Report No. 35749-BR Brazil Inputs for a Strategy for Cities A Contribution with a Focus on Cities and Municipalities (In Two Volumes) Volume II: Background Papers November 10, 2006 Brazil Country Management Unit Finance, Private Sector and Infrastructure Management Unit Latin America and the Caribbean Region Document of the World Bank Inputs to a Strategy for Brazilian Cities Page i This volume comprises ten background papers prepared as contribution to the main report. The authors of the papers are the following: Chapter Title Author Chapter 1 Urbanization, Growth, and Welfare in Brazil Somik Lall Chapter 2 City Performance and Policy Actions Somik Lall Chapter 3 Urban Policies and Slum Formation Somik Lall Chapter 4 The Evolution of Brazilian Municipal Finances, Fernando Blanco 2000-2004 Second Draft/January 2006 Chapter 5 Municipal Credit Markets, Issues and options Benjamin Darche Chapter 6 Efficiency of Brazilian Municipalities Suhas Paradekar Chapter 7 Main Aspects of the Regulatory Framework Edesio Fernandez Governing Urban Land Development Processes in Brazil Chapter 8 Land Markets in Brazil: Capturing Land Value Fernanda Furtado and Pedro to Finance Infrastructure Improvement Jorgensen Chapter 9 Urban Land Use Regulation in Brazilian cities ­ Paulo Avila Impact on Urban Land Markets and access of Low Income People to Land and Housing Chapter 10 Brazil's Urban Land and Housing Markets David Dowall How well are they working? Inputs to a Strategy for Brazilian Cities Page ii ABBREVIATIONS ABS Asset Backed Securities ADR Age Grade Distortion Rate BACEN Banco Central (Central Bank) BNDES National Development Bank CDO Collateralized Debt Obligation CEF Federal Loan and Savings Bank (Caixa Econômica Federal) CMN National Monetary Council CVM Comisión de Valores Inmobiliarios (Stock Exchange) DEA Data Envelopment Analysis FAT Fundo de Amparo ao Trabalhador (Workers' Support Fund) FDH Free Disposal Hull FDIC Fondos de Investimentos dos Direitos Creditos FGTS Workers Severance Fund (Fundo de Garantia do Tempo de Serviço) FJP Joao Pinheiro Foundation FPE Fundo de Participacao Estadual FPM Fundo de participacao municipal FRL Fiscal Responsibility Law GDP Gross Domestic Product HDI Human Development Index IBGE Brazilian Institute of Statistics and Geography (Instituto Brasileiro de Geografia e Estatística) ICM Imposto sobre valor de mercadorias (Value added tax) IMR Infant Mortality Rate IPEA Institute for Applied Economic Research (Instituto de Pesquiza Econômica Aplicada) IPTU Urban Property Tax (Imposto Predial Territorial Urbano) MDF Municipal Development Fund MIC Middle Income Countries MP Market Potential NCR Net Current Revenues NGO Non-Governmental Organization OECD Organization for Economic Cooperation and Development OODC Otorga Onorosa do Direito de Construir (Sale of Building Rights) PFI Private Finance Initiative PNAD Pesquisa Nacional aos Domicilios (National Household Survey) PNUD United Nations Development Program (Programa das Nações Unidas para o Desenvolvimento) RM Região Metropolitana (Metropolitan Region) SPE Special Purpose Entity STN Secretaria do Tesouro Nacional (Treasury Secretariat) TC Tribunal de Contas ZEIS Zone of Special Social Interest (Zona de Especial Interesse Social) Inputs to a Strategy for Brazilian Cities Page iii Table of Contents 1. Urbanization, Growth, and Welfare in Brazil ...............................................................................................1 Chapter 1. Urban Growth and Competitiveness................................................................................................2 Urban Growth Patterns.......................................................................................................................................2 Patterns of Income Growth.................................................................................................................................7 Specialization Across Cities.............................................................................................................................10 Industrial Decentralization ...............................................................................................................................14 Summary of Findings .......................................................................................................................................16 2. City Performance and Policy Actions...........................................................................................................19 Background ......................................................................................................................................................19 Measuring City Growth....................................................................................................................................19 Model and Estimatio Estrategy ........................................................................................................................21 Demand Side ....................................................................................................................................................21 Population Supply ............................................................................................................................................22 Determinants of Growth...................................................................................................................................22 Policies Favoring Secondary Cities..................................................................................................................29 Summary of Findings .......................................................................................................................................30 3. Urban Policies and Slum Formation.............................................................................................................33 Slum Formation Across Cities..........................................................................................................................34 Housing Supply and Slum Formation ..............................................................................................................37 Heterogeneous Housing Supply Elasticities.....................................................................................................38 Findings from Empirical Analysis....................................................................................................................40 Summary ..........................................................................................................................................................45 References ........................................................................................................................................................47 4. The Evolution of Brazilian Municipal Finances, 2000-2004.......................................................................50 Introduction ......................................................................................................................................................50 The Evolution of Municipal Revenues.............................................................................................................57 The Evolution of Municipal Expenditures .......................................................................................................64 Municipal Expenditures by Economic Category..............................................................................................64 Municipal Expenditures by Function ...............................................................................................................71 Conclusions and Policy Implications ...............................................................................................................79 5. Municipal Credit Markets.............................................................................................................................89 Introduction ......................................................................................................................................................89 Current Status of the Municipal Credit Market: Developments in the Demand and Supply of Municipal Credit................................................................................................................................................................90 Monitoring Sub-National Compliance with the Fiscal Responsibility Law.....................................................91 Municipal Capital Revenues and Expenditures................................................................................................93 New Borrowing Instruments and Lending Institutions to Assist Subnational Governments to Raise Capital.94 Public Private Partnerships and the Municipal Credit Market .........................................................................95 Development of Municipal Credit Markets in Brazil.......................................................................................96 Current Capital Market Conditions ..................................................................................................................96 Municipal Credit Instruments...........................................................................................................................98 The Use of Asset Backed Securities (ABS) to Further Develop the Municipal Credit Market .....................100 Monitoring the Municipal Credit Market.......................................................................................................101 6. Efficiency of Brazilian Municipalities.........................................................................................................105 Abstract ..........................................................................................................................................................105 Introduction ....................................................................................................................................................105 Why Discuss Municipal Efficiency?..............................................................................................................105 Problems in the Measurement of Efficiency ..................................................................................................106 Empirical Frontier Estimation of Efficiency ..................................................................................................107 Inputs to a Strategy for Brazilian Cities Page iv Organization of this Paper..............................................................................................................................108 Literature on Frontier Estimation of Municipal Efficiency............................................................................108 General Municipal Efficiency ........................................................................................................................109 Municipal Efficiency for Specific Services....................................................................................................110 Municipal Efficiency in Brazil .......................................................................................................................110 Human Development in Brazilian Municipalities..........................................................................................111 Human Development Index: IDH-M and Current Spending per Capita ........................................................112 Graphical Presentation of the Data.................................................................................................................113 Results from FDH Analysis of Efficiency......................................................................................................117 Comparing North and South on IDH-M 2000................................................................................................117 Efficiency for IMR 2000 with Physical Input Variables................................................................................120 Comparison between 1991 and 2000..............................................................................................................120 FDH after Correction for Contextual Variables: Understanding Drivers of Efficiency................................122 Outcome Variable of Interest .........................................................................................................................122 Evoluation of Age-Grade Distortion Rate (ADR) between 1991 and 2000...................................................123 The Effect of Contextual Variables...............................................................................................................125 Efficiency Analysis with and without Contextual Variables..........................................................................128 Municipal Efficiency: Possible Effects of Consortia and Municipal Councils .............................................130 References ......................................................................................................................................................135 7. Main Aspects of The Regulatory Framework Governing Urban Land Development Processes..........137 1 Executive Summary.....................................................................................................................................137 2 General Remarks: the Broad Legal Context of Urban Land Policies..........................................................138 3 Urban Policy in the 1988 Federal Constitution ...........................................................................................140 4 The 2001 Urban Policy Law: A new statute for Brazilian Cities................................................................144 5 Prospects for Progressive Urban Planning and Participatory Urban Management .....................................146 6 The main Institutional Actors......................................................................................................................150 7 Main Legal Aspects of Land Titling and Regularization Programmes .......................................................151 8 Main Legal Aspects of Land Registration...................................................................................................155 9 The Ongoing Process of Land Law Review: Proposed Changes to Federal Law No. 6,766/1979 ............158 10 Final Remarks............................................................................................................................................163 11 Recommendations for Policy Makers........................................................................................................166 12 References .................................................................................................................................................167 8. Land Markets in Brazil: Capturing Land Value to Finance Infrastructure Improvements................168 Urban Land Value Capture.............................................................................................................................170 Conceptualization...........................................................................................................................................170 Theoretical And Historical Aspects Applied To Brazil And Latin America..................................................172 Experience In Latin America And Possibilities In Brazil ..............................................................................175 Basic Chronology...........................................................................................................................................175 Experiences ....................................................................................................................................................177 Using Land Value Capture Instruments For Financing Urban Infrastructure In Brazil .................................183 The Present Basic Infrastructure Deficit in Brazil and some Basic Ways for Dealing with it.......................183 Analysis of the Modalities of Land Value Capture in Latin America............................................................186 Questions Related to the Urban Land Market In Brazil .................................................................................191 Basic Features.................................................................................................................................................191 Problems of Self-Sustainability In Basic Infrastructure Provision................................................................194 Regularization of Informal Settlements, Production of Urbanized Land and Costs Recovery ......................196 Brazil's Recent Experience ............................................................................................................................196 Value Capture For Financing Infrastructure In Informal Settlements: Two Basic Questions .....................204 Final Considerations and Recommendations .................................................................................................209 Bibliography...................................................................................................................................................214 9. Urban Land Use Regulations in Brazilian Cities.......................................................................................216 Inputs to a Strategy for Brazilian Cities Page v Summary ........................................................................................................................................................216 Reasons for Controlling Urban Land Markets ...............................................................................................220 The Classical Urban Economics Approach and the Form of the City............................................................222 Brazilian Cities Regulation Framework an Overview....................................................................................225 Some Stylized Facts about Brazilian Cities....................................................................................................230 Urban Development Controls and Effects on Land Prices.............................................................................237 Can Land Controls Loosening Improve Formal Urbanized Land Production?..............................................245 Conclusions and General Implications...........................................................................................................253 References ......................................................................................................................................................255 10. Brazil's Urban Land and Housing Markets...........................................................................................257 Introduction ....................................................................................................................................................257 Characteristics of Well-Fnctioning Urban Land and Housing Markets .........................................................258 Is there a Brazilian Paradox?..........................................................................................................................259 Caveats about the data Used in this Paper......................................................................................................260 Performance of Brazil's Urban Land and Housing Markets during the last half of the Twentieth Century..261 How Large is Brazil's Informal Housing Sector? ..........................................................................................268 The Urban Land use Consequences of Urbanization .....................................................................................274 Looking Forward: Brazil's Future Urban Housing Needs and Prospects for Reaching them?......................280 What can be done to improve Urban Land and Housing Market Outcomes? ................................................282 References ......................................................................................................................................................283 List of Boxes Box 4.1 São Paulo Indebtedness......................................................................................................................57 Box 4.2: The Effects of Iintergovernmental Transfers on Recipient's Tax Effort and Expenditure...............62 Box 4.3 Social Security Imbalance in Municipal Finances.............................................................................66 Box 4.4 Budget Rigidity..................................................................................................................................71 Box 5.1 Government of Brazil Real Denominated Global Bond...................................................................101 Box 5.2 Mexico's Municipal Credit Market ..................................................................................................102 List of Charts Chart 9.1 Common Instruments of Urban Policy used in Brazilian Cities....................................................227 Chart 9.2 Main Planning and Land use Regulation Standards in used in Brazilian Cities............................228 List of Figures Figure 1.1: Urban and Rural Population Dynamics (population in thousands)..................................................1 Figure 1.2: Metro Areas, 2000 ...........................................................................................................................1 Figure 1.3: Urban Agglomerations by Population Size......................................................................................3 Figure 1.4: Population Growth in Urban Agglomerations by Region................................................................4 Figure 1.5: Individual City Size Growth between 1970 and 20001)..................................................................5 Figure 1.6: Changes in Population Rank Order between 1970-2000 .................................................................7 Figure 1.7: Relative Income Level and City Population in 1970 and 2000........................................................8 Figure 1.8: Annual Income Growth and Initial Income Level for 1970-2000 .................................................10 Figure 1.9: Annual Income Growth and Initial Income Level for 1991-2000 .................................................10 Figure 1.10: Index of Specialization by Agglomeration Size Group ...............................................................13 Figure 1.11: City Specialization in 2000..........................................................................................................14 Figure 1.12: Sector Employment Shares by Agglomeration Size Distribution................................................15 Figure 3.1 Cities with the Fastest Slum Formation between 1980 and 2000 ...................................................35 Figure 3.2 Slum Dweller Growth and City Population Growth between 1980 and 2000 ................................36 Figure 3.3 Slum Dweller Growth and Formal Housing Stock Growth between 1980 and 2000 .....................37 Figure 3.4 Housing Supply Elasticity and Slum Formation.............................................................................39 Inputs to a Strategy for Brazilian Cities Page vi Figure 4.1 Municipal Tax Revenue Collection by Municipal Size, 2005 ........................................................60 Figure 4.2 Municipal Current Revenues per Capita by Municipal Size (Reais of 2004).................................61 Figure 4.3 Municipal Capital Revenues per Papita by unicipal Size, 2003 (Reais of 2004)...........................64 Figure 4.4: Current Expenditure Per Capita by Municipal Size, 2003 (Reais of 2004) ..................................68 Figure 4.5 Capital Expenditures Per Capita by Municipal Size, 2003 (R$ of 2004).......................................69 Figure 4.6 Municipal Expenditure Composition, 2000-2004...........................................................................70 Figure 4.7 Municipal Deb Composition (% average 2000-04.........................................................................76 Figure 6.1 Basic IDH-M 2000 Graph for all municipalities...........................................................................113 Figure 6.2 Graph Showing Small and Very Small (below 20,000 Population) Municipalities in Green.......114 Figure 6.3 Graph Showing NE Municipalities in Blue, Others Red ..............................................................115 Figure 6.4 Small Alagoas Municipalities with Blue Cross, Small Rio Grande do Sul ..................................115 Figure 6.5 Alagoas Small Municipalties with a Green `S', Medium Municipalities with a Purple `M'........116 Figure 6.6 IMR 2000 Graph for all Municipalities ........................................................................................119 Figure 6.7 Basic IMR 2000 Graph for all Municipalities...............................................................................121 Figure 6.8 Evolution of Age-Grade Distortion (ADR) All Brazil: 1991 - 2000 ...........................................124 Figure 6.9 Evolution of Age-Grade Distortion (ADR) North-East: 1991 - 2000..........................................125 Figure 6.10 Evolution of Age-Grade Distortion (ADR) South: 1991 - 2000...............................................126 Figure 6.11 Basic IDH-M 2000 Graph for all Municipalities ........................................................................132 Figure 6.12 Basic IDH-M 2000 Graph for all Municipalities ........................................................................132 Figure 6.13 Basic IDH-M 2000 Graph for all Municipalities ........................................................................133 Figure 6.14 Basic IDH-M 2000 Graph for all Municipalities ........................................................................133 Figure 6.15 Basic IDH-M 2000 Graph for all Municipalities ........................................................................134 Figure 9.1 Simple supply and demand curves................................................................................................230 Figure 9.2 Households Density Gradients for 10 Brazilian cities in 2000 ....................................................236 Figure 9.3 Marginal effects over high land prices likelihood in 2003...........................................................244 Figure 10.1 Median Housing Prices and Median Household income, Middle income Countries, 1998........259 Figure 10.2 Percent Distribution of Urban and Rural Population ..................................................................261 Figure 10.3 Urban and Rural Population Trends in Brazil, 1950-2000..........................................................262 Figure 10.4 Private Investment in Housing is Robust and Increasing in Real Terms ....................................266 Figure 10.5 Defining Informal Housing is Complicated................................................................................270 Figure 10.6 Level of Informal Varies Widely Across Brazil, 2000 ...............................................................270 Figure 10.7 Number of Favela Dwelling Units in Rio de Janeiro, 1900-1991...............................................271 Figure 10.8 Low-Income Does Not Entirely Explain Informality .................................................................273 Figure 10.9 Trends in Public Sector Gross Fixed Capital Formation.............................................................273 Figure 10.10 Residential Land is Expensive Relative to GDP per Capita .....................................................275 Figure 10.11 Spatial Distribution of Population Change: ..............................................................................277 Figure 10.12 Spatial Distribution of Change in Urban Land Development:..................................................278 List of Tables Table 1.1: City Size Distribution........................................................................................................................3 Table 1.2: The 7 Fastest Growing Cities between 1970 and 2000* ...................................................................4 Table 1.3: Spatial Gini Coefficients in 1970 and 2000 ......................................................................................5 Table 1.4: The Transition Matrix .......................................................................................................................6 Table 1.5: Convergence of Per Capita Income1) .........................................................................................9 Table 1.6: The Concentration of Industry in 2000 ...........................................................................................12 Table 1.7: Employment Share by Industry in 1970 and 2000..........................................................................15 Table 1.8: Variable Definitions and Data Sources ...........................................................................................17 Table 1.9: Correlation Coefficients ­ Wages Versus Income ..........................................................................17 Table 2.1 Demand Side: Determinants of Income Per Worker........................................................................23 Table 2.2 Population Supply ............................................................................................................................24 Inputs to a Strategy for Brazilian Cities Page vii Table 2.3 City Size Growth Equation...............................................................................................................26 Table 2.4 Decomposition of City Size Growth ................................................................................................27 Table 2.5 Regression of City Growth Residuals ..............................................................................................28 Table 2.6 Policy Simulation: Favoring Largest Cites Versus Smallest Ones..................................................30 Table 3.1: Slum Formation Across City Sizes, 2000 .......................................................................................35 Table 3.2 Formal Housing Stock Growth Equation .........................................................................................41 Table 3.3 Slum Growth Equation.....................................................................................................................42 Table 3.4 Slum Growth Equation (II)...............................................................................................................44 Table 4.1 Municipal Fiscal Balances, 2000-2004 (Billion of Reais of 2004) ..................................................53 Table 4.2 Current and Primary Balance of 2000-03 (Billion of Reais of 2004)..............................................54 Table 4.3 FRL and Financial Indicators...........................................................................................................55 Table 4.4 FRL and Financial Indicators by Municipal size 2003.....................................................................56 Table 4.5 Interest and Investment Coverage by Municipal Size......................................................................56 Table 4.6 Municipal Revenues, 2000-2004 (Billion of Reais of 2004)............................................................58 Table 4.7 Municipal Revenues by Municipal Size, 2003 (Billion of Reais of 2004).......................................58 Table 4.8 Municipal Current Revenues, 2000-2004 (Billion of Reais of 2004) ..............................................59 Table 4.9 Current Revenue Composition (%) Brazil, 2000-04 ........................................................................60 Table 4.10 Current Revenue Composition (%) by Municipal Size, 2003........................................................61 Table 4.11 Current Revenues Growth by Municipal Size, 2000-2004 (%)......................................................63 Table 4.12 Capital Revenues, 2000-2004 (Billion of Reais of 2004) ..............................................................63 Table 4.13 Capital Revenue by Municipal Size, 2003 (Billion of Reais of 2004)...........................................63 Table 4.14 Municipal Expenditures, 2000-2004 (Billion of Reais of 2004)....................................................64 Table 4.15 Municipal Expenditures by Municipal Size, 2003 (Billion of Reais of 2004) ...............................65 Table 4.16 Municipal Current Expenditures, 2000-2004 (Billion of Reais of 2004).......................................67 Table 4.17 Current Expenditures Growth by Municipal Size, 2000-2004 (%)................................................67 Table 4.18 Municipal Capital Expenditures, 2000-2004 (Billion of Reais of 2004)........................................68 Table 4.19 Capital Expenditures Growth by Municipal Size, 2000-2004 (%).................................................69 Table 4.20 Municipal Expenditures by Function, 2002-2004 (Billion of Reais of 2004)................................72 Table 4.21 Municipal Expenditures Composition by Municipal Size (%).......................................................73 Table 4.22 Municipal Expenditures Per Capita by Municipal Size -2003 (R$ of 2004)..................................74 Table 4.23 Municipal Consolidated 2000-2004 (Billion of R$ of 2004) .........................................................76 Table 4.24 Municipal Consolidated Debt by Municipal Size -2003 (R$ of 2004)...........................................77 Table 4.25 Number of Municipalities by Net Consolidated Debt to Net Current Revenue Ratio ...................78 Table 4.26 Population and GDP by Municipal Size.........................................................................................84 Table 4.27 Municipalities with FRL Indebtedness Indicator Below 1.2..........................................................85 Table 5.1 Total Investments of Institutional Investors on November 2003/January 2004...............................98 Table 6.1 FDH Scores of Output Efficiency for Small Municipalities in Alagoas ........................................117 Table 6.2 Means from IDH-M 2000 Efficiency Analysis for North and South .............................................118 Table 6.3 Means from IMR00 Efficiency Analysis for Maranhao and Sao Paulo.........................................119 Table 6.4 Means from IMR00 Efficiency for Maranhao and Sao Paulo........................................................120 Table 6.5 Comparing IDH-M and Expenditures between 1991 and 2000 .....................................................121 Table 6.6 Description of Variables Ued in Regression ..................................................................................127 Table 6.7 Results of Regression of DEL-AD714...........................................................................................128 Table 6.8 Comparing Mean FDH Efficiency Scores with and without contextual variables.........................129 Table 6.9 Efficient municipalities With and Without Contextual Variables..................................................130 Table 6.10 Existence of Municipal Consortia and Education Councils .........................................................131 Table 8.1 Housing Deficit I Brazil ................................................................................................................184 Table 8.2 Housing Deficit in Brazil................................................................................................................185 Table 8.3 Estimate of Recovery Potential of Costs for Putting in Basic Infrastructure .................................186 Table 8.4 National Distribution of Housing Loans ........................................................................................199 Table 8.5 World Bank National Housing Bank Contract No. 165/BR...........................................................200 Inputs to a Strategy for Brazilian Cities Page viii Table 8.6 Brazil Evolution of Contract Operations Regarding Urbanized Plots by Region ..........................200 Table 8.7 Brasilia, Ciritibaand Recife 2002/2003 ..........................................................................................204 Table 9.1 Basic Facts about 10 Brazilian Cities in 2000................................................................................232 Table 9.2 Households Density and Concentration Index Gradients for 10 Brazilian Cities in 2000 .............235 Table 9.3 Linear Regression Results for Log-Land Prices for Residential Plots ...........................................241 Table 9.4 Probit regression results for land prices likelihood above mean....................................................243 Table 9.5 Land Consumption in Urban Development under Different Standards .........................................247 Table 9.6 Feasibleness of Urbanized Land and Housing Production.............................................................251 Table 10.1 How Do Brazilian Cities Compare to Cities in Other Countries (1990s).....................................260 Table 10.2 Decade-by-Decade Change in Urban and Rural Population ........................................................262 Table 10.3 Urban Population Trends in Brazil's 15 Largest Metropolitan Regions, 1950 to 2000...............263 Table 10.4 Urban Polulation Charge in the 15 Largest Metropolitan Areas, 1950-60 to 1991 - 2000 ..........264 Table 10.5 Permanent Dwelling Units for 15 Largest Metropolitan Regions and Decade by Decade ..........265 Table 10.6 Trends in Household formation 15 Largest Metropolitan Regions, 1970 - 2000.........................267 Table 10.7 Ratio of Change in Permanent Dwelling Units to Changes in the Number of Households .........268 Table 10.8 Total Dwelling Units and Those Lacking Adequate Infrastructure..............................................272 Table 10.9 Population, Urban land Use and Gross Population Density in Latin American Cities ................275 Table 10.10 Trends in Population and Built up Area, Selected Brazilian Cities, 1991 and 2000..................276 Table 10.11 Population Density Gradients in Selected Brazilian Cities 1991 and 2000................................280 Table 10.12 Projections of Brazil's Total, Urban and Rural Population 2000 - 2030....................................281 Inputs to a Strategy for Brazilian Cities Page 1 1. Urbanization, Growth, and Welfare in Brazil1 by By Somik Lall 1.1 Brazil had undergone a phenomenal change in its spatial structure. Over the last 30 years, the share of population living in urban areas rose from 56% in 1970 to 82% in 2002. The urban system has also changed, as new urban forms, cities and metropolitan regions have emerged exploiting the economic and social potential awakened by liberalization, democratization and improvement in infrastructure. Cities are an integral part of Brazil's landscape. Not only does the majority of the population live in urban areas, the entire growth in population that is expected over the next three decades will be in cities (Figure 1.1). This will add about 63 million people to Brazil's cities, and the urban population will cross 200 million. While over 35 million people live in the three largest metro areas, there are about 34 million people who live in 15 metro areas of a million to 5 million, and 10 million who live in medium metro sized areas (500,000 ­ 1 million). (Figure 1.2). Figure 1.1: Urban and Rural Population Dynamics Figure 1.2: Metro Areas, 2000 (population in thousands)2 250000 200000 onita 150000 Total Urban opulP100000 Rural 50000 0 1950 1960 1970 1980 1990 2000 2010 2020 2030 1.2 While urbanization is accompanied by economic growth (90% of the country's GDP is generated in cities), not all urban dwellers benefit from the growth process. Limited opportunities due to human capital deficiencies, limited assets, locational disadvantages, as well as land, credit and housing market distortions lead to pockets of poverty in cities. According to the national household survey (Pesquisa Nacional por Amostra de Domicilios, PNAD), there were 18.3 million urban poor in Brazil in 1996, which accounted for more than half of Brazil's overall poor population. The urban poor also receive disproportionately low access to services. While 86 % of the overall urban population has access to solid waste management services, only 56.4% of the poor do. This problem of unequal access to services varies by city size. 1The findings reported here are a result a joint research program between the World Bank and the Instituto de Pesquisa Econômica Aplicada (IPEA), Brasilia. Preliminary findings from this research have been presented at the World Bank/ IPEA urban research symposium in Brasilia and also discussed with representatives from the Ministries of Cities and Territorial Integration. The research program has been partly funded through a World Bank Research Grant and by the Urban Cluster of the World Bank's Latin America and Caribbean Region. Main contributors to the research are: Somik Lall (TTL), Uwe Deichmann, Hyoung Wang (World Bank); Alexandre Carvalho, Daniel Da Mata (IPEA, Brasilia), J. Vernon Henderson (Brown University) and Christopher Timmins (Duke University). 2Source: UN World Urbanization Prospects 2003. Inputs to a Strategy for Brazilian Cities Page 2 1.3 Across the urban system, there is a need for strategic interventions to meet backlogs in infrastructure, service delivery, and amenity provision, as well as to enhance growth and reduce poverty in Brazil. Policy instruments and programs need to be prepared in a rapidly changing environment. They will also need to be tailored for different size cities. As a basis of such interventions, there is a need first to undertake a diagnostic on the performance of the urban system and provide a framework around which it will be possible to develop and evaluate strategic interventions. 1.4 Part I aims at providing this diagnosis and framework. The work is organized around three topics: how cities grow (in population and productivity) and become more competitive ­ Chapter 1; the impact of policy interventions on city performance ­ Chapter 2, and factors that explain slum formation across cities­ Chapter 3. Chapter 1. Urban Growth and Competitiveness 1.5 In Brazil, public debate has recently centered on the role of the urban system in driving regional economic dynamics. In particular, various levels of government have examined the potential for balancing growth by promoting "secondary cities". The concern is, both, to distribute economic gains more broadly and to relieve the increasing strain experienced by fastest growing cities. This debate occurs at the national level, where the focus is on second tier cities in the lagging regions of the North and Northeast, as well as at the regional level, where states promote development of smaller and medium sized towns. 1.6 The objective of Chapter 1 is to contribute to this debate by analyzing the dynamics of the Brazilian urban system over the last several decades. The analysis is mostly descriptive and focuses on two aspects of urban growth--population and income of agglomerations--that can be consistently measured over the last three to four decades. We begin by describing urban growth patterns between 1970 and 2000 and then investigate two processes in the productive sectors of the Brazilian economy that have accompanied the maturing of the Brazilian urban system. These are industrial specialization and employment deconcentration both across the urban hierarchy and within agglomerations. Finally, we provide a simple analysis of some of the proximate correlates of urban growth. Urban Growth Patterns 1.7 Our examination of urban growth patterns in Brazil focuses on changes in population size and economic productivity. Both are interrelated indicators of city "success". In the presence of free movement of labor and capital, factors of production will move to the areas that promise highest returns. Workers and employees will therefore seek out places in which they can maximize wages given their skills and experience. Successful cities are also able to provide infrastructure and administrative support to businesses which will enhance productivity and, in turn, raise wages. High quality public services and amenities will also attract new residents, especially higher skilled workers that add disproportionately to productivity gains. 1.8 Defining Urban Areas: Before describing population and income dynamics for the urban system, it is essential to develop a working definition of city, urban area or agglomeration, since there is no official statistical or administrative entity in Brazil that reflects the concept most appropriate for economic analysis: a contiguous built up area that operates as a functional economic entity. Socioeconomic data in Brazil tend to be available for municipios, the main administrative level for local policy implementation and management. Municipios, however, vary in size. In 2000, São Paulo municipio had a population of more than ten million, while many other municipios had only a few thousand residents. Furthermore, many functional agglomerations consist of a number of municipios, and the boundaries of these units change over time. Our analysis therefore adapts the concepts of agglomerations from a comprehensive urban study by IPEA, IBGE and UNICAMP (2002) resulting in a grouping of municipios to form 123 urban agglomerations (Figure 1.3). Details about the geographic definitions employed and construction of the database are included in. Throughout this part of the report we refer to these units of analysis as agglomerations, urban areas, or cities. Inputs to a Strategy for Brazilian Cities Page 3 Figure 1.3: Urban Agglomerations by Population Size Source: IPEA, IBGE 1.9 Population growth is occurring across the Brazilian urban size distribution (Table 1.1, see also Lemos et al. 2003). Of the 123 major urban agglomerations in Brazil, only three were above 2 million people in 1970 versus ten in 2000. In the middle of the size distribution in 2000, there were 52 agglomerations with population between 500,000 and 2 million people compared to 25 in 1970. Since we are limiting analysis to cities that were agglomerations in 1991, we cannot track dynamics at the lower end of the distribution. This is because our set includes cities that were not yet agglomerations in 1970, while excluding cities of similar size in later years. However, among the 72 agglomerations that had at least 100,000 people in 1970 (Table 1.2), the average population more than doubled from 553,000 to 1,250,000 over the thirty year period. Table 1.1: City Size Distribution Population size 1970 1980 1991 2000 > 5 million 2 21) 32) 3 2 million - 5 million 1 3 7 7 1 million - 2 million 4 5 5 8 500,000 - 1 million 5 10 15 14 250,000 - 500,000 16 21 23 30 100,000 - 250,000 44 43 44 46 < 100,000 51 39 26 15 Total number of cities 123 123 123 123 Average size 350,857 507,242 657,602 788,222 Min 20,864 41,454 76,816 86,720 Max 8,139,705 12,588,745 15,444,941 17,878,703 1) "São Paulo" and "Rio de Janeiro" . 2) "Porto Alegre" is newly added. Inputs to a Strategy for Brazilian Cities Page 4 Table 1.2: The 7 Fastest Growing Cities between 1970 and 2000* Top 7 Cities Region Population Population Annual pop in 1970 in 2000 growth (1970-2000, %) Campo Grande Central-West 140,233 663,621 5.2 Cuiabá Central-West 226,437 1,051,183 5.1 Brasília Central-West 761,961 2,965,951 4.5 Goiânia Central-West 450,538 1,651,691 4.3 Manaus North 534,060 1,865,901 4.2 Petrolina Northeast 122,900 428,841 4.2 Grande Vitória Southeast 385,998 1,337,187 4.1 Average of the top 7 cities 374,590 1,423,482 4.5 Average of others (65) 571,805 1,231,759 2.5 Total (72) 552,631 1,250,398 2.7 * For the cities with population greater than 100,000 in 1970. 72 cities meet this cutoff criterion. 1.10 Geographically, the strongest population growth has been in the North and Central West regions (Figure 1.4). Growth has been slowest in the South and Southeast, where rapid urban expansion occurred in an earlier period. The Central-West region experienced the second highest urban population growth (4.9 percent annually), but has only 11 agglomerations--compared to 60 in Southeast, and 25 and 24 in the Northeast and South, respectively. In Table 1.2 we list the seven fastest growing cities between 1970 and 2000 among the 72 existing cities in 1970. Over the period the average annual city population growth of the top seven cities was 4.5 percent, considerably higher than for all other cities with population above 100,000 in 1970. Most of the high growth agglomerations (four out of seven) are located in the Central-West region. The fastest growing agglomeration was Campo Grande, with an increase from 140,000 in 1970 to 664,000 in 2000 (5.2 percent annually). Like Campo Grande, the seven fastest growing agglomerations did so from relatively small based populations in 1970, except for Brasília (762,000) and Manaus (534,000). Figure 1.4: Population Growth in Urban Agglomerations by Region 6 00 20- 5 70 19,etar 4 htworg 3 2 % alun 1 An 0 North Northeast Central- Southeast South Total West Total Urban Source: Population Censuses of 1970 and 2000. 1.11 Figure 1.5 shows that, overall, the initial agglomeration size in 1970 does not influence population growth afterwards. There is a positive relationship between agglomeration growth and its manufacturing share in non- agricultural employment. Growth is also positively related to the average years of schooling in 1970 which is used as a measure of human capital accumulation in a city. Regional differences, after controlling for initial size and Inputs to a Strategy for Brazilian Cities Page 5 education are important in explaining city size growth as indicated by the Wald test that shows that regional dummies are jointly significant. These differences may be due to institutional factors or from natural advantages. Figure 1.5: Individual City Size Growth between 1970 and 20001) annual pop growth(70-00) vs. ln(pop in 1970) 5 .0 00) 4 70-(ht .0 3 ow .0 gr 2 pop .0 1 .0 annual 0 11 12 13 14 15 16 ln(pop in 1970) pop growth, North pop growth, Northeast pop growth, Southeast pop growth, South pop growth, West-Central Fitted values 1) For the cities with population greater than 100,000 in 1970. 72 cities meet this cutoff criterion. 1.12 The rapid growth of Central-West cities is related to changes in their industrial composition.3 As shown in Annex 2, the three fastest growing agglomerations have experienced rapid increases in the employment shares of food and beverage manufacturing, business services (finance service, transportation/ ware housing/ communication services, commerce and construction) and public services including education and health services. It suggests that their success in attracting new residents comes from their roles as hubs for serving rural demand in the rapidly expanding soybean growing regions (Motta, Muelle and Torres, 1997). 1.13 Table 1.3 shows the spatial Gini coefficients (Krugman, 1991) for the country and each of the regions for 1970 and 2000. These coefficients are a measure of inequality of population distribution across the 123 agglomerations. The larger the coefficient, the further is the urban system from an equal size distribution. Overall, the spatial Ginis have increased slightly over the period, which is mainly due to the downward movement of small size cities. While the highly concentrated Southeast region has virtually no change in spatial inequality around 0.76, there has been a significant increase in spatial inequality in the Central-West region, which had been the least concentrated in 1970. As a result, the entire southern region, including the Southeast (0.76), South (0.66) and Central-West (0.58), was more spatially concentrated in 2000 than the North (0.46) and Northeast (0.57). Table 1.3: Spatial Gini Coefficients in 1970 and 2000 1970 (a) 2000 (b) (b-a) Total (123) 0.692 0.700 0.008 North (3) 0.456 0.463 0.007 Northeast (25) 0.561 0.569 0.008 Southeast (60) 0.760 0.761 0.001 South (24) 0.626 0.658 0.032 Central-West (11) 0.441 0.583 0.142 Number of cities in parentheses. 3We exclude Brasília, since its growth is mainly due to its role as the capital city in Brazil. Inputs to a Strategy for Brazilian Cities Page 6 1.14 Another way to examine changes in agglomeration size in Brazil is via a transition matrix. It helps examine the degree of mobility of cities up and down the urban hierarchy and test for the stationarity of 123 agglomerations (Eaton and Eckstein, 1997; Dobkins and Ioannides, 2001). Following Black and Henderson (2003), we divide the 1970 agglomeration size distribution into five groups or cells containing approximately 35%, 30%, 15%, 10% and 10% of all cities starting from the bottom, with fixed relative cell cut-off points.4 Table 1.5 presents the resulting transition matrix. The transition probabilities of the transition matrix, Pjk, are calculated as the total number of cities moving from cell j to k over three decades divided by the total number of cities starting in cell j in the three decades. Diagonal elements are the probabilities of staying in the starting state, and off-diagonals the probabilities of moving lower or upper cells. Table 1.4: The Transition Matrix cell in t+1 (2000) 5 (smallest) 4 3 2 1 (largest) 5 0.987 0.013 0.000 0.000 0.000 t 4 0.183 0.720 0.098 0.000 0.000 in 3 0.000 0.091 0.800 0.109 0.000 Cell (1970) 2 0.000 0.000 0.029 0.882 0.088 1 0.000 0.000 0.000 0.000 1.000 1.15 The probability of staying in the same state is the highest at 100 percent for the cities at the top of the hierarchy (cell 1), which implies no downward mobility for the largest agglomerations. Also the mobility is extremely low for the smallest agglomerations in cell five. This extremely high probability of smallest cities in cell five staying in the same state (98.7 percent) is quite different from the finding of Henderson and Wang (2004).5 The cities in the middle portions of the hierarchy have a relatively high degree of mobility moving up and down in response to changing demands of their products, product readjustment, and local entrepreneurship or lack thereof. In particular the lower-medium size cities in cell 4 have only 72.0 percent probability of staying in the same state and the probability of moving down a state exceeds that of moving up (18.3 percent versus 9.8 percent). However the upper-medium size agglomerations in cell two have a higher probability of moving up a state than moving down (8.8 percent versus 2.9 percent). The stationarity of the transition matrices is barely accepted,6 implying the city size distributions evolve over time according to a homogeneous stationary first-order Markov process. Figure 1.6 provides a continuous view of the dynamics of city rankings between 1970 and 2000. The largest changes are among the middle and lower ranked agglomerations. 4The relative size (city population/mean(city population)) upper cut-off points are 0.256, 0.469, 0.812, 1.340 and the maximum. 5For the metro areas of the world with population over 100,000, the probability of the smallest cities staying in the same state was 78 percent, and that of the largest cities 96 percent (Henderson and Wang 2004). 6The 2 statistic is 27.07 with 40 degrees of freedom (p-value 0.059). Inputs to a Strategy for Brazilian Cities Page 7 Figure 1.6: Changes in Population Rank Order between 1970-2000 140 120 100 2000 k anr 80 noi atlupoP 60 40 20 0 0 20 40 60 80 100 120 140 Population rank 1970 Patterns of Income Growth 1.16 The second aspect of city performance is investigated here relates to economic performance. Average household income is used as a proxy for productivity increases, since neither firm level factor productivity, nor data on real wage rates are consistently available for the time period 1970-2000. However, income and wages are strongly correlated in both levels and rates of growth for the years in which both are available at the municipio level (1991 & 2000). Annex 1 provides details.7 1.17 During the period 1970-2000, Brazil's economic performance has fluctuated considerably, ranging from economic boom in the 1970s to a sharp decline in the 1980s and a recovery in the 1990s. We focus on the broad trends between 1970 and 2000. The first pattern discussed here is that relative to the national average, wages are higher in larger agglomerations. Figure 1.7 plots per capita income levels relative to national averages in 1970 and 2000 against the agglomeration populations in those periods. The figure and the corresponding OLS regression result indicate a positive relationship between the per capita income level and the size of a city. A Chow test shows no statistical difference between the 1970 and 2000 patterns. 7As a second caveat, our data represent "nominal" incomes per capita, not "real" incomes. While the average agglomeration income figures have been adjusted for inflation over time, they do not reflect purchasing-power-parity [PPP] estimates across space--i.e., they do not consider local price indexes. Housing costs vary significantly across cities, reflecting commuting costs and rent gradient shifts. As land prices rise, asset price increases will spill over into the prices of retail goods sold in the city. If everyone is a home owner, as land prices rise, residents will recoup implied rent increases in the form of returns on land investment. But many people working in Brazilian cities are renters. Thus a rise in "nominal" incomes may overstate the rise in "real" incomes that translates into tangible welfare gains. Despite these qualifications, however, we believe that the broad patterns discussed in the following paragraphs hold. Inputs to a Strategy for Brazilian Cities Page 8 Figure 1.7: Relative Income Level and City Population in 1970 and 2000 relative income vs. ln(pop) in 1970, 2000 3 age er av onali 2 nat/emocni 1 ai aptc per 0 10 12 14 16 18 ln(pop) per capita income/national average in 1970 Fitted values per capita income/national average in 2000 Fitted values Pooled 1970 & 2000 OLS Results Dependent variable: ln(Income / Average agglomeration income) Coefficient t-value ln(Population) 0.130 4.04 Adj R2 = 0.09 ln(Population)*Year2000 Dummy ­ 0.037 -0.85 N = 246 Dummy for year2000 0.369 0.69 Constant ­ 0.542 -1.41 North South egare 3 av naloi 2 atn/emocni 1 a apticrep 0 10 12 14 16 10 12 14 16 ln(pop) per capita income/national average in 1970 Fitted values per capita income/national average in 2000 Fitted values Graphs by regions 1.18 At the same time, trends in income growth indicate a convergence process, i.e., agglomerations with lower wages in 1970 are growing relatively faster. Recently, Andrade et al. (2004) tested income convergence across Brazilian municipalities from 1970 to 1996.8 Their empirical finding suggests a club convergence (a conditional convergence) between the agglomerations in the poorer Northern region (the North and the Northeast) and the richer Southern region (the Southeast, the South and the West-Central). 8They evaluated convergence of Brazilian municipalities by directly examining the cross-section distribution of income, suggested by Quah (1993, 1997). Inputs to a Strategy for Brazilian Cities Page 9 1.19 reports the OLS estimation results for convergence across urban agglomerations. The speed of convergence is calculated using the coefficient estimate and is reported in the last row in the table.9 The results strongly suggest " convergence" across Brazilian agglomerations. The speed of convergence, when regional dummies are added, is stable around 3.4 percent, which is slightly higher than other countries.10 In the last two columns, we examine the possibility of conditional convergence between agglomerations in the Northern and the Southern regions as a group. The coefficient of the Southern dummy is significantly positive when the same speed of convergence is assumed (column (3)), potentially indicating a higher steady-state growth rate of the Southern region cities. This is consistent with the finding of Andrade et al. (2004). However, overall, we cannot reject the hypothesis of identical speed of convergence and steady state growth rates between the two regions (column (4)). 1.20 Figure 1.8 and Figure 1.9 confirm these findings. Figure 1.8 shows a negative (linear) relationship between the annual income growth rate between 1970 and 2000 and the log of per capita income level in 1970. Figure 1.9 shows the pattern between 1991 and 2000. The fitted lines of the Northern and the Southern regions seem to have a similar slope but different intercepts. Table 1.5: Convergence of Per Capita Income1) (1) (2) (3) (4) (Chow test) Basic Basic equation Basic equation Basic equation equation + 5 regional + 2 regional + 2 regional dummies2) dummies2) dummies2) ln(income in 1970) -0.015 -0.021 -0.022 -0.018 (-10.21) (-12.63) (-13.62) (-5.43) ln(income in 1970)* Dummy(south) -0.005 (-1.20) Constant 0.104 0.136 0.126 0.111 (14.66) (16.72) (18.32) (7.77) Dummy(south) 0.011 0.031 (6.75) (1.84) Number of observations 123 123 123 123 Adj. R2 0.46 0.60 0.60 0.61 Speed of convergence (%) 2.03 3.44 3.47 1) Dependent variable = (1/30)*ln[income(2000)/income(1970)] 2) Five regional dummies correspond to the North, the Northeast, the Southeast, the South and the West-Central regions, with the West-Central as a base. Two regional dummies are for the northern (the North and the Northeast) and the south (the others) regions, with the northern regions as a base. The estimated coefficients for five regional dummies in eq. 2 are not reported. 3) t-values are in the parentheses. 9The speed of convergence is calculated using the formula of Barro and Sala-i-Martin (1995), such that ( ) b^ = -1 -e- T where b is the coefficient estimate and T=30. ^ T 10Most published studies have investigated convergence across administrative regions rather than urban areas. With regional dummies added, the convergence speed across U.S. states for 9 subperiods between 1880 and 1990 was 1.9 percent, Japanese prefectures for 7 subperiods between 1930 and 1990 was 2.3 percent, and European regions for 4 subperiods between 1950 and 1990 was 1.9 percent (Barro and Sala-i-Martin (1995)). Inputs to a Strategy for Brazilian Cities Page 10 Figure 1.8: Annual Income Growth and Initial Income Level for 1970-2000 annual income growth(70-00) vs. ln(income in 1970) 6 .0 00)-07(htwo 4 .0 gr 2 e .0 mocnil 0 nua an 20-. 3.5 4 4.5 5 5.5 6 ln(per capita income in 1970) annual income growth(70-00), North Fitted values annual income growth(70-00), South Fitted values Figure 1.9: Annual Income Growth and Initial Income Level for 1991-2000 annual income growth(91-00) vs. ln(income in 1991) .06 )00 91-(ht .04 ow gr e .02 mocni 0 ual ann 20-. 4.5 5 5.5 6 6.5 ln(per capita income in 1991) annual income growth(91-00), North Fitted values annual income growth(91-00), South Fitted values Specialization Across Cities 1.21 Urban productivity is influenced by economic composition. Both, a concentration in closely related industries (localization economies) and a diversity of economic activities (urbanization economies) tend to enhance the productivity of urban areas. As a country develops, industrial deconcentration tends to increase as a result of improvements in transport, utilities and communication. In earlier stages of development, most modern economic activity is located in one or a few centers where scarce labor and capital can be employed most productively. Manufacturing and higher end services will spread to smaller cities in later stages, allowing these places to specialize in sectors where they have a comparative advantage. As these cities continue to grow in size, Inputs to a Strategy for Brazilian Cities Page 11 other modern-sector activities will locate there, resulting in a diversified economy that offers greater economic opportunity and a lower susceptibility to sector specific downturns. Table 1.6 shows urban concentration for each two-digit level industry across the urban hierarchy. The concentration index shown here has a value of zero if an industry is spread evenly across all cities according to their sizes, as is typical for personal and retail services. If an industry is highly concentrated it has a value approaching two. 1.22 The first column reports the measure of industry concentration, Gj of each 2-digit level industry j. Industry concentration is relatively low for "ubiquitous" industries which are hard to transport and available in many places. Food and beverage manufacturing (0.0042), Metal products manufacturing (0.0046), Furniture and miscellaneous manufacturing (0.0041), and service industries (excluding Finance service) are in this category. The concentration is higher for the natural resource based industries (Tobacco product (0.3698) and Leather products (0.2013)) and the technology intensive industries (Electrical and electronic machinery/equipment (0.0417) and Transportation equipment (0.0486)). 1.23 The third column of Table 1.6 shows the shares of each industry in the total employment of all urban agglomerations, and the last five columns show the relative importance of each sector in urban agglomerations of a given size category. Shares above 100 percent indicate that the industry is more prominently represented in that group of agglomerations compared to the national average share. Several patterns emerge. First, high and medium-high technology industries are concentrated in large cities (Publishing and printing, Chemical products, Electrical and electronic machinery/equipment and Transportation equipment). In particular, computer related industries and financial services are heavily concentrated in large cities. Second, medium technology industries are relatively more concentrated in medium size cities (Textile products and Pulp and paper products). Third, low technology industries that are usually related to natural resource extraction are concentrated in small cities (Agriculture and forestry, Mining and wood products). Finally, "ubiquitous", industries producing non-tradable goods and services are fairly evenly distributed across the urban hierarchy. Overall, among 123 cities in Brazil 65 percent of national employment is concentrated in the 15 largest cities, whereas the 57 smallest cities accommodate only eight percent of national employment. Inputs to a Strategy for Brazilian Cities Page 12 Table 1.6: The Concentration of Industry in 2000 Share in Share relative to the national average, %1 2-digit Industry Classification Gj national Cell 1 Cell 5 employ- (large Cell 2 Cell 3 Cell 4 (small ment cities) cities) Agriculture and forestry 0.0452 5.0 63.0 118.0 172.2 177.6 238.7 Fishing 0.0500 0.2 83.8 145.3 126.8 173.7 76.7 Mining 0.0240 0.3 77.6 98.8 118.2 135.0 238.3 Food and beverage manufacturing 0.0042 2.3 88.9 111.8 128.8 116.6 129.7 Tobacco product manufacturing 0.3698 0.1 128.8 60.8 60.6 22.7 27.2 Textile product manufacturing 0.0102 3.4 88.4 107.6 161.6 115.7 107.5 Leather processing and products manufacturing 0.2013 1.0 101.6 34.7 43.2 286.8 95.7 Wood products manufacturing 0.0199 0.4 80.3 110.5 137.5 134.4 179.0 Pulp, paper and paper products manufacturing 0.0296 0.3 103.5 60.6 144.1 84.5 99.0 Publishing, printing, reproduction of recordings 0.0282 0.9 118.4 72.7 73.3 53.7 55.8 Coal products, petroleum refining, alcohol prod. 0.0262 0.1 110.2 66.0 66.1 109.9 96.0 Chemical products manufacturing 0.0291 1.0 120.0 67.9 62.3 67.2 50.4 Rubber and plastics product manufacturing 0.0484 0.7 114.0 79.6 102.5 55.4 49.9 Metal product manufacturing 0.0046 2.9 95.3 104.5 144.9 73.1 107.3 Machinery and equipment manufacturing 0.0185 0.8 104.6 103.6 116.9 74.5 59.6 Electrical, electronic machinery & equipment 0.0417 0.5 123.3 69.8 70.4 31.1 41.7 Transportation equipment manufacturing 0.0486 1.1 123.6 48.9 65.1 40.0 69.5 Furniture and miscellaneous manufacturing 0.0041 1.5 98.2 98.3 120.4 98.6 97.2 Finance service 0.0230 2.0 121.3 72.9 60.7 48.5 49.0 Transportation, warehouses, communication 0.0030 6.9 109.2 86.5 79.8 85.5 77.5 Commerce 0.0003 21.4 100.2 101.5 99.0 103.4 94.3 Construction 0.0005 8.7 99.6 101.3 99.7 103.1 99.3 Domestic service 0.0010 9.1 99.8 101.7 91.6 102.8 105.7 Public service 0.0063 6.1 99.6 121.8 84.5 87.7 95.5 Education service 0.0006 6.7 99.0 110.9 98.7 94.7 97.3 Health service 0.0020 4.8 108.2 97.5 80.2 72.9 78.8 Other service 0.0013 5.0 106.2 94.5 84.5 90.6 81.5 Other industry 0.0013 6.8 103.7 96.5 88.4 96.4 90.0 (High tech industry) 2 (0.8) (126.2) (69.6) (54.1) (35.1) (30.3) Number of cities 123 15 14 17 20 57 Employment share in a cell 100.0 65.2 12.2 8.2 6.4 8.0 1The relative size cutoff points are calculated for 1970 city size distribution to be divided into five cells containing approximately 35%, 30%, 15%, 10% and 10% of all cities starting from the bottom. 2High tech industry covers; (i) Manufacture of machines and equipment of computer science (CNAE 30000); (ii) Activities of computer science - exclusive maintenance and clerical repairing of machines and computer science (CNAE 72010); and, (iii) Maintenance of machines clerical and computer science (CNAE 72020). 1.24 Another dimension of manufacturing deconcentration is decreasing specialization of cities over time and city size. Smaller and medium size cities tend to be fairly specialized, for instance in food and beverage production, textiles, shoes, or pulp and paper products. Bigger cities tend to have a more diverse industrial base, with providers of niche products and services who can find a market in a large agglomeration. High-tech, specialized production and complex business services also tend to be found more in larger cities, since they require an educated, highly skilled workforce that is attracted to places that offer a greater range of amenities. As development proceeds, manufacturing processes become more complex with more stages of production and Inputs to a Strategy for Brazilian Cities Page 13 greater out-sourcing. This allows smaller and medium size cities to capture some of these activities and become more diverse. 1.25 Specialization and diversity is measured by (Henderson, Lee and Lee 2001). k SPi =( sij - Ej )2 j=1 , where Ej is the share of industry j in national employment, sij is the share of industry j in total employment of agglomeration i, and the sum is over k industries locally. The index measures for each industry how much the local production share differs from the national share. If all industries mimic the national share the index has a value zero and the city is perfectly diverse. A highly specialized city would have an index approaching two. Figure 1.10: Index of Specialization by Agglomeration Size Group 1980 2000 0.07 xed 0.06 in 0.05 n io 0.04 at 0.03 0.02 ecializpS0.01 0 Smallest Largest Total City size category Source: Brazilian Population Censuses of 1970 and 2000; urban size categories as defined before. 1.26 Figure 1.10 shows how this index varies across the urban hierarchy in Brazil. Specialization has decreased across all size categories in the last 20 years. As expected, the specialization index is negatively related to agglomeration size. As a city becomes bigger, diversification increases. In 2000, the specialization index in the largest agglomerations (0.0047) is just 28 percent of that in the bottom agglomerations (0.0166). Figure 1.11 and the corresponding regression also show a significant negative relationship between agglomeration specialization and size in 2000. A Chow test shows no statistical difference between the Northern and the Southern regions. Inputs to a Strategy for Brazilian Cities Page 14 Figure 1.11: City Specialization in 2000 ln(city Specialization) vsln(pop in 2000) -2 )niota -3 lizaicepSytic(ln -4 -5 -6 -7 10 12 14 16 18 ln(pop in 2000) ln(city specialization), North Fitted values ln(city specialization), South Fitted values Pooled 1970 & 2000 OLS Results Dependent variable: ln(Specialization index for 2000) (1) (2) ln(Population 2000) -0.358 -0.502 (-5.28) (-3.37) ln(Population)*South dummy 0.151 (0.91) Dummy for South region -2.413 (-1.12) Constant -0.220 1.997 (-0.25) (1.03) Adj. R2 0.18 0.22 t-values in parentheses Industrial Decentralization 1.27 As the urban system develops, typically manufacturing decentralizes out of the biggest cities first into their suburbs and nearby ex-urban transport corridors and then into smaller cities, with their lower cost of living, lower wages, and lower rents (Henderson et al. 1995, Deichmann et al. 2005). Decentralization as noted above is spurred by inter-city and hinterland infrastructure investment and increasing overall sophistication of the labor force. In a modern system of cities the share of manufacturing in local economic activity tends to rise as we move down the urban hierarchy. As part of a domestic product cycle, traditional standardized products are manufactured in smaller cities and more high tech, innovative products in the biggest cities. In contrast to manufacturing, as we move down the urban hierarchy, the share of business services such as financial and legal activities in local economic activity declines. Conversely, the ratio of manufacturing to business services falls as we move up the urban hierarchy, reflecting the service orientation of bigger cities (Kolko 1999). 1.28 As suggested by theory, Brazil has experienced a manufacturing decentralization process between 1970 and 2000. In Table 1.7 we list the ratios of employed population working in the secondary and the tertiary industries in 1970 and 2000 (see also Figure 1.12). We group agglomerations into five size groups as before based on the relative population cutoff points. A comparison between 1970 and 2000 employment shares shows a typical manufacturing decentralization process, albeit less than we anticipated. In 1970 the secondary industry share in overall local employment was positively related to agglomeration size. Similarly, the manufacturing share in total non-agricultural employment increases from 28.9 percent among the smallest agglomerations to Inputs to a Strategy for Brazilian Cities Page 15 34.7 percent in the top group as we move up the urban hierarchy in 1970. But by 2000 there is a dramatic drop in the manufacturing share of the cities in the top two cells to about 25 percent. While the manufacturing share drops in smaller cities as well, the decline is more modest, so that by 2000 smaller cities have more local manufacturing concentration than bigger ones. We therefore observe decentralization of manufacturing industry out of big cities. Table 1.7: Employment Share by Industry in 1970 and 2000 Agglomeration size groups1 Number Employment share, % Core area's Core of cities secondary area's Secondary Tertiary a + b ×100 a industry tertiary industry (a) industry share2) industry (b) share2) 1970 Largest: 1.340 pop/mean 12 30.9 58.1 34.7 64.2 76.1 0.812 pop/mean < 1.340 12 23.4 53.8 30.3 58.0 72.1 0.469 pop/mean < 0.812 19 24.1 46.1 34.4 69.7 71.9 0.256 pop/mean < 0.469 36 19.7 46.1 30.0 83.8 86.7 Smallest: pop/mean < 0.256 44 18.9 46.5 29.0 . . Total 123 27.8 54.8 33.7 66.6 77.1 2000 Largest: 1.340 pop/mean 15 24.1 71.2 25.3 47.0 61.3 0.812 pop/mean < 1.340 14 23.0 69.8 24.8 60.0 70.9 0.469 pop/mean < 0.812 17 27.3 63.1 30.2 61.1 70.3 0.256 pop/mean < 0.469 20 24.6 65.3 27.3 79.8 81.4 Smallest: pop/mean < 0.256 57 24.0 62.8 27.7 . . Total 123 24.2 69.4 25.9 55.2 66.5 1) The relative size cutoff points are calculated for 1970 city size distribution to be divided into 5 cells containing approximately 35%, 30%, 15%, 10% and 10% of all cities starting from the bottom. 2) Core area's secondary (tertiary) industry share (%) is the ratio of the secondary (tertiary) industry employment in core areas to the total secondary (tertiary) industry employment in an agglomeration. The ratio of the number of suburb to core areas (MCAs) in each cell is 13.7, 4.0, 3.4 and 0.9 for 1970, and 11.7, 4.0, 2.5 and 1.5 from the top cell to cell 4. The last cell with relative population less than 0.256 is not calculated since the core areas have too small suburb areas (on average 0.3). Figure 1.12: Sector Employment Shares by Agglomeration Size Distribution Secondary sector Secondary sector Tertiary sector Tertiary sector Core area's secondary industry share Core area's secondary industry share 90 90 80 80 70 70 tnecr 60 60 50 50 40 tnecr 40 Pe 30 Pe30 20 20 10 10 0 0 Smallest Largest Total Smallest Largest Total City size category City size category Source: Brazilian Population Census of 1970 and 2000 Inputs to a Strategy for Brazilian Cities Page 16 1.29 Overall, the share of tertiary industry has increased rapidly from 54.8 percent in 1970 to 69.4 percent in 2000. The city size decomposition of the tertiary industry shows concentration of service sector employment in bigger cities. As we move up the urban hierarchy, the service industry share in 2000 increases from 62.8 percent among the smallest agglomerations to 71.2 percent among the largest. The manufacturing industry share in non- agricultural employment is highest in the medium size cities (30.2 percent in 2000). It suggests that manufacturing decentralization was relatively more intense from large to medium size cities. 1.30 Industrial decomposition within agglomerations shows a pattern of manufacturing suburbanization from the core to the suburb areas. The manufacturing industry share in the core areas, relative to the total city manufacturing industry employment, decreased from 64.2 percent in 1970 to 47.0 percent in 2000 (Table 1.7). The relative manufacturing employment share of the core areas decreases as a city becomes bigger. Manufacturing suburbanization is more distinct in bigger agglomerations and this process is also observed in the tertiary sector. The suburbanization of service industry shows a similar pattern as those of manufacturing. The service industry has experienced an overall increase in suburbanization over the period, and the suburbanization is relatively more intense in the largest cities. Still, overall, the service industry in 2000 is more concentrated than manufacturing in the core areas (66.5 percent versus 55.2 percent). Summary of Findings 1.31 In this chapter, we describe patterns of population and income growth across urban agglomerations in Brazil. In general, the Brazilian urban system follows a dynamic trajectory that has also been found in other countries. Urban growth happens throughout the urban system, but with regional differences in magnitudes. In particular, cities in the Central-West and North have recently grown faster than the already established urban agglomerations in the traditional industrial regions of the south. Per capita incomes tend to be larger in bigger cities, a pattern that has not changed over the three decades since 1970. However, there is some indication of income convergence with smaller, lower income cities experiencing relatively faster income growth. 1.32 Cities of different size tend to show a different mix of economic activity. Small urban areas are dominated by non-tradables sectors and lower level services. Some small and medium size agglomerations also host industries that depend on the natural resource base. Medium size cities are typically more specialized in a few industries such as textiles and pulp and paper products. Large agglomerations are much more diversified with a mix of higher technology manufacturing and specialized business services. These require increased labor skills and yield higher profits which translate into higher wages, which in turn attract qualified workers. 1.33 Within larger agglomerations, manufacturing sector tends to be decentralized. As land prices and congestion in the center increase, enterprises move out. Rather than moving into smaller towns, where wages are low, they locate in the periphery of large cities to continue to reap agglomeration benefits, such as proximity to buyers, suppliers and specialized services. Annex 1: Data Sources and Definitions 1.34 There is no official definition of "city" or "agglomeration" in Brazil. The lowest administrative level consists of more than 5000 municipios. However, these vary greatly in size and many functional economic and population agglomerations consist of a number of municipios. In this paper, we therefore follow the example of a study of Brazilian urban dynamics by IPEA, IBGE and UNICAMP (2002). It defined agglomerations based on their place in the urban hierarchy from "World Cities" (Sao Paulo and Rio de Janeiro) to subregional centers. For each agglomeration, this study identified the municipios that were a functional part of the urban area. The municipios belonging to each agglomeration were then further classified into eight categories according to how tightly they are integrated in the agglomeration, from "maximum" to "very weak". The main criteria used in these classifications were centrality, function as a center of decision making, degree of urbanization, complexity and diversification of the urban areas, and diversification of services. These were measured by a range of census and Inputs to a Strategy for Brazilian Cities Page 17 other variables such as employed population in urban activities, urbanization rate, and population density. We modified this classification slightly by also including smaller municipios to existing agglomerations if their population exceeded 75,000 population and more than 75 percent of its residents lived in urban areas in 1991, or if they were completely enclosed by an agglomeration. 1.35 The agglomeration definitions developed by IPEA, IBGE and UNICAMP (2002) are based on the Brazilian Bureau of Statistics (IBGE) Population Census of 1991 and the Population Count of 1996, while our study captures dynamics from 1970 to 2000. During this time, many new municipios were created by splitting or re-arranging existing ones. In fact, the number of municipios increased from 3951 to 5501 during these three decades. In order to create a consistent panel of agglomerations for the 1970 to 2000 period, we therefore used the Minimum Comparable Area (MCA) concept. MCAs group municipios in each of the four census years so that their boundaries do not change during the study period. All data have then been aggregated to match these MCAs. The resulting data set represents 123 urban agglomerations that consist of a total of 447 MCAs. 1.36 The sources for the majority of data employed in this paper are the Brazilian Bureau of Statistics (IBGE) Population and Housing Censuses of 1970, 1980, 1991 and 2000. We used the full Brazilian census counts to get information about total population and housing conditions (urbanization rate). Other data were collected only for a sample of households. We used this census sample information for income, industrial composition, education, piped water provision, and electricity availability. The sample sizes varied across census years (1970: 25 percent; 1980: 25; 1991: 12.5; 2000: 5). but all are representative at the municipio level, and thus are also reliable at the MCA level employed in this study. Table 1.8 reports the variables, their source and the years available. Table 1.8: Variable Definitions and Data Sources Data Source Years Population Population Censuses 1970, 1980, 1991, 2000 Urbanization rate Population Censuses 1970, 1980, 1991, 2000 Income per capita (monthly deflated to 2000 values) Population Censuses (sample) 1970, 1980, 1991, 2000 Industrial composition (percentage of labor force employed Population Censuses (sample) 1970, 1980, 1991, 2000 in an economic activity sector) Years of schooling (average) Population Censuses (sample) 1970, 1980, 1991, 2000 Percent of households with access to piped water provision Population Censuses (sample) 1970, 1980, 1991, 2000 Percent of households with access to electricity Population Censuses (sample) 1970, 1980, 1991, 2000 Income vs. wages 1.37 As discussed in the text, per capita income is not the preferred proxy for productivity growth as it includes not only real wage income, but also transfer payments and dividends or capital gains that were not necessarily generated locally. However, we used income because consistent wage data are available only for 1991 and 2000, and the overall quality of income data is better than the wage information. Overall, we do not believe that the use of income data significantly affects our analysis, since the correlation of income and wage data, both in terms of levels and growth rates, is very high for the two years in which both are available Table 1.9 Table 1.9: Correlation Coefficients ­ Wages Versus Income Wage 1991 Wage 2000 Income 1991 Income 2000 Wage growth Income growth Wage 1991 1.000 Wage 2000 0.953 1.000 Income 1991 0.993 0.948 1.000 Income 2000 0.948 0.989 0.958 1.000 Wage growth -0.238 0.051 -0.227 0.032 1.000 Income growth -0.419 -0.147 -0.412 -0.148 0.937 1.000 Inputs to a Strategy for Brazilian Cities Page 18 Annex 2: Industrial Composition of the 3 Fastest Growing Cities in the West-Central (excl. Brasília)* 2000 1980 Change in Employment Employment Ratio relative 2-digit Industry Classification industrial share of the 3Ratio relative share of the 3 to the nationalcomposition cites in West-to the nationalcites in West- average, % (a-b)/b*100 Central average, % (a) Central (b) Agriculture and forestry 6.56 132.02 12.09 145.78 -9.44 Fishing 0.09 36.29 0.45 62.98 -42.38 Mining 0.23 80.90 1.41 42.54 90.19 Food and beverage manufacturing 2.70 119.10 1.26 35.95 231.29 Tobacco product manufacturing 0.02 23.18 0.60 81.32 -71.49 Textile product manufacturing 4.24 125.71 14.22 142.59 -11.83 Leather processing and products manufacturing 0.46 46.21 11.96 162.38 -71.54 Wood products manufacturing 0.54 123.33 6.07 118.66 3.93 Pulp, paper and paper products manufacturing 0.10 28.22 3.95 138.48 -79.62 Publishing, printing and reproduction of recordings 0.73 81.47 0.11 17.10 376.33 Coal products, petroleum refining and alcohol production 0.06 54.91 1.58 113.73 -51.72 Chemical products manufacturing 0.54 56.12 5.40 116.85 -51.97 Rubber and plastics product manufacturing 0.19 27.14 7.44 98.62 -72.48 Metal product manufacturing 1.80 61.81 2.45 57.27 7.92 Machinery and equipment manufacturing 0.35 43.81 15.24 160.52 -72.70 Electrical and electronic machinery, equipment manufacturing 0.11 22.78 2.61 122.68 -81.43 Transportation equipment manufacturing 0.22 20.69 1.41 73.94 -72.02 Furniture and miscellaneous manufacturing 1.30 86.67 3.07 107.53 -19.40 Finance service 1.62 78.99 0.18 32.82 140.63 Transportation, warehouse and communication service 6.01 86.57 0.88 30.70 182.02 Commerce 23.63 110.36 0.15 13.58 712.47 Construction 9.11 104.90 0.09 23.48 346.66 Domestic service 9.48 104.32 1.22 84.86 22.93 Public service 7.47 122.18 1.28 24.11 406.80 Education service 6.68 99.65 0.38 18.18 448.03 Health service 4.35 90.02 0.73 42.37 112.44 Other service 5.28 105.68 3.06 43.08 145.28 Other industry 6.16 90.35 0.72 107.37 -15.85 (High tech industry) 0.61 74.66 * The 3 fastest growing cities in the West-Central region are Campo Grande, Cuiabá and Goiânia. Inputs to a Strategy for Brazilian Cities Page 19 2. City Performance and Policy Actions by By Somik Lall 2.1 Relevant Issues: Which factors make some cities perform better than their peers? What is the impact of infrastructure and service delivery on urban growth? How does it impact large metro areas? What are the prospects for secondary city growth? Is the development of secondary cities fiscally viable? What is the role of institutions and governance in promoting urban growth? Background 2.2 Why are some cities more successful than their peers? Is the `success' of individual cities driven by factors mostly external to any city's immediate control (location, growth in market potential, being a port in a period of national trade growth, national level decentralization and improved governance), or do individual city policies and politics influence growth and development? Disentangling the relative contribution of regional and local efforts is important for understanding the potential of alternate policy interventions for stimulating growth of cities across the national urban system. At this time, there is little clarity on the effectiveness of local and national policy environments on urban growth in Brazil, as well as most developing countries. 2.3 The scale of urbanization and the distribution of population across the urban hierarchy (as discussed in Chapter 1) will challenge policy makers to devise appropriate policies for cities of different sizes. Across the urban system, there will be need to meet backlogs in infrastructure, service delivery, and amenity provision, as well as accommodate further growth. In addition to population increases across the urban system, fiscal and administrative decentralization has increased the role of individual cities in attracting investments and in providing services that are responsive to the needs of local residents. Brazil is one the most decentralized among developing countries. The 1988 Constitution established municipalities as the third level of government, and provided states and municipalities with more revenue raising power and freedom to set tax rates. However many local governments have limited administrative and institutional capacity, and have not been able to effectively use their autonomy to improve service delivery or attract new investment. A recent study by the World Bank (World Bank 2002) identifies that maximizing urban competitiveness from agglomeration economies and minimizing congestion costs from negative externalities are key challenges facing national and local governments in Brazil. 2.4 Under this backdrop of rapid population growth and decentralization of administrative and fiscal responsibilities, it becomes essential to identify what types of interventions stimulate growth of individual cities. In addition, we want to find out the consequences of favoring investments in secondary cities on aggregate efficiency and economic growth. There is an ongoing debate in Brazil's policy circles that the largest agglomerations have become too big leading to significant negative externalities of crime, social conflict, and high land costs, and policies should be designed to actively stem the growth of these large agglomerations and favor investments in secondary cities. It is however not clear if net agglomeration economies in large cites can be offset by incentives and other measures to divert growth to smaller cities. Measuring City Growth 2.5 To examine these issues, we consider a model of a city, which consists of a demand side--what utility levels a city can pay out--and a supply side--what utilities people demand to live in a city. For the empirical analysis, we construct a dataset of Brazilian agglomerations to examine city growth between 1970 and 2000. Much of the underlying data come from the Brazilian Bureau of Statistics (IBGE) Population Censuses of 1970, 1980, 1991, and 2000. he details of the dataset are provided in Chapter 1, Annex 1. 2.6 Urban growth is represented by both individual city productivity growth and city population growth, which are different indicators of city "success" and represent two interconnected dimensions of successful urban Inputs to a Strategy for Brazilian Cities Page 20 growth. However before we can look at any individual city's success, we need to understand the broader context, in which the economy as a whole is changing. Cities from an economic perspective represent the way modern production is carried out in a country and, as such, reflect what is occurring in the country as a whole. 2.7 Production composition of cities varies by city size, where different types of goods are best produced in bigger versus smaller cities. If national output composition changes, altered by changing trade demand or domestic demand that changes with economic growth, then demand moves away from goods produced in smaller types of cities and those cities will suffer a setback. Some will falter; others will adjust what they produce and perhaps upgrade, moving up the urban hierarchy. Which ones adjust well may depend on "luck", but it may also depend on observable attributes such as education of the labor force. A better educated labor force may allow for more nimble adjustment and up-scaling of products produced what is called the reinvention hypothesis. Similarly the skill composition of the labor force will vary across cities in systematic ways, as output composition and skill needs vary. More generally, national productivity growth comes from productivity growth within cities, which engender the close social-spatial interactions inherent in innovation, knowledge accumulation and technological improvements. To understand individual city success, we need to account for the external, national factors driving urban changes, as well as to understand the sources of local productivity growth. 2.8 At the same time we need to be able to measure when cities are being "successful" versus less successful and what drives success. Much of success may be driven by conditions external to the city, as just noted. In addition to demand changes, changes in national institutions, for example providing smaller cities with greater autonomy in local public sector decision making and greater access to fiscal resources may make it easier for smaller cities to finance the infrastructure and public sector services demanded by firms (transport and telecommunications) and by higher skilled workers (e.g., better schools) and compete successfully with bigger cities for certain industries. For terms of city level conditions, better run cities with more efficient use of public sector revenues will be more attractive to both firms and migrants. And better run cities will co-ordinate better with local businesses to help service their needs and make them more productive. So part of measuring city success is measuring what local producer and consumer amenities are valued and what cities are better at providing these amenities. 2.9 In related work, Glaeser et al. (1995) examined how urban growth of the U.S. cities between 1960 and 1990 is related to various urban characteristics in 1960, such as their location, initial population, initial income, past growth, output composition, unemployment, inequality, racial composition, segregation, size and nature of government, and the educational attainment of their labor force. They showed income and population growths are (1) positively related to initial schooling, (2) negatively related to initial unemployment, and (3) negatively related to the initial share of employment in manufacturing. Racial composition and segregation are not correlated with later city population growth. Government expenditures (except for sanitation) are also not associated with subsequent growth. However, per capita government debt is positively correlated with later growth.11 2.10 In a long run analysis, Beeson et al. (2001) examine the location and growth of the U.S. population using county-level census data from 1840 and 1990. They showed access to transportation networks, either natural (oceans) or produced (railroads), was an important source of growth over the period.12 In addition, industry mix (share of employment in commerce and manufacturing), educational infrastructure, and weather have promoted population growth. 2.11 In a recent paper for developing countries, Au and Henderson (2005b) take a slightly different approach. They model and estimate net urban agglomeration economies for cities in China, which can be postulated by 11They attributed this correlation to higher expected growth which made it cheaper to borrow, or government invest heavily in infrastructure to serve that growth. 12Transportation network is represented by a group of dummy variables indicating ocean, mountain, confluence of two rivers, railroads, and canals. Inputs to a Strategy for Brazilian Cities Page 21 inverted-U shapes of net output or value-added per worker against city employment. They find urban agglomeration benefits are high ­ real incomes per worker rise sharply with increases in city size from a low level, level out nearer the peak, and then decline very slowly past the peak. The inverted-U shifts with industrial composition across the urban hierarchy of cities. Larger peak sizes are for more service oriented cities, but smaller for intensive manufacturing cities. In addition, (domestic) market potential and accumulated FDI per worker have significant and beneficial effects on city productivity, measured by value-added per worker. However, percentage of high school graduates, distances to a major highway and to navigable rivers, and kilometers of paved road per person have no effects, once market potential is controlled for. Model and Estimatio Estrategy 2.12 The model that is used to examine determinants of city growth consists of a demand side--what utility levels a city can pay out--and a supply side--what utilities people demand to live in a city. We estimate aspects of the demand and supply side; and then a reduced from equation that describes city sizes and their growth. In the end the focus is on the last item. Demand Side 2.13 The demand side is given by the schedule of utility levels a city can offer workers, as city size increases. A prime determinant of that is income, I, which consists of wage income and income from rents and other non- labor sources. In addition in an indirect utility function we also have a vector of items, Qi , such as commuting costs, housing rents, local taxes, and local public services and amenities, so that Ui =U(Ii,Qi) D (1) For wage income there is a wage rate component and then a work effort component discussed momentarily. The wage rate component comes from value of marginal productivity relationships, where wi = w(MPi,ri,ei, Ni) (2) 2.14 In (2) r is the rental rate on capital, e is the quality or education level of workers, MP is market potential reflecting the demand for a city's output and hence the price it receives, and N is a measure of scale, such as city employment. MP from the new economic geography and monopolistic competition literature has a specific form with components we can't measure. We make two adjustments. First we use "nominal" market potential, which is simply the distance discounted sum of total incomes of all MCAs in Brazil for city i , or MPi = TI j (3) j, ji ij 2.15 TI is total income andij represents the transport cost between i and j.13 The calculation of market potential is described in Appendix A, where we use distance as the measure of transport costs. However travel times and costs vary by more than distance. Brazil for 1968, 1980 and 1995 has a measure of the transport cost from each city to its state capital. We divide that variable by distance from the city to the state capital to get a city specific measure of local unit transport costs which producers in a city face in selling in the local region. The variable "inter-city transport costs",ii , will be determined by intercity road infrastructure investment. 2.16 The major items from urban theory affecting worker well-being, apart from the wage rate are rents and commuting costs. Commuting costs are time costs, of which part will be reflected in lost work time or energy for work, and part in out-of-pocket commuting costs. So total wage income is a function of both the wage rate and 13The MCAs (Minimum Comparable Areas) are groups of municípios. The detailed description is in Appendix C. Inputs to a Strategy for Brazilian Cities Page 22 hours and energy available to work, where the later will be negatively affected by commuting times. Housing costs are tricky, since higher housing rents are also reflected in higher non-labor income earned by landowners. 2.17 For demand side estimation, what we know from the data is total income per worker in each city. We model that as a function of the determinants of the wage rate and then factors affecting work time/energy and housing rental income. Both are a function of city size. In sum we estimate: Ii = I D(MPi,ii,ei, Ni) (4) 2.18 The scale variable, N, captures three things, scale externality effects on wage rates, increasing housing rental incomes, and reduced work time/energy. As such its sign is uncertain--if cities are at a size where the commuting cost aspects of urban living weigh heavily, at the margin increases in scale could detract from incomes. That will be the case in our estimation (which is also good for "stability" given supply curves are upward sloping--being on the rising part of the "demand curve" can be problematical and also makes sign interpretations in the city size equation more difficult as discussed later). Population Supply 2.19 The population supply relationship we estimate has population supplied to a city increasing in utility offered per worker, which we approximate by income per worker. This will tell us the supply elasticity of people to a city. In addition supply is shifted by attributes, Zi , of the surrounding area--or substitutes of places to work for population in the area. We have supply to a city of population from nearby rural areas. It is decreasing in surrounding rural incomes where we use a gravity measure of surrounding rural incomes, and it is increasing in surrounding rural population supply where again we use a gravity measure of surrounding rural population. The calculation details are in Appendix B. The supply equation is given by Ni = NS (U s(Ii),Zi), where NS /I > 0, NS /Z > 0 (5) Note the inverse we will use later is Ii = I S (Ni,Zi) where I S /N > 0, I S /Z < 0. (6) Determinants of Growth 2.20 Urban Demand: Results from estimating the demand side model are presented in Table 2.1, pooling three years (1980, 1991, and 2000). In columns 1 and 2 the scale measure is total workers in each city. In column 3, population instead of total workers is used to represent urban scale. In column 4, we provide the effects on outcomes of a one standard deviation increase in covariates. All variables have significant impacts on total income per worker. For market conditions, average schooling and ln(market potential), a one standard deviation increase (1.26 and 1.01) raises total income per worker by 33% and 11% respectively. (Of course for covariates in log form we already have elasticities.) For ln(number of workers), we have the classic simultaneity problem where larger cities per se are "more efficient" so we get a positive coefficient in OLS, but in fact we expect negative scale effects at the margin, because we should be on the downward sloping portion of inverted U's (of Inputs to a Strategy for Brazilian Cities Page 23 income against city size).14 A reduction of one standard deviation (-1.13) in ln (number of workers) increases total income per worker by 18%, with similar results when the size measure is population (column 3). Table 2.1 Demand Side: Determinants of Income Per Worker (robust standard errors in parentheses) Dependent variable: ln(income per capita) (1) (2) (3) (4) The effects of 1 GMM-IV OLS GMM-IV s.d. increase in covariates of (1) Average Schooling 0.263*** 0.236*** 0.255*** 0.331 (0.018) (0.025) (0.017) ln(market potential) 0.108*** 0.015 0.095*** 0.109 (0.020) (0.013) (0.017) ln(no. workers) -0.158*** 0.002 -0.141*** -0.179 [ln(population) for (3)] (0.029) (0.011) (0.025) ln(inter-city transport costs) -0.075** 0.034 -0.055** -0.033 (0.029) (0.027) (0.028) state capital dummy 0.158** 0.040 0.169** (0.068) (0.054) (0.067) ln(distance to São Paulo) -0.075*** -0.077*** -0.071*** -0.077 (0.006) (0.009) (0.006) time dummies yes yes yes Observations 369 369 369 R2 0.853 Hansen J statistic (overidentification test) 3.781 3.829 (p-value) (0.436) (0.430) Average of Partial R2 0.572 0.575 Average of Partial F's 58.10 57.16 *** significant at 1% level; ** significant at 5% level; * significant at 10% level. 2.21 Other variables may reflect policy conditions. Cities further from Sao Paulo, over and above declines in market potential, suffer. While this could reflect some aspect of Sao Paulo's huge, modern business service sector market that is critical to access for other cities, it might reflect other items like cost of capital or state provided production amenities that respectively rise and fall as one moves further from the center of the political elites and power in Sao Paulo. The intercity-transport cost variable, reflecting relative investments of transport infrastructure is significant where a reduction of one standard deviation (-.344) increases total income per worker by 3.3%. For intercity-transport costs we use the 1980 value for years 1980 and 1990; and we use the 1995 value for 2000. We give zero values to ln(intercity-transport costs) of state capital cities and insert a dummy for sate capitals. Note the coefficient on state capitals of .16 is much larger than would be expected (.075*.68= .051), if we assigned state capitals mean unit transport costs (.68), given that latter variable has a coefficient of .075. This may suggest state capitals are larger than expected, perhaps favoring themselves (or being favored by the national government) with investment in unobserved production amenities. 14Theory suggests that, under free migration within a country, if particular cities are not a their peak of inverted U's, they will be to the right of the peak, due to either "stability" conditions in migration-labor markets or conditions on what constitutes a Nash equilibrium in migration decisions (Au and Henderson, 2004; Duranton and Puga, 2004). Inputs to a Strategy for Brazilian Cities Page 24 Table 2.2 Population Supply (robust standard errors in parentheses) Dependent variable: ln(population) (1) (2) (3) (4) (5) GMM-IV OLS GMM-IV GMM-IV GMM-IV (1980) (1991) (2000) Ln(income per capita) 2.997*** 1.813*** 2.700*** 2.937*** 3.335*** (0.413) (0.378) (0.402) (0.417) (0.420) Ln(rural income opportunities: -6.892*** -4.152*** -7.645*** -6.236*** -6.678*** market potential) (0.997) (0.819) (1.195) (0.850) (0.965) Ln(rural pop. supply market 7.555*** 4.878*** 8.330*** 6.901*** 7.325*** potential) (0.936) (0.752) (1.133) (0.788) (0.908) time dummies yes yes yes yes yes Observations 369 369 123 123 123 R2 0.745 Hansen J statistic (overidentification test) 1.341 2.401 0.161 2.444 (p-value) (0.854) (0.663) (0.997) (0.655) Average of Partial R2 0.708 0.735 0.693 0.701 Average of Partial F's 47.26 30.14 36.70 41.02 *** significant at 1% level; ** significant at 5% level; * significant at 10% level. 2.22 Supply Side: Results for population supply are provided in Table 2.2. Again, for the estimation we pool three years (1980, 1991, and 2000). Columns 1 and 2 give the GMM-IV and then OLS results. In column 1, a 1% increase in a city's total income per capita increases city population by 3%, suggesting an elastic supply curve, but one that is definitely not perfectly (or highly elastic). The gravity measures of surrounding rural population supply and rural income opportunities have the expected opposite effects with similar magnitudes. A 1% increase in surrounding rural population supply increases city population by 7.6%, and a 1% increase in surrounding rural income opportunities decreases city population by 6.9%. Thus, city populations are very sensitive to rural population supply and earning opportunities. 2.23 In columns 3-5, we present supply elasticities by year. The point estimates of the income coefficient increases over time, indicating increasing mobility. But as just noted even in 2000, the elasticity, 3.3, is far from a perfect mobility elasticity.15 2.24 City Growth: A growth formulation allows us (i) to separate out labor force quality improvements from the effect of base period education on local technology (knowledge accumulation spillovers); (ii) to difference out time invariant unobservables affecting city size that are difficult to instrument for; and (iii) to incorporate adjustment processes, where city growth is affected by base period size and industrial composition, as well as changing economic conditions. Table 2.3 shows the growth results pooling 1991-1980 and 2000-1991 differenced equation years for equation (7). Covariates are differenced; in addition, before differencing, we now normalize market potential measures with the mean for that year to emphasize how each city's relative conditions are changing over time. For differenced intercity-transport costs, we use the difference between 1995 and 1980 for 15Under perfect labor mobility, we expect a horizontal population supply curve. All the cities offer the same utility level, and city sizes are only determined by demand-side factors. Inputs to a Strategy for Brazilian Cities Page 25 2000-1991; and the difference between 1980 and 1968 for 1991-1980. All covariates, except changes in rural population supply have strong and expected sign coefficients. The poor performance of rural population supply is most probably due to the limited variance in relative market potential measures over time and their high degree of correlation (negative between changes in rural income and population supply conditions, as would be expected). Controlling for population allows for dynamic adjustment to steady state levels from the base, and introducing industrial composition allows for adjustment relative to changes in national output composition. Initial city size has a negative coefficient, suggesting either mean reversion or some conditional convergence in population growth across cities, or both. Also, cities with high manufacturing ratios in the base period experience faster growth. We also find that once base period population and industrial composition are controlled for, state capitals are growing faster than other cities, perhaps reflecting favoritism as noted earlier. 2.25 For changes in basic demand and supply conditions, we find that decreases in rural income opportunities and increases in market potential of goods and labor force quality (measured by changes in educational attainment) increase the growth rate of city population. As a new effect, educational attainment in the base period increases city population growth rates afterwards, confirming spillover effects of knowledge accumulation. Reductions in inter city-transport costs have a fairly strong effect on city population growth rates: a 10% decrease in inter city unit transport costs increases city population growth by 1.4% over a decade. In the next section we discuss in more detail the magnitudes of effects on growth of different covariates. 2.26 In Table 2.3, columns 3 and 4, we introduce one additional local characteristic to the base specification, the base period homicide rate, an amenity which may affect city growth. Results suggest that higher homicide rates have a detrimental effect on city growth. For example, a 10% increase in base period homicide rates reduces city growth by 1.5% over the next decade. Inputs to a Strategy for Brazilian Cities Page 26 Table 2.3 City Size Growth Equation (robust standard errors in parentheses) Dependent variable: ln(population) (1) (2) (3) (4) GMM-IV OLS GMM-IV OLS ln(rural pop. supply market -0.160 0.221** -0.126 0.272*** potential) / mean (0.264) (0.079) (0.218) (0.092) ln(rural income opportunities: -0.517*** -0.059 -0.672*** -0.098** market potential) / mean (0.095) (0.035) (0.110) (0.035) ln(market potential) / mean 2.874*** 0.853*** 2.723*** 0.694*** (0.670) (0.159) (0.863) (0.130) Average schooling (t-1) 0.056*** 0.023* 0.093*** 0.040*** (0.020) (0.012) (0.019) (0.011) Average schooling 0.474*** 0.098*** 0.432*** 0.098*** (0.100) (0.032) (0.120) (0.033) ln(inter-city transport costs) -0.138** -0.089** -0.105** -0.077** (0.061) (0.040) (0.052) (0.035) state capital dummy 0.211*** 0.129*** 0.150*** 0.112*** (0.048) (0.036) (0.031) (0.030) ln(population) (t-1) -0.051*** -0.019* -0.060*** -0.026*** (0.009) (0.010) (0.009) (0.009) Manu / service (t-1) 0.129*** 0.101*** 0.076* 0.087*** (0.037) (0.018) (0.040) (0.022) ln(homicide / pop) (t-1) -0.150*** -0.093*** (0.035) (0.025) time dummies yes yes yes Yes Observations 246 246 245 245 R2 0.384 0.431 Hansen J statistic (overidentification test) 10.502 8.038 (p-value) (0.232) (0.430) Average of Partial R2 0.469 0.457 Average of Partial F's 499.0 1160.4 *** significant at 1% level; ** significant at 5% level; * significant at 10% level. 2.27 Decomposing City Growth: In Table 2.4, we decompose the city population growth results of Table 2.3, column 1 into contributions of each covariate. We focus on the covariates which are statistically significant. The contribution of each covariate is calculated as a fitted value (the mean value multiplied by the estimated coefficient) relative to the sum of all the fitted values. Column 5 shows the overall contributions for all cities. There is a strong negative effect of city size in base period (-24%). The key positive component to growth comes from increases in market potential (108%); much of what happens to cities is determined by conditions external to them--demand for their products as driven by what is evolving geographically around them. Changes in educational attainment (21%), along with base period's educational attainment (10%) which affects local technology growth also matter. Inputs to a Strategy for Brazilian Cities Page 27 Table 2.4 Decomposition of City Size Growth Coef. of ( ) Decomposition of city growth a Table 2.5 (1), Mean bi (ai ×bi / c), % (ai ) Total Large Small Large Small citiesb citiesb Total citiesb citiesb No. cities 123 61 62 ln(city pop) 0.226 0.264 0.188 ln(rural income -0.517 1.000 0.991 1.009 -19.3 -19.0 -19.7 opportunities) / mean ln(market potential) / 2.874 1.000 1.003 0.997 107.5 106.9 108.0 mean Average schooling (t-1) 0.056 4.568 4.773 4.366 9.6 9.9 9.2 Average schooling 0.474 1.208 1.215 1.201 21.4 21.4 21.5 ln(inter-city transport -0.138 -0.215 -0.191 -0.239 1.1 1.0 1.2 costs) State capital dummy 0.211 0.171 0.344 0.000 1.3 2.7 0.0 ln(population) (t-1) -0.051 12.339 13.172 11.520 -23.5 -24.9 -22.1 Manu / service (t-1) 0.129 0.406 0.428 0.385 2.0 2.0 1.9 c = a i×bi 2.674 2.695 2.654 i sum 100.0 100.0 100.0 a. Means are for 2000-1991 and 1991-1980. For average schooling (t-1), it is for 1991 and 1980. b. We define large (small) cities if they have greater (less) than median city population in each year. 2.28 The estimated effects of market potential and technology spillovers support the new economic geography emphasis on local markets and the endogenous growth literature emphasis on human capital accumulation. These results are also consistent with cross country findings in Henderson and Wang (2005).16 Columns 6 and 7 compare city growth decompositions of large versus small cities. We find no major difference in these effects across city size. 2.29 Decomposition of City Growth Residuals: We now use the residuals from the GMM estimations in Table 2.3, column 3, and examine if they have any systematic association with time invariant local characteristics. Our main interest is in examining if local management or governance, and inter industry linkages are associated with city growth. In principle, autonomous local government would actively work to provide local public goods for its constituents, and develop policies to stimulate growth and manage externalities. For our analysis, we have 16Henderson and Wang (2005) analyzes how urbanization in a country is accommodated by increases in numbers versus population sizes of cities. Using a worldwide dataset on all metro areas over 100,000 population from 1960-2000, they show market potential, educational attainment, and the degree of democratization strongly affect growth in both city numbers and individual city sizes. Inputs to a Strategy for Brazilian Cities Page 28 two measures of local government efforts: (1) existence of laws to collect property [IPTU] taxes and (2) percentage of population under land zone laws.17 Table 2.5 Regression of City Growth Residuals (Robust standard errors in parentheses) Dependent variable: Residuals of Table 5 (3) OLS OLS Laws to collect property tax 0.014 -0.086 (0.047) (0.055) % of population under 0.036* 0.031* land zone law (0.020) (0.017) Public industry capital / -0.666 -0.034 total industry capital in 1980 (0.504) (0.748) (No. formal firms / 0.955** No. workers in formal firms) (t-1) (0.387) time dummy Yes No Observations 245 122 R-squared 0.017 0.041 *** significant at 1% level; ** significant at 5% level; * significant at 10% level. 2.30 In terms of inter industry linkages; we expect a clustered or densely populated region to provide a rich environment for competition and collaboration among firms and workers in the region, which leads to economic growth. As Saxenian (1994) observes, regional development is more distinct in a region consisting of many small size, more competitive firms than that of a few large firms.18 We measure the density of economic activities by ln(no. firms relative to workers) = ln(no. formal firms / no. workers in formal firms). We also experimented with the ratio of public industrial to private industrial capital in 1980 (the only year we have it recorded) to see if cities which are more state capitalist are less efficient, or grow more slowly.19 2.31 The basic estimation results from decomposing the residuals are reported in Table 2.5. Due to the lack of longitudinal data for local characteristics, the estimation result should be interpreted as associations of contemporary variables rather than a causal relationship. Column 1 is for city growth residuals in (2000-1991) 17Those two measures are as of 1999. 18Saxenian (1994) examined different regional economic performances between Silicon Valley in California and Route 128 in Massachusetts. Dense social networks and open labor market in Silicon Valley have facilitated informal communication and collaborative practices, and produced a regional network-based industrial system. The Route 128 region, in contrast, is dominated by autarkic (self-sufficient) corporations that internalize a wide range of productive activities. She concluded that this difference in regional socio-economic structure accounts for the divergent prosperity of two regional economies, in spite of their common origins in postwar military spending and university-based research, and even though they enjoyed roughly the same employment levels in 1975. 19La Porta and López-de-Silanes (1999) showed privatization in Mexico in 1980s an 1990s led to a significant improvement in firm performance, as profitability increased 24 percentage points and converged to levels similar to those of private firms. Inputs to a Strategy for Brazilian Cities Page 29 and (1991-1980) regressed against time invariant city characteristics. Column 2 is for (2000-1991) adding to regressors a density measure of economic activities in 1991.20 2.32 We find that population growth is higher in cities with better enforcement of land use and zoning laws ­ the estimates suggest that city growth is associated with increases in the percentage of city population under land zone laws.21 However, we do not find any statistically significant association between city growth and existence of laws to collect IPTU (property tax). This is most likely because there is almost no variation in the IPTU collection data ­ most cities have laws to collect the property tax. A richer set of inter industry linkages is also associated with growth ­ the OLS coefficient for the number of (formal) firms relative to (formal) workers is statistically significant and has the expected sign. A higher number of firms relative to workers stimulate competition and collaboration among firms and workers in a city, and is associated with higher city growth. The coefficient of public industry capital ratio has a negative sign suggesting a detrimental effect on city growth. However, it is not statistically significant. Policies Favoring Secondary Cities 2.33 Using the results from the regressions of city growth, let us consider the following policy experiment. There is considerable policy debate in Brazil that investments need to be directed towards secondary cities to stimulate local economic development and limit the growth of the largest metropolitan areas. However, the impact of these initiatives on overall economic growth and urban efficiency is unclear. 2.34 Suppose the Brazilian government invests in transportation infrastructure in order to decrease inter-city transport costs. An issue is whether favoring investments in small cities vis-à-vis large cities increase overall productivity growth, and therefore higher overall economic growth in Brazil. To make the analysis tractable, we first assume that the amount of transportation investment to reduce one unit of inter-city transport cost (per mile) is proportional to city population. So one unit decease in inter-city transport costs for a city of 1 million is assumed to cost the same amount of government expenditure as those for 10 cities of 100,000 people. 2.35 In 2000, the largest city, São Paulo, has 17.9 million residents, which is equivalent to the total population of the 88 smallest cities (Table 2.6). The total population of the 7 largest cities is the same as that of remaining 116 small cities (Our data consist of 123 cities). Our assumption says that total transportation investment needed to decrease one unit of transport costs for São Paulo will also reduce one unit of transport costs for the 88 smallest cities, if invested in those cities. 20We only have 1991 and 2000 data for ln(no. formal firms / no. workers in formal firms). 21We can get a similar result when we use a dummy variable indicating more than 50% of population is under land zone laws. Inputs to a Strategy for Brazilian Cities Page 30 Table 2.6 Policy Simulation: Favoring Largest Cites Versus Smallest Ones (1 standard deviation decrease in inter-city transport costs in 2000) Total urban income relative to the baseline Comparison income (%) (b-a, %p) Favoring largest Favoring smallest cities (a) cities (b) 1 largest vs. 88 smallest 102.514 102.809 0.296 2 largest vs. 104 smallest 103.823 104.366 0.543 3 largest vs. 109 smallest 104.951 105.239 0.287 4 largest vs. 112 smallest 105.658 105.987 0.330 5 largest vs. 113 smallest 106.202 106.294 0.091 6 largest vs. 115 smallest 106.560 106.944 0.384 7 largest vs. 116 smallest 107.114 107.408 0.294 2.36 Table 2.1 describes the determinants of income per worker, in which average schooling, market potential, city population, and inter-city transport costs affect income per worker. From this equation, we can calculate the total urban income in Brazil, s. t. 123 total urban income = income per workeri × no. workersi i=1 123 X b^ i GMM × no. workersi. i=1 2.37 Now suppose the government invests in transportation infrastructure. In Table 2.6, we compare the effect on total urban income of investments favoring big cities versus small cities. The first column is the total urban income relative to the baseline income when infrastructure investments favor largest cities, specifically a 1 standard deviation (.8) decrease in inter-city unit transport cost of largest cities. The baseline income is the predicted value of Table 2.2 (3). The second column is the total urban income when the same amounts are invested in the smallest cities to decrease those cities' transport cost by the same magnitude (.8). We experiment with several combinations of cities in Table 8. 2.38 The simulation results show that there are very small differences in total urban income from favoring small cities vis-à-vis large cities. These income differences range around 0.1 ~ 0.5%p of total urban income growth in 2000. The difference is highest when we favor the 104 smallest cities vis-à-vis than the largest two cities (.543%p). These results tell that there are no major gains in terms of overall urban income from diverting investments from the largest cities to secondary cities. Summary of Findings 2.39 In this chapter, we examine the determinants of Brazilian city growth between 1970 and 2000. For the analysis, we construct a dataset of 123 agglomerations, and examine factors that influence wages and labor supply. Our main findings are the following: Increases in rural population supply is a major driver of city growth. Inter-regional transport improvements that lead to increases in the market potential of goods and reduce inter city transport costs stimulate growth. In fact, increases in market potential have the strongest impact on city growth. Inputs to a Strategy for Brazilian Cities Page 31 Improvements in labor force quality and the spillover effects of knowledge accumulation (measured by initial levels of education attainment) also have strong growth impacts. 2.40 In terms of inter regional transport improvements, the Brazilian government has made significant investments in infrastructure to integrate the national economy and lower business costs in peripheral regions. Most of the improvements in the road network occurred between the 1950s and 1980s, leading to significant reduction in transportation and logistics costs. Castro (2002) measures the benefits of improvements in highway infrastructure from 1970-1995 as the change in equivalent paved road distance from each municipality to the state capital of São Paulo, accounting for the construction of the network as well as the difference in vehicle operating costs between earth/gravel and paved roads. He shows that transport cost reductions were quite significant for the Northern region and Central region state of Mato Grosso, with numbers varying from 5,000 to 3,000 equivalent kilometers of paved road. Average reductions fall to the 1,000 km range in the Central region states of Goiás and Mato Grosso do Sul, the southern states, and the coastal northeastern states. Using this measure, Castro (2002) finds that the reduction in interregional transport costs was one of the major determinants of both the expansion of agricultural production to the central regions of Brazil after the 1960s as well as increases in the country's agricultural productivity. 2.41 Moving from regional initiatives to city level characteristics and innovations, our main findings are that local land use and zoning enforcement is positively associated with city growth, where as over-accumulation of public industrial capital which appears to crowd out private capital is negatively associated with growth. City growth is lower in cities which have high homicide rates. Appendix A. Market potential measures (1) Basic Market Potential Market potential of agglomeration i is defined as the sum of its member MCAs' market potential. Therefore the market potential of agglomeration i in year t is 3659 yj (t)× popj (t) -1 . kii j=1 ( Ad ) ki , j where yj t is per capita income of MCA j in year t, and popj t population of MCA j in year t. di ( ) ( ) is the , j distance between MCA i and j (100 miles). The distance of own MCA di ( )is the average distance to city ,i 2 area center, which is equal to . is assumed to be 2, is 0.3 (0.22 between two port cities), and A is such 3 that Adi,j =1 for the smallest land area city (Au and Henderson, 2004; Hummels, 2001). 0.3 (2) Incomes offered in local rural areas competing with own city for local population The gravity measure of surrounding rural per capita incomes is a market potential measure of agglomeration i in year t , such that rural3659 GDPj (t)/ rural popj (t) . kii j=1 ji ( Ad ) -1 ki , j Inputs to a Strategy for Brazilian Cities Page 32 The MP calculation does not include the rural per capita MCA incomes of the same agglomeration. All parameters are the same as (1). Rural GDPs of (1970, 1980, 1985, and 1996) are assigned to those of (1970, 1980, 1991, and 2000). (3) Potential supply of people to the city from local rural areas The gravity measure of surrounding rural population is also a market potential measure of agglomeration i in year t , such that rural 3659 popj (t) . kii j=1 ji ( Ad ) -1 ki , j The MP calculation is the same as (2). (4) Market potential measure of agricultural land availability The agricultural land market potential is calculated in the same way as (1), such that agri 3659 land j (t) -1 kii j=1( Ad ) ki , j where agri land j t is agricultural area of MCA j in year t. All parameters are the same as previous ones. ( ) Inputs to a Strategy for Brazilian Cities Page 33 3. Urban Policies and Slum Formation by Somik Lall 3.1 Relevant Issues: What is the impact of policy interventions on household welfare, especially among slum dwellers? How effective are the urban poverty-targeted programs? 3.2 An estimated one-third of all urban residents live in informal settlements or slums ­ the vast majority in developing countries. Globally, almost one billion people live in slums (United Nations, 2003). In some developing country cities, one-half or more residents live in these inadequate settlements. Conditions in such areas vary widely from dismal, temporary shelter in squatter settlements to relatively well-constructed, informal housing that may persist for many decades. Common characteristics include uncertain tenure status, poor basic services such as water and sanitation, low-grade construction and overcrowded living conditions. Apart from physical deprivation, slum dwellers also often face more subtle disadvantages such as poor labor market integration and the social stigma attached to an inferior residential location. Children living in slums are deprived of access to good quality education and health services, which are not located in reasonably proximity of these settlements. 3.3 With continuing rapid growth of urban areas, improving the life of slum dwellers is a high priority for national and city governments and the international community. The Millennium Development Goals, for instance, advocate significant improvements in the lives of at least 100 million slum dwellers by 2020 (United Nations, 2005). However, at this time there is very little evidence on factors that influence the formation of slums and identifying the local and national policy environments that influence the capacity of cities to manage the growth of slums. In particular, it is essential to know if and how public policy can influence slum formation. The focus in this chapter is on the link between informal settlements (slums) and the supply response of the formal housing market. The proposed argument is that slum formation increases when housing supply in the formal housing market is inelastic and cannot accommodate increases in overall housing demand. 3.4 Across developing countries, housing demand has a number of regularities, which vary with income within and across countries in predictable ways (Malpezzi and Mayo, 1987). However, housing supply has been understudied so far and there is no firm consensus on the price elasticity of housing supply and the nature of housing supply (Green, Malpezzi, and Mayo, 2005). Recent research finds that housing supply elasticity varies significantly across cities within a country and across countries, and that these differences are mainly from restrictive zoning and other land use regulations (Saks, 2005; Glaeser, Gyourko, and Saks, 2005a,b; Green, Malpezzi, and Mayo, 2005; Quigley and Raphael, 2005). Most existing empirical studies are focused on developed countries, particularly the United States, and therefore market clearing is implicitly assumed: housing prices and housing stock adjust to external shocks, and housing market is in equilibrium. However, this market clearing assumption cannot be held in many developing countries where the capacity of the formal housing market is so limited that the urban poor and even middle income households resort to informal housing solutions. 3.5 We distinguish formal and informal housing sectors. When the formal housing sector in a city cannot provide enough housing stock due to various bottlenecks, city residents will solve housing demand from the informal sector and slums grow. In this regard, the price elasticity of housing supply measures the capacity of the formal housing sector to absorb urban migrants into the system. Inelastic housing supply limits housing stock adjustment in response to urban migration and expansion, and therefore bring about massive slum dwellings. We develop a slum formation model which accounts for this housing market disequilibrium. This model provides a framework for the empirical work. We estimate a reduced form equation for the formal housing market, and then a slum growth equation. By extending the slum growth equation, we examine the effects of local characteristics and local land use regulations on housing supply elasticity and slum growth in cities. Inputs to a Strategy for Brazilian Cities Page 34 For the empirical analysis, we construct a dataset of Brazilian agglomerations from 1980 to 2000. Much of the underlying data come from the Brazilian Bureau of Statistics (IBGE) Population Census of 1980, 1991, and 2000. For the estimation, we make use of OLS and nonlinear OLS estimations and report robust standard errors when residuals are tested to be heteroskedastic. Our main findings are that the imputed price elasticity of housing supply in the Brazilian formal housing market is very inelastic, and therefore it serves as a major determinant of slum growth. In terms of local characteristics, we find that a city has inelastic housing supply and a higher slum growth rate, if a city (1) is in a semi-arid area, (2) has a centralized urban form where city population is concentrated in city centers rather than dispersed across city areas, (3) has higher transport / commuting costs, and (4) has minimum lot size regulations and land zoning laws which are supposed to be restrictive and excessively regulate urban land use. Similar to previous studies on the housing market in the United States, we draw the conclusion from Brazil that heavily regulated cities exhibit low housing supply elasticities. In addition, we find that the identified local factors, which are supposed to reduce housing supply elasticity, increases the growth of slums. It confirms our proposed linkage between housing supply elasticity and slum growth in cities. To the best of our knowledge, this is the first attempt at modeling slum growth in relation to housing market and examining the effects of local characteristics on housing supply elasticity and slum growth. Slum Formation Across Cities 3.6 The urbanization and industrialization processes in Brazil began in the 1930s. As coffee prices were severely depressed after the Great Depression, the lack of foreign exchange from coffee export restricted the imports of industrial products and created new domestic demand for industrial production and urban labor. This demand for urban labor and the coffee driven agricultural collapse spurred massive urban migration.22 However, the urbanization process has been largely uncontrolled.23 The access to urban land and housing has relied mostly on (1) the division of central and peripheral land (loteamentos) and (2) the invasion of public and private urban land (favelas). As a result, in main cities modern central areas are surrounded by irregular and illegal settlements which lack in drainage and sewerage systems, health and education facilities, and green spaces. Public transport is insufficient and expensive, and the quality of life in slums is very low. 3.7 There are three forms of slums in Brazil: Favelas, cortiços, and irregular/clandestine loteamentos. Favelas are the most popular form of urban slums in Brazil.24 They are precarious human settlements resulting from the invasion of both public and private urban areas. Those invaded areas are generally located close to city centers, but mostly unsuitable for human occupation due to geographical and ecological factors. They lack in almost every element of urban infrastructure and collective equipment. 3.8 Cortiços are high density collective housing in city centers. They are old and subdivided into small rooms with many fire and explosive hazards, few bathrooms, no formal rental relationships, no proof of payments, and often run by intermediaries connected to the police and criminals (Saule 1999). Irregular and clandestine loteamentos are usually developed in peripheral areas irregularly, if not also illegally. Irregular land divisions are in precarious technical conditions, and not registered in the public registry office. Loteamentos developed in areas of contested ownership are called "clandestine." They differ from favelas, since the occupiers have bought their plots from whoever presented themselves as landowners, and in most of cases paid all due taxes. 3.9 We review overall slum formation in relation to city growth. The slum data are from the Brazilian Bureau of Statistics (IBGE) Population Censuses of 1980, 1991, and 2000.25 It defines "subnormal 22Krueckeberg and Paulsen (2000). 23Prior to the promulgation of the 1988 Constitution there was no adequate legal planning framework and corresponding institutional apparatus in force (Fernandes, 1997). 24The name comes from a mountain (Morro de Favela) in the center of Rio de Janeiro, occupied by squatters in 1906. 25Throughout the paper, we will refer the reader to Appendix B for a discussion of data sources and variable definitions. Inputs to a Strategy for Brazilian Cities Page 35 agglomeration" as a set of (slums or related) houses that occupies other people's land (either public or private) and that, by and large, are structured in a disordered and densely way and that lack public services and utilities. The sample consists of 123 agglomerations, which includes 447 MCAs (Minimum Comparable Areas). In the empirical analysis, we focus on 72 agglomerations (335 MCAs) where slum data are observed for the sample period. Table 3.1 shows that the largest cities have higher absolute levels of slum formation in 2000. The four largest cities have about 9.1% of their residents living in sub standard dwellings. Slum formation is lower as one goes down the urban size distribution. 3.10 Figure 3.1 displays the ten fastest slum growth cities, which tend to be located along the coastline. This suggests that local and regional characteristics may influence city slum growth. Interestingly, Figure 3.2 and corresponding OLS regression show no statistically significant relationship between slum growth and city population growth. All these suggest that slum formation is a complicated process influenced by various city characteristics, rather than simply proportional to city size growth itself. Table 3.1: Slum Formation Across City Sizes, 2000 no. no. slum no. housing no. formal city size no. population dwellers Slum share houses / no. cities (a), (b), (b/a, %) stock, total houses 1000 1000 1000 (%) Largest (5.301 pop/mean) 4 39,095.6 3,566.4 9.12 11,409.6 91.78 (1.340 pop/mean < 5.301) 11 25,260.3 1,583.4 6.27 6,786.9 94.18 (0.812 pop/mean < 1.340) 14 11,682.8 302.5 2.59 3,160.4 97.62 (0.469 pop/mean < 0.812) 17 7,699.9 154.7 2.01 2,098.7 98.20 (0.256 pop/mean < 0.469) 20 5,879.1 81.1 1.38 1,624.3 98.80 Smallest (pop/mean < 0.256) 57 7,333.5 87.8 1.20 2,046.8 98.86 Total 123 96,951.3 5,775.9 5.96 27,126.6 94.51 Figure 3.1 Cities with the Fastest Slum Formation between 1980 and 2000 Inputs to a Strategy for Brazilian Cities Page 36 Figure 3.2 Slum Dweller Growth and City Population Growth between 1980 and 2000 gr_nslum vs. gr_n .2 .1 muls _n gr 0 1-. 0 .02 .04 .06 pop growth rate, annual:1980-2000 slum dweller growth rate, annual:1980-2000 Fitted values OLS: ln slum _ dwellers2000 (1/ 20) (1/ 20) = -2.11*ln + 0.109 slum _ dwellers1980 pop2000 pop1980 (-1.45) (2.92) N=35, adj R2 = 0.031, t-values in parentheses. 3.11 We propose that slums grow when the capacity of the formal housing market is limited and cannot accommodate an increase in overall housing demand. The capacity of the formal housing market or the housing supply adjustment in response to housing demand changes can be represented by the price elasticity of housing supply. Then a corollary of our proposition is that slum growth rates will differ across cities as cities have heterogeneous housing supply elasticities due to different local backdrops and housing supply bottlenecks. 3.12 Figure 3.3 and corresponding OLS result show a statistically significant and negative relationship between the slum growth and the growth of formal housing stock for 1980-2000.26 For example, the two cities in the bottom right of Figure 3.3 are Cuiabá and Campo Grande. These two cities successfully increased formal housing stock at the annual growth rates of 5.9% and 5.6% respectively between 1980 and 2000, and were able to manage slum formation (slum growth rates are -3.5% and -6.9% annually). 26Formal housing stock is defined by the difference between the number of total housing units and the number of slum units. Inputs to a Strategy for Brazilian Cities Page 37 Figure 3.3 Slum Dweller Growth and Formal Housing Stock Growth between 1980 and 2000 gr_nslum vs. gr_fh .2 .1 muls _n gr 0 1-. .02 .04 .06 .08 formal house growth rate, annual:1980-2000 slum dweller growth rate, annual:1980-2000 Fitted values OLS: ln slum _ dwellers2000 (1/ 20) (1/ 20) = -2.99*ln + 0.163 slum _ dwellers1980 formal _houses2000 formal _ houses1980 (-2.32) (3.48) N=35, adj R2 = 0.114, t-values in parentheses. Housing Supply and Slum Formation 3.13 We postulate that slums are a response to close the gap between overall housing demand and formal housing supply of the city. If formal housing supply is elastic, a rise in housing demand will be accommodated by an increase in housing supply without significant rise in housing price. However, if housing supply is inelastic, housing demand increase cannot be met by sizable housing supply increase and therefore housing price will rise significantly. It forces city residents to resort to informal housing solutions and slums grow. 3.14 The slum formation model in relation to the housing market is formulated as follows. The share of people living in slums of city i in year t can be approximated such that slum _ shareit = (Nit - FNit )/ Nit ln Nit - ln FNit . (1)27 After solving for ln FNit and substituting we can get the following equation: 27slum _ shareit = (Nit - FNit )/ Nit =1- FNit / Nit -ln(FNit / Nit )= ln Nit - ln FNit . Inputs to a Strategy for Brazilian Cities Page 38 slum _ shareit = 01 +10 2 1 +1 (2) 31 + 3 lnYit - 31 ln FHit + ln Nit . 3.15 In the model, slums are created to close the gap between the overall housing demand and the formal housing supply in a city. This specification is also consistent with a general agreement that, in principle, city (population) growth could be completely determined by other variables, with the housing supply simply responding to those factors.28 Slum growth is an outcome of city population growth and housing supply adjustment. By log transformation and simple manipulation, we get the following slum formation equation.29 ln(slum_ shareit )= + 0b1 + ln -1 + 0 1 2 1 3 1 3 1 lnYit - + b1 ln FHit + 1 1 ln Nit (3) , where b1 1/ 1. An advantage of this transformation from eq. (4) to eq. (5) is that we can directly estimate key parameter values. The demand and supply shift terms of the formal housing market (0,0 ) may vary across years. With a simple approximation such that 0 =0 +0 t and 0 = 0 + 0 t in year t, and first differencing, we can obtain a formal housing stock growth equation and a slum growth equation. lnFHit = 01 +10 21 31 1 +1 +1 +1 lnYit + 1 +1 lnFNit (2') ln(slum_shareit)= + 0b1+ 0 2 1 3 1 1 lnYit - +b1lnFHit + 1 1 lnNit (5') Heterogeneous Housing Supply Elasticities 28Glaeser, Gyourko, and Saks (2005b). 29By substituting b1 1/ 1 , we can get. slum _ shareit = 0 10 2 lnYit - + 1 1 b1 ln FHit + ln Nit . Or 3 + 3 b1 + 3 3 3 slum _ shareit = 1 0 2 1 3 1 3 ln Nit f (b1). 1 + 0b1 + 1 lnYit - + b1 ln FHit + 1 1 3 Since ln slum _ shareit = ln ( ) 1 +ln ( f (b1)) ( ( )) f (b1)-1, 3 and ln f b1 ln(slum_ shareit )= + 0b1 + ln -1 + 0 1 2 1 3 1 3 1 lnYit - + b1 ln FHit + 1 1 ln Nit. Inputs to a Strategy for Brazilian Cities Page 39 3.16 The price elasticity of housing supply is reported to vary significantly across cities within a country and across countries, as listed above. The effects of different housing supply elasticities on housing supply adjustment and slum growth are illustrated in Figure 3.4. Suppose two cities which are identical except for different housing supply elasticities: (a) a city of low price elasticity of housing supply and (b) a city of high price elasticity of housing supply. And assume that there is the same population growth due to migration N .30 The ( ) population growth increases overall housing demand and therefore pushes up housing prices. Suppose two cities have the same housing price increase P . Even though two cities show the same population growth and ( ) housing price increase, the housing supply adjustment and slum growth will differ between two cities due to different housing supply elasticities. Figure 3.4 Housing Supply Elasticity and Slum Formation FH0 D FH1 D FHie S D FH1 D P P FH0 S FNie FNe FHe P0+P P0 FHie FHie+FHie FH FHe FHe+FHe FH (a) a city of inelastic housing supply (b) a city of elastic housing supply 3.17 For the city of inelastic housing supply, the supply adjustment responding to housing price increase P ( ) is limited and only accommodate a small fraction of population growth FNie . However, the city of elastic ( ) housing supply responds to the same housing price rise by a significant increase in housing stock, and therefore accommodate more city population FNe > FNie . Since N = FN + slums , the slum growth in the city ( ) of elastic housing supply is lower than that of inelastic housing supply slumse < slumsie . ( ) 3.18 As illustrated in Figure 3.4, the price elasticity of housing supply may differ across cities in year t such that 1 = 1 . We postulate that the price elasticity of housing supply in city i in year t 1 1/b1 ( ) ,it ,it ,it is 30As discussed above, in principle, city (population) growth could be completely determined by other variables, with the housing supply simply responding to those factors. Inputs to a Strategy for Brazilian Cities Page 40 influenced by various local characteristics in the base year including local regulatory and policy environments. We approximate it by 1 1 b1 =0 + ,it kXk,i(t-1 . ) (6) ,it k , where Xk ,i(t-1)are city characteristics in the base year (t-1) which influence housing supply adjustment. Substituting eq. (6) into eq. (5'), we can get a reduced form slum growth equation incorporating local characteristics which affect housing supply adjustment. ln(slum_shareit)= 0 ( ) 1 + 0 0 + 0 k Xk,i(t-1 +2 lnYit ) k 1 (7) - +0 lnFHit - k Xk 1 ( ) 3 1 ,i(t-1)lnFHit + lnNit. k 1 3.19 An advantage of this transformation is that from eq. (7) we can directly estimate parameters of 0 and k by nonlinear OLS estimation. In this way, we can examine the effects of specific local characteristics on local housing supply elasticity and slum growth. 3.20 If the parameter estimate of k is significantly negative, it suggests that the local characteristics Xk ,i(t-1) increases (formal) housing supply adjustment and therefore reduces slum growth. In the same way, significantly positive parameter estimate of k suggests that the local characteristics Xk ,i(t-1) is a main bottleneck to elastic housing supply adjustment and therefore a major determinant of slum growth. Local policy priority to manage slum growth should be focused on eliminating major bottlenecks and improving the housing supply capacity of cities to absorb the city poor into formal housing markets and provide them better tenure status, better basic services such as water and sanitation, better labor market integration, and higher quality of life. 3.21 For the city of inelastic housing supply, the supply adjustment responding to housing price increase P ( ) is limited and only accommodate a small fraction of population growth FNie . However, the city of elastic ( ) housing supply responds to the same housing price rise by a significant increase in housing stock, and therefore accommodate more city population FNe > FNie . Since N = FN + slums , the slum growth in the city ( ) of elastic housing supply is lower than that of inelastic housing supply slumse < slumsie . ( ) Findings from Empirical Analysis 3.22 Table 3.2 is the results from estimating the formal housing stock growth equation. It assumes the price elasticity of housing supply is the same across Brazilian cities: homogeneous housing supply elasticity. It pools two periods of formal housing stock growths (2000-1991 and 1991-1980), and reports OLS results with Breusch- Pagan /Cook-Weisberg test for heteroskedasticity. We cannot reject the null hypothesis of homoskedasticity. Column 1 is for all the cities in the sample (123 cities), and column 2 is a subset of 72 cities where we observe slums data. We focus on column 2 results for comparison with the slum growth model results which use a sample of 72 cities. All coefficient estimates in column 2 are significant and have expected signs. Growths in income per capita and city population in formal houses increase the growth rate of formal housing stock in a city. Inputs to a Strategy for Brazilian Cities Page 41 Table 3.2 Formal Housing Stock Growth Equation Dependent variable: (1) (2) ln(formal housing stock) OLS OLS ln(income per capita) 0.020 0.051*** (0.016) (0.020) ln(no. people in formal Houses) 0.958*** 0.948*** (0.017) (0.019) time dummies Yes Yes Observations 246 144 (no. cities) (123 cities) (72 cities) Adjusted R2 0.945 0.956 Test for heteroskedasticitya 2(1) 1.60 0.99 (p-value) (0.206) (0.321) *** significant at 1% level; ** significant at 5% level; * significant at 10% level. 3.23 As illustrated in Figure 3.4, if a city's housing supply is elastic, an outward shift in housing demand results in a larger increase in housing stock and a relatively modest housing price increase. However, if housing supply is inelastic, we expect a shortage of housing stock and housing price will rise significantly. In this regard, the level of price elasticity of housing supply has an important policy implication. However, we cannot directly measure the housing supply elasticity due to a standard identification problem.31 Malpezzi and Mayo (1997) and Malpezzi and Macleannan (2001) solved this problem by assuming the housing demand elasticities to be in a certain range. 3.24 Malpezzi and Mayo (1997) suggested that reasonable bounds for the price elasticity of housing demand would be between -0.5 and -1.0, and the long-run income elasticity between 1.0 and 1.5. Malpezzi and Macleannan (2001) also proposed similar bounds for the United States and the United Kingdom: between -0.5 and -1.0 for the price elasticity of demand, and between 0.5 and 1.0 for the long-run income elasticity of demand. Table 5 calculates the imputed price elasticity of housing supply (1) based on coefficient estimates. The calculation is from a range of assumptions about housing demand elasticities (1,a2 ) mentioned above. We assume the price elasticity of housing demand (1) to be between -0.5 and -1.0, and the income elasticity of housing demand (2 ) between 0.5 and 1.5. The imputed price elasticity of housing supply based on the coefficient estimates of column 2 in Table 3.2 is reported in column 3. It turns out to be very inelastic ranging between 0.02 and 0.1. Inelastic housing supply suggests that the formal housing market in Brazil could not respond to outward shifts of housing demand from urban migration and created massive slum dwellings in cities. 3.25 When comparing other countries in column 5 and afterwards, the imputed Brazilian housing supply elasticity is similar to those in Malaysia and South Korea, which were regarded to have restrictive regulatory environments. It is quite different from the elastic housing supply of the United States where housing supply 31In eq. (2'), we have 4 parameters (1,a2,3,1) and 2 coefficient estimates 21 , 1 + 1 1 + 1 31 . Inputs to a Strategy for Brazilian Cities Page 42 elasticity ranges between 6 and 19. However, this interpretation has its limitation that the housing supply in Brazil is measured by the number of houses rather than the amount of housing services including quality adjustment. 3.26 Results for slum growth estimation are reported in column 1 of Table 3.3. Again, for the estimation we pool two periods (2000-1991 and 1991-1980) and focus on 72 cities where we observe slum data. Since the test for heteroskedasticity shows OLS residuals are not homoskedastic, we report robust standard errors which allow for differences in the variance / standard errors due to arbitrary within-region correlation.32 We assume the observations may be correlated within five regions, but would be independent between regions.33 As shown in da Mata et al. (2005), clustered estimation results are also robust to residual spatial dependence.34 We use the same error specification for all slum growth estimations. Table 3.3 Slum Growth Equation (robust standard errors in parentheses) Dependent variable: (1) (2) (3) ln(share of people in slum dwellings) OLS Nonlinear OLS Nonlinear OLS ln(income per capita) 0.022** 0.019** 0.018** (0.008) (0.006) (0.005) ln(no. formal houses) -0.513*** -0.505*** -0.498*** (0.097) (0.105) (0.109) ln(city population) 0.480*** 0.477*** 0.476*** (0.089) (0.095) (0.095) k : Dummy for semi arid area 0.124*** 0.127*** (0.024) (0.017) Std. dev. / mean of 0.014 altitude (0.048) 0 0.315*** 0.318*** (0.013) (0.011) time dummies Yes Yes Yes Observations 144 144 144 R2 0.608 0.621 0.621 *** significant at 1% level; ** significant at 5% level; * significant at 10% level. 32Breusch-Pagan / Cook-Weisberg test for heteroskedasticity rejects the null hypothesis of constant variance ( 2(1)=59.59, p-value = 0.00). 33Five regions are North, Northeast, Southeast, South, and Central-West regions. 34da Mata et al. (2005) finds that clustered GMM and OLS results are very similar to two-step spatial GMM and spatial OLS ones which were developed by Conley (1999). Inputs to a Strategy for Brazilian Cities Page 43 3.27 All variables in column 1 of Table 3.3 have strong and expected sign coefficients. Slum growth is positively related with the growth of a city's total income per capita and city population growth, and negatively related with the growth of formal housing stock. Income growth increases housing demand, and assuming upward sloping housing supply curve it raises housing price. As raised housing prices are not affordable to some urban poor households, they resort to informal housing solutions. However, this effect is not strong: a 10% increase in per capita income raises slum growth by 0.2 % over a decade. Both city population growth and growth of formal housing stock have fairly strong effects on slum growth rates in similar magnitude: a 10% increase in formal housing stock decreases slum growth by 5% over a decade, and a 10% increase in city population raises slum growth by 5% over a decade. 3.28 The imputed price elasticity of housing supply (1), which is calculated based on coefficient estimates of column 1 of Table 3.3, is listed in column 4 of Table 5. The calculation is also based on a range of assumptions about housing demand elasticities, mentioned above. The imputed housing supply elasticities range between 0.47 and 0.50, which are higher than those from the formal housing stock growth estimation, but still very inelastic. Two results confirm that the price elasticity of housing supply in Brazil is very inelastic and therefore causing a major bottleneck in housing stock increases in response to urban migration. Slum growth is a natural outcome of this inability of the formal housing sector to increase housing supply. 3.29 Inelastic housing supply adjustment in response to urban migration brings about a shortage of housing units and the urban poor will resort to informal houses in order to solve housing problem. The elasticity of housing supply in a city or the capacity of a city to increase housing stock in response to housing demand shifts will be determined by various city characteristics, such as geographical constraints, densification and centralization of city population and economic activities within a city, infrastructure provisions including transportation systems, and local land use regulations. 3.30 In Table 3.3 and Table 3.4, we examine how various city characteristics influence the capacity of cities to increase housing supply in response to city population increase and therefore the capacity of cities to manage slum growth. Columns 2 and 3 in Table 3.3 examine the effects of geo-climate constraints on housing supply adjustment and slum growth. A dummy for semi-arid areas has a statistically significant and positive coefficient, suggesting cities in semi-arid areas have lower housing supply elasticities and higher slum growth rates. Provision of serviced land with water and sanitation will be difficult in semi-arid areas, and therefore restrict housing supply expansion. Availability of land for housing will also be limited, if a city is located in hilly areas. We do not have an exact measure of geographical steepness or variation in altitude within a city, but can obtain a proxy by calculating the coefficient of variation of average altitudes in MCAs within an agglomeration.35 However, the estimate in column 3 is not significant. It may be because our proxy variable poorly measures the actual land availability. 3.31 In Table 3.4, we examine the effects of city characteristics in detail. First, we look at slum growth in a context of residential location decision. A standard urban residential choice and spatial equilibrium model in monocentric cities predicts higher transport costs push people to live close to city centers where they commute to work. High concentration of city population around city centers limits land availability for housing, and therefore will cause inelastic housing supply adjustment. That negative effects can be alleviated, if jobs are decentralized from city centers, or if a city is polycentric rather than monocentric. 35The coefficient of variation in a distribution is defined as its standard deviation divided by its mean. Inputs to a Strategy for Brazilian Cities Page 44 Table 3.4 Slum Growth Equation (II) (robust standard errors in parentheses) Dependent variable: (1) (2) (3) ln(share of people in slum dwellings) Nonlinear OLS Nonlinear OLS Nonlinear OLS ln(income per capita) 0.014* 0.017** 0.016** (0.006) (0.005) (0.006) ln(no. formal houses) -0.340** -0.343** -0.318** (0.113) (0.103) (0.104) ln(city population) 0.481*** 0.493*** 0.496*** (0.076) (0.073) (0.068) k : semi arid area dummy 0.137*** 0.119*** 0.116*** (0.025) (0.017) (0.015) Std. dev. / mean of -0.031 -0.040* -0.042** Population in a city (t-1) (0.018) (0.018) (0.016) ln(inter-city transport 0.232*** 0.247*** 0.220*** costs, t-1) (0.021) (0.020) (0.036) state capital dummy 0.221*** 0.233*** 0.201*** (0.034) (0.025) (0.037) % of population under 0.067** 0.063** minimum lot size regulationsb (0.017) (0.022) % of population under 0.066 land zoning laws (0.035) 0 0.333*** 0.339*** 0.338*** (0.015) (0.017) (0.016) time dummies Yes Yes Yes Observations 144 144 144 R2 0.655 0.662 0.667 *** significant at 1% level; ** significant at 5% level; * significant at 10% level. 3.32 We measure transport costs per a unit distance by the variable "inter-city unit transport costs" which is the transport cost from each city to its state capital divided by distance between two cities. Even though it is not an Inputs to a Strategy for Brazilian Cities Page 45 exact measure of "inner-city" unit transport costs, it represents the quality of transportation infrastructure in a city and therefore can be a proxy for inner-city transport costs. Since we only have 1968, 1980, and 1995 transport cost data, we use 1980 values for slum growth for 1991-1980, and use 1995 values for 2000-1991. We give zero values to ln(inter-city transport costs) of state capital cities and insert a dummy for state capitals. We measure the city population decentralization by the coefficient of variation of city population, again using MCA level populations within an agglomeration. A higher value represents more job decentralization within a city. 3.33 Column 1 in Table 3.4 indeed shows the patterns we propose. Most of variables are significant and have expected signs: only the coefficient of variation of city population is not significant by a narrow margin, but do have an expected sign.36 However, as shown later, when we control for other city attributes, it becomes significant. Inter-city transport costs in base year reduce housing supply elasticity and increase slum growth rate. Slum growth can be reduced if a city has a more dispersed population distribution. Columns 2 and 3 in Table 3.4 examine the effect of land use regulations on housing supply elasticity and slum growth. We use (i) percentage of city population under minimum lot size regulations below 125m2, (ii) percentage of city population under urban land zoning laws. Minimum lot size regulations restrict the construction of substandard small size housing units which are mainly for the poor, and therefore limit housing supply adjustment and slums grow. Land zoning regulation itself may not bring about inelastic housing supply, if well planned and efficiently implemented. However, in most of developing countries, zoning regulations are too restrictive and rigid to incorporate ever-changing city dynamics, in particular, during massive urban migration. Ill- maintained excessive zoning regulations will limit efficient use of urban land which needs redevelopment and recycling on a regular basis. In this respect, we expect a positive coefficient estimate. 3.34 We use land use regulations as of 1999. Since municipalities enacted land zoning laws and parcel laws, where minimum lot size regulations are stipulated, way back from 1979, we assume those two variables have a historical perspective influencing housing supply adjustment in our sample period. The coefficient estimates of the minimum lot size regulation variables in columns 2 and 3 have statistically significant and positive signs, suggesting a negative effect of minimum lot size regulations on housing supply elasticity and therefore it increases slum growth rate. The coefficient estimate of percentage of population under urban land zoning laws has an expected positive sign, but fails to be significant at a narrow margin (p-value is 0.13). Summary 3.35 Our main findings are the following: The price elasticity of housing supply in the Brazilian formal housing market is very low. Inelastic housing supply significantly limits housing supply adjustment in response to housing demand increases, and therefore slums grow. Natural constraints such as semi-arid areas limits a city's capacity to increase housing supply and raises slum growth rate. A standard residential choice model in monocentric cities explains how urban forms and transport costs influence housing supply response and slum formation: high transport costs and large concentration of city population in city centers reduce housing supply expansion and slums grow. Urban land use regulations have strong negative effects on housing supply elasticity, and therefore the slum growth is higher in highly regulated cities. 3.36 These findings have significant policy implications for local governments and urban planning authorities, as improving the life of slum dwellers and managing slum growth become high priorities for national and city governments and the international community. Local government's role should be as facilitator rather than 36P-value is 0.16. Inputs to a Strategy for Brazilian Cities Page 46 provider, encouraging development and involvement of the private sector in the housing provision. Local governments need to overhaul and reform inappropriate, excessively detailed and inflexible regulatory and legal frameworks. Existing urban planning strategies and regulations should also be reviewed to utilize the importance of urban form and transportation infrastructure in relation to housing supply elasticity and slum growth. Inputs to a Strategy for Brazilian Cities Page 47 References Alesina, A. and D. Rodrik (1994), "Distribution Politics and Economic Growth," The Quarterly Journal of Economics, 109, 456-490 Angel, S. (2000). Housing Policy Matters: A Global Analysis, Oxford University Press. Beeson, P., D. DeJong and W. Troesken (2001), "Population Growth in U.S. Counties, 1840-1990," Regional Science and Urban Economics, 31, 669-699. Barro, R.J. and Sala-i-Martin, X. (1995). Economic Growth. New York: McGraw-Hill. Black, D. and V. Henderson (2003), Urban Evolution in the USA, Journal of Economic Geography, 3, 343-372. Blanchard, O.J. and L.F. Katz (1992), Regional Evolutions. Brookings Papers on Economic Activity, 0, 1:1-61. Conley, T. (1999), "GMM Estimation with Cross Sectional Dependence," Journal of Econometrics, 92, 1-45. Da Mata, D., U. Deichmann, V. Henderson, S. Lall, and H. Wang (2005). Examining the Growth Patterns of Brazilian Cities. Mimeo. Duranton, G. and D. Puga (2004), "Micro-Foundations of Urban Agglomeration Economies," in J. V. Henderson and J.F. Thisse (eds.) Handbook of Regional and Urban Economics, Vol 4. North-Holland. Eaton, J. and Z. Eckstein, (1997). Cities and growth: Theory and evidence from France and Japan, Regional Science and Urban Economics, Elsevier, vol. 27(4-5), pages 443-474. Fernandes, E. (1997), "Access to Urban Land and Housing in Brazil: Three Degrees of Illegality," Lincoln Institute of Land Policy Working Paper #WP97EF1. Ellison, G. and E. Glaeser (1997), Geographic concentration in US manufacturing industries: a dartboard approach, Journal of Political Economy, 105:5, 889-927. Galor, O. and J. Zeira (1993), "Income Distribution and Macroeconomics," Review of Economic Studies, 60, 35- 52. Glaeser E., J. Scheinkman and A. Shleifer (1995), "Economic Growth in a Cross-Section of Cities," Journal of Monetary Economics, 36, 117-143. Glaeser, E., M. Kahn, and J. Rappaport (2000). "Why Do the Poor Live in Cities?" NBER Working Paper #7636. Glaeser, E. and J. Gyourko (2002), "The Impact of Zoning on Housing Affordability," NBER Working Paper #8835. Glaeser, E., J. Gyourko, and R. Saks (2005a). "Urban Growth and Housing Supply," NBER Working Paper #11097. Glaeser, E., J. Gyourko, and R. Saks (2005b), "Why Have Housing Prices Gone Up?" The American Economic Review, 95, AEA Papers and Proceedings, 329-333. Green, R., S. Malpezzi, and S. Mayo (2005), "Metropolitan-Specific Estimates of the Price Elasticity of Supply of Housing, and Their Sources," The American Economic Review, 95, AEA Papers and Proceedings, 334-339. Harris Dobkins, L. and Y. Ioannides (2001), Spatial interactions among U.S. cities: 1900-1990, Regional Science and Urban Economics, 3, 6:701-731. Henderson, J. V. and H.G. Wang (2003), Urbanization and City Growth, Brown University, mimeo, http://www.econ.brown.edu/faculty/henderson/papers/Urbanization_and_City_Growth0803.pdf. Henderson, J.V. and C.C. Au (2004), How Migration Restrictions Limit Agglomeration and Productivity in China, Journal of Economic Development, forthcoming. Inputs to a Strategy for Brazilian Cities Page 48 Henderson, J.V., T. Lee, and Y.-J. Lee (2001 ) Scale Externalities in a Developing Country, Journal of Urban Economics, 49, 479-504. Henderson, V., Kuncoro, A., and M. Turner (1995), Industrial development in cities. Journal of Political Economy, 103, 1067-1090. Hummels, D. (2001), "Toward a Geography of Trade Costs", Purdue University, mimeo. IPEA, IBGE, and UNICAMP (2002), Configuração Atual e Tendêncies da Rede Urbana, Serie Configuração Atual e Tendêncies da Rede Urbana, Instituto de Pesquisa Econômica Aplicada, Instituto Brasileiro de Geografia e Estatistica, Universidade Estadual de Campinas, Brasilia. IPEA, IBGE, and UNICAMP (2002), Configuração Atual e Tendêncies da Rede Urbana, Serie Configuração Atual e Tendêncies da Rede Urbana, Instituto de Pesquisa Econômica Aplicada, Instituto Brasileiro de Geografia e Estatistica, Universidade Estadual de Campinas, Brasilia. Kolko, J. (1999), Can I get some service here: Information technology, service industries, and the future of cities, Harvard University mimeo. Krugman, P. (1991), Geography and Trade, Cambridge, Mass.: MIT Press. Krueckeberg, D. and K. Paulsen (2000), "Urban Land Tenure Policies in Brazil, South Africa, and India: An Assessment of the Issues," Lincoln Institute Research Report #LP00Z26. La Porta, R. and F. López-de-Silanes (1999), "The Benefits of Privatization: Evidence From Mexico," The Quarterly Journal of Economics, 114, 1193-1242. Lemos, M,. Moro, S., Biazi, E., Crocco, M. (2003). A Dinâmica urbana das Regiões Metropolitanas Brasileiras. Economia Aplicada, 7, 1:213-244. Malpezzi, S. and S. Mayo (1987), "The Demand for Housing in Developing Countries: Empirical Estimates from Household Data," Economic Development and Cultural Change, 35, 687-721. Malpezzi, S. and S. Mayo (1997), "Getting Housing Incentives Right: A Case Study of the Effects of Regulation, Taxes, and Subsidies on Housing Supply in Malaysia," Land Economics, 73, 372-391. Malpezzi, S. and D. Maclennan (2001), "The Long-Run Price Elasticity of Supply of New Residential Construction in the United States and the United Kingdom," Journal of Housing Economics, 10, 278-306. Mayer, C. and C. Somerville (2000), "Residential Construction: Using the Urban Growth Model to Estimate Housing Supply," Journal of Urban Economics, 48, 85-109. Morandi, L. and E. Reis (2004), "Estoque De Capital Fixo No Brasil, 1950-2002," Anais do XXXII Encontro Nacional de Economia, Proceedings of the 32th Brazilian Economics Meeting. Motta, D., Muelle, C., Torres, M. (1997). A Dimensão Urbana do Desenvolvimento Econômico-Espacial Brasileiro. Institute of Applied Economic Research (IPEA) Working Paper No. 530, IPEA, Brasilia. Pereira, F., Lemos, M. (2003). Cidades Médias Brasileiras: Características e Dinâmicas Urbano-Industriais. Pesquisa e Planejamento Econômico, 33, 1:127-165. Persson, T. and G. Tabellini (1994), "Is Inequality Harmful for Growth?," American Economic Review, 84, 600- 622. Quigley, J. and S. Raphael (2005), "Regulation and the High Cost of Housing in California," The American Economic Review, 95, AEA Papers and Proceedings, 323-328. Saks, R. (2005), "Job Creation and Housing Construction: Constraints on Metropolitan Area Employment Growth" FEDs Working Paper #2005-49. Inputs to a Strategy for Brazilian Cities Page 49 Saxenian, A. (1994), Regional Advantage: Culture and competition in Silicon Valley and Route 128, Harvard University Press. Shimer, R. (2001), "The Impact of Young Workers on the Aggregate Labor Market," The Quarterly Journal of Economics, 116(3), 969-1007. United Nations (2003). World Urbanization Prospects. Weil, D. (2005). Economic Growth, Addison-Wesley. Inputs to a Strategy for Brazilian Cities Page 50 4. The Evolution of Brazilian Municipal Finances, 2000-2004 by Fernando Blanco Introduction 4.1 The fiscal situation of Brazilian municipal governments has been experiencing a significant improvement. The turnaround on the fiscal stance resulted in the generation of increasing fiscal balances. As a consequence fiscal sustainability indicators improved. This improvement has been observed at the aggregate level as well as at the individual one. In fact, the great majority of the municipal government units have adopted a tight fiscal stance that allowed them to improve their financial indicators. 4.2 However, it is important to point out that the fiscal situation of the largest municipalities has suffered some deterioration. The importance of these municipalities on the aggregate municipal level has partially masked the very sound fiscal performance observed for most of the Brazilian municipalities. 4.3 The fiscal adjustment of the public sector, the enactment of the Fiscal Responsibility Law in 2000 and the imposition of strong credit access restrictions have favored the improvement of municipal fiscal accounts. First, municipal governments have accompanied the adjustment efforts of the federal government to generate increasing primary surpluses, which have in turn contributed to the overall improvement in Brazil's fiscal accounts. 4.4 Besides the own improvement on fiscal balances, the fiscal adjustment has promoted positive developments in municipal public finances. The difficult financial situation stimulated the enhancement of tax collection effort which was expressed in a strong increase on municipal tax revenues and made municipalities less dependents on transfers from federal and state governments. 4.5 On the expenditure side, municipalities have tried to interrupt the increasing path of personnel expenditures with partial success. Lack of new enrollment processes, better control on payroll bills and a conservative wage adjustment policy were adopted to keep personnel expenditures under control. The outsourcing of some services has shifted part of the personnel expenditures to the other current expenditures category (non-personnel operating expenditures), promoting a seeming reduction of personnel expenditures. However the social security benefits to retired municipal employees and constitutional obligations of minimum levels of expenditures continue to exert an upward pressure on municipal expenditures. The fall on non-personnel operating expenditures also reflects the adoption of a series of measures aimed at improving the public spending control of purchase and procurement systems that reduced the operating costs. Finally, debt service has not exerted a significant pressure on municipal finances, except for the largest municipalities. The low level of indebtedness of most of Brazilian municipalities and the debt renegotiation agreements between the National Treasury and about 180 largest municipalities that set a ceiling for debt service payments (13% of net revenues) explains the low pressure of this expenditure category on the public finances of most of the municipal government units. 4.6 The increase of current account savings allowed the improvement of municipal financial indicators. The interest payment coverage and the investment coverage indices exhibited significant improvements which indicate that municipalities are able to generate sufficient cash flows to pay their debt service obligations and finance most of their investment expenditures. 4.7 Second, the Fiscal Responsibility Law (FRL) has decisively favored the fiscal adjustment of municipalities. With the landmark Fiscal Responsibility Law, Brazil has made great strides toward institutionalizing fiscal discipline at all three levels of government. The prohibition of refinancing operations between level of government, the establishment of ceilings for key fiscal variables and the enhancement of fiscal transparency promoted a revolution in Brazilian public finances. Additionally, the establishment of ceilings for Inputs to a Strategy for Brazilian Cities Page 51 indebtedness, for credit operations, for personnel for expenditures, debt service payments and the observance of the golden rule has reinforced the tendency for the adoption of prudent fiscal stances by municipalities. 4.8 Besides the consolidation of fiscal responsibility, the Fiscal Responsibility Law fostered the strengthening of fiscal transparency and planning. The FRL requires the government units the public disclosure of fiscal and planning information including the Multi Annual Plans (PPA), Budgetary Guidelines Laws (LDO); the Annual Budget Law (LOA), and the bimonthly Summary Budget Execution Reports (RREO) and Fiscal Management Reports (RGF). 4.9 In order to fulfill the FRL this requirements, municipalities have continuously improved the quality of its fiscal accounts information. Progress has also been made in the standardization of norms and accounting procedures used in the reports of fiscal management and budget execution by an increasing number of the municipal governments. Also, the coverage of the fiscal information was amplified with the inclusion of the indirect administration in the fiscal accounting systems of municipalities. 4.10 Fiscal planning was also strengthened. The FRL stimulated a closer relationship between multi-annual plans, budgetary guidelines and annual budget laws and also a closer integration of budgeting, accounting and financial execution which should enhance the efficiency of municipal resource management and the creditworthiness of the municipal governments that were successful in the adoption of these FRL requirements. 4.11 Third, the imposition of hard domestic credit supply constraints has guaranteed the generation of municipal positive primary results trough the constraint of municipal public investment. Basically, the CMN resolutions establish two type of limitations: i) set a ceiling of 45% of equity exposure of domestic financial institutions to the borrowing of the public sector and ii) set global limits of domestic credit to the public sector. The first limitation is particularly binding for the Caixa Econômica Federal (CEF) and Banco Nacional de Desenvolvimento Econômico e Social (BNDES), two federal financial institutions which constitute the most important source of credit supply for municipal governments. Given the very low global ceilings for domestic credit to public entities, the second type of limitation in fact has blocked the access to credit by municipalities37. 4.12 As a consequence domestic credit operations to municipalities were drastically reduced inhibiting municipal investment expenditures. The domestic credit restrictions constrained the municipal investment expenditures and amplifying the gap between municipal investment needs and resource availability. Additionally, given the overall public sector fiscal adjustment, the federal and state governments have reduced the capital transfers to municipalities deepening the gap making current account savings the main financing source for municipal investments. 4.13 The need positive primary balances in a context of increasing budget rigidity led to the reduction of the fiscal space for investments. The increasing share of mandatory expenditures as personnel expenditures and debt service obligation and the establishment of new earmarking schemes in social sectors have reduced the space of maneuver for expenditure cuts, forcing municipal governments to concentrate the expenditure adjustment on discretionary expenditures affecting infrastructure investments. 4.14 Despite the overall improvement of Brazilian municipal finances important risks still remains. At this respect, the major source of vulnerability of municipal finances continues to be personnel expenditure which is the most important expenditure category. As mentioned above, the Fiscal Responsibility Law restriction on personnel expenditures which set a ceiling of 60% of Net Current Revenues (NCR) for this expenditure category, forced municipalities to adopt a tight control on personnel expenditures. However, as the personnel expenditure containment measures adopted and their effect are temporary, more structural measures need to be implemented to guarantee that the increasing trend of this expenditure item will not be resumed in the medium term. 37The National Monetary Council issued resolutions 2827 (March 2001), 2920 (December 2001), 2954 (April 2002), 3049 (November 2002), and subsequent resolutions - issued in 2003-2004 ­ with the objective to limit the access to credit operations by government entities. The subsequent resolutions renewed the period of these credit restrictions and allowed some exceptions for the finance of specific infrastructure investments (sanitation). Inputs to a Strategy for Brazilian Cities Page 52 4.15 In particular, the upward trend of social security benefits for retired public employees continues to be the structural threat for the continuity of the personnel expenditure containment. The municipal employees' social security system is a pay as you go system and presents very large actuarial deficits that indicate potential increase on personnel expenditures in the future. 4.16 The imbalances in the social security system are a major preoccupation affecting the financial health of all three administrative levels - central, state and municipal - in Brazil. Over the past few years, two Constitutional Amendments were introduced in an attempt to reduce this imbalance by modifying the rules governing social security. However, the impact on municipal personnel expenditures of should be limited and further reform measures need to be adopted in order to reduce the imbalances of the social security system for municipal employees. 4.17 Other source of budget rigidity also comes from the proliferation of constitutional revenue earmarking schemes directed to privilege budget allocations in social areas. The enforcement of the achievement of minimum levels of expenditure in health and education is forcing municipalities to expand personnel expenditures as these sectors are labor intensive, pressuring the reduction of current account savings and reducing the fiscal space for municipal infrastructure investments. 4.18 The positive prospects of the Brazilian economy depict a favorable evolution for the municipal fiscal and financial indicators in the next years. In particular, a higher economic growth (with stronger effects on revenues than expenditures) vis a vis the population growth (with stronger effects on expenditures than revenues) would favor fiscal sustainability. 4.19 But the guarantee of municipal fiscal sustainability critically depends on the enhancement of municipal revenue collection and on the containment of current expenditures. In the revenue side, many municipal governments have been implementing programs to enhance the efficiency of the municipal revenue services obtaining very good results. However, there is a large space for further improvements especially for medium size municipalities. Also, the disappointing performance of user charges and cost recovery suggest the possibility of substantial gains on this revenue source in the future. 4.20 On the expenditure side, the control of the growth of personnel expenditures and the rationalization of government purchases would consolidate the adjustment and improve the quality of municipal service delivery. The containment of personnel expenditures is the main challenge of municipalities as it constitutes a serious obstacle that threat their fiscal sustainability. Also, the improvement of public sector management in order to increase the efficiency of tax collection, generate other own revenue sources and to promote a more efficient use of municipal resources would allow municipalities to consolidate their fiscal adjustment and to expand and improve the public services delivery. 4.21 This report contains six sections including this introductory part. The second describes the evolution of the municipal fiscal balances and of the main Fiscal Responsibility Law indicators during the period 2000-2004. The third section analyzes the evolution and composition of municipal revenues. The fourth relates the evolution of municipal expenditures by economic categories and functions. The five section analyses the evolution of municipal indebtedness and describes the mechanisms that restrict credit access by municipalities. The final section summarizes the main overall findings and makes a number of policy implications.The Fiscal Adjustment of Brazilian Municipalities, 2000-200438. 4.22 During the period 2000 to 2004, Brazilian municipal government level fiscal accounts presented a solid improvement expressed in the generation of increasing fiscal balances. Total balance passed from R$ 4.2 billion 38For the analysis of the municipal finances two samples were used. For the analysis of the evolution of municipal finances during the period 2000-2004, given the data availability limitations, a sample of 3,208 municipalities were used to make consistent inter temporal comparisons. These 3,028 municipalities respond for 78 percent of the Brazilian population. For the cross comparisons by municipal population size, it was selected the year of 2003 as it has the largest sample (4,967 municipalities) representing 95 percent of the Brazilian population. Inputs to a Strategy for Brazilian Cities Page 53 or 0.24 percent of GDP in 2000 to R$ 7.8 billion or 0.44 percent of GDP in 2004. In terms of current account balances, municipalities passed from a surplus of R$ 12.4 billion or 0.7 percent of GDP in 2000 to a current account surplus of R$ 18 billion or 1.06 percent of GDP in 2004. Also, Table 4.1 reveals that the primary balance passed from R$ 6 billion or 0.34 percent of GDP in 2000 to a surplus of R$ 8.6 billion or 0.48 percent of GDP. 4.23 The improvement resulted from a higher increase in revenues than in expenditures. During the period 2000-2004, total revenues increased by 21 percent, with an increase of current revenues of 23 percent and a decrease of capital revenues of 5 percent. On the expenditure side, total expenditures grew by 18% with current expenditures increasing by 18 percent and capital expenditures growing by 20 percent. The increase of the primary balance resulted from an increase of 19% of primary revenues vis a vis an increase of 17 percent of primary expenditures. 4.24 As shown by Table 4.1, the fiscal adjustment was based on the strong increase of current revenues enabling municipalities to increase their ability to generate current account savings. The increased current savings allowed municipalities to meet its debt service obligations and to finance most of their investment expenditures as capital revenues have stagnated during the period. 4.25 It is important to remark, that the improvement of municipal fiscal balances happened in a period of a disappointing economic performance (average growth rate of 2.6 percent) which made more difficult the adjustment. In fact, the slowdown of the economic activity in 2003 determined a temporary interruption of the increasing trend of fiscal balances. The strong recovery of the economic activity in 2004 led to a vigorous revenue increase of 11% of municipal revenues consolidating the fiscal adjustment that had been implementing in the previous years. Table 4.1 Municipal Fiscal Balances, 2000-2004 (Billion of Reais of 2004) 2000 2001 2002 2003 2004 1. Total Revenues 96.8 101.3 111.2 108.1 117.6 2. Total Expenditures 92.6 96.7 105.2 103.7 109.8 3. Total Balance (1-2) 4.2 4.6 6.0 5.4 7.8 In Percent of GDP 0,24% 0,26% 0,34% 0,28% 0,44% 4. Current Revenues 92.8 98.7 106.6 105.3 113.8 5. Current Expenditure 80.4 85.6 89.8 90.2 95.1 6. Current Balance (4-5) 12.4 13.1 16.8 15.1 18.7 In percent of GDP 0,7% 0,74% 0,95% 0,85% 1,06% 7. Capital Revenue 4.0 2.6 4.6 2.8 3.8 8. Capital Expenditure 12.2 11.1 15.4 13.5 14.7 9. Capital Balance (7-8) (7.8) (8.5) (10.8) (10.3) (10.9) In Percent of GDP (0.46%) (0.48%) (0.61%) (0.57%) (0.62%) 10. Financial Revenues 1.3 2.0 3.7 4.5 3.7 11. Financial Expenditures 3.0 3.9 4.3 4.3 4.4 12. Primary Revenues (1-10) 95.5 99.3 107.5 103.6 113.9 13. Primary Expenditures (2-11) 89.6 92.8 100.9 99.4 105.4 Primary Balance (11-13) 5.9 6.5 6.6 4.2 8.5 In Percent of GDP 0,34% 0,37% 0,37% 0,24% 0,48% Sample of 3,028 municipalities. 4.26 The improvement of fiscal balances has not been homogeneous among municipal government units. Small and medium size municipalities experienced a stronger improvement than the largest municipalities. In fact, Inputs to a Strategy for Brazilian Cities Page 54 the municipalities with population superior to 1 million of inhabitants have worsened their fiscal balances. The importance of these municipalities in the aggregate municipal level buried the significant gains obtained by the rest of municipal government units39. 4.27 Excluding the municipalities with more than 1 million of inhabitants, the rest of municipalities increased their total balances by R$ 5 billion, their current balances by more than R$ 6 billion and their primary balances by R$ 3 billion. Table 4.2 Current and Primary Balance of 2000-03 (Billion of Reais of 2004) 2000 2003 Population Total Current Primary Total Current Primary 0 to 5,000 0.05 0.39 0.09 0.41 0.80 0.40 5,000 to 20,000 0.15 1.40 0.35 1.43 3.02 1.51 20,000 to 150,000 0.68 3.66 1.06 2.53 6.72 2.41 150,000 to 1,000,000 1.02 3.51 1.21 1.62 4.70 1.38 More than 1,000,000 2.55 5.42 3.72 (0.11) 3.35 (0.02) Total 4.46 14.4 6.42 5.88 18.6 5.69 Sample of 4,965 municipalities. 4.28 As a result of its general sound fiscal performance, the aggregated municipal level has complied with the fiscal limits established by the Fiscal Responsibility Law (FRL). In fact, aggregate municipal FRL indicators are well below the legal ceilings. Net consolidated debt as a proportion of net current revenue, used for compliance of the Fiscal Responsibility Law, has fluctuated around 35 percent, far below the legal ceiling of 120 percent. 4.29 Table 4.3 also shows that municipalities have kept within the other limits laid down by the Fiscal Responsibility Law (FRL). In 2000-2004, overall personnel costs represented about 46 percent of net current revenue, also below the ceiling of 60 percent of net current revenue. 4.30 Debt service as a proportion of the net current revenue has oscillated around 4 percent, well below the legally-imposed limit of 11.5 percent. The lack of access to credit operations, made the indicator credit operations to net current revenue to vary around the 1 percent level while the corresponding FRL limit is 16%. 4.31 Besides the Fiscal Responsibility Law figures, other two financial indicators were included to evaluate the cash flows generation of the municipal governments. The first is the interest coverage, which represents the current account cash flow without interest payments, that is the flow of funds generated internally, divided by the interest payments. The minimum level recommended for this indicator is 1.5. The second indicator is the investment coverage, which represents the amount of investment financed with cash flow (current account balance). The minimum level recommended for this indicator is 30%. 4.32 The Brazilian municipal governments have both indicators well above the minimum levels recommended for these two financial indicators. In the case of the interest coverage, Table 4.3 shows that municipalities had large ability to generate cash flows to finance interest payments on their debts. However, it is important to mention that the low level of indebtedness of most of the Brazilian municipalities more than the ability to generate current account cash flow justifies the high value of the interest coverage. 4.33 Also, municipalities were able to integrally finance their investment expenditures as the investment coverage indicator is higher than 1. Again, it is important to mention that the fiscal adjustment and the restrictions to credit determined a low level of investment expenditures which explains the municipal government's high investment coverage. 39From the 5,500 Brazilian municipalities, more than 1,300 have population below 5,000 inhabitants, 2,700 municipalities have population between 5,000 to 20,000 inhabitants, 1,300 municipalities have population between 20,000 to 150,000 inhabitants, 132 municipalities have population between 150,000 to 1,000,000 and 12 municipalities have a population above 1,000,000 of inhabitants. Inputs to a Strategy for Brazilian Cities Page 55 Table 4.3 FRL and Financial Indicators 2000 2001 2002 2003 2004 Debt/NCR* (%) 35.3 29.2 35,7 35,8 - Personnel Expenditures/NCR (%) 43.5 43.4 45,9 45,5 47,4 Debt Service/NCR (%) 3.1 4.1 4,3 4,4 4,1 Credit Operations/NCR (%) 1.1 0.6 0,7 0,9 1,1 Interest Coverage 8.3 6.9 8.5 7.8 9.1 Investment Coverage 1.3 1.6 1.30 1.39 1.52 *Source: National Treasury. Sample of 3,215 municipalities. 4.34 The heterogeneity among Brazilian municipalities is more evident in the case of the Fiscal Responsibility Law and the financial indicators described above. In particular the extremely concentrated weight of the largest municipalities deserves some attention. 4.35 The municipalities with population higher than 1 million of inhabitants respond for almost 80 percent of municipal debt, for 35 percent of municipal personnel expenditures, for 60 percent of municipal debt service, for 70 percent of credit operations and for 28 percent of municipal investment expenditures. 4.36 Table 4.4 shows that indebtedness is a problem only for the largest municipalities as the net debt to net current revenues for small and medium size municipalities are well below the limits established by the Fiscal Responsibility Law40. For large municipalities, it is important to remark that the debt indicator is high because the high indebtedness of São Paulo (251 percent of net current revenue) and the importance of São Paulo on municipal accounts. (Box 4.1). 4.37 Personnel expenditure to net current revenue ratio is distributed more homogeneously among municipal government units varying around 45 percent without a clear association with the population size. However, there are striking differences among active and inactive personnel expenditures. Municipalities with population lower than 1 million of inhabitants are responsible for 72 percent of the municipal active personnel expenditures while municipalities with more than 1 million of inhabitants respond for 61 percent of the inactive personnel. 4.38 These figures have important policy implications: the weight of active personnel expenditures in small size municipalities results from the lack of scale economies and would constitute a real threat to municipal finances as high active personnel implies high social security benefits in the future. For large size municipalities, the high retired personnel expenditures implies the need of containment of the deficit of the social security system of municipal employees and a conservative employment policy in order to maintain under control personnel expenditures. 4.39 Large differences are observed with relation to the debt service and credit operations to net current revenues indicators. Table 4.5 shows that debt obligations are not a problem at all for small size municipalities and that only large size municipalities have some access to credit operations. 4.40 A very interesting result is the low levels of the indicators of indebtedness and debt service to net current revenues ratios compared with the Fiscal Responsibility Law ceilings which is observed for the great majority of Brazilian municipalities with the exception of the largest municipalities. This fact indicates that the FRL ceilings were set looking at the indebtedness situation of the largest municipal governments. In turn, the large difference between the indicators observed in small size municipalities and the FRL ceilings would stimulate a quick and large increase of indebtedness as the space for additional indebtedness allowed by the FRL is substantial. Thus, this observation shows a weakness of the FRL in the sense that it set identical ceilings for very different municipalities. 40. Because the lack of availability of disaggregated municipal debt information, the net debt used in table 4.4 was calculated by WB staff and presents some differences with the Fiscal Responsibility Law indicator presented in table 4.3. Inputs to a Strategy for Brazilian Cities Page 56 4.41 The interest and investment coverage depict a similar pattern. There is a negative correlation between interest and investment coverage with the population size. The very low or almost inexistent indebtedness of small size municipalities makes them able to face the interest payments obligations. Given the lack of access to credit operations, small municipalities can finance the totality of their investment expenditures. Larger municipalities can also pay their interest payment obligations and finance their investment expenditures with more difficulty. Table 4.4 FRL and Financial Indicators by Municipal size 2003 0 to 5,000 5,000 to 20,000 to 150,000 to More than Total 20,000 150,000 1,000,000 1,000,000 Debt/NCR (%) FRL = 120% 4.7 12.2 14.3 29.6 136.1 49.6 Personnel Exp/NCR (%) FRL =60% 43.8 45.7 46.3 47.5 46.9 46.2 Debt Service/NCR FRL=11.5% 1.5 1.9 2.4 3.0 9.4 3.8 Credit Op/NCR FRL = 16% 0.2 0.2 0.3 0.5 2.4 0.7 Interest Coverage 119.1 86.8 41.7 12.4 2.9 9.0 Investment Coverage 1.7 1.7 1.6 1.5 1.0 1.4 Sample of 4,965 municipalities. 4.42 It is important to remember that even among the largest municipalities there are marked heterogeneities. In particular, the importance and high indebtedness of São Paulo and Rio de Janeiro municipalities generate an upward bias on municipal Fiscal Responsibility Law and financial indicators. Table 4.5 Interest and Investment Coverage by Municipal Size 2000 2003 Population Interest Investment Interest Investment 0 to 5,000 45.13 0.74 116.63 1.74 5,000 to 20,000 30.86 0.72 84.87 1.67 20,000 to 150,000 23.86 0.95 41.03 1.59 150,000 to 1,000,000 10.37 1.15 12.39 1.49 More than 1,000,000 5.49 1.94 2.93 1.00 Total 9.00 1.18 8.87 1.43 Sample of 4,965 municipalities. Inputs to a Strategy for Brazilian Cities Page 57 Box 4.1 São Paulo Indebtedness As São Paulo net consolidated debt represents more than 70 percent of Brazilian municipal net consolidated debt, its evolution has a strong impact on the aggregated indebtedness indicators of Brazilian municipalities. At the end of 2004, São Paulo net consolidated debt was equal to 251 percent of net current revenue, far above the Fiscal Responsibility Law ceiling of 178%, which should reach 120% of net current revenue in 2016. More than its high level, the problem is that the indebtedness has followed an increasing trend in the last years. Federal Senate Resolution 40/2001, complying Fiscal Responsibility Law, establishes a ceiling for net consolidated debt to net current revenues of 120 percent for municipal governments. For municipalities above this ceiling, a decreasing path for the next fifteen years was set. In 2001, São Paulo municipal government was above the 120 percent ceiling by 73 percent of net current revenue. As a result, the legal requirement establishes that São Paulo municipality indebtedness level should decrease by 4.9 percentage points each year since then to achieve 120 percent in 2016. In contrast to this legal requirement, from 2001 to 2004, there was a significant indebtedness increase. During this period, indebtedness level rose by 54 percentage points of the net current revenues (the legal requirement was an accumulated reduction of 14.7 percent of net current revenue). Several factors contributed to this unsustainable trend: automatic rollover of part of due interest of debt contracts with the National Treasury, debt indexation above consumer price index and the generation of net borrowing requirements (deficits). The debt refinancing agreement between federal government and regional and local governments set a maximum amount of debt service to be paid 13 percent of real net revenue, for a set of debt contracts. Any due debt service above this would be incorporated to the debt stock. As a result of this ceiling, amortization payments were very low, around 1 percent of debt stock each year. At the same time, new credit operations were contracted with an average increase of 2 percent on debt stock per year from 2002 to 2004. Thus, as new credit operations were higher than the amortization payments, the municipal debt increased. Also, debt service ceiling at renegotiation agreement meant an automatic refinancing of part of interest due, the capitalization of interest payments also contributed to the indebtedness increase. Debt indexation was another factor for the debt increase. As 95 percent of contractual debt is indexed by IGP (90 percent) and the exchange rate (6 percent), this indexation was responsible for an important part of the increase on the debt level. Although its aim was only to correct the stock level according to inflation, the general price index (IGP) presented in the last years a higher elevation than other price indexes such as consumer indexes and gross domestic product (GDP) deflator. Finally, the positive contribution of net borrowing needs implied that there was no fiscal effort to decrease the indebtedness level during the period. The Evolution of Municipal Revenues 4.43 As mentioned above, the improvement in the municipal fiscal balances was based on the good performance of revenues, in particular on the increase of current revenues. From 2000 to 2004, total revenues Inputs to a Strategy for Brazilian Cities Page 58 grew by 21 percent with current revenues increasing by 23 percent while capital revenues experienced a fall of 5 percent. The strong increase of municipal current revenues needs to be remarked. Brazil's accumulated GDP growth was of 9.5 percent during the period, indicating that municipal current revenues grew 13 percent above GDP during this period. 4.44 On the contrary, the declining performance of capital revenues reflects the hard credit restrictions imposed by the federal government on credit operations. At the same time, capital transfers from federal and state governments also suffered a strong containment given the fiscal adjustment effort of the three levels of government. Table 4.6 Municipal Revenues, 2000-2004 (Billion of Reais of 2004) 2000 2001 2002 2003 2004 Current Revenues 92.8 98.7 106.6 105.3 113.8 Capital Revenues 4.0 2.6 4.6 2.8 3.8 Total 96.8 101.3 111.2 108.1 117.6 Current Revenues per capita (R$) 772 821 887 876 946 Capital Revenues per capita (R$) 33 22 38 24 32 Total ­ per capita (R$) 805 843 925 900 978 Sample of 3,028 municipalities. 4.45 The classification of municipalities by population size shows large disparities in two dimensions. First, the municipalities with population larger than 1 million respond for more than 25 percent of municipal total, current and capital revenues while the smallest municipalities have a very low participation in municipal revenues (about 4 percent). Second, despite their low participation on municipal revenues, these smallest municipalities have municipal revenues per capita 20 percent higher than the 12 largest municipalities and 50 percent higher than the national average. 4.46 Clearly, the concentration of municipal revenues in the largest municipalities results from the concentration of population, tax bases and economic activity in these cities while the intergovernmental transfers system is the main cause for the high municipal revenues per capita of small municipalities. Table 4.7 Municipal Revenues by Municipal Size, 2003 (Billion of Reais of 2004) 0 to 5,000 5,000 to 20,000 to 150,000 to More than 20,000 150,000 1,000,000 1,000,000 Brazil Current Revenues 4.7 19.2 39.4 33.6 33.2 130 Capital Revenues 0.14 580.6 0.87 0.7 1.1 3.4 Total 4.8 19.8 40.2 34.2 34.4 133.5 Current Revenues per capita (R$) 1,202 738 718 793 1,027 816 Capital Revenues per capita (R$) 36 22 16 16 35 21 Total ­ per capita (R$) 1,238 760 734 810 1,063 837 Sample of 4,965 municipalities. 4.47 Analyzing the components of current revenues it is possible to observe that the three main categories exhibited a strong growth. Tax revenues increased by 22 percent, intergovernmental transfers by 20 percent while other current revenues by 36 percent. Tax revenues and intergovernmental transfers exhibited the same growth pattern growing permanently until 2003, when the economic slowdown determined the interruption of the increasing trend. In 2004, the strong recovery of the economic activity fostered the strong revenue growth. An Inputs to a Strategy for Brazilian Cities Page 59 important implication of this pattern is that the structure of current municipal revenues makes them highly correlated with the economic activity cycle. The low importance of property taxes is behind this high sensitiveness of municipal revenues to the economic activity. 4.48 Municipal taxes exhibited a very strong growth of 28 percent while user charges declined by 15 percent. The good performance of tax revenues can be explained mostly by the incorporation of the federal Income Tax (IRRF) of municipal employees as a municipal tax since 2002. The two most important taxes collected by the municipalities, the urban property tax (IPTU) and the tax on services (ISS) also presented a strong growth of 15 percent and 21 percent respectively. 4.49 Even excluding the effect of the incorporation of the IRRF as municipal own tax revenue, the growth rates of IPTU and ISS were well superior to the growth rate of the Brazilian economy, fact that indicates the enhancement of tax collection efficiency. In the last years, many municipal governments launched an aggressive program to enhance the efficiency of the municipal revenue services. Investments in software and the modernization of administrative processes were responsible for the improvement of tax collection effort. In the opposite way, the performance of user charges was disappointing suggesting the weakness of municipal cost recovery policies and the space for a strong improvement of this revenue source in the future. 4.50 Intergovernmental transfers increased by 20 percent. Most of the increase of intergovernmental transfers responds to the increase of multi-governmental transfers. In particular, the re-classification of the FUNDEF (Basic Education Fund) as a multi-governmental transfer explains the strong increase of this category. Federal transfers and state transfers experienced a more modest behavior, specially the transfers from state governments. Federal transfers grew by 8 percent while state transfers decreased by 1.3 percent. The federal transfers from the Municipal Participation Fund (FPM ­ Fundo de Participação Municipal- revenue sharing of federal income and industrial product taxes) increased by 20 percent. At the state level, the most important transfer is the revenue sharing of the ICMS, which increased by 18 percent. Other transfers both from the federal and from the state governments decreased by 3.5 percent and 46 percent respectively41. 4.51 Other current revenues exhibited the best performance among current revenue categories (36 percent). This category includes returns of financial assets, tax in arrears recovery, fines and other minor revenue items. Table 4.8 Municipal Current Revenues, 2000-2004 (Billion of Reais of 2004) 2000 2001 2002 2003 2004 Current Revenues 92.8 98.7 106.6 105.3 114.8 Tax Revenues 20.1 20.8 23.4 23.3 24.6 Taxes 17.3 18 20.7 20.7 22.2 IPTU 7.1 7.4 7.8 7.9 8.2 ISS 8.5 9.0 9.2 9.1 10.3 Other taxes (IRRF- ITBI) 1.7 1.6 3.7 3.7 3.7 User Charges 2.8 2.8 2.7 2.6 2.4 Intergovernmental Transfers 60.4 64.3 67.8 65.8 72.5 Federal 26.7 28.5 26.3 25.3 28.9 FPM 12.9 14.3 16.1 15.0 15.6 Other 13.8 14.2 10.2 10.3 13.3 State 32.4 34.5 30.6 30.2 32.0 ICMS 22.5 24.1 25.0 25.2 26.6 Other 9.9 10.4 5.5 5.0 5.4 Sample of 3,028 municipalities. 41Again, the re-classification of Fundef as a multi-governmental transfers explain the strong fall of other state transfers. Inputs to a Strategy for Brazilian Cities Page 60 4.52 The similar growth pattern of municipal tax revenues and intergovernmental transfers maintained practically constant the municipal current revenue structure. The participation of tax revenues was constant at 21 percent of current revenues while the better performance of other current revenues led to a higher participation of this revenue category. Table 4.9 Current Revenue Composition (%) Brazil, 2000-04 2000 2001 2002 2003 2004 Current Revenues 100,00 100,00 100,00 100,00 100,00 Tax Revenues 21,61 21,08 21,93 22,09 21,55 Intergovernmental Transfers 65,13 65,10 63,57 62,46 63,73 Other Current Revenues 13,26 13,81 14,50 15,45 14,72 Sample of 3,028 municipalities. 4.53 The high participation of intergovernmental transfers on municipal current revenues shows a high dependence of municipal finances on transferred revenues. However, it is important to stand out that most of the transfers are constitutionally sanctioned (FPM, ICMS and IPVA); other are legally regulated transfers (SUS ­ the Health Unique System ­ and FUNDEF ­ the Educational Fund). Both are established at the Constitution but the amount of transfers is regulated by infra constitutional legislation. Only a small fraction of intergovernmental transfers are voluntary transfers. This may indicate less risk on the revenue side than it seemed at first sight. In any case, the dependence of municipal finances on the tax collection effort and tax policies of federal and state governments which are clearly not controlled by the municipality, makes municipal finances vulnerable as revenue increases not only depends on municipal own tax collection effort but also on the these transfers. 4.54 The classification of municipalities by size of population shows again the strong concentration of tax revenues inn the largest municipalities. The 12 largest municipalities respond for almost 50% of municipal tax revenues while the smallest municipalities collect only 1% of municipal tax revenues. Figure 4.1 Municipal Tax Revenue Collection by Municipal Size, 2005 1% 4% 19% 48% 28% 0 to 5,000 5,000 to 20,000 20,000 to 150,000 150,000 to 1,000,000 More than 1,000,000 4.55 Disparities are also evident using per capita figures but in this case, favoring small municipalities. Current revenues per capita in small municipalities are 50 percent higher than in medium size municipalities and about 15 percent higher than in the largest municipalities. 4.56 However, tax revenues per capita in the largest municipalities are nine times the tax revenue. On the opposite direction, the redistributive nature of federal transfers and the distortions provoked by the rules of Inputs to a Strategy for Brazilian Cities Page 61 distribution that favor the creation of small municipalities explain the fact that federal transfers per capita in the smallest municipalities are five times the federal transfers per capita in the largest municipalities and three times the medium size municipalities. Transfers per capita from state governments are more equally distributed among different size of municipalities. Figure 4.2 Municipal Current Revenues per Capita by Municipal Size (Reais of 2004) 1,400 1,200 1,000 800 600 400 200 0 0 to 5,000 5,000 to 20,000 to 150,000 to More than Brazil 20,000 150,000 1,000,000 1,000,000 Tax Revenue Federal Transfers State Transfers Current Revenue 4.57 The strong disparities described above can be expressed in terms of the composition of municipal current revenue. Own revenues (tax revenues plus other current revenues) represent only 8.8 percent of current revenues of small municipalities. A very low participation of own revenue sources can also be observed for municipalities with population up to 150,000 inhabitants. On the contrary, Table 4.10 shows the extreme dependence of small and medium size municipalities on intergovernmental transfers, with a weight of transfers varying between 91 percent to 74 percent levels. On the other side, largest municipalities have a more balanced current revenue composition with own resources responding for 55 percent of municipal current revenues. Table 4.10 Current Revenue Composition (%) by Municipal Size, 2003 0 to 5,000 5,000 to 20,000 to 150,000 to More than 20,000 150,000 1,000,000 1,000,000 Brazil Current Revenues 100,00 100,00 100,00 100,00 100,00 100,00 Tax Revenues 3,52 5,83 12,14 21,63 36,56 19,58 Intergovernmental Transfers 91,11 87,12 73,93 62,25 45,30 66,17 Other Current Revenues 5,37 7,05 13,93 16,12 18,14 14,25 Sample of 4,965 municipalities. Inputs to a Strategy for Brazilian Cities Page 62 Box 4.2: The Effects of Iintergovernmental Transfers on Recipient's Tax Effort and Expenditure While the Brazilian intergovernmental transfer system promote the decentralization of public resources and the reduction of regional fiscal disparities, the design of the Brazilian intergovernmental transfer system generates distortions on the behavior of recipient units. In particular, the predominance of intergovernmental transfers in the finance structure of municipalities is associated to: i) a low interest on exploiting own revenue sources (tax revenue); ii) expenditure expansions as a result of the disconnection between cost and benefits of public goods and expenditures; iii) low quality of public expenditures by the recipient units explained by the lack of control and transparency of the transferred resources and iv) the creation of a large number of new local governments stimulated by the federal transfers to municipalities distribution rules. Municipalities that rely on transfers as their main revenue source are not concerned with the enhancement of its tax collection efficiency, given the low importance of own tax revenue. Blanco (1998) estimated the effect of transfers in the tax collection effort of states and municipalities and found a strong negative correlation between the share of transfers on current revenue and the tax collection efforts. The automatic nature of the intergovernmental transfers generates a perception that the transferred resources are assured, perception that accentuates the negative effect on the recipient's tax collection effort. Because transfers are (supposedly) not financed by local taxpayers, there is an undervaluation of the cost of public goods. As a result, there is an expansionary bias of transfers when compared with own revenue sources. Known as the flypaper effect, the finding that one additional unit of revenue coming from transfers have a stronger effect on expenditures than an additional unit of income of taxpayers, was confirmed by a large set of empirical evidences. Blanco and Carvalho (2000) found strong evidences of the presence of the flypaper effect in the municipal public finances in Brazil. A study on the impact of the revenue structure of municipalities on the expenditure composition showed that the share of intergovernmental transfers on the available revenue is positively correlated with overhead expenditures and negatively correlated with municipal infrastructure and social expenditures. This result can be explained because small municipalities (that are highly dependent on intergovernmental transfers) have high overhead cost given the difficulties to explore scale economies. Finally, Shikida (1997) demonstrated that the fiscal decentralization process trough the strengthening intergovernmental system promoted by the Constitution of 1988 and the distribution rules of the federal transfer to municipalities (FPM) resulted in the creation/ emancipation of municipalities. The higher federal transfers per capita received by small municipalities created incentive for the emancipation of municipalities. From 1988 to 1997, more than 1,400 municipalities were created. In summary, while recognizing that the intergovernmental transfer system is the most important mechanism for the decentralization and the redistribution of fiscal resources in a federation characterized by strong regional disparities as Brazil, the redesign of the rules to alleviate the negative incentives provoked by transfers constitute a challenge for the Brazilian fiscal federalism. 4.58 However, small and medium size municipalities have improved their tax collection performance more than large municipalities. Table 4.11 shows a strong negative correlation between municipal size and the increase of tax revenue during the period 2000 to 2004. It is noteworthy the improvement of tax revenue collection of small and medium size municipalities which exhibited an accumulated growth varying between 78 percent to 42 percent, well above the national average increase. The growth of intergovernmental transfers was higher in medium size municipalities as well as the growth of other current revenue. The higher growth of tax revenues vis a vis the growth of intergovernmental transfers should reduce the dependence of small municipalities on transfers which is a good development. 4.59 Table 4.11 also reveals that medium size municipalities experienced the higher growth of current revenues which resulted from a good tax revenue performance, a significant increase of intergovernmental Inputs to a Strategy for Brazilian Cities Page 63 transfers and an exceptional performance of other current revenues. On the opposite side the largest municipalities' current revenues had a disappointing performance that explains the worsening of their fiscal situation. Table 4.11 Current Revenues Growth by Municipal Size, 2000-2004 (%) 5,000- 20,000- 150,000- More 0-5,000 20,000 150,000 1,000,000 than Brazil 1000000 Current Revenues 26.06 28.17 35.36 19.01 14.25 22.61 Tax Revenues 78.11 48.10 42.28 28.46 12.59 22.31 Intergovernmental Transfers 24.52 26.60 30.05 13.98 10.48 19.97 Other Current Revenues 27.74 33.22 63.07 27.52 29.39 36.09 Sample of 3,028 municipalities 4.60 On the capital account side, capital revenues experienced a strong volatility during 2000-2004, varying between R$ 4.6 billion (highest level in 2002) and R$ 2.6 billion (lowest level in 2001) without a defined trend. In 2004, capital revenues were 5 percent lower than the observed level in 2000. Its main components, credit operations and capital transfers experienced a similar volatile behavior. 4.61 The domestic supply credit constraints imposed by the National Monetary Council (CMN) since 2001 practically blocked the access to credit operations by municipalities, while the fiscal adjustment of the federal government and state governments limited the capital transfers from these higher levels of governments to municipalities. Other capital revenues which mainly include municipal asset sales were kept at low levels. Table 4.12 Capital Revenues, 2000-2004 (Billion of Reais of 2004) 2000 2001 2002 2003 2004 Capital Revenues 4.02 2.62 4.61 2.83 3.81 Credit Operations 1.01 0.56 0.73 0.88 1.25 Capital Transfer 2.28 1.62 3.38 1.27 2.05 Other / Asset Sales 0.73 0.44 0.50 0.69 0.51 Sample of 3,028 municipalities. 4.62 Using the stratification by size of population, it is possible to see a very similar pattern than the observed for the municipal current revenues. Capital revenues are highly concentrated in the largest municipalities which respond for 30 percent of them. Credit operations are even more concentrated in the largest municipalities which contracted more than 70 percent of municipal credit operations during the period. In fact, small municipalities do not have access to credit operations. Capital transfers are more evenly distributed among municipalities while asset sales are also highly concentrated in the largest municipalities. Table 4.13 Capital Revenue by Municipal Size, 2003 (Billion of Reais of 2004) 0 to 5,000 5,000 to 20,000 to 150,000 to More than 20,000 150,000 1,000,000 1,000,000 Brazil Capital Revenues 0.14 0.58 0.87 0.70 1.13 3.42 Credit Operations 0.01 0.03 0.10 0.17 0.60 0.91 Capital Transfers 0.11 0.47 0.65 0.39 0.15 1.77 Other / Asset Sales 0.02 0.08 0.11 0.14 0.38 0.74 Sample of 4,965 municipalities. 4.63 The per capita figures also show a similar pattern than the observed for municipal current revenues. While capital revenues per capita are higher for the smallest and largest municipalities, credit operations per capita in the largest municipalities are seven times the same figure for small municipalities, confirming the lack of access to Inputs to a Strategy for Brazilian Cities Page 64 credit operations by small municipalities. Capital transfers per capita exhibit a similar pattern than current transfers per capita: are higher in small municipalities (six times than the capital transfers to largest municipalities). In summary, large municipalities tend to use credit operations to finance investment expenditures while the only source of capital revenues for small municipalities is the capital transfers from the federal and state governments. Figure 4.3 Municipal Capital Revenues per Papita by unicipal Size, 2003 (Reais of 2004) 40 35 30 25 20 15 10 5 0 0 to 5,000 5,000 to 20,000 to 150,000 to More than Brazil 20,000 150,000 1,000,000 1,000,000 Capital revenues Credit Operations Capital Transfers Other The Evolution of Municipal Expenditures Municipal Expenditures by Economic Category 4.64 As mentioned in section A, the improvement in municipal fiscal accounts resulted from the lower increase of expenditures than of the observed in the revenue side. From 2000 to 2004, total expenditures revenues grew by 18 percent. Current expenditures increased by 18 percent while capital expenditures grew by 20 percent. As in the case of revenues, Table 4.14 shows that between 2000-2004 it was observed a permanent expenditure increase which was interrupted in 2003 as a consequence of the economic stagnation that imposed a fall in municipal revenues. In 2004, the recovery of economic activity fostered a strong revenue increase that allowed municipalities to increase their expenditures. Table 4.14 Municipal Expenditures, 2000-2004 (Billion of Reais of 2004) 2000 2001 2002 2003 2004 Current Expenditures 80.4 85.6 89.8 90.2 95.1 Capital Expenditure 12.2 11.1 15.4 13.5 14.7 Total 92.8 96.7 105.2 103.7 109.8 Current Expenditures ­ per capita (R$) 669 691 704 686 703 Capital Expenditures ­ per capita (R$) 101 90 121 103 109 Total ­ per capita (R$) 770 781 825 789 812 Sample of 3,028 municipalities. Inputs to a Strategy for Brazilian Cities Page 65 4.65 The classification of municipalities by population size also shows a similar pattern than the observed in the revenue side. First, it is possible to observe a strong concentration of expenditures in the largest municipalities which execute more than 25 percent of municipal expenditures while the small municipalities have an insignificant participation in municipal expenditures (about 3 percent). Second, as in the case of revenues, the small municipalities have high expenditures in per capita terms, with total, current and capital expenditures per capita about 50 percent higher than the national average and even higher than the per capita expenditures observed for the largest municipalities. Table 4.15 Municipal Expenditures by Municipal Size, 2003 (Billion of Reais of 2004) 0 to 5,000 5,000 to 20,000 to 150,000 to More than 20,000 150,000 1,000,000 1,000,000 Brazil Current Expenditures 3,91 16.0 32.5 28.8 29.9 111.2 Capital Expenditures 0.55 2.24 5.16 3.78 4.59 16.3 Total 4.45 18.3 37.7 32.6 34.5 127.5 Current Expenditures per capita (R$) 909 564 543 624 845 639 Capital Expenditures per capita (R$) 127 79 86 82 130 94 Total ­ per capita (R$) 1036 644 629 706 975 732 Sample of 4,965 municipalities. 4.66 The analysis of the evolution of the main components of current expenditures reveals that personnel expenditures and interest payments experienced the strongest growth, 40 percent and 37 percent, respectively, while other current expenditures have stagnated. 4.67 The growth of personnel expenditures can be explained by two main reasons. First, the constitutional requirement to achieve minimum levels of expenditure in health and education (15 percent and 25 percent of net current revenue) forced municipalities to contract new staff in both sectors. Small municipalities were more affected by the compliance of these constitutional requirements. The levels of expenditure in health and education of small municipalities observed in the previous years were lower than the required by this constitutional norm. Thus, the accomplishment of these requirements forced small municipalities to accelerate the increase of expenditures in both sectors that are labor-intensive. As a consequence, personnel expenditures of small municipalities suffered an upward pressure generated by the constitutional obligation of minimum expenditure levels in health and education. 4.68 Second, the increasing imbalance of the social security system of municipal employees determined the increase of personnel expenditures. In this case, large municipalities are most affected by the weight of the social security benefits to retired employees42. 4.69 Thus, the containment of the growth of personnel expenditures which is one of the main challenges of municipalities confronts serious obstacles that threat their fiscal sustainability. Despite the fact that most of the municipalities are reasonably below the Fiscal Responsibility Law ceiling for the personnel expenditures to net current revenue ratios, the increasing trend of this indicator observed in the last years (see Table 4.3 above) needs to be curbed. As mentioned above, other constitutional norms as the establishment of minimum levels of expenditures in certain sectors and the impossibility of reducing the deficit of the social security system for municipal employees constitute obstacles that act against the accomplishment of the FRL requirements in this issue. 42Personnel expenditures figures would be underestimated as many municipalities are classifying personnel expenditures in other current expenditures under the category of outsourced services. Inputs to a Strategy for Brazilian Cities Page 66 Box 4.3 Social Security Imbalance in Municipal Finances Social security imbalance constitutes a major risk for the financial situation of the three the levels of federation in Brazil. In the case of municipalities, the aggregated deficit of the social security system for municipal employees has followed an increasing path reaching 0.4 percent of GDP in 2003. At the individual level, the municipal lending projects that are being implemented by the World Bank, have revealed very large social security actuarial deficits. For example, the municipal governments of Belo Horizonte had in 2004 an actuarial deficit of R$ 2.5 billion or 120 percent of net current revenue; Uberaba of R$ 100 million or 40 percent of net current revenue and the municipality of São Paulo of R$ 34 billion or 270 percent of net current revenues. The high actuarial deficits indicate a very likely increase in personnel expenditures which will put additional pressure on municipal finances harming the ability to generate current account savings. As the social security imbalances is a major concern for the overall public sector, in the last years, two Constitutional Amendments tried to reduce this imbalance, changing social security rules (the Constitutional Amendment n 20, in 1998 and the Constitutional Amendment n 41 in 2003). The objective of both constitutional amendments was to reduce the long run deficit of the socials security system through the increase of the retirement age, the time of contribution and the imposition of a new system for new entrants. All these measures will have medium and long run effects that will reduce the actuarial deficit of the system. For the short run, the Constitutional Amendment No 41 established the possibility of the collection of social security contributions charged over retirement and pension benefits. Benefits lower than a threshold level will be exempted from this contribution. This floor is the same for federal, state and municipal governments. The establishment of this contribution represented a great advance to the federal government, but, for states and especially for municipalities the gain is much smaller. Given that municipal social security benefits are lower than the conceded at the federal and the state levels, most of the municipal social security benefits will be exempted from the contribution, reducing the gain of municipal governments on this matter. The high actuarial deficits of the social security systems for the three levels of government suggest the need for another round of constitutional reform on this area. 4.70 Interest payments also experienced a strong growth. The reason for this increase was the carrying out of the debt renegotiation agreements signed by the National Treasury (STN) with 180 municipal governments in 2001 (MP 2185-38-01). The agreements consist in the refinancing of municipal debt by the federal government and the rescheduling of debt obligations with the National Treasury in better conditions than the observed in the period previous to the debt renegotiation. Despite these agreements set a limit of 13 percent of net revenues for debt service obligations, allowing the capitalization of interest payments that exceed this limit, in fact these agreements implied higher interests payments by the municipalities that had debt refinancing agreements with the National Treasury as in the past municipalities capitalized interest debt almost integrally. 4.71 Other current expenditures were maintained constant. This category includes goods and services purchases (operating costs excluding personnel expenditures). The stagnation of this category is surprising as municipalities have outsourced services shifting expenditures that were previously considered personnel expenditure to the other current expenditures category. Other municipalities simply re-classified part of the payroll bill of the new staff contracted in the health sector as other current expenditures. Thus, operating expenditures that are genuinely non personnel expenditures (material, goods and non personnel services purchases) may have suffered a compression to compensate the increase of personnel services included in this category. 4.72 At this respect, it is important to mention that the rationalization of government purchases trough the improvement of procurement methods, the use of electronic tools and other management measures have been Inputs to a Strategy for Brazilian Cities Page 67 progressively adopted by large municipalities and can justify part of the reduction of operating costs. However, the advance of personnel expenditures included in the other current expenditures reinforces the concern of the increasing and excessive level of personnel expenditures in municipal finances. Table 4.16 Municipal Current Expenditures, 2000-2004 (Billion of Reais of 2004) 2000 2001 2002 2003 2004 Current Expenditures 80.4 85.6 89.8 90.2 95.1 Personnel 34.2 36.0 45.5 45.7 47.9 Interest Payments 1.70 2.23 2.28 2.28 2.34 Other Current Expenditures 44.5 47.4 41.9 42.2 44.8 Sample of 3,028 municipalities. 4.73 The stratification of municipalities by population size shows that the largest municipalities have experienced the highest growth of current expenditures, having contributing negatively to the overall fiscal situation of the municipal government level. The good revenue performance of medium size municipalities allowed them to expand current expenditures. Table 4.17 reveals that personnel expenditures and interest payments experienced a very strong growth in the largest municipalities which was compensated by a strong compression of other current expenditures. In any case, Table 4.17 also confirms that expansion of personnel expenditures is a common problem for all municipalities independent of their size and that interest payments from municipal debt are concentrated in the largest municipalities. Table 4.17 Current Expenditures Growth by Municipal Size, 2000-2004 (%) 0-5000 5000-20000 20000- 150000- More than 150000 1000000 1000000 Brazil Current Expenditures 13.37 15.15 23.78 14.00 19.32 18.29 Personnel 20.26 25.37 32.09 31.68 66.87 39.93 Interest Payments -12.44 -28.37 10.68 10.29 50.27 37.77 Other Current Exp 7.80 6.56 15.90 -0.31 -10.20 0.87 Sample of 4,965 municipalities. 4.74 In per capita terms, again the smallest and largest municipalities have per capita current expenditures that are 50 percent higher than the Brazilian average. Figure 4.4 reveals that with the exception of interest payments, the same is observed for the different components of current expenditures. As mentioned above the absence of economies of scale on small jurisdictions justifies the high per capita personnel expenditures of the smallest municipalities. On the other side, the heavy weight of social security benefits to retired municipal employees justifies the high per capita personnel expenditures in the largest municipalities. 4.75 Other current expenditures are also much higher in the smallest and largest municipalities than national averages. Finally, Figure 4.4 shows that interest payments have relevance only for the largest municipalities. Inputs to a Strategy for Brazilian Cities Page 68 Figure 4.4: Current Expenditure Per Capita by Municipal Size, 2003 (Reais of 2004) 1000 900 800 700 600 500 400 300 200 100 0 0 to 5,000 5,000 to 20,000 to 150,000 to More than Brazil 20,000 150,000 1,000,000 1,000,000 Current Expenditures Personnel Interest Payments Other Current Exp 4.76 On the capital account, debt amortizations and investment expenditures experienced a strong growth (54 percent and 29 percent, respectively), while other capital expenditures suffered a significant fall of 68 percent. 4.77 As mentioned above, the debt refinancing agreements signed by 180 municipalities with the National Treasury explains the increase of municipal governments' amortization expenditures. Even with the favorable conditions of municipal debt rescheduling which set a ceiling for debt services, the agreements have forced municipalities to, differently from the previous period when municipalities had been rolling over their debts, to pay on time their debt amortizations to the National Treasury. Investment expenditures followed a very volatile pattern. 4.78 Table 4.18 shows that investment suffered a strong fall in 2001 reaching their minimum level (R$ 8.3 billion) during the period. In 2002, municipal investments substantially grew by 50 percent achieving their maximum level (R$ 12 billion) in the last five years. In 2003, investment expenditures fell by 20 percent. Finally, in 2004, the strong improvement of revenues allowed the recovery of expenditures investments. 4.79 The volatility of investment expenditures, the fall of other capital expenditures and the containment of other current expenditures imply that these categories can be more easily adjusted by municipal governments as the other main categories, personnel expenditures, interest payments and amortizations are of mandatory nature, having their own dynamics being out the control of municipal authorities. Table 4.18 Municipal Capital Expenditures, 2000-2004 (Billion of Reais of 2004) 2000 2001 2002 2003 2004 Capital Expenditures 12.2 11.1 15.4 13.5 14.7 Investment 9.55 8.39 12.8 10.8 12.3 Amortization 1.34 1.72 1.96 1.94 2.06 Other 1.30 1.02 0.68 0.73 0.42 Sample of 3,028 municipalities. 4.80 The stratification by population size shows that investment expenditures experienced a larger increase in medium size municipalities. Given that medium size municipalities experienced a substantial increase of current revenues and generated large current account savings, they were able to increase investment expenditures. The lack of access to credit explains the stagnation of investments in small municipalities. On the contrary, the access Inputs to a Strategy for Brazilian Cities Page 69 to credit operations (even if restricted) allowed largest municipalities to increase investment expenditures, even when current balances have deteriorated. 4.81 Debt amortizations increased in all municipalities but the increase was stronger in larger municipalities which confirm the enforcement of the contracts agreed between the National Treasury and 180 medium and large municipalities. Finally, cuts in other capital expenditures were applied in a similar magnitude independent of the size of municipalities. Table 4.19 Capital Expenditures Growth by Municipal Size, 2000-2004 (%) 0-5,000 5,000- 20,000- 150,000- More than 20,000 150,000 1,000,000 1,000,000 Brazil Capital Expenditures -2.46 8.42 33.82 27.10 12.89 20.93 Investment 0.33 9.95 38.03 30.90 28.23 28.35 Amortization 0.38 25.05 56.53 44.67 72.90 53.33 Other -63.74 -71.06 -74.75 -56.84 -67.91 -67.34 Sample of 4,965 municipalities. 4.82 Per capita capital expenditures present the same pattern observed in the revenue side and in the current expenditures. The smallest and the largest municipalities have total capital expenditures and investment expenditures that are about 50 percent higher than the national average figures. Differently, amortization payments and other capital expenditures are much higher in the largest municipalities. Figure 4.5 Capital Expenditures Per Capita by Municipal Size, 2003 (R$ of 2004) 140 120 100 80 60 40 20 0 0 to 5,000 5,000 to 20,000 to 150,000 to More than Brazil 20,000 150,000 1,000,000 1,000,000 Capital Expenditures Investments Amortizations Other 4.83 The higher growth of personnel expenditures, interest payments and amortizations when compared with the growth of investments and the compression of other current and capital expenditures reflects the increasing rigidity of municipal expenditure composition. Figure 4.4 shows the increasing importance of personnel expenditures and the debt service obligations which make municipal expenditures more rigid. The share of these mandatory expenditure items increased from 40 percent in 2000 to 48 percent in 2004. Given that an increasing part of other current expenditures is becoming mandatory (for example the inclusion of some personnel expenditures, mentioned above), the expenditure rigidity would be more severe than at a first look. Inputs to a Strategy for Brazilian Cities Page 70 4.84 On the opposite direction, the share of discretionary expenditures that includes investments and other capital expenditures as well as part of other current expenditures has been progressively reduced. 4.85 The increasing expenditure rigidity constitutes a real threat for the consolidation of the municipal fiscal adjustment and the continuity of municipal goods and service provision. As described above, the fiscal adjustment of municipalities was based on revenue increases and the control of expenditures categories that are more easily controlled by municipal authorities as investments and part of other current expenditures. 4.86 In summary, the sustainability of the fiscal adjustment of municipalities will depend on the ability to curb the increasing trend of mandatory expenditures. In particular, the containment of personnel expenditures will allow municipalities to enhance their ability to generate current account savings to finance investment expenditures. 4.87 Given the lack of access to credit operations and the fall of capital transfers from state and federal governments positive current account balances constitutes the only source to finance investment expenditures for small municipalities. Thus, on the expenditure side, the high level of personnel expenditures per capita suggests that small municipalities can control the expansion of this expenditure category without affecting municipal services delivery. 4.88 The heavy weight of the social security benefits to retired personnel and survivals and of debt services payments also reinforces the need for a strict control of personnel expenditures and for the rationalization of municipal goods and services purchases in large municipalities. Figure 4.6 Municipal Expenditure Composition, 2000-2004 100% 80% 60% 40% 20% 0% 2000 2001 2002 2003 2004 Personnel Interest Payments Amortizations Other Current Exp Investments Other Capital Exp Inputs to a Strategy for Brazilian Cities Page 71 Box 4.4 Budget Rigidity Budget rigidity arises from both sides of the budget. On the revenue side, constitutional or legal revenue earmarking mechanisms guarantee that certain type of government revenues should finance specific expenditures. As a result, it is forbidden the use of any resource legally earmarked to specific expenditures to other type of expenditure. Intergovernmental transfers and earmarking mechanisms to guarantee the financing of education and social security and assistance expenditures are the most relevant sources of revenue rigidity. On the expenditure side, there are expenditures whose execution is considered mandatory and include legal (constitutional) obligations of the government such as interest payments personnel expenditures and entitlements such as social security and assistance benefits. Of course, there are some overlaps between the revenue rigidity and the expenditure rigidity which mean that there is a portion of mandatory expenditure that is financed with earmarked revenues. The degree of budget rigidity is extremely high. In 2003, for example, 80% of the federal revenues were earmarked for specific purposes, whereas mandatory expenditures represented 89% of total non-financial expenditures by the federal government. Taking into account the overlapping mentioned above, the "fully free" portion of the executed budget in 2002 was just 4.5% of the non-financial revenues, or 8.1% of non-financial expenditures. Municipalities also suffer a strong rigidity. On the revenue side, the constitutional norms establishing minimum levels of expenditure in specific sectors (15 percent of municipal revenues in health and 25 percent in education) are increasing the budget rigidity. On the expenditure side, the increase of personnel payments observed in small and medium size municipalities also contributed to increase the expenditure rigidity at the municipal level. Additionally, the weight of debt services in highly indebtedness municipalities also contributes to increase municipal budget rigidity. Whatever the way to measure the budget rigidity (portion of earmarked revenues on total revenues or portion of mandatory expenditures on total expenditures) and taking into account the overlap between both types of rigidities, it is clear that the portion of the budget that the government can allocate "freely" within an annual budget process is quite limited. Municipal Expenditures by Function 4.89 Analyzing the evolution of municipal expenditures by economic functions for the period 2002-2004 it is possible that the municipal expenditure composition have not suffered significant changes. Over-head expenditures increased their participation on overall municipal expenditures from 28 percent to 29 percent of municipal expenditures while infrastructure functions reduced their share on municipal expenditures from 20.5 percent to 20 percent. Social expenditures preserved its participation at about 50% of municipal expenditures. The higher participation of overhead expenditures was the result of their higher expansion (6.5 percent) compared with the growth observed for total expenditures 4.2 percent). On the opposite direction the low growth of infrastructure expenditures (1.3 percent) justifies the fall in their share on municipal expenditure. As social expenditures grew at the same rate than total expenditures, they maintained their participation on municipal expenditures. Inputs to a Strategy for Brazilian Cities Page 72 Table 4.20 Municipal Expenditures by Function, 2002-2004 (Billion of Reais of 2004) Function 2002 % of % of % of % Growth total 2003 total 2004 total 02/04 Over-Head 30.4 28.3% 29.9 28.4% 32.3 29% 6.5 Administration 19.6 18.3% 19.2 18.2% 20.2 18.2% 3.1 Social Security 5.5 5.1% 5.4 5.1% 6.6 5.9% 19.6 Interest Payments 5.2 4.9% 5.3 5.0% 5.5 5.0% 5.7 Social 52.4 48.9% 52.2 49.5% 24.6 48.9 4.2 Health 22.1 20.6% 22.3 21.2% 24.0 21.5% 8.6 Education 26.1 24.3% 25.8 24.5% 26.5 23.7% 1.5 Social Assistance 3.1 2.9% 3.2 3.0% 3.1 2.8% 1.3 Housing 1.1 1.1% 0.9 0.9% 1.0 0.9% -13.3 Infrastructure 22 20.5% 20.8 19.7 22.3 19.9 1.3 Urbanism 12.8 11.9% 12.3 11.7% 13.5 12.1% 5.5 Sanitation 3.4 3.2% 3.0 2.9% 2.8 2.5% -16.3 Other Infrastructure* 5.8 5.4% 5.5 5.2% 5.9 5.3% 2.2 Other** 2.5 2.3% 2.5 2.3% 2.5 2.2% 0 Total 107.2 100.0 105.3 100.0 111.7 100.0 4.2 *Other infrastructure aggregates expenditures on agriculture, environment, energy, industry and commerce, communications and transport. ** Other expenditures are public security, science and technology, citizenship rights and sport and leisure Sample of 3198 municipalities. 4.90 Despite some stability on municipal expenditures composition among broad functions, strong changes can be observed for less aggregated categories. In particular, it is remarkable the high growth of social security benefits (almost 20 percent) which has led to the increase in the participation of this type of expenditure from 5 percent to 6 percent of municipal expenditures. On the social areas, health expenditures experienced a significant growth (9%) which increased their share on municipal expenditures while education expenditures' participation fell slightly. Also, the strong fall on housing expenditures has reduced the participation of this function on municipal expenditures. For the infrastructure side, it is worth to remark the strong fall on sanitation expenditures (-16%) which explains the decreasing importance of this function on municipal expenditures. 4.91 Analyzing the municipal expenditure composition according to the size of municipalities it is possible to observe substantial differences in the way in which municipalities allocate their resources among functions. Large municipalities allocate almost one third of their expenditures to overhead costs. Medium size municipalities allocate more than 50 percent of their expenditures to social areas and small municipalities proportionally dedicate more resources to infrastructure sectors. Inputs to a Strategy for Brazilian Cities Page 73 Table 4.21 Municipal Expenditures Composition by Municipal Size (%) 150,000 More 5,000 to 20,000 to to than Function 0 to 5,000 20,000 150,000 1,000,000 1,000,000 Brazil Over-Head 27.2% 25.4% 25.8% 28.1% 31.2% 27.8% Administration 23.8% 21.7% 20.8% 20.3% 12.8% 18.7% Social Security 1.8% 1.8% 2.4% 4.0% 9.3% 4.6% Interest Payments 1.6% 1.9% 2.6% 3.8% 9.1% 4.5% Social 49.7% 54.4% 53.0% 51.6% 45.2% 50.58% Health 18.4% 19.4% 20.1% 24.7% 20.1% 21.1% Education 26.5% 30.3% 29.0% 23.4% 21.3% 25.6% Social Assistance 4.0% 4.1% 3.5% 2.7% 2.5% 3.1% Housing 0.7% 0.7% 0.5% 0.8% 1.4% 0.8% Infrastructure 21.6% 18.6% 19.1% 17.8% 21.3% 19.4% Urbanism 7.0% 8.4% 11.2% 11.8% 13.6% 11.5% Sanitation 1.6% 1.8% 3.0% 3.2% 2.9% 2.8% Other Infrastructure 13.0% 8.4% 5.0% 2.8% 4.9% 5.2% Other 1.6% 1.5% 2.1% 2.5% 2.4% 2.2% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Sample of 3198 municipalities. 4.92 Within each broad category significant differences can also be observed. Small municipalities allocate a significant share of their resources to administration expenditures while social security benefits and interest payments are very important in large municipalities. This finding confirms the pattern observed above which indicates that small municipalities has heavy active personnel expenditures while large municipalities have heavy retired personnel expenditures. Also, Table 4.21 confirms that debt burden is only important in large municipalities. 4.93 In social areas, it is possible to observe strong differences in the allocation to the education area. Municipalities with more than 5,000 to 150,000 inhabitants allocate 30 percent to education expenditures while the largest municipalities allocate only 21 percent to education. With the exception of medium to large municipalities (between 150 thousand to 1 million inhabitants), health expenditures represent about 20 percent of municipal expenditures. Housing expenditures are very low, but the largest municipalities allocate a substantially higher share to this type expenditure than the rest of municipal governments. 4.94 In the infrastructure sectors, it is important to note that the high share of infrastructure expenditures by the smallest municipalities is due to the heavier allocation (13 percent) of municipal resources to other infrastructure sectors (agriculture, environment, energy, etc). For urbanism and sanitation expenditures it is possible to observe a positive correlation between municipality size and the share dedicated to both infrastructure sectors. 4.95 In per capita terms, it is possible to observe the same pattern identified in the previous sections: in all the functions analyzed the smallest and largest municipalities have per capita expenditures that are much higher than Inputs to a Strategy for Brazilian Cities Page 74 the Brazilian average. Table 4.22 reveals that the smallest and largest municipalities have overhead expenditures 50 percent higher than the observed in all Brazilian municipalities. In infrastructure sectors, this figure achieves 75%. As mentioned above the absence of economies of scale on small jurisdictions justifies the high per capita expenditures on overhead and infrastructure. Differences in municipal social expenditures per capita are less pronounced. Smallest and largest municipalities spend in per capita terms 25 percent more than the Brazilian municipalities' average. 4.96 More detailed analysis show that smallest municipalities have very high administration expenditures per capita while in large municipalities, the high levels of expenditure in social security and interest payments (3 times the national average in both cases) explain the high per capita expenditures in overhead functions. 4.97 The four sub sectors of social areas exhibit the same pattern mentioned above with the smallest and largest municipalities spending more resources in these areas. In the infrastructure sector, again it is noteworthy to mention that other infrastructure expenditures are responsible for the high expenditure per capita of the smallest municipalities. Differently, urbanism and sanitation expenditures per capita maintain the same positive correlation with the size of municipalities as the observed in absolute terms (see Table 4.20). Table 4.22 Municipal Expenditures Per Capita by Municipal Size -2003 (R$ of 2004) 150,000 More 5,000 to 20,000 to to than Function 0 to 5,000 20,000 150,000 1,000,000 1,000,000 Brazil Over-Head 309 179 177 217 332 223 Administration 269 153 143 157 136 150 Social Security 21 12 16 31 99 36 Interest Payments 18 13 18 29 97 36 Social 561 383 364 398 482 405 Health 208 136 138 191 214 169 Education 301 213 199 180 227 205 Social Assistance 45 29 24 21 26 25 Housing 8 5 3 6 15 7 Infrastructure 244 131 131 137 227 155 Urbanism 80 59 77 91 144 92 Sanitation 17 12 20 25 31 22 Other Infrastructure 147 59 34 22 52 41 Other 18 11 15 19 25 18 Total 1,132 703 688 771 1,066 800 Sample of 3198 municipalities. Inputs to a Strategy for Brazilian Cities Page 75 The evolution of municipal indebtedness and the restricted access to credit43 4.98 As a result of the fiscal discipline, the enforcement of the Fiscal Responsibility Law and the hard credit access restrictions, Brazilian municipalities have maintained their indebtedness indicators at low levels. The Fiscal Responsibility Law indebtedness indicator, the net consolidated debt to net current revenue ratio increased slightly from 0.51 in 2000 to 0.54 in 2004. The good revenue performance expressed in a growth of 17 percent between 2000 and 2004 also contributed to keep this indebtedness ratio at a level well below the legal limit of 1.244. 4.99 In absolute terms, from 2000 to 2004 the municipal consolidated debt grew by 32 percent in real terms. However, it is important to mention that this growth was concentrated in 2002 when it increased 17 percent. The exchange rate shock of this year was the main responsible for the strong debt increase45. In 2003, the municipal consolidated debt fell 5 percent as a result of the strong fiscal adjustment and the lack of access to credit operations. In 2004, the improvement of the economic situation and some relaxation of the restrictions to credit access allowed municipalities to contract credit operations that resulted in an increase of consolidated debt of 6 percent. 4.100 The most important component of municipal debt, the domestic contractual debt (share of 62 percent), has increased by 28 percent from 2000 to 2004. However, as mentioned above, this debt, which is basically debt with the federal government (National Treasury and Federal Banks), reached its highest level in 2002 due to the exchange rate devaluation (see previous footnote). 4.101 External debt is the smallest component of municipal debt (share of 5 percent) and it suffered a strong fall of 20 percent for the whole period that was accelerated after the devaluation of 2002 (from 2002 to 2004, it declined by 35 percent). 4.102 Table 4.23 shows that the social security contributions in arrears (share of 6 percent), other liabilities (share of 16 percent) and floating debt (share of 12 percent) have exhibited a strong volatility. In the case of social security contributions in arrears, the increasing coverage of municipal accounts (the progressive inclusion of indirect administration entities) resulted in the incorporation of debt with the National Institute of Social Security of these indirect entities by their respective municipalities. Currently, the federal government is negotiating with the Brazilian municipalities' association an overall term for the repayment of this debt. The specific terms of this negotiation should define the effect of the recognition of these liabilities on future debt service payments46. 4.103 In the case of other liabilities which are basically debt with suppliers, the growth of this type of debt reflects its utilization as a finance mechanism given the difficult access to credit operations. Besides the normal budget execution problems that result in floating debt, the importance of this type of debt can be also explained by its increasing utilization as an informal way of financing. 4.104 In terms of net consolidated debt, as the financial assets hold by municipalities strongly increased during 2000 to 2004, the growth of the net consolidated debt was 24 percent lower than the observed increase of consolidated debt. Municipalities have accumulated large financial assets which explain the fall of net consolidated debt. Positive financial results explain the accumulation of financial assets. Given the high interest rates paid by the federal government bonds, the financial revenues have became an important source of current revenues for subnational governments. 43Given the lack of detailed information on debt, the information presented in this section was produced by the WB staff using information of the Patrimonial Balances of municipalities. 44. Brazilian GDP accumulated growth from 2000 to 2004 was 9 percent, thus municipal revenues grew 90 percent above real GDP. 45. As mentioned in Box 4.1, debt stock of the most indebted municipalities (São Paulo, Rio among the most important) is very influenced by exchange rate. In 2002, São Paulo net consolidated debt rose by 20.2% in constant prices, mainly due to the exchange rate devaluation which promoted a big increase of the General Price Index (IGP) which is the index used to correct the municipal debt with STN. 46. The contingent nature of this type of debt recommends some warning with the accuracy of the numbers. Inputs to a Strategy for Brazilian Cities Page 76 Table 4.23 Municipal Consolidated 2000-2004 (Billion of R$ of 2004) 2000 2001 2002 2003 2004 Consolidated Debt 54,0 60,1 70,3 66,8 71.0 Contractual 37.8 38,7 50.0 44,7 47,1 Domestic 34,7 35,4 43,2 41,9 44,6 External 3,1 3,3 3,7 2,8 2,4 Social Security contributions in arrears 3.0 1,9 3,2 7,7 3,5 Other liabilities 4,2 10,9 12,7 6,9 13.6 Floating Debt 8,9 8,5 7,4 7,4 6,8 Deductions 7,4 13.6 15.0 13.6 13.2 Financial Assets (7,33) (13,6) (15,0) (13,6) (13,2) Loans (0,02) (0,02) (0,03) (0,02) (0,02) Net Consolidated Debt 46,6 46,4 55,3 53,2 57,8 Net Current Revenue 92.0 98.4 101,3 99,5 107,6 Consolidated Net Debt/Net Current Revenue 0.51 0.47 0.55 0.53 0.54 Sample3,028 municipalities. Figure 4.7 Municipal Deb Composition (% average 2000-04 140 120 100 80 60 40 20 0 0 to 5,000 5,000 to 20,000 to 150,000 to More than Brazil 20,000 150,000 1,000,000 1,000,000 Capital Expenditures Investments Amortizations Other 4.105 Analyzing the municipal debt by the size of municipalities, it is possible to confirm that indebtedness is highly concentrated in the largest municipalities which account for 66 percent of total municipal consolidated debt and 77 percent of total municipal net consolidated debt. Inputs to a Strategy for Brazilian Cities Page 77 4.106 The high concentration of debt in large municipalities also is confirmed in per capita figures. The consolidated debt per capita of the largest municipalities is more than ten times the per capita debt of the smallest ones. The net consolidated debt per capita of the largest municipalities is 40 times the observed for the smallest one. Also, it is important to mention that differently from most of the revenue and expenditure per capita figures where largest and smallest municipalities had the highest levels, in the case of per capita debt there is a positive relationship between indebtedness per inhabitant and the size of municipalities. 4.107 In terms of debt composition, given the lack of access to credit, the smallest municipalities the main component of the debt of small municipalities are the floating debt and other liabilities. For this type of municipalities, domestic contractual debt is very low and they do not have external debt. Differently, the debt of medium to large municipalities rely is mainly domestic contractual debt. For the largest municipalities domestic contractual debt represents 74 percent in the largest municipalities and external debt accounts for 6 percent of the consolidated debt of the largest municipalities. The social security debt has a similar weight in all the groups, varying between 10 and 15 percent of municipal debt. Table 4.24 Municipal Consolidated Debt by Municipal Size -2003 (R$ of 2004) 0 to 5,000 5,000 to 20,000 to 150,000 to More than Brazil 20,000 150,000 1,000,000 1,000,000 Consolidated Debt 0.536 3.210 8.520 12.1 48.0 72.4 Contractual 0.056 0.407 1.976 5.536 38.24 46.14 Domestic 0.055 0.402 1.880 5.410 35.6 43.3 External 0.001 0.005 0.076 0.126 2.64 2.84 Social Security contributions in arrears 0.074 0.552 1.090 0.138 5.50 8.54 Other liabilities 0.104 0.806 2.590 2.700 2.08 8.28 Floating Debt 0.302 1.440 2.940 2.520 2.24 9.44 Deductions 0.422 1.609 4.894 4.748 4.53 16.19 Financial Assets (0.422) (1.600) (4.890) (4.580) (4.52) (16.0) Loans (0.000) (0.009) (0.004) (0.168) (0.01) (0.19) Net Consolidated Debt 0.114 1.600 3.630 7.380 43.5 56.2 Net Current Revenue 3.943 16.21 33.95 29.42 29,68 113.2 Consolidated Net Debt/Net Current Revenue 0.03 0.1 0.1 0.25 1.47 0.50 Memo items: Consolidated debt per capita 136.31 123.51 155.54 286.87 1,485.55 454.65 Net Consolidated debt per capita 28.98 61.60 66.18 174.59 1,345.59 352.97 Sample of 4,965 municipalities. 4.108 The problem of excessive indebtedness, expressed in a net consolidated debt to net current revenue ratio higher than the FRL ceiling of 1.2, also affects a very few number of municipalities. In fact, in 2003 only 41 out of 4,965 municipalities or 0.8 percent had the indebtedness FRL indicator above the legal limit. From this sample, 4,894 municipalities or 98.5 percent of Brazilian municipalities have an indebtedness ratio lower than 1 and 4,548 municipalities or 92 percent have a FRL debt indicator of 0.5, well below the legal limit of 1.2. Inputs to a Strategy for Brazilian Cities Page 78 Table 4.25 Number of Municipalities by Net Consolidated Debt to Net Current Revenue Ratio DCL/RCL 0 to 5,000 5,000 to 20,000 to 150,000 to More than Brazil 20,000 150,000 1,000,000 1,000,000 0.00 ­ 0.25 1,024 1,888 885 76 5 3,878 0.25 ­ 0.50 111 316 214 27 2 670 0.50 ­ 0.75 22 129 80 11 2 244 0.75 ­ 1.00 8 54 34 5 1 102 1.00 ­ 1.20 3 15 5 6 1 30 More than 1.20 4 15 14 7 1 41 Total 1172 2417 1,232 132 12 4,965 4.109 Restricting to the group of municipalities with more than 150,000 inhabitants which are the relevant group for potential credit operations by the Bank, given the scale of the projects that can be financed, 132 out of the 144 municipalities or 94.5 percent of this group have a FRL debt indicator below the legal ceiling and able to contract credit operations. Thus, the problem of high indebtedness affects only a very restricted group of municipalities. As a result, it is possible to conclude that the Fiscal Responsibility Law ceiling on indebtedness is not a binding constraint for most of the Brazilian municipalities. (See Annex 2 with the complete list of municipalities with population higher than 150,000 inhabitants that are below the FRL limit). 4.110 However, there are other mechanism that effectively constraints the ability of municipalities to contract credit operations. The debt renegotiation agreements that the National Treasury and 180 municipalities agreed under the resolution MP2185 of 2001 condition the contract of credit operations by this set of municipalities. This legal instrument regulates the refinancing of the debt of 180 municipalities, and defines a debt chronogram which prohibits the issuance of municipal bonds during the time of the agreement. 4.111 New credit operations are also conditioned to a satisfactory accomplishment of the chronogram of debt service. In fact, the agreements establish the collateralization of resources to ensure debt service and the corresponding authorization to the federal government to withhold transfers mandated by the Constitution. 4.112 Even more important, in addition to the above controls on the demand for credit given by the FRL and the debt renegotiation agreements, the resolutions of the National Monetary Council (CMN) constitute the most important mechanism for the restraining of the credit supply to public sector entities. Basically, the first resolution, 2827 establishes the 45% exposure limit of domestic financial institutions to the borrowing of the public sector and a global limit of domestic credit to the public sector. Further resolutions altered the article 9 of the Resolution 2827 which defined the global limit and allowed some operations in specific sectors (sanitation and energy infrastructure projects). 4.113 Specifically, the first control channel was to limit the exposure of domestic financial institutions to the public sector borrowers. The exposure limit was set by Resolution 2827 (March 2001) to 45 percent of equity--a limitation that is particularly binding for the Caixa Econômica Federal (CEF) and Banco Nacional de Desenvolvimento Econômico e Social (BNDES) and other state government development banks which constitute the most important source of credit supply for subnational governments. 4.114 The second control channel was to establish a global limit of domestic credit to public entities. The article 9 of the 2827 CMN resolution set a global limit for domestic credit operations to public entities of R$1 billion. Subsequently this limit was further reduced to R$ 200 million by Resolution 2954 of 2002. Exceptions have been made for certain sectors, for example, Resolution 3153 of December 2003 exempted credit operations for sanitation projects from this limit and established a limit of domestic credit operations to public sector entities of R$ 1.1 billion directed to finance sanitation projects. Inputs to a Strategy for Brazilian Cities Page 79 4.115 The very low limits established by the CMN resolution practically eliminated the access to credit operations by municipalities. Demand for credit operations was estimated at about R$ 10 billion, which implies that the infrastructure needs are many times higher than the global limits established by the CMN. 4.116 External borrowing also faces hard constraints. External credit operations need to be approved in a first step by the COFIEX (External Finance Commission) which is composed by the Planning and the Finance ministries who evaluates the adequacy of the projects to be financed. The second step is to obtain the sovereign guarantee by the National Treasury. Any external credit operation to public entities including subnational needs the guarantee of the Treasury. To give the sovereign guarantee, the National Treasury verifies the accomplishment of the targets defined in the debt renegotiation agreements and evaluates the indebtedness and repayment capacity of the borrower. 4.117 In summary, restrictions on the demand and supply side have exerted an efficient control that practically eliminated the access to credit operations by municipalities and were responsible for the slow growth of municipal indebtedness since 2000. Conclusions and Policy Implications 4.118 Since 2000 municipal governments have adopted a sound fiscal performance that has been expressed in the generation of increasing fiscal balances. The improvement of fiscal balances has not been homogeneous among municipal government units. Small and medium size municipalities experienced a stronger improvement than the largest municipalities. In fact, fiscal balances of the municipalities with population superior to 1 million of inhabitants have been reduced. 4.119 The overall fiscal adjustment of the Brazilian public sector, the enactment of the Fiscal Responsibility Law in 2000 and the imposition of strong credit access restrictions have fostered the improvement of municipal fiscal balances. 4.120 Municipal governments have accompanied the adjustment efforts of the federal government to generate increasing primary surpluses, which in turn contributed to the overall improvement in Brazil's fiscal accounts. 4.121 The Fiscal Responsibility Law (FRL) has encouraged the fiscal adjustment of municipalities. The establishment of ceilings for indebtedness, for credit operations, for personnel for expenditures, debt service payments and the observance of the golden rule reinforced the adoption of prudent fiscal stances by municipalities. Besides the consolidation of fiscal responsibility, the Fiscal Responsibility Law promoted the strengthening of fiscal transparency and planning at the municipal level. 4.122 Finally, the imposition of hard domestic credit supply constraints has guaranteed the generation of municipal positive primary results. 4.123 As a result of its general sound fiscal performance and the hard credit access restrictions, the aggregated municipal level has complied with the fiscal limits established by the Fiscal Responsibility Law (FRL). The aggregate municipal FRL indicators are well below the legal ceilings and even more important, most of the FRL indicators did not exhibit a clear deteriorating trend during the last five years. 4.124 Net consolidated debt as a proportion of net current revenue, used for compliance of the Fiscal Responsibility Law, has fluctuated around 35 percent, far below the legal ceiling of 120 percent. Debt service as a proportion of the net current revenue has oscillated around 4 percent, well below the legally-imposed limit of 11.5 percent. The lack of access to credit operations, made the indicator credit operations to net current revenue to vary around the 1 percent level while the corresponding FRL limit is 16%. Differently, despite being below the ceiling of 60 percent, there was observed an increasing trend for personnel expenditures to net current revenue ratio which grew from 43 percent in 2000 to 47 percent in 2004. In fact, the most important risk for municipal finances is the difficulty to curb increasing trend of personnel expenditures. Additionally other financial indicators reveal that most of the Brazilian municipalities have a strong capacity to generate cash flows to face their debt service obligations and to finance a substantial part of their investment expenditures. Inputs to a Strategy for Brazilian Cities Page 80 4.125 In general, small and medium size municipalities have better FRL and financial indicators than the largest municipalities. In particular: the high indebtedness is a problem only for the largest municipalities; the lack of access to credit operations by small municipalities makes them to finance with current account balances the totality of their investment expenditures; the largest municipalities has ability to pay their interest obligations on debt, however credit operations are needed to finance part of their investment needs; the growth of personnel expenditures is a common problem for all municipal governments; however, the source of the excess of personnel expenditures is different. Small municipalities have an excess of active personnel expenditures while the largest ones have excess of social security benefits to retired personnel. 4.126 The improvement in the municipal fiscal balances was based on the good performance of revenues, in particular on the increase of current revenues. The strong increase of municipal current revenues happened in a period of low economic growth which makes more remarkable this achievement. On the contrary, the declining performance of capital revenues reflects the hard credit restrictions imposed by the federal government on credit operations. At the same time, capital transfers from federal and state governments also suffered a strong reduction which resulted from the fiscal adjustment effort of both levels of government. 4.127 The good performance of municipal tax revenues indicates the enhancement of tax collection efficiency. In the last years, many municipal governments launched an aggressive program to enhance the efficiency of the municipal revenue services. Investments in software and the modernization of administrative processes were responsible for the improvement of tax collection effort. 4.128 Medium size municipalities experienced the higher growth of current revenues which resulted from a good tax revenue performance, a significant increase of intergovernmental transfers and an exceptional performance of other current revenues. Small and medium size municipalities have improved their tax collection performance more than the. On the opposite side the largest municipalities' current revenues had a disappointing performance that partially explains the worsening of their fiscal situation. 4.129 Despite the higher increase of tax revenues in small and medium size municipalities there remain striking differences in per capita terms. Tax revenue per capita of the largest municipalities is nine times the tax revenue of the smallest one. On the opposite direction, the redistributive nature of federal transfers and the distortions provoked by the rules of distribution that favored the creation of small municipalities explain the fact that federal transfers per capita in the smallest municipalities are five times the federal transfers per capita to the largest municipalities and three times the transfers per capita to the medium size municipalities. 4.130 As a consequence, the composition of municipal current revenue reveals strong differences. There is a severe dependence of small and medium size municipalities on intergovernmental transfers, with a weight of transfers varying between 91 and 74 percent of current revenues. The dependence on intergovernmental transfers in the municipalities with population higher than 150,000 is much lower varying between 62 and 45 percent of current revenues. 4.131 The high significance of intergovernmental transfers on municipal current revenues represents a major risk on the revenue side in the sense that reductions of tax revenues of the federal and state governments shared with municipalities would have a significant impact on the finances of these recipients units. The constitutional nature of most of the intergovernmental transfers received by municipalities alleviates in some extent the vulnerability of municipal finances derived from the extreme dependence on this revenue source. In any case, and independent of the size of the municipalities, the enhancement of the tax collection effort of the municipalities would increase the ability to generate current savings and expand municipal investments while would reduce the dependence on transfers from higher level of government. Inputs to a Strategy for Brazilian Cities Page 81 4.132 Municipal capital revenues have not only stagnated but also have suffered a strong volatility. Credit operations are highly concentrated in the largest municipalities which responded for 70 percent of them. In fact, small municipalities do not have access to credit operations. However, small municipalities are financed by capital transfers from the federal and state governments. 4.133 The stagnation of capital revenues resulted from the domestic supply credit constraints imposed by the National Monetary Council (CMN) which practically blocked the access to credit operations by municipalities. In turn, the fiscal adjustment of the federal government and state governments limited and made more uncertain the capital transfers from these higher levels of governments to municipalities. 4.134 Municipal expenditures have experienced a lower increase of expenditures than of the observed in the revenue side. The increase of municipal expenditures was commanded by the increase of mandatory expenditures. Personnel expenditures, interest payments and amortizations experienced a strong increase. On the contrary, investment and other current expenditures have been reduced in the last years. 4.135 As a consequence, the evolution of the municipal expenditure composition shows an increasing rigidity on the expenditure side which is characterized by a rising share of mandatory expenditures that is reducing the space of maneuver of municipalities for sustaining their fiscal adjustment effort. In particular, the increasing trend of mandatory expenditures obligates municipalities to raise the tax burden or to cut investment expenditures as the available adjustment options. 4.136 Thus, the increasing expenditure rigidity constitutes a real threat for the consolidation of the municipal fiscal adjustment and the continuity of municipal goods and service provision which requires the expansion of the fiscal space for municipal investment expenditures. 4.137 The rationalization of government purchases trough the improvement of procurement methods, the use of electronic tools and other management measures offer a potential way for the reduction of current expenditures opening space for investment expenditures in a context of hard fiscal constraints. 4.138 The stratification of municipalities by population size reveals a strong concentration of expenditures in the largest municipalities and high expenditures per capital in the smallest and largest municipalities. This finding is valid for all the expenditures categories. 4.139 It is natural that the much higher financial capacity justifies the concentration of municipal expenditures in large municipalities. The high per capita expenditures observed in the smallest municipalities results from the lack of economies of scale as municipal goods and services provision can not justify this high expenditure per capita. In particular, the existence of fixed costs explains the high per capita expenditures in the smallest municipalities which are mostly financed by intergovernmental transfers. 4.140 The prospects for municipal finances in the next years are positive. This expectation is based on two important assumptions: an economic growth higher than population growth and that the responsible fiscal stance will remain. 4.141 A higher economic growth is responsible for a higher rate of increase on revenues than on expenditures, which depends more on population growth. There is a strong consensus indicating a strong improvement of the economic scenario for Brazil in the next years. At the same time, the enhancement of tax revenue collection efficiency can reinforce the positive effect of economic growth on municipal revenues. Additionally, other current revenues as user charges or cost recovery need to be exploited in order to finance municipal investments. 4.142 On the expenditure side, as mentioned above the interruption of the increasing expenditure rigidity constitute the main challenge for the next years. In particular, the reduction of the weight of personnel expenditures on municipal finances and the rationalization of operating costs would expand the current narrow space for municipal investment expenditures. 4.143 Even with the increase of the fiscal space for municipal investment, the large demand for the expansion of the municipal infrastructure and strengthening of service delivery indicates the need for developing instruments to finance municipal investments. The strong fiscal adjustment effort, the development of an institutional framework Inputs to a Strategy for Brazilian Cities Page 82 for fiscal responsibility and the improvement of fiscal transparency and fiscal planning fostered by the Fiscal Responsibility Law would improve the creditworthiness of Brazilian municipal governments opening space for the development of a municipal lending market. 4.144 At this respect, given the strong heterogeneity among Brazilian municipalities a compound approach needs to be implemented. Figure 4.5 depicts the most relevant financial characteristics and constraints of the municipalities by population type that can orient the definition of a lending policy to municipal governments. The criteria used to classify municipalities are: indebtedness, investment coverage, the ability to collect tax revenues, municipal public sector management and the weight of personnel expenditures on municipal finances. Each indicator is measured in each one of the axis of the radar. The rating for each dimension varies from 0 (worst) to 5 (best). Thus, the municipalities that occupy a bigger area of the radar are in a better situation. 4.145 Overall, municipalities with a population size between 150 thousand to 1 million are in a better financial position as they exhibit a more balanced rating schedule. They have a low level of indebtedness, a large ability to generate cash flows to finance investments, an intermediate tax collection effort and management capacity and have the same problem with the excessive weight of personnel expenditures as the rest of Brazilian municipalities. As mentioned above, the largest municipalities have a high indebtedness problem that limits their access to credit operations. Small municipalities have a very low tax revenue collection effort and a poor management capacity. Inputs to a Strategy for Brazilian Cities Page 83 Figure 6: Municipal Financial Typology by population size Large municipalities ( >1mi) Municipalities with pop 150th to 1mi Indebtedness Indebtedness 5 5 4 4 3 3 2 2 Personnel Investment Coverag Personnel Investment Coverage 1 1 0 0 Management Revenue Collection Management Revenue Collection Municipalities with pop 20th to 150th Municipalities with pop 5th to 20th Indebtedness Indebtedness 5 5 4 4 3 3 2 2 Personnel Investment Coverag Personnel Investment Coverag 1 1 0 0 Management Revenue Collection Management Revenue Collection Municipalities with 0 to 5th pop Indebtedness 5 4 3 2 Personnel Investment Coverage 1 0 Management Revenue Collection Inputs to a Strategy for Brazilian Cities Page 84 4.146 In summary, municipalities up to 1 million inhabitants have a large space for further indebtedness and they have ability to generate current account savings that guarantees a strong repayment capacity. Also, lending programs to finance investments in these municipalities would be accompanied by a public sector management component directed to the rationalization of expenditures and the enhancement of revenue collection effort. Annex: Brazilian Municipalities Background Information (to be expanded) 4.147 Brazil is a federal system comprising the federal government, 26 state governments and 5,506 municipalities. Differently from other federal countries, since the Constitution of 1988, municipalities are autonomous entities having the same status enjoyed by state governments. 4.148 The municipal level is dominated by small government units. More than 4,000 municipalities have population below 20,000 inhabitants, 1,300 municipalities have population between 20,000 to 150,000 inhabitants, 132 municipalities have population between 150,000 to 1,000,000 and 12 municipalities have a population above 1,000,000 of inhabitants. Figure 1: Brazilian municipalities by population size, 2004 3,000 2737 2,500 2,000 1,500 1314 1305 1,000 500 132 12 0 0-5,000 5,000-20,000 20,000- 150,000- More than 150,000 1,000,000 1,000,000 4.149 Table 4.26 shows strong disparities in terms of population and GDP. Small municipalities, which represent 75 percent of the municipal units, respond for 19 percent of the Brazilian population and for 14 percent of the national GDP. The municipalities with more than 150,000 inhabitants, which represent 3 percent of Brazilian municipalities, respond for almost 50 percent of population and 52 percent of Brazilian GDP. Table 4.26 Population and GDP by Municipal Size Population in Share of Total GDP ­ 2002 (Billion Share of Total GDP 2000 (million) Population of R$ of 2004) 0-5,000 3.9 2.5 32.5 2.1 5,000-20,000 26.0 16.3 177.3 11.3 20,000-150,000 54.8 34.4 478.6 30.5 150,000-1,000,000 42.3 26.5 469.2 29.9 More than 1,000,000 32.3 20.3 409.1 26.1 Total 159.3 100.0 1,566.7 100.0 Source: IBGE. 4.150 For the analysis of the municipal finances two samples were used. For the analysis of the evolution of municipal finances during the period 2000-2004, given the data availability limitations, a sample of 3,208 municipalities were used to make consistent inter temporal comparisons. These 3,028 municipalities respond for 78 percent of the Brazilian population. For the cross comparisons by municipal population size, it was selected the Inputs to a Strategy for Brazilian Cities Page 85 year of 2003 as it has the largest sample (4,967 municipalities) representing 95 percent of the Brazilian population. Table 4.27 Municipalities with FRL Indebtedness Indicator Below 1.2 Municipality Population Net Net Primary NCD/NCR Debt service / NCR consolidated current Balance debt (NCD) revenue (PB) (NCR) R$ Million (FRL Ceiling= 1.2) (FRL Ceiling=0.115) RIO BRANCO/AC 253,059 57.2 149.0 4.7 0.38 0.01 ARAPIRACA/AL 186,466 18.0 95.7 1.3 0.19 0.01 MACEIO/AL 797,759 17.2 410.4 61.9 0.04 0.05 MANAUS/AM 1,405,835 31.0 816.0 83.0 0.04 0.03 MACAPA/AP 283,308 28.9 126.5 6.4 0.23 0.00 CAMACARI/BA 161,727 13.3 235.4 -10.7 0.06 0.06 FEIRA DE 480,949 63.8 150.3 7.6 0.42 0.02 SANTANA/BA ILHEUS/BA 222,127 45.7 90.0 8.2 0.51 0.03 ITABUNA/BA 196,675 111.1 106.1 10.6 1.05 0.05 VITORIA DA 262,494 57.9 116.3 4.7 0.50 0.03 CONQUISTA/BA CAUCAIA/CE 250,479 -12.2 99.0 10.0 (0.12) 0.01 FORTALEZA/CE 2,141,402 -76.8 1,357.1 -49.3 (0.06) 0.02 JUAZEIRO DO 212,133 19.7 81.0 1.4 0.24 0.02 NORTE/CE MARACANAU/CE 179,732 -15.4 131.1 14.8 (0.12) 0.01 SOBRAL/CE 155,276 13.1 122.3 0.7 0.11 0.01 CACHOEIRO DE 174,879 32.6 87.3 6.4 0.37 0.04 ITAPEMIRIM/ES CARIACICA/ES 324,285 37.2 96.1 -0.5 0.39 0.02 SERRA/ES 321,181 126.6 201.1 19.0 0.63 0.03 VILA VELHA/ES 345,965 60.8 162.0 -7.8 0.38 0.02 VITORIA/ES 292,304 68.1 422.6 26.7 0.16 0.03 ANAPOLIS/GO 288,085 50.9 150.0 14.3 0.34 0.04 APARECIDA DE 336,392 0.2 117.4 12.7 0.00 0.01 GOIANIA/GO GOIANIA/GO 1,093,007 492.6 915.1 42.1 0.54 0.02 IMPERATRIZ/MA 230,566 35.8 101.3 3.3 0.35 0.02 SAO LUIS/MA 870,028 108.8 544.7 12.2 0.20 0.02 BELO 2,238,526 884.6 2,057.4 26.1 0.43 0.04 HORIZONTE/MG BETIM/MG 306,675 65.7 385.5 16.3 0.17 0.03 DIVINOPOLIS/MG 183,962 39.4 115.9 2.6 0.34 0.03 GOVERNADOR 247,131 12.0 158.6 3.2 0.08 0.02 VALADARES/MG IPATINGA/MG 212,496 -6.2 195.2 20.7 (0.03) 0.05 JUIZ DE FORA/MG 456,796 51.2 337.1 28.2 0.15 0.02 MONTES 306,947 73.3 154.4 10.1 0.47 0.03 CLAROS/MG RIBEIRAO DAS 246,846 7.9 60.6 -14.9 0.13 0.03 NEVES/MG SANTA LUZIA/MG 184,903 26.9 76.4 -6.9 0.35 0.02 SETE LAGOAS/MG 184,871 47.8 105.4 7.8 0.45 0.05 UBERABA/MG 252,051 16.4 202.1 7.7 0.08 0.02 UBERLANDIA/MG 501,214 23.5 376.6 -9.2 0.06 0.02 CAMPO 663,621 46.4 561.6 25.4 0.08 0.02 GRANDE/MS Inputs to a Strategy for Brazilian Cities Page 86 Municipality Population Net Net Primary NCD/NCR Debt service / NCR consolidated current Balance debt (NCD) revenue (PB) (NCR) DOURADOS/MS 164,949 98.7 117.6 6.7 0.84 0.03 CUIABA/MT 483,346 344.7 342.7 44.1 1.01 0.10 RONDONOPOLIS/M 150,227 47.2 109.8 10.1 0.43 0.04 T VARZEA 215,298 44.0 118.4 17.9 0.37 0.04 GRANDE/MT ANANINDEUA/PA 393,569 6.9 102.5 9.6 0.07 0.01 BELEM/PA 1,280,614 28.0 679.0 21.7 0.04 0.02 MARABA/PA 168,020 4.4 92.0 1.0 0.05 0.02 SANTAREM/PA 262,538 5.7 97.7 19.9 0.06 0.02 CAMPINA 355,331 -15.7 161.7 -55.0 (0.10) 0.04 GRANDE/PB JOAO PESSOA/PB 597,934 47.4 366.5 43.2 0.13 0.02 CABO DE SANTO 152,977 5.8 102.4 10.4 0.06 0.01 AGOSTINHO/PE CARUARU/PE 253,634 2.0 91.0 -1.2 0.02 0.01 JABOATAO_DOS 581,556 85.8 164.9 13.6 0.52 0.02 GUARARAPES/PE OLINDA/PE 367,902 64.2 101.4 3.3 0.63 0.03 PAULISTA/PE 262,237 -3.6 82.9 -2.7 (0.04) 0.04 PETROLINA/PE 218,538 81.6 93.0 3.3 0.88 0.08 RECIFE/PE 1,422,905 186.7 1,021.3 -8.6 0.18 0.02 TERESINA/PI 715,360 -38.4 391.5 16.9 (0.10) 0.03 CASCAVEL/PR 245,369 23.3 138.4 -2.4 0.17 0.05 COLOMBO/PR 183,329 -26.4 91.2 5.0 (0.29) 0.05 CURITIBA/PR 1,587,315 488.7 1,977.5 134.4 0.25 0.04 FOZ DO IGUACU/PR 258,543 69.2 232.8 16.2 0.30 0.07 GUARAPUAVA/PR 155,161 -5.9 88.1 -7.2 (0.07) 0.01 LONDRINA/PR 447,065 79.1 391.4 -26.9 0.20 0.02 MARINGA/PR 288,653 117.5 248.3 23.2 0.47 0.05 PONTA GROSSA/PR 273,616 97.0 159.5 -11.2 0.61 0.05 SAO JOSE DOS 204,316 -14.6 193.2 4.7 (0.08) 0.04 PINHAIS/PR BARRA MANSA/RJ 170,753 -1.6 121.8 11.7 (0.01) 0.04 BELFORD ROXO/RJ 434,474 3.0 127.4 11.5 0.02 0.01 DUQUE DE 775,456 -5.1 497.3 -0.9 (0.01) 0.00 CAXIAS/RJ ITABORAI/RJ 187,479 5.6 97.0 -4.6 0.06 0.00 MAGE/RJ 205,830 34.7 89.7 0.3 0.39 0.01 NILOPOLIS/RJ 153,712 21.9 54.4 0.9 0.40 0.04 NITEROI/RJ 459,451 64.8 455.0 -6.0 0.14 0.02 NOVA 173,418 53.1 127.3 -2.2 0.42 0.00 FRIBURGO/RJ NOVA IGUACU/RJ 920,599 -48.4 221.3 18.1 (0.22) 0.01 PETROPOLIS/RJ 286,537 44.8 264.6 7.5 0.17 0.01 RIO DE JANEIRO/RJ 5,857,904 5005.5 6,607.3 810.1 0.76 0.10 SAO GONCALO/RJ 891,119 0.1 245.2 18.4 0.00 0.01 SAO JOAO DE 449,476 -34.3 134.8 -0.6 (0.25) 0.01 MERITI/RJ VOLTA 242,063 248.3 290.5 18.3 0.85 0.01 REDONDA/RJ MOSSORO/RN 213,841 61.6 130.1 2.2 0.47 0.02 NATAL/RN 712,317 13.4 426.4 24.4 0.03 0.02 Inputs to a Strategy for Brazilian Cities Page 87 Municipality Population Net Net Primary NCD/NCR Debt service / NCR consolidated current Balance debt (NCD) revenue (PB) (NCR) PORTO VELHO/RO 334,661 298.0 190.7 8.5 1.56 0.03 BOA VISTA/RR 200,568 16.5 156.4 3.4 0.11 0.03 ALVORADA/RS 183,968 -12.9 63.4 1.5 (0.20) 0.07 CANOAS/RS 306,093 -65.7 297.1 20.7 (0.22) 0.01 CAXIAS DO SUL/RS 360,419 -115.3 426.8 21.8 (0.27) 0.02 GRAVATAI/RS 232,629 137.2 118.3 -13.6 1.16 0.05 NOVO 236,193 39.1 179.2 10.8 0.22 0.07 HAMBURGO/RS PASSO FUNDO/RS 168,458 5.4 105.3 2.2 0.05 0.04 PELOTAS/RS 323,158 126.0 216.9 34.7 0.58 0.06 PORTO ALEGRE/RS 1,360,590 490.9 1,747.7 41.5 0.28 0.04 RIO GRANDE/RS 186,544 -8.4 119.8 3.3 (0.07) 0.01 SANTA MARIA/RS 243,611 -14.4 138.3 -5.9 (0.10) 0.03 SAO LEOPOLDO/RS 193,547 1.0 159.1 12.3 0.01 0.03 VIAMAO/RS 227,429 56.8 80.1 -1.6 0.71 0.05 BLUMENAU/SC 261,808 57.5 292.3 28.3 0.20 0.03 CRICIUMA/SC 170,420 -7.7 116.5 22.3 (0.07) 0.04 FLORIANOPOLIS/SC 342,315 107.5 322.2 45.2 0.33 0.03 JOINVILLE/SC 429,604 -67.8 438.3 62.3 (0.15) 0.02 LAGES/SC 157,682 4.5 106.6 4.2 0.04 0.00 ARACAJU/SE 461,534 63.2 336.3 71.6 0.19 0.02 AMERICANA/SP 182,593 122.7 170.0 -7.6 0.72 0.04 ARARAQUARA/SP 182,471 3.7 171.1 0.1 0.02 0.00 BARUERI/SP 208,281 -9.6 457.7 20.0 (0.02) 0.00 BAURU/SP 316,064 117.7 225.5 -7.8 0.52 0.02 CARAPICUIBA/SP 344,596 83.2 85.8 -6.5 0.97 0.01 DIADEMA/SP 357,064 322.4 283.1 19.8 1.14 0.05 EMBU/SP 207,663 -0.5 103.0 6.7 (0.00) 0.00 FRANCA/SP 287,737 85.5 169.8 -8.9 0.50 0.03 GUARUJA/SP 264,812 359.7 331.3 -4.7 1.09 0.02 GUARULHOS/SP 1,072,717 569.4 905.1 -6.4 0.63 0.02 ITAPEVI/SP 162,433 19.4 104.8 11.4 0.18 0.02 ITAQUAQUECETUB 272,942 37.3 115.1 -1.6 0.32 0.00 A/SP JACAREI/SP 191,291 20.1 157.7 20.7 0.13 0.03 JUNDIAI/SP 323,397 154.7 409.5 8.8 0.38 0.06 LIMEIRA/SP 249,046 44.3 205.0 14.6 0.22 0.03 MARILIA/SP 197,342 31.8 215.3 13.5 0.15 0.02 MAUA/SP 363,392 209.8 233.0 20.1 0.90 0.09 MOGI DAS 330,241 35.0 208.1 32.4 0.17 0.04 CRUZES/SP PIRACICABA/SP 329,158 76.7 296.7 3.2 0.26 0.02 PRAIA GRANDE/SP 193,582 -18.0 228.5 6.6 (0.08) 0.03 PRESIDENTE 189,186 51.9 164.2 24.4 0.32 0.03 PRUDENTE/SP RIBEIRAO 504,923 -38.8 518.4 -149.4 (0.07) 0.01 PRETO/SP RIO CLARO/SP 168,218 -5.8 140.4 11.1 (0.04) 0.04 SANTA BARBARA 170,078 34.6 101.6 3.1 0.34 0.04 D'OESTE/SP SANTO ANDRE/SP 649,331 475.6 456.3 27.3 1.04 0.00 SANTOS/SP 417,983 224.7 570.5 83.2 0.39 0.02 Inputs to a Strategy for Brazilian Cities Page 88 Municipality Population Net Net Primary NCD/NCR Debt service / NCR consolidated current Balance debt (NCD) revenue (PB) (NCR) SAO BERNARDO 703,177 120.5 1,020.1 2.2 0.12 0.01 DO CAMPO/SP SAO CARLOS/SP 192,998 99.7 163.0 5.6 0.61 0.06 SAO JOSE DO RIO 358,523 16.0 309.9 -22.9 0.05 0.02 PRETO/SP SAO JOSE DOS 539,313 -355.3 687.7 39.9 (0.52) 0.04 CAMPOS/SP SAO VICENTE/SP 303,551 45.8 204.1 16.7 0.22 0.02 SOROCABA/SP 493,468 -35.1 475.5 4.3 (0.07) 0.03 SUMARE/SP 196,723 140.5 126.2 7.7 1.11 0.07 SUZANO/SP 228,690 4.9 161.2 5.7 0.03 0.03 TABOAO DA 197,644 -9.6 158.8 2.9 (0.06) 0.01 SERRA/SP TAUBATE/SP 244,165 5.7 214.5 9.7 0.03 0.01 Inputs to a Strategy for Brazilian Cities Page 89 5. Municipal Credit Markets By Benjamin Darche Introduction 5.1 States and municipal governments in Brazil require a substantial amount of resources. The financing burden for municipalities to fund this investment is increasing as the Federal government continues to incrementally shift health, education, and other social expenditure assignments to state and local governments. To meet even a small portion of the urban infrastructure investment demand, local governments can accelerate their investment requirements by borrowing. But the Federal government has severely constrained the demand and supply of credit to local governments due to the current administration's tight fiscal and monetary policies that support long-term price stability. Can Brazil move to prudently expand borrowing for municipal investment without repeating the collapse of the municipal credit market experienced from 1989 through the late 90s? If so, what steps can it take now to develop a prudent future capital market for municipal finance? 5.2 This paper will briefly address these questions and expand on the findings of the World Bank's 2001 report, "Brazil: Financing Municipal Infrastructure; Issues and Options". 5.3 The World Bank's 2001 report. The Banks report on Municipal Finance in Brazil presented the state of municipal credit market. The conditions of the municipal credit market present in the Bank's previous report are much the same today: · Low outstanding municipal debt (12%) as a percentage of total public sector debt; · High level of creditworthiness among medium and small municipalities; · A well developed private banking system; · Long history of municipal borrowing · Expanding sub-national capital expenditure in education, health housing and other social services; · Increasing investments in municipally owned water, sewer, solid waste companies as service areas grow, service demand increases, and companies defer annual maintenance to limit user fee increases; · Continued reliance on subsidized Federal government bank loans to provide credit at below market rates for municipal loans; · Restrictive legal and regulatory environment inhibiting municipal credit; · Significant regulatory and administrative burdens for the federal government to monitor the municipal credit market; · Limited investments in municipal infrastructure through public private partnerships primarily due to regulatory and user fee limitations. But rising federal and local government expectations that the new PPP law will encourage further private activity to help meet the growing infrastructure demand. 5.4 Legal borrowing and lending retrictions created by the municipal debt crisis. The mixed condition of the current municipal credit market described in the bullets above arose out of the local government debt crisis in the mid-1990s. The Bank's 2001 report showed that as of September, 1999, the outstanding stock of municipal debt reached $R 28.7 billion. State capitals held 75.1% of this debt with the cities of Rio de Janeiro and Sao Paulo holding 63.6% of the total. In contrast, the remaining municipalities had a small debt burden, significant borrowing capacity and an excellent loan repayment record. However, the legislation restricting municipal borrowing does not distinguish between good and bad municipal borrowers. It has severely limited local governments from borrowing, regardless of risk, and most banks, except the federally owned banks CEF and BNDES, from lending to municipalities. Municipal governments continue to rely on Federal government banks and other "government to government" lending for investment capital. 5.5 CEF and BNDES interest rate subsidies. On going interest rate subsidies for municipal loans by federal banks reinforces the government's role in municipal lending and further delays development of a self-sustaining Inputs to a Strategy for Brazilian Cities Page 90 municipal credit market. Although the federal banks include a risk factor in their calculation of interest rates municipalities pay on their loans, the high level of interest rate subsidies, absorbed by the FGTS and FAT social welfare funds, keeps municipal interest rates significantly below the Government of Brazil's (GOB) domestic borrowing rate. The combination of low rates in a restrictive borrowing environment has created "innovative" borrowing solutions and legal structures that distort the development of a risk-based private market for municipal debt. However, these same structures can also lead the prudent development of a municipal credit market, as we describe in the paper's second section. 5.6 Report organization. The report has two main sections. Section 1.0 presents the current status of the municipal credit market and the changes in the demand and supply of municipal credit since the Bank's 2001 report. Some of the demand side municipal credit issues include: · A brief review of the legal and regulatory factors constraining borrowing and lending; · Volatile capital expenditures by local government and bank lending for infrastructure investments since 2001; · Burdensome government mandated municipal credit market disclosure; · The creation of new types of local government financing institutions and organizations that can enhance the municipal market, but also act to delay adoption of market competitive rates supported by institutional investors. Supply side municipal credit issues the report will discuss include: · The impact of CMN and BACEN legal resolutions that limit financial institution municipal lending; · Continued dominance of CEF and BNDES in the municipal loan market; · The development of new instruments to fund municipal investments; · The potential impact of the new Public Private Partnership (PPP) law on the municipal credit market. 5.7 Section 2.0 presents the current status of the capital markets and the elements of the current municipal capital market that inhibit or are conducive to the development of market based municipal credit. It describes some of the new CEF municipal credit structures and variations of these structures based on the rapid development of Asset Backed Securities (ABS) in the domestic bond market. It provides a discussion of: · The status of Brazil's Capital Markets with an emphasis on the rising role of institutional investors and the potential to develop new municipal borrowing instruments; · The role of credit ratings in development of a municipal credit market; · How the GOB can shift some of the current disclosure and monitoring burden from STN to the private sector to encourage development of a prudent municipal credit market; · Some of the transition steps the GOB may take to move toward a private, risk based municipal credit market. Current Status of the Municipal Credit Market: Developments in the Demand and Supply of Municipal Credit 5.8 Legal and regulatory supply and demand constraints. The local government debt crisis in the mid-1990s initiated a flurry of legal activity in subsequent years to control the issuance of sub-national debt and reduce the fiscal impact to the GOB of profligate sub-sovereign government spenders. The crisis was triggered by the mountain of outstanding state debt that contributed to more than one third of Brazil's total outstanding consolidated public debt in 1997. The federal government negotiated agreements with the major state debtors to restructure their outstanding obligations, primarily state bonds, which amounted to 11.5% of Brazil's GDP at the time. The restructuring agreements, sanctioned by Law 9496 of 1997, required states to comply with fiscal and Inputs to a Strategy for Brazilian Cities Page 91 debt indicators and other structural reforms to address the causes that lead to the federal bailout and restructuring. Senate Law 78, passed in 1999, applied these restrictions to all sub-national governments. The law included several targets such as debt service to current revenue ratios; personnel expenditures as a percent of current revenue; privatization of state enterprises and other reforms described in the Bank's previous report47. It also prohibited states to issue bonds while their restructured debt to the federal government remains outstanding. Similar legislation passed in 2001 to restructure the debt of 180 municipalities also prohibits the issuance of bonds by municipal government until they retire their outstanding restructured debt. The state-federal agreements on fiscal adjustment and structural reform established the foundation for subsequent laws that limited sub-national borrowing. 5.9 Fiscal Responsibility and other laws restricting the supply and demand of municipal credit. To consolidate the various laws and restrictions on sub-national borrowing and to help buttress Brazil's progress to maintain fiscal discipline and price stability, the Senate passed the Fiscal Responsibility Law (FRL) in 2000. The FRL further strengthened the previous Senate laws and other resolutions controlling sub-national debt obligations. It established debt policies for sub-national borrowing including debt and fiscal indicators and other provisions that make borrowing more prudent and efficient48. It also included other financial policies that attempt to control borrowing such as a prohibition of debt refinancing between different government levels and identifying financing sources for capital investment based on annual budget targets. However the FRL left the specific quantitative indicator controlling sub-national borrowing for future Senate Resolutions. Senate Resolutions 40 and 43 of 2001 established the fiscal and debt targets and prohibited specific types of borrowing. Some of the critical municipal credit indicators and borrowing restrictions are as follows: For New Borrowing: o Maximum debt service cannot exceed 11.5% of Net Current Revenues (NCR)49; o Total borrowing in the budget year cannot exceed 16% of NCR; o Total outstanding stock of debt cannot be greater than twice NCR for states and 1.2 times for municipalities; · General Sub-national Borrowing Restrictions: o State and municipal governments cannot issue bonds until 2010 except to refinance principal on existing outstanding debt; o They cannot borrow to refinance arrears on existing outstanding debt; o They cannot assume payment arrears from suppliers in the form of promissory notes or other forms of debt; o Higher level governments cannot lend to lower level governments. · Other Indicators and Borrowing Restriction: o Personnel expenditures to NCR must be less than 60% for the federal gand 60%. These legal supply and demand restrictions to control municipal debt are some of the most stringent in Latin America50. Monitoring Sub-National Compliance with the Fiscal Responsibility Law 5.10 To help implement the FRL's sub-national borrowing restrictions, the CMN and BACEN have issued several regulations governing the supply of municipal credit offered by financial institutions to local governments. The CMN establishes lending policies and the BACEN and STN execute the policies through issuance of regulatory limits and loan approvals. While maintaining the spirit of fiscal control provided in the FRL, the plethora of CMN Resolutions issued since the FRL gives state and local government a certain amount of 47See Brazil, Municipal Finance, Issues and Options, Box 4.3, Controls on Sub-national borrowing, page 36. 48ibid, page 40. 49Net Current Revenues are very similar to RLR. 50See Freire, Huertas, Darche (1998) "Subnational Aceess to the Capital Markets: The Latin American Experience", page 12. Inputs to a Strategy for Brazilian Cities Page 92 borrowing "wiggle room" for specific investments. It has also led to the creative use of new "borrowing" instruments that have the potential to move municipal obligations toward the private capital markets, but may also support current expenditure rather than investment. We will return to this development in Section 2.0. For purposes of the current legal and regulatory review, one result of the myriad regulations governing sub-national borrowing is the lack of criteria to select priority sub-national government loan applications to allocate the global cap. It delays the distribution of municipal credit based primarily on credit risk and capital market discipline. Supply Restrictions - CMN and BACEN Resolutions. CMN's 1999 Resolution 2827 set the initial financial institution lending limit to sub-national governments up to 45% of the institution's total equity. It established a global limit of R$ 1 billion lending to sub-national governments in 2001, reduced to R$ 200 million by Resolution 2954 in 2002. However, CMN Resolution 3153 in 2003 raised the ceiling on previous CMN Resolution lending limits to R$ 1.1 billion for sanitation projects which was increased to $R 2.2 billion by Resolution 3331 in 2005. The Ministry of Finance expects private banks to provide $R 200 million of this amount51. 5.11 In spite of the increases in lending limits, there is a significant waiting time for STN approval for municipal credits. The backlog of loan applications waiting for STN approval is substantial. Local governments pejoratively call it the "filon" (long queue). The borrowing restrictions combined with a burdensome application and approval process has led to stagnant municipal lending between 2001 and 2004. 5.12 Burdensome municipal credit approval process. The STN has a manual52 that explains the burdensome municipal credit approval process. It outlines the documentation required for AROs and internal and external municipal borrowing following FRL and Senate Resolution 40 and 43 requirements. It shows how the local governments should construct the various debt and fiscal indicators provided in the Senate Resolutions; evidence of the required Tribunal de Contas budget reviews; documentation of approvals by the authorizing body (such as the City Council in the case of municipal borrowing); and other procedures to comply with the legal morass that controls municipal borrowing. The reporting requirements include: 1. A bi-monthly report on its "financial and actuarial position" that shows: · a budget balance sheet; · revenues in the reporting period and an update of year-end forecasted revenues; · expenditures in the reporting period and an update of year-end forecasted expenditures; · revenues from credit operations and all debt service payments. 1. A trimester Financial Management Report that includes: · Compliance with legal debt service and other fiscal ratios; · Personnel expenditures segregated by active and inactive (pensioners) personnel; · Financial guarantees provided by the local government; · Progress on annual budget targets and the Net Worth of the local government. 2. An annual Budget and Financial Management Report that incorporates items 1 and 2 above. 5.13 Required reports and their relationship to municipal credit analysis. The reports provide good information for legal municipal debt compliance. STN data collection is a good instrument to insure standardization of local government reporting. However, the STN manual is not a toolkit to show local governments how to assess their creditworthiness using "best practice" credit criteria, such as international credit rating agency criteria that use additional factors such as the economic base, political and administrative setting to give a broader view of the willingness and ability of municipal governments to repay their debt. The borrower is only required to disclose information pertaining to the laws governing municipal credit. Lenders, such as CEF, rely on their own risk analysis to determine the general creditworthiness of municipal government. To move the 51Valor Economico Nov 25, 2005. 52Ministerio de Fazenda (2005), "OPERACÕES DE CRÉDITO DE ESTADOS E MUNICÍPIOS MANUAL DE INSTRUCÃO DE PLEITOS" - MIP.- Inputs to a Strategy for Brazilian Cities Page 93 municipal market toward self-monitoring in the future, the legal financial reporting requirements for municipal governments should be streamlined with many of the elements of the current reports and documentation shifted to a market regulating process. This will most likely only occur as the municipal credit market shifts from government bank lending to private creditors in a competitive interest rate environment. 5.14 Subsidized interest rates. Lack of risk criteria to determine interest rates for municipal loans is prevalent in the municipal credit market. The 2001 report illustrated that the majority of municipal credits have subsidized interest rate loans from the federal government banks CEF and BNDES. This is still the norm, albeit at a lower lending volume due to legal credit supply restrictions. There is a small component of the CEF and BNDES interest rate that is attributed to municipal credit risk in these loans, but this is offset by the substantial base rate subsidy. CEF and BNDES are able to provide this subsidy because their cost of funding for municipal loans, deposits in FGTS and other social welfare funds, earn below market interest rates. This creates a web of supply and demand subsidies that distorts the municipal credit market and further removes sub-national governments from competitive interest rates and market discipline. Municipal Capital Revenues and Expenditures 5.15 Rigid controls on municipal credit operations contributed to stagnant growth in capital revenues by sub- national governments between 2000 and 2004 and were concentrated primarily in the largest municipalities. A recent Bank report53 of a sample of 3,028 municipalities showed a small average annual decline in total capital revenues (credit operations, intergovernmental transfers, and asset sales) between 2000 and 2004 from $R 4.02 to $R 3.81 billion. Credit operations initially declined from $R1.01 to $R 0.56 billion between 2000 and 2001, but gradually increased to 1.25 billion by 2004. Over 30% of capital spending was concentrated in Brazil's municipalities with populations over 1,000,000. Credit operations showed a more pronounced concentration in these municipalities. They absorbed over 70% of municipal credit in the period and had per capita borrowing seven times the per capita amounts of the smaller municipalities54. 5.16 Capital investment expenditures. Although capital revenues were relatively stagnant between 2000 and 2004, capital investment expenditures increased 28.7% in the period from $R 9.55 to $R12.3 billion, but showed significant year-to-year volatility, falling to $R 8.39 billion in 2001 and rising over 50% to $R 12.8 billion in 2002. Medium size municipalities showed a larger increase over the period, probably due to the significant rise in their net current balance as tax income increased at a more rapid pace than current expenditures and low debt service payment requirements. Small cities that lacked access to credit operations showed stagnant investment levels. The largest municipalities increased investment expenditure through credit operations in spite of borrowing restrictions55 imposed by the FRL and their debt restructuring agreements with the federal government. 5.17 Constraint on medium and small municipal credit. The current restrictions on municipal borrowing apparently have limited borrowing by medium and small municipalities despite their substantial debt capacity. The top quartile of municipalities with 50,000 or more inhabitants (122 of 491) demonstrate considerable debt capacity; their Debt Service to Current Revenue ratio was only 2.1% using 2003 STN data, substantially below the 11.5% regulatory limit.56 Political influence of the large municipalities for access to credit may crowd out small and medium size municipalities in the competition for scarce municipal credit. They get pushed to the back of the "filon". 53World Bank (2005) The evolution of Brazilian Municipal Finances, 2000-2004 First Draft/November 2004. 54 ibid, pg. 19. 55ibid, pg 24. 56Vetter, David (2005) "The Role of Private Subnatinal Credit Markets in making Land Development More Affordable", pg. 16. Inputs to a Strategy for Brazilian Cities Page 94 5.18 Although large municipalities continue to finance some of their capital investments with CAXIA and BNDES loans, they still need to raise capital for infrastructure investments. To gain access to additional funding, some larger municipalities have creatively developed new capital formation instruments and institutions. New Borrowing Instruments and Lending Institutions to Assist Subnational Governments to Raise Capital 5.19 The rigid municipal credit controls have spawned creative measures by sub-national governments to raise capital. Two of the more original developments are the securitization of oil royalties and the transformation of state Municipal Development Funds into "Entidades Non-Dependentes" (non-dependent entities). 5.20 Oil royality FDIC. The State of Rio de Janeiro sold future oil royalties via "Fondos de Investimentos dos Direitos Creditos" (FDIC) to raise capital. FDICs are a new financing vehicle that has deepened the domestic capital markets by providing companies with alternatives to bank credit57. They were introduced by the BACEN in 2001 and are governed by the CVM, the Securities and Exchange Commission. FIDCs are structured as open or closed-end mutual funds, wherein investors purchase "shares" of assets owned by the FDIC in a "true sale". Since an FDIC transaction is considered as a sale of assets, rather than a debt obligation, these instruments do not fall under municipal credit legal and regulatory restrictions. In addition to Rio de Janeirio, other state and municipalities with oil royalties are considering sales of their future oil royalty assets. 5.21 The advantage of the oil royalty FDIC is that it diversifies local government funding sources via a competitive capital market instrument, albeit at higher interest rates than CAXIA or BNDES subsidized rates. The fact that a state sold a FDICs at non-subsidized interest rates shows that they are willing to raise funds in the competitive capital market. It also probably indicates the difficulty they have to raise capital for their investment needs. A disadvantage is that the proceeds from the sale may not be used for investment purposes, but to support current budget deficits. If this is the case, the local government is selling future generation's assets for current consumption - an unwise policy that will also hinder future economic growth by reducing infrastructure investment. 5.22 Another possible disadvantage of this instrument is that it circumvents the legal and regulatory controls on municipal borrowing and may exacerbate a local government's fiscal condition by spending on current consumption rather than investment. However, oil royalty FIDCs may actually be a positive development that moves local governments toward market competitive market rates and market discipline. The market will price the political, economic and other risks associated with the sale of a future oil revenues. The FIDCs, as an asset backed security (ABS), can spawn other ABS instruments related to the municipal market, as more fully discussed in Section 2.0. 5.23 Entidades Non-Dependentes. The second variation in the municipal credit market that took place since the Bank's 2001 Report is the transformation of municipal development funds to "Entidades Non-Dependents". The passage of the legal and regulatory restrictions on municipal loans eliminated state Municipal Development Funds from lending to municipalities. To circumvent this obstacle, many states have restructured their MDFs into Entidades Non-Dependentes. These are non-bank organizations that are created with the remaining capital from liquidated MDFs. They are then eligible to provide capital to municipal governments for their infrastructure investments. 5.24 The advantage of the Entidades Non-Dependentes is that provide an much needed source of capital to creditworthy municipalities. The disadvantage is that they continue to reinforce the role of government in the municipal credit market, albeit in a similar fashion as the former MDFs. The danger is that these organizations become more like state development banks with potential political interference that may distort credit supply. As a "stop-gap" measure to help finance much need municipal infrastructure, these organizations are a good 57Moody's Special Report, Structured Finance "Growth of FIDCs in Brazil: Current Outlook and Featured Structural Mechanisms" (October 27, 2004). Inputs to a Strategy for Brazilian Cities Page 95 development. But there should be some mechanisms to shift these entities to the private credit markets as quickly as possible. Public Private Partnerships and the Municipal Credit Market 5.25 The factors impeding the growth of private participation in infrastructure have not significantly changed since the Bank's 2001 report. It listed a number of legal, regulatory and financial issues that, for the most part, continue to hamper PPP progress, especially in the water and sanitation sectors: · Legal ­ ownership of assets between many state and municipal water companies is still unclear with on- going court cases attempting to resolve asset ownership and other issues. Lack of a law that clearly defines water and sewerage asset ownership is a significant impediment to greater private participation. · Regulatory - there is still no standardized national law for water and sewer concessions, no clear legislation on ownership of existing assets or tariff adjustment. Concession contracts have improved, but often contain clauses on tariff adjustment that are left unclear; · Financial - Most municipal water companies remain in poor financial condition. Tariffs do not support operations and political interference in tariff adjustments has not substantially abated since 2001 report. 5.26 In spite of these obstacles, the government continues to encourage private participation in public services. The Congress passed Act 11.079 on December 30, 2004 (the "PPP Law") that allows for more flexibility in the construction and operation of public infrastructure projects by the private sector. It expands upon the 1995 concession law (8987) and allows for direct government financial support of public projects in "administrative" (without user fees) and "sponsored" (with user fees) concessions. "Ordinary" non-recourse funded concessions that do not have any direct government financial support are not subject to the new law and remain regulated by the 1995 Act (8987). 5.27 PPP law and the PFI Concept. The PPP law is similar to the PFI concept that first emerged in the United Kingdom and that has spread to other countries in Europe, Australia, New Zealand and South Africa. Several other Emerging Market countries are in various development phases of creating PFI Units to expand traditional concession projects to include "administrative" as well as "sponsored" PPPs. This is a natural outcome of the process of private participation in infrastructure. Many traditional non-recourse infrastructure projects, especially in the water and sanitation sector, could not support private returns on equity with socially constrained tariffs. These projects required some type of government financial support to be commercially viable. In the UK, the government expanded the traditional user fee concession project concept to include private construction and maintenance projects for public facilities such as prisons, schools and health clinics, built and operated by the private sector, but financially supported with local government budget revenues. The use of direct government budget revenues has also led to new debt instruments in the UK capital markets to finance PPP projects58. Likewise, it may be possible to create new debt obligation instruments for municipal PPP projects in Brazil's capital markets more fully discussed in Section 2.0. 5.28 On-going PPP risks. Like similar PFI laws around the world, Brazil's PPP federal legislation establishes the contract guidelines for federal, state, federal district and municipal PPPs. They provide for sharing legal, economic, and financial and other risks between the government and private contractor and present the general terms and conditions of government financial support for PPP projects. 5.29 The law is a good beginning and should result in further private participation in the construction, operation and maintenance of public service projects. But like existing concession projects in Brazil, the PPP law only provides contract guidelines. Individual PPP contract and concession agreements may be deficient and require substantial improvement to properly allocate risks between the public and private sectors. 58Standard & Poors (2005) "Public Private Partnerships Global Credit Survey 2005". Inputs to a Strategy for Brazilian Cities Page 96 5.30 Several issues listed above that currently plague concession contracts might similarly affect projects regulated by PPP Law 11.079. This is especially the case for "sponsored" concessions that have user fees that guarantee project debt and private contractor equity returns. For example, the law allows user fees to automatically adjust tariffs based on indices included in the contract. However, it also gives the local government the opportunity to approve the tariff increase if it publishes the reasons in the official press 15 days after presentation of the private contractor's payment invoice59. This introduces political risk for "sponsored" or user fee concessions such as a water supply project and will influence the extent to which private companies will invest in and creditors lend to PPP projects. Unless the GOB improves the current legal and regulatory framework for water and other user fee related projects, PPP Law sponsored concessions will most likely have the same tariff adjustment risks as current concession projects. 5.31 The expansion of government financial support in PPP Law project can mitigate tariff adjustment concerns to some degree through debt instruments that make government payments "water tight". This will depend on the degree to which creditors can enforce the legal documents supporting the security structure for a PPP debt obligation that includes direct government financial support in addition to tariffs. The federal government may consider developing regulatory guidelines for the new PPP law so that state and municipal governments that introduce their own PPP laws use a standardized approach for the legal documents related to PPP debt obligations. 5.32 Using the PPP Law to expand the municipal credit market. The PPP law allows for further development of the municipal credit market in several ways. First, as a "private" company, the administrated or sponsored concession PPP is not subject to the regulatory limits restricting borrowing by local governments or bank lending limits to local governments. Second, local governments can provide budgetary financial from revenues or through a variety of other financing mechanisms from: · A PPP Guarantee Fund; · Other special funds established in law for PPPs; · Private insurance company surety bonds; · International financial institution guarantees; · Other mechanisms permitted by law60. Section 2.0 will discuss some possible capital market options to finance PPP projects. Development of Municipal Credit Markets in Brazil Current Capital Market Conditions 5.33 The initial defaults in the sub-sovereign bond market in the late 1980s and associated bail out of state and municipal contracted indebtedness by the federal government in 1993 and 1997 effectively closed down the domestic sub-sovereign bond market. The market remains closed to general sub-sovereign debt as stipulated in the FRL legal and other regulatory impediments. However, it is open to debt obligations issued by private utility companies, PPP project debt financing, and refinancing of outstanding contracted debt by local government. 5.34 As discussed in Section I above, CEF and BNDES dominate bank lending to local governments for urban infrastructure improvements. CEF has several urban development programs funded by various federal government ministries, the World Bank and IDB. In 2004 the federal government, CEF and BNDES financed a total of $R 7.7 billion of urban infrastructure of which $R 5.9 billion was for water and sanitation programs61 . CEF provided about $R 4 billion since 2002 to fund these programs in large, medium and small municipal 59 Law 11.079, 30/12/04, Chapter I, Article 4, item x.1. 60Law 11.079, 30/12/04, Chapter III, Article 8. 61Interview with Rogerio Tavares, CEF. Inputs to a Strategy for Brazilian Cities Page 97 governments. BNDES targets it's lending primarily to the larger private utilities or other PPP projects on a limited or non-recourse basis. Subsidized interest rates are an obstacle to commercial bank participation in the municipal credit market. CEF and BNDES have programs to integrate private commercial banks into their urban infrastructure lending programs, but commercial bank lending remains modest. Most private banks are not willing lend to municipalities for the limited spreads required in the BNDES and CEF subsidized programs when they can earn several times that amount in the private market and often at lower risk. The subsidized interest rates provided by CEF and BNDES through the FGTS and FAT funding are significant obstacles in attracting private banks to the municipal credit market. Until the federal government shifts the subsidy arrangements for essential public service investments, such as water and sanitation projects, bank lending for these projects will remain mostly with private banks, for the most part62. 5.35 Domestic bond market. Municipal borrowing in the domestic bond market is done almost exclusively by state and municipal electric and water utilities that are privately held. Municipal utilities from Sao Paulo, Minas Gerais, Parana, Ceara, and Brasilia have issued about $R 2.4 billion of bonds63. Due to the strict limitations on bond sales by state and local governments, there have been no sub-sovereign domestic bond issues in the past few years except for the state of Rio de Janerio's oil royalty FDIC, discussed above, and the of Rio Grande do Sul's $R 100 million structured financing backed by ICMS taxes sold last October. The inability of sub-sovereign governments to enter the domestic market with "plain vanilla" municipal bonds has led to the emergence of ABS type structures. The ABS market in Brazil has shown explosive growth in the past few years which now includes structured financings for sub-sovereign governments. 5.36 The domestic bond market in Brazil continues to expand and diversify with new structured financing instruments using a variety of trade receivable, credit card, and commercial consumer loans and, in the municipal market, oil royalties and ICMS taxes to back the bonds. These asset-backed instruments, developed in mature capital markets in the US, and have recently expanded into Emerging Market countries' domestic capital markets. Brazil's domestic capital market continues to broaden and deepen with these and other new instruments like those issued by the GOB, public utilities, and the FIDCs. A major driving force in this development is demand created by institutional investors. 5.37 Institutional Investors64. In most domestic capital markets, the demand for high quality fixed income investments by institutional investors motivates the structural innovations of new financing mechanisms that take place in the market. As the pension systems in the Emerging Markets shift from "pay-as-you-go" government funding to public and private professional fund management, the demand for longer term, high quality investments will grow. 5.38 As shown in Table 5.1, pension funds and insurance companies have almost US$86 billion invested in fixed income securities in Brazil65: 62MDFs are another source of funding for public services. Another paper will address developments in the MDF municipal credit market. 63Moodys (1 de Dezembro de 2005) "Lista De Ratings Da Moody's Para O Brasil" . 64 For purposes of this report, we refer to institutional investors as the managers of Brazil's public and private pension funds and insurance companies. 65From Vetter (2005), "The Role of Private Subnational Credit Markets in Making Land Development More Affordable", pg.14. Inputs to a Strategy for Brazilian Cities Page 98 Table 5.1 Total Investments of Institutional Investors on November 2003/January 2004 Institutional Investors Total Investments (Billions of US$) Mutual Funds 186.4 Private Pension Funds 69.4 Insurance Companies 16.0 5.39 Changing Composition of Institutional Investor Portfolios. Given the relative immature state of Brazil's domestic market, most institutional investors rely on GOB securities institutional investors for their portfolio investments. GOB bonds comprise approximately 75% of pension funds fixed income portfolios66. However, the proportion of GOB bonds to total portfolio investments is changing as the domestic market expands with new fixed income instruments such as FDICs and other structured finance products. Institutional managers seeking to prudently diversity their portfolios are significant investors in high quality FDICs and other structured finance bonds. These bonds are almost always rated above investment grade (BBB-) and often carry the same AAA national scale rating as GOB's securities. Institution investor portfolio managers are driving much of the current demand for the new structured finance instruments such as FDICs and other asset backed securities and are a potential source of demand for the structured finance and other instruments that can help develop a future municipal bond market. But direct investment in municipal credits by either institutional investors or banks will not be significant until the GOB restructures its policy of subsidized interest rate lending by CEF and BNDES to put municipal credits on a level playing field with other corporate and public fixed income instruments. 5.40 Subsidized interest rates and development of new market instruments for municipal credit. Demand for municipal assets from institutional investors and commercial banks will grow when subsidized rates for municipal credits converge with market rates and the municipal credit market adopts "best practice" credit analysis and disclosure procedures. 5.41 Reduction of subsidized lending will require substantial shifts in FGTS and FAT funding policies from subsidized funding to market funding for CEF and BNDES municipal loans. A myriad of complex inter-related financial institution, capital market, municipal service, regulatory and political issues will influence this event, which is unpredictable. Until it happens, an alternative approach is to incrementally integrate the private markets into the currently subsidized municipal credit markets. CEF and BNDES have begun to do this as the GOB continues its financial institution reforms to move government owned banks toward market competition. They have expanded their lending instruments to incorporate elements of private markets into their municipal credit operations. The next section discusses these developments and suggests further incremental steps CEF and BNDES may consider to accelerate private participation in municipal credit. Municipal Credit Instruments 5.42 This section will present new municipal credit structures that CEF and BNDES have developed to incorporate private banks into their lending programs since the Bank's 2001 report. This is an encouraging step that shows a GOB commitment to cautiously re-integrate private credit into the municipal credit market67. On the domestic capital market side, several new structures have emerged that may help diversify the municipal credit market. These include the FDIC and other asset backed structures mentioned above. The section will introduce 66IMF (2005) Chapter II, Global Financial Capital Market Developments. 67Private banks and the domestic bond market investors were very active in municipal credit until the GOB bailed out these investors during the local government debt crisis in the late 1980s and early 1990s. See "Brazil: Issues in Fiscal Federalism", op.cit., pages 6-10. Inputs to a Strategy for Brazilian Cities Page 99 some additional capital market structures that may be appropriate at this time to support future municipal bond issues. 5.43 The small but encouraging developments in the municipal credit market must still overcome substantial difficulties for Brazil to attract private creditors to the municipal credit market. Critical judicial, bankruptcy and "moral hazard" issues persist, as indicated in the Bank's 2002 report on fiscal federalism68. There are also critical credit analysis, disclosure, and monitoring obstacles that may prevent private creditors from moving back to the municipal credit market. We present some suggestions for incorporating improvements to the current municipal credit market activities that will hopefully provide private creditors, especially institutional investors, with the confidence to become more active in the municipal credit market. 5.44 CEF and BNDES Private Municipal Credit Structures. CEF is exploring several new debt structures that shift its lending instruments toward the private markets and directly involve the FGTS fund69. These structures directly link FGTS funding with municipal creditors. One structure is to integrate commercial banks into CEF's municipal loan program. In this case, private financial institutions act as a "retail" bank that on-lends wholesale FGTS funds to municipalities with CEF acting as fiscal agent rather than in a traditional lending role. 5.45 Another CEF structure uses an SPE created for a PPP project, such as a water supply treatment facility. In this case the SPE borrows directly from FGTS or FAT to finance the investments. A variation of the latter mechanism is similar to a "sale-leaseback" structure common in the US municipal and European bond markets. In this case the SPE is a Trust that "buys" water concession assets from the water company. The water company builds the plant with the sale proceeds and leases it back from the Trust. It then receives periodic principal and interest installment payments from the company. In Brazil this structure can only be legally transacted through Certificados de Receiveis Imobiliários70 (CRIs) issued by the Trust. 5.46 Credit enhancement. The Dexia report describes variations of these structures using credit enhancement instruments such as Letters of Credit, full or partial credit guarantees from Brazilian or foreign financial institutions and monoline insurance companies. Credit enhancement shifts some or all of the municipal credit risk from private banks or institutional investors to the credit guarantor. This is a useful mechanism that has several potential benefits especially in the evolution of nascent municipal credit markets: · Provides comfort to investors that are not familiar with municipal risk; · Allows municipality's that might otherwise not be able to enter the market to obtain credit at affordable rates; · Provides support for new types of financing instruments such as "pooled" financing, securitization and other structures that might not otherwise be marketable. 5.47 In addition to lowering borrowing costs for municipalities, credit enhancement provided by international financial institutions can disseminate useful knowledge for domestic capital market participants (commercial banks and institutional bond investors) about municipal credit risk. Local investors can build their credit analysis capacity and become more comfortable with municipal credit risk by learning "best practice" municipal credit analysis techniques. Introduction of credit ratings by companies using international credit rating methodologies accepted by global investors should accelerate the local investor learning process for municipal risk. 68Op cit., page 20. 69Standard & Poors rated the debt a notch below the domestic currency BB rating because the bonds are at parity with other external debt and would be accelerated like other outstanding international bond issues in the event of a default. See Standard and Poors, "The Credit Implications of Local Currency Financing", October 5, 2005. 70This is a fixed income instrument secured by real property. Inputs to a Strategy for Brazilian Cities Page 100 The Use of Asset Backed Securities (ABS)71 to Further Develop the Municipal Credit Market 5.48 Securitization of CEF/FGUTS and BNDES/FAT assets using Collateralized Debt Obligations (CDOs). CEF and BNDES can expand their new municipal credit instruments discussed above by structuring an ABS for the domestic capital market. Basic options include: · Securitization of CEF,s direct loan portfolio; · Securitization of SPE/Trust CRIs that finance PPP projects; · Securitization using Credit Enhancement 5.49 These new instruments use the same concepts as Rio de Janeiro and Rio Grande do Sul's oil royalty and ICMS tax revenue securitizations recently issued in the domestic capital market. In a new collateralized debt obligation (CDO72) structure, CEF/FGTS and BNDES/FAT sell a portfolio of their direct municipal loans to an FDIC in the domestic capital market. This will release capacity for the banks to continue lending to local governments under the 45% regulatory limit. 5.50 The sale of a CEF or BNDES CDO is indirect funding of municipal credits by the domestic bond market. It introduces a diverse portfolio of municipal credits to investors and expands the use of credit ratings into the municipal credit market73. 5.51 CDO familiarize the market with small and medium municipal credit risk. The credit rating of a municipal CDO will require a "shadow" or underlying rating of each of the municipalities in the total loan portfolio. This will allow CEF and BNDES staff to compare the results of their risk analysis methods with international credit rating methodologies when structuring a CDO. The CDO also shifts CEF municipal risk for seasoned loans to institutional investors. Credit enhancement, partial or full, can also play a role in this structure depending on market conditions and the underlying municipal credits. 5.52 CEF will most likely not be able to use a CDO structure to sell its loan portfolio until the gap between subsidized loans and market rates shrinks. It could only sell its municipal loan portfolio in today's market at a loss because it would have to over-collateralize its offering with additional cash flow from the underlying loan debt service payments to compensate for the gap between current market rates and its municipal portfolio rates. However, it may be possible to overcome this obstacle through issuance of a Real denominated international security. 5.53 A Real denominated CEF CDO issued in the global and domestic market may mitigate this problem and attract global and domestic investors at yields that do not require over collateralization, in the same way the 12.75% GOB global issued last October attracted domestic buyers. The Real global bond was over-subscribed even though domestic 8 year GOB bonds were selling at a yield of about 15.5% yield at the time (see Box 5.1). It may be possible to structure a CEF CDO with credit enhancement features that raises the credit quality of the pool 71An Asset Backed Security is comprised of a portfolio of individual assets such as trade, credit card, consumer loans, commercial loans, other debt obligations (generically referred to as Collateralized Debt Obligations ­ CDOs) that are bundled together and sold to investors. The cash flows from the underlying assets are passed though to bondholders who own an interest in the assets. 72A CDO is a generic name of a package of loans and/or bonds that are sold as a single fixed income security. The cash flow from the underlying loans provides the interest and principal payments for the fixed income security. Often a CDO will have senior and subordinated tranches of debt obligations within the same security that reflect their priority position to receive the underlying cash flow payments. 73International credit rating agencies have issued sub-sovereign credit ratings in Brazil primarily for electric utilities and or future flow deals such as the State of Rio de Janeiro and Rio Grande do Sul's structured financings. Fitch rated Rio de Janeiro's oil royalty FIDC AAA(bra) and Moody's an A3.br rating for Rio Grande do Sul's securitization of ICMS tax revenues. They do not have general municipal credit ratings due to the lack of general municipal credits in the domestic market. Inputs to a Strategy for Brazilian Cities Page 101 of municipal loans to provide greater comfort to institutional investors while matching the yield requirements for this type of ABS. Box 5.1 Government of Brazil Real Denominated Global Bond The Government of Brazil issued a 10 year, 3.4 billion Real denominated bond in October with a yield of 12.75%. This extended the GOB's domestic bond yield curve from 8 to10 years and gave the GOB an average maturity of 21 months for its debt. The 12.75% coupon for this ground breaking issue significantly lowers the government's borrowing cost. Foreign investors seeking higher fixed income yields in a relatively low interest rate market had understandably stronger demand for the issue than domestic investors who could purchase 30 day GOB bonds at a rate of about 18% throughout 2005. Nevertheless, the GOB was able to reduce it domestic currency cost of capital and deepen the domestic capital market with this Real denominated issue.74 5.54 Securitization of CRIs. It may also be possible for CEF to sell its CRIs in the market by bundling these securities and selling shares in an FIDC mutual fund. A critical question is whether CEF has a sufficient volume of CRIs to package and sell as a FIDC or other asset backed instrument to institutional investors. There are several other legal and market questions that this type of instrument would have to address. Nevertheless, it may be possible to introduce institutional investors to the credit characteristics of PPP projects and municipal utility risk by bundling the CRIs of smaller municipalities into a pool of securities. Again, credit enhancement may play a roll in this structure. Monitoring the Municipal Credit Market 5.55 Progress toward broader, deeper self-sustaining municipal credit market based on competitive interest rates will depend, to a large degree, on the pace of the reforms that shift the supply of municipal credit from government controlled banks to private financial institutions and institutional investors in the bond market. In the previous section we discussed some of the incremental changes in CEF and BNDES municipal lending programs that incorporate private banks. We also presented some of the new ABS instruments that can support progress in the municipal bond market. These incremental steps are a good beginning, but additional financial institution, capital market and regulatory reforms are required to accelerate the shift to a self-sustaining municipal credit market. What are some of the regulatory reforms the GOB may consider? In this section we present suggestions for reforms in municipal credit risk analysis, disclosure and reporting, and surveillance. The objective is to move toward a self-regulating municipal credit market governed by prudential regulatory guidelines that maintain a sound municipal credit system with competitive interest rates. The elements of this system should be in place and further encourage the government in order to shift the supply of municipal credit from subsidized CEF and BNDES lending to competitive market rates. 5.56 A good example of the confluence of reforms and development of a self-sustaining municipal credit market is Mexico, as shown in Box 5.2. 74Standard & Poors (October, 2005), "The Credit Implications of Local Currency Financing" .. Inputs to a Strategy for Brazilian Cities Page 102 Box 5.2 Mexico's Municipal Credit Market Much like Brazil, government banks in Mexico provided a substantial portion of municipal credit in the past. Local governments were not considered creditworthy; there was no secure source of debt repayment and very little lending by commercial banks to municipalities. The bond market for municipal debt did not exist. As Mexico's economy stabilized, and banking, financial markets, securities regulation, pension, fiscal and decentralization reforms proceeded, the nature of the municipal credit market changed. It shifted from the predominately government development bank supply of credit to a private supply of credit. Municipal creditworthiness improved through fiscal and decentralization reforms that allowed for the transparent transfer of federal government shared tax revenues to the states. Capital market regulation allowed sub- sovereign governments to issue bonds, but only if they received a credit rating from 2 international credit rating agencies, among other regulatory controls. Banking regulations required all banks to use Basel II principles to assess portfolio credit risk to determine capital reserve requirements, including loans to municipalities. Private pension funds established through pension reforms created a new supply of capital. Capital market prudential regulations allowed pension funds to purchase municipal securities. The confluence of these reforms with Mexico's stable macroeconomic growth exploded the municipal credit market. Since the on-set of these reforms, local governments have raised US$10 billion from the domestic bond market for infrastructure investments.75. 5.57 One of the key elements of a shift toward private municipal credit markets is to move monitoring municipal financial health by the government to the private market. 5.58 Since the sub-sovereign debt crisis, monitoring municipal credit has shifted almost entirely to the government. Prior to the crisis there were almost no controls on municipal borrowing. Banks and the bond market relied on government as the lender of last resort. Their instincts were correct. When local governments threatened debt defaults, the federal government stepped in and converted the private municipal debt into federal municipal debt. This caused a major pendulum swing in "monitoring" (which was non-existent for the most part) municipal credit back the other direction to significant government intervention in limiting municipal borrowing and lending. The laws limiting the supply and demand of municipal credit also required the federal government to monitor municipal credit. STN became the executing agency for these controls as discussed in the first Section. Is it time for the pendulum to begin its move back to a self-regulatory municipal credit market? 5.59 In the ideal regulatory situation, the government's role is to provide the prudential legal and regulatory rules that govern market interactions between the suppliers of municipal credit and municipal borrowers. The suppliers of credit are responsible for evaluating the municipal credit risks associated with the sale of loans and purchase of bonds. Municipalities follow government's regulatory procedures, but it is incumbent on investors to confirm that local government complies with the rules governing their sale of securities and purchase of loans. This is the current regulatory situation in the mature municipal capital markets in the OECD countries, including Mexico. How can Brazil move toward less government intervention in municipal credit market regulation? 5.60 Below we present recommendations for changes in three critical elements of municipal credit market regulation: · Credit risk analysis and credit ratings; · Disclosure documents, financial accounting and auditing; · Municipal credit monitoring. 5.61 Bank Credit analysis. The BACEN should establish regulatory requirements for municipal credit suppliers in the banking system that supports a healthy municipal credit market. This includes detailed regulations for commercial and government banks to develop an acceptable methodology that measures municipal credit risk and the amount of capital reserves banks need to set aside to cover this risk. This methodology should 75Billand, C (2005) " Municipal Finance: Increasing Local Government Resources To Fund Multi-Sectoral Facilities", page 4. Inputs to a Strategy for Brazilian Cities Page 103 not only be consistent with Basel II accords, as is CEF's current municipal credit risk methodology, but be more broadly defined and use the criteria developed by international credit rating agencies that asses municipal credit risk. This type of regulatory guideline is very different from the credit evaluation of secured and unsecured corporate, real estate and other commercial loans that follow Basel II accords. Governments have issued regulatory guidelines to implement the Basel II principles that assess portfolio credit risk, including Brazil, but many countries do not have specific guidelines for municipal credit risk. These regulatory guidelines should use "best practice" credit risk evaluation criteria established by the three primary international credit rating agencies that are active in the domestic sub-sovereign capital markets throughout the world. These companies have developed a national scale credit rating for Brazil's capital market in addition to their standard global rating scale. The national scale ratings have a direct relationship to global scale ratings that give investors information to compare the unique characteristic of Brazil's municipal credit system to the credit risks associated with municipalities around the world. 5.62 Credit ratings for municipal bonds. Like Securities and Exchange Commissions in other Latin American countries, the CVM requires credit ratings by qualified credit rating agencies for debt instruments registered on the stock exchange. But there are no specific regulations governing the qualifications of credit rating companies that evaluate municipal credit risk. To accelerate the shift toward a self-sustaining prudent municipal credit market, the CVM may consider modifying its current credit rating requirements to a more specific rating requirement for municipal obligations, as in the case of Mexico. 5.63 Another way to more broadly introduce a range of municipal credit ratings into the current bond market is through the offering of a CEF CDO, as described above. International credit rating agencies would undertake the CDO rating. These rating agencies use a complex rating methodology for an ABS security. They prepare a cash flow model that tests the ability of the portfolio's combined debt service loan payments to meet the ABS debt service payment. They assign an underlying rating to each of the municipalities to assess the probability of default of each loan in the ABS portfolio and, consequently, the probability that the CDO will default. 5.64 The primary advantages of the CEF CDO, from a credit rating perspective, are: · To familiarize the capital markets with a range of individual municipal credit risks that will facilitate municipal market development; · To provide CEF with a credit rating benchmark to compare its credit evaluation methodology with the international credit rating agency approach. 5.65 Like other aspects of regulatory control, the principles guiding the credit rating, its methodology and the quality of the firm producing the ratings are critical to maintaining a healthy capital market. The CVM may consider promulgation of regulatory guidelines to approve credit rating companies that perform a municipal credit rating analysis. These guidelines should follow international "best practices" approach for credit evaluation as discussed in the previous paragraph. 5.66 Data quality and audits. An important element of municipal credit ratings is the data required to produce debt and other financial operations indicators, key criteria of a general municipal credit rating. The reliability of the quantitative data that comprises the indicators, usually obtained from local government financial reports, is critical. Municipalities in Brazil are required to produce a substantial amount of timely financial data for STN to construct the indicators that measure municipal compliance with FRL and other laws, as discussed in Section 1.0. This data is not audited by an independent external company, but by a government agency, the Tribunal de Contas. 5.67 The Tribunal de Contas, by law, audits local government financial statements and budgets. It does not have sufficient trained staff in municipal accounting to provide detailed quality audits for Brazil's 5,500 municipalities. The GOB may consider shifting some of the municipal auditing burden from the Tribunal de Contas to private accounting and auditing firms. The Tribunal de Contas can either out-source the audits to private firms, or municipalities can substitute the Tribunal de Contas audits with government approved private accounting firms. This activity should be coordinated with the CVM to insure that any changes it makes in Inputs to a Strategy for Brazilian Cities Page 104 disclosure, accounting and auditing requirements for municipal obligations registered on the stock exchange is consistent with the out-sourcing of this activity by the Tribunal de Contas. 5.68 External audits by private accounting firms can start on a pilot basis and slowly expand as the municipal credit market shifts from CEF and BNDES lending to private market lending. The cost of the audit is borne by the municipality seeking a loan or listing a bond on the stock exchange. 5.69 Disclosure documents. Quality disclosure documents are critical for the development of a prudent private municipal credit market. Clearly, quality disclosure documentation did not exist prior to the sub-sovereign debt crisis. The CVM has regulations regarding disclosure for debt obligations listed on the stock market. But, like quality municipal credit ratings, these are general guidelines for fixed income obligations registered with the CVM. CVM may consider preparation of disclosure document guidelines that insure that the critical risks associated with a municipal debt offering are provided to investors. This is an important to build the institutional investor base for a future municipal bond market. 5.70 In addition to the quality and content of the disclosure, the guidelines may also consider regulations to update the disclosure document with any information that creates a material change in the financial condition of the municipality. This will help support a secondary market for trading municipal obligations. Of course a secondary market cannot develop without a robust primary market, which does not currently exist. Nevertheless, if the CVM considers changes in regulations governing municipal disclosure, it should also include requirements that support secondary market trading. Again, this process should be coordinated with the STN, the government agency responsible for the monitoring the municipal credit market. As STN shifts its role from primary monitor to supervisor of the monitoring process, as recommended below, it can supervise the implementation of the primary and secondary market regulations prepared by the CVM. As the municipal credit market evolves, self- regulating mechanisms, such as those that have evolved in the US and Europe capital markets, would replace the STN. 5.71 Monitoring municipal credit. The STN has established a country-wide municipal reporting system and website that collects data and evaluates municipal compliance with financial and debt indicators required by law. This is a costly operation to build and maintain. In addition, it further entrenches a government bureaucracy in the administrative activities to monitor municipal credit. It may also inhibit the shift toward the market's self- regulation. We suggest that the STN consider out-sourcing this activity to private firms as an initial step to shift municipal credit regulation from the government to a self-regulating activity. The firm would also move the municipal financial data from the STN website to a non-government site for public disclosure. This site can be expanded to act as a repository for municipal disclosure, including any disclosure documents required by CVM for municipal bonds, as discussed above. STN would remain the government regulator of bank compliance with municipal credit law. However, it would use the non-government web site and its access to bank portfolio data to assess bank compliance with GOB's banking laws that pertain to municipal loans, in addition to Basel II requirements. 5.72 To implement this recommendation, STN issues a Request for Proposals (RFP) from private firms to collect municipal financial data and maintain the municipal credit website. The firm provides the STN with the periodic municipal financial reports required by law and identifies any municipality that may be approaching non- compliance with the legal indicators. This is the initial step in the shift of STN as the primary municipal credit regulator to a self-regulating mechanism. When municipalities return to the domestic bond market, the CVM would already have an existing mechanism to supervises municipal credits and may further allow incremental changes toward a self-regulating industry commonly found in mature capital markets in other countries. Inputs to a Strategy for Brazilian Cities Page 105 6. Efficiency of Brazilian Municipalities by Suhas Parandekar* Abstract 6.1 This paper attempts to provide a methodological contribution regarding the study and improvement of municipal efficiency in Brazil. Drawing on the previous literature regarding municipal efficiency, and the particular problems surrounding its measurement, the paper looks to a simple graphical presentation of data regarding municipal efficiency. Instead of drawing a regression line through the scatter cloud of variables, the frontier estimation methodology drops an envelope over the cloud and measures the distance of municipalities on the interior of the cloud from the outer surface, taken to be the efficiency frontier. The main thrust of the paper is that efficiency measures are delicate, but they can be explained in intuitively appealing ways to policy makers in the municipalities. With efficient municipalities being benchmarked under the method, the idea would be for discussion amongst the local actors to derive lessons regarding the institutional structure and other mechanisms that enable some highly efficient municipalities to achieve superior results while using the same or lower amount of resources as compared to municipalities whose efficiency can be improved to reach the frontier. Introduction 6.2 This paper is motivated by the aim to provide a straightforward measure of municipal performance in Brazil, concentrating on the efficiency with which that performance is obtained. Data collection and availability have advanced rapidly in the past few years, and detailed data about Brazilian municipalities is now available on the internet and through other media. A natural question associated with the ready availability of the data concerns its use. Data is of course used by researchers to undertake all manner of complex research, which is a good thing, but an even better thing would be to contribute to the actual improvement in municipal efficiency, through an active involvement of mayors and others who are responsible for implementing policy on the ground. This paper provides one example of how municipal level data can be put to use to encourage policy discussion regarding the efficiency of municipal performance, as a step towards enhancing that efficiency. Why Discuss Municipal Efficiency? 6.3 A process of improvement in accountability at all levels of government has been steadily taking place in Brazil, even as the news has been dominated somewhat by recent political scandals. Under the institution of fiscal federalism, municipal accountability works in two directions - towards the higher level of governments that provide transfers to the municipalities, as well as to the voting citizens who elect the local government. It is in the interest of both higher levels of government and the voters that municipal resources be used as efficiently as possible. Elected representatives and appointed officials in the local government (prefeitura} from the mayor (prefeito) on downwards, are themselves interested in improving performance, but how to achieve that improvement in performance is not a simple question to answer. In the vast tropical heartland of Brazil, good __________________ *Thanks are due to Marize Santos and Alfonso Trevigniani for their help in data collection, and to Cassia Miranda for her help in putting together the references. Thanks to Paul W. Wilson and the Free Software Foundation's GNU Project for making freely available the software to run frontier estimation models. This document was prepared using another free software. LaTeX2e, initially developed by Leslie Lamport. Inputs to a Strategy for Brazilian Cities Page 106 governance practices that enable more effective or efficient provision of local public services often remain confined to the municipality that has put such practices into use, though they could potentially also be adapted and adopted by other municipalities. 6.4 It is unlikely that practices of local government that lead to improved efficiency would be known, let alone emulated by others, unless first there is an identification of such practices. The process of benchmarking and systematizing the diffusion of good practices is first dependent on the identification of possible municipalities where such practices might be found. Once reasonably robust indices of municipal efficiency are developed by researchers, a discussion of the efficiency of municipal performance, involving not only researchers, but also policy practioners and civil society at large would be required to uncover the practices that lie behind the efficient outcomes. Other levels of government and associations of municipalities can then play an instrumental role in disseminating information about municipal performance and good practices, as well as providing a mechanism for vetting of that information. 6.5 The involvement of practitioners on the ground is important because it is easy to make mistakes in efficiency analysis. In particular, the identification of certain municipalities as more efficient than others cannot be taken lightly because of the deep political implications of this analysis. This has a very important methodological implication that drives the approach chosen by this paper. This implication concerns the need for the construction of the efficiency index to be as simple and transparent as possible. Thus, if there is a mistake either in the data or in the researcher's programming, this should be easily identifiable. It is perhaps not a coincidence that in most published research on the kind of efficiency analysis that is relevant to this paper, the authors also provide the raw data that they have used together with the paper itself. Problems in the Measurement of Efficiency 6.6 Measurement of municipal efficiency as a concept is easier to talk about than it is to implement. Efficiency means getting the most output for a given input, or getting to a given level of output with a minimum use of inputs. This much is simple, but then but how does one go about defining the outputs and inputs? Should we consider the general welfare of the municipality as an output, or should one restrict oneself to public services? If we are interested in helping to identify municipalities where it is good to live, to work and to run businesses, clearly we should be interested in the overall level of welfare. If we are interested in a particular aspect of public policy, say whether chartered public hospitals do better than private hospitals, or whether more educated teachers are conducive to better learning, we should probably look at performance regarding only those services. 6.7 If we do decide to examine the provision of multiple services, we would need a plausible story regarding the weighting scheme between the services, and define how exactly the service is to be measured. Of course, even though data availability has improved greatly in the recent past, we are restricted by the availability of data, as in any other empirical endeavour. Then there is the problem of deciding between looking at a static picture, where one considers municipal production as a sort of maintenance of a given flow at a point of time; the analysis would be a "freeze frame" photograph of that flow. Alternatively, one could look at a more realistic dynamic concept of stocks and flows, where efficiency is concerned with progress through time. 6.8 A similar concern about definition and data holds for the input side. Public expenditures (whether financed through local taxes, transfers or user charges) are only a part of what can be considered as inputs on the production of locally provided public services. Thus, a mayor may reasonably claim that the apparently poor performance of his municipality is a result of a complex interplay of historical and social forces that have resulted in a municipality where poverty, unemployment and crime is rampant, in spite of the best efforts of the government at all levels. In fact, this kind of argument is often heard as a reason for poor performance across many levels, ranging from the growth performance of a nation, to the poor reading performance of children in a school located in a poor neighborhood. Of course, at some stage the outcome is endogenous to a particular public policy, but it is very difficult in practice to determine where policy responsibility ends and bad luck begins. Inputs to a Strategy for Brazilian Cities Page 107 6.9 Excuses are of course not the exclusive domain of the public sector. The reason of environmental factors as an explanation for poor results is put forth with equal enthusiasm by a salesman who fails to meet his quarterly sales target, or the chief executive of a giant corporation defending her lack of performance. There is a huge economic literature on information asymmetry, with the associated policy lessons from the literature on mechanism design. In the case of the salesman, the solution is to provide a minimum assured salary, but link the bulk of his remuneration to a commission on sales. Similarly, a portion of CEO compensation is generally linked to company performance through stock options. The literature on mechanism design has the potential to provide insights to the area of municipal performance76. Any decent measure of municipal efficiency needs to address the issue of environmental factors, and even more so if the objective is forward-looking, to understand how efficiency might be improved. 6.10 Another relevant issue regarding the identification of inputs for efficiency measurement is the fact that counting public expenditures is only one way of accounting for inputs. Thus, we should also be looking at the use of physical or real inputs - hospital beds, nurses, educational establishments and so on that are used to produce health and education services. Indeed, differences in efficiency depending on whether the input chosen is expenditure or physical inputs would be revealing regarding the remuneration or price of the inputs. However, measuring physical inputs brings its own attendant problems of the measurement of quality of those inputs. 6.11 Even if we arrive at a tentative understanding regarding the inputs and outputs, a vexing issue that would remain concerns the universe of the municipalities whose efficiency is to be analyzed. Again, most people would agree that sensible efficiency comparisons can only be made between more or less homogenous units that face similar questions of allocations, institutional mechanisms and incentives for public workers and beneficiaries. However, the answer to what constitutes a reasonably homogeneous set of municipalities is not so easy to establish. 6.12 Assuming that reasonable compromises are possible regarding the variables to be used to perform the efficiency measurement, the next question is to determine the methodology to be used. In this regard, we are fortunately on safer ground, thanks again to the exponential increase in computing power over the past few decades that enables one to crank through computationally intensive empirical techniques through brute force rather than mathematical artifice. Empirical Frontier Estimation of Efficiency 6.13 Earlier approaches to efficiency analysis were typically based on the parametric statistical estimation of a production function such as the ubiquitous Cobb-Douglas production function or the more general Constant Elasticity of Substitution (CES) production function. Most typically applied to production process in firms, this kind of analysis sought to arrive at estimates of regression coeffients that would reveal insights about returns to scale, and efficiency analysis could be performed as a derived product of the regression equations. A more recent literature - the so called empirical estimation frontier takes a more direct and intuitively simple approach. 6.14 Rather than make assumptions about the possible theoretical shape and location of the production function, this literature does not assign any restrictions on the nature of the underlying technology of the production function. The method side-steps the question of whether a particular unit is located on or off a theoretically defined production function by focusing attention exclusively on the empirical evidence. 6.15 The approach is particularly attractive when, as in the case of Brazilian municipalities, the definition of inputs and outputs is tentative at best. It would be extremely difficult to come up with a plausible assumption regarding a smooth and well-behaved concave production surface with these alleged inputs and outputs. Indeed, a look at the production surface of municipal performance would probably reveal a picture not of Euclidean smoothness, but of considerable ruggedness, like a mountain range. Indeed because of the multiple influences from measured and unmeasured inputs on the output, with a complex of local forces in addition to global ones, it would be most likely that as in the case of most real mountains, the production surface would have a fractal 76An excellent example can be obtained from Gasparini and Ramos, 2004. Inputs to a Strategy for Brazilian Cities Page 108 dimension. We explore this notion later in this paper. For now, it should just be noted that empirical frontier estimation in its basic form avoids the need for making any sample-based statistical inference. 6.16 The methodology of empirical frontier estimation is best understood with some very elemental graphs (see the next section of this paper). Essentially, instead of estimating a conditional expectation function relating the output to an input through the scatter plot, as in a statistical regression, the empirical frontier estimation seeks to find out the envelope of the entire data cloud. We are not seeking to establish a possible statistical general relation between the input and the output; we are merely interested in identifying who it is that lies on the empirical efficiency frontier, and how far within that envelope are the other municipalities. It is possible that the theoretical efficiency frontier lies even further out, but we are not interested in its location. We are only interested in showing what a real life municipality has actually achieved, and we seek to generate a healthy discussion about how other municipalities can possibly seek to repeat the performance of some of their peers. 6.17 The most basic technique is called the Free Disposable Hull (or FDH); adding the assumption of convexity of inputs would invoke the technique of Data Envelopment Analysis (DEA). Formulations of the DEA method can alternatively impose the restriction of constant, decreasing or increasing returns to scale. Even the FDH method is not completely assumption free, which does constitute a weakness that should be known when using the method to measure municipal performance. This method assumes that inputs are freely disposable, meaning that if a unit is using an excessive amount of one or more inputs, it can get rid of these inputs without incurring any cost. In the Operational Research language of linear programming, the slack has not a price associated with it. In terms of implementing the method, FDH is nothing else but the combination of multiple linear programs, one for each of the units in the analysis. The production frontier is constructed piece-wise out of the output of the various linear programs, which is what makes it computationally intensive. Organization of this Paper 6.18 This paper has four remaining sections. In Section77, we present a brief literature review. Section 3 presents some graphs. The graphs build up successively to the argument that we should examine efficiency more closely within a reasonably homogeneous group defined by groups of municipality defined by size, within a particular state. In Section 4, we present an FDH analysis of some representative groups of municipalities, with alternative specifications. Finally, in Section 5 we examine the influence of environmental or contextual variables on municipal efficiency and perform a tentative investigation of the on efficiency of the specific institutional arrangement of sectoral municipal consortia across neighboring municipalities and of consultative councils within a municipality. Through all the presentation that follows, it bears reiteration that the objective of this paper is not to provide a definitive or final conclusion about municipal efficiency in Brazil; rather, the aim is to stimulate discussion amongst policy makers in municipalities and the policy thinkers who help them. Literature on Frontier Estimation of Municipal Efficiency 6.19 The empirical literature using frontier estimation techniques has literally exploded in the past few years, and there are hundreds, if not thousands of papers available on the subject. In terms of a mathematical introduction to the topic, a useful source is Fried, Lovell and Schmidt (Eds.), 1993 [2]. Other useful textbook length treatments include Charnes,Cooper, Lewin and Seiford (Eds.), 1995 [3] and Zhu, 2002 [4], the last one in the form of a tutorial for practitioners, distributed with an easy to use software that adds on to Microsoft Excel. A periodically updated list of books on frontier estimation is available online at the website DEA Zone. The same website is also linked to a very large list of papers. For the purpose of this paper, the next sub-section of this introductory section focuses attention briefly on the available papers that have used frontier estimation to compare municipal efficiency. 77However, this need not be a fatal assumption. We can still explore options regarding improving efficiency that do not involve reducing inputs, which is generally politically infeasible at least in the short and medium term for municipalities when human resource inputs are considered. Inputs to a Strategy for Brazilian Cities Page 109 6.20 The following brief review of the literature is by no means comprehensive. The papers reviewed are roughly divided into three groups for purpose of classification, this grouping being only one of other possible forms of organization. In the first group are papers that present efficiency as a general concept, taking into consideration all municipal expenditures. The second group consists of papers that examine specific services, such as solid waste collection or education. Finally, a list of papers is discussed that have looked at efficiency in Brazilian municipalities. General Municipal Efficiency 6.21 Borger and Kerstens, 1996 [5]is an early paper examining efficiency in Belgian local governments. The authors analyze total current expenditures as the input for a set of five outputs for a single cross-section of 589 Belgian municipalities for the year 1985. The five outputs considered are (i) the number of beneficiaries of minimal subsistence grants; (ii) the number of students enrolled in local primary schools; (iii) the surface of public recreational facilities; (iv) total population; and (v) the fraction of the population older than 65. The authors note that the overall population is not an `output' in the traditional sense, but is a proxy for administrative tasks undertaken by the municipality, such as maintaining the register of births and deaths, etc., and because direct quantitative data on the magnitude of such tasks was not available. Similarly, the population of older residents serves as a proxy for the supply of social services to the elderly. This kind of pragmatic treatment of inputs and outputs is fairly common in the literature. 6.22 As the title of their paper suggests, the authors wish to compare the findings of efficiency measures calculated by 5 alternative methodologies - two of which, DEA (with variable returns to scale) and FDH are of our special interest. The two principal findings of this paper are worthy of note. First, they find that the assessment of efficiency, generally presented as scores from 0 to 1 varied widely depending on the method used. The estimated mean efficiencies for the 589 municipalities considered as a group, ranged from 0.57 to 0.94, surely a striking difference. The authors also run a second stage regression that seeks to relate each of the five efficiency scores to a mix of variables such as the per capita personal income of the municipality, the population density, and variables intended to capture the political context. Their second main finding is that the coeffients in those regressions are close to one another, even though the efficiency scores vary widely from one another. 6.23 Worthington (2000[6] examines 177 local governments in the State of New South Wales, Australia, again for a single cross-section (1993). The author uses a detailed specification of inputs and outputs, with an attempt at separating the effects of physical inputs and input prices (including average municipal salaries). The noteworthy outcome of this paper for our purpose is the author's comment about how the mathematical programming method is the method of choice when the objective of the exercise is to ``benchmark local governments and peer groups for comparison." The author also notes the special circumstances regarding measurement of efficiency for local government - ``Thus, in cases where the usual axioms of production activity breakdown (i.e. profit maximization) then the programming approach may offer useful insights into the efficiency of these types of industries. This is especially the case with local public sector activities." 6.24 In a related paper, Worthington and Dollery (2002)[7] examine the impact of contextual information, or the so-called `non-discretionary' inputs on the efficiency of the same universe of NSW local governments. The two basic approaches are (i) to include the non-discretionary inputs as if they were regular inputs; and (ii) to use the efficiency scores from a first-stage DEA or FDH analysis using only the discretionary inputs, and use a second stage regression analysis with the efficiency score as the dependent variable and the non-discretionary inputs as independent variables. This solution seems somewhat arbitrary, though the choice depends on whether the objective of the analysis is to benchmark specific units and generate discussion about their performance, or whether it is to establish general patterns regarding the relationship between the various inputs and outputs. 6.25 A number of papers in fact use the two-stage approach. Most papers present the second stage regression as a tobit, given that the efficiency score is truncated at the upper bound of 1. In an innovative variation, Balaguer-Coll, Prior and Tortosa-Ausina, 2003 [8] use nonparametric kernel regression at the second stage in their analysis of Valencia (Spain) municipalities. Their paper includes representations of the three dimensional Inputs to a Strategy for Brazilian Cities Page 110 graphs of joint densities that are most revealing about the underlying relationships. Conceptually, their approach is also more consistent, since they are not really interested in arriving at an inference about a single measure of a regression coefficient. The powerful visual presentation of the output in this paper deserves to be replicated in other contexts. Three more papers round up the first group of papers considering municipal efficiency in general - Michailov, Tomova and Nenkova, 2003[8]; Afonso and Fernandes, 2003 [10]; Loikkanen and Susiluoto, 2004 [11]. These papers look at municipal performance, respectively in Bulgaria, in the Vale do Tejo region of Portugal, and in Finland. The notable aspect of the paper looking at the Finnish municipalities is that it is the only paper considered of the lot so far that repeats the analysis for multiple years between 1994 and 2002. However, rather than presenting an analysis of dynamics, the authors look at average scores for each municipality across each of the years, perhaps in an attempt to enhance the robustness of the results. Municipal Efficiency for Specific Services 6.26 We look at four papers in this group, though it bears repetition that the list of papers is not comprehensive. Two of these papers look at efficiency in the municipal function of collection of solid waste. An example from Australia is discussed in Worthington and Dollery 2000 [12]] and one from Spain in Bosch, Pedraja and Suárez-Pandiello, 2001 [13]. An investigation of efficiency regarding municipal water services can be obtained from Woodbury and Dollery, 2003 [14]. An interesting variation from the study of rubbish is an excellent analysis regarding Norwegian Local Secondary Schools, obtained from Borge and Naper, 2005 [15]. This paper has important features that can be copied for other contexts of efficiency analysis. 6.27 The choice between including contextual variables (per capita income and so on) as inputs in the frontier estimation or as explanatory variables in a second stage tobit regression is arbitrary and does not have a sound conceptual basis. Borge and Naper neatly address the issue by doing the regression first. They are measuring performance in terms of test scores in municipality managed schools. The authors first run a regression of the scores on the socio-economic characteristics of the students (the contextual variable), together with dummy variables for each municipality. They do not run into the Neyman Pearson incidental parameters problem because they use data on more than 50,000 students across 426 municipalities. The authors then use the coefficients on the municipal dummies as the output variable for municipal effiency, this variable representing municipal performance purged of the influence of the quality of residents. In a similar manner, Grosskopf, Hayes, Taylor and Weber, 2000 [16] studied efficiency in school districts in Texas. They use the residuals from test-score regressions as the output, with various institutional and administrative variables as the inputs. Municipal Efficiency in Brazil 6.28 Three sets of papers are available that have used frontier estimation to analyze aspects of municipal efficiency in Brazil. Two of these papers study specific services. Seroa da Motta and Moreira, 2004 [17] study efficiency in the sanitation sector to understand how aspects such as jurisdiction and the nature of management affect efficiency. Marinho, 2003 [18] study municipal efficiency for the health sector for 74 of the 91 municipalities in the State of Rio de Janeiro. The outputs are hospitalizations and consultations per capita, and the hospital mortality rate. Inputs are beds, hospitals and health clinics per capita and the duration of hospital stay and consultations. The author also run a second stage regression on efficiency scores with population and GDP as the dependent variables. The paper's findings relate to the importance of neighborhood spillovers and of municipal consortia to derive economies of scale and capture some of the externalities in an optimal way. The interesting feature of the analysis is that the second stage regressions are run separately for the two groups of efficient and inefficient municipalities. It appears that OLS regressions are used, and it is not clear why the author does not use a Chow test to test for structural differences in the regression coefficients. The presentation of results includes a useful table depicting the relative contribution of each input/output to the efficiency score. 6.29 A set of three related papers examines general efficiency of municipalities for all Brazil. Sampaio de Sousa and Ramos, 2001 [19] study all Brazilian municipalities but in the presently referenced work report findings only for the 1,429 Northeast and 1,334 Southeast municipalities for which data was available. The input Inputs to a Strategy for Brazilian Cities Page 111 variable is municipal current spending for the year 1992, in 1992 cruzeiros, as obtained from the secretariate of the national treasury (STN). Five variables, for the year 1991, are used as outputs, "carefully selected" in the authors' words: (i) total resident population; (ii) domiciles with access to safe water; (iii) domiciles served by garbage collection; (iv) illiterate population; and (v) enrollment in primary and secondary municipal schools. Three methodologies are used, DEA with constant and variable returns to scale, and FDH. The authors then analyze the distribution of the efficiency scores by various groupings of municipalities, and examine the issue of returns to scale, finding that increasing returns to scale are prevalent for municipalities of size smaller than 30,000 inhabitants. An important conclusion of the authors for the present paper regards their finding comparing the DEA and FDH methodology: "Moreover, the convexity of the data set that characterizes the DEA methods is, clearly, in our case, an unjustified assumption when analyzing the efficiency of the Brazilian municipalities". They further add about FDH: "Furthermore, instead of calculating an abstract frontier by referring to a fictitious combination of municipalities, as DEA methods do, this procedure builds up its cost-efficiency frontier by contrasting actually observed municipalities." 6.30 Two other papers (not a comprehensive list) use Brazilian municipal data to make excellent methodological contributions to the general literature. The related papers, Stosic and Sampaio de Sousa 2003 [20] and Sampaio de Sousa and Stosic 2003 [21] examine the issue of leverage of outlier municipalities on the distribution of efficiency scores for other municipalities. The idea is that if a very small number of municipalities outperform all the others, that performance is probably due to a special reason, and it should generally not be expected that other municipalities would be able to emulate such a performance. The method is conceptually quite powerful, it basically consists of taking each municipality out of the dataset in turn and running the DEA (in the case of these papers) again on the new set of municipalities, now reduced by 1, and compare the change in efficiency scores for other municipalities. This method is termed as `jacknifing'. With more than 5,000 municipalities in Brazil, the process is so computationally intensive that it would take (as of 2003), according to the authors' calculations, roughly 7 months on a regular desktop computer (a 1Ghz Pentium III), though figures for other kinds of computers are not mentioned. The authors suggest a bootstrapped random re-sampling approach of 10% of the dataset, and provide findings in support of using such a methodology combining the jacknife and the bootstrap. 6.31 An interesting aspect of the discussion is the evidence the authors present of the need to look at the dataset closely, especially at the fringes of the distribution of the input and output variables (these are fairly standard ones such as current spending for the municipality as an input; student enrollment as an output, to mention two examples). They discuss how the data included municipalities, with the specific names, that reported spending per capita of R$6.00 and R$0.32 and so on, while the dataset mean was R$45.21, all figures in current 2001 Reales. They also note the example of how one municipality, the city of Iperó of about 18,000 habitants had an enrollment per school of 2165 students, roughly 12 times the national average78. A similar discussion relates to the authors' investigation of 20 of the 50 most "inefficient" municipalities in all of Brazil as identified by the DEA model. It turns out that half of those 20 municipalities receive substantial royalties on oil and water. The authors note that their per capita spending levels are very high (3 to 8 times the national average), but those increased costs are not reflected in the services considered. Which begs the question, and then what does that extra earning get for the residents? In any event, this kind of qualitative discussion is definitely needed to extract the maximum value from the preceding quantitative analysis of efficiency scores. Human Development in Brazilian Municipalities There are two parts in this section. We present the reasons for using the two main variables that we look at to analyze efficiency, and we then present some graphical results from the data. 78The authors do no report of any attempt o communicate with the municipal authorities or with the federal Ministry of Education that provided the data to corroborate the information about the schools in Iperó. Inputs to a Strategy for Brazilian Cities Page 112 Human Development Index: IDH-M and Current Spending per Capita 6.32 We choose to look at the Human Development Index for Brazilian municipalities as the output variable of interest for three reasons. First, we hold the belief that the idea of measuring municipal performance needs to be holistic if it is meant to be a contribution to the overall task of economic development in Brazil. Suppose there were a very high level of technical efficiency of service provision in a particular municipality, but the overall standard of living were miserable? What good would that be? Municipal authorities could not use that data to go to say, a group of business investors, and argue in favour of investment in their municipality. On the other hand, it is difficult to imagine how a municipality would have superior indicators on overall welfare, and do very poorly on public services. It does remain a possibility, and by all means, further research can investigate a more delimited aspect of performance. In fact, it is hoped that the findings of this paper would generate a discussion not just in the research community, but amongst practitioners in municipalities, about what parameter would they like to be measured on, and one can speculate that there would be a distribution of preferences. Thus, this paper looks at only one of many possible perspectives on municipal performance. 6.33 A second reason is the pragmatic one on data availability. Thanks to the efforts of the UNDP and other organizations, a fairly high quality dataset on the Human Development Index (IDH-M as the Portuguese acronym) is readily available at the following website: Atlas do Desenvolvimento Humano. Data is available for all municipalities for the two years of 1991 and 2000, a long enough span to permit comparison and analysis. In fact, many other researchers have used the same data set, so this paper would be a useful complement to those efforts. The overall municipal index, IDH-M, (M for Municipality) is a 0-1 normalized index. It is comprised of the simple, unweighted average of three 0-1 normalized indices - IDH-L, or longevity index that measures normalized life expectancy at birth as an indicator of health and wellness; IDH-E (E for Education) that combines the gross enrollment rate for habitants aged 7-22, and the literacy rate, the two combined with respective weights of 1 and 2 and normalized; and finally, IDH-R (R for Renda per capita, or Per Capita Income), again normalized. As with any such index, various criticisms can be made about the choices made, but we are not aware of all the factors considered by the researchers who constructed that index, and take the construction of the index as a given. The same UNDP website also provides the data of the various sub-indices, and future research can very well consider alternative considerations of an index from the sub-indices or from other basic parameters. 6.34 A third reason for choosing the IDH-M as our output variable of choice is methodological. The FDH method can of course handle multiple outputs and inputs. However, when one moves beyond a one-output one- input measurement of efficiency, it loses some of its direct intuitive appeal. As a beginning at least, it is safer to stick to such a simple conceptualization. One of the reasons for the complications when considering multiple outputs or multiple inputs is the issue of weighting between different outputs or inputs. The structure of the weights under a DEA or FDH model depends on the underlying structure of the data, rather than being explicitly assigned by the researcher and the weights are not uniform across the municipalities. Without delving into a mathematical explanation, essentially the outcome is that if a particular municipality does relatively much better (i.e., higher for an output, or lower for an input) on any one indicator, that indicator is assigned a higher weight for that municipality. This system is fair, because the methodology of course does not choose between municipalities, and being `rewarded' on one dimension may be balanced by being `penalized' on another dimension. However, the upshot is that it makes the efficiency scores unstable. It is not clear at all that all variables on the output dimension should be treated so equally, but then if one begins to make discretionary choices about the weighting scheme, the method becomes subject to further criticism. In a methodology that is rather sensitive to such specification issues, we decided to stick with a simple formulation. 6.35 As the single input, we consider municipal current spending per capita. It is the variable most often used in the literature that has been cited, and we do not have any particular reason to deviate, even though it is a far from perfect measure on the input side. The single most damaging criticism of our choice in the case of this paper is that it is contemporaneous. In this section we look at the year 2000 for the IDH-M and the current expenditure per capita. It is clear that current municipal spending does not immediately produce better human development; one should rather look at a combined series of lagged current and capital expenditures, together with adequate Inputs to a Strategy for Brazilian Cities Page 113 controls for the contextual variables that are given to a municipal administration. However, we leave that task for future research. In defense of the pragmatic choice made here, we can only say the following things. First, that even if current expenditures do not fit neatly into a production model, they do reflect an aspect of maintenance of the given level of human development in terms of municipal effort, be it in the areas of health and education services, or more generally for economic development. Second, current expenditures per capita do not usually exhibit wild swings from one year to another, so even if we considered a combined lagged aggregate, it is probable that the cross-sectional variation would not be drastically altered. Finally, later in this paper, we do explore a similar analysis for the year 1991, thus providing at least a partial dynamic aspect of efficiency. Graphical Presentation of the Data 6.36 There are a little over 5,500 municipalities in Brazil, but the STN dataset on current public spending for the year 2000 is available for only 4,617 of those municipalities. The mean municipal current spending of that subset is R$451 (in current reales), the minimum was R$0.25 and the maximum of R$2,817. As referred to earlier, the frontier estimation methods are extremely sensitive to outliers, and some of the very low numbers on spending may indicate data inaccuracies or other issues not related to efficiency. We would like to eliminate these kinds of extreme low-spending municipalities from the sample, as their behavior is clearly not going to be replicable in any general sense. 6.37 The problem is to determine the point of cut-off below which the expenditure information would not be relevant. Rather than pick an absolute number, we decided that it would be less arbitrary to pick the first percentile as the cutoff and drop the bottom 1% of municipalities. In the case of the dataset being considered, the bottom 1% consisted of municipalities reporting current spending of less than R$111.2 per capita. For the sake of symmetry, we also eliminate the top 1% of the sample, with the 99th percentile cutoff being at R$1288.92. It should be noted that for DEA and FDH methods, where we are not making statistical inferences, unlike for regression approaches, this kind of purposive elimination is perfectly valid. We are not interested in determining the value of a regression coefficient, for which it would not be correct to reduce the sample in this way. Figure 6.1 shows a plot of the IDH-M against the spending per capita for all the 4,525 municipalities now in the sample. 6.38 Our objective is not to draw a regression line through this scatter cloud (with or without control variables in orthogonal dimensions) to derive a single value of a B coefficient linking the y-variable to the x-variable, Figure 6.1 Basic IDH-M 2000 Graph for all municipalities Inputs to a Strategy for Brazilian Cities Page 114 but to draw an envelope over the scatter cloud, and measure the distance of individual points relative to the axes and the envelope. But we are not yet ready for that step. Figure 6.2 and Figure 6.3 shows the same scatter plot, but with a third dimension added through the use of color (would probably have to be changed in the eventual report version of this paper to pattern). Figure 6.2 shows the plot with the added information about the size of the municipalities, with green dots representing small municipalities (those below size of 20,000 habitants) and the purple dots representing not small municipalities (above 20,000 habitants). Figure 6.3 shows the plot with the added information about the region of the municipalities, with blue dots representing municipalities in the North- East(NE) of Brazil and red representing all the other regions combined. Figure 6.2 Graph Showing Small and Very Small (below 20,000 Population) Municipalities in Green 6.39 It can be noted from the graphs that in the case of size, the smear appears to be from left to right, with the small municipalities tending to have higher expenditures per capita, but there is a lot of overlap, especially on the vertical dimension. The graph by region on the other hand is quite astounding (at least for the uninitiated) in the extent to which it shows the clear demarcation between the North-East and other regions of Brazil. There is some overlap, but the smear clearly runs from top to bottom, with most NE municipalities occupying a position of low expenditures and low IDH-M. 6.40 In terms of efficiency analysis, where we seek to identify the empirical efficiency frontier, marked by those municipalities that get the most output for a given level of input, or least input for a given level of output, the figures provide an important insight. If the purpose of the efficiency analysis is to move towards helping the inefficient municipalities become more efficient, it would be very important to do so within reasonably homogeneous groups. We next turn to an examination of the scatter plot at a slightly more disaggregated level. We want to compare a section of the small municipalities (those between 5,000 and 20,000 habitants) in two arbitrarily selected States from the South and North-East - Rio Grande do Sul and Alagoas. Inputs to a Strategy for Brazilian Cities Page 115 Figure 6.3 Graph Showing NE Municipalities in Blue, Others Red 6.41 The remarkable aspect of Figure 6.4 is how the municipalities of the two states separate out so neatly, as if there were very little to compare between them. The figure serves to show that it would not be fair to classify Alagoan municipalities as inefficient in comparison to those from Rio Grande do Sul, due to the intervention of a host of historical and other factors, not accounted for in this simple two-dimensional analysis. Specifically, this paper makes the case that if Alagoan municipalities are exhorted to be more efficient, this should be done in a fair comparison to the municipalities in the same state.79 We next turn to look in Figure 6.5 at two continguous municipal sizes, the small municipalities between 5,000 and 20,000 habitants; and medium municipalities between 20,000 and 150,000 habitants. Figure 6.4 Small Alagoas Municipalities with Blue Cross, Small Rio Grande do Sul 79It may also be meaningful to compare within regions, given the ineguities in Brazil are driven so much by region. Even though municipalities in the Brazilian federation are not subservient hierarchically to the State government, the State does provide a range of contextual variables that would be important to efficiency. This is the reason we prefer to make municipal efficiency comparisons within a State. Inputs to a Strategy for Brazilian Cities Page 116 6.42 Lines are added to Figure 6.5 to depict the envelope of the two groups of small and medium municipalities, in a manner that attempts to illustrate the intricacies of the FDH method and the manner that it can be applied for the purpose of encouraging discussion amongst the policy makers of the municipalities in the State of Alagoas. The first thing to note is that there is an outlier small municipality of Satuba (Population 12,555, IDH- M of 0.705, Municipal Expenditure per capita of R$ 204.37) that far outperforms the rest of the small municipalities. This fact is noted, and the envelope for small municipalities (black line) is constructed without Satuba. The envelope for medium municipalities (red, dashed line) is constructed on the basis of the entire available sample for medium municipalities, as no such particular outlier is observed in their case. The efficiency frontier for medium municipalities lies outside the one for small municipalities. Figure 6.5 shows qualitative evidence in support of size groupings within a State, if sufficient numbers are present. Figure 6.5 Alagoas Small Municipalties with a Green `S', Medium Municipalities with a Purple `M' 6.43 The FDH efficiency scores in the output dimension measure the relative position of a municipality vertically from the envelope, and the efficiency score on the input dimension measure the scores horizontally from the envelope. For the purpose of this paper, we focus on the output efficiency, under the assumption that it would be very difficult for municipalities to achieve efficiency by reducing the use of inputs, due to various institutional and political factors, but that it would be feasible, and indeed desirable that municipalities could strive to achieve a better output for the same level of input. To conclude this section of the paper, as an illustrative table to complement the mainly graphical presentation in this section, Table 6.1 depicts the efficiency scores for the most and the least efficient of the small municipalities in the State of Alagoas. Inputs to a Strategy for Brazilian Cities Page 117 Table 6.1 FDH Scores of Output Efficiency for Small Municipalities in Alagoas Municipality Expenditure IDH-M 2000 Efficiency Score Efficient Municipalities Satuba 204.37 0.705 1.00 Barra de S~ao Miguel (AL) 422.87 0.639 1.00 Coit´e do N´oia (AL) 201.48 0.569 1.00 Estrela de Alagoas (AL) 131.61 0.545 1.00 Maribondo (AL) 276.81 0.636 1.00 Olho d' ´ Agua das Flores (AL) 208.02 0.606 1.00 Paripueira (AL) 254.94 0.617 1.00 5 Least Efficient Municipalities Inhapi (AL) 269.66 0.515 0.83 Roteiro (AL) 299.05 0.522 0.82 Branquinha (AL) 292.09 0.513 0.81 Po¸c~o das Trincheiras (AL) 311.54 0.499 0.78 Porto de Pedras (AL) 304.89 0.499 0.78 MEAN FOR 44 municipalities 321.88 0.569 0.91 6.44 The average level of efficiency is 91%, but at the same time, the average IDH-M on 2000 for Alagoas was only 0.57, as compared to a Brazil average of 0.70 and a Southern region average of 0.77. This appears to suggest that while efficiency could be improved in small municipalities in Alagaos, the potential improvement in efficiency would still leave Alagoas behind in terms of human development. While this might seem to be an obvious conclusion that did not require so much of data analysis, it is also the case that discussion about small Brazilian municipalities is often dominated by discussion of their poor institutional capacity. The data examined in this section shows that the story of poor capacity may not be universal. However, deeper analysis would be required as definitive conclusions cannot be drawn from a single illustrative example. Results from FDH Analysis of Efficiency 6.45 In this section we look at some illustrative examples regarding the application of the FDH methodology to study municipal efficiency in Brazil. Comparing North and South on IDH-M 2000 6.46 We look closely at municipal efficiency within two regions, the North and South. Frontier estimation requires a certain minimum number of municipalities, if there are too few in any group, there may be a chance that they may be declared "efficient by default", meaning that it may not be so much that they are efficient as the fact that there are no comparator municipalities. To avoid the problem, we only look at those State and municipal size groupings which reach beyond a handful. Of the 7 States in the North and the 3 States in the North, we cover a total of 352 municipalities in the North, and 1,135 municipalities in the South. In each case, we carry out the efficiency analysis to determine the distance of each municipality from the frontier, without any convexity assumption, i.e., the envelope would similar to the one in Figure 6.5. However, in this section we do not look at the individual municipalities for outliers to provide a correction as we did for the case of small municipalities in Alagoas. We do present the key results in a set of tables in Appendix 1, and the entire results are available in electronic form by requesting to the author. As was stated in the introductory section, the purpose of this exercise is not so much to provide a definitive word on municipal efficiency as it is to engender a discussion amongst the practioners. Size Groupings are Very Small (below 5,000); Small (5,000 to 20,000); Medium (20,000 to 150,000); Large (150,000 to 1,000,000) and Very Large (1,000,000). Inputs to a Strategy for Brazilian Cities Page 118 6.47 Table 6.2 provides a summary of the analysis carried out. The table shows the disparities between the North and South in terms of the output variable being considered here, IDH-M for 2000. However, expenditures in the South tend to be higher as well. The point to note from the table is that within the defined comparator groups, inefficiency is not widespread amongst the municipalities. The municipalities in the States in the North are only slightly less efficient than the ones in the South, though overall outcomes are distinct. Also, it should be noted that within the same State, very small and small municipalities represent much higher expenditures as compared to larger ones, without showing a commensurate improvement in the IDH-M indicator. However, within their individual groupings the average inefficiency is not larger for the smaller municipalities. This could be considered to be an interesting finding. The process of the generation of smaller municipalities in Brazil has stopped, but it seems difficult that it would be politically feasible to consolidate municipalities in Brazil. However, smaller municipalities can get together in consortia with other smaller municipalities or larger municipalities. If smaller municipalities are not inefficient in a deep structural way, and these results seem to indicate that efficiency is not so poor, the mechanism such as consortia may prove to be very helpful. Table 6.2 Means from IDH-M 2000 Efficiency Analysis for North and South State Group Number IDH-M Expenditure Efficiency RONDONIA All 44 0.709 310.92 0.93 ACRE All 15 0.622 326.12 0.87 AMAZONAS All 27 0.637 320.41 0.88 RORAIMA All 13 0.677 244.73 0.93 PARA All 31 0.668 269.30 0.89 AMAPA All 5 0.731 246.36 NA TOCANTINS V Small 44 0.662 537.89 0.94 TOCANTINS Small & Med 38 0.693 308.87 0.88 PARANA V Small 91 0.730 643.27 0.91 PARANA Small 217 0.734 428.38 0.92 PARANA Med & larger 79 0.770 370.44 0.94 SANTA CATARINA V Small 104 0.778 648.43 0.94 SANTA CATARINA Small 136 0.789 397.66 0.93 SANTA CATARINA Med & larger 48 0.820 393.91 0.96 R G DO SUL V Small 193 0.779 704.04 0.92 R G DO SUL Small 177 0.778 459.23 0.93 R G DO SUL Med & larger 90 0.805 384.14 0.96 3.1 Comparing Maranhão and São Paulo on IMR 2000 6.48 We turn to a similar analysis now as in Table 6.2, to examine the Infant Mortality Rate (IMR) and Health expenditures.Figure 6.6 depicts the graph for all 4517 Brazilian municipalities in the sample, the data source is the same UNDP database for IMR and STN for health expenditures. As before we eliminate the bottom and top 1% of the sample on health expenditure per capita, the cutoffs being R$ 12.64 and R$ 291.17, in current 2000 reales. Mean municipal health expenditures (classified in the database as health and sanitation) being R$88.70 and mean IMR being 31.44, with a minimum IMR of 5.38 and maximum IMR of 109.67. For the purpose of efficiency analysis, we take the reciprocal of the IMR as the `output', multiplied by 1000 to retain the notion that more of the output is better. The IMR is of course defined as the number of deaths below 1 year of 1000 live births for the concerned municipality, and is a standard indicator of health status. In this example, we do not present further graphics after Figure 6.6, but delve straight into the efficiency analysis. We look at a comparison between two states, Maranhão in the North-East and São Paulo in the South-East. 6.49 In the case of Maranhão, we look at two groups, comprised of very small and small municipalities in one group (68 municipalities), and medium and large in the other (47 municipalities). In the case of São Paulo, we look at four groups. The first three groups are constituted respectively of very small (170 municipalities), small Inputs to a Strategy for Brazilian Cities Page 119 (221 municipalities) and medium municipalities (186 in number). The last group consists of 37 large and very large municipalities, including So Paulo and Guarulhos, though it might well be argues that very large municipalities belong to a class of their own and should not be compared in the same group as large municipalities. Table 3 presents the mean results and tables in Appendix 2 present selected municipal level results. Figure 6.6 IMR 2000 Graph for all Municipalities 6.50 The notable fact from Table 6.3 is the low level of efficiency as compared to the earlier results for IDH- M. There are multiple possible results for this finding, which should be considered with care. First, the STN data on municipal expenditures by area may of less accuracy than the data on overall expenditures. There are accounting issues of earmarked transfers from the federal government and programs such as the family health program that has Table 6.3 Means from IMR00 Efficiency Analysis for Maranhao and Sao Paulo State Group Number IMR Expenditure Efficiency MARANHAO V. Small & Small 68 60.97 58.77 0.68 MARANHAO Medium & Large 47 58.00 63.39 0.67 SAO PAULO V. Small 170 15.91 163.26 0.54 SAO PAULO Small 221 15.61 117.69 0.54 SAO PAULO Medium 186 14.85 114.13 0.52 SAO PAULO Large & V. Large 37 16.13 157.25 0.78 been accredited with a large positive impact on the reduction of infant mortality80. The private expenditures on health are also not recorded here. Second, a number of municipalities have low health expenditures but have high health performance as they are the so called `dormitory municipalities', close to a larger metropolitan city that provides needed health services in the area. On the flip side, municipalities with high expenditure per habitant may, in fact, be serving a bigger population. Finally, it is a characteristic of frontier estimation techniques that the 80See Macinko, Guanais and Marihho de Souza 2006, [22]. The authors report that after controlling for a set of other variables, a 10% increase in the coverage of the Family Health Program led to a decline of 4.6% in the IMR. Inputs to a Strategy for Brazilian Cities Page 120 variation in efficiency scores closely follows the variation in the input and output variables. As an output variable, even a glance at the mean IMR shows a much greater variation than was the case for IDH-M. Efficiency for IMR 2000 with Physical Input Variables 6.51 To test the hypotheses regarding the ostensibly low efficiency of health performance (as measured by the IMR), we examine an alternative specification of inputs. Information is available from the IBGE of a number of physical inputs that go into health services. We choose the following, that includes both public and private resources in the municipality - number of health centres (of all kinds), number of doctors, number of nurses, and number of hospital beds. We also use IBGE data about contemporaneous (year 2000) percentage of households with access to safe drinking water and the percentage of households benefiting from garbage collection services. The municipal level tables are presented in Appendix 2, and a summary of the means is presented in Table 6.4. Table 6.4 Means from IMR00 Efficiency for Maranhao and Sao Paulo State Sample Centre Doctor Nurse Bed Agua Lixo IMR EFF Maranh~ao 1 37 5 9 3 33 16% 22% 62.73 0.86 Maranh~ao 2 45 20 100 26 208 29% 44% 57.45 0.88 S~ao Paulo 1 14 2 10 2 32 97% 99% 14.79 0.93 S~ao Paulo 2 119 4 24 4 44 97% 99% 15.73 0.88 S~ao Paulo 3 175 17 141 20 194 97% 98% 14.73 0.78 S~ao Paulo 4 37 128 2,365 437 1,433 98% 99% 16.13 0.91 6.52 As expected, the efficiency scores for health performance go up when compared to using recorded public municipal health and sanitation expenditures. However, the magnitude of the increase is considerable, with the least efficient group, very small municipalities in the State of Maranho indicating a mean efficiency level of 0.88. The remarkable change in efficiency scores indicates an important methodological point. It has been stated before that a useful purpose of efficiency analysis is to publicize and foment public discussion of efficiency scores as a precursor to the identification of high performing municipalities that set benchmarks for other municipalities to emulate. If the benchmark municipalities are obtaining higher performance with the same level of resources, it is probable that such performance is due to better institutional features and behavioural incentives. 6.53 In some cases, analytical quantitative research may be able to diagnose the drivers of such performance. However, it is likely that the complexity of local conditions account for the variation in the performance, and the true reasons for superior performance can only be encountered through interaction with the local actors - the so called positive deviance' methodology, as described by Pascale and Sternin, 2005 [23]. Even if one were to restrict attention to available quantitative data, and not go to the next stage of policy discussion with practitioners, it remains the case that efficiency scores need to be reasonably robust. The illustrative example considered here for health performance underscores the critical importance of the specification of inputs and outputs, as efficiency scores may alter radically from one specification to the next. In policy terms, the results from the IMR analysis appears to reconfirm the finding that enhancing municipal efficiency would only go so far in terms of improving human development in Brazilian municipalities. Comparison between 1991 and 2000 6.54 So far the presentation has been of a static nature, examining the efficient frontier, as it existed in 2000. However, a most interesting discussion is the manner in which the shape and location of the frontier shifts over the years. Since the empirical frontier estimation method is not based on a theoretical production function, it is intriguing why such an attempt to trace out the movements of municipalities has not been presented before in the literature. In this sub-section we undertake a graphical examination of the movement of the efficiency frontier Inputs to a Strategy for Brazilian Cities Page 121 considering the IDH-M between 1991 and 2000, tracked against per capita municipal current expenditures in constant 2000 reales to enable comparison. Table 6.5 Comparing IDH-M and Expenditures between 1991 and 2000 Obs MUNICIPIO EXPER91 IDHM 91 EXPER00 IDHM 00 1 Bel´em (PA) 264.160 0.767 275.723 0.806 2 Manaus (AM) 351.246 0.745 308.257 0.774 3 Salvador (BA) 275.517 0.751 315.851 0.805 4 Fortaleza (CE) 260.732 0.717 340.005 0.786 5 Recife (PE) 462.506 0.740 465.784 0.797 6 Goi^ania (GO) 318.530 0.778 528.879 0.832 7 S~o Paulo (SP) 775.973 0.805 549.181 0.841 8 Guarulhos (SP) 540.284 0.762 559.520 0.798 9 Belo Horizonte (MG) 413.931 0.791 581.166 0.839 10 Rio de Janeiro (RJ) 723.605 0.798 679.026 0.842 11 Curitiba (PR) 450.747 0.799 779.434 0.856 12 Porto Alegre (RS) 525.328 0.824 816.746 0.865 6.55 A better depiction of the data above can be obtained by looking at the figure showing the same data. In Figure 6.7, the points of origin of 1991 are denoted by a + sign and the point of destination of 2000 denoted by a sign. Of course, we do not the know the precise trajectory between 1991 and 2000; the path may well have been wavy or looped or jagged. To facilitate the reading of the graph, the cross is in blue color and the triangle in green. Appendix 2 provides detailed graphs with labels indicating each municipality. The lines that go up vertically pertain to the North and North-East municipalities. The lines for the two South municipalities slope to the right, while Rio de Janeiro and São Paulo show leftward slopes, marked in the case of São Paulo. On the whole, Table 6.5 and Figure 6.7 show an overall positive picture regarding efficiency, or at least one that is not declining. A more rigorous quantitative investigation would surely provide some interesting conclusions. Figure 6.7 Basic IMR 2000 Graph for all Municipalities Inputs to a Strategy for Brazilian Cities Page 122 FDH after Correction for Contextual Variables: Understanding Drivers of Efficiency Outcome Variable of Interest 6.56 In this section, we examine the evolution of a particular outcome variable of interest. The variable is chosen from the education sector, but the methodology would be applicable to other output variables, such as the ones considered in the previous section, namely the IDH and the IMR. The literature regarding municipal performance insofar as education is concerned generally looks at enrollment, i.e the number of students enrolled. However, the use of enrollment as an output variable when coverage is nearly universal seems to be an unnecessary compromise. Coverage in Brazil is nearly universal for Primary Education in Brazil, which is the level of education where Municipalities predominate in the provision of educational services. Variation in the number of children enrolled is therefore only indirectly an effect of municipal performance, through possible Tiebout effects and stochastic demographic variations related to performance in the long run. In view of the policy focus of this paper, which seeks to generate and inform the debate about municipal performance, an education outcome variable is required that is more directly related to actions that are undertaken or neglected by municipal administrations. 6.57 It is generally regarded that value added standardized test scores of students are the best possible measure of the quality of education services, though there is some dissension amongst scholars about this point of view. Thus one measures student achievement at the beginning of an educational course, and then measures the achievement of the same students in a valid, reliable and accurate way after the completion of the course. Without getting into the merits or otherwise of this `ideal' measure of educational quality, the fact of the matter is that this kind of data is simply not available for Brazilian municipalities. Also, even though coverage of Primary Education in Brazil is universal, primary completion is not universal. Typically, because of a combination of disadvantaged family contexts including low income, and poor quality of service provision, students begin to lag behind in their studies, end up having to repeat the same grade rather than moving to the next one at the end of the year, and after a few repeated episodes of repetition, drop out of school altogether. Even the children who may repeat grades but not drop out are victims of low self-esteem and motivation. The probability for further education is hampered by grade repetition, and an individual's productivity and labor market prospects are also damaged. Grade repetition as a phenomenon is one that is eminently treatable by policy action on the part of the municipal administration running municipal schools. The effects of municipal action would clearly be stronger for municipal schools, because more policy variables are under municipal control. However, in collaboration with State governments in Brazil, municipalities can and do play a role even in the grade repetition in State run schools, and arguably also in private schools, because of the flows of students from one system to another, and the overall context of educational service provision in a municipality. 6.58 Students repeat grades because they do not learn the material in the curriculum designed for them within the year. There may be multiple factors that affect grade repetition, or educational service provision more generally. These factors can be classified into roughly three groups: First, the education of a child is a joint outcome of a child's experience at school and at home; after all, a child spends 5 or 6 times more of his or her time at home or away from school rather than in the school, and there is only so much individualized attention that can feasibly be provided at school. Second, a lot depends on the physical availability and quality of educational inputs at the school - simple things such as the physical layout of the school building in terms of sound protection and light availability, furniture, sports fields and equipment, trees and gardens and recreational areas in general, textbooks, other didactic material, the presence of teachers and so on. Third, there are a number of educational policy variables that specifically affect grade repetition, in addition to policy effects through the provision of inputs - the entire system of training and pedagogical and administrative support provided by the municipal administration to schools, the human resource policies in place for the municipality, the presence of remedial programs for children from disadvantaged backgrounds, policies to engender community participation and collaboration, cooperation with other levels of government, with neighbouring municipalities, and so on. Inputs to a Strategy for Brazilian Cities Page 123 6.59 In this last section of the paper, we provide a preliminary examination of a particular way to measure grade repetition, the Age Grade Distortion Rate. The data is readily available from the same source, the Atlas of Human Development for Brazil, prepared by the UNDP with other collaborating organizations. The variable in question is the Age Grade Distortion Rate of 1 year or more. It captures the cumulative effect of grade repetitions, because in fact it is this cumulative effect that is most damaging to children, rather than a possibility isolated case of repetition in one year. It also provides a unique combination of quantitative and qualitative factors regarding service provision. The indicator used here measures the percentage or proportion of students who have repeated at least one grade. It takes as the denominator all the children enrolled in a certain age group; here the age group of 7 to 14 years old. As the numerator, which is a subset of the denominator group of children, it uses the number of children who are behind their cohort, by 1 or more years, due primarily to repetition. It is an indicator that does not pose problems of aggregation across schools within a municipality and does not require assumptions regarding the population for reference, as is the case, for instance, with enrollment rates. A municipality that provides better educational services (quantitatively and qualitatively) would have a lower Age Grade distortion rate, and poor service provision would likely show up in a higher Age Grade distortion rate, though two such municipalities of the same size and in the same State might have identical enrollment. 6.60 Since the level of the Age-Grade distortion rate (henceforth ADR) itself may be a manifestation of a long past trajectory, from a policy variable, it would be useful to examine the change or evolution in the Age-Grade distortion rate, over a sufficiently long but not too long period of time. We use the change in the Age-Grade distortion rate, between the two years of 1991 and 2000, for the municipality as whole, that is not restricting attention to municipal schools, which would account for an average roughly between one-half and two-thirds of the enrollment for that age group for the period under consideration. In fact, there was an accelerated process of municipalization of education in that time period, which provides further reason to consider all schools in the municipality rather than just municipal schools. World Bank, 2002 [24] provides a detailed analysis of the context and consequences of this process of municipalization. Evoluation of Age-Grade Distortion Rate (ADR) between 1991 and 2000 6.61 Considering the 5508 municipalities for which data is available, the ADR for Brazil was 45.67 in 1991, and it declined to 28.88 in 2000, still considerably high, but representing a dramatic improvement. The remarkable fact is that the reduction of the ADR took place for all of Brazil; of the more than five thousand municipalities in Brazil, only two reported an increase in the ADR over the period 1991-2000. A look at the evolution of the municipal distribution between 1991 and 2000 provides an interesting picture, as seen in Figure 6.8 6.62 The municipal distribution for 1991, shown in blue with right slanting lines filling the kernel density function has two distinctive peaks; it is interesting to note that with the leftward shift in 2000, the peak on the right (showing much higher age-grade distortion) has flattened considerably. The graph for All Brazil is actually a combination of two groups of unimodal distributions. Two representative groups are shown as Figure 6.9 and Figure 6.10, pertaining respectively to the municipalities in the North-East and South, respectively. Inputs to a Strategy for Brazilian Cities Page 124 Figure 6.8 Evolution of Age-Grade Distortion (ADR) All Brazil: 1991 - 2000 6.63 It should be noted that compared to a Brazil average of 45.67 for the ADR in 1991, the mean for the North-East was 61.94, as the distribution moved to the left, the variance increased, and the mean of 46.69 for 2000, still higher than what it was for Brazil in 1991. At the other extreme one finds (Figure 6.10) that the mean ADR for South Brazil declined from 28.03 to 13.91. The graph also shows how the variance was reduced, even though there is a tendency for a fat right tail to remain. 6.64 In terms of analysis of municipal efficiency, the interesting question to answer concerns the relation between reductions in the ADR to the expenditure for educational services incurred by municipalities. If it were possible to identify municipalities that were able to achieve higher levels of reduction in the ADR for given level of expenditures, the institutional arrangements and policies of those efficient municipalities would serve as benchmarks or guideposts for other municipalities. As has been stated before, the ADR is a good bellwether of educational services, and through educational inputs and policies, it is possible to improve the provision of educational services. Inputs to a Strategy for Brazilian Cities Page 125 Figure 6.9 Evolution of Age-Grade Distortion (ADR) North-East: 1991 - 2000 The Effect of Contextual Variables 6.65 In the literature review section of this paper, we mentioned the problem faced by researchers attempting empirical production frontier analysis. Contextual variables affect the output variable, but it is conceptually difficult to integrate such contextual variables as 'inputs'. Second stage regressions of efficiency scores on the contextual variables have their own problem, because of truncation problems and lack of clarity about the meaning of the efficiency score that excludes contextual variables. We adopt a pragmatic approach, that is closest to Grosskopf, Hayes, Taylor and Weber (2000) [16]. We choose a set of readily available contextual variables that would affect the output independently of our chosen policy variable of municipal education expenditures as an input. We run an OLS regression of the reduction in ADR, the variable called DEL_AD714, as the dependent variable on a set of regressors that serve as our contextual variables (detailed below). We then take the residuals from that regression, and add on the residuals to the mean level of DEL_AD714, a step that follows Grosskopf, et. al and that is required because we need an output variable that is above zero for all cases. This adjusted ADR or ADJ_ADL serves as the output variable that in a sense has been purged of the influence of the contextual variables. This would become clearer after a presentation of our regression results. Inputs to a Strategy for Brazilian Cities Page 126 Figure 6.10 Evolution of Age-Grade Distortion (ADR) South: 1991 - 2000 6.66 First of all, even though municipalities are autonomous of the State government, and also because we choose the reduction in ADR for all schools as the output of choice, we include dummy variables for each State, to take account of the fixed effect of the State. Using São Paulo as the excluded dummy, we have a set of 25 dummies for the 26 States in Brazil, for our database of 4605 municipalities for which education expenditure data is available. We use five other contextual variables - RPC91, the per capita income of the municipality in 1991, as an indicator of the general economic development of the municipality; PERQMW91, the percentage of the individuals in the municipality who in 1991 lived in households where the income per person was less than a quarter of the existing minimum wage - this serves as an indicator of income inequality; AN_INST that depicts the year of formation of the municipality, to capture the effect of the splintering of municipalities and the consequent institutional dismembering; DIST_CAPITAL that measures the geographical distance of the municipal centre from the State Capital; and TOTPOP_91, the total population of the municipality in 1991. Table 6.6 presents the means and standard deviations the variables used in the regression. Inputs to a Strategy for Brazilian Cities Page 127 Table 6.6 Description of Variables Ued in Regression VARIABLE DESCRIPTION MEAN STANDARD DEVIATION D11 Dummy (RONDONIA) 0.0095548 0.0972913 D12 Dummy (ACRE) 0.0032573 0.0569862 D13 Dummy (AMAZONAS) 0.0058632 0.0763550 D14 Dummy (RORAIMA) 0.0028230 0.0530628 D15 Dummy (PARA) 0.0065147 0.0804588 D16 Dummy (AMAPA) 0.0010858 0.0329368 D17 Dummy (TOCANTINS) 0.0184582 0.1346159 D21 Dummy (MARANHAO) 0.0254072 0.1573754 D22 Dummy (PIAUI) 0.0249729 0.1560593 D23 Dummy (CEARA) 0.0280130 0.1650279 D24 Dummy (RIO GRANDE DO NORTE) 0.0269273 0.1618884 D25 Dummy (PARAIBA) 0.0390879 0.1938253 D26 Dummy (PERNAMBUCO) 0.0295331 0.1693137 D27 Dummy (ALAGOAS) 0.0191097 0.1369254 D28 Dummy (SERGIPE) 0.0117264 0.1076633 D29 Dummy (BAHIA) 0.0699240 0.2550466 D31 Dummy (MINAS GERAIS) 0.1730727 0.3783512 D32 Dummy (ESPIRITO SANTO) 0.0162866 0.1265894 D33 Dummy (RIO DE JANEIRO) 0.0193268 0.1376859 D41 Dummy (PARANA) 0.0849077 0.2787746 D42 Dummy (SANTA CATARINA) 0.0631922 0.2433347 D43 Dummy (RIO GRANDE DO SUL) 0.1001086 0.3001773 D50 Dummy (MATO GROSSO DO SUL) 0.0156352 0.1240728 D51 Dummy (MATO GROSSO) 0.0212812 0.1443359 D52 Dummy (GOIAS) 0.0464712 0.2105262 RPC91 Income per Capita 1991 131.9422085 74.1038848 PERQMW91 % Individuals P1). 119Recife's master plan is being revised and these coefficients will be defined in 1, with the possibility to reach up to 3, even though local real state developers don' t agree with it. 120Not accidentally, one of the prime designations of the mechanism of selling extra construction rights was solo criado (created land). Inputs to a Strategy for Brazilian Cities Page 230 Figure 9.1 Simple supply and demand curves. 9.85 If zoning controls are effective in avoiding negative externalities, the characteristics responsible for the attractiveness of regulated zones are preserved and, consequently, demand and land prices are kept high. As such, regulation is able to affect housing prices not only by reducing housing supply, but also by affecting positively other variables that cause land prices to be high. A new supply curve is represented by S3 and the new equilibrium point is C. At this point the housing produced is H3 and the new price per sq mt is P3 that is greater than P2 and P1. 9.86 This is a simplified model that illustrates possible effects from regulation over housing supply and prices. Even if extra building rights are conceded to developers to produce more housing units to respond to demand, the costs from that will be integrated in final housing prices and the effects pointed above would be still present and the effects on prices could trigger mechanisms of social exclusion. The gains from preventing negative externalities and from selling extra building rights, in turn, could be exceeded by social costs from spatial segregation and operational costs from providing urban services in a sprawled city. Some Stylized Facts about Brazilian Cities 9.87 Among the several features presented by Brazilian cities there is a remarkable spatial pattern of population distribution in which low income people are located far from the central areas, in opposition to the central location of the high income people. As seen in Section 2, there are many reasons that can be detached to explain location and high land prices at the central areas. A possible one, which certainly is not the unique, is that spatial improvements and land use controls carried out by public intervention preferably at the central areas, while peripheral areas were neglected, reinforced their advantages driving urban formal markets to them. As a result of land high prices, associated with high transportation costs, central areas were occupied by high income groups, while low income population were pushed to the periphery. 9.88 It is possible to measure the pattern of distribution of households across urban areas as function of their distance to the city center. Table 9.1 and Table 9.2, and Figure 9.2, show basic facts about a sample made of 10 Inputs to a Strategy for Brazilian Cities Page 231 Brazilian cities121 that are cores of large Brazilian metropolitan areas. Note that the results presented are related to the cities, not to the entire metropolitan area. 9.89 The sample presents cities from different Brazilian regions that grew under different contexts. Some of them faced consistent industrialization process and became the core of large metropolitan regions as São Paulo, Rio de Janeiro and Belo Horizonte. São Paulo and Rio de Janeiro are the largest cities in Brazil and are among the largest cities in Latin America. In fact, these cities polarized intense population flows from overall country and experienced fast urbanization processes during the second half of the 20th Century. 9.90 Others also developed important industrial parks; however, the trend of their influence was exerted mainly at a regional level like Porto Alegre, Curitiba, Salvador and Recife. These cities also experienced fast urban growth during the same period. Other cities faced intense urban growth until the end of the century like Belém and Fortaleza and are located in regions where per capita income is very low. Finally, Brasília122, a city built to be the capital of the country and conceived under the modernist city paradigms. The principles of the modernist city prescribed a city organized in different land use sectors, not mixed, where the use of car would be largely stimulated. As such, the city adopts a very restrictive control on land use in order to preserve its features, since it was assigned as Cultural Heritage of the World by UNESCO. 9.91 The calculations were made by using the households' information from the Demographic Census of 2000 carried out by the Brazilian Bureau of Statistics, IBGE. As such, the distribution of households by income was calculated with respect to the medium wage earned monthly by the person in charge of the household, expressed in terms of Minimum Salary (MS)123. The geographic level of aggregation of the data was the neighborhoods in all cities, with the exception of São Paulo, where was used Districts (96) and Brasilia, where the data were gathered in sub-areas, in a total of 56, since the city does not adopts a neighborhood division and the territorial unit, the Administrative Regions, are not suitable for our objectives. 9.92 As seen in Table 9.1, the cities located at North and Northeast regions have the lower GDP per capita. Not accidentally, more than 60% of their households are in the band of 3 MS or less of medium earnings of the person in charge of the household. In addition, if were considered the wages up to the band of 5 MS, the proportion of households is above 75% in Belém, Fortaleza and Salvador, and about 72% in Recife, the highest ratios presented in the sample. 9.93 On the other hand, the ratio of households in the superior band, above 20 Minimum Salaries (MS), is below than 5% in those three cities and about 6.6% in Recife, the lowest ratios found. The cities located in the Southeast and South regions presented a proportion smaller than 46% of households within the 3 MS band or less, while the households within the band of 20 MS and above was at least 9%. 9.94 Albeit the highest GDP per capita, the amount of households in Brasilia124 that is within the lowest wage band is proportionally similar of that in cities from the South and Southeast Regions of our sample. However, Brasilia and Porto Alegre concentrated about 11.6% of the households within the superior band of wages, the highest ratio found. 9.95 Despite the total of households in Brasilia is similar to other cities, like Curitiba or Fortaleza, the distortion of their distribution across city is remarkable. The medium distance of households to the city center for all cities of the sample, except São Paulo, Rio de Janeiro and Brasília, is about 7.4 kilometers. São Paulo and Rio de Janeiro that are in the largest metropolitan areas of the country have average distance of about 13.6 km and 19.1 km, respectively. Brasilia, in turn, presents a mean of 19.2 km per household to the city center. Note that, 121Belém, Fortaleza, Recife, Salvador, Belo Horizonte, Rio de Janeiro, São Paulo, Curitiba, Porto Alegre e Brasília. 122Even though there are people who admit the use of the name 'Brasilia' only to designate the Plano Piloto, that is, the area object of the heritage assignment, here Brasilia is used to represent the set of urban areas within the Federal District since they form, in fact, a unique city despite its fragmented structure. 123The monthly minimum salary at the census date was R$ 151.00 equivalent about USD 83.50. 124In Brasilia the presence of the Federal Government and public enterprises whose financial records are registered in local GDP can overestimate this indicator. Inputs to a Strategy for Brazilian Cities Page 232 while São Paulo and Rio de Janeiro have about 3 and 1.8 million of households, Brasilia has only about 0.5 million of households, similar to cities like Curitiba, Fortaleza or Belo Horizonte, for example, that present a average distance to the city center of about 7 km per household. Table 9.1 Basic Facts about 10 Brazilian Cities in 2000 Total Households concentration (%) and average distance (km) per household to the GDP per permanent city center according to the income of the person in charge of the household capita (R$) households City 2000 up to 3 MS 3 to 5 MS 5 to 10 MS 10 to 15 MS 15 to 20 MS above 20 MS Belém 4,287.00 Households 296,352 62.7% 12.5% 13.4% 4.1% 3.0% 4.3% Average distance 7.7 km 8.6 km 7.9 km 6.8 km 5.7 km 4.7 km 3.7 km Fortaleza 4,516.00 Households 526,079 65.8% 11.1% 11.7% 4.0% 3.0% 4.5% Average distance 7.9 km 8.6 km 7.9 km 6.8 km 5.8 km 5.3 km 4.8 km Recife 6,585.00 Households 376,022 61.4% 10.4% 12.8% 4.9% 3.9% 6.6% Average distance 6.7 km 6.9 km 6.8 km 6.4 km 6.1 km 5.9 km 5.8 km Salvador 3,926.00 Households 651,278 62.8% 12.4% 12.9% 4.3% 3.1% 4.6% Average distance 8.0 km 8.5 km 7.8 km 7.0 km 6.5 km 6.3 km 6.0 km Belo Horizonte 7,130.00 Households 628,442 45.7% 15.0% 18.6% 6.6% 5.2% 8.9% Average distance 6.9 km 7.9 km 7.6 km 6.5 km 5.4 km 4.8 km 4.0 km Rio de Janeiro 9,818.00 Households 1,802,34 7 42.4% 15.6% 20.8% 7.0% 5.4% 8.8% Average distance 19.1 km 22.0 km 20.4 km 18.0 km 15.1 km 13.4 km 12.1 km São Paulo 12,154.00 Households 2,985,97 7 40.1% 17.9% 21.0% 6.4% 5.2% 9.4% Average distance 13.6 km 15.7 km 14.7 km 12.8 km 10.7 km 9.5 km 8.1 km Curitiba 8,087.00 Households 471,163 37.0% 18.0% 22.2% 7.4% 6.0% 9.3% Average distance 7.3 km 8.6 km 8.1 km 6.9 km 5.6 km 4.8 km 4.0 km Porto Alegre 8,764.00 Households 440,557 35.8% 15.5% 22.0% 8.3% 6.9% 11.5% Average distance 7.2 km 8.8 km 7.9 km 6.6 km 5.6 km 5.3 km 4.9 km Brasília 14,223.00 Households 547,656 43.4% 13.6% 18.3% 7.2% 5.9% 11.6% Average distance 19.2 km 23.0 km 22.2 km 19.4 km 15.5 km 12.7 km 9.3 km Data Source: Census 2000, IBGE. 9.96 In addition, the mean distance to the city center of households in the band of 3 SM or less in Brasília is 23 km, the highest, followed by Rio de Janeiro and São Paulo with 22 and 15.7 km, respectively. In the other cities the average distance of the households of this band is around 8.3 km. Note that the average distance to the city center decreases monotonically as the mean wage of the person in charge of the household increases. As such, the Inputs to a Strategy for Brazilian Cities Page 233 distance of the poorest households from the city center in Brazilian cities is greater than overall mean distance and, in some cases, more than twice the mean distance of the richest households. 9.97 Overall mean distance of households within the band of 20 SM or above is about 4.7 km. The exceptions are São Paulo and Rio de Janeiro that presented a mean distance of about 8 and 12 km per household, respectively and Brasília that presented an average distance of about 9.3 km. The medium distances presented by Brasilia are compared only with São Paulo and Rio de Janeiro, even though the amount of households in it is smaller than in those cities. This is because not only the original conception of the Plano Piloto, characterized by a less compact built area intensive in land consumption and preserved by strong land use controls, but mainly for planning policies that produced a fragmented urban structure that abate overall densities. 9.98 The effects of households distribution in these cities also is noted when the gradient of household density is calculated for all the cities and all the salary bands. The gradient is a formalized exponential negative function (D=D0 . e-Gd) that is largely employed to assess the rate in which density varies with distance. The expression is easily calculated using empirical data for densities at the several locations of the city with relation their distance to the city center and taking their logarithms in a linear regression. 9.99 Table 9.2 shows that except for Brasilia, overall household density decreases from city center towards the outer limits of the cities. Porto Alegre and Curitiba presented density gradients of -0.187 and -0.151, respectively. This signifies that household density decreases at a rate of 19% per kilometer from the city center in Porto Alegre and 15% in Curitiba. These two cities are the most centralized of the sample. São Paulo and Rio de Janeiro, the largest cities of our sample, are the most decentralized and presented rates of 4.8% and 6.1%, respectively. The distance is a significant regressor at a 95% level in all the 9 cities. 9.100 However, regressions' outcomes for the adjusted-R² lower than 50% show that the variation of the distance explains only partially the variation of the households' density from the city center. In other words, other variables influence the spatial distribution of households across urban areas. Cities that have strong topographic and geographic constrains as Recife, Salvador, Belo Horizonte and Rio de Janeiro, for example, showed low values for the adjusted-R² while the more compact ones presented higher values, like Curitiba, Porto Alegre and Belém. Thus, variables that identify other aspects of the cities could be added to the model to improve its explanatory force. Nevertheless, at this point, our interest is in assessing the effect of distance. 9.101 Brasilia, in turn, presents a different pattern of household distribution. The distance is not significant in explaining distribution of households across its urban area. This fact reflects the pattern of urban growth that the city faced since its inauguration in 1960. Land allocation in Brasilia is driven not by market forces, but is commanded by a bureaucratic decision system that is unusually facilitated by the public property of the majority of land available for urban development. In addition, the functionalist spatial features of the Plano Piloto is preserved by very restrictive land use controls that block the development of the empty spaces and changes in the exclusive land uses prescript by the original plans. 9.102 The high land prices presented by the exclusive residential areas within the Plano Piloto can be pointed as a result of this context in which are associated a commanded land allocation and very restrictive land use controls that constrain housing supply within the preserved area. Further, while low densities are maintained within central areas the greatest densities are found far from the city center in settlements produced by the government to absorb city growth, in special that represented by the low income population. Brasilia, due to its context, cannot be considered a good example to represent a typical Brazilian pattern of urban growth. However, in spite of its unique spatial characteristics, the Brazilian capital can be useful to study the effects of commanded land allocation and strong land use controls. 9.103 Table 9.2 also shows regression outcomes for the variation of a household concentration index in each wage bands by the distance from the city center. The household concentration index is a ratio between the relative participation of the amount of households in each salary band within each neighborhood and the relative participation of the neighborhood's households in the total of the city. The outcomes show that in all the cities of the sample the proportion of poor households increase with distance, at a rate that varies from 1.9% in Belém to Inputs to a Strategy for Brazilian Cities Page 234 10.4% in Curitiba per kilometer. Curitiba, Belo Horizonte and Recife were the cities that presented the highest variation rate, that is, the poorest households are more decentralized, with rates about 10%. The variation rate is high in Porto Alegre and Brasilia as well, of about 8% and 7% respectively. 9.104 The distance as a regressor is significant at a 95% level for all cities of the sample. The adjusted-R² showed better results than the regressions for overall density of households' variation. In cities like Curitiba and São Paulo the adjusted-R² reached values near 50%. In others, like Porto Alegre and Brasilia, at least 33% of decentralization of the poorest households is explained by distance. In cities like Recife, Rio de Janeiro and Belo Horizonte the adjusted-R² presented low values, due to the reasons already mentioned. The largest cities of São Paulo and Rio de Janeiro presented low rates of poor household decentralization, maybe because of the extension of their urban areas as well as the existence of poor households near central areas, such as favelas and cortiços. 235e MS R² 0.094 0.227 0.155 0.082 0.304 0.119 0.579 0.738 0.249 0.268 Pag 20 Adjusted above Gradient -0.032 (0.0053) -0.206 (0.0000) -0.264 (0.0001) -0.106 (0.0034) -0.261 (0.0000) -0.047 (0.0000) -0.144 (0.0000) -0.275 (0.0000) -0.140 (0.0000) -0.100 (0.0000) R² MS 0.103 0.281 0.136 0.118 0.215 0.132 0.682 0.677 0.338 0.244 Adjusted Salaries) 20 to 2000 15 in Gradient -0.030 (0.0037) -0.185 (0.0000) -0.208 (0.0002) -0.107 (0.0005) -0.171 (0.0000) -0.036 (0.0000) -0.097 (0.0000) -0.195 (0.0000) -0.119 (0.0000) -0.068 (0.0000) (Minimum Cities R² MS 0.151 0.292 0.128 0.135 0.171 0.124 0.732 0.670 0.324 0.127 household Adjusted 15 the to Brazilian of 10 10 Gradient -0.030 (0.0005) -0.152 (0.0000) -0.156 (0.0002) -0.089 (0.0002) -0.120 (0.0001) -0.025 (0.0000) -0.066 (0.0000) -0.134 (0.0000) -0.087 (0.0000) -0.037 (0.0033) for charge in R² MS 0.278 0.258 0.066 0.210 0.010 0.027 0.283 0.315 0.027 0.003 Adjusted person 10 Gradients the to 5 of Gradient -0.030 (0.0000) -0.089 (0.0000) -0.066 (0.0072) -0.063 (0.0000) -0.023 (0.1817) -0.006 (0.0271) -0.017 (0.0000) -0.043 (0.0000) -0.020 (0.0797) -0.003 (0.6793) Index bands R² MS 0.012 0.004 0.001 0.505 0.153 0.045 0.161 0.380 0.198 0.235 Adjusted earnings 5 to 3 Concentration Medium Gradient -0.005 (0.1824) -0.012 (0,2223) 0.005 (0.7588) -0.018 (0.0176) 0.055 (0.0002) 0.007 (0.0062) 0.021 (0.0000) 0.056 (0.0000) 0.036 (0.0000) 0.046 (0.0001) and R² Density Adjusted 0.147 0.230 0.133 0.038 0.236 0.128 0.473 0.514 0.331 0.353 MS 3 < 0.019 Gradient (0.0006) 0.061 (0.0000) 0.097 (0.0002) 0.029 (0.0343) 0.097 (0.0000) 0.019 (0.0000) 0.047 (0.0000) 0.104 (0.0000) 0.080 (0.0000) 0.074 (0.0000) Households ___________________________________________________________________________________ 9.2 = R² R² Cities Density Adjusted 0.493 0.200 0.045 0.218 0.079 0.222 0.285 0.392 0.466 0.002 Table Gradient Brazilian Households Density IBGE. Gradient -0.098 (0,0000) -0.114 (0,0000) -0.089 (0,0230) -0.107 (0,0000) -0.096 (0,0066) -0.048 (0,0000) -0.061 (0,0000) -0.151 (0,0000) -0.187 (0,0000) -0.007 (0,7434) for 2000 Census Strategya t]) t]) t]) t]) t]) t]) t]) t]) t]) t]) > > > > > > > > > > to [|T| [|T| [|T| [|T| Horizonte [|T| Janeiro [|T| Paulo [|T| [|T| Alegre [|T| [|T| y de Source: Cit Belém (Prob Fortaleza (Prob Recife (Prob Salvador (Prob Belo (Prob Rio (Prob São (Prob Curitiba (Prob Porto (Prob Brasília (Prob Data Inputs 25 236e 12 18 20 Pag 15 ertnecyt retnec 15 12 ci eht tyic e ot th 9 ecna to 01 stiD ecnats 6 SALVADOR CURITIBA Di 5 3 0 0 0 0 0 0 14 12 yti10sneD sdlohesuoH 80 60 40 20 40 35 30 y tisne25 D sdl20 ohesuoH 15 10 5 0 15 45 04 2000 in 12 35 retn 03 9 cities ceyt retnec ci 25 ytic eht e th ilian ot 20 to 6 cena PAULO RECIFE stiD 15 ecnats Di Braz 3 SÃO 01 10 5 for 0 0 80 70 60 50 40 30 20 10 0 0 0 ytisneD sdlohesuoH 10 90 80 70 60 50 40 30 20 10 ytisneD sdlohesuoH 4 Gradients 55 02 40 05 54 35 Density 15 04 retn 30 ret ert 53 ce cen encyt 25 ytic ci 30 ytci eht eht 10 eht ot ot 25 ot 20 JANEIRO ce ce Households antsiD 20 DE 15 cenatsiD an 51 stiD 5 BRASÍLIA 9.2 FORTALEZA 01 10 RIO 5 5 0 0 Figure 0 0 0 0 0 140 120 100 80 60 40 20 0 14 12 10 80 60 40 20 ytisneD sdlohesuoH ytisneD sdlohesuoH 55 50 45 40 35 30 25 20 15 10 5 0 ytisneD sdlohesuoH ___________________________________________________________________________________ 0 15 Cities 52 05 12 r ert 043 encytic tenec 20 ert Brazilian 9 tyic enc for eht 51 ytic the ot eht to ALEGRE ot 0 cena 6 ecn ec BELÉM stiD HORIZONTE 10 tasiD antsiD Strategya 021 3 5 BELO PORTO to 0 0 0 0 0 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 10 90 80 70 60 50 40 30 20 10 ytisneD sdlohesuoH ytisneD sdlohesuoH ytisneD sdlohesuoH Inputs Inputs to a Strategy for Brazilian Cities Page 237 9.105 Note that, with the exception of Recife, cities that have low GDP per capita like Belém, Fortaleza and Salvador presented the lowest rates of decentralization of poor households comparatively with cities of the same size. On the other hand, as the mean wage of household chief increases the households became more and more centralized. In the band of 20 minimum salaries and above, the pace of concentration towards central areas is very high in all cities, exceeding the decentralization of poorest households. 9.106 Some results for high income household centralization are among the more consistent of the calculations and in cities like São Paulo and Curitiba the adjusted-R² reaches 67% in some cases (band salaries above 10 minimum salaries). In cities like Belo Horizonte, Brasilia and Porto Alegre the adjusted-R² was about 25% or above in the highest wage band. However, in cities like Rio de Janeiro, Salvador, Fortaleza and Recife the adjustment is weak and the distance has not much force to explain rich households' concentration at the central areas. This can be caused by the constraints imposed by the site or the attractiveness of coastal locations for the richest families. However, the results presented are robust, mainly in the extreme bands, that is, in the poorest and richest bands. 9.107 Despite the attractiveness that central areas have normally and the role of public investments in provide better services at the central areas the role of commuting costs should not be neglected in impeding the suburbanization of the medium and high income groups in Brazil. Certainly, fixed and variable material costs in having cars surpass the economy in nonmaterial commuting costs relative to time in Brazil. The marginal costs from moving to suburbs are very high comparatively with marginal gains and accordingly so, these groups do not suburbanize. 9.108 The use of cars did not encourage the suburbanization of medium and high income families in Brazilian cities, even when cars became affordable for everyone. The late improvement of infrastructure and urban services in places marked by poverty and the operation of more efficient transportation modes were not able to change trends historically stamped in city. While high income groups were impelled to centralize by formal markets, which preferred the more profitable areas, the lack of affordable housing at the central areas for low income population made them to solve their housing demands at the periphery through irregular developments and self- construction. 9.109 Informality is still crescent in Brazil and the amount of people living in slums and illegal developments, produced mainly in land neglected by formal markets is increasing. Despite land in risky areas or allocated for ecological purposes are not priced by formal markets, since they are not legally available for development, they have a value considering their location within urban structure. The opportunity cost of not allocate them for urban development is lower than the benefits generated overall city in allocating them for other uses. Nevertheless, this opportunity cost can be high for people excluded from formal markets, since housing is a basic need. Hence, undeveloped land in serviced areas or near job opportunities, given to ecological uses, for example, is attractive for very poor people and become a target of informal occupancy. Urban Development Controls and Effects on Land Prices 9.110 Distance to the city center used as unique regressor to explain the variation in spatial distribution of households is not very efficient, even though it is significant. Hence, models that use other variables in their specification should be more efficient in explaining housing location patterns and land prices. Considering the objectives of this paper we are interested in describing the effects that land use regulations exert on land prices. Considering the lack of extensive surveys on land markets in Brazil and the lack of available data that combines land prices with other variables, we take advantage from the existence of a database used previously in a study conducted by Serra and Dowall (2003) about land prices in three Brazilian cities, to produce new calculations. 9.111 The database is compounded of residential land prices gathered through a systematic survey with real estate brokers covering several types of residential plots in several geographic zones of the metropolitan areas of Brasilia, Curitiba and Recife. They bring residential land prices by distance from the city center and information if plots are legally titled; have access to infrastructure, identified by the existence of paved roads, and if plot is under Inputs to a Strategy for Brazilian Cities Page 238 or over 500 square meters in size. The database brings land prices collected in 2001 and 2003 for Brasília and Recife's metropolitan areas and in 2000 and 2002 for the metropolitan area of Curitiba. 9.112 The choice of these three cities is justified because of their pace of fast urban growth and, considering their different contexts and conditions of growth, the different responses to poor access to land and housing. In addition, they are cities located in different geographic regions of Brazil and where planning policies differed so as to be able to evaluate how different regulations affect land markets (Serra et. al., 2003). 9.113 Differently from the former calculations the database is referred to the metropolitan area of the three cities. The database includes the areas of each city limited by the commuting distance, defined as the distance in which a family could look for housing in the next ten years. In Recife, the data area covered 2,742 km², including a total population of 3.2 million people in fourteen municipalities. The Brasilia database covered, besides the Federal District, five other municipalities125 with a population of 2.4 million people in 2000 and an area of 7,619.2 km². The Brasilia's database does not include land prices within 10-kilometer radius from the city center since special features of Plano Piloto cause prices to be very high, which could distort the results. The area covered by Curitiba's database was composed of thirteen municipalities within an area of about 2,082 km² and a population of 2.6 million people in 2000 (Serra et. al., 2003). 9.114 Brasilia was inaugurated in 1960 and, since then, counts with a very controlled land market commanded by the government. As already mentioned, the government has the ownership of majority of the land for urban development and the city grew according to the guidelines established by the several plans produced during the city existence. Along the commanded land allocation land use in the inner city is very restrictively regulated since it was listed as an urban site of great interest due to its spatial conception. 9.115 The public property of land allowed the implementation of several public housing programs directed to poor population originating several urban areas settled far from the central areas separated by large empty areas126. Even though these programs have improved poor access to land and housing for the poor, one of the main objectives of them was to avoid the undeveloped land of the central areas become targets of irregular occupancy. Thus, the restraints on land markets in association with planning policies that guided housing programs yield a spatial structure in which population densities are greater in peripheral areas than in central areas. 9.116 Curitiba, the capital of the state of Paraná is the core of an important metropolitan area of the South Region, having a large concentration of industries. Its first zoning legislation is dated from 1953, as a consequence of an urban plan developed ten years before (Plano Agache) that established several urban growth and land use guidelines. In 1966 was passed the Master Plan of the city that had among its aims, push urban growth out from the city center by limiting its growth and encouraging business occupancy along the main arterial roads. Curitiba is worldwide known by its public transportation system and land use planning associated with the mobility and accessibility allowed by transportation. 9.117 But, even counting with a robust urban planning legislation Curitiba was not prepared to deal with the fast urbanization caused by its economic growth, which generated the sprawling of peripheral areas by informal settlements. Since then, local authorities have implemented some regularization and urban improvement programs that aim to integrate irregular settlements to the legal city, regularizing, upgrading spaces and providing infrastructure. In addition, some programs are designed to assist residents relocated from hazardous areas both psychologically and technically (Serra et. al., 2003). 9.118 Recife, one of the main urban centers of the Northeast Region concentrates within its metropolitan area a large number of industries and firms. Nonetheless, the city has one of the highest rates of poverty in Brazil, with 125The municipalities considered are Águas Lindas, Santo Antônio do Descoberto, Novo Gama, Valparaíso and Cidade Ocidental, all in the state of Goiás. 126The largest urban concentration of the Federal District with 43% of the overall population is formed by the urban areas of Taguatinga, created in 1958, and Ceilandia, Samambaia and Recanto das Emas created from 1971 to 1993 and situated about 30 km from the Plano Piloto that has less than 10% of the population (Census data, IBGE 2000). Inputs to a Strategy for Brazilian Cities Page 239 many of its residents living in inadequate areas with the lack of infrastructure and under inappropriate housing conditions. For example, within the metropolitan area about 55% of households presented infrastructure inadequacies according to the Census of 2000. From this amount, about 34% were in the band of 3 minimum salaries or less of the earnings of the person in charge of household. 9.119 The history of urban growth in Recife is, in part, the history of occupancy of inadequate areas by the poor people. Since early days poor people built their shelters on the hills that surround the city, while central areas were occupied preferably by the richest people. The built of mocambos, a kind of shanty, was prohibited in central areas and during the 1940's they were removed from the inner city. The city has a significant set of urban renovation plans elaborated during the first half of 20th century by renowned city planners who traced the guidelines of urban growth that was followed by several administrations. 9.120 Recife has a solid tradition in designing instruments for popular housing and urban land tenure. The Social League Against Mocambos127 (Liga Social Contra os Mocambos), for example, was created in the end of the thirties to built popular shelter for poor people that lived in degraded housing units, mainly in the central areas. The city also initiated an innovative and pioneering experience in 1987 aimed to regularize informal settlements named Plan of Regularization of Special Zones of Social Interest (Plano de Regularização de Zonas Especiais de Interesse Social, PREZEIS). This plan is responsible in giving low income occupants of irregular areas security of tenure, the right to receive infrastructure and urban services access and allows them to participate in decision-making at the neighborhood and city levels (Serra et. al., 2003). 9.121 Despite the different contexts of urban growing of these cities some effects produced are similar, like the pattern of spatial distribution of high and low income households. The urban structure produced under their particular contexts reflects, among other factors, planning priorities adopted and the effects produced on urban land markets in time. 9.122 The database available do not allow a strict evaluation of the effects of intervention on land markets since they have not a detailed register of land use controls adopted within each urban zone. In addition, the database also does not cover different moments in order to follow the changes on land use controls in time. The two moments available for 2000-2001 and 2002-2003 are too close to permit such analysis. Nevertheless, alternative calculations could be made to extensively analyze how urban characteristics impressed in the spatial structure affect land prices in a static approach. Thus, in the lack of detailed database, the results reported here must be considered exploratory and indicative for more studies. 9.123 For assessing these effects were made linear regression calculations (OLS) by taking the log-land price of residential land prices as dependent variable and distance from the city center, existence of legal title (dummy), existence of infrastructure (dummy for paved roads) and plot size (1 if it is over 500 square meters in size) as independent variables. The regressions also include a dummy variable that estimates the combined effect on residential land prices of legally titled plot and infrastructure. In addition, an independent dummy variable was included to represent, by hypothesis, the urban features that are prevailing within a reliable distance-radius from the city center which affect positively land prices in overall urban area. 9.124 This variable is a proxy to estimate the effects on land prices from unobserved variables present within central areas that are associated with land use controls along city development. This notion is grounded in the fact that historically central areas has been more regulated than others, which has produced specific conditions that affect land prices128. Controls over urban development are extensively adopted to prevent central areas from excessive densification that result in negative externalities like infrastructure overcharging, for example. Thus, it is assumed that a wide range of special features present within central areas are being preserved or created by 127Mocambo is a kind of shanty that precedes the favelas (slums). 128In this model regulation is considered as exogenous and spatial features that increase land prices are determined ex-post land use controls adoption. On the other hand, in some cases land use controls can be considered endogenous being adopted to keep urban qualities determined ex-ante. Regulations within Brasilia's central areas can be considered endogenous, while in Curitiba and Recife can be considered exogenous. Inputs to a Strategy for Brazilian Cities Page 240 regulation, and hence, they can be correlated. In maintaining these characteristics by using zoning ordinances and land use controls, the attractiveness of these areas are kept high increasing land prices that, eventually, exclude low income segments from these areas. 9.125 The cut measure of distance to define the proxy for regulation was set based on the study conducted by Serra and Dowall (2003) that reports the dynamics of real estate development in the three cities. They quote that urban development and changes in population concentration in Recife and Curitiba occurred beyond 10 kilometers from the city center between 1991 and 2000. In 2000 44% of the total metropolitan population in Recife and 58.5% of total population in Curitiba lived within this radius. In Brasília less than 50% of the metropolitan population was located within 10 to 25 kilometer from central areas. In addition, population changes in Brasilia occurred outside a 20 kilometers radius from the central areas. 9.126 Hence, we established the cut distance radius for the dummy variable in 10 kilometer for Recife and Curitiba, and 20 kilometer for Brasilia, considering that the sprawling pattern of population change in these cities can be caused, among other factors, by planning restrictions on development in areas adjacent to the city center or redevelopment within them, forcing residential growth to the peripheral areas. 9.127 Error! Reference source not found. shows the results of the log-land price linear regressions using the database of the three cities for residential land prices. In order to do not discharge available information the regression's calculations considered all variables except distance, the quantitative variable which was changed by the central dummy variable. The outcomes for the regressions show that overall adjustments are good in predicting land prices, with except for Brasilia that presented low adjusted-R², mainly in 2001. Note that in 2001 only the central dummy variable was significant for residential land prices in Brasilia. Because of that, the adjustment of the regression was too low, since the adjusted-R² was only about 2.4%, and the outcomes for 2001 are not much consistent. A possible explanation is that land prices in Brasilia are too high overall city, not only within the central areas. Inputs to a Strategy for Brazilian Cities Page 241 Table 9.3 Linear Regression Results for Log-Land Prices for Residential Plots BRASÍLIA CURITIBA RECIFE 2001 2003 2000 2003 2001 2003 Sample size 176 181 1926 1.926 2.504 2.504 Adjusted R² 0.0244 0.359 0.574 0.572 0.450 0.461 Constant 4.1312 2.9360 2.6472 2.5686 3.0527 2.8952 t-stat. 15.529 6.346 54.443 52.572 85.958 83.107 (p-value) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) Regulation proxy 0.4836 1.4029 1.6114 1.6162 1.0216 1.0270 t-stat. 2.368 8.430 42.757 42.905 35.083 35.986 (p-value) (0.0190) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) Infrastructure 0.3468 1.5032 0.8953 0.8830 0.6134 0.6131 t-stat. 1.099 3.017 16.697 16.380 14.961 15.240 (p-value) (0.2733) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) Title 0.3286 0.3937 0.0719 0.0720 0.1910 0.1905 t-stat. 0.953 0.817 1.321 1.322 4.467 4.512 (p-value) (0.3420) (0.4149) (0.1864) (0.1862) (0.0000) (0.0000) Infrastructure and title -0.5899 -0.5199 -0.0364 -0.0346 0.0125 0.0065 t-stat. -1.470 -0.993 -0.480 -0.455 0.216 0.114 (p-value) (0.1434) (0.3223) (0.6313) (0.6489) (0.8291) (0.0000) Plot size 0.1867 -0.9779 -0.4431 -0.4234 -0.0562 -0.0548 t-stat. 0.887 -4.152 -10.414 -9.923 -1.948 -1.931 (p-value) (0.3762) (0.0001) (0.0000) (0.0000) (0.0515) (0.0535) 9.128 The values of the constant in the model predicts land price of a plot without legal titling, without infrastructure, less than 500 sq mt and located outside the radius defined for the regulation proxy variable. In Brasilia a plot with these characteristics and outside the 20-kilometer radius was priced about R$ 18.84 per sq mt in 2003. In Curitiba, a plot with the same features was R$ 13.04 per sq mt while in Recife it was about R$ 18.09 both outside the 10-kilometer perimeter. Note that the predicted price per sq mt in Brasilia is higher than in Curitiba and Recife at a distance that is two times larger than in these cities. 9.129 Title and infrastructure associated was not significant in regressions. In addition, in Brasilia (2003) and Curitiba (2000 and 2002) the presence of legal title by itself also was no significant at a level of 95%. This fact can reflect the power of other variables in increasing land prices, overwhelming the effects of titling since tenure land is not the greatest problem in the context of these metropolitan areas. On the other hand, in Recife, due to the context of poverty and irregularity, the variable for plots legally titled were significant in 2001 and 2003, increasing land prices about 19% overall city. These results show that variation on land prices in formal and informal markets are not very affected by legal tenure of land. 9.130 Data from the Deficit Habitacional no Brasil (2005) show that in Brasilia and in the 5 municipalities considered in this study, the total of inadequacy in housing tenure in 2000 was about 2.1%, while in Curitiba metropolitan area was 7.2%. In Recife's metropolitan area in 2000, households with titling inadequacies were about 10.6%. In addition, while in Recife and Curitiba title inadequacies affected more intensively households with 3 minimum salaries of maximum familiar median income, presenting respectively 66% and 45% of the households without titles, in Brasilia this problem affected especially the households with familiar median income above 5 minimum salaries. 9.131 All other variables coefficients had the expected signal and show significant influences over land prices. Infrastructure increases significantly residential land prices. In Brasilia, albeit the largely covering in water and sanitation facilities, provision of infrastructure adds almost 350% in land prices, even in plots outside the 20- kilometer radius around the city center. Plots located outside the Federal District, where infrastructure and title Inputs to a Strategy for Brazilian Cities Page 242 problems are more intense are priced at very low values. In Curitiba provision of infrastructure increases land prices about 143% while in Recife adds 84% in residential land prices. 9.132 In fact, the lack of infrastructure provision is the major inadequacy that affects urban households. Data from the Deficit Habitacional no Brasil (2005) show that about 21% of urban households in Brasilia and those 5 municipalities, presented at least one problem related with infrastructure provision, while in the Curitiba's metropolitan area this problem affected about 15.2% of overall households. In Recife's metropolitan area about 55% of the households presented infrastructure problems in 2000. In all these cities the lack of infrastructure provision in absolute terms was more intense within their respective municipalities. Nevertheless, in proportional terms infrastructure inadequacies affected more than 60% of the households within other metropolitan municipalities. In addition, infrastructure inadequacies affected especially the households with median familiar income up to 3 minimum salaries. 9.133 The results for plot size also had the expected signal and were significant at a 95% level. That is, large size plots were priced in 2000-2001 and 2002-2003 below small plots per square meter. In Brasilia, large plots were priced about 37% below small plots, while in Curitiba and Recife's metropolitan areas they were, respectively, 65% and 95%cheaper than plots with less than 500 sq mts. One possible reason for that is the low price of land at the fringes of urban areas inducing large land consumption per plot, or the existence of land with rural use. 9.134 The proxy dummy variable for regulation was significant in all cities and in both moments 2000-2001 and 2002-2003 added high values in land prices by the simple fact if plot is located within the radius specified for the variable definition. In Brasilia in 2003, for instance, a plot without title and infrastructure and under 500 sq mt in size located within the 20-km radius from the city center was priced about 300% above an equivalent plot placed outside that perimeter. In Curitiba, the amount added in land prices located within the 10-kilometer radius was about 400%, while in Recife it was about 179% in both in 2000-2001 and 2002-2003. Once more, value added in Brasilia's land price is too high comparatively to Curitiba and Recife, since the effect is calculated for a radius distance of two times larger than in Brasilia. 9.135 These results are robust to demonstrate that unobserved urban features present in central areas that can be correlated to planning controls affect land prices positively. In order to assess how these variables affect individually residential land prices across urban areas were calculated the marginal effects of each one of them on the likelihood of residential land prices is over a median land price value through Probit regressions. In the model specification was included the quantitative variable for the distance from the city center to calculate how marginal effects act over land prices as distance changes. The median land prices were calculated from the database used in the former regressions. In Brasilia (2003) overall mean plot prices was 143.57 reais per square meter, while in Curitiba (2002) it was 71.13 reais and in Recife (2003), 70.69 reais. 9.136 Table 9.4 shows the results from the Probit Model regressions. The model is highly predictive for the three cities, about 79% for Recife and over 80% for Brasilia and Curitiba. Title was no significant for Brasília and Curitiba, as well as the dummy for infrastructure and title combined in the three cities. As distance from the city center increase, the probability of plots are priced over the average price is decreased about 18% in Brasilia and 16% in Curitiba, which demonstrate how concentrated within central areas are land priced above mean in these cities. On the other hand, in Recife, an increase in distance by 1 kilometer from central areas increase the chance of land is priced above average only by 2%. 9.137 Residential land prices in Recife can be more balanced overall urban area and land priced over mean is not so concentrated as in Brasilia or Curitiba. While there are several factors that can affect land prices across urban area, per capita income may be one variable to explain lower land prices in Recife. In addition, presence of the Special Zones of Social Interest (ZEIS) can perform an important role in regulating land prices. At present days, there are 66 areas recognized as ZEIS in Recife, in all of the six political-administrative regions (RPA) of the city, corresponding of about 6% of municipality's surface. Inputs to a Strategy for Brazilian Cities Page 243 Table 9.4 Probit regression results for land prices likelihood above mean BRASÍLIA 2003 CURITIBA 2002 RECIFE 2003 Mean land Price R$143.57 R$71.13 R$ 70.69 Sample size 181 1.926 2.504 Adjusted measure 81.2% 84.3% 78.8% Predicted 0 69 788 992 Predicted 1 78 836 981 Constant 3.2079 1.3148 -0.8887 t-stat. 2.560 6.939 -7.732 (p-value) (0.0105) (0.0000) (0.0000) Distance -0.1795 -0.1591 -0.0187 t-stat. -5.876 -13.707 -4.065 (p-value) (0.0000) (0.0000) (0.0000) Regulation proxy -1.7187 0.5120 1.3235 t-stat. -3.064 4.279 16.103 (p-value) (0.0022) (0.0000) (0.0000) Infrastructure 3.0166 1.1392 0.8316 t-stat. 2.695 9.645 9.908 (p-value) (0.0070) (0.0000) (0.0000) Title 0.7845 0.0972 0.2653 t-stat. 0.717 1.017 3.095 (p-value) (0.4736) (0.3090) (0.0020) Infrastructure and title -1.0947 -0.0447 0.0429 t-stat. -0.946 -0.275 0.363 (p-value) (0.3444) (0.7835) (0.7167) Plot size -1.5196 -0.5291 -0.1216 t-stat. -4.233 -6.347 -2.057 (p-value) (0.0000) (0.0000) (0.0397) 9.138 The estimated coefficients from Probit Model do not show directly the variation on likelihood of land is priced above the average value due to a change in the independent variables. This effect is better estimated by calculating the marginal effects of each variable individually over this probability as distance changes. Figure 9.3 shows the plots of marginal effects of the independent variables for the three cities with distance from the city center. As previously mentioned, infrastructure and title combined is not significant at 95% in all cities as well as title is not significant in Brasilia and Curitiba. Note that the likelihood of land prices is higher than the mean decreases with distance in all cities. Inputs to a Strategy for Brazilian Cities Page 244 Figure 9.3 Marginal effects over high land prices likelihood in 2003. BRASÍLIA 2003 PROB LAND PRICE > AVERAGE Marginal effects over probability land price higher than average 1.0 0.9 0.8 0.7 0.6 bo 0.5 Pr 0.4 0.3 0.2 0.1 0.0 0 3 5 8 10 13 15 18 20 23 25 28 30 33 35 38 40 43 45 distance dd Pavement Title Pavement +Title Plot Size CURITIBA 2002 PROB LAND PRICE > AVERAGE Marginal effects over probability land price higher than average 1.0 0.9 0.8 0.7 0.6 bo 0.5 Pr 0.4 0.3 0.2 0.1 0.0 0 3 5 8 10 13 15 18 20 23 25 28 30 33 35 38 40 distance dd Pavement Title Pavement +Title Plot Size RECIFE 2003 PROB LAND PRICE > AVERAGE Marginal effects over probability land price higher than average 1.0 0.9 0.8 0.7 0.6 bo 0.5 Pr 0.4 0.3 0.2 0.1 0.0 0 3 5 8 10 13 15 18 20 23 25 28 30 33 35 38 40 43 45 48 50 53 distance dd Pavement Title Pavement +Title Plot Size Inputs to a Strategy for Brazilian Cities Page 245 9.139 Infrastructure and regulation had the largest marginal effects in increasing probability of land prices are above the mean as distance decreases from the city center in Curitiba and Recife. In Brasilia, infrastructure had the greatest marginal effect, overwhelming land located at the central areas and plot size influences. Influence from infrastructure is great, even at large distances, until 45 kilometers from the city center, which is the limit between the Federal District and other municipalities, where the lack of infrastructure is troublesome. On the other hand, the role of the central unobserved features in affecting likelihood of land is priced above mean seems highly concentrated within the 20-kilometers radius in Brasilia. 9.140 Inside Federal District, virtually all urban areas are strongly regulated and, consequently differences in prices due to different locations are not so distinguished due to regulation. Land prices overall city is high, reflecting other factors as, for example, the high income per capita or the fact that the city is the national capital. Nonetheless, prices of serviced land and land without infrastructure present great difference inside the Federal District. Serviced land is priced almost 3 times above land without infrastructure, even though median land price of plots without infrastructure in Brasilia is higher than land without infrastructure in Curitiba and Recife comparatively. As such, serviced plots effects surpass the influence of other variables in Brasilia. 9.141 Plot size had positive impact, but not as strong as the other variables. The influence of large plots in increasing the likelihood on prices are above mean is crescent as the city center is closer, from about 20-kilometer in Brasilia and Curitiba. In Recife, the increase on this probability is very low, less than 20% at the city center. Probably the fact that large plots are scarce near to the central areas makes them to be more valorized than small plots in those areas. 9.142 Recife presents interesting results. Even though the influence of the central dummy variable and infrastructure are positive and crescent towards central areas, the maximum likelihood of residential land is priced above mean is less than 70% at the city center. In Brasilia and Curitiba this likelihood is very close to 100%, highly affected by those variables. In Recife, central plots had the strongest marginal effect over this chance, while in the other cities infrastructure affected more intensively land prices as plots are closer to the city center. 9.143 Note that, while in Curitiba the increase on land prices are above the mean is crescent from 30 kilometers towards inner city, reaching 100% at the city center, in Recife this chance is maintained beyond 50 kilometers, even though at a very low level. Further, prices of plots located at the central areas can affect land be priced above average about 30% at 50-kilometers from the central areas, while in Curitiba this possibility is null at the same distance. 9.144 This result is striking and not easily explained depending on more studies. Probably, a combination of overall low income of its population, urban structure conditions and public policies of ZEIS can affect land market operation. Firstly, the conditions in affording housing prices can control land prices, since the latter reflects housing demand factors. Secondly, in spite of the largest number of households in Recife's metropolitan area, the constraints imposed by geographical conditions produced a more sprawled urban structure comparatively to Curitiba, where distribution of poor and rich households is more balanced (see Table 9.1and Table 9.2). 9.145 Finally, public programs on regularization of informal settlements (PREZEIS) can improve access to land for poor people nearby central areas reducing distortions on land prices due to exclusiveness produced in these areas by planning controls. These programs, pursuing to guarantee tenure rights, can affect land markets by increasing the supply of land that can be traded in formal markets. Can Land Controls Loosening Improve Formal Urbanized Land Production? 9.146 In the previous section was found that residential land prices overall urban areas are affected by infrastructure provision, plot size, title (only in Recife) and land located at the central areas where planning controls are supposed to be correlated with unobserved features that increase land prices. Land use and design parameters enforced by urban legislation along time, especially in the central areas of the cities analyzed, may have limited responsiveness of housing supply to demand increasing housing and land prices. Consequently, due Inputs to a Strategy for Brazilian Cities Page 246 to the irreproducibility of the advantages of some urban locations, especially in a context in which public investments neglect peripheral areas, inner cities are preferably occupied by high income segments of the population, while poor people solve their housing demands in informal settlements or in precarious areas at the peripheries. 9.147 Urban legislation in Brazil establishes several standards and requirements for producing urban developments that are aimed to achieve an ideal pattern of urban space. On the other hand, the same legislation allows relax the parameters in order to facilitate the production of affordable housing and serviced land for low income population. Further, titling and urbanization programs of informal settlements also relax urban standards as a necessary step for regularization. 9.148 Deregulation of planning ordinances is suitable to remedy critical situations crystallized in space and, barely have been heard that relaxing these ordinances have encouraged formal market to produce serviced land and affordable housing for low income population. Housing demands of these groups are left to public sector to solve that acts only when critical situations are installed. However, costs for urbanizing and regularizing informal settlements are pointed as being higher comparatively with costs of providing urbanized land as a preventive policy. While the costs per household in urbanizing programs is from U$ 50 to U$ 70 per sq mt, the costs of urbanized land by private developers is about U$ 25 per sq mt, including the profit (Smolka, 2003). 9.149 There are a large range of factors that hinder formal production of urbanized land and affordable housing for low income families ahead of the obvious drift of developers for the more profitable market segments and budget limitations of poor people. One of these factors is the role that planning controls play in increasing production costs in formal markets. 9.150 From the findings of previous sections and using the database available for residential land prices for Brasilia, Curitiba and Recife, were made some simulations to calculate economic feasibleness of a project that, hypothetically, is aimed to produce 100 urban plots according to the parameters established by the urban legislation currently in force in these three cities. 9.151 Cities in Brazil often adopt minimum standards based on the Federal Law number 6,766 that rules land subdivision and registration proceedings for urban developments. The minimum plot size generally adopted is 125 sq mt with minimum plot frontage of 5 m. Another standard adopted is a compulsory donation of land for public use like green spaces, squares, roads and public facilities. This amount was defined in 35% at least by the Federal Law 6,766, but it was changed and left to municipal authorities to decide this amount according to specific parameters. However, the majority of Brazilian cities still assume 35% as minimum. 9.152 Many cities also set several parameters in roads dimensioning. In fact the dimension of roads depends on several factors as hierarchical category, population density, immediate land use, design patterns and others that imply different consumption of space. Normally, local roads, which access directly residential plots are sized at 7 mt in width for vehicles, and sidewalks between 1.5-2 mt. In areas of social interest these standards are relaxed and are admitted 5 mt for roads width and 1 mt for sidewalks, in general. The surface used for roads is included into the public area requirements. 9.153 Given these parameters it is possible trivially calculate the amount of land required for develop 100 plots. Table 9.5 show the results from data described above considered in four contexts: enforcing standards and relaxing standards to build a single family unit per plot (columns 1 and 2) and to build 2 housing units per plot (columns 3 and 4), in other words, increasing density twice. Restriction on density increasing is the effect of several controls adopted, like floor area ratio (FAR) limits, maximum construction coefficients, minimum rate of plot surface per housing unit or, explicit density limits. The data rows of the table show the parameters considered in both enforced and loosened situations and outcomes are shown in the results rows of the table. 9.154 Column 1 shows that under basic standards, there is an increase of 53.8% in land consumption per plot produced or a single family unit house, with relation the useful land required per plot. Relaxing these parameters the increase of land consumption is reduced to 43%. In the first situation is produced 100 plots consuming near one hectare of land, while under loosened standards are used only 0.43 ha. Inputs to a Strategy for Brazilian Cities Page 247 Table 9.5 Land Consumption in Urban Development under Different Standards 1 2 3 4 Specifications Basic Relaxed Basic Relaxed standards standards standards standards Plot units developed 100 100 100 100 Minimum plot size (sq mt) 125.0 60.0 125.0 60.0 Minimum lot frontage (mt) 5.0 5.0 5.0 5.0 Road width (mt) 7.0 5.0 7.0 6.0 data Sidewalk width (mt) 2.0 1.0 2.0 1.5 Total spaces for public use (%) 35.00% 30.00% 35.00% 30.00% Basic Total housing units per plot 1 1 2 2 Total housing produced 100 100 200 200 Land area required for plots (sq mt) 12,500 6,000 12,500 6,000 Land area for roads and sidewalks (sq mt) 2,295 1,530 2,385 1,988 Total land area required for development (sq mt) 19,230.8 8,571.4 19,230.8 8,571.4 Land area required for public use (sq mt) 6,730.8 2,571.4 6,730.8 2,571.4 Ratio of land for public use 35.00% 30.00% 35.00% 30.00% Ratio of land for roads 11.9% 17.9% 12.4% 23.2% Total consumption of land per housing Results unit (sq mt) 192.3 85.7 96.2 42.9 Increase in land consumption per housing produced due to land use standards 53.8% 42.9% -23.1% -28.6% Source: Author's calculation 9.155 Considering that a housing unit under popular standards is about 60 sq mt, when plots are produced with an area larger than that, the amount of land added is appropriated by few plots. The extra private land added is useful to get better life conditions by improving ventilation or insolation, for example. However, alternative designs could utilize less land for housing unit without loss in quality of life, but improving land allocation. For instance, in order to produce 100 plots of 85 sq mt under relaxed standards the amount of land added per plot is also about 43% over useful plot area, and the development will have only 0.6 ha. 9.156 When density is doubled the amount of land consumed per housing unit under enforced parameters is reduced in 23.1%, indicating a more efficient land allocation in housing production. Relaxing the standards in column 4 land area consumption per house unit produced decreases about 28.6% with relation the housing area. Overall reduction of land consumption by building 2-households per plot under the relaxed standards is about 78%with relation the original standards. 9.157 This trivial exercise shows that the jointly effects of land use controls over land consumption should be not neglected, even though the individual effects can be apparently unimportant in face of the benefits expected from design standards adoption. These benefits, like the improvement of quality of life should be considered in a comprehensive approach of the city, since the costs of that are shared for everyone. 9.158 Based in these findings, were formulated a model to evaluate the economic feasibleness of producing urbanized land and affordable housing under different design parameters situations. Since land regulation affects markets differently depending on market conditions, were analyzed economic flow of costs and revenues to produce urbanized plots and housing in Brasilia, Curitiba and Recife in four different situations. The situations considered enforced and loosened parameters in producing and selling only urbanized land and in producing and selling urbanized land and housing at popular pattern. Inputs to a Strategy for Brazilian Cities Page 248 9.159 Calculations considered a production of 100 plots by changing only design parameters, as showed in columns 1 and 4 of Table 9.5, maintaining all other possible variables constant (ceteris paribus). In addition, in loosened contexts the time to approve projects and to construct infrastructure were also changed. Projects are approved during the first year and the infrastructure can be provided progressively in a period of 8 years. 9.160 In the economic flux of formal process of urban development the initial investment was considered as the amount of revenue that land owner would receive if land were used to produce informal plots129. In other words, is considered the opportunity cost of land owner in using land to produce irregularity or to produce urbanized and registered plots. In latter, developer would acquire land from owner paying at least the expected amount that land would rent if plots were traded in informal market. In addition, informal developer does not allocate enough land for public use, neither for roads nor for other uses like green spaces or squares. As such, the amount of plots produced by informality is larger than in formal market, even under relaxed parameters. 9.161 Thus, while formal developer produces 100 plots under enforced parameters, informal developer produces 140 plots of same size, that is, an increase of 40% above formal production. When parameters are relaxed, increase in informal production is only of 25%, that is, informal developer gets only 25 more plots than formal developer, since land enforced for public use is reduced. 9.162 Prices used in calculations are from the residential land prices database of Brasilia, Curitiba and Recife in 2003, at locations further than 10 kilometer in Curitiba and Recife and ahead of 20 kilometer in Brasilia, as discussed in the previous section. At this distance was taken the mean price per square meter of land with and without infrastructure and title. Here the interest is in knowing economic feasibleness in producing regular land at the distances where those cities are experiencing transformation. 9.163 Prices per sq mt of plots without title and infrastructure, considered as informal, are lower than serviced and titled plots at the same location. In Brasília, the mean price per sq mt of plots without title and infrastructure is equivalent to 65% of serviced and titled plots, while in Curitiba and Recife these prices are, respectively, 59% and 61% of land prices of plots with title and infrastructure. These prices are considered too high, since there is no investment made by the informal developer that would justify the profit achieved. 9.164 Production process is considered within a discrete time period that involves the time required to acquire land, the initial time; the time necessary to approve projects, to develop land and build houses and to sell the units produced. The time zero starts in 2003 when developer decides to produce formal housing and acquire land from land owner. The final time is 2013, when developer has sold all plots and housing. 9.165 The time necessary to approve the projects was defined in two years. However, this process can take much more time. In Rio de Janeiro this time can reach 4 years and in Brasilia it can take more than 5 years. Developer must accomplish several steps to approve projects. First, urban design projects have to be approved by local authorities. Secondly, infrastructure projects have to be approved in each one of the several companies that are in charge of electricity, sanitation and drainage facilities provision. 9.166 In addition, projects have to be licensed by environmental authorities to be authorized. About three stages are necessary to achieve environmental approval. Firstly is required a previous license that is followed by an installation license, when is required the elaboration of detailed environmental studies and reports on environmental impacts resulted from the development. Finally is conceded an operation license, when the development is allowed to be implemented. In short, the complete process for environmental approving is costly and time consuming. 9.167 Even though all these proceedings are argued as necessary, time-consuming bureaucracy and obstructive official routine increase costs and risks that jeopardize feasibleness of formal production. In addition to the ordinary opportunity costs involved in time-consuming bureaucracy that can be estimated at the market real interest rates over initial investments; changes in construction costs previously calculated, changes in 129Informality is used here to designate plots produced without infrastructure and title. Inputs to a Strategy for Brazilian Cities Page 249 macroeconomic context and appearance of new enforcements along time can increase risks and uncertainty of business. 9.168 In deciding to initiate a development business sales velocity is a factor that has great impact in decision taking. The impact of sales velocity can be calculated only in some cases, since there is no general pattern to estimate needs of capital in each case. However, considering the market segment that would be attended by small and medium builders and developers, the velocity of sales is critical because business feasibleness depends on the cash flow. Great profit margins are in the past and building firms, especially small and medium ones need to keep a strict control on cash-flow. In addition, projects that are not considered highly profitable result in very high loan costs that are impracticable for small and medium builders because of the interest rates. 9.169 The time assumed to build the development is that defined by legislation in a maximum period of 4 years for developer construct roads and infrastructure in order to register the development according to the Federal Law 9,785/1999 that modifies the article 18 of the Federal Law 6,766/1979). Subsequently, from the second period, developer sells the plots during a period of 8 years, considering the time need to receive the value from plots sold to families whose median income is about 3 minimum salaries of 2003 and can compromise only about 25% or 30% of their familiar budget with monthly payments. 9.170 For simplicity, the taxes that occur along the formal process of development construction were considered into a ratio of 25% (BDI) added over construction costs to cover overall builder profit, administration costs and taxes. Construction costs of infrastructure and housing at popular standards per square meter (60 sq mt) were calculated from PINI's construction national survey made monthly in several cities in Brazil and released in Construção Mercado (PINI Publishers). Prices were taken in Reais in August, 2003. 9.171 The costs for infrastructure provision were also calculated from PINI's construction national survey taking in account water, sewage, pavement and drainage, and electricity provision. There are few specialized publications dealing with costs for infrastructure provision, although some studies had been conducted about this theme. These studies are dedicated mainly in assessing urbanization costs of slums, which involves additional costs from removing shanties from risky areas, recuperation of degraded areas and others. In addition, costs can vary from several factors such as topography and grid layout. The costs calculated and adopted in this study are very similar to those found in literature130. 9.172 Table 9.6 shows the results for the feasibleness of a hypothetical builder that would intend to invest in projects of constructing urbanized plots and housing for low income families in Brasilia, Curitiba and Recife under different regulation contexts. In order to evaluate economic feasibleness of the different alternatives is used two indicators, Net Present Value (NPV) and Interest Rate of Return (IRR). Basically, NPV shows the value of a cash flow in future discounted back to the present time by a percentage that is the minimum desired rate of return. As such, this percentage is the interest rate that equals the cost of capital in time. Projects can be discarded if NPV from a cash flow is below zero (NPV<0). If NPV is negative, it signifies that the rent yield by the investment is less than the opportunity cost of the capital. 9.173 IRR is based on the same principle and represents the discount rate that results in a null net present value for the cash flow of the investment. The IRR is the true interest yield expected from an investment expressed as a percentage. The investment will be attractive if it is positive, that is, if IRR is above the interest rate of the cost of capital. By comparing investments, the best will be that has the highest IRR. 9.174 In calculation for the Net Present Value (NPV) was adopted the mean interest rate from the difference between the basic interest rate (SELIC) and the inflation (IPCA) from 2003 to 2005. Thus, the discount rate was defined as 11.5% during the period and it is assumed that all costs involved will be constant in the investments flow, along the future moments. 9.175 The results show that there is no feasible business in producing formal urbanized and titled plots facing the profits from informal plots production in any of the three cities, considering the residential land prices with 130For a broad discussion about urbanization costs of slums see ABIKO et al., 2005. Inputs to a Strategy for Brazilian Cities Page 250 and without infrastructure. Indicators, internal rate of return (IRR) and net present value (NPV), are negative if considered only production and sales of urbanized plots. In addition, there is no pay back for the investments. Although the revenues earned from selling urbanized plots in all cities would be sufficient to cover urbanization and titling costs, they do not cover the opportunity cost involved in abandoning the production of plots without infrastructure and title. 9.176 Several reasons contribute for these results. First of all, and most important, market prices of land without infrastructure is too high, comparatively with urbanized land, becoming informal business very attractive. Once housing demand is high, even land without infrastructure and title reaches high prices in urban land market. Note that the revenues from selling plots without infrastructure and title would be enough to provide both services and a good profit for developer. In other words, the prices of land in informal market embody in advance, the expected rent from future regularization and provision of infrastructure by the government. This rent is appropriated totally by the informal developer without investments. 9.177 Second, the opportunity cost of capital in Brazil, rated at 11.5% per year (real interest rate) can be a potential factor to hinder investments. This rate increases the cost of capital for medium and small developers, for example, in financing construction costs, mainly in long term loans. Because of that, NPV is negative and investment is considered not attractive. 251e Pag Relaxed standards 100 60 30 200 8,571.40 49.15 30.09 121.91 1,738.57 27,280.51 -225,500.40 -198,240.58 -423,740.99 -5,456,100.00 -5,879,840.99 294,903.62 6,820,125.01 7,115,028.62 -238,614.28 -7.80% - -1,818,374.75 26.90% Recife Basic standards ,728,050.01 ,582,826.28 410,062.49 100 125 35 100 19,230.80 49.15 30.09 121.91 3,174.47 27,280.51 -525,137.19 -329,639.07 -854,776.26 -2 -3 614,382.55 3, 4,024,445.04 -463,453.89 -6.10% - -1,301,875.44 8.20% 494.16 Relaxed standards 100 60 30 200 8,571.40 41.87 24.71 165.38 1,686.66 32,488.49 -185,177.76 -201,740.97 -386,918.73 -6,497,700.01 -6,884,618.74 251,297.46 8,122,124.99 8,373,422.45 205,135.13- -9.60% - -2,089, 32.80% Curitiba Production Basic standards ,248,850.00 ,009,265.61 061,062.50 Housing 100 125 35 100 19,230.80 41.87 24.71 165.38 3,126.43 32,488.49 -431,235.29 -329,180.32 -760,415.61 -3 -4 523,536.37 4, 4,584,598.87 -407,091.24 -7.20% - -1,193,962.79 17.50% and Land de Relaxed standards 100 60 30 200 8,571.40 104.93 67.97 268.01 1,752.88 33,623.99 -509,130.98 -228,889.68 -738,020.66 -6,724,800.00 -7,462,820.66 629,664.80 8,406,000.00 9,035,664.80 -414,826.50 -2.40% - -2,405,112.39 19.90% Brasilia Urbaniz of Basic 929.92 standards 90% 100 125 35 100 19,230.80 104.93 67.97 268.01 3,170.70 33,623.99 -1,185,645.88 -343,870.64 -1,529,516.52 -3,362,400.00 -4,891,916.52 1,311,801.67 4,202,999.99 5,514,801.65 -847, -2.50% - -1,981,040.91 5. Feasibleness mt) 9.6 . qs % (R$) mt) 60 11.5 Table mt) (R$) (R$/sq (R$/unit ___________________________________________________________________________________ (R$/sq revenues) costs mt) title production) inflation):- (R$) (sq title and plots title rate Cities standard and ar and sales production 1 unit) informal (R$) tion interest (R$) velopment ructure astructure popul Brazilian a gn housing housing asic de infr at costs housing (B plots and for te for infrast housing costs sales and ra mt) ) cost and (2003-2013) %( or title urbaniza,y nd land la plots (sq with- without- (R$/plot) $] abandoni [R in nda title ST (IRR) (IRR) (NPV) rn (NPV) rn interest Strategya size area produced required titled from urbanized (R$/plot price price costs production cost RESUL retu retu (opportunit of value of to developed plot area d y units construction S and value urbanized DATA public sales lan land selling rate rate units of land costs investment urbanization costs housing urbanization, NUEE revenues FLOW Discount- by y present back selling present 1 Inputs BASIC Plot Minimum Ratio Housing Total Market Market Titling Urbanization Housing COSTS Initial (opportunit Total Total Total Total REV Urbanized Housing Total CASH Onl Net Internal Pay By Net Internal Obs: Inputs to a Strategy for Brazilian Cities Page 252 9.178 Total earnings from selling urbanized land in all cities are not sufficient to cover the costs of providing infrastructure and title, the costs of capital and to make developer gives up from informal land development. Thus, developer is encouraged to produce plots without infrastructure and title. 9.179 Loosening standards does not improve economic feasibleness in producing urbanized land in the three cities studied. Under relaxed standards design parameters are changed and time enforced to develop land and provide infrastructure is enlarged at the same time. The NPV increases under relaxed standards since the amount of land necessary to produce the same amount of plots is reduced, decreasing investments. Note that urbanization costs per plot are lower about 45% in relaxed standards situation comparatively to enforced parameters situation. This agrees with experience elsewhere that urbanization costs increase with the size of the area occupied, even the amount of served plots are the same, due to the extension of the infrastructure systems. 9.180 On the other hand, IRR decreases because a problem of scale and market context. The reduction in urbanization costs about 45% do not compensates the ratio between these costs and the revenues earned from selling urbanized plots under relaxed parameters, since the size of plots is reduced (from 125 sq mt to 60 sq mt) and total revenues from plots sales also is reduced. Under the enforced parameters, infrastructure costs is equivalent about 23.8% of the revenues from plots sales, while under relaxed parameters, these costs is about 26.3% of the revenues. Here a loss of scale due to plots size and the revenues earned from plots sales (priced per sq mt) is responsible for the internal rate of return decrease. 9.181 Even under relaxed standards plots at a very small size can be counterproductive. This suggests that there is an optimum relation between plot size, considering its value per square meter, and the costs of providing infrastructures for single family housing typologies, since these costs cannot be priced directly from the number of households serviced131. 9.182 The effect of enlarging time to construct and register a development can be positive in increasing economic feasibleness of urbanized developments. For example, in Recife's market context when enforced time to construct and register developments is changed from 4 to 5 years, maintaining all the other conditions unchanged, the NPV is increased about 8%. Relaxing this time can be an alternative to encourage urbanized and titled land production by formal market, since developers can invest in infrastructure progressively, adjusting investments to the payment capacity of low income dwellers. This allows developer to get rid of high cost of capital investing resources only when necessary and using own revenues from sales in infrastructure provision. 9.183 It is interesting to compare the findings reported above with the results from the cash flows of housing construction at popular patterns. In all cities NPV is negative at a high level, even though the internal rate of return (IRR) is positive. Under the regulated parameters NPV was smaller than under loosened standards due to the same reasons previously reported. The IRR is positive but at a low level indicating that housing production with urbanized land can be more profitable than produce only urbanized land in the context studied. 9.184 However, when parameters are relaxed in order to permit an increase in density, the internal rate of return indicates that housing construction can be highly profitable. Clearly there is a gain of scale and earnings from sales of popular housing compensate total costs at a rate about 33% in Curitiba, 27% in Recife and 20% in Brasilia. In addition, by relaxing parameters, the investments had reduced the pay back time in two years comparatively with that under enforced standards. However, at the discount rate used (11.5% per year) the opportunity cost of capital is too high to encourage developer to produce housing at popular standards, as indicated by the negative net present values in all situations. 9.185 Albeit the simplifications assumed here the model of cash-flow presented is useful to evaluate the expected effects in urbanized land and housing production under different contexts of land use controls. In fact, in some contexts, land use regulations can hinder production of affordable housing for low income population and deregulate urban markets can encourage developers to produce more housing. However, although relaxing land 131There are several factors that can affect urbanization costs, as development size, grid layout, topography, density, and others (ABIKO et. all, 2005). Inputs to a Strategy for Brazilian Cities Page 253 use regulation can be important to reduce urbanization costs, get better scale of housing production, and to enhance land allocation it cannot be taken as sufficient to guarantee, by itself, the enlargement of housing supply and improvement in the conditions of poor people access to urbanized land and housing. Experiences in other countries, has demonstrated that a total liberalization of urban markets can be counterproductive, worsening conditions of poor people access to affordable housing. Conclusions and General Implications 9.186 Discussion on the limits of urban planning controls and land use regulations on urban land markets is polemic and polarized. In one side there are the ones who defend free urban markets as a way to get informality reduced by improving housing supply and so decreasing housing prices. On the other side there are the ones who defend public intervention in order to guarantee an adequate and sustainable urban development by correcting market failures and preventing diseconomies to arise in urban environment. 9.187 Surely, both are correct, in principle, but in fact not all public policy makers recognize the necessity of increasing and diversifying housing supply for low income segments using formal markets. This is a more effective strategy to deal with poor people access to housing and prevent informal settlements to arise than work on crystallized situation of informality in a reactive way. 9.188 Land use regulations that are excessively demanding, plenty of formalisms, constitute elements that increase costs of infrastructure and housing production. Ordinances regarding roads design, for example, based on the premise of intense use of cars generate extensive land consumption, even in areas where car ownership is low. Although adoption of minimum design standards can improve performance of some spatial features and is important to enforce provision of public spaces, these demands eventually are not accomplished by low income segments pushing them to irregularity, informality and even illegality. 9.189 Limitation on densities also is a critical factor since it hampers housing supply that increases housing and land prices mainly in areas where housing demand is great. Thus, despite its benefits, limitation on densities can distort land prices, constraining affordable housing production. In addition, depending on the specific urban market conditions, constraints on the formally housing production can affect land prices even of plots produced without infrastructure and title, encouraging informal production. In addition, expectations on future regularization make land prices in informal market to increase, rent that is totally appropriated by informal developers. 9.190 Roughly, planners ignore costs of public intervention on real estate markets and side effects are usually examined ex post. As a result, during revision of planning legislation and evaluation of the outcomes obtained by public policies on land use is usual to create new enforcements and ordinances to deal with negative effects from previously controls adopted. Systematic adoption of discretionary rules in controlling urban development without a comprehensive cost-benefit analysis has implied in effects that distort urban market operations and increase operational costs of the city. 9.191 Planning controls that neglect opportunity costs of an efficient land allocation can result in constraints on housing supply inducing urban sprawl that, in turn, increase costs of transportation and infrastructure provision, for example. As such, it is important to distinguish the areas where controlling urban development produces more benefits than costs and identify who will pay for that, directly or indirectly. 9.192 Formal production of serviced land also is hindered by the blockages imposed by excessive requirements and bureaucratic proceedings for approving, producing and registering developments. Bureaucracy increases time consuming that, in turn, increases costs of capital and uncertainty on business. Time can affect economic feasibleness of real estate formal production, mainly for small and medium construction firms, due to the high costs of capital. In order to improve participation of small and medium firms in urban development it is important to reduce red tape costs and uncertainty by becoming bureaucracy proceedings more efficient and transparent. 9.193 In order to do so it is necessary to break the traditional paradigm of a commanded managing approach to urban development based on a strong public intervention. Less restrictive and a more permissive land use Inputs to a Strategy for Brazilian Cities Page 254 ordinances can improve formal land development and affordable housing production. In this sense a strategic interaction between the private sector and the government should be searched in order to lead to a committed action between these agents where the results achieved tend to be more effective. In addition, while discretionary and punitive ordinances tend to eliminate private incentives for urban development, tax and fiscal policies may be more effective in managing expectations. 9.194 Associations and partnerships involving public sector, developers and land owners could improve the feasibleness for private participation in upgrading and regularization programs, as well as in producing serviced land and housing for low income groups in a preventive way. These associations could be strategically important to discourage the production of informality by the side of supply. At the same time, improvements on credit for low and medium income segments could be useful to incentive housing self-construction, since one of the reasons for the low economic feasibleness of popular housing construction is the cost of capital. 9.195 Enlarging the time of infrastructure construction also is important to fit the financial conditions of the market to financial needs of construction. Adjust the construction time to the payment condition of poor people is fundamental to balance cash flows of small firms, avoiding them to get loans at high interest rates. Production of progressive developments, through semi-urbanized plots, for example, could improve the conditions of small firms in participating in urban development under competitive conditions. In addition, production of serviced land in stages seems to be an important strategy to adjust land development to the capacity of payment of low income families. 9.196 Alternative building typologies that allow increase densities is an important step to obtain gains of scale in serviced land and housing production, decreasing overall costs. However, this can be limited by the single family plot 'culture', even among low income groups, and by urban legislation that does not predict some alternative forms of shared property of land. Several intermediate typologies of housing buildings and land subdivision, different from the single family plots that are intensive in land consumption or the vertical condominiums that may overcharge infrastructure systems, are not regulated by the legal framework currently in force. 9.197 It is interesting to mention that the new regulatory set of instruments for intervening on urban markets offered by the City Statute, a Federal Law passed in 2001, seems to be more positive and adjusted to the uncertainty and dynamic of the markets than the former static legislation structured on technocratic approaches. The statute intends to assume the 'urban management' aspect more than 'planning' the shape of the city, that is, urban management is seen as a continuous process and not as a goal itself. Further, the statute offers new possibilities of public and private partnerships in urban development, broadening possibilities in sharing among public and private sectors responsibilities in producing the city. 9.198 Relaxing urban land use ordinances can play an important role in improving the conditions to extent formal housing production for the low income segments, but it is not sufficient. Totally deregulated markets in contexts where are large the income differences and the social inequalities have shown to be counterproductive, since market interactions can contribute to wealthy concentration. In asymmetric contexts the presence of government as a regulatory agent seems to be important to correct market failures and redistributing the costs and the benefits from urban development. In this way, fiscal policies seem to be more effective than discretionary ordinances in controlling urban development and land use. 9.199 Finally, improvements on low income segments access to housing and serviced land should be viewed as a necessary step to reduce poverty. However, programs that pursue to facilitate or improve access of poor people to land and housing should be accompanied by a comprehensive approach that combines concerns on an extensive access to overall urban markets. Producing housing for low income groups by concentrating poverty in locations far from job opportunities, for example, is not effective in the long run since it increases other costs for those who cannot afford them feeding back mechanisms of exclusion. Hence improvements on overall social conditions, like health, educational and employment have to be comprehended along housing policies in order to guarantee their sustainability. Inputs to a Strategy for Brazilian Cities Page 255 References ABIKO, Alex; CARDOSO, Luiz Reynaldo de Azevedo, RINALDELLI, Ricardo; HAGA, Heitor Cesar Riogi. Basic costs of slum upgrading in Brazil. Third Urban Research Symposium on "Land Development, Urban Policy and Poverty Reduction". World Bank and IPEA (Instituto de Pesquisa Econômica Aplicada). Brasília, 2005. ALONSO, William. Location and Land Use. Harvard University Press, 1964. BERTAUD, Alain and RENAUD, Bertrand. Socialist Cities Without Land Markets, in Journal of Urban Economics 41, 137-151, 1997. BERTAUD, Alain and MALPEZZI, Stephen. "The Spatial Distribution of Population in 48 World Cities: Implications for Economies in Transitions". The Center for Urban Land Economics Research, The University of Wisconsin, Madison, Wisconsin. 2003. BERTAUD, Alain and MALPEZZI, Stephen. "Measuring the Costs and Benefits of Urban Land Use Regulation: A Simple Model with an Application to Malaysia". Journal of Housing Economics, Vol. 10, pp 393-418, 2001. BRASIL. Estatuto da Cidade (2002). "Estatuto da Cidade: Guia para Implementação pelos Municípios e Cidadãos". Brasilia: Câmara dos Deputados, Coordenação de Publicações, 2005. DOWALL, David E. and CLARK, Giles. "A Framework for Reforming Urban Land Policies in Developing Countries". The International Bank for Reconstruction and Development/The World Bank, Washington. Urban Management Programme Discussion Paper, August, 1996. EVANS, Alan W. "Building Jerusalem: Can Land Use Planning Affect Economic Growth?". Lincoln Institute of Land Development. Conference Paper, 2002. FUNDAÇÃO JOÃO PINHEIRO. Centro de Estudos Políticos e Sociais. "Déficit Habitacional no Brasil". Belo Horizonte, 2001. Centro de Estatística e Informações. Déficit Habitacional no Brasil. Belo Horizonte, 2005. GLAESER, Edward L.; KAHN, Mathew E.; RAPPAPORT, Jordan. "Why do the Poor Live in Cities?". Working Paper 7636, Cambridge, National Bureau of Economic Research: NBER Working Paper Series, April 2000. 35 p. GLAESER, Edward Land GYOURKO, Joseph. "The Impact of Zoning on Housing Affordability". Working Paper 8835, Cambridge, National Bureau of Economic Research: NBER Working Paper Series, March 2002. IBAM. "Estudo de Avaliação da Experiência Brasileira sobre Urbanização de Favelas e Regularização Fundiária". Rio de Janeiro: Instituto Brasileiro de Administração Municipal: Assessoria Internacional. Projeto nº 17.408. Volumes 1 e 2. Mimeo, outubro, 2002. LEME, Maria Cristina da Silva, org. Urbanismo no Brasil: 1895 ­ 1965. 1 ed. São Paulo: Studio Nobel; FAUUSP; FUPAM, 1999. LEROY, Stephen F., SONSTELIE, Jon. "Paradise Lost and Regained: Transportation Innovation, Income, and Residential Location". Journal of Urban Economics 13, 67-89 (1983). MALPEZZI, Stephen. "The Regulation of Urban Development: Lessons from International Experience". University of Wisconsin Center for Urban Land Economic Research. Wisconsin-Madison CULER working papers No. 99-07. Draft. July, 1999. Inputs to a Strategy for Brazilian Cities Page 256 MALPEZZI, Stephen; GUO, Wen-Kai. "Measuring `Sprawl': Alternative Measures of Urban Form in U.S. Metropolitan Areas. The Center for Urban Land Economics Research. Revised, January 15, 2001. MANDELL, Paul. "A Teoria e a Prática do Planejamento Urbano Brasileiro". Revista Geografia e Meio Ambiente. Número 6 (11-12): 207-219, outubro 1981. MILLS, Edwin S.; HAMILTON, Bruce W. Urban Economics. 5 ed. New York: Addison-Wesley Educational Publishers, 1994. O'SULLIVAN, Arthur. Urban Economics. 4th Ed. Europe: Irwin/Mcgraw-Hill, International Edition, 2000. PENDALL, Rolf. "Local Land Use Regulation and the Chain of Exclusion". Journal of the American Planning Association; Spring 2000; Vol. 66, No. 2; pp 125-142. RICHARDSON, Harry Ward. Economia urbana. Rio de Janeiro: Interciência, 1978. SABATINI, Francisco. 1998. "Land market reform and residential segregation in Chile", Conference paper, código CP98A11. Cambridge MA: Lincoln Institute of Land Policy. Available in: http://www.lincolninst.edu/pubs/dl/817_Sabatini.pdf. SERRA, M. V.; DOWALL, David; MOTTA, Diana and DONOVAN, Michael. "Urban Land Markets and Urban Land Development: An Examination of Three Brazilian Cities: Brasília, Curitiba and Recife". Institute of Urban and Regional Development. University of California at Berkeley. Working Paper 2004-03 SMOLKA, Martim O. "Informalidad, pobreza urbana y precios de la tierra." En Land Lines Newsletter, volumen 15, número 1. Cambridge, MA: Lincoln Institute of Land Policy. Enero, 2003. SMOLKA, Martim O. and DAMASIO, Claudia P. "El urbanizador social: un experimento del politica del solo en Porto Alegre". In Land Lines Newsletter, volume 17, number 2. Cambridge MA: Lincoln Institute of Land Policy. April, 2005. SMOLKA, Martim O. and SABATINI, Francisco. "El debate sobre la liberalización del mercado de suelo en Chile". In Land Lines, volume 12, number 1. Cambridge MA: Lincoln Institute of Land Policy. Enero, 2000. UNDP. Human Development Report 2004: Cultural liberty in today's diverse world. United Nations Development Programme, New York, USA, 2004. VILLAÇA, Flávio. Espaço Intra-Urbano no Brasil. São Paulo, SP Studio Nobel, FAPESP, Lincoln Institute, 1998. Inputs to a Strategy for Brazilian Cities Page 257 10. Brazil's Urban Land and Housing Markets How well are they working? by David E. Dowall Acknowledgements The author would like to thank Pedro Peterson for his research assistance to support the preparation of this paper. Valuable comments and suggestions were provided by Mila Freire, Edesio Fernandes, Paul Avila and Fernanda Furtando and from participants at a World Bank-Lincoln Institute Seminar in Brasilia on March 6, 2006. Greg Ingram and Martim Smolka of the Lincoln Insitute provided detailed comments on the paper. Any errors that remain are the responsibility of the author. Introduction 10.1 This paper uses a macro, national-level perspective to assess urban land and housing market outcomes across Brazil. It is based on available empirical data from IBGE, field studies, the Fundacion Joao Pinhero, and other sources. The paper starts by posing and answering the following questions: What are the characteristics of well-functioning urban land and housing markets? How well are Brazil's urban land and housing markets performing relative to other countries? It then proceeds to provide a assessment of urban land and housing market outcomes in Brazilian cities. The paper concludes by exploring a range of opportunities for enhancing urban land and housing market outcomes. 10.2 This paper is one of four papers prepared under a collaborative World Bank-Lincoln Institute of Land Policy project. The other papers are: · Paulo C. Avila, "Urban Land Use Regulations in Brazil: Land Market Impacts and Access to Housing." · Fernanda Furtado and Pedro Jorgensen, "Value Capture in Brasil: Issues and Opportunities." · Edesio Fernandes, "Legal Aspects of Urban Land Development in Brazil." 10.3 Each paper takes a distinct perspective on the overall topic of urban land policy in Brazil. Paulo Avila's paper reviews the various models of urban land use planning and regulation in Brazilian cities. He then analyzes the effects of planning regulations, titling and infrastructure provision on residential land prices, and the efficiency of residential land subdivision. Avila's paper is one of the few quantitative econometric and financial analyzes of urban land and housing markets, building on the previous work of Serra, Dowall, Motta and Donovan [2004]. His analysis indicates that land use planning regulations and infrastructure provision significantly and positively affect urban residential plot prices. 10.4 Fernanda Furtado and Pedro Jorgensen's papers explores the concept of land value capture--the range of tax and policy instruments that can be used to generate public resources to fund public investments to support urban development. These instruments work by assessing fees, taxes and charges on the incremental increase in land values generated by public investments. Furtado and Jorgenson outline eight types of value capture models and illustrate how they might be used to finance, in whole or in part, the costs upgrading informal settlements throughout Brazil. 10.5 Edesio Fernandes' paper presents an historical analysis of land and property legislation in Brazil which provides a thorough understanding of the role of federal legislative actions from the early twentieth century to the significant policy reforms of the past 10 years, culminating in the promulgation of the City Statute [2001]. Fernandes's paper discusses the fundamental issues surrounding informality and lack of secure land tenure in favelas and irregular settlements. He outlines issues and opportunities for reforming land titling and registration systems in Brazil and discusses how these reforms could contribute to the regularization and upgrading of low- income settlements embedded in Brazil's vast system of cities. Inputs to a Strategy for Brazilian Cities Page 258 10.6 The present paper attempts to make the case for reforming urban land and housing policies in Brazil, by arguing that the historical as well as current performance of Brazil's urban land and housing markets are below their potential. As a consequence, urban land and housing markets are not providing sufficient housing opportunities for low- and middle-income families, and contribute to a growing housing deficit and widespread housing informality [FJP, 2002 and 2005]. The paper attempts to make the case that although dwelling unit production is satisfactory relative to household formation, the provision of infrastructure and urban services is unsatisfactory. Characteristics of Well-Fnctioning Urban Land and Housing Markets 10.7 Urban land and housing markets should efficiently allocate land and housing resources between suppliers and demanders. Housing supply should reasonably match the housing demands of households in terms of prices, locations and quality attributes. In most market economies, private production (from large merchant builders to self-built housing to informally provided housing in favelas and irregular settlements) is the predominant mode of housing production. Aside from a few countries, such as Singapore, public provision of housing is miniscule relative to overall production. The full range of housing supply, including new as well as existing units should provide households with affordable options for purchase as well as rental. Depending on household incomes and housing prices, the private real estate markets typically produce housing that is affordable to households to the 30th to 40th percentile of the income distribution [Dowall, 1989 and 1990]. Households with lower incomes, typically rent accommodations, share housing with extended families or postpone forming households. Some are fortunate to get housing assistance from government sources. 10.8 Achieving this level of performance requires that housing markets produce housing that is priced between 3 and 6 times total household income. Middle and low-middle income households should be able to afford such units by saving money for down payments and taking out mortgages from housing lenders. Unfortunately, housing supplies are frequently constrained and housing prices are much higher in relation to income. This is due to restrictive land use regulations, complex land titling and registration, lack of investment in basic infrastructure to serve residential development projects and limitations on the availability of construction and borrower financing. 10.9 In middle-income developing countries housing price to income ratios vary considerably. As household incomes rise, the variation of the ratio diminishes as housing and real estate markets mature and broaden their range of housing products (and prices). In cases where formal housing production is constrained, house price to income ratios increase. Figure 10.1illustrates the relationship between housing price to income ratios and household incomes for a 27 middle income countries.132 It is based on tabulations of the World Bank's housing Indicators program. The data were collected in 1998, and are based on data from a sample of large cities in each country [WDR, 2000]. The ratio of median housing prices to median household income ranges from a low of 1.7 for Poland to 20 for Lithuania. Brazil has a ratio of 12.5. This is higher than all Central and Latin American countries included in the data series. Only five countries have higher ratios than Brazil--Panama, Serbia and Montenegro, Latvia, Cote d'Ivoire and Lithuania. On the other hand 11 of the 27 countries have ratios below 6, suggesting good performance. 132Middle income countries, as defined by the World Bank, have per capita Gross National Incomes ranging from $826 to $10,065 (in 2004 dollars). This is further divided into low-middle-income ($826-$3255) and upper-middle-income ($3256- $10,065). Inputs to a Strategy for Brazilian Cities Page 259 Figure 10.1 Median Housing Prices and Median Household income, Middle income Countries, 1998 25 e 20 m conI d ol usehoH 15 ot ceirP gn 10 usioHfo oitaR 5 0 0 2000 4000 6000 8000 10000 12000 14000 16000 Median Household Income USD Source: World Bank, World Development Report, 2000. Is there a Brazilian Paradox? 10.10 To motivate the reader, I would like to suggest that Brazil urban housing market suffers from a paradox-- housing is expensive relative to income (See Figure 10.1) and its lacks infrastructure services and secure land tenure. The private sector is capable of producing satisfactory numbers of dwelling units, despite the fact that the public sector is not capable of producing enough infrastructure services or planning and approving enough residential subdivisions to support housing development. The result is an urban land and housing market paradox--expensive housing lacking water and sanitation, secure land tenure,133 adequate circulation and common areas for schools and parks. Table 10.1 compares the housing characteristics of Brazilian cities with those in other countries134, and lends some credence to the paradox. In Brazilian cities, 93 percent of the housing stock is classified as permanent; this is significantly higher than the comparable rate for low-middle-income countries 86 percent. On the other hand, Brazil does poorly with respect to the percentage of housing units with piped water connections--64 percent versus 74 percent for cities in low-middle-income countries. At the same time, its portion of unauthorized housing units 23 percent is well below levels found in other low-middle-income countries--36 percent. So the overall scorecard for Brazil is again a paradox--both good--a relatively low rate of unauthorized housing and a high portion of permanent structures; and bad--a relatively low level of access to water supply. Compared to other Latin American countries Brazil ranks poorly in terms of providing infrastructure to support residential development [UNECLAC, 2003].135 133According to the World Bank's Doing Business survey, Brazil ranks eighth out of nine countries on ease of property registration [Doing Business Survey, 2005]. 134The World Bank classifies Brazil as a low-middle-income country. 135The percentages in Table 10.1 have limitations. They are based on binary definitions of service access and do not reflect poor quality of service, such as water supply limits to 3-4 hours per day. Inputs to a Strategy for Brazilian Cities Page 260 Table 10.1 How Do Brazilian Cities Compare to Cities in Other Countries (1990s) Cities in Percentage of Percentage of Percentage of Average Housing Units that Housing Units with housing units per are Permanent Piped Water that are captia Structures unauthorized GNI, 2004, US$ Low-income 67 56 64 507 countries Low-middle-income 86 74 36 1,686 countries Brazilian Cities 93 64 23 3,000 Middle-income 94 94 20 4,769 countries Middle-high-income 99 99 3 16,046 countries High income 100 100 0 32,112 countries Source: UNCHS An urbanizing world Global Report on Human Settlements, 1996. Caveats about the data Used in this Paper 10.11 In Brazil like most other developing countries, housing and urban planning experts constantly discuss the informal housing crisis--slums, shanty towns, squatter settlements and the like. Many settlements take on iconic positions--Cairo's "City of the Dead"--a squatter settlement encamped on top of one of the city's largest cemeteries; "Smokey Mountain"--a massive slum located on top of Manila's main garbage dump; or Mumbai's Dharavi--"Asia's biggest slum." These settlements are horrific manifestations of society's inability or unwillingness to address the housing needs of low-income residents. 10.12 Urban planners and housing policy professionals and advocates are fully justified in voicing outrage about these terrible conditions. But at the same time they fail to provide any systematic assessment of actual urban land and housing market outcomes in developing country cities. This paper attempts to bridge this gap by providing a quantitative assessment of Brazil's urban land and housing markets. 10.13 There are a number of important caveats that I need to offer before proceeding. First of all this paper starts by taking an integrated approach to evaluating Brazil's urban land and housing markets. It looks at the entire spectrum of housing units, both formal and informal; this includes dwelling units located in fully approved housing projects--subdivisions and apartment complexes--as well as favelas, and irregular and illegal settlements. This definition is broad and incorporates a wide range of housing conditions, and has the advantage of allowing one to make a macro-level assessment of overall housing supply and demand. How many total units are produced in Brazil over a year? How many new households are formed each year? How many units need to be replaced due to deterioration, demolitions and change of use? As will be explained below, total housing production of both formal and informal dwelling units is slightly less than new household formation [World Bank, 2002 ]. 10.14 The second caveat relates to the definition of informality. Our review of the literature on housing informality indicates that it is based on three distinct but interdependent factors--type of land tenure; access to infrastructure; and physical characteristics of settlements and dwelling units. As is commonly the case in many Inputs to a Strategy for Brazilian Cities Page 261 countries, census data on informal housing stocks is highly inaccurate. Some countries ignore informal housing altogether, others grossly undercount it. Brazil is no exception and data from IBGE are problematic. In order to maintain the empirical mode of analysis, I have chosen to define informal housing based on the most inclusive single measure--access to infrastructure services. This definition allows widespread measurement of stock and flow trends for municipalities and metropolitan regions over time. However, it may understate informality by excluding cases where urban services are available, but where households lack secure and legal land title or that the subdivisions where the housing units are located are poorly planned and executed. With these caveats in mind, the next sections of the paper map out a broad assessment of Brazil's urban land markets. Performance of Brazil's Urban Land and Housing Markets during the last half of the Twentieth Century 10.15 At the country level, Brazil has undergone a massive shift in the spatial patterns of its population. Between 1950 and 2000, the country added 117,600,000 persons, approximately 2.4 million per year. More dramatically, the spatial structure of the population shifted from being predominately rural to urban. As this section will illustrate, the most challenging period of rapid urbanization has passed. In the 1990s population and household growth slowed as Brazil passed through its urban transition. Using IBGE census data, Figure 10.2and Figure 10.3 illustrate that in 1950, about 64 percent of Brazil's population was located in rural areas and 36 was located in urban areas. By 1980, the pattern was completely reversed--32 percent rural and 68 percent urban. Since then, urban population dominance has increased, and by 2000, approximately 81 percent of the Brazilian population lived in cities, and 19 percent lived in rural areas. Figure 10.2 Percent Distribution of Urban and Rural Population 90.0% 80.0% 70.0% 60.0% tnecr 50.0% Percent Urban Percent Rural Pe40.0% 30.0% 20.0% 10.0% 0.0% 1950 1960 1970 1980 1990 2000 Year Source: IBGE, 2005. Inputs to a Strategy for Brazilian Cities Page 262 Figure 10.3 Urban and Rural Population Trends in Brazil, 1950-2000 180,000,000 160,000,000 140,000,000 120,000,000 onital 100,000,000 Rural Urban opuP 80,000,000 60,000,000 40,000,000 20,000,000 0 1950 1960 1970 1980 1991 2000 Year Source: IBGE, 2005. 10.16 In absolute terms the increase in urban population has been enormous. Table 10.2 shows that between 1950 and 2000, the country's urban population increased by 118,914,548, while at the same time its rural population slightly decreased by -1,314,502. While some of these changes reflect alterations of administrative boundaries and definitions of what constitutes an urban place, they overwhelmingly reflect massive rural to urban migration-- on average, cities in Brazil added 2,378,291 persons per year between 1950 and 2000. 10.17 Rural-urban migration was particularly strong in the 1950s and 1960s, reflecting the country's emerging economic growth, and social transformation. During the 1970s, 1980s and 1990s, rural-urban migration slowed and as a consequence urban population growth slowed as well. In percentage terms annual urban population growth has ranged from a high of 3.0 percent during the 1950s to a low of 1.4 percent Table 10.2 Decade-by-Decade Change in Urban and Rural Population Population Change Annual Percent Change Total Urban Rural Total Urban Rural 1950-60 18,126,060 12,520,143 5,605,917 3.0% 5.2% 1.6% 1960-70 23,068,580 20,781,950 2,286,630 2.9% 5.2% 0.6% 1970-80 25,863,669 28,351,425 -2,487,756 2.5% 4.4% -0.6% 1980-91 27,822,769 30,554,581 -2,731,812 2.1% 3.3% -0.7% 1991-00 22,718,968 26,706,449 -3,987,481 1.4% 2.2% -1.2% 1950- 117,600,046 118,914,548 -1,314,502 2000 2.4% 4.1% -0.1% Source: IBGE, 2005. during the 1990s. This decline in the percentage rate of growth is common throughout Latin America as rural areas depopulate and as overall rates of natural population increase slow. However, in absolute terms, annual urban population growth continued to grow up until the 1990s and will continue to do so in the future, but it will be driven mainly by natural population increase and less by rural-urban migration. 10.18 Rural areas of Brazil have actually been losing population since the 1970s and contain about 10,000,000 fewer persons in 2000 than in 1970. On the other hand urban areas have been increasing rapidly since the 1950s, growing from 18.8 million persons in 1950 to 137.7 million in 2000--more than a sevenfold increase. Annual Inputs to a Strategy for Brazilian Cities Page 263 urban population growth has ranged from approximately 1.25 million during the 1950s to a peak of 3 million during the 1980s. During the 1990s, the annual rate of growth has slightly declined to 2.7 million persons. Urbanization of Brazil's 15 largest Metropolitan Regions 10.19 Urbanization trends can be disaggregated to examine population growth in Brazil's fifteen largest metropolitan areas. Table 10.3 presents tabulations of population trends for Brazil's 15 largest metropolitan regions from 1950 to 2000. Over the fifty year period, these cities accounted for a decreasing share of total urban population, falling from 54.8 percent of total urban population in 1950 to 42.8 percent in 2000--indicating a deconcentration of urban population. 10.20 However, despite the declining share, absolute population change has been significant. Table 10.4 presents population increases for the fifteen metropolitan areas by decade from 1950-60 to 1991-2000. Population growth in the 15 metropolitan areas was the greatest during the 1970-80 decade when they added a total of 12.6 million persons. Since then the absolute decadal increases have declined, and during 1991-2000 period stood at 9.2 million. This is consistent with their decreasing share of total urban population, these 15 metropolitan areas accounted for a relatively declining share of country-wide increases in urban population, falling from 52 percent of the total increase during the 1950s to 34.6 percent during the 1990s. What these trends show, is that over the 50 years, urbanization has gradually slowed in Brazil's 15 largest metropolitan areas, This is due to two factors-- urban growth is shifting to areas outside the boundaries of the 15 metropolitan areas and that second tier metropolitan areas are accounting for an increasing share of population increase. Table 10.3 Urban Population Trends in Brazil's 15 Largest Metropolitan Regions, 1950 to 2000 Metropolitan Region Total Population 1950 1960 1970 1980 1991 2000 Belém 268,252 422,648 669,768 1,021,473 1,401,305 1,795,536 Belo Horizonte 565,970 990,055 1,719,490 2,676,352 3,515,542 4,349,425 Brasilia 141,742 537,492 1,176,908 1,601,094 2,051,146 Curitiba 333,138 554,515 875,269 1,497,352 2,061,531 2,726,556 Fortaleza 464,507 699,262 1,091,117 1,651,744 2,401,878 2,984,689 Goiânia 82,826 196,596 442,790 827,446 1,230,445 1,639,516 Grande São Luís 119,785 180,747 302,609 498,958 820,137 1,070,688 Grande Vitória 123,281 213,449 410,103 744,744 1,126,638 1,425,587 Maceió 178,705 240,733 357,514 522,173 786,643 989,182 Natal 169,293 245,303 373,754 554,223 826,208 1,043,321 Porto Alegre 842,390 1,263,401 1,751,889 2,468,028 3,230,732 3,718,778 Recife 843,409 1,275,125 1,827,173 2,386,453 2,919,979 3,337,565 Rio de Janeiro 3,178,310 4,869,103 6,891,521 8,772,277 9,814,574 10,894,156 Salvador 463,545 739,799 1,147,821 1,766,724 2,496,521 3,021,572 São Paulo 2,662,776 4,791,245 8,139,705 12,588,745 15,444,941 17,878,703 Total 15 Metros 10,296,187 16,823,723 26,538,015 39,153,600 49,678,168 58,926,420 Total Brazil Urban 18,782,891 31,303,034 52,084,984 80,436,409 110,990,990 137,697,439 15 Metros as a percent or total urban 54.8% 53.7% 51.0% 48.7% 44.8% 42.8% Source: IBGE, 2005. Inputs to a Strategy for Brazilian Cities Page 264 Table 10.4 Urban Polulation Charge in the 15 Largest Metropolitan Areas, 1950-60 to 1991 - 2000 Metropolitan Region Change in Population 1950-60 1960-70 1970-80 1980-91 1991-2000 Belém 154,396 247,120 351,705 379,832 394,231 Belo Horizonte 424,085 729,435 956,862 839,190 833,883 Brasilia 141,742 395,750 639,416 424,186 450,052 Curitiba 221,377 320,754 622,083 564,179 665,025 Fortaleza 234,755 391,855 560,627 750,134 582,811 Goiânia 113,770 246,194 384,656 402,999 409,071 Grande São Luís 60,962 121,862 196,349 321,179 250,551 Grande Vitória 90,168 196,654 334,641 381,894 298,949 Maceió 62,028 116,781 164,659 264,470 202,539 Natal 76,010 128,451 180,469 271,985 217,113 Porto Alegre 421,011 488,488 716,139 762,704 488,046 Recife 431,716 552,048 559,280 533,526 417,586 Rio de Janeiro 1,690,793 2,022,418 1,880,756 1,042,297 1,079,582 Salvador 276,254 408,022 618,903 729,797 525,051 São Paulo 2,128,469 3,348,460 4,449,040 2,856,196 2,433,762 Total 15 Metros 6,527,536 9,714,292 12,615,585 10,524,568 9,248,252 Total Brazil Urban Population Change 12,520,143 20,781,950 28,351,425 30,554,581 26,706,449 Percent 15 of Total 52.1% 46.7% 44.5% 34.4% 34.6% Source: IBGE, 2005. Housing demand and Housing Production in Urban Brazil 10.21 Housing demand is determined by population growth, household formation, income, and requirements to replace old dilapidated housing stock and replace housing units removed from the stock. Housing production trends in Brazilian cities has largely followed trends in urbanization and overall production of formal and informal housing has reasonably paced increases in household growth. 10.22 Table 5 presents trends in housing units by metropolitan region for census years 1970 to 2000 for Brazil's fifteen largest metropolitan areas. During the 30 year period, informal and formal housing stock increased from 5.4 to 16.5 million units--a gross increase of 11.2 million units. On an annual basis this is 373,000 units a year. For all urban areas in Brazil, the total housing stock increased from 10.5 to 38.7 million between 1970 and 2000. This is approximately 940,000 units per year. Overall, this is a remarkable level of residential construction and investment, although, as we will explain below, much of it is produced through informal channels and is not supplied with adequate infrastructure and secure land titling. It is also significant that persons per household declined dramatically over the 30 year period, falling from 5.0 persons per unit to 3.6 persons per unit, a percent decrease of 28 percent. Inputs to a Strategy for Brazilian Cities Page 265 Table 10.5 Permanent Dwelling Units for 15 Largest Metropolitan Regions and Decade by Decade Metropolitan Region Number of Dwelling Units 1970 1980 1991 2000 Belém 105,675 184,364 292,218 419,791 Belo Horizonte 319,386 568,116 858,303 1,189,609 Brasilia 99,303 253,950 386,396 556,762 Curitiba 178,338 342,427 543,032 790,982 Fortaleza 188,412 320,663 523,219 731,278 Goiânia 83,514 180,810 312,228 467,227 Grande São Luís 49,228 90,563 167,174 249,682 Grande Vitória 74,579 161,041 279,674 401,091 Maceió 66,028 104,667 176,051 247,536 Natal 65,023 109,867 183,440 260,220 Porto Alegre 380,128 630,867 936,221 1,153,274 Recife 332,871 481,456 678,819 873,407 Rio de Janeiro 1,489,189 2,152,226 2,743,178 3,302,119 Salvador 205,588 353,789 581,080 807,352 São Paulo 1,721,964 2,999,178 4,083,306 5,079,188 Total of the 15 MR 5,359,226 8,933,984 12,744,339 16,529,518 Persons per dwelling unit 5.0 4.4 3.9 3.6 Total Urban 10,501,000 18,364,477 28,532,388 38,678,933 Metropolitan Region Change in Number of Dwelling Units 1970-80 1980-1991 1991-2000 1970-2000 Belém 78,689 107,854 127,573 314,116 Belo Horizonte 248,730 290,187 331,306 870,223 Brasilia 154,647 132,446 170,366 457,459 Curitiba 164,089 200,605 247,950 612,644 Fortaleza 132,251 202,556 208,059 542,866 Goiânia 97,296 131,418 154,999 383,713 Grande São Luís 41,335 76,611 82,508 200,454 Grande Vitória 86,462 118,633 121,417 326,512 Maceió 38,639 71,384 71,485 181,508 Natal 44,844 73,573 76,780 195,197 Porto Alegre 250,739 305,354 217,053 773,146 Recife 148,585 197,363 194,588 540,536 Rio de Janeiro 663,037 590,952 558,941 1,812,930 Salvador 148,201 227,291 226,272 601,764 São Paulo 1,277,214 1,084,128 995,882 3,357,224 Total 15 MR 3,574,758 3,810,355 3,785,179 11,170,292 Total Urban 7,863,477 10,167,911 10,146,545 28,177,933 Source IBGE, 2005. 10.23 Regardless of whether these units are located in legal or illegal residential subdivision, or favelas, the increases in housing stock are impressive. They represent significant financial accomplishments of households, Inputs to a Strategy for Brazilian Cities Page 266 especially for low and moderate income households. Figure 10.4 illustrates country-wide (urban and rural) private gross residential capital outlays and per capita outlays in constant 1999 Reais [IBGE, 2005].136 As it shows, spending has been robust and has increased in per capita real terms from R$ 131.4 in 1970 to R$ 310.0 in 2000. Despite the ups and downs of the Brazilian economy during the 1980s private investment in housing has increased on a decade-by decade basis. In constant Reais, private residential investment has increased 4.3 times between 1970 and 2000. Figure 10.4 Private Investment in Housing is Robust and Increasing in Real Terms 350 60,000.00 300 0) 50,000.00 10 9= 99 atipaCreP 250 (1 40,000.00 s aieR 200 siaeR 30,000.00 ssorGt 150 ntat nsoC 20,000.00 100 tansnoCfo 10,000.00 50 snoilli M 0 0.00 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 Year Per Capita Gross Residential Investment Source: Suzigan, W. A indústria brasileira : origem e desenvolvimento. São Paulo: Brasileiense, 1986; Abreu, M. P. Verner, D. Long-term brazilian economic growth 1930-1994. Paris: OECD, 1997. (Development Centre Studies. Long-term growth series/OCDE); IBGE, Diretoria de Pesquisas, Departamento de Contas Nacionais. 10.24 How adequate has this spending been in terms of providing sufficient housing stock for new households? The question can be partially answered by comparing the relationship between housing production and increases in households. Table 10.6 presents estimates of increases in household formation for the 15 major metropolitan regions from 1970 to 2000. Table 10.6 reveals that household formation has been robust in the 15 metropolitan areas. Between 1970 and 2000, these 15 metropolitan regions added approximately 10.6 million households. In total, the number of households in all urban areas of Brazil increased by 27.2 million over the 30 year period-- about 900,000 households per year. As pointed out above, a main factor of increased household formation is the reduction in persons per household. With smaller persons per dwelling unit (and by extension persons per household) a falling household size means that the number of households per 1000 population will increase. Its is interesting to note that the 28 percent decline in persons per dwelling unit reflects a flexible response in housing supply to accommodate more households per 1000 of population.137 136The figures pertain to fixed capital only and do not include land, operating or maintenance costs. 137If housing supply was tightly constrained, we would expect to see a stable or increasing number of persons per dwelling unit as people delayed household formation, doubled up with other households or extended families. Inputs to a Strategy for Brazilian Cities Page 267 Table 10.6 Trends in Household formation 15 Largest Metropolitan Regions, 1970 - 2000 Metropolitan Region Households 1970 1980 1991 2000 Belém 128,063 219,200 332,063 477,536 Belo Horizonte 328,774 574,324 833,067 1,156,762 Brasilia 102,771 252,555 379,406 545,518 Curitiba 167,355 321,320 488,514 725,148 Fortaleza 208,627 354,452 569,165 793,800 Goiânia 84,663 177,564 291,575 436,041 Grande São Luís 57,860 107,073 194,345 284,757 Grande Vitória 78,414 159,816 266,976 379,145 Maceió 68,358 112,054 186,408 263,080 Natal 71,463 118,932 195,784 277,479 Porto Alegre 334,969 529,620 765,576 989,037 Recife 349,364 512,114 691,938 887,650 Rio de Janeiro 1,317,690 1,882,463 2,325,728 2,897,382 Salvador 219,469 379,125 591,593 803,610 São Paulo 1,556,349 2,701,447 3,659,939 4,754,974 Total 15 5,074,190 8,402,060 11,772,078 15,671,920 Total Urban 17,610,993 25,156,482 37,843,782 44,857,290 Metropolitan Region Household Change 1970-80 1980-91 1991-2000 1970-2000 Belém 91,137 112,863 145,473 349,473 Belo Horizonte 245,550 258,742 323,695 827,988 Brasilia 149,784 126,851 166,111 442,747 Curitiba 153,965 167,194 236,633 557,792 Fortaleza 145,825 214,714 224,635 585,174 Goiânia 92,900 114,011 144,467 351,378 Grande São Luís 49,212 87,273 90,412 226,897 Grande Vitória 81,403 107,160 112,170 300,732 Maceió 43,696 74,354 76,672 194,722 Natal 47,468 76,852 81,695 206,016 Porto Alegre 194,651 235,957 223,460 654,067 Recife 162,751 179,824 195,712 538,286 Rio de Janeiro 564,772 443,266 571,653 1,579,691 Salvador 159,657 212,467 212,017 584,141 São Paulo 1,145,098 958,491 1,095,036 3,198,625 Total 15 3,327,870 3,370,018 3,899 ,842 10,597,730 Total Urban 7,545,489 12,687,300 7,013,508 27,246,297 Source IBGE, 2005. 10.25 Table 10.7 compares the housing stock increases of Table 10.5 with the increases in households presented in Table 10.6. Focusing on the 15 largest metropolitan areas, the 11.2 million housing stock increases between Inputs to a Strategy for Brazilian Cities Page 268 1970 and 2000 closely tracked the 10.6 million-increase in households. The overall ratio of housing stock increase to household increase for the 15 metropolitan areas is 1. 1--suggesting that 1.1 housing units were added to the stock of the 15 metros for every 1 household increase. Closer inspection of the ratio across the metropolitan areas reveals that 10 of the 15 metros are producing relatively more housing units per increase in household. On the other hand, housing markets in the metropolitan regions of Belem, Fortaleza, Grande Sao Luis, Maceio, and Natal are not producing enough units to accommodate new household formation. Table 10.7 Ratio of Change in Permanent Dwelling Units to Changes in the Number of Households Change in Permanent Dwelling units/Change in Households Metropolitan Region 1970-80 1980-1991 1991-2000 1970-2000 Belém 0.86 0.96 0.88 0.90 Belo Horizonte 1.01 1.12 1.02 1.05 Brasilia 1.03 1.04 1.03 1.03 Curitiba 1.07 1.20 1.05 1.10 Fortaleza 0.91 0.94 0.93 0.93 Goiânia 1.05 1.15 1.07 1.09 Grande São Luís 0.84 0.88 0.91 0.88 Grande Vitória 1.06 1.11 1.08 1.09 Maceió 0.88 0.96 0.93 0.93 Natal 0.94 0.96 0.94 0.95 Porto Alegre 1.29 1.29 0.97 1.18 Recife 0.91 1.10 0.99 1.00 Rio de Janeiro 1.17 1.33 0.98 1.15 Salvador 0.93 1.07 1.07 1.03 São Paulo 1.12 1.13 0.91 1.05 Total 15 MR 1.07 1.13 0.97 1.05 Total Urban 1.04 0.80 1.45 1.03 Source: Tables 10.5 and 10.6. 10.26 These ratios are very impressive, given the fact that they incorporate housing stock demolitions and removals. The net increase in the stock has, with the exception of the 1980s kept pace with strong household formation, driven by both population increases and smaller average household size. 10.27 Our first, level evaluation of Brazil's housing market indicates that there is a strong private (informal and formal) sector and that housing production is substantial. Private Gross Fixed Capital formation in the housing sector has increased by more that 4 fold in constant terms. On a per capita basis, real constant reais investments in housing have increased by about 2.35 times between 1970 and 2000. But as we shall see, most of the housing stock increases are in informal settlements with limited infrastructure services available. How Large is Brazil's Informal Housing Sector? 10.28 The previous section outlined the overall performance of Brazil's urban land and housing market, looking at both the formal and informal sectors of housing production and consumption. This section explores the role and performance of the informal sector in producing housing in Brazilian cities. Inputs to a Strategy for Brazilian Cities Page 269 10.29 As noted in the introduction to this paper, defining and systematically exploring informal housing is problematic [Pontual, 2005 and Pontual and Serra, 2005]. In the case of Brazil, there are widely differing estimates of housing informality both in term of the size of the informal housing stock and the rate at which informal housing units are added to the supply of housing. What defines informality? Informal housing can be defined along three main conceptual lines: security of land tenure; access to infrastructure services; and the physical characteristics of the settlement and the housing structures in it. Informal land subdivisions are a predominant component of informal housing provision. In the Brazilan case there are two types of informal land subdivisions--illegal subdivisions and clandestine subdivisions. 10.30 Illegal subdivisions are produced by a landowner or his agent. The subdivision of the parcel typically is done without government permission (approval of subdivision plan), lack of a legal physical cadastre identifying plots, and incomplete infrastructure provision. Purchasers of such lots will usually build housing over a 2-5 year period and given the lack of legal status will construct housing without obtaining building permits and inspections. 10.31 Clandestine subdivisions refer to settlements that are produced on land not owned by the developer or real estate agent. It is common but not impossible for clandestine subdivisions to be located on government land. Houses in clandestine subdivisions usually do not have secure tenure and usually do not have complete urban infrastructure services.138 Favelas are also invasions of land, but the subdivision of the land is typically unorganized--and does not follow a plan. Plots in favelas do not have legal title and nor do they have access to services. 10.32 The physical characteristics of informal settlements vary considerably. In clandestine subdivisions and favelas, housing construction can range from very poor ­temporary arrangements to reasonable good conditions-- brick walls, concrete floors and tin roofs. Condition varies by the age of the settlement--newer ones are more precarious and more established settlements have better housing conditions. Over time virtually all settlements go through an incremental process of upgrading. Some of this upgrading is self organized and some is based on government programs, where government agencies work with residents of informal settlements to provide secure tenure, make infrastructure investments in water, wastewater collection and treatment, drainage, electricity and solid waste collection. These programs also include assistance to homeowners to make improvements to their houses. Even, in cases where governments do not support or sanction upgrading, community-based efforts are organized to improve conditions through self-help activities. The overall result is that in most metropolitan regions, the stock of informal housing is constantly changing through additions, resettlements and upgrading efforts. 10.33 Figure 10.5 illustrates how the three dimensions of informality can be combined to categorize housing settlements and housing production into formal and informal classifications. Unfortunately, in terms of empirical data, Brazilian statistics on informal housing stock are incomplete and in some cases misleading. Census data from IBGE on housing units combines informal and formal units and does not provide any basis for distinguishing between the two types. The work of the Fundacion Joao Pinheiro [2002 and 2005] also does not shed much light on this matter. While their extensive research on Brazil's housing deficit provides specific tabulations of inadequate housing, over crowding, lack of access to infrastructure, and excessive rental payments these figures cannot be aggregated into overall estimates of informal housing stocks. 10.34 IBGE does however collect information on whether the housing units have access to infrastructure services and on the physical conditions of each dwelling unit, and tabulations of the number of households where the occupant has legal right to the structure, but not the land. But here again, the tabulations cannot be aggregated without the risk of significant double counting [IBGE, 2000]. 138For example, some favelas in Rio de Janiero (such as Favela da Rochinha) have most services, but still lack formal title. Also, as mentioned earlier, classifying settlements as either having or not have infrastructure services is problematic since this binary treatment does not capture the variable quality of infrastructure services. Inputs to a Strategy for Brazilian Cities Page 270 Figure 10.5 Defining Informal Housing is Complicated Urban Services Physical Characteristics Secure Land Title 10.35 As Figure 10.5 shows, informality can be limited to lack of infrastructure, lack of secure land title and poor physical conditions of housing and settlement layout. Quite often housing informality occurs with combinations of two or three of the above conditions. Since IBGE does not have data on land tenure, we have only two of the three variables necessary to measure informality. 10.36 Reliance on access to services and physical conditions while foregoing information on land tenure, is likely to undercount the stock of informal dwelling units in Brazil's urban areas. Unfortunately, we simply do not know how serious the under estimation is. If the incidence of dwelling units with infrastructure, good physical conditions and lack of secure land tenure is low, then the underestimation will be low. On the other hand if there are substantial numbers of units in cities that lack secure land title but have infrastructure and are in good physical condition, then the underestimation will be large. 10.37 Discussions with housing and land tenure experts in Brazil indicate the range of underestimation probably varies from city to city, with it being higher in the North and Northeast, where land titling and registration are less common [conversation with Edesio Fernandes, March 6, 2006]. In addition, many housing experts have noted that the IBGE data on access to infrastructure and physical conditions are inaccurate and that they frequently undercount informal housing. Despite the limitations with IBGE's data on informal housing, their estimates of housing units with access to infrastructure may provide a useful picture of housing conditions in Brazilian cities and therefore we will use them as a proxy for informal housing. 10.38 Figure 10.6 provides a tabulation of the percent of housing units without urban infrastructure services, by major metropolitan region in Brazil, based on the 2000 census. The figures range from over 10 percent for Sao Paulo to nearly 55 percent for Recife. Figure 10.6 Level of Informal Varies Widely Across Brazil, 2000 60 al 50 mrofnI 40 30 cent 20 erP 10 0 a lo re BelemFortalez cife or o Ri Re Salvad Ho rizonte oPau ritiba Cu toAleg Brasilia lo Sa Por Be Source: FJP, 2005. Inputs to a Strategy for Brazilian Cities Page 271 10.39 Figure 10.7 provides an example of changes in the informal housing stock in Rio de Janiero. Informal housing increased from virtually zero in 1900 to over 225,000 units in 1991. Since the 1960s, the rate of growth has slowed, but it is still increasing, and overspilling into outlining areas [O'Hare and Barke, 2003]. As a result, the proportion of Rio's housing stock that is located in favelas is declining. In 1970 about 13.5 percent of the housing stock was located in favelas, whereas by 1991 the portion had slightly declined to 12 percent, which is roughly consistent with the percentage indicated in Figure 10.6 Figure 10.7 Number of Favela Dwelling Units in Rio de Janeiro, 1900-1991 mulS 250000 in 200000 esaerc sti Un 150000 in evit gnille 100000 lau Dw 50000 muC 0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1991 Source: Development Planning Unit, Understanding Slums: Case Studies For The Global Report 2003. 10.40 Table 10.8 provides an estimate of informal housing stock for both 1991 and 2000 which is based on access to adequate infrastructure. The table enumerates formal and informal housing stock for 1991 and 2000 and it provides estimates of the net flow of formal and informal dwelling units for the 10 largest metropolitan areas in Brazil and other urban areas. The overall portion of informal units has increased from 13 to 23 percent. In some cities, Brazilia, Belem and Recife, the portion of informal units has doubled. In others--Curitiba, Salvador, and Sao Paulo it has remained constant. However, experts familiar with Salvador indicate that the ratio of unserviced informal housing is grossly underestimated. [comment by Ivo Imparato at World Bank Seminar on March 6, 2006]. 10.41 These data provide a rough estimate of relative contribution of formal and informal housing production in Brazil's urban areas between 1991 and 2000. The most important result of the tabulations presented in Table 10.8 is that the informal sector accounted for over half--56 percent of the increase in Brazil's urban housing stock between 1991 and 2000. Out of the total 10 million units increase in permanent dwelling units between 1991 and 2000, informal production accounted for 5.6 million units. 10.42 Table 10.8 also suggests that informality is now more prevalent outside the 10 largest metropolitan regions. In 1991, informal housing accounted for 13.7 percent of the total housing stock outside the 10 largest metropolitan areas in Brazil. In 2000, the figure increased to 24.1 percent. By 2000, 22.9 percent of the urban housing stock in Brazil could be classified as informal (lacking access to infrastructure). 10.43 Looking at the net flow of informal housing production between 1991 and 2000, in the 10 largest metropolitan regions, informal unit change accounted for 43.1 percent of the total increase. Put another way, between 1991 and 2000, 4 out over every 10 units developed in the 10 metropolitan areas were without infrastructure access. In Brazil's smaller metropolitan areas and cities, informal production accounted for 63.7 percent of total net housing production. This indicates that informality is growing rapidly in small and medium sized cities--between 1991 and 2000, the portion of housing units lacking infrastructure increased from 14 to 26 Inputs to a Strategy for Brazilian Cities Page 272 percent. In 2000, Brazil's urban housing stock totaled 44.8 million units. Of these, 10.3 million units were informal lacking in access to infrastructure. Table 10.8 Total Dwelling Units and Those Lacking Adequate Infrastructure 1991 1991 1991 2000 2000 2000 Informal Total Informal Percent Total Informal Percent Increase Metropolitan Region Permanent Dwelling Of Permanent Dwelling of as a % of Total Dwellings* Units** Total Dwellings* Units*** Total Increase Belem 274,186 38,386 14.0% 416,176 193,271 46.4% 109.1% Fortaleza 479,852 146,355 30.5% 723,197 333,262 46.1% 76.8% Recife 605,880 181,764 30.0% 859,574 459,352 53.4% 109.4% Salvador 547,678 124,323 22.7% 796,200 180,904 22.7% 22.8% Belo Horizonte 822,147 229,379 27.9% 1,295,824 214,114 16.5% -3.2% Rio de Janiero 2,753,543 273,669 9.9% 3,252,659 654,324 20.1% 76.3% Sao Paulo 3,967,579 273,669 6.9% 4,992,570 571,466 11.4% 29.1% Curitiba 508,699 72,744 14.3% 776,060 108,938 14.0% 13.5% Porto Alegre 840,660 81,544 9.7% 1,112,752 162,856 14.6% 29.9% Brasilia 363,222 6,538 1.8% 777,473 205,787 26.5% 48.1% Total Metropolitan Regions 11,163,447 1,428,371 12.8% 15,002,485 3,084,274 20.6% 43.1% Other Metropolitan Regions 23,571,268 3,224,240 13.7% 29,774,255 7,176,802 24.1% 63.7% Total Urban Brazil 34,734,715 4,652,611 13.4% 44,776,740 10,261,076 22.9% 55.8% Source: * Census Table 2432; **Fundação João Pinheiro (FJP), Centro de Estatística e Informações (CEI), Table 10.4, 2002. ***Fundação João Pinheiro (FJP), Centro de Estatística e Informações (CEI). Déficit Habitacional no Brasil ­ Municípios Selecionados e Microrregioes Geográficas, 2005. 10.44 Compared to other Latin American countries, Brazil ranks poorly in terms of access to infrastructure. According to a survey by the United Nations Economic Commission of Latin America and the Caribbean [2004] it ranked 8th out of 13 in terms of the percent of dwelling units with access to piped water, ranked 11th out of 13 with respect to sewerage collection and treatment connections, and ranked 5th out of 14 with respect to access to electricity.139 These are not impressive standings, and they reflect the limited options open to low and medium income households to secure shelter. 10.45 Despite high levels of private investment in residential construction, urban housing production in Brazil is predominantly based on informal housing construction. Based on available data, more than half--56 percent of the housing stock increase between 1991 and 2000 was informally provided (see Table 10.8). This is largely a reflection of the failure of formal urban housing and land markets to generate sufficient supply at affordable prices. However, informality is not simply a manifestation of low incomes. As Figure 10.8 illustrates, levels of informality are not highly correlated with incomes. Informality varies considerably within a narrow range of metropolitan areas with GDPs between Reais 4,000 to 6,000. 139With a per capita GNI of $3000, Brazil ranks below, Mexico, Argentina, Chile, Uruguay and these countries score higher on infrastructure access. However, some lower income countries such as Honduras and Guatemala, El Salvador, and Nicaragua score higher than Brazil on water and sanitation. Inputs to a Strategy for Brazilian Cities Page 273 Figure 10.8 Low-Income Does Not Entirely Explain Informality 70 60 50 40 30 20 10 0 0 2000 4000 6000 8000 10000 12000 14000 16000 GDP Per Capit a, 2 0 0 0 Source: FJP, 2005. 10.46 The most important obstacles to increased supply are lack of serviced subdivided land. Public infrastructure services are not expanding fast enough to meet housing production and there are over 10 million units that do not have access to adequate infrastructure. Figure 10.9 illustrates that public sector investment in infrastructure has not kept pace with housing production. Public sector gross fixed capital formation has lagged behind. As a consequence, much of Brazil's housing production is delivered without the support of public infrastructure services. Figure 10.9 Trends in Public Sector Gross Fixed Capital Formation 5.00000 4.50000 4.00000 3.50000 00 3.00000 1970=1. 2.50000 dexnI 2.00000 1.50000 1.00000 0.50000 - Year Public Sector GFCF Private Residential GFCF Source: Suzigan, W. A indústria brasileira : origem e desenvolvimento. São Paulo: Brasileiense, 1986; Abreu, M. de P. ; Verner, D. Long-term brazilian economic growth 1930-1994. Paris: OECD, 1997. (Development. Centre Studies. Long-term growth series/OCDE); IBGE, Diretoria de Pesquisas, Departamento de Contas Nacionais. Inputs to a Strategy for Brazilian Cities Page 274 10.47 If present trends continue, Brazil's urban housing stock will become increasingly dominated by informal production. While there will be some modest increase in slum upgrading and regularization that will move informal units into the formal category, it is quite likely that the overall proportion of informal urban dwelling units in Brazil will increase over the next several decades. In fact, if the trends in informal and formal housing production that took place between 1991 and 2000 continue, the Brazil's urban informal housing stock can be expected to increase to 35 percent overall, by 2030. 10.48 One of the most significant consequence of urbanization and housing construction is the spatial development of cities. Overtime as cities grow and expand, their spatial structure changes [Angel, et. al., 2005]. Motorization and increasing use of automobiles, is now one of the principal factors driving low density metropolitan development. As the next section illustrates, Brazilian cities are decentralizing and consuming more land per persons added. The Urban Land use Consequences of Urbanization 10.49 Brazil's rapid urbanization has profoundly shaped the physical development of its cities and metropolitan regions. Since, urban population growth must be supported by urban land, as cities grow, their urban areas (built up areas) increase in size. Table 10.9 provides summary statistics on the built-up areas and population densities for selected Brazilian and Latin American cities. As the table illustrates, gross population densities in Latin American cities range from 35 persons per hectare in Curitiba to a high of 101 persons per hectare in Rio de Janiero. 10.50 The urban development challenges posed by increasing urban population growth are substantial. Additional population requires additional housing stock, water supply and wastewater treatment, solid waste collection, schools, health facilities, streets, transport and employment opportunities. All of this requires land to support such development. In fact, the supply of serviced land is one of the principal determinants of urban land market performance. When the supply of serviced land, is limited, urban land prices are typically high relative to income and economic activity. This makes housing and non-residential real estate more expensive. Figure 10 provides a tabulation of land prices relative to GDP per capita in three Brazilian cities Brasilia, Curitiba and Recife. As it illustrates, in all three cities, the price of 100 square meters of serviced residential land roughly equals the per capita GDP of the metropolitan area. Inputs to a Strategy for Brazilian Cities Page 275 Table 10.9 Population, Urban land Use and Gross Population Density in Latin American Cities City Year Population Urban Land Gross Source Use, Hectares population density/ urbanized hectare Bogota 1990 5,484,200 158,700 34.6 Brinkhoff, 2003 Brasilia 2000 2,403,000 61,648 39.00 Serra, Dowall, Motta, Donovan 2005 Buenos 1990 7,974,000 115,700 68.9 Alain Bertaud, 20004 Aires Caracas 1990 1,822,465 43,300 42.1 Brinkhoff, 2003 Curitiba 2000 2,594,000 109,629 23.7 Serra, Dowall, Motta, Donovan 2005 Mexico City 1990 8,235,700 149,900 54.9 Brinkhoff, 2003 Recife 2000 3,339,000 88.6 Serra, Dowall, Motta, 37,669 Donovan 2005 Rio de 1990 5,480,800 54,265 101.0 Alain Bertaud Janeiro Santiago 1990 4,518,100 55,700 81.1 Simmonds and Hack, 2000 Sao Paulo 1990 15,416,400 203,800 75.7 Simmonds and Hack, 2000 Sources: http://alain-bertaud.com/, Thomas Brinkhoff, http://www.citypopulation.de/index.html; M.V. Serra, David E. Dowall, Diana Motta and Michael Donovan, Urban Land Markets and Urban Development: An Examination of Three Brazilian Cities: Brasilia, Curitiba and Recife. Brasilia: IPEA, 2005; and Roger Simmonds and Gary Hack, Global City Regions: Their Emerging Forms, London: Spon, 2000. Figure 10.10 Residential Land is Expensive Relative to GDP per Capita 16000 14000 12000 10000 GDP per capita Reais 8000 6000 Price of 100 Meter Plot 4000 2000 0 Brasilia Curitiba Recife Source: Serra, Dowall, Motta and Donovan, 2005. Inputs to a Strategy for Brazilian Cities Page 276 10.51 Household earning incomes below the GDP average are forced out of the formal market and must seek shelter in informal settlements, and generate overcrowding as households share dwellings. It is no coincidence that informal housing production, despite rigorous enforcement in the center of Brasilia is higher than in Curitiba. In the case of Recife, the very high rates of informality are due to both affordability gaps and limited land for residential development [Serra, Dowall, Mota, Donovan, 2004]. 10.52 Recent research on land markets in Brasilia, Curitiba, Recife, and Sao Paulo provides some indication of the relationship between population growth and urban land development. Table 10.10 presents data on these patterns for the four metropolitan areas. Using population and land use data from 1991 and 2000, the table illustrates the clear and direct relationship between population growth and urban land development. Depending on the metropolitan region, each additional 1000 person increase in population requires between 6 and 37 hectares of land to be developed. The amount of land that is needed depends on a range of factors such as the population per household, the density of residential development (houses per hectare), the extent to which new population in accommodated through urban redevelopment of older buildings and the additional demand for urban development that comes from nonresidential uses such as commercial and industrial activities. In the cases of both Recife and Sao Paulo, development is taking place at higher population densities. This is most likely due to denser residential development whether formal or informal. However, over time the overall density of metropolitan areas declines. Table 10.10 Trends in Population and Built up Area, Selected Brazilian Cities, 1991 and 2000 Metro Area 1991 2000 1991 2000 Change in Change Hectares per Population Population Built-up Built-up Population in Built- 1000 Area Area up Area population hectares hectares Hectares increase Brasilia 1,592,000 2,403,000 40,213 61,648 811,000 21,435 26.4 Curitiba 2,051,000 2,594,000 89,659 109,629 543,000 19,970 36.8 Recife 2,917,000 3,339,000 31,559 37,669 422,000 6,110 14.5 Sao Paulo 10,730,000 15,416,000 126,350 155,430 4,686,000 29,076 6.2 Total/ Average 17,290,000 23,752,000 287,781 364,376 6,462,000 76,591 11.9 10.53 In this section, we examine the spatial structure of the three cities--Brasilia, Curitiba and Recife, looking at the distribution of population and the compactness of urban land development. Examination of the spatial distribution of population in the three cities provides the opportunity to compare and contrast the overall compactness of urban development. We measure compactness by calculating the cumulative percentage of total population located within specific radii of the city center. Compactness will change over time depending on the spatial distribution of residential development taking place between 1991 and 2000. 10.54 Figure 10.11arrays the spatial distribution of population change for the three cities between 1991­2000 according to seven distance bands, expressed in terms of distance (kilometers) from the city center. In order to foster comparison, the bands are defined to reflect the overall spatial distribution of the three cities. Inputs to a Strategy for Brazilian Cities Page 277 Figure 10.11 Spatial Distribution of Population Change: Brasília, Curitiba and Recife, 1991 - 2000 50.0 45.0 40.0 35.0 egat 30.0 Brasília cenreP 25.0 Curitiba 20.0 Recife 15.0 10.0 5.0 0.0 0-5 5.1-10 10.1-15 15.1-20 20.1-25 25.1-30 30+ Distance (km) 10.55 Change in population between 1991 and 2000 reveals several interesting results. The first and most dramatic finding is that Brasília's population is distributed quite differently than Curitiba's and Recife's--most of its population is concentrated far from the city center. In 1991, over half (53.6%) of Brasília's metropolitan population was located more than 25 kilometers from the city. By 2000, the percentage had declined somewhat, to 50%, but still remained distinctly different from the spatial patterns in the other two cities. The percentage of population located within 10 kilometers of Brasília's center averaged about 8% for both 1991 and 2000. 10.56 In sharp contrast, in 1991 nearly 70% of Curitiba's population resided within 10 kilometers of the city center. By 2000, Curitiba's population had begun to decentralize and 58.5% of the total metropolitan population was located within 10 kilometers of the center. Peripheral population in Curitiba was low in comparison to Brasília--less than 6% in 1991 and less than 9% in 2000 of the total population residing more than 25 kilometers from the central city. 10.57 In Recife, the patterns are similar to Curitiba. In 1991, over 48% of the population resided within 10 kilometers of the city center. In 2000, the portion was 44%. Recife's peripheral population was about the same as Curitiba's and well below that of Brasília. In 1991, 8.5% lived more than 25 kilometers from the city center. In 2000, the figure increased to 9.2%. 10.58 The spatial distribution of population in the three cities between 1991 and 2000 largely reflected the baseline spatial structure of 1991. In Brasília, about half of the population growth took place in areas more than 25 kilometers from the center. It is significant to note that approximately 27% of the population change took place in the distance band of 20.1­25 kilometers--reflecting the growth in the area northeast of the city center. This decentralized, sprawling pattern of population change in Brasilia suggests that planning restrictions and government ownership of land introduces profound distortions into Brasilia's urban land market. Since development is blocked in areas adjacent to the city center, residential growth is forced to the periphery. It is interesting to contrast this with both Curitiba and Recife, where land use regulations are far less stringent. 10.59 In Curitiba, population growth moved out beyond 10 kilometers from the city center. Between 1991 and 2000, nearly half of the increase took place in areas between 10 and 20 kilometers from the city. This suggests that Curitiba has been relatively successful in achieving compact development--channeling growth into areas that are contiguous to existing urban areas. Compact development is not necessarily high density. In the case of Inputs to a Strategy for Brazilian Cities Page 278 Curitiba, the city used 37 hectares of land for each additional 1000 persons--this is much more land than in Brasilia, which used 26 hectares. 10.60 In Recife, approximately 58% of the increase in population between 1991 and 2000 occurred between 10.1 and 20.1 kilometers from the city center. Like Curitiba, Recife's growth has been compact, moving out beyond the densely developed core. But unlike them, it is developing at a much higher density--it used about 15 hectares per 1000 increase in the population. 10.61 Figure 10.12illustrates the change in urban developed land between 1991 and 1997/2000 for the three cities. In the core of Brasília (within 5 kilometers), less than 10% of the total urban land area is developed.140 In contrast, over 90% of the land in the core of Curitiba is developed. In Recife, about 80% of its developable core is urbanized. In Brasília, net new urban development in the core--conversion of vacant land to urban uses--is effectively zero (1 hectare). In Curitiba, net urban development in the core increased by 14 hectares, and in Recife, was the greatest increase at 48 hectares. 10.62 As far as urban land development beyond the core, Curitiba's and Recife's urban development is concentrated in the 10- to 25-kilometer bands. Between 1991 and 2000, 81% of Curitiba's change in developed, urbanized land was located in this 10­25 kilometer band. In Recife, 73% was similarly located. In contrast, in Brasília, less than 50% was located within 10 to 25 kilometers. In fact, approximately 53% of urban land development in Brasília between 1991 and 1997 took place beyond 25 kilometers from the city center-- suggesting that Brasília is sprawling. 10.63 What are the implications of these alternative forms of urban land development in the three cities? There are three important issues that emerge from this comparison. First, cities that sprawl--such as Brasília-- consume more land per person than those that develop compactly. Brasília developed 19,620 hectares of land to accommodate 811,000 persons--24 hectares per 1,000 additional persons. In contrast, Recife developed 6,738 hectares of land to accommodate 422,000 additional persons--16 hectares of land per 1,000 persons. However, Curitiba developed 19,220 hectares of land to accommodate 543,000 additional persons--35 hectares of land per 1,000 persons suggesting that Curitiba experienced substantial low-density development. Figure 10.12 Spatial Distribution of Change in Urban Land Development: Brasilia, Curitiva and Recife, 1991-1997/2000 50.0 45.0 40.0 35.0 30.0 Brasília 25.0 centagereP Curitiba 20.0 Recife 15.0 10.0 5.0 0.0 -5.0 0-5 5.1-10 10.1-15 15.1-20 20.1-25 25.1-30 30+ Distance (km) 140 The total area of the core is 7,850 hectares--*radius2. Inputs to a Strategy for Brazilian Cities Page 279 10.64 A second factor is the welfare implications of forcing population to travel greater distances to the center of the city. As Bertaud and Buckley have suggested for India, low-density urban sprawl introduces significant transportation costs on residents. A good comparative measure of compactness is the average per capita distance from the city center [Bertaud 2001]. This is calculated as the weighted average distance of each population in each zone. In 2001, the average per capita distance for Brasília was 24.3 kilometers; for Curitiba it was 11.2 kilometers; and for Recife it was 13.1 kilometers. In all cases, the average per capita distance to the city center increased between 1991 and 2001. In 1991, Brasília's average was 22.5 kilometers, Curitiba's was 9.75 kilometers, and Recife's was 12.62 kilometers. In a recent paper, Bertaud and Bruckner [2004] illustrated that cities with restrictive development controls take up more space and have higher commuting costs. Given the fact that distances are approximately twice as great in Brasília than in Curitiba or Recife, there is clearly a compelling case for assessing the welfare implications of the capital's dispersed spatial structure.141 10.65 The third impact is that more compact development economizes on urban infrastructure costs, whereas low-density sprawling development typically requires higher infrastructure costs per capita.142 The experience in Curitiba and Recife is consistent with empirical research on patterns of population density in Latin America and worldwide show that over time, population densities decline. As Ingram points out: Over time, a universal finding is that metropolitan populations have become more decentralized (population density gradients become flatter)--due to the effects of increases in income (promoting housing consumption) and improvements in transport performance (higher speeds and lower costs relative to incomes). Population growth in large cities usually does not increase the population density of high density areas, but promotes densification of less-developed areas and expansion at the urban fringe [Ingram, 1998 pp. 1021-2]. 10.66 Density gradients measure the relationship between population density and distance from the city center. Normally, as cities expand, population density gradients "flatten out" as people move to suburban rings of the metropolitan area to find housing [Mills 1972]. This flattening out is the result of two changes in the gradient-- first, the population at the center declines, and second, there is a decline in the rate at which population density falls with distance from the city center. Empirical research has shown that the following simple exponential function provides a reasonable basis for describing the pattern of declining population density in metropolitan areas: Dx = D0e-gx where Dx is the population density at x kilometers from the city center, D0 is the population density at the center of the city, and g is a population density gradient parameter to be estimated from the data. 10.67 Table 10.11 presents the results of separate regression models estimating the population density gradients for a range of Brazilian cities. Intercept data and gradients are presented for two time periods. In all cases, the gradients "flatten out" over time. With the exception of Recife, the intercept population density (the estimated population density in the city center) decreases over time, suggesting that residential occupancy decreases in the center--perhaps signaling conversion to non residential uses or residential population shifts to newer outlying areas. The increase in central city population in Recife, although modest, may suggest that the preservation of high-density favelas in ZEIS areas near the city center is an effective means for preserving residential areas in central cities. 141 In fact, average distance per capita figures for other national capitals, such as Moscow (10.57 km), Paris (10.24 km), and London (12.63 km), are less than half of Brasília's despite the fact that they have larger populations. 142See Robert W. Burchell et al. The Costs of Sprawl. TCRP REPORT 74. New Brunswick New Jersey: Center for Urban Policy Research, 2000. Inputs to a Strategy for Brazilian Cities Page 280 10.68 The flattening out of population density gradients, has important implications for urban land management. As cities grow, the amount of land supply needed per person will increase. Therefore, looking toward the future, cities in Brazil will spatially expand, as densities decrease. This increase in urban population will generate considerable demand for urban land and infrastructure services. 10.69 Sprawl also poses a major challenge for metropolitan management and planning institutions. If the population growth of Brazil's largest metropolitan areas is spilling over into outlying municipalities, central city governments like Rio de Janerio and Sao Paulo are losing their control of spatial development policies and infrastructure investment decisions. Table 10.11 Population Density Gradients in Selected Brazilian Cities 1991 and 2000 Population Density Gradients in Selected Brazílian Cities, 1991 and 2000 CITY YEAR INTERCEPT Source (D0)* GRADIENT (g) Belo 1991 122 -0.082 Avila Horizonte 2000 113 -0.052 Curitiba 1991 140 -0.201 Dowall 2000 124 -0.166 Fortaleza 1991 206 -0.166 Avila 2000 171 -0.108 Porto Alegre 1991 166 -0.187 Avila 2000 158 -0.168 Recife 1991 165 -0.076 Dowall 2000 179 -0.073 Rio de 1991 169 -0.040 Avila Janiero 2000 148 -0.029 Salvador 1991 219 -0.146 Avila 2001 198 -0.100 Sao Paulo 1991 200 -0.073 Avila 2000 154 -0.049 * Density is persons per hectare Source: Dowall, 2004 and Avila, 2005. Looking Forward: Brazil's Future Urban Housing Needs and Prospects for Reaching them? 10.70 Projections of future urban population growth for Brazil suggest robust growth [UN ECLAC, 2004]. As illustrated in Table 10.12, between 2000 and 2030 Brazil's total population is projected to increase by 65,961,000, reaching 235,505,000. All of this increase will occur in urban areas, as rural hinterlands are expected to continue losing population. Total urban population will increase from 138,000,000 in 2000 to 215,000,000 in 2030 an increase of 77,000,000--this is like adding 7 Rio de Janiero's over the 30 year period. On an annual basis the increase in urban population will average over 2,500,000 persons per year--almost like adding a Curitiba each year. These are huge numbers that imply massive challenges for city planning and public sector capital investment programming. Inputs to a Strategy for Brazilian Cities Page 281 Table 10.12 Projections of Brazil's Total, Urban and Rural Population 2000 - 2030 Population Year Total Urban Rural 2000 169,544,443 137,697,439 31,847,004 2005 186,405,000 157,041,000 29,364,000 2010 198,497,000 171,904,000 26,593,000 2015 209,401,000 185,052,000 24,349,000 2020 219,193,000 196,573,000 22,620,000 2025 227,930,000 206,557,000 21,373,000 2030 235,505,000 214,940,000 20,565,000 Annual Percent Change Year Total Urban Rural 2000-05 2.0% 2.8% -1.6% 2005-10 1.3% 1.9% -1.9% 2010-15 1.1% 1.5% -1.7% 2015-20 0.9% 1.2% -1.4% 2020-25 0.8% 1.0% -1.1% 2025-30 0.7% 0.8% -0.8% Source ECLAC, United Nations 2004 Accommodating Urban Growth: How much Urban Land Supply is Needed? 10.71 We can roughly approximate the urban land supply requirements to accommodate future urban population growth in Brazil. Estimates are based on combinations of Table 10.10 and Table 10.12, using the overall average 11.9 hectares of built up area to support a 1000 person increase in urban population, then the total urban land requirements to accommodate 77 million persons is approximately 916,300 hectares or 9,163 square kilometers. Put another way, accommodating this urban population grow will require a built up area equivalent to 7 Sao Paulos. 10.72 Of course, this estimate is speculative. It may be possible to accommodate the population growth at higher densities, by redeveloping inner city areas with housing, and by increasing the density of suburban development [Dowall and Treffeisen, 1991]. By shifting away from single family dwelling units (in both formal and informal settlements) to mid rise condominiums and more compact low-rise residential development, per capita urban land requirements can be reduced.143 For example if the urban land supply requirements per 1000 person could be reduced by about 25 percent only 9 hectares of urban land would be required for each 1000 persons (111 persons per hectare. This would reduce the aggregate land supply requirement to 693,000 hectares-- 6,930 square kilometers. However, increasing density will make it more difficult for the informal sector to operate since higher density multifamily units will be needed. In order for this approach to work--such housing must be affordable to low and moderate income households. This suggests that the government should concentrate its efforts on providing urban infrastructure to land suitable for development. 143See Burchell, et. al. Ibid. Inputs to a Strategy for Brazilian Cities Page 282 What can be done to improve Urban Land and Housing Market Outcomes? 10.73 The Government of Brazil, in partnership with local governments, non-governmental organizations and the private sector, could do much to foster increased production of affordable housing. This section outlines what such a strategy might look like. 10.74 First and foremost, the urban land and housing strategy should be multi-faceted and similar to policy models used by public health professionals--it should include both "curative" and "preventive" programs. The curative aspects of the strategy would focus on upgrading and improving housing conditions in informal areas. Preventive strategies should be implemented to reduce the growth of informal areas--this requires opening up more land for residential development, providing public infrastructure and facilities, and creating incentives for the provision of low and moderate income housing. Both approaches are needed. On its own, the curative approach will not succeed. While existing favelas and irregular settlements can be upgraded, this approach does not prevent the formation of new informal settlements--these will continue to expand as long as urban land and housing markets fail to produce affordable housing. 10.75 Effective upgrading programs should include community participation, provide secure land tenure, and give access to critical residential infrastructure--water, wastewater collection and treatment, drainage, electricity, schools and clinics as well as parks and recreation facilities. Large-scale programs such as Sao Paulo's Guarapiranga project have been largely successful and provide useful models for replication [City of Sao Paulo, 2000]. However, due to their complexity, they are difficult to implement and replicate [Cohen, 1983]. This suggests that more work is needed to design more efficient and simpler procedures as well as generating more professional expertise about upgrading. 10.76 Preventing the continued expansion of informal housing requires that Brazil's urban land and housing markets start producing more housing and providing more affordable housing that is located within reasonable commuting distances to jobs. If this can be accomplished, then the demand for informal housing should decline as households shift to less expensive formal housing. What would it take to achieve such a result? 10.77 First, cities and metropolitan regions need to better understand how their land and housing markets operate. Urban planners, housing specialists and policy makers need better empirical data on urban land and housing markets--both current demand and supply information on land and housing prices and projections of future housing and urban land requirements to accommodate demographic and economic growth [Dowall and Clarke, 1991]. 10.78 Second, these data and projections should be used to prepare master plans for cities and metropolitan regions. The plans should ensure that adequate supplies of serviced urban land are available to support residential demand. This will require pro-poor land use plans and zoning regulations [UN Habitat, 2004]. Lands should be targeted for residential development and tax incentives should be used to encourage owners to bring land to the market for residential development. Governments will need to provide the funding for infrastructure provision so that developers will be encouraged to construct housing. 10.79 Third, massive investments in private infrastructure are needed to foster residential subdivision development. The Government of Brazil and State and local governments need to develop more fiscal resources to finance infrastructure. This can be accomplished through a range of policy interventions including, levying user and beneficiary charges, and implementing value capture programs as outlined by Furtado and Jorgensen [2006]. 10.80 Fourth, land subdivision and building regulations should be reviewed to assess their impacts on housing costs. Subdivision standards frequently impose excessive standards on developers--large minimum lot sizes, high land dedication requirements and investments in non-essential infrastructure [Avila, 2006]. Building codes often prove costly and impose too much of a burden on low and moderate income households [Dowall, 1992]. One interesting model is Colombia's "minimum norms" for low income settlements [Carroll, 1980]. Another possibility is to create a zoning classification that permits the development of sites and services projects--this Inputs to a Strategy for Brazilian Cities Page 283 would in effect legalize irregular settlements, if they met basic standards for circulation, plot size and layout [UN Habitat, 2004]. 10.81 Fifth, the government needs to develop cost effective and replicable models for land titling and registration. These issues and policy reforms are comprehensively outlined by Fernandes [2006]. 10.82 Taken together as a package, these five initiatives could foster increased affordable land and housing production. To launch this effort, the central government needs to articulate a policy framework and then to collaborate with local governments to design and implement plans and programs. Over time, the framework as well as specific programs should be evaluated and modifications should be made as necessary. References Angel, S., Sheppard, S., & Civco, D. [2005]. The Dynamics of Global Urban Expansion. Washington, DC: World Bank. Avila, P. C. "Urban Land Use Regulations in Brazil: Land Market Impacts and Access to Housing." Lincoln Institute-World Bank, 2006. Avila, P. C. & Mandell, P. [2005]. Interpretation of Population Density Gradients: A Brazilian Perspective. Third Urban Research Symposium on "Land Development, Urban Policy and Poverty Reduction," Brasilia. Bertaud, A. [2001] Metropolis: A Measure of the Spatial Organization of 7 Large Cities. Accessed from website: http://alain-bertaud.com/ Bertaud, A. Robert Buckley and Kate Owens [2003], Is Indian Urban Policy Impoverishing? Washington: World Bank: Urban Research Symposium. Bertaud, A. and Jan Brueckner, [2003] "Analyzing Building-Height Restrictions: Predicted Impacts, Welfare Costs, and a Case Study of Bangalore, India." Accessed from website: http://alain-bertaud.com/ Brinkhoff,Thomas http://www.citypopulation.de/index.html; Burchell, Robert, et. al. [2000] The Costs of Sprawl. TCRP Report 74. New Brunswick New Jersey: Center for Urban Policy Research. Carroll, A. [1980]. Pirate Subdivisions and the Market for Residential Lots in Bogotá [Working Paper No. 435]. Washington, DC: World Bank. City of Sao Paulo, [2000] Guarapiranga: Urban and Environmental Rehabilitation in the City of Sao Paulo. Cohen, Michael, [1983] "The Challenge of Replicability: Towards a New Paradign for Urban Shelter in Developing Countries," Regional Development Dialogue, Vol. 4, No. 1, Spring. Development Planning Unit, Understanding Slums: Case Studies For The Global Report 2003. Inputs to a Strategy for Brazilian Cities Page 284 Dowall, D. E. [1992] "The Benefits of Minimal Land Use Regulations in Developing Countries," Cato Journal, Vol. 12, No. 2, 1992, pp. 413-423. Dowall, D. E. [1992]"A Second Look at the Bangkok Land and Housing Market," Urban Studies, Vol. 29, No. 1, 1992, pp. 25-38. Dowall, D.E. [1989] "Bangkok: A Profile of an Efficiently Performing Housing Market, Urban Studies, Vol. 26, No. 3, 1989, pp. 327-339. Dowall, D., & Clarke, G. [1991]. A Framework for Reforming Urban Land Policies in Developing Countries [Policy Paper]. Washington, DC: Urban Management Program. Dowall, D. & Treffeisen [1991]. Spatial Transformation in Cities of the Developing World: Multinucleation and land-capital substitution in Bogotá, Colombia. Regional Science and Urban Economics, 21, 201-224. Economic Commission for Latin America and the Caribbean [2004]. Progress of the Latin American And the Caribbean Region towards Sustainable Development in the Issues: Human Settlements, Sanitation, Water. Regional Preparatory Process for the Twelfth Session of the Commission on Sustainable Development [unpublished]. Fernandes, Edesio "Legal Aspects of Urban Land Development in Brazil." Lincoln Institutee-World Bank, 2006 Fundação João Pinheiro [2002]. Déficit Habitacional no Brasil 2000. Informativo CEI ­ Demografia. Belo Horizonte, Brazil: Government of Minas Gerais. Fundação João Pinheiro [2005]. Déficit Habitacional no Brasil: Municípios Selecionados e Microregiões Geográficas, 2a Edição. Belo Horizonte, Brazil: Government of Minas Gerais. Furtado, Fernanda and Pedro Jorgensen, "Value Capture in Brasil: Issues and Opportunities." ." Lincoln Institutee-World Bank, 2006 Ingram, G. [1998]. Patterns of Metropolitan Development: What Have We Learned? Urban Studies, 35, 1019-1035. Instituto Polis [2001] The Statute of the City. Nairobi: UN-Habitat. Macedo, J., O'Dell, W., & Smith, M. [2005]. Affordable housing needs assessment methodology: The adaptation of the Florida model to Brazil. In M.V. Serra & D. Motta (Eds.), Enabling Strategy for Moving Upgrading to Scale in Brazil. Washington, DC: Cities Alliance. O'Hare, Greg and Michael Barke, [2003] "The Favelas of Rio de Janiero: A Temporal and Spatial Analysis," Geoforum V. 56, pp. 225-240. Pontual, R. [2005]. Sistema estatístico de informações habitacionais: Proposta para criação de um sistema nacional de informações habitacionais. In M.V. Serra & D. Motta (Eds.), Enabling Strategy for Moving Upgrading to Scale in Brazil. Washington, DC: Cities Alliance. Inputs to a Strategy for Brazilian Cities Page 285 Pontual, R. & Serra, M. V. [2005]. Análise da habitação informal: A Evolução da habitação informal em três metrópoles do Brasil ­ Recife, Curitiba, e Distrito Federal. In M.V. Serra & D. Motta (Eds.), Enabling Strategy for Moving Upgrading to Scale in Brazil. Washington, DC: Cities Alliance. Serra, M. V., Dowall, D., Motta, D., Donova, M. [2005]. Urban Land Markets and Urban Land Development: An Examination of Three Brazilian Cities: Brasília, Curitiba and Recife [Working Paper No. 1082]. Brasília: Instituto de Pesquisa Econômica Aplicada. Simmonds, Roger and Gary Hack, [2000] Global City Regions: Their Emerging Forms, London: Spon. United Nations-Habitat [2004] Pro-Poor Land Management. Nairobi. World Bank, [2000] World Development Report. Washington. World Bank [2005] Doing Business Survey. Washington. World Bank [2002] Urban Housing Sector Reform in Brazil: World Bank Policy Note, September, 2002. World Bank [2002] Brazil: Progressive Low-Income Housing: Alternatives for the Poor. Washington. World Bank [1989] Malaysia ­ The Housing Sector: Getting the Incentives Right. Washington.