level comparable to the total value of investments realized in this sector by BNH and other public sector agencies in the year 1976. The former (simulation) amount is 6% higher than the latter (Tables 26 and 39). In comparison to public sector outlays derived for urban adequate housing from the Type II simulations (Table 37), the amounts are small; and consequently, they also turn out to be a small fraction of projected GDP under the same assumptions used before (Table 43) -- particularly beyond the first decade considered. I/ ConcerTlinO i-il' ;u3sumption of constant unit costs over time, data for U.S. urban households indicate that service charges per household for water and sewerage services, plus telephone service, increased by less than 2.6% (total) in real ter.is during the period 1960 to 1977 (Source: U.S. Department of Commerce, Statistical Abstract of the United States, 1977, Washington, D.C. 1977, Table 771, p. 479.) - 104 - Table 39: URBAN ACCESS TO WATER, TYPE II SIMULATIONS; AVERAGE ANNUAL PUBLIC SECTOR OUTLAYS BY DECADE, 1970-2000 1970 1980 1990 2000 A. Stocks (in year) Dwellings (in thous.) 10,276 16,518 22,763 29,813 Coverage target 0.54 0.80 0.95 1.00 Dwgs. w/access to water (in thous.) 5,549 13,214 21,625 29,813 1971-80 1981-90 1991-00 B. Flows (in decade) Av. annual net additions (in thous.) a/ 766 841 819 Av. annual public sector outlays (in mills US$'s) b/ 383 420 409 a/ In constant 1976 dollars; at US$500 per addition connection. 5. Urban Sewerage 177. Type II simulations on urban sewerage are analogous to those dealing with access to urban water. However, instead of adopting the Type I optimistic simulation as Type II targets, these are stipulated at far more ambitious levels. Instead of rising from 22% in 1970 to 65% in 2000, the coverage ratio is set at its upper bound by the end of the century and assumed to rise linearly from its 1980 level reached under the Type I pessimistic simulation. As in the case of urban water, and for the same reasons, all general network additions are assumed to be financed by the public sector. Unit investment cost to build or extend a general network per additional connection is set at US$650 for the entire period. Again, this amount is higher than the numbers commonly used by BNH is project staff in their projections, which range up to US$625. 178. Simulation results show average annual public sector outlays for this service growing from a somewhat lower level than urban water in the 1971-80 period to over twice the urban water amount in the last decade con- sidered (Table 40). Unlike the urban water case, the derived average annual - 105 - outlay for the 1971-80 period is substantially greater than total annual investments by BNH and other public sector sources for this purpose reported for the years 1974 through 1976 (Table 31). When related to projeuetd GDP, in the same way as with other services above, public sector outlays from urban sewerage simulations rise from less than 0.2% of GDP in the 1971-80 period to almost 0.3% in the two remaining decades in the century (Table 43). Table 40: URBAN SEWERAGE TYPE II SIMULATIONS; AVERAGE ANNUAL PUBLIC SECTOR OUTLAYS BY DECADE; 1970-2000 1970 1980 1990 2000 A. Stocks (in year) Dwellings (in thous.) 10,276 16,518 22,763 29,813 Coverage target 0.22 0.35 0.68 1.00 Dwgs. w/sewerage (in thous) 2,261 5,781 15,479 29,813 1971-81 1981-90 1991-00 B. Flows (in decade) Av. annual additions (in thous.) 352 970 1,433 Av. annual public sector outlays (in mills. US$'s) a/ 229 630 932 a/ In constant 1976 prices; at US$650 per additional connection. 6. Rural Water and Sewerage Coverage ratios for both water and sewerage in rural areas in 1970 were low in comparison to most other services under consideration (6% and 26% respectively). To date, public sector involvement in the provision of thesA services has been negligible. Yet, Type I optimistic simulations suggest that coverage ratios by the year 2000 may conceivably approach 50% for water and 80% for sewerage in rural areas. If current policy initiatives to increasingly involve the public sector in rural sanitation were to grow into forceful and effective programs, one could expect considerable improvement over these ratios. - 106 - 180. The difficulty and expense of providing these two particular services increase as the agglomerations of dwellings to be served decrease below a certain point in either size or density. This feature is dealt with in Type II simulations by assuming that coverage will proceed so that very small or sparsely built rural agglomerations and isolated dwellings are left for the final decade. Then it is assumed that unit costs entailed in covering the relatively isolated dwellings are drastically higher than for the rest. The simulations explore the financial implications of attaining 100% coverage in these two services by the year 2000 even though it appears unlikely that such targets will be actually met. 1/ As in the case of rural adequate dwellings, the number of rural water and sanitation connections which the public sector would have to finance if the Type II targets were to be met are derived by difference from the numbers that would be attained under the Type I pessimistic coverage -- the latter being attained, by assumption, without public sector involvement. 181. BNH and Bank estimates indicate that potable water can be provided to rural communities containing between 100 and 400 dwellings at a cost ranging from $430 to $480 per dwelling, and a latrine at a cost of around $70 each. 2/ To err on high side, however, the simulation assumes unit costs of $500 per dwelling for water and $100 for a latrine for all dwellings in rural communi- ties of around 100 households and over. 182. Not all rural dwellings, however, are clustered in agglomerations of 100 or more units, nor is it reasonable to assume that this will ever be the case. The cost of providing potable piped water and a latrine to isolated dwellings, or even small clusters of fewer than say, 25 units, is likely to have a very large range of variation depending on the location of the dwellings, and the average unit cost per connection is likely to be substantially greater than US$500. On the other hand, the proportion of such isolated dwellings to all rural dwellings is very unlikely to be greater than, say 10%. Such higher costs are taken into account by assuming that the services will be brought to the isolated 10% of all rural households during the 1991-2000 period, and financed by the public sector, at a unit cost of $2000 per water connection and $400 per sanitary installation. This would have the effect of increasing (weighted) average unit cost for both isolated and non-isolated dwellings during this last period from $500 to $741 per water connection and from $100 to $201 per latrine. 1/ The rationale for making such an implausible assumption is simply to explore whether even if coverage were to reach such limits, money would be a problem. 2/ For example, the Paraguasu Rural Development project in the state of Bahia, appraised by the Bank in 1978, has a component for these services in communities of this size class. Since this project area is characterized by an unusually deep water table, the unit cost figure for water used in the project is at the upper end of the range ($480). - 107 - 183. Simulation results for rural water indicate that the required public sector outlays would be small compared to other services considered thus far (Table 41). There would remain below 0.1% of GDP for ech of the two decades from 1981 to 2000 (Table 43). Corresponding costs for rural sanitation remain below 0.03% of GDP (Table 42 and 43). Table 41: RURAL WATER, TYPE II SIMULATIONS AVERAGE ANNUAL PUBLIC SECTOR OUTLAYS BY DECADE, 1970-2000 1970 1980 1990 2000 A. Stocks (in year) Households (in thous.) 7,703 8,488 9,751 11,343 Coverage ratio 0.06 0.18 0.59 1.00 Dwellings with water (in thous.) 462 1,528 5,739 11,343 1971-80 1981-90 1991-00 B. Flows (in decade) Av. annual net additions (in thous.) a/ 352 457 Av. annual public sector outlays (in mills US$'s) b/ 176 339 a/ Additions under Type II coverage target in excess of those that would be forthcoming under Type I pessimistic coverage ratio. b/ In constant 1976 dollars; at US$500 and US$741 during 1981-90 and 1991-2000 respectively per public sector financed net addition. - 108 - Table 42: RURAL SANITATION, TYPE II SIMULATIONS; AVERAGE ANNUAL PUBLIC SECTOR OUTLAYS BY DECADE, 1970-2000 1970 1980 1990 2000 A. Stocks (in year) Households (in thous.) 7,703 8,488 9,751 11,343 Coverage ratio 0.26 0.40 0.70 1.00 Dwellings with sanitation (in thous.) 2,003 3,995 6,826 11,343 1971-80 1981-90 1991-00 B. Flows (in decade) Av. annual net additions (in thous.) a/ - 253 337 Av. annual public sector outlays (in mills US$'s) b/ 25 67 a/ Additions under Type II coverage target in excess of those that would be forthcoming under Type I pessimistic coverage ratio. b/ In constant 1976 dollars; at US$100 and US$201 during 1981-90 and 1991-2000 respectively per public sector financed net addition. - 109 - 7. Summary 184. The conclusion emerging from Type II simulations is that the financial outlays by the public sector required to substantially accelerate the expansion of basic housing services in Brazil during the next two decades and approach full coverage by century's end are not likely to present a serious problem. Based on assumptions that systematically sought to produce upwardly biased projections of such public sector outlays, the simulations suggest that less than 1-1/4% of GDP would be required during each of the two decades between 1980 and 2000 (Table 43). Moreover, there is no a priori reason to believe that this share of GDP would have to be substantially exceeded in any particu- lar year during the entire simulation period. By comparison, aggregate BNH applications of funds in the year 1976 equaled 1.7% of GDP. Table 43: TYPE II SIMULATIONS: SUMMARY OF PUBLIC SECTOR OUTLAYS AS PERCENT OF GDP BY DECADES, 1971-2000 Services 1971-80 1981-90 1991-00 Dwellings Urban 0.51 0.57 0.52 Rural 0.00 0.13 0.11 Sub-Total, Dwellings 0.51 0.70 0.63 Water Urban 0.28 0.18 0.12 Rural 0.00 0.08 0.10 Sub-Total, Water 0.28 0.26 0.22 Sewerage Urban 0.17 0.28 0.28 Rural 0.00 0.01 0.02 Sub-Total, Sewerage 0.17 0.29 0.30 All Services Urban 0.96 1.03 0.92 Rural 0.00 0.22 0.23 Grand-Total, All 0.96 1.23 1.15 Note; Figures are ratios of accumulated public sector outlays for each service over 10-year period, to GDP accumulated annually over same 10-year period using 4% growth rate beginning in 1980. 185. This hardly means that the coverage targets set for Type II simula- tions are likely to be met or exceeded but simply that "money does not seem to be the problem" in the future. The main factors constraining progress in this - 110 - area are likely to be institutional and those relating to the allocation of scarce non-financial resources, particularly, top and middle level managerial talent. Analysis of the policy, program and administrative measures that could be taken to deal with such constraints lies beyond the scope of the simulation exercise. 186. Another conclusion arising from the Type II simulation is that of the three housing sub-sectors included in the exercises (dwellings, water and sewerage), dwellings would take the lion's share in attempting to approach full coverage in the three sub-sectors by century's end. This share would probably exceed 50% of all public sector outlays for these housing related services over the remaining period to the year 2000. In turn, urban dwellings would probably absorb over 80% of the total dwellings' share. - ill - APPENDIX A Housing Services and Per Capita Income: Historical Regularities 1. Historical trends in housing service indicators reviewed in the first section of this Annex suggest certain patterns of association between the relative availability of such services and per capita income. This appendix explores the extent to which the two sets of variables have been statistically related in the past. This may provide some basis on which to extrapolate likely future trajectories of these indicators in response to alternative assumptions concerning per capita income growth in the coming decades under the Continuismo scenarios. 2. Clearly, the availability and spatial distribution of housing related services in Brazil has been and will continue to be governed by a multiplicity of factors. Public sector investment decisions, technological options and the household distribution of income within regions are surely very important among them -- and their role is explicitly taken into account elsewhere in this Annex. At this point, however, it will be assumed for simplicity that: (a) Public sector investment in the provision of housing related services in any given state is positively related to the per capita income of the state (via tax revenues collected from each state); and (b) That technologies for the production of these services become simultaneously available to all states or that their dissemi- nation reaches the higher per capita income states in sequence to their per capita income ordering. 3. The very wide divergence between urban and rural areas in the value of the housing related indicators in specific years noted above suggests taking into account the urban-rural distinction in the estimate of relation- ships between the housing indicators and per capita income. Similarly, these historical trends revealed a tendency for spatial (regional and locational) differentials to narrow over time. Category rankings by indicator scores, both at the beginning and end of the periods considered, generally coincides with their ordering by per capita income. 1/ 4. These observations support common sense in suggesting a non-linear relationship between indicator value and per capita income, in which the value of any such indicator increases at a decreasing rate as per capita incomes increases. This condition is accommodated by the semi-logarithmic form: 1/ Per capita income ranks across categories for the years 1949, 1959 and 1970 can be inferred from Tables II.5, page 26 and II.9, page 31 in Constantino Lluch "Growth and Distribution in Brazil: Some Long-Run Trends", IBRD/DRC, January 4, 1978. - 112 - z = a + b ln y; o C z 6 1, where the dependent variable z is the proportion of dwellings in a particular spatial category having some specified housing service; y is per capita income and a and b are parameters. It is simple to work with and parameters are easy to interpret. It is the basic specification used inthe regressions reported below. 5. Table Al presents the results of cross section regression estimates for the year 1970 in which state level subdivisions of Brazil are taken as units of observation. 1/ The Federal District (Brasilia and surrounding area) has been excluded from all regressions and Guanabara has been excluded from regressions pertaining to rural areas. In all of them, the dependent variable is the proportion of dwellings inthe specified location having the indicated service, and the independent variable is the logarithm of statewide per capita income in 1970. Therefore, the regression coefficients can be interpreted as the percentage point difference inthe value of the corresponding indicator associated with a 1% difference in state per capita income. 2/ All regression coefficients are statistically greater than zero at the 1% confidence level, and with the exception of the regression of the proportion or rural dwellings with any sanitation device (equation 5), the specified relations accounted for 78% or more of the interstate variations in the level of the selected indi- cators. The parameter value associated with equation 1 ("urban water") and I/ These are: 1. Amazonas (including Rondonia, Acre and Roraima) 2. Para (including Amapa) 3. Maranhao 4. Piaui 5. Ceara 6. Rio Grande do Norte 7. Paraiba 8. Pernambuco (including Fernando de Noronha) 9. Alagoas 10. Sergipe 11. Bahia 12. Minas Gerais 13. Espirito Santo 14. Rio de Janeiro 15. Guanabara 16. Sao Paulo 17. Parana 18. Santa Catarina 19. Rio Grande do Sul 20. Mato Grosso 21. Goias 2/ For example, b=0.26 in equation 1 indicates that, on the average, if state i's per capita income was 10% greater than state j's in 1970, one could expect that the percentage of urban dwellings connected to a general water network or having piped water would be approximately 2.6 percentage points higher in state i than in state j. Table A.1: CROSS SECTION ESTIMATES OF THE RELATIONSHIP BETWEEN INCOME PER CAPITA AND HOUSING SERVICES INDICATORS AT THE STATE LEVEL, 1970a/ Equation Dependent Variable A-Intercept B-Slope!b/ F-Ratio No. (Percentage of dwellings in specified location with indicated service) 1 Water General Network or Pipe d Urban -2.5337 0.2617 135.5 0.88 2 General Network or Well, Rural -3.8926 0.3629 62.5 0.78 CZ 3a General Network or Piped-form Well, Urban and Rural -3.1223 0.2968 136.2 0.88 3b General Network or for Rural Dwellings, w from a Well -3.2980 0.3166 190.2 0.90 Sanitation 4 General Sewerage Network or Sceptic Tank, Urban -2.4943 0.2434 36.4 0.85 5 Any Sanitation Device, Rural -0.7576 0.0717 21.3 0.54 6 General Sewerage Network, Sceptic Tank or, for Rural only, Any Device, Urban and Rural -2.4422 0.2484 89.8 0.82 a/ Number of observations (N) = 20 (state-level units) for equations 2 and 5, where Guanabara is excluded; in all others, N=21. b/ Independent variable is the logarithm of statewide per capita income in 1970, for all equations. c/ "Piped" Corresponds to census category "Canalizacao interna" Sources: Dependent variable: IBGE, Censo Demographico, 1970 (State Volumes); per capita income: FGV, "Contas Nacionais", Conjuntura Economica, July, 1977. - 114 - 4 ("urban sanitation") turned out to be quite close: b=0.26 and 0.24 respectively and a=-2.53 and -2.49 respectively. At the same time, levels for corresponding indicators derived from the 1970 census show that the proportion of urban dwellings with general network water, as well as the proportion with piped water is considerably higher than the proportion of urban dwellings connected to a sewerage network or with a septic tank (Table 3 and 4 in this annex). Thus, the pattern of dissemination of both types of housing service in urban areas associated with variation in per capita income seems to be quite similar, while the extent of dissemination is clearly led by the water related services. This finding is consistent with the post-1970 PLANASA experience of lagging accomplishments in the implemen- tation of its sewerage relative to its drinking water objectives for urban areas established in 1971 (see Section D.2 of this annex). 6. Judged in terms of both F-ratios and R2 values, the regression results reported in Table A.1 indicate strong statistical association between the selected housing indicators and state per capita income in 1970, partic- ularly for relations including urban areas -- either alone or in aggregation with rural areas. To explore the extent to which those results might be useful in projecting indicator values at the regional level, such values were derived for urban water and for urban sanitation from regression equations 1 and 4, respectively, for the year 1970. These were then compared with corresponding value derived directly from the 1970 census tabulations. The results of this exercise are reported in Table A.2. 7. The comparison between the predicted and observed value for the two selected indicators show small difference throughout all regions. The maximum proportional difference between a 12% prediction overestimate in the case of safe urban water in the Northeast. For both indicators, the predicted values for the Northeast and Frontier exceed the observed and conversely for the Southeast. Thus, particularly in the case of urban water in the Northeast, the results indicate that the region was not only lagging the other regions in terms of this indicator, but that in 1970 it was also lagging in terms of the level that could be expected given its per capita income on the basis of the overall statistical norm. - 115 - Table A.2: OBSERVED AND PREDICTED VALUES OF CERTAIN HOUSING INDICATORS BY REGION 1970 Condition Value of indicator Equation Indep. var._ "Predicted" ObservedS! (3)/(4) No.a (%) (M) A. Water, Urban: (1) (2) (3) (4) General Network of piped All regions 1 12.007 0.61 0.61 1.00 Northeast 1 11.058 0.36 0.32 1.12 Southeast 1 12.321 0.69 0.71 0.97 Frontier 1 11.417 0.45 0.42 1.07 B. Sanitation, Urban: General Sewerage or Sceptic Tank All regions 4 12.007 0.43 0.44 0.98 Northeast 4 11.058 0.20 D.19 1.05 Southeast 4 12.321 0.50 0.53 0.94 Frontier 4 11.417 0.28 0.26 1.08 a/ From Table A.1 above. b/ Value substituted for ln y in the corresponding regression equation. For each region, this value is the natural logarithm of the region's per capita income in 1970 (from Table c/ For indicator A, Tables A.3 and A.4; for indicator B, Tables A.4 and B.4. - 116 - Table A.3: POPULATION, INCOME AND PER CAPITA INCOME BY REGION, 1960 AND 19701/ Population Income Per Capita Income (Units) (Persons X 103) (Current Cr$ x 106) (Current Cr$) (1) (2) (3) 1970 All regions 92,601.4 151,724 1,638.46 Northeast 28,112.4 17,838.4 634.54 Southeast 56,350.0 125,487 2,244.67 Frontier 8,139.0 7,398 908.96 1/ Cash income only. Note: Distrito Federal Excluded Sources: Col (1): Demographic Censuses, 1960, 1970. Col (2): Conluntura Economica 31:7 (July 1977) pp. 106-111. Col (3): (Col (2)/Col (1) X 10 - 117 - Table A.4: PERMANENT DWELLINGS AND DWELLINGS WITH CERTAIN HOUSING SERVICES, BY REGION AND LOCATION, 1970 (In thousands dwellings units) All regions2J Northeast Southeast Frontier 1. Permanent dwellings 17,529.2 5,140.6 10,986.9 1,401.7 2. In urban areas 10,181.0 2,162.1 7,408.7 610.2 3. In rural areas 7,348.2 2,978.5 3,578.2 791.5 4. Urban dwellings with general network or piped water 6,232.2 692.5 5,280.9 258.8 5. (4) / (2) 0.61 0.32 0.71 0.42 a/ Federal District excluded; thus some cells differ from Appendix B, Table B.3. Source: IBGE, Demographic Census, 1970, State Volumes. - 118 - APPENDIX B Statistical Appendix - 119 - Table B.1 ALL NON INSTITUTIONAL DWELLINGS AND THOSE CLASSIFIED AS "DURABLE"2! BY REGION AND URBAN-RURAL LOCATION, 1972 (Thousands of Dwellings) All locations Urban Rural Regions All Durable All Durable All Durable Northeast 5721.7 4213.1 2378.3 2240.4 3143.4 1972.7 Southeast 12376.6 11680.0 8801.7 8439.0 3574.9 3241.0 Federal District N.A. N.A. 130.0 97.6 N.A. N.A. a/ For the definition of "durable dwellings," see text. Source: F. IBGE, Indicadores Sociais (Rio de Janeiro: 1977). Table 4, p. 165 - 120 - Table B.2 DWELLINGS WITH WATER SUPPLY SERVICES BY REGION AND URBAN-RURAL LOCATION, SELECTED YEARS, 1950 - 1973 (Thousands of Dwellings) Service, Region and Location 1950 1960 1970 1973 A. General Network All Regions n.a. 2,845.6 5,784.2 7,808.4-/ Urban n.a. n.a. 5,592.5 7,643.6 Rural n.a. n.a. 191.7 164.8a/ Northeast n.a. 222.2 636.1 1,024.3 Urban n.a. n.a. 619.8 1,003.4 Rural n.a. n.a. 16.3 20.9 Southeast n.a. 2,550.1 4,853.3 6,337.2 Urban n.a. n.a. 4,691.0 6,193.9 Rural n.a. n.a. 162.3 143.3 Frontier n.a. 73.4 294.8 n.a. Urban n.a. n.a. 281.7 446.3 Rural n.a. n.a. 13.1 n.a. B. Piped Water All Regions 1,563.3 n.a. 5,863.0 7,968.7-a/ Urban 1,474.7 n.a. 5,409.4 7,311.8 Rural 88.6 n.a. 453.6 656.9-2 Northeast 135.3 n.a. 571.4 857.5 Urban 129.2 n.a. 546.7 822.8 Rural 6.1 n.a. 24.7 34.7 Southeast 1,384.9 n.a. 5,025.1 6,763.4 Urban 1,304.5 n.a. 4,622.8 i Rural 80.4 n.a. 402.3 621.5 Frontier 43.1 n.a. 266.5 n.a. Urban 41.0 n.a. 239.9 347.1 Rural 2.1 n.a. 26.6 n.a. a! Excludes rural Frontier. Sources: For 1950, 1960, 1970: Demographic Censuses; For 1973: F.IBGE, Indicadores Sociais (Rio de Janeiro: 1977) Table 5, P. 167. - 121 - Table B. 3 PERMANENT DWELLINGS BY REGION AND URBAN-RURAL LOCATION, SELECTED YEARS: 1950 - 1973 (Thousands of Dwellings) Region and Location 1950 1960 1970 1973 All Regions 10,046.2 13,497.8 17,628.7 19,401.7!/ Urban 3,730.4 6,350.1 l1l,276.3 12,574.8a/ Rural 6,315.8 7,147.7 7,352.4 6,826.9 Northeast 3,569.7 4,233.1 5,140.6 5,855.3 Urban 991.3 1,481.7 2,162.1 2,635.8 Rural 2,578.4 2,751.4 2,978.5 3,219.5 Southeast 5,800.7 8,281.6 10,986.9 12,601.5 Urban 2,555.3 4,520.0 7,408.7 8,999.3 Rural 3,245.4 3,761.5 3,578.2 3,602.2 Frontier 675.8 983.2 1,500.9 n.a. Urban 183.8 348.4 705.3 939.7 Rural 492.0 634.8 795.6 n.a. a/ Excludes rural Frontier Source: For 1950, 1960, 1970: Demographic Censuses; For 1973: F.IBGE, Indicadores Sociais (Rio de Janeiro: 1977). Table 5, p. 167 - 122 - Table B.4 DWELLINGS WITH SANITATION SERVICES BY REGION AND URBAN-RURAL LOCATION, SELECTED YEARS, 1950 - 1973 (Thousands of Dwellings) Service, Region and Location 1950 1960 1970 1973 A. Any Sanitary Device All Regions 3,317.6 6,731.6 10,674.6 13,098.7-/ Urban 2,659.0 n.a. 8,793.0 10,981.1 Rural 657.8 n.a. 1,881.6 2,117.6a Northeast 4,505.0 1,020.3 1,493.9 1,830.7 Urban 389.7 n.a. 1,313.5 1,634.3 Rural 60.8 n.a. 180.4 196.4 Southeast 2,703.0 5,282.7 8,355.6 10,412.0 Urban 2,148.2 n.a. 6,864.3 8,494.8 Rural 554.8 n.a. 1,491.3 1,917.2 Frontier 164.1 428.5 825.1 n.a. Urban 121.9 n.a. 615.2 852.0 Rural 42.2 n.a. 209.9 n.a. B. Sewerage Network or Sceptic Tank_ All Regions n.a. 3,212.3 4,684.3 7,587.4' Urban n.a. n.a. 4,539.2 7,246.6 Rural n.a. n.a. 145.1 341.2-' Northeast n.a. 257.1 408.8 891.7 Urban n.a. n.a. 399.0 836.4 Rural n.a. n.a. 9.8 55.3 Southeast n.a. 2,868.0 4,086.8 6,364.0 Urban n.a. n.a. 3,958.9 6,078.2 Rural n.a. n.a. 127.9 285.8 Frontier n.a. 87.2 188.7 n.a. Urban n.a. n.a. 181.3 | 332.0 Rural n.a. n.a. 7.4 n.a. a/ Excludes rural Frontier. Sources: For 1950, 1960 and 1970: Demonraphir Censuses; For 1973: F.IBGE, Indicadores Sociais (Rio de Janeiro: 1977) Table 5, p. 167. - 123 - Table B.5 HOUSING SERVICES UNIT COST ASSUMPTIONS FOR SIMULATIONS (in 1976 US$) Urban Rural Roads and drainage 600 a/ Site and services b/ 1,100 Construction materials c/ 2,100 2,100 Sub-total 3,800 2,100 Water d/ 500 Sewerage d/ 650 Water and excreta disposal - for agglomerated dwellings e/ 500 - for isolated dwellings f/ 2,000 a/ Not applicable b/ Include: land purchase and terracing for individual lot (of approximately 100 m2) and on lot connections to water and sewerage networks c/ To build a 25 m2 "enibryo" structure d/ Unit investment cost to built or extend general network e/ Sewerage facilities consist of latrines f/ Water is pumped from artesian well or river; sewerage facilities consist of latrine. - 124 - Table B.6 TYPE II SIMULATIONS; SUMMARY OF AVERAGE ANNUAL PUBLIC SECTOR OUTLAYS BY SERVICE AND DECADE (in millions of US$) a/ Services 1971-80 1981-90 1991-00 Dwellings: Urban 703 1,255 1,695 Rural 0 286 357 Sub-Total, Dwellings 703 1,541 2,052 Water: Urban 383 420 409 Rural 0 352 457 Sub-Total, Water 383 772 866 Sewerage: Urban 229 630 932 Rural 0 25 67 Sub-Total, Sewerage 229 655 999 All Services Urban 1,315 2,305 3,036 Rural 0 663 881 Grand-Total, All 1,315 2,968 3,917 a/ Constant 1976 dollars.