35692 Authorized v 1 Disclosure Decentralized Public the poor Authorized Disclosure Public Authorized Disclosure Public Authorized Disclosure Public Decentralized the poor h service delivery for Portada_Falsa.indd 2 Portada_Falsa.indd 7/27/06 2:11:08 AM 7/27/06 2:11:08 AM Decentralized Service Delivery for the Poor Copyright © 2006 by The International Bank for Reconstruction and Development / The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All Rights Reserved Manufactured in Mexico City First Printing: August, 2006 The findings, interpretations, and conclusions expressed in this book are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. The World Bank encourages dissemination of its work and normally will promptly grant permission to reproduce portions of the work. For permission to reprint individual articles or chapters, please fax a request with complete information to Gabriela Aguilar, Communications Officer ­ The World Bank Colombia and Mexico Fax (55) 54804222. All other queries on rights and licenses should be addressed to the Communications Officer, The World Bank Mexico at the above faxed to (55) 54804222. Printing and Manufactured in Mexico City / 2006 The World Bank Mexico Edition: Marcela Rubio Sanchéz Design: Alejandro Espinosa Photographic: Adrian Mealand ; The World Bank's photo library The World Bank Mexico World Bank Decentralized Service Delivery for the Poor. ­ Mexico : The World Bank, 2006. 2 vol. il. ­ (Report No. 35692) Content: vol. 1. Core report. ­ vol. 2. Background papers 1. Education ­ Decentralization ­ Mexico. ­ 2. Health ­ Decentralization ­ Mexico. ­ 3. Infrastructure (Social) Decentralization ­ Mexico. ­ Rural Development Mexico 351.0972 / W67 / 2006 Preface The collaboration between the Government of Mexico (GOM) and the World Bank on poverty has four objectives: (i) to assist the GOM in improving the effectiveness of its programs in reducing poverty; (ii) to support the analysis and design (or redesign) of specific policy instruments; (iii) to build capacity, especially in evaluation techniques and processes; and (iv) through workshops, to share best practice in poverty reduction policies in a national and international context. The combination of demanddriven analytical work and capacity building is designed to strengthen the government's institutional capacity to implement policies that bring about effective poverty reduction results on the ground. The results of the first phase of the Programmatic Poverty Work of the World Bank in Mexico were published in 2004: Poverty in Mexico: an Assessment of Conditions, Trends, and Government Strategy (World Bank, 2004). The second report was published in 2005: Income Generation and Social Protection for the Poor, which summarized and consolidated the findings of three World Bank studies on poverty issues in Mexico written as part of the second phase work: Urban Poverty, Rural Poverty, and Social Protection. This report is structured in two Volumes. Volume I is a crosssectoral report-- the Core Report. The first part includes the program and performance overview which describes the programs (including their poverty goals) and their process of decentralization, the overview of expenditures, the access of the poor to these services, and the evidence on the quality of services. The second part of the Core Report is on the institutional dimensions of decentralized service delivery which includes the WDR 2004 type framework of accountability, adapted to the situation in Mexico. Volume II contains the sectoral pieces. The third phase of the Poverty Work was coordinated by Steven Webb and Gladys LópezAcevedo. Team members included Marcela Rubio, Mark V. Hagerstrom, Harry Patrinos, Vicente GarcíaMoreno, April Harding, María Luisa Escobar, Anna Wellenstein, Luis Andrés, Angélica Núñez, José María Caballero, David Rosenblatt, Christian González, and Odracir Barquera. Specific inputs were provided by Armando Arredondo, Jesús Álvarez, John Scott, Carlos Mancera, Michael Walton, and Carlos Moreno. Elizabeth Forsyth assisted with the editing. This report has greatly benefited from the comments received from the Social Cabinet, Presidencia, the Ministry of Finance, and the peer reviewers: Margaret Grosh (Lead Economist), Yasuhiko Matsuda (Senior Public Sector Specialist) Alain de Janvry (Professor from the University of California at Berkeley and Consultant), and Maria Emilia Freire (Regional Advisor). Participants at the review meetings provided comments, including Guillermo Perry, Humberto López, and Ariel Fiszbein. Special thanks to Michael Walton for his insightful comments to this report. Decentralized Service Delivery for the Poor i We would also like to acknowledge the valuable contributions made by our Government counterparts, especially: Antonio Ruiz (SAGARPA); Félix Vélez, Gustavo Merino, Miguel Angel López, and Germán Palafox (SEDESOL); Rafael Freyre, Felicia Knaul, and Silvia Schmelkes (SEP); Debora Schlam, María Cecilia Hernández, José Luis Meixueiro, Patricia Cravioto, and Miguel Angel González (SHCP); Tábata Vilar (Social Cabinet); Eduardo González Pier, Héctor Peña, and Rafael Lozano (SSA); and state government officials who provided their support. The Vice President for Latin America and the Caribbean is Pamela Cox; the Country Director for the Mexico and Colombia Unit is Isabel Guerrero; the Sector Director for Poverty Reduction and Economic Management in Latin America and the Caribbean is Ernesto May; Sector Leader for PREM in the Mexico and Colombia Unit is David Rosenblat; and the Task Managers are Steven Webb and Gladys LópezAcevedo. ii Decentralized Service Delivery for the Poor TABLE OF CONTENTS EXECUTIVE SUMMARY xi CHAPTER 1. Decentralization, Poverty, and Development in Mexico 1 I. Service Areas in this Report 2 II. International Experience with Decentralization of Services: From Grand 8 Hopes to Complex Realities III. Mexico's Historical Process of Decentralization 13 IV. Common Challenges across Sectors 18 CHAPTER 2. Fiscal Overview 19 I. Mexico's Decentralization Process and Normative Principles 21 II. Revenue and Expenditure Trends 23 III. Education 27 IV. Health 34 V. Social Infrastructure Fund (FAIS) 37 VI. Alianza Para el Campo: Rural Development Program 40 VII. Toward a New Fiscal Federalism that Will Improve the Access of the 42 Poor to Opportunities CHAPTER 3. Performance of the Sectors 45 I. Education 47 II. Health 57 III. Basic Infrastructure 69 IV. Rural Development 74 V. A Common Challenge for Evaluation and Accountability 82 ANNEX 3.1. Analysis of the Performance of the Sectors 84 ANNEX 3.2. Institutional Arrangements After Decentralization 86 CHAPTER 4. Institutional Dimensions of Decentralized Service Delivery 91 I. Accountability Frameworks Adapted to Mexico 92 II. Education 93 III. Health 99 IV. Social Infrastructure 108 V. Rural Development 114 VI. Conclusions 123 ANNEX. Background Notes on Expenditure Decentralization and Poverty 125 Alleviation in Mexico REFERENCES 151 Decentralized Service Delivery for the Poor iii LIST OF TABLES Table 2.1. Programs as a Share of the Federal Budget, 2004 20 Table 3.1. Summary--Performance of the Sectors 47 Table 3.2. Average Schooling of 2565 YearOlds, by Income Decile, 19842004 48 Table 3.3. Enrollment Rates in Basic Education, 2004 50 Table 3.4. State Actions as Determinants of Student Learning 57 Table 3.5. Maternal Mortality Rate in OECD Countries, Early 2000's 58 Table 3.6. Life Expectancy at Birth, 19902005 61 Table 3.7. Indicators of Quality, by Health System, 2003 62 Table 3.8. Perceived Quality of Care by Health Care Provider, 2000 62 Table 3.9. Access to Services, 19922004 69 Table 3.10. Access to Services, by Group of Municipalities, 2000 71 Table 3.11. FAIS Allocations, by Sector, 1999, 2001, and 2003 72 Table 3.12. Public Infrastructure Investment, by Sector, Excluding 74 Telecommunications, 2003 Table 3.13. Distribution of Rural Population, Beneficiaries, and Funds of the 77 Rural Development Program, by Marginality of Their Area of Residence Table 3.14. Characteristics of the Rural Development Program Beneficiaries, by 77 Type, 2004 LIST OF BOXES Box 1.1 Decentralization and Accountability for Results in Canada and Germany 10 Box 2.1 Formulas for Allocating Funds to the State and Municipal Governments 24 Box 2.2 Per Student Allocation Transfers in Colombia 33 Box 3.1 Implementing a Nationally Coordinated System for Citizen Identification 82 and Exchange of Fiscal and Social Information: Lessons from Argentina and Chile Box 4.1 Local Institutional Innovations in Education: The Case of Colima 95 Box 4.2 Local Institutional Innovations in Education: The Case of Aguascalientes 97 Box 4.3 International Experiences on School Autonomy 98 Box 4.4 Local Institutional Innovations in Health: Several States 101 Box 4.5 Separating Funders from Providers in Health: The International 106 Experience Box 4.6 Comparing Institutional Arrangements: The Case of Puebla and Veracruz 111 Box 4.8 State Experiences in the Implementation of Alianza 121 iv Decentralized Service Delivery for the Poor LIST OF FIGURES Figure 1. Accountability Relationships with Political Parties, National Provider xiv Organizations, and Subnational Governments Figure 2. Math and Reading Test Results and Expenditures per Student, 2001 xix Figure 1.1. CONTIGO Poverty Reduction Strategy 3 Figure 1.2. Map of Decentralization and Social Programs 4 Figure 1.3. Accountability Relationships among Citizens, Policymakers, and 12 Providers of Services Figure 1.4. Percent of Municipalities and States Governed by Parties Other than 15 the PRI, 19852003 Figure 2.1. Federal Government Expenditures, 2004 19 Figure 2.2. Concentration Coefficients, National, 2000 and 2002 21 Figure 2.3. States' Public Resources per Capita, 2004 26 Figure 2.4. States' Public Resources per Capita, 1992 26 Figure 2.5. Expenditure Trends in Education, Health, FAIS and Alianza 27 Figure 2.6. Education Expenditures, 19902004 29 Figure 2.7. Principal Federal Expenditures on Basic Education, 2004 30 Figure 2.8. State Expenditure from Own Resources on Basic Education as a 31 Percentage of Total Public Expenditure on Basic Education, Average 19982001 Figure 2.9. Public Expenditure per Student on Basic Education, 2002 31 Figure 2.10. Math and Reading Test Results and Expenditures per Student, 2001 32 Figure 2.11. Health Expenditures, 19982004 35 Figure 2.12. Distribution of Federal Health Resources per Population without 36 Social Security Coverage, 2004 Figure 2.13. Distribution of Health Sector Expenditures per Capita, 2004 37 Figure 2.14. FAIS Expenditures, 19982004 38 Figure 2.15. Distribution of FAIS per Capita, 2004 39 Figure 2.16. Federal Expenditures on Social Infrastructure 40 Figure 2.17. Decentralized Alianza Funds per Head of Rural Population, Federal 41 and State Contributions, Average 20012005 Figure 2.18. Federal Expenditures of Principal Programs in Agriculture, 1996 42 2004 Figure 3.1. Population Over 15 Years Old without a Lower Secondary Education, 48 2000 Figure 3.2. Coverage Ratio of Basic Education, 3 to 15 year olds, 19912004 49 Figure 3.3. Math Scores in PISA and Dispersion across Countries, 2003 51 Figure 3.4. Performance in Mathematics in PISA and GDP per Capita, 2003 52 Figure 3.5. Means of the Math Test Scores in PISA, by State, 2003 53 Figure 3.6. Test Results by Type of School, PISA 2003 54 Figure 3.7. Performance Advantage of Different Types of Schools Compared with 55 Telesecundaria Decentralized Service Delivery for the Poor v Figure 3.8. Life Expectancy and Mortality, 19902004 58 Figure 3.9. Infant Mortality Rate by Municipality, 2000 59 Figure 3.10. Share of Households with Catastrophic Health Expenditures, 1992 59 2004 Figure 3.11. Population by Institution of Affiliation, 2004 60 Figure 3.12. Population without Social Security Coverage, 2000 61 Figure 3.13. Health Indicators, 19902003 63 Figure 3.14. Health Indicators for Select States, 19902001 65 Figure 3.15. Share of Households with Catastrophic Health Expenditures in 66 Selected States, 2003 Figure 3.16. Distribution of Seguro Popular and Oportunidades Beneficiaries of SSA 67 Health Services and of the Uninsured Population Figure 3.17. Concentration Coefficients for Seguro Popular and Other Health 68 Programs and the Uninsured Population Figure 3.18. Households with Access to Piped Water, Sewage, and Electricity, 70 2000 Figure 3.19. Correlation of FISM Allocations and Basic Infrastructure, 1999, 2001, 73 and 2003 Figure 3.20. Agricultural Labor Productivity in Latin American and Caribbean 75 Countries, 2002 Figure 3.21. Public Expenditure in Agriculture per Agricultural Worker in LAC 75 Countries, Average 19962000 Figure 3.22. Distribution of 2004 Alianza Beneficiaries, by Program and Type of 78 Beneficiary Figure 3.23. Distribution of the Rural Development Program Funds and Type of 79 Beneficiary, 2004 Figure 3.24. Distribution of the Rural Development Program Beneficiaries and the 79 Rural Population (Families) Figure 3.25. Percentage Increase in Employment and Gross Income of Alianza 80 2002 Beneficiaries between 2002 and 2005 Figure 3.26. Percentage Increase in Assets and Technology Level of Alianza 2002 81 Beneficiaries between 2002 and 2005 Figure 4.1. Channels of Influence in the Mexican Education Sector 95 Figure 4.2. Channels of Influence in the Mexican Health Sector: SSA 100 Figure 4.3. Channels of Influence in the Mexican Health Sector: SSA with Seguro 103 Popular Figure 4.4. Channels of Influence in the Mexican Health Sector: IMSS 104 Oportunidades Figure 4.5. Channels of Influence in Mexico's Social Infrastructure: FAISFISM 114 Figure 4.6. Channels of Influence in Mexico's Rural Development: Alianza's Rural 117 Development Program vi Decentralized Service Delivery for the Poor Acronyms and Abbreviations AGEs Apoyos a la Gestión Escolar School Management Support Program CIDRS Comisión Intersecretarial para el Desarrollo Rural Sustentable InterSecretarial Commission for Sustainable Rural Development CDI Comisión Nacional para el Desarrollo de los Pueblos Indígenas National Commission for the Development of Indigenous Peoples CMDRS Consejos Municipales de Desarrollo Rural Sustentable Municipal Councils for Sustainable Rural Development CNA Comisión Nacional del Agua National Water Commission CONAFE Consejo Nacional de Fomento Educativo National Council for Educational Support CONAPO Consejo Nacional de Población National Population Council COPLADEMUN Consejos de Planeación Municipal Municipal Planning Councils DGE Dirección General de Evaluación General Direction for Evaluation ENHRUM Encuesta Nacional de Hogares Rurales de México National Rural Households Survey of Mexico ENIGH Encuesta Nacional de Ingresos y Gastos de los Hogares National Household Income and Expenditure Survey FAEB Fondo de Aportaciones para la Educación Básica Basic Education Fund FAETA Fondo de Aportaciones para la Educación Técnica y de Adultos Technical and Adult Education Fund Decentralized Service Delivery for the Poor vii FAIS Fondo de Aportaciones para la Infraestructura Social Social Infrastructure Fund FAM Fondo de Aportaciones Múltiples Multiple Uses Fund FAO Food and Agriculture Organization FASSA Fondo de Aportaciones para los Servicios de Salud Health Services Fund FISE Fondo para la Infraestructura Social Estatal State Social Infrastructure Fund FISM Fondo para la Infraestructura Social Municipal Municipal Social Infrastructure Fund GDP Gross domestic product IMSS Instituto Mexicano de Seguro Social Mexican Institute for Social Security INEE Instituto Nacional para la Evaluación de la Educación National Institute for Education Evaluation INEGI Instituto Nacional de Estadística, Geografía e Informática National Institute of Statistics, Geography, and Information ISSSTE Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado Security and Social Services Institute for the State Workers LCF Ley de Coordinación Fiscal Fiscal Coordination Law LDRS Ley de Desarrollo Rural Sustentable Law of Sustainable Rural Development OLS Ordinary least squares ORFIS Órgano de Fiscalización Superior Comptroller's Office viii Decentralized Service Delivery for the Poor PAN Partido Acción Nacional National Action Party PAFEF Programa de Apoyos para el Fortalecimiento de las Entidades Federativas Support Program for the Strengthening of the States PAPIR Programa de Apoyo a los Proyectos de Inversión Program for Investment Projects Support PEC Programa de Escuelas de Calidad Program of Schools of Quality PISA Program for International Student Assessment PRD Partido de la Revolución Democrática Democratic Revolution Party PRI Partido Revolucionario Institucional Institucional Revolutionary Party PRODESCA Programa de Desarrollo de Capacidades Program for the Development of Capacities PROFEMOR Programa de Fortalecimiento de Empresas y Organizaciones Rurales Program for the Strengthening of Firms and Rural Organization PRONASOL Programa Nacional de Solidaridad National Solidarity Program SAGARPA Secretaría de Agricultura, Ganadería, Desarrollo Rural, Pesca y Alimentación Ministry of Agriculture, Livestock, Rural Development, Fishery, and Alimentation SEDESOL Secretaría de Desarrollo Social Ministry of Social Development SEP Secretaría de Educación Pública Ministry of Public Education SHCP Secretaría de Hacienda y Crédito Público Ministry of Finance and Public Credit Decentralized Service Delivery for the Poor ix SNTE Sindicato Nacional de Trabajadores de la Educación Teachers' Union SRA Secretaría de la Reforma Agraria Ministry of the Agrarian Reform SSA Secretaría de Salud Ministry of Health x Decentralized Service Delivery for the Poor DECENTRALIZED SERVICE DELIVERY FOR THE POOR EXECUTIVE SUMMARY Mexico has reorganized many of its government programs over the last dozen years, among others, by decentralizing them to states and municipalities, although that process is far from complete. Simultaneously, Mexico has achieved major expansion in the coverage of these social programs, especially for the poor. Currently, the biggest challenge is to improve program quality and accountability, which remain low by Organization for Economic Cooperation and Development (OECD) standards and about average for Latin America. For this report--the third phase of a programmatic series of studies on poverty in Mexico--Mexico's Social Cabinet requested that the Bank analyze service delivery of a specific list of programs that are essential to the government's development strategy in the context of decentralization. The elements of the government's strategy considered here--basic education, health care for the population without social security coverage, local social infrastructure, and local rural development--all aim directly or indirectly to reduce poverty. Both social and economic services shape the opportunities of poor people. Social services contribute to human capital formation and risk management, while economic services shape access to markets. · Basic education includes preschool, primary school (grades 1­6), and lower secondary school (grades 7­9). Public schools account for 92 percent of enrollment in basic education. The Fondo de Aportaciones para la Educación Básica (FAEB, the Basic Education Fund), which is an earmarked transfer to the states, is the main source of funding for basic education. FAEB plus the federal programs considered in this report--the Programa de Escuelas de Calidad (PEC) and the Consejo Nacional de Fomento Educativo (CONAFE)-- account for more than 15 percent of total public spending. FAEB and state spending cover mainly personnel costs. The PEC brings funds directly from the federal government to schools subject to certain conditions. CONAFE is a federal program that serves marginalized rural areas. · Health for the population without social security coverage is provided by public programs that are principally directed to the poor; providers include the Ministry of Health (SSA), Instituto Mexicano de Seguro Social (IMSS) Oportunidades, and Seguro Popular. Spending for the population without social security coverage constitutes about 5 percent of total public spending. The Fondo de Aportaciones para los Servicios de Salud (FASSA, the Health Services Fund) transfers earmarked funds to the states for running the Decentralized Service Delivery for the Poor xi decentralized health care system. Ramo 12 is a federal item that supports SSA, the health component of Oportunidades, as well as all other federal health funding including Seguro Popular, the national health institutes, and federal hospitals. · Local social infrastructure is provided through the Fondo de Aportaciones para la Infraestructura Social (FAIS), the third largest decentralized fund, which uses a transparent povertyweighted formula to distribute resources for investment to states and municipalities. The financial resources go initially to the state treasuries, with strict conditions to pass along the majority of the funds to municipalities, under the title Fondo para la Infraestructura Municipal (FISM). The rest of the resources remain at the state level under the title Fondo para la Infraestructura Social Estatal (FISE). FISM accounts for almost 2 percent of public spending. Federal and local governments are also able to spend more than the allotted distribution on local infrastructure provided that they have access to other financing resources. This occurs especially in wealthier municipalities. · Local rural development is conducted through Alianza para el Campo, the large scale farm modernization program in Mexico. The Rural Development Program--a municipalized program of Alianza--provides funding to support rural productive investment. The Rural Development Program is targeted to farmers with small holdings. About 0.2 percent of the federal budget is allocated to the Rural Development Program. Participation varies greatly according to state, as does states' own funding. Some of the states and municipalities have independently developed their own programs that augment and, to some extent, fill in gaps for the federal programs reviewed here. There has been little or no coordination among these programs. During the 1990s, Mexico made major policy changes that accelerated the decentralization of programs and financing; at the same time, democracy was taking root at all levels of government. Since then, decentralization of policies and programs has changed little. Although the decentralization agenda remains incomplete, some states and municipalities have made good progress in implementing the reforms of the previous decade and making improvements in the local practice. Many other states and municipalities, due to politics, weak administrative capacity, and socioeconomic conditions, have not taken the initiative to make creative improvements in the federalized programs. Rather, they remain agents to disburse federal money, struggling to deal with national unions and federal regulations. They do not implement the programs to their full potential which is necessary to address local problems such as development and poverty reduction. Since the wider trends in the national politics will continue to dictate the overall degree of decentralization (and probably not lead to significant changes in the near future), the way forward for improving the poverty xii Decentralized Service Delivery for the Poor reduction effects of these programs should focus on creating incentives for improving the quality of services, broadening access to these programs by the poor, and improving transparency with respect to inputs and results. Even though the federal government favors decentralization, there is no unified strategy--the commitment to decentralization varies from ministry to ministry. Within each sector, different decentralization processes coexist.1 States and municipalities are attempting to fill in the blanks of the strategy, sometimes in an ad hoc manner, as their participation shapes the process in their own jurisdictions. While there are often clear steps that would improve the quality of services, many of the national policies and organizational features needed to create more capacity and set better incentives to encourage the "right steps" are lacking. Decentralization can potentially bring benefits, but there are tradeoffs. On one hand, there are potential gains from local knowledge and greater responsiveness to local preferences. On the other hand, there are potential losses from lack of coordination, local capture, or higher inequality in the quality and coverage of services across jurisdictions. The relevant dimensions of gain or loss from the decentralization processes analyzed here include, first, whether the decentralization processes improved performance, in terms of the quality and coverage of services for the poor; second, whether decentralization engendered greater local efforts to mobilize resources or to use them more efficiently; third, whether decentralization allowed or engendered better accommodation of programs to local conditions; and fourth, whether decentralization improved transparency and accountability of service providers to the government or to citizenbeneficiaries. The second, third, and especially the fourth criterion not only are valued for their own sake, but also reflect the features of institutional design that are important determinants that influence the degree to which decentralization achieves its aims. Yet it should be noted, in most cases, there is not sufficient data to attribute specific outcomes to specific features of the decentralization processes. The most that one can report are the trends that occurred in the period since the decentralization processes started. Institutional Conceptual Framework To analyze how servicedelivery institutions may affect quality and efficiency, the report principally uses a framework developed in the 2004 World Development Report (WDR). This framework is adapted to the institutional context prevalent in the different sectors analyzed here, taking into account that each sector has gone through different stages in the decentralization process. In our basic conceptual framework, the 1 Administrative decentralization started around the 1990s. The main events for the purposes of this study are explained in the annexes to Volume I, mainly in the background notes on expenditure decentralization and poverty alleviation; see subsequent chapters for more detail. Decentralized Service Delivery for the Poor xiii main actors in every country are the citizenbeneficiaries and taxpayers (darkly shaded box at the bottom left of Figure 1), the government (darkly shaded box at the apex), and the service providers (darkly shaded box at the bottom right). The administrative process creates a compact--a structure of incentives, financial arrangements, and accountability mechanisms--between the government and the service providers who offer services and expect remuneration. The political process creates another compact between the government, which collects taxes, and their citizens, who expect benefits and hold the government accountable. One can then look at the chain of accountability from provider to government to citizens. Improving the effectiveness of this chain is one way to improve service outcomes. There is also a shorter potential route of accountability, from providers directly to citizenbeneficiaries. Making this short route effective requires adequate local administrative autonomy to accommodate to local conditions, sufficient information for beneficiaries to evaluate service provider performance, and incentives for providers to respond to the influence of beneficiaries. For these programs to achieve their potential in poverty reduction, incentives are needed for targeting and efficiency, and local providers need to have authority and flexibility to respond to these incentives. Local authorities need to have or to develop institutional capacity to carry out these mandates. Figure 1. Accountability Relationships with Political Parties, National Provider Organizations, and Subnational Governments Federal Government National Political Union Party State or Municipality Families/ Citizens Service voters Providers / Political Influence Taxes/Resources Sectoral Policy and Operational Practices (including wage negotiations) Beneficiary voice: the Short Route of Accountability To adapt the framework to the decentralized institutional setting of service provision in Mexico, we need to add three actors--the three lighter boxes included in the diagram. The state and municipal governments add an extra step in the funding and xiv Decentralized Service Delivery for the Poor accountability between the federal government and the service providers, but they might also allow a shorter long route of accountability if they take adequate control of the service provision and are responsive to the local citizenry. The political parties in Mexico come in as separate actors because of their strong role in integrating the system (especially in the decades of oneparty rule) and in managing the careers of the politicians who run the governments. With no reelection allowed in Mexico, pleasing the party is the main currency used by officials (elected or appointed) to assure their future. An analysis of the political parties is beyond the scope of this report, except to note that they are important intermediaries in the system. To some extent they facilitate democratic communication from voters to leaders, and to some extent they represent their own interests in ways that hinder such communication. The national unions (teachers, health workers, farmers) and other national corporatist organizations (for example, the association of construction contractors) have a traditionally significant role in and strong links to the political parties in Mexico's corporatist structure. This report does not analyze political parties, but rather the relationships among the state, providers, unions, and citizens or clients. Unions and corporatist organizations make a positive contribution in representing key stakeholders and in integrating the national development strategy. However, they also represent a private interest that may conflict with the public interest, and there is the possibility that they may disrupt the delegation of authority to subnational governments and the necessary accountability for service quality and efficiency. Even the customized WDR framework cannot capture all of the important institutional dimensions. We must recognize, for instance, the need to develop and retain competent personnel at local levels, the need to consolidate federal programs and streamline their regulations so that local government can readily access them, and the need to create effective linkages among federal programs and between them and locally run programs. Decentralization and Performance In all sectors, coverage, equity, and targeting of services have improved, especially for the poor, and there is evidence of quality improvements in some cases. Quality generally remains low by OECD standards and remains roughly on par with Latin America. Service performance varies widely across the country, even after accounting for the correlation with the state income per capita, which indicates that other important factors are also influencing performance. Nonetheless, there is very little information at the state and municipal levels on the performance of the programs analyzed here, making it difficult to establish a link with decentralization. In education, the gap in primary education coverage between the poor and the rich states has narrowed since decentralization. Equity has improved, perhaps due to Decentralized Service Delivery for the Poor xv the cumulative effects of compensatory programs.2 However, efforts to increase lower secondary schooling for the poor confront significant challenges. Quality of education is low by international standards even after controlling for Mexico's level of income. This is particularly true for the poor, which exacerbates the problems of access. The rapid expansion of the telesecundaria system is worrisome; although it has helped to improve access to school for the rural poor and accounts for 20 percent of total lowersecondary enrollment (and a much higher share in rural areas), students in the telesecundaria have recorded low levels of achievement. Despite important institutional limitations in the decentralization process, local innovations have helped both rich and less wealthy states, improve their basic education systems. States that follow this pattern include Aguascalientes, Colima,3 Nuevo León, and Quintana Roo. The first three states plus the Distrito Federal rank at the top of the PISA scale, while the last state is around the national average. These states are working to improve learning achievement on several fronts, taking initiatives such as improving pedagogy practices, evaluation, and the hiring of teachers. In addition, some federal educational programs are stimulating discussion about how to improve school management and learning achievement. In the area of health, indicators of outcome, such as life expectancy and infant mortality, have improved in the last decade, due in part to the success of centralized programs, such as the vaccination program. However, maternal mortality has remained at very high levels according to international standards. The Millennium Development Goal on maternal health proposed to reduce maternal mortality by twothirds from 1990 to 2015. However, with a reduction of only 30 percent in the last 15 years, it is unlikely that Mexico will be able to achieve this Millennium Development Goal. There are a lot of data about the overall population, but the existence of several different public systems of provision (IMSS, SSA, ISSSTE, IMSSOportunidades, and state health systems) make it difficult to establish a relationship between systems and outcomes. The following results are from a health analysis based on a sample of six states (Arredondo 2005). First, across states, there seem to be important differences in health outcomes and indicators, and the differentiation is not necessarily between the poor and the rich states. Second, in high to mediumincome states such as Baja California Sur, Jalisco, and Tabasco, the evolution of health performance has been neither even nor significantly better than in the rest of the country. Third, these states that seem to be far advanced in the decentralization process are investing considerable resources in their health systems, and have been able to reallocate resources relatively effectively (preventive versus curative expenditures). More data and analysis are necessary to assess the progress of decentralization and its effects in these states and in others. A new federal program-- 2 The Ministry of Education's compensatory education interventions target schools in disadvantaged rural areas and increase resource allocations for those schools to give students more equal opportunities. Compensatory education programs tend to decrease dispersion in scores. Perhaps the cumulative impact of compensatory education from basic education is carried over to lowersecondary schools and explains, in part, the high degree of equity observed in PISA results. 3 Colima is the best performer but only ranks 14th in GDP per capita. xvi Decentralized Service Delivery for the Poor Seguro Popular--distributes funds to states based on the location of patients who choose to affiliate and is helping to expand coverage among the population without social security coverage and to reduce catastrophic health expenditures. Coverage of social infrastructure has improved, although it continues to be low for the poor. FAIS is an important source of financing for local social infrastructure investment in the poorest municipalities, and it is, in general, well targeted according to the level of marginality. However, the significance of FAIS in the overall availability of municipal financial resources varies by sector and size of municipality. The sectoral allocation of the FAIS investment (water versus roads versus housing upgrading and so forth) is not generally correlated with lack of coverage in that sector, although this finding varies by state and sector. A potential explanation is that municipalities have several sources of funds for investment in social infrastructure in addition to FAIS. Another explanation is that FAIS institutional arrangements vary from state to state. In productionoriented programs in rural areas, investment is high but not notably effective. This is due to the proliferation of segmented federal programs with no coordination, the lack of clear political accountability for rural development because responsibilities are diffused among various federal Secretarías and state governments, and a host of problems in program design and implementation. Additionally, the poverty impact of these programs, and its impact on rural development, is difficult to assess because much of it is indirect, depending on its effect on employment and wages. There is evidence for Mexico and other Latin American and Caribbean countries that agricultural growth has a strong effect in reducing rural poverty, and there is also evidence that the rural poor in Mexico derive a large part of their income from nonfarm sources. Therefore, programs oriented to increasing employment and income opportunities in the farm and rural nonfarm sectors are, in principle, prorural poor, but the targeting of these programs may be problematic. Thus, a problem with Alianza's Rural Development Program is that it concentrates on small farmers, that is, on producers who already have access to at least some productive assets and in turn are wealthier than the poorest in rural Mexico. As a consequence, even if the targeting of the program is good in terms of "farmers" (80 percent of beneficiaries belong to the two smallest categories of the five types of farmers considered by the Food and Agriculture Organization), it is not designed to reach the poorest rural families. Funding, Equity, and Performance The incentive for raising revenues or using transfers efficiently at the state or local level is driven by the current arrangements for taxes and transfers between the states and the central government.4 The system has its foundations in the early 1990s 4 The two main categories of transfers to the states are participaciones and aportaciones. Participaciones were originally state and municipality revenues whose collection was delegated to the federal level in the Fiscal Decentralized Service Delivery for the Poor xvii and earlier (Fiscal Coordination Law), when the interinstitutional relationships described above were heavily dominated by the power of one political party that controlled the federal and almost all state governments and drew its power base from various organized interests, including public employee labor unions. States and municipalities are heavily dependent on transfers from the center. A standard result from the literature on fiscal decentralization is that the dependence on transfers reduces the direct accountability between subnational governments and their citizenclients. This lack of accountability can result in less effort to mobilize local revenues or to use transfers efficiently. A negative effect of decentralization over the past decade seems to be a reduced tax effort by municipal governments, which declined from 0.17 percent of GDP in 1990 to 0.07 percent of GDP in 2004. The amount of transfers to municipalities has grown strongly, but the resources are coming to them with less discretion and more by formula. Since the participaciones formula gives a slight reward for local fiscal effort, and state debt bailouts and discretionary transfers from the president virtually ended after 1998, the marginal incentives for local fiscal effort have improved. But the big increase in total federal transfers to municipalities has had an apparently negative effect on the total tax effort. States have not changed their tax effort appreciably, perhaps because the law does not give them much scope to collect their own revenue. Even though no claims of causality can be made, it is illustrative to report key trends in efficiency observed since the beginning of the decentralization. In resource use, there is little evidence of greater efficiency as a result of decentralization, but it varies by sector. For instance, the bulk of education spending is budgeted and distributed "inertially," based on the stock of physical capital and employees belonging to the federal system before decentralization. This allocation does not consider the demand of clients, the productivity of suppliers, or poverty. Moreover, the expansion of enrollment nationally has been proportionately less than the increase in the number of teachers and schools and slower than the increase in spending. This increase in spending per student since decentralization is explained mainly by the increase in teachers' salaries.5 Crosssectional data show that teacher salaries are not correlated with student achievement.6 Therefore, it is likely that this higher expenditure has not Pact for tax efficiency reasons. Most of these transfers are distributed under Ramo 28. Aportaciones were conceived as federal money earmarked to pay for (formerly) federal commitments and transferred to the states and municipalities together with those commitments (for example, education and health). These funds, formerly under Ramo 26, now go out under Ramo 33. For an extensive review see World Bank (2004a; 2001a; 2001b). 5 International comparisons suggest that higher shares of nonsalary spending are associated with better performance. An area for future research is to obtain data on nonsalary spending by state, in order to assess whether spending is correlated with performance. 6 The subnational governments received transfers equal to about twice as much per capita in real terms in 2004 as in 1992, before the major sectors began to be decentralized (1993) and the last main revision of the participaciones formula was finalized (1994). xviii Decentralized Service Delivery for the Poor affected performance. Moreover, standardized educational testing results across states are not at all correlated with spending per student (see Figure 2).7 Figure 2. Math and Reading Test Results and Expenditures per Student, 2001 Lectura Matemáticas 500 435 SIN 495 SIN 430 490 TAM 425 HGO HGO 485 SLP CHIH 420 AGS DGO DGO o* GTO QRO QR COAH o* OAXSLP TLAX 480 CHIS TLAX 415 NL BC GTO JAL CHIS NAY TAM NAY omedi 475 OAX NAL BCS COL COL omedi 410 CHIHNL COAH preaj AGSM MVERICH OR SON QRO NAL MOR JAL 470 YUC preaj 405 M ICH SON ZAC BCS untP 465 PUE ZAC 400 PUE YUCQR BC M EX untP GRO 460 TAB 395 CAM CAM M EX 455 390 VER 450 385 GRO TAB 445 380 6 7 8 9 10 11 12 13 14 15 6 7 8 9 10 11 12 13 14 15 Gasto por alumno en 2001 (miles de pesos) Gasto por alumno en 2001 (miles de pesos) Source: SEP. In health, one cannot attribute evidence of improvements in outcome indicators to decentralization: performance varies greatly across states, particularly among those states that have gone farthest with decentralization, and there is little information at the state level on the nonfiscal factors that might influence performance. With respect to spending, health expenditures have been increasing in real terms at a rate of 9.6 percent a year. Federal transfers (FASSA) and federal spending (Ramo 12) have increased from MxP$1,186 per capita for the population without social security coverage in 1992 to MxP$2,319 per capita for the population without social security coverage in 2004. In the mid1990s, during the 1994 crisis and shortly after, there was a substantial drop in spending, followed by a recovery. As in the case of education, the bulk of spending is budgeted and distributed "inertially" across states, based on the stock of physical capital and employees belonging to the federal system before decentralization. This leads to inequity, because states that are well endowed receive large portions of FASSA funds. Although 20 percent of FASSA is allocated through a formula that takes into account state deficits or surpluses in health care, this does not compensate fully for the unequal distribution of services, nor does it give incentives for states to improve the efficiency of their systems. Recent initiatives are trying to reverse these inequities and inefficiencies. In social infrastructure, there are some indications of improved efficiency since decentralization. Most states have developed their own planning and implementing 7 The tests were not given in the years prior to decentralization, so we do not know how changes in spending over time correlate with test results. More statistically sophisticated analysis, taking account of the local socioeconomic conditions and the basic organizational features of the states, confirms the insignificance of per capita funding levels for explaining variation in test performance across states. See the discussion in the education chapter in Volume II and Quality of Education, World Bank, 2006b. Decentralized Service Delivery for the Poor xix models for FAIS, which require careful analysis in order to identify and disseminate best practices. In rural development, there are no clear standards for output and thus no concrete indicators of efficiency. Most public spending for agriculture and rural development is not efficient in terms of its direct effects on poverty, since most public spending does not go to the poorest rural citizens. According to targeting criteria based on state marginality, the decentralized funding of Alianza is regressive. Whether it is effective for developing agriculture is beyond the scope of this report. Although the variance in performance typically is not correlated with the variance in federal or total funding levels (as mentioned in the context of efficiency), the geographic variation across states and municipalities in funding per capita or per user is substantial (usually by more than a factor of two from lowest to highest). This variance in funding may potentially generate differences in coverage and quality. The best program in this regard is FAIS, whose distribution follows a transparent formula based on population and the marginality index. There is still room for improvement because the municipalities do not necessarily distribute the projects to benefit their poorest communities and because some states apply an alternative formula to the one used by the federal government to distribute FAIS resources to their municipalities, with the result that some municipalities with similar degrees of poverty in different states do not receive a similar level of funding thru FAIS. In basic education and health for the population without social security coverage, the bulk of the distribution (FAEB and FASSA in Ramo 33) is done according to the distribution of teachers, doctors, and nurses in the federal facilities before the decentralization.8 While this was a practical starting point for the national agreements and the transition, it has led to widely varying funding levels per student and per population without social security coverage, disparities that have increased over time. The disparities are not correlated with poverty levels, but rather with the absolute state population (larger states get less per capita) and with the size of the states' own systems before decentralization. States that had--and made the fiscal effort to pay for--larger systems of their own usually had small federal systems per capita before, and so have received less federal financing since decentralization. The Rural Development Program is allocated to states according to a complex formula that includes variables such as rural population, agricultural GDP, harvested land, irrigated land, number of production units, and state contribution. State trust funds are responsible for the financial management and, through technical committees, for the approval of farmers' proposals. Since 2003, part of the funding of the Rural 8 The federal government has been allocating resources to compensate for inequities through the CONAFE and PEC programs in education and the Seguro Popular program in health. xx Decentralized Service Delivery for the Poor Development Program started going to municipal authorities. States generally complain that excessive rules and regulations hinder the implementation and adaptability of the program. Accommodation to Local Conditions The decentralization process in Mexico is far from complete. One root of this is the continued near monopoly of the federal government in raising taxes. Second, even though around half of programmable public sector funds pass through the state and municipal governments, the federal government retains a dominant role in directing how the decentralized resources are used. The means of influence differ, but much of the underlying political economy is similar. Powerful interests, such as some unions, have a close relationship to the federal government, built up during the 20th century under a corporatist singleparty state, and they prefer to avoid dealing with the independent and politically competitive governments of the 32 states.9 In the education and health sectors, the unions for teachers, doctors, and nurses negotiate the salaries and rules for personnel with the federal government, and then the federal government transfers the money to states to pay these workers. States have little power to manage personnel which seriously limits their options for managing the sector or for accommodating to local circumstances. In the decentralized programs for municipal infrastructure and rural development, there is little ex ante earmarking of funds to be devolved to municipalities; in most areas where such funds could be used, they are dominated by other federal programs in water, electrification, roads, and agricultural investment. To some extent, the municipal programs allow the local governments to put money where they feel the federal government has missed an opportunity, but some of the devolved funds come back under federal control as required counterpart funds to the federal programs. The central government has imposed many restrictions and rigidities on the implementation of municipalized rural development programs, making it difficult for the local governments to operate. Furthermore, it is difficult to design strategically sound rural development programs with investments clustered around local development axes. Incentive for Local Performance International experience shows a variety of incentive mechanisms for improving local performance. Unfortunately, most of them are absent in the decentralization arrangements in Mexico. 9 The leadership of the national teachers union has opposed decentralization of the union as a potential threat to its influence and importance. Decentralized Service Delivery for the Poor xxi At the aggregate political level, in most other countries local politicians have the incentive of reelection, which makes them accountable to local voters. There have been proposals, but no actions, to allow reelection at the municipal level in Mexico where the single threeyear term discourages projects with longerterm benefits. Coahuila is the one exception to this rule. At the aggregate fiscal level, locally raising a substantial share of fiscal resources for local spending, as is the case in Brazil (Southern states), Canada, and the United States, creates another accountability relationship with local taxpayer voters. In some larger cities of Mexico, local revenue generation occurs, with beneficial incentives for efficiency and accountability. However, smaller municipalities lack the capacity and states lack the fiscal space to replicate this effort. Additionally, the growth of federal largess with transfers has further undercut the incentive for local fiscal efforts. Providing technical assistance and legal tax authority (to states) as a substitute for increased federal transfers would help improve incentives. In education and health, many countries have formulas whereby the funding follows the location of students' attendance or patients' use of services, with perhaps some simple rulebased adjustments to account for demographic characteristics, like age and geographic dispersion. Such arrangements create beneficial incentives in several dimensions. First, there is an incentive for equity, as the formula for distributing national resources creates an expectation for similar methods of distribution locally--for example by facility. Second, such formulas create incentives and space for local governments to choose more efficient combinations of inputs. Third, the formulas reward facilities and local governments that improve school attendance and make their health services more attractive. Box 2.2 in Chapter 2 describes how Colombia has converted its basic education transfers to a capitation basis, after starting from a distribution based on the historical location of teachers and health workers, as Mexico has. Accountability and Transparency Since the beginning of the decentralization process, institutional change in these programs has allowed for an improvement of the accountability mechanisms in two ways--first, by forging shorter routes of accountability through some of the state and municipal governments that have responsibility for managing service delivery in education, health, and municipal infrastructure services and, second, by forging direct links of accountability between clients and providers in some schools, health facilities, and municipal service agencies. Although it is hard to prove definitively, anecdotal evidence supports the presumption of this report that more transparency and accountability will lead to better social services and infrastructure for the poor. The challenge of improving the transparency of the decentralized programs has two aspects: clarity on the allocation of resources and evidence of the results. The allocation of resources for the sectors under consideration is certainly more transparent than it was 12 years ago when most of the decentralization began. A politically more xxii Decentralized Service Delivery for the Poor independent national Congress leaves much less of the budget to executive discretion. The budget--proposed and approved--and the various reports from the executive branch now provide detailed public information about distribution of resources across programs and geographic areas. Some resources, including the funds for social infrastructure and rural development considered here, are distributed by formula to states and within states to municipalities. People can see where the money is going geographically and debate the pros and cons of the formula. On local decisions regarding the use of financial resources, transparency varies, but at least some places are utilizing participatory platforms to choose projects. Among them are the Municipal Planning Councils (COPLADEMUN) for social infrastructure and the Municipal Councils for Sustainable Rural Development (CMDRS) for agricultural and rural development. The electoral democracy at the local level could be mutually reinforcing of these institutions, if they facilitate the attainment of technically sound and popularly supported spending decisions and avoid being captured by local elites. In health and especially education, the national unions strongly assert their own interests, sometimes at the expense of service quality and other citizenbeneficiary interests. Nonetheless, the variety of statelevel experience in Mexico shows the potential, partially achieved in some states, to make incremental improvements through negotiation with the unions. As states achieve more fiscal autonomy by raising revenues locally, they have more flexibility to negotiate such agreements. Transparency about results is also improving, although information is still scarcer than for resource allocation. In education, Mexico now uses the international standard PISA test, which provides information on individual student achievement, along with data on socioeconomic and geographic background. Identification of the policies and innovations in states with high average scores and low dispersion could be valuable knowledge for the others. (See the sectoral piece on education in Volume II and World Bank, 2006b.) Available analysis also shows that the special federal programs in education that build local accountability of schools to the parents and community--Apoyos a la Gestión Escolar (AGEs of CONAFE) and perhaps PEC--are associated with higher school achievement, even after controlling for socioeconomic factors. These programs do not create new schools; rather, they increase the resources and transform the incentives in schools that remain in the traditional framework of state management and federal funding for teachers' salaries. In health, making each provider accountable for outcomes will require information on the health outcomes in each of the client populations--information that is still missing. SSA has begun collecting key data on the quality of health care with three evaluations, in 2003, 2004 and 2005, which will be published in August 2006, on issues such as whether facilities are opened and staffed on schedule, how long patients Decentralized Service Delivery for the Poor xxiii have to wait for procedures, what procedures are performed, what share of procedures are successful, and whether medicines are available. Some evidence suggests that the IMSSOportunidades health system, which is still operated centrally, but only in about half of the states, provides better service than the staterun SSA programs. Some states have combined the management of the two programs: Campeche, Chiapas, Oaxaca, Yucatán, and partially Veracruz. Seguro Popular is the major recent innovation in the health sector institutions. The program has recorded success in shielding families, especially the poor, from catastrophic health expenditures. Participation has grown quickly since 2003 and is slightly more progressive than participation in SSA (which is already propoor), but less progressive than in other programs such as Oportunidades. With the goal of having funds follow the patient; the program has been successful in assigning funds according to the number of beneficiaries. The next step is to direct funding to the facility where the patient goes, a change that is scheduled for 2010. The states have to contribute a share of the costs, which gives them incentives for efficient management. Much of the expenditure under the program goes for medicines, and the part used for personnel has been used to hire shortterm contract workers, which should give administrators some leeway in managing the facility. To improve monitoring and influence the quality of service, Seguro Popular has a federal accreditation program for states and facilities; there is not yet evidence on how well this works. No progress has been made on creating a culture of prepayment, as the states are listing almost all of the registrants as the extreme poor, so that they make no payment. Some states with IMSSOportunidades facilities have started to fund them with Seguro Popular as well, which has the potential to create an effective bridge with the state SSA system. For local infrastructure, almost all municipalities report what they do with the resources from FISM to state governments, however only some states report this information or information on FISE funds to the federal government, as required by law. The lists of investments undertaken appear reasonable--water and sanitation, local roads, and water appear to be most common. FAIS allocations in the water and electricity sectors appear to be correlated with lack of access to these services, although a similar correlation was not found in the case of sanitation investments. Many other programs (mostly federal, some local) also fund such infrastructure. So, without knowing the total array of investments coming into each locality (which no one keeps a record of), it is not possible to assess whether the resources from FISM (generally a minority of the total) are being well allocated. For municipal investment in rural development, information is even scarcer about both the local needs and the actual allocation of investments. The detailed and usually delayed federal regulations on the use of funds seem to have the primary effect xxiv Decentralized Service Delivery for the Poor of making project execution difficult and inefficient. The complexity also makes it unnecessarily difficult for municipal rural councils to understand and implement the program. Policy Challenges and Options for the Future To date, decentralization in Mexico has been haphazard. Fiscal decentralization of transfers and spending has gone quickly, for political reasons, while taxation has remained centralized. Although local governments write the check, they have insufficient authority to manage the sectors for results. In education and health, most of the money goes to workers whose contracts are set at the national level. In rural development, complex national rules constrain local programming, and in municipal infrastructure as well as rural development, the much larger direct federal programs often overshadow any results produced by the local programs. For these and other reasons, institutions and incentives for accountability have remained weak in all the program areas examined here. True accountability is not possible without adequate authority. Decentralization has worked better in those services, states, and municipalities where there was effective (efficient, propoor) filling of the gaps in the federal programs. This implies a need both for systematic shifts to a more rational decentralization process and for a process of learning from successes and failures, in particular whether successes are products of good design or local context. Some welldeveloped states such as Aguascalientes do well in all services, but other states with comparable income levels do well in a few sectors, but not in others. The differences in their reactions to the decentralization challenge provide scope for learning from good and bad experiences. While the variety of circumstances and results precludes a uniform conclusion about decentralization, part of the way forward towards improved service delivery seems to entail loosening of regulations about what the local governments do and giving the federal government a stronger role in setting standards of outcomes (for example, core curricula, engineering standards), undertaking monitoring and evaluation, and requiring that the local governments disclose to their citizens the financial arrangements and outcomes of the sectors and programs they manage. While looser regulation means less earmarking and direction within each sector, international experience in Canada and Europe shows that sectoral earmarking can be loosened (without leading to underfunding) as the citizens develop clear expectations that the local government has the responsibility for providing the service. This phenomenon is starting to happen in Mexico, most clearly with FAIS, which never had tight earmarking. It also has occurred in education and health, such as when people Decentralized Service Delivery for the Poor xxv complain to their governor about a rise in maternal mortality or a problem with schools. Concerning incentives created by the fiscal system, international experience suggests a number of policy options for improving the mobilization of local resources, efficiency of resource use, and distribution across jurisdictions: · Incentives for local revenue generation appear to be weak in Mexico: own revenues are low and have declined since the increase in federal transfers during the past decade. At the municipal level, the mayors in poor municipalities have incentives mainly to "bring resources down" by influencing federal and statelevel resources, instead of mobilizing local revenues (Grindle, 2004). This is bad for overall revenue mobilization and for strengthening the local tax contract. The situation of the states is no better. The increasing role of Congress has tended to magnify this: while the Support Program for the Strengthening of the States (PAFEF) is still relatively small, it adds another incentive for states to look "up" for resources rather than to generate revenues locally. · Despite the shift to a more formuladriven allocation of resources for municipal investment, with greater automaticity of federal resources (which improves incentives at the margin), the wealth effect of the large increase in the volume of (inframarginal) resources has decreased the incentives for mobilizing local revenue in many places. · For both states and municipalities, the federal government would do well to slow, if not reverse, the growth of total transfers, using them selectively to motivate nationally desirable practices, to compensate poor areas, and to encourage local revenue efforts as a way to find additional resources for public purposes, including the povertyreducing programs discussed here. · For municipalities, which have a good potential tax base (property tax), the federal government could support technical assistance to strengthen the cadastres, property registries, and collection of property taxes. The Ministry of Finance was considering such a project in 2005 and should return to this idea. · For states, which lack a good tax base, the government several times has proposed allowing them to have a final sales tax on the same base as the federal value added tax (VAT). The Convención Nacional Hacendaria (CNH, National Fiscal Convention) also has recommended it, but agreement has never been reached about how the VAT and the tax participation system would be adjusted, in order to accommodate the proposal. Perhaps the next government will take up a similar idea. xxvi Decentralized Service Delivery for the Poor · Implementing the CNH recommendation to expand the tax base of the state governments would give states' fiscal authority to respond to citizen demands for improved services and give citizens more interest in monitoring the efficient use of the taxes they pay. · The CNH in 2004 recommended moving to a distribution formula for federal resources (aportaciones) based on the number students or clients. Implementing the CNH recommendation to distribute education and health resources, at least at the margin, based on the location of students and clients, respectively, would contribute both to improved equity and to citizens' understanding that resources come to a state in response to the needs of their citizens rather than the bargaining power of a national union. · Further fixes to the FAIS formula should take careful consideration of how much the rich states should be expected to help their own poor.10 Better and more efficient services are in many instances related to the flexibility and authority that local governments have to adapt programs to local needs and to take other steps to improve efficiency and quality. International experience shows the importance of this (World Bank 2004c). A number of specific actions would allow greater accommodation to local conditions. In some cases, they involve changing the intergovernmental compact, so that state or local governments have more freedom to respond to local citizens. In other cases, citizens would gain more scope for expressing their preferences from having more freedom to choose their service provider (for example, Seguro Popular). · Programs like AGEs and PEC and other measures to strengthen accountability of schools to parents have the potential to help improve the quality of Mexico's basic education and thus to reduce poverty. Further analysis needs to identify what aspects of these programs have the beneficial impact and how this could expand to a wider circle of schools. It is important to promote educational models that improve pedagogical practices, evaluation at the local level, and accountability, so that resources are allocated based on performance. · In education and health, the federal government could encourage states to take the initiative in the negotiations with the union by setting the standards of quality without specifying details of method, by allowing more state discretion to improve such things as school curriculum and health payment 10 We do not know how much rich states are allocating their untied (nonFISM) resources to help their own poor. The question we pose here is how much they should be allocating. Do they ever graduate from needing federal assistance to helping their own poorer regions? Decentralized Service Delivery for the Poor xxvii systems, by promoting general fiscal autonomy, and by disseminating information on best practices. · SSA should allow greater flexibility for management of health units at the local level, so that they can respond to local needs and make the best use of the decentralization process. Accountability, both upward and downward, should continue to be fostered to ensure that the flow of resources to the states is based on performance and to strengthen the hand of citizens in exerting constructive political influence on their local governments. While these measures do not guarantee that the local governments will respond appropriately to citizens' recommendations, they should help strengthen the accountability relationship that arises out of the democratic process. · An attractive option for reform in the health sector, with its multiplicity of programs and providers, would be to have the affiliation of each client indicate the source of funding that she or he could then take to the most suitable provider.11 Seguro Popular aims to have portability of funding in the near future, and the other programs could arrange similar portability as well. Having funding follow the client would, based on international experience, enhance the power of the clients and the accountability of the providers. · As clients gain flexibility to seek the best provider, the providers--hospitals and clinics--will need and demand more flexibility and authority to improve their services. · For FAIS and FISM, some states, like Puebla and Zacatecas, have instituted their own rules to assure that poor communities receive adequate consideration within their municipalities. These experiences should be studied in more depth, and best practices should be disseminated. · In some sectors, FAIS resources seem to be well targeted according to the needs of the municipalities. However, an analysis of the efficiency of FAIS would require knowledge of local needs, conditions, and resources. In any case, allowing reelection of mayors, as recommended by CNH in 2004, would improve continuity and probably contribute to more efficient investment plans with a longer time horizon. 11 Initially, the portability of Seguro Popular funding might be limited to SSA and IMSSOportunidades facilities, since access to the insurance (without social security coverage)based facilities (for example, regular IMSS, ISSSTE) is now linked to employment status. While virtually all the other reform options discussed here could go with or without the others, separating the provider from the financing source would affect the design of some of the other proposed reforms. xxviii Decentralized Service Delivery for the Poor · For rural development programs, as an alternative to municipalization, Mexico might consider consolidating most rural development productive programs at the state level, as places like Germany, Italy, and Spain do successfully. In this model, federal regulation of local program design would be relaxed, focusing instead on the transparency of information to the local population. Transparency of information is a key factor for improving accountability in each program. Many states and some municipalities have followed, and some even have surpassed, the federal example of enacting the Federal Law of Access to Information and setting up an agency to implement it. Federal actions in the areas discussed here should support this trend. Below is a list of specific suggestions, which includes both improved information on resource allocation and improved measurement of performance. For example, the specific transfer programs might include disclosure of information as a condition for receiving at least some of the funds. · In education, the government should make the data available publicly down to the school level. Information for each school on the budget and staffing levels and the attendance of students and staff should also be made available to the local community and thus to the wider public. · SSA should continue its programs to improve information about health, making publications more prompt and providing clearer information on the linkages between outcomes and the agencies responsible for each segment of the population. · To strengthen capacity and accountability of the COPLADEMUN and the CMDRS, a combination of fiscal incentives for transparency (like increments to transfers being conditional on timely publication of information), technical assistance to build capacity, and budget flexibility to build and implement a mediumterm program at the municipal level is needed. The federal government should disseminate information on the best practices in terms of planning and allocation of FAIS resources. · Relying on the judgment of the local citizenry, enhancing the information available to them, and strengthening the capacity of local government to respond could help to improve the choice of local social infrastructure. Major improvement will require more than simple transparency measures, however, such as more incentives for raising local revenue, as mentioned above. · If the government decides to pursue the strategy of support for small and medium farmers via local governments--states or municipalities--the most viable option is to shift the regulatory emphasis on disclosure of information. Decentralized Service Delivery for the Poor xxix For instance, one option is to make the money for each year available once the municipality has fully disclosed the information on the program in the previous year (eventually including performancemonitoring data) and has made public its plan for investment in the current year. Some beneficial changes in program design or sectoral management are not specific to the institutional arrangements of decentralization. For example: · For the rural development program, expanding the access for landless rural families would improve the probable impact on poverty reduction. This could be done either by facilitating the access of landless farmers to land (as in the program for young landless farmers recently started by the Secretaría de la Reforma Agraria, SRA) or through much more support for rural nonfarm activities. · In education, experience internationally and within Mexico indicates the benefit of improving (a) the pedagogical process (the content of textbooks, teaching process, and use of didactic material, for example); (b) the hiring, assignment, and reassignment of personnel (for example, to avoid transferring teachers during the school year); and (c) the testing and monitoring of results. Looking across all programs and states, time is a factor leading to improvement in federal programs such as AGEs in CONAFE. The longer a state is running a program, the better the results will be on average. Thus patience is needed: · As long as a program is showing some improvement, even if it has not yet achieved full expectations in a couple of years, it should not be scrapped or allowed to die with the change of administrations. It would be better to continue with it and make marginal improvements in design, while the local and federal authorities learn how to operate it better. Whatever the benefits of the individual programs in this report, one must also consider the context of overall public spending and the pattern of earmarking so many resources for particular uses by subnational governments. As these commitments become politically, if not legally, inflexible, few resources are left for any government to implement a broad economic development plan. The social programs that benefit the poor directly will work best to reduce poverty when they are in the context of overall strong economic growth. Any wholesale reversal of decentralization in Mexico seems politically unlikely. The increased democracy of state and municipal levels has contributed to the decentralization and seems likely to sustain it politically. So trusting more in local xxx Decentralized Service Delivery for the Poor democracy may help to improve the performance of decentralized service delivery. Ideas for reasserting central control run up against the evident failure of this strategy in rural development where detailed rules for decentralized federal programs have made local implementation administratively difficult. Moreover, the federal authorities lack the information and political mandate to positively influence the substance of what the local governments are doing. Further improvements to decentralization in Mexico seem to be more about increasing the transparency of both resource allocation and intergovernmental division of responsibilities, along with the overall task of deepening democracy and strengthening governmentsociety links. The analysis presented in this report does not pretend to be exhaustive of decentralized service delivery for the poor. Important areas for future research include the following: (i) analysis of the compliance of the law or of the detail regulations in programs; (ii) systematic comparisons of state interventions in the sectors analyzed in this report; (iii) analysis of the implications and requirements to implement per capita funding in health and education, and freedom to choose providers; and (iv) systematic compilation of information at the state and municipal level on the performance of the programs. Decentralized Service Delivery for the Poor xxxi xxxii Decentralized Service Delivery for the Poor CHAPTER 1 DECENTRALIZATION, POVERTY, AND DEVELOPMENT IN MEXICO This report, the third in a series on poverty in Mexico, examines the relationship between decentralization and the provision of key social and productive services-- education, health, social infrastructure, and rural development--viewing the overall performance of these services over the past decade as potential contributors to poverty reduction. While it is not possible to identify the effect that the decentralization of these programs has had on poverty reduction, the report does discuss trends in the performance of these programs, which should affect poverty, and the strengths and weaknesses of the decentralization process with regard to accountability, incentives for performance, and allocation of effective authority. The firstphase report dealt with the overall conditions and influences on poverty, including an evaluation of the CONTIGO strategy. It found that Mexico has achieved major progress in some nonmonetary dimensions of wellbeing, but much weaker progress on others, such as the income of the poor. Despite the recent gains, particularly for the extreme poor, poverty remains a challenge. CONTIGO provides a sound conceptual framework for the government's poverty reduction strategy. There have been notable successes in specific programs as well, especially Oportunidades. But many challenges remain for implementing the principles of the CONTIGO framework across government programs. The quality of services is a major problem in many sectors. Social policies for meeting the basic needs of the extreme poor are well developed, but policies for promoting their income growth are not. A broader agenda for the moderate poor is pending. Most of the extreme and moderate poor fall outside the formal social protection system and face significant risks, for example, from health crises, unemployment, or lack of income in old age. And a wide range of institutional issues have yet to be tackled, such as strengthening accountability, especially under decentralized structures. The secondphase report dealt with the determinants of rural and urban incomes and management of household risks. The poor are very heterogeneous, and, among other dimensions of poverty, geographic location matters in the design of appropriate poverty alleviation interventions. The urban poor are limited to lowquality jobs marked by low productivity and limited social protection. To continue supporting the rural poor to move out of poverty, it is important to increase agricultural productivity, especially for small and mediumsize farmers. In addition, Mexico's health system does a poor job of addressing the risks facing the poor. Decentralized Service Delivery for the Poor 1 To follow up on the results of the previous reports, this report focuses on the delivery of public services and on centerlocal relations in this area. It does so for two reasons. First, both social and economic services shape the opportunities of poor people. Social services contribute to human capital formation and risk management, while economic services shape access to markets. The government provides an important subset of these services, and achieving better access and better quality of services is a major goal of public action. Second, a large and rising proportion of government services are being delivered, to varying degrees, under decentralized auspices. This is the result of Mexico's particular pattern of decentralization, especially in the past decade or so. Which services matter most? The report selects areas of particular importance to the poor in the domain of social services (education and health) and economic services (local infrastructure and services related to rural development and agriculture). The report does not address the provision of preventative communal public health or other municipal services in large cities, since those are rarely financed with earmarked federal resources, or the provision of a social safety net, since the federal government operates the main program in this area--Oportunidades. This chapter provides an overview of the sectors and programs analyzed in this report. Also, this chapter discusses the framework and history of decentralization of these programs. I. SERVICE AREAS IN THIS REPORT The government strategy for poverty reduction, CONTIGO, includes two types of interventions: broad based and targeted (see Figure 1.1). Broadbased social expenditures are not specifically targeted to the poor, but the CONTIGO framework recognizes that they provide opportunities for the poor. Within broadbased interventions, the government distinguishes the areas of social security, health care, education, job training, and housing. 2 Decentralized Service Delivery for the Poor Figure 1.1. CONTIGO Poverty Reduction Strategy Social Security Health (for open population) Education (basic) Broad policies Job training (general population) Housing Human capital development CONTIGO (that is, Oportunidades) strategy Targeted policies Employment and income (poor population) opportunities (for example, Alianza's Rural Development) Development of physical capital (for example, FAIS) Recognizing that broadbased interventions alone do not suffice to help the poor emerge from poverty, the Government of Mexico has also implemented three groups of targeted interventions specifically designed for the poor. The first group promotes investments in the human capital of the poor, with a focus on education, health, and nutrition. The second aims to improve employment opportunities for the poor. The third targets poor areas in order to provide them with better services and physical capital. One can see these programs of federal origin in a broader context by considering also the programs that individual states and municipalities create. Figure 1.2 shows them arrayed by the originating level of government and also by whether the benefits go more to individuals or to communities. Decentralized Service Delivery for the Poor 3 Figure 1.2. Map of Decentralization and Social Programs "Topdown" decentralization Federalization of Education (F) Oportunidades (F) Health Decentralization (F) PEC (F) Social Infrastructure Funds (F) Municipios Saludables (F) Exogenous Endogenous Social Community Patronage Action Promoting values and social groups Attention to Vulnerable Groups Educational Improvement (M) Sanitary Prevention (M) Social and Family Entrepreneurship Employment Poverty (M) "Bottomup" (F) Federal Programs welfare actions (M) Municipal Programs Source: Cabrero Mendoza 2003. Basic Education Around 92 percent of students in basic education attend public schools, so it is not a targeted program. Nonetheless, the majority of these students come from poor families, and the basic education program is the largest federal spending program. Although the biggest gaps in educational coverage for the poor are at uppersecondary and university levels, improving the quality of primary education and access to lower secondary education remain priorities, both for their own sake and to provide a foundation for the poor to move up the educational ladder (World Bank, 2004a). The Fondo de Aportaciones para la Educación Básica (FAEB, Basic Education Fund) of Ramo1 33 transfers to states the resources to pay teacher salaries, through an agreement reached in 1993. As an earmarked expenditure, this absorbs more than onethird of all resources transferred to the state governments. Federal programs that supplement the basic education program include Oportunidades, Consejo Nacional de Fomento Educativo (CONAFE), and the Programa 1 The Ramos are the budgetary codes. 4 Decentralized Service Delivery for the Poor Escuelas de Calidad (PEC, Program of Quality Schools). Oportunidades provides a subsidy directly to poor households on condition that the children attend schools. PEC was started on a pilot basis, bringing funds directly from the federal government to schools. Three parties are required to participate: (a) the administrators of the individual school must develop and implement a fiveyear school improvement plan, (b) the state must match the federal contribution at a ratio of 1:2, and (c) the school community must make a financial contribution (which can come from a variety of sources, including parents, municipalities, and the private sector) and participate in the planning process. Communities voluntarily selfselect for the program. CONAFE supports about 25 percent of schools--those in marginalized rural areas--in the development of curricula, didactic materials, and textbooks in an indigenous language and in Spanish to facilitate bilingual education. Health Care for the Population without Social Security Coverage Despite progress in health, half the population remains uninsured and a tenth is without access to health care. For the rural poor, the priority is to provide access to a basic package of preventive and curative health care. Extreme poverty brings with it a high level of mortality and morbidity. Moreover, when families without social security coverage face a severe health shock, they incur catastrophic health costs, which cause them to become poor. Whereas only 2.7 percent of the insured suffered excessive health shocks2 in 2004, 5.6 percent of the uninsured did. The second most important fund in Ramo 33 is the Fondo de Aportaciones para los Servicios de Salud (FASSA) or Health Services Fund. FASSA was established in 1997 to finance states, which became responsible for paying the salaries of health workers who attend the population without social security coverage. The public programs that go mostly to the population without social security coverage (mainly the poor) include the Ministry of Health (SSA), Instituto Mexicano de Seguro Social (IMSS)Oportunidades, and Seguro Popular. The Ministry of Health has an extensive geographic coverage--more than 12,000 health clinics in the 32 states. IMSS Oportunidades provides health services in marginalized rural and urban communities, using facilities and personnel arranged by IMSS but not part of the mainstream IMSS system. Seguro Popular purchases services from SSA and IMSSOportunidades. Seguro Popular aims to improve equity both at the state and at the household levels. At the state level, it seeks to achieve equality in public health spending per capita across states (as well as across public health insurance schemes). At the household level, it aims to achieve full coverage of the population without social security coverage, minimizing vulnerability to catastrophic or impoverishing health expenditures and, more generally, ensuring equality of basic health opportunities. The program intends to reach these 2 Excessive health expenditures include catastrophic (over 30 percent of household disposable income) and impoverishing (those that are not catastrophic yet, but cause a family to fall below the poverty line) health expenditures. Decentralized Service Delivery for the Poor 5 objectives by 2010. Evaluation of the program's impact is under way. However, preliminary results indicate that Seguro Popular's targeting could be substantially improved, for example, by following closely the targeting of Oportunidades. Social Infrastructure in Marginal Municipalities In Mexico, there are large differences in standards of living among states as well as among municipalities within states. These differences could be due to differences in the characteristics of the households living in various areas, such as level of education and occupation, but they could also be due to characteristics of the areas themselves, such as infrastructure. More precisely, the characteristics of an area can have both direct geographic effects (significant geographic determinants of income or consumption that are observed even after controlling for a wide range of household characteristics that may be due to climate, access to markets, demand for labor, or industry concentration) and indirect geographic effects (which may affect household endowments, such as availability of schools or opportunities) on the wellbeing of its inhabitants. According to previous research, there are indeed both direct and indirect geographic effects on standards of living at the state level in Mexico. The geographic effects are relatively stable over time and coherent with observed migration. This provides some rationale for investing in the physical capital and social infrastructure of poor areas, since area characteristics do matter for poverty reduction. The modalities of providing and funding social infrastructure vary widely in Mexico. During the early 1990s, investments in social infrastructure were driven by the Programa Nacional de Solidaridad (PRONASOL), which was devoted largely to basic social infrastructural investment. One of the program's most remarkable elements was its demanddriven mechanism, using the organized participation of local communities, which bypassed municipal and state governments. PRONASOL was discontinued following accusations of intensive political misuse. In 1996 and 1997, a public formula was used for the first time to distribute social infrastructure investments through a fund named Fondo de Desarrollo Social Municipal, which was later institutionalized under the Fiscal Coordination Law of 1998 and renamed the Fondo de Aportaciones para la Infraestructura Social (FAIS) or Social Infrastructure Fund. FAIS, the third largest fund in Ramo 33, uses a transparent povertyweighted formula to distribute resources for investment capital to states and municipalities. The amount was initially decided annually in the budget process, but starting in 1998 the Fiscal Coordination Law specifies that FAIS should be 2.5 percent of the projected recaudación federal participable. The money goes initially to the state treasuries, but with strict conditions to pass along 87.8 percent to municipalities, under the title Fondo para la Infraestructura Municipal (FISM), with the rest remaining at the state level under the title Fondo para la Infraestructura Social Estatal (FISE). The distribution across states and to the 6 Decentralized Service Delivery for the Poor municipalities within states is set by formulas in the federal law. FISE has no earmarking (it has to be used for regional or intermunicipal projects), and FISM can be used in any of the following areas: water, sewerage and sanitation, municipal urbanization, rural electrification, basic health infrastructure, basic education infrastructure, housing improvement, rural roads, and rural productive infrastructure. There is no tight monitoring of these funds, and most municipalities also put funds from other sources into these areas. Decentralized Rural Development To overcome poverty, the rural poor need more than transfers like those from Oportunidades or private remittances; they also need sustainable income and employment opportunities. Many federal programs are oriented to this. Indeed, the federal government spends an enormous amount in rural areas. In the 2005 expenditure budget, MxP$ 147 billion were allocated to rural areas under the many programs included in the Programa Especial Concurrente. Of this total, MxP$ 48 billion are classified as production support. Hence, an estimate of around US$ 4.5 billion goes annually to support production in rural areas. Alianza para el Campo is the major farm modernization program in Mexico. It is targeted to poor producers and is the only productionoriented rural program that has been partly decentralized to the municipal level. It has been operating since 1996 at an annual cost of more than US$1 billion and represents 17 percent of the Secretaría de Agricultura, Ganadería, Desarrollo Rural, Pesca y Alimentación (SAGARPA) budget. It is a demanddriven program for investment support, including three major subprograms: Fomento Agrícola, Fomento Ganadero, and Desarrollo Rural (agricultural, livestock, and rural development, respectively). All farmers and farm associations are eligible for this program, which subsidizes variable percentages (often 50 percent) of productive investments in approved projects. Alianza's Rural Development Program consists of three programs. The first is the Programa de Apoyo a los Proyectos de Inversión (PAPIR), an investment fund that supports income and employmentgeneration investments by small producers and other rural dwellers. PAPIR provides subsidies to lowincome rural residents to improve their farms or carry out other productive investments. Until 2005 there were two modalities: (1) support for productive projects, which is restricted to producer groups of six or more members, and (2) support for individual producers. A maximum of 30 percent of PAPIR resources in each state could be allocated to individual producers. In the case of groups, the subsidy for ordinary beneficiaries ranges between 50 and 70 percent, decreasing with the amount of subsidy, and between 50 and 90 percent for priority groups.3 For individual producers, it ranged between 50 and 53 percent. Since 2005 only groups, not 3 The priority groups are youth, women, indigenous peoples, elderly producers, and the handicapped. Decentralized Service Delivery for the Poor 7 individuals, are eligible for support. In 2005 PAPIR accounted for 72 percent of the MxP$ 2.2 billion of federal funds allocated to the Rural Development Program. The other parts of the Rural Development Program are the Programa de Desarrollo de Capacidades (PRODESCA) and the Programa de Fortalecimiento de Empresas y Organizaciones Rurales (PROFEMOR). PRODESCA provides training and technical assistance to help producer groups formulate and implement projects financed by PAPIR. PROFEMOR supports producer organizations and provides technical support to the rural development councils through coordinadores técnicos (technical coordinators). II. INTERNATIONAL EXPERIENCE WITH DECENTRALIZATION OF SERVICES: FROM GRAND HOPES TO COMPLEX REALITIES Decentralization is an ongoing fact of life in Mexico, driven by political as much as by economic reasons. Decentralization became popular in the development community in the 1980s and 1990s. It was often viewed as a solution to the inefficient and inequitable provision of services. To understand the particular features of Mexico's process, this section outlines two of the main arguments for decentralization in the development literature and experience and then relates these to the more complex experience in practice. We highlight two broad categories, both because advocates of decentralization consider them to be important and because they provide a useful intellectual basis for the analysis in this report. Public Finance and the Optimal Location of Service Delivery The first category of issues flows from general concerns about the underlying production functions for delivery. Where preferences or conditions vary across geographic jurisdictions, local service delivery has a potential advantage to the extent that local organizations have better information and can adapt the mix or design of services to local conditions. However, this local advantage has to be set against the possible advantages of delivery by higher levels of government to the extent that there are economies of scale in production or management, there are spillovers in costs or benefits, and provision has the feature of a public good. Other than of a clearly economywide nature--macroeconomic policy and national defense, for example--most public services are at least partially suited to local delivery. In education, curriculum design and testing are public goods and should be primarily national concerns. In contrast, school management requires good local knowledge and should be run locally. In health, treatment norms are public goods, and immunization has externalities; both are typically run nationally. Health centers, in contrast, can be run locally. In infrastructure, a country's arterial transport network has both major spillovers and economies of production and should be under central management. Having common engineering standards for the whole country makes sense and is generally defined nationally. In contrast, lower levels of government are 8 Decentralized Service Delivery for the Poor typically best at running local urban and rural roads. In agriculture, pricing policy is intrinsically national, as is most, though not all, research, whereas water management is a local concern, and extension services may work best locally. There is a tendency to consider whether a whole sector is or should be decentralized, when the more important question is how central and local public action can best complement each other and how the division of labor should be defined in organizational terms. The optimal level of decentralization need not correspond to the core public organizational structure. In education, the most local unit for management is the school, not the municipality. In river basin development, it is typically a geographic region that encompasses more than one municipality and often crosses state lines. This needs to be taken into account in designing management structures. Finally, there can be differences in the capacity or responsiveness of different levels of government. Local government may have potentially more information on local preferences and conditions, but weaker capacity to use this information. This takes us to the second consideration. Agency and Accountability The second category of issues is the extent to which different levels of government respond to local or national concerns. Government action is embedded within a polity and is a form of "agency" problem (see Bardhan and Mookherjee 2005 for a recent review of the relevant literature). Governments hold power and take action formally on behalf of the citizenry, whether in authoritarian or democratic settings. But for a variety of reasons, governments may be imperfect agents of the wishes of citizens; in other words, they may be incompletely accountable to all citizens. They may lack information, they may reflect the interests of some groups more than others (as is classically the case in unequal, stratified societies), or officeholders may pursue their own interests. Decentralization can make a difference in how governments respond to these influences and thus in how they behave. When a national government devolves financing and service delivery responsibilities, it can do so with the requirement and expectation of accountability either back up to the national level or downward to the local citizens. Both strategies have demonstrated successes internationally. As Box 1.1 indicates, Germany follows more a strategy of upward accountability with many administrative controls, while Canada has sought to create more accountability at the local level. Both strategies work best when there is good administrative capacity, and both Canada and Germany are ahead of Mexico in that regard. The Canadian model has less complexity, requires less in terms of obedience to authority, and allows for utilizing the administrative capacity that may be strong in some places; thus it may be more suited to Mexico. Decentralized Service Delivery for the Poor 9 Box 1.1. Decentralization and Accountability for Results in Canada and Germany The constitutions of all federations allocate powers either for selfrule or shared rule, with the degree and type of decentralization varying widely. In Australia, Canada and the United States, subnational governments have substantial legislative authority, especially on matters of expenditure. Germany, in contrast, follows a model of administrative federalism, where virtually all legislative power rests with the center, and the laender have almost all responsibility for implementation and administration. In the Canadian federation, provinces have the authority and responsibility for most of both expenditure and revenue. They levy their own personal and corporate income taxes and their own sales taxes, and they control the natural resources within their borders. With most provinces raising most of their own revenue, both fiscal and political electoral pressures give incentives for officials to assure the quality and efficacy of services. In Germany, the federal legislation sets the standards for decentralized services, and most of the revenue to pay for them comes from taxes set nationally but collected locally. Equalization formulas for redistributing tax revenue per capita dilute the incentives for local tax effort. Nonetheless, the culture of bureaucratic competence and compliance with rules, as well as the political accountability of local officials for implementing the programs, makes the production of services generally efficient and effective enough to meet federal standards. In both countries, the formal accountability is chiefly through the long route of government, with Canada having more of the shorter long route going through local governments. Neither country seems to have a lot of formal programs to create the short route to accountability, but the general vigor of civil society in both countries assures that some of this accountability takes place. Sources: Giugale and Webb 2000; Webb and Zapata 2006. Citizens may influence the performance of local government through two processes, following Hirschman's classic approach: through "exit" (linked to the mobility of investment and population or labor) and through "voice" (the direct expression of citizen influence on government within a jurisdiction). The "exit" route works through the mobility of capital or citizens: the possibility of attracting more capital or workers can induce local governments to offer better services--in local infrastructure, government processing of regulations, and schooling, for example. The threat of losing mobile capital and more educated workers is, conversely, an incentive for local governments to maintain or improve the quality of services. Thus Nuevo León may seek to provide better services than Estado de México in order to attract more foreign and local investment. This is the essence of the idea that federalism may promote a virtuous cycle of competition, applied, for example, as an interpretation of China's successful development in the 1980s and 1990s (Qian and 10 Decentralized Service Delivery for the Poor Weingast 1997). More recently, there are reports of intense competition among Indian cities in their effort to provide a better business environment for the informatics and softwarerelated industries. This interjurisdictional competition can also take a less benign form for economic development when it involves tax incentives, as in the "fiscal wars" waged by Brazilian states seeking to attract foreign direct investment (Tendler 2000). Mexico has largely avoided such competition by discouraging local taxes since the midtwentieth century (Díaz Cayeros and McLure, 2000). "Voice" can work through a variety of mechanisms, including elections that vote a party out of office, exposés and critiques in the media, citizen complaints, participation in parent schooling committees, street protests, and even, in a few extreme cases, lynching of the mayor.4 Advocates of decentralization argue that local government is closer to the people and so is responsive to their interests and needs, including those of poor people within their jurisdiction. This is linked to the desirability of fostering greater participation, both in formal local democratic processes and in the interaction between social mobilization and the design and implementation of more open government policy. In the past 15 years, Mexico has greatly increased the competitiveness of elections and the possibility of party change at all levels of government. This has directly increased voice and amplified the effect of other modes of voice. Multiple Relationships of Accountability It is useful to analyze the effects of decentralization in terms of the relationships of "accountability" among different levels of government and citizens and within the national and local context. This can be done through a network of relationships of accountability (see Figure 1.3 for a stylized framework). The central idea is that the nature, extent, and quality of service delivery to a particular group of citizens are mediated by relationships of accountability. In this aggregate picture, there are four such relationships: · Between citizens and policymakers ("voice"). Policymakers decide policies, while citizens, to varying degrees, influence them through a variety of political channels. · Between policymakers and provider organizations ("compact"). Policymakers do not themselves deliver services, but they have contractual relationships with providers, including, perhaps, the executing parts of a ministry, a parastatal, or a regulated private company. · Between the managers of the provider organization and frontline workers ("management"). While this relation occurs within an organization--from the 4 Grindle (2005) opens her study of Mexican municipal development with a number of anecdotes in this vein. The lynching occurred in Ilave, Peru, in April 2004. See also the film "La Ley de Herodes" for a fictional account of the cycle of mayoral action and eventual citizen revolt in the heyday of the Partido Revolucionario Institutional (PRI) in Mexico. Decentralized Service Delivery for the Poor 11 Ministry of Education to teachers, for example--a set of factors influences the translation of managerial decisions and the behavior of teachers, nurses, or extension agents. · Between clients and frontline workers ("client power"). The actual transaction occurs between "streetlevel bureaucrats" and citizens, who are themselves a heterogeneous group. This is an interaction between people--in the classroom, clinic, farm, town hall--and the power, or lack of it, of the client can influence the behavior of the worker. Figure 1.3. Accountability Relationships among Citizens, Policymakers, and Providers of Services The State Politicians Policymakers Long Route of Accountability Compact Voice Short Route of Accountability Citizens/Clients Providers Non poor Poor Client Power Frontline Organizations Coalition/Inclusion Management Services Source: World Bank 2004b. As highlighted in Figure 1.3, the clientpower relationship can be termed the "short" route to accountability, as exemplified by the effects of direct community participation or parent councils. The route via citizen voice, policymakers, and provider behavior can be termed the "long" route. While these work in different ways, they are often complementary, rather than substitutes. The compact relationship is not only concerned with accountability for performance, but also with the degree of autonomy and authority that governments (especially in a multilevel situation like Mexico's) have to manage the providers. How does this prism help to interpret the mixed experience of decentralization? It shows that service delivery is far from a technocratic activity; rather service delivery is embedded in a set of social relationships, embodied in formal and informal institutions, and these relationships are both reflected in and influenced by action at the central and 12 Decentralized Service Delivery for the Poor local level. Where policymakers respond to narrow elites, the design of policy will be biased against poorer groups, and implementation will be responsive to such elites. More broadly, in many Latin American countries, including Mexico, government action has historically been embedded within structures of corporatism and clientelism. The design of government policy and the behavior of service providers have aimed largely to protect the jobs of and the services to members within corporate structures and to provide the discretionary patronage associated with powers of office. Chapter 4 customizes the 2004 World Development Report (WDR) framework to reflect the multiple levels of government and other players, but even this cannot capture all the important institutional dimensions. For instance, the local governments need to develop and retain competent personnel. There has been a major increase in the number and complexity of federal programs that rely on local governments to deliver, which reflect a need to consolidate federal programs and streamline regulations, so that local government can more readily access them. Also there is a need to create effective linkages among federal programs and between them and the locally run programs. Mixed Experiences in Decentralization While enthusiasm for decentralization has been high in many quarters, experiences have been mixed. There have been impressive successes--for example, the oftcited cases of Porto Alegre, Brazil, and Bogotá, Colombia--but there have also been many cases in which decentralization has entailed poor performance by local government or had no apparent impact. Decentralization clearly is not itself the solution for service delivery, whether for the poor or for the nonpoor. Nevertheless, it may be a desirable and effective instrument of policy. How it works will depend on both its context and design and, in particular, how the relationships described above interact. III. MEXICO'S HISTORICAL PROCESS OF DECENTRALIZATION Mexico has a longstanding formal commitment to decentralization, embodied in the 1917 Constitution's explicit endorsement of "free municipalities"--or municipios libres--and its federal character. Despite this formal commitment, for much of the 20th century Mexico was highly centralized, both in political and administrative terms. Political centralization was part and parcel of the dominance of the Partido Revolucionario Institucional (PRI), which effectively had exclusive power at all levels of government from the end of the 1920s until the early 1980s and held power at the national, federal level until 2000. The PRI was a hierarchical party, tightly allied to a range of vertical, corporatist structures for organizing society, notably among worker groups. Administrative centralization was reflected in the concentration of power in the national executive branch of government--in particular, the presidency--with very limited devolution of tax and spending authority to state and municipal levels. Decentralized Service Delivery for the Poor 13 The PRI's control of all government structures for many decades reinforced this centralizing tendency. An example of this is the PRI's management of subnational offices. While elections were held at state and municipal levels, the PRI managed both the choice of candidates and, to a large degree, the electoral process itself in the beginning of the 1970s. Ironically, the noreelection rule (threeyear terms for mayors and sixyear terms for state governors) only served to increase the PRI's hierarchical power. This rule was originally conceived as a democratizing move, but in practice it made politicians dependent on the party rather than their electorates: the PRI managed candidacies and appointments as part of an overall strategy of vertical control. Just as centralization reflected the intertwining of political and administrative structures, so decentralization of service delivery was a product of two interrelated processes: the gradual transition from an authoritarian to a democratic polity and a series of specific decisions on taxes, sectoral spending, and associated laws and regulations. Democratization started at the local level, as manifested by real competition between political parties. Whereas the presidency remained in PRI hands until 2000, a sprinkling of municipalities and states were controlled by other parties beginning in the late 1980s. By 1997, more than 30 percent of municipalities and more than 20 percent of states were run by either the Partido Acción Nacional (PAN) or the Partido de la Revolución Democrática (PRD), and by 2003, these proportions had risen to more than 40 and 50 percent, respectively (see Figure 1.4) 14 Decentralized Service Delivery for the Poor Figure 1.4. Percent of Municipalities and States Governed by Parties Other than the PRI, 19852003 Source: Grindle 2005. The midterm elections of July 1997 also saw the PRI lose its majority in the lower house of the national Congress. The shift in power from the executive to the legislative branch continued in 2000. This is relevant because regional interests, as well as some important corporate interests (the teachers union and farmers groups), wield significant influence within the Congress. Administrative decentralization started before the democratization process. For the purposes of this study, the main events were as follows. The 1978 Fiscal Coordination Law set the legal framework for centerlocal relations. Its central feature was that states and municipalities would forgo collecting a set of federal taxes, levies, and contributions in return for receiving an agreed share of total revenues. The shares of the participaciones are legally owned by the states and municipalities. They are calculated under transparent formulas and passed on to lower levels of government in the federal budget under Ramo 28. They constitute the major source of nonearmarked funds for almost all state and municipal governments, accounting for around a third of total revenues at both levels.5 5 Only Nayarit and the Distrito Federal have comparable levels of own revenues. Decentralized Service Delivery for the Poor 15 Lower levels of government also receive resources from individual federal ministries and agencies under specific convenios. These are specific agreements involving the transfer of resources to states for defined uses (at times with a requirement to pass portions on to municipalities). Unlike participaciones, they are federal resources and subject to supervision, audit, and sanction by federal authorities. They have the advantages and disadvantages of falling under the discretionary power of federal bodies: this gives them considerable flexibility as a tool for sharing program management and providing incentives for lower levels of government. But they lack the transparency and clarity of formuladriven allocations and provide opportunities for rent seeking and influence peddling, enabling lower levels of government to seek additional revenues from higher levels of government rather than encouraging them to generate funds locally. Convenios are used in a variety of areas and are still relevant to many areas of spending today, including all decentralized spending under agriculture, for example, but they constitute only about 3 percent of the federal budget. Beginning in the early 1990s, there was growing recognition of the inefficiencies of a service delivery system dominated by federal spending. Changes were initially made in a number of sectorspecific areas. The three most important (and relevant to this study) were the following: · The PRONASOL program, introduced in 1989 by President Salinas de Gortari for the delivery of social and economic infrastructure, was a federal program under the Ministry of Social Development that encouraged local participation while largely bypassing local government structures. · The 1992 Acuerdo Nacional para la Modernización de la Educación Básica¸ signed by the Ministry of Public Education (SEP, Secretaría de Educación Pública), the national teachers union, and the state governments, involved the transfer of financial resources for running basic schooling. · The 1996 Acuerdo Nacional para la Descentralización de Salud, signed by the Ministry of Health, the health sector unions, and the state governments, involved the transfer of financial resources for running the governmentmanaged part of the health system. These programs were incomplete as they responded to partial objectives. And in the case of PRONASOL, there was a significant backlash, owing to its blatant political misuse.6 Indeed in 1996, under President Zedillo, a new formula was introduced under which twothirds of PRONASOL's budget was distributed to lower levels of government. The PRI's loss of dominance in the lower house of Congress in 1997 and rising demands, especially from nonPRI governors, for greater transparency and less 6 While statistically it shows up as a tool for support of the PRI (Díaz Cayeros and Silva 2004), Salinas used PRONASOL as a way to build a system of personal loyalties to himself, outside of and counterbalancing traditional PRI structures. 16 Decentralized Service Delivery for the Poor discretion in centerlocal budgetary relations laid the basis for a more comprehensive change. The December 1997 reform of the LCF introduced a new instrument: sector specific transfers to subnational governments, or aportaciones, to be passed from the federal budget under a new budgetary code, Ramo 33. These transfers are intermediate in character between participaciones and convenios. Like participaciones, they are assigned by predictable rules or even transparent formulas, which provide state and municipal governments with a stable and certain amount of resources. But like convenios, they are earmarked for specific sectoral uses, initially under five headings: · FAEB, which subsumed the 1992 agreement on basic education · FASSA, which subsumed the 1993 agreement on the government segment of the health system · FAIS, the successor program to PRONASOL for the provision of local social and economic infrastructure · Fondo de Aportaciones Múltiples (FAM), which subsumed two federal programs for school feeding and school construction · Fondo de Aportaciones para la Educación Técnica y de Adultos (FAETA), which took over federal programs in technical and adult education. Smaller programs were added later. This report focuses on the first three, which account for the bulk of resources transferred to state and local governments under Ramo 33 and whose clients mostly are in the poorer deciles of the income distribution.7 It assesses the overall design of decentralization for four broad service categories-- education, health, local infrastructure, and rural development (outside of Ramo 33)--that together involve financing from a variety of sources: participaciones, convenios, aportaciones, and own funding from local resource mobilization. 7 The two main categories of transfers to the states are participaciones and aportaciones. Participaciones were originally state and municipality revenues whose collection was delegated to the federal level in the Fiscal Pact for tax efficiency reasons. Most of these transfers are distributed under Ramo 28. Aportaciones were conceived as federal money earmarked to pay for (formerly) federal commitments and transferred to the states and municipalities together with those commitments (for example, education and health). These funds, formerly under Ramo 26, now go out under Ramo 33. Since 1998, Congress has been incorporating in the budget the Programa de Apoyo para el Fortalecimiento de las Entidades Federativas (PAFEF) without any clear mechanism in the allocation of these resources across states. The initial purpose of PAFEF was to strengthen the education infrastructure in those states in which the state governments were not allocating enough resources. Therefore, these resources were earmarked for education infrastructure if the states did not allocate adequate resources for this; otherwise, they could use it for other type of infrastructure. In 2000, the rules changed so that a state could use it for infrastructure investment or to cover financial imbalances from the pension fund accounts. In 2002, Congress formalized this program, which now will be disbursed under Ramo 39, and modified its rules so that states can use these resources for infrastructure and debt relief. The size of the program is relatively small, about 1.5 percent of the recaudación federal participable. Decentralized Service Delivery for the Poor 17 IV. COMMON CHALLENGES ACROSS SECTORS Mexico's decentralization process has been characterized by three features: strong presence of unions in health and especially education, the lack of clear and stable definitions regarding the responsibilities of each level of government in the provision of services, and emphasis on the expenditure side of the federal budget. These three factors have shaped outcomes, both at the macro and microeconomic levels. To move forward, the roles and responsibilities of each level of government will have to be clearly defined. Since the 1997 reforms to the LCF, the federal government has initiated new programs or continued with old ones in areas that clearly should be the responsibility of subnational governments. This generates duplication of administrative costs and sends a confusing signal to citizens as to the level of government responsible for services. It can also generate perverse incentives that induce subnational governments to shirk some of their responsibilities and shift costs to the federal government, while reaping political benefits. The proliferation of programs and duplication of tasks are not the most efficient and effective ways to use resources, hold subnational governments to their responsibilities, and induce them to develop the administrative and technical skills required for these tasks. All levels of government should engage in a collaborative process with clear roles and responsibilities based on their comparative advantage. In some sectors, ongoing discussions are taking place about the level of decentralization. Further discussion clearly is needed based on the national and international experience of programs. For example, in education should the model be to decentralize at the school level, while strengthening capacity at all levels of government for planning, supervision, and so forth? In rural development the right model may be to decentralize not to the municipality but perhaps to the micro region. Institutional and administrative fragmentation and duplication of programs have made the decentralization process even harder. For example, in rural development, the lack of coordination of federal programs and policies makes it difficult to undertake planning and allocate resources at the state level. In health, systemic fragmentation and rigid organization of public provision have led to "producer immunity" from emerging accountability instruments. These characteristics add to the cost of the system and constrain the emergence of choice and competition as sources of pressure to perform. They further block professional management of public facilities because the absence of authority limits the facility's director to simple administration. 18 Decentralized Service Delivery for the Poor CHAPTER 2 FISCAL OVERVIEW This chapter reviews the distribution of federal, state, and municipal funds for basic education, health for the population without social security coverage, social infrastructure, and agricultural development programs in the context of decentralization. The federal government spends about 22 percent of its budget on fiscal transfers for basic education, health for the population without social security coverage, social infrastructure, and a rural development program (see Figure 2.1 and Table 2.1). Basic education and health are some of the most progressive programs in Mexico, giving more benefits to the poor than to the nonpoor (see Figure 2.2). Figure 2.1. Federal Government Expenditures, 2004 14,000 12,000 Other Federal Expenditures 10,000 Direct Federal ta Expenditures PAFEF 8,000 capi per P R28 Mx 6,000 Rural Dev. 0042 Social Development Intergovernmental 4,000 Transfers Health Direct Federal 2,000 Expenditures Education Transfers Uninsured Transfers 0 Total Federal Total Federal Education Health Population Social Rural Spending Spending Development Development Note: The intergovernmental transfers are conformed by Ramo 28, Ramo 33, and PAFEF. Source: SHCP and World Bank staff calculations. Decentralized Service Delivery for the Poor 19 Basic education, health for the population without social security coverage, and social infrastructure are financed through Ramo 33, which consists of eight funds. The most significant fund under Ramo 33 is FAEB, which transfers resources for the payment of teacher salaries through an agreement reached in 1993. As an earmarked expenditure, this absorbs more than onethird of all resources transferred to the state governments. The second most important fund is the FASSA, which was established to finance states' decentralized responsibility for paying the salaries of health workers who attend the population without social security coverage. In addition, the states and municipalities receive an amount of funding for social infrastructure through FAIS. Table 2.1. Programs as a Share of the Federal Budget, 2004 Program Percent of the budget Basic education 15.3 FAEB 11.8 Distrito Federal 1.2 Other 2.3 Health for the population without social security 4.6 coverage FASSA 2.5 Ramo 12 1.7 IMSSOportunidades 0.4 FAIS 1.7 Alianza para el Campo 0.5 Rural Development Program 0.2 Other 0.3 Source: SHCP. Alianza para el Campo is the major farm modernization program in Mexico, operating since 1996, with an annual cost of more than US$1 billion and representing 17 percent of SAGARPA's budget. It is a demanddriven investment support program, including three major subprograms: Fomento Agrícola, Fomento Ganadero, and Desarrollo Rural. All farmers and farm associations are eligible for this program, which subsidizes variable percentages (often 50 percent) of productive investments in approved projects. 20 Decentralized Service Delivery for the Poor Figure 2.2. Concentration Coefficients, National, 2000 and 2002 ISSSTE Pensioners ISSSTE Active Workers GINI 2000 Tertiary Edu. ISSSTE Health 2002 VAT Fiscal Spending IMSS Active Workers IMSS Pensioners PEMEX Health Hospital Health Electricity Subsidy IMSS Health Upper Sec. Edu. Maternal Health Primary Health TOTAL Lower Sec. Edu. PROCAMPO Primary Edu. Preschool SSA Oportunidades 0.8 0.6 0.4 0.2 0.0 0.2 0.4 0.6 0.8 Note: The most common measure of absolute progressiveness is the concentration coefficient (C), or quasi Gini, which is a Gini measure derived from these curves. C is defined in the (1, 1) interval, where, in the case of transfers, negative (positive) values represent progressive (regressive) allocations. Source: SHCP, ENIGH, and World Bank calculations. I. MEXICO'S DECENTRALIZATION PROCESS AND NORMATIVE PRINCIPLES Over the past 21 years, since the administrations of de la Madrid and Salinas, Mexico has embarked on a policy of increasing decentralization. This process accelerated with the advent of the Zedillo administration in December 1994. A permanent feature of this process has been the lack of a clearly designed decentralization strategy. Although nominally the states and the municipalities have autonomy to formulate their own budgets and spend their own funds any way they wish, in reality the lack of revenue autonomy and the preponderance of conditional and categorical grants with high levels of micro management from the upper level of government make Mexico resemble a centralized state that is pursuing deconcentration rather than decentralization. The driving forces for true decentralization have been the advent of political pluralism in addition to the need to address the rising level of public debt held by some states and the need to improve efficiency and rationalize the delivery of services in key sectors, such as social welfare, education, health, and transportation. The end result of this process has been a system of intergovernmental transfers that seeks to satisfy eight normative principles. First, intergovernmental grants should complement the associated expenditure and tax allocations in ways that make the overall system resonate well with the needs and underlying social values of the Mexican federation and society. For example, the Decentralized Service Delivery for the Poor 21 importance of relatively equal access to certain national public goods like education implies that the federation will earmark grants for those purposes. Second, intergovernmental grants should deliver the conception and degree of equity embedded in the Mexican "political contract." This means that the equalizing or horizontal transfers to states should be tailored to deliver this degree of subnational equity. As an important aside, equity does not relate to the grants in isolation: it relates to the sum of the transfers and all other state revenues. That is, the states' overall fiscal capacity should be brought up to the equity standard. Third, the transfer arrangements should be predictable over time. The set of transfers should be designed within a multiyear framework in order to allow the states a corresponding multiyear planning and budgeting horizon. Although absolute predictability cannot be guaranteed, an acceptable initial compromise might be a federalstate agreement over three or five years with provisions requiring adequate notice for any changes. The fourth principle is a variant of the last: formulas should determine transfers as much as possible. This implies that the procedures for calculating the formulas should be open and transparent, such that they can be reproduced by third parties. To facilitate this, Mexico should consider establishing a formal or informal agency representing both the federal government and the states. Australia's Commonwealth Grants Commission is the preeminent exemplar of a formal, quasiindependent agency, while Canada's technical working group of federal and provincial officials represents the informal variant. An important role for formulabased transfers is to depoliticize, at least to a degree, the overall transfer regime. These formulas will presumably be altered over time as conditions merit. But at any point in time, the degree of objectivity and transparency of formulabased transfers tends to defuse what is, in effect, a zerosum redistributional game. The fifth principle relates to the nature of the conditions attached to grants associated with expenditure areas. Consistent with meeting the expenditure goals, the conditions should respect state priorities. For example, different states may be able to satisfy these conditions in ways that relate to the particular social, cultural, and economic needs of their respective constituencies. This argues not for a onesizefitsall approach to conditionality but rather for conditionality defined in terms of "equivalencies." In the fiscal federalism literature, this has come to be associated with the term "competitive federalism" or "horizontal competition," which means allowing states sufficient flexibility to design their own bundle of goods and services consistent with the agreedupon conditions or, preferably, equivalencies. This should tilt the federation in the direction of dynamic efficiency, since demonstrably superior approaches to policy design in some states will be copied by other states. The focus here is on conditional grants: we take it as axiomatic that unconditional grants will be spent in ways that reflect state priorities. 22 Decentralized Service Delivery for the Poor The sixth principle relates to the objective of tax decentralization. Whatever the approach finally taken to equalization (or to the amelioration of horizontal imbalances), the incentives in this system should ensure that states that increase their tax effort or increase the effectiveness of their collection should be able to retain a meaningful share of the additional revenue collected. Even in developed federations one can find equalization programs that embody 100 percent (that is, confiscatory) taxation with respect to subnational government revenue increases--transfers fall fully in line with revenue increases. The relevant formulas should prevent this and should retain some incentive for local tax effort. The seventh principle also anticipates greater tax decentralization. The overall system of grants should be designed in a way that anticipates and accommodates the desire of states to enhance tax revenue. This means that the transfer system should allow an increase in states' access to taxes accompanied by an appropriate decrease in federal taxes and transfers to the states, thus making new tax room available to the states in an overall revenueneutral manner. Any such design should respect principles three and six in the process. Finally, but hardly exhaustively, the design of a new grant should take into account the status quo. Part of the rationale here is that the current set of transfers presumably embodies the implicit social contract. This must inform the evolution of transfers. Furthermore, to be politically feasible, any novel transfer arrangement must be accompanied by adequate and acceptable arrangements for transition from the old to the new system. II. REVENUE AND EXPENDITURE TRENDS Since taxing power was centralized in the middle of the twentieth century, almost all of the resources available to Mexico's states and municipalities have come from the federal government. Thus the distribution of subnational government resources is an active concern of the federal government, which seeks to assure that states can meet their spending obligations, especially in executing social programs decentralized since the early 1990s, and that subnational governments have appropriate incentives for efficiency, fiscal discipline, and tax effort. The transfers from the federal level come in two main forms: discretionary participaciones and sectorally earmarked aportaciones. Participaciones--the 20 percent of federal domestic tax revenue that is shared with all the states and municipalities--have accounted for almost 40 percent of state and municipal revenue in recent years. Their advantage from the state's point of view is the complete flexibility in their use. Aportaciones in the seven funds of Ramo 33 constitute almost 45 percent of state revenues and more for most municipalities. Some are automatically transferred to the municipalities; the largest ones are transferred to the states and are earmarked for Decentralized Service Delivery for the Poor 23 payrolls in education and health. Box 2.1 further describes the formulas for the allocation of resources. Box 2.1. Formulas for Allocating Funds to the State and Municipal Governments The Fondo de Aportaciones para la Educación Básica (FAEB) appropriation is determined each year in Congress based on wage negotiations between the teachers union, and the federal government, and the registry of staff and schools transferred to state governments. Each year the state secretariats of education negotiate with the federal secretariat in order to get more teaching positions added to their registry. Data on enrollment and schoolage population enters into the negotiation, but there is no transparent formula for the allocation of money or for the new teaching positions. The distribution of resources from the federal government to the states is based on the registry. The Fondo de Aportaciones para los Servicios de Salud (FASSA) appropriation is determined in a way similar to the FAEB but it is based on the registry of staff and medical infrastructure transferred to states governments. The distribution of resources from the federal government to the states has to be done according to a formula which depends upon the difference between the minimum accepted expenditure per population without social security coverage and the states own expenditure on the sector (Fiscal Coordination Law). The minimum accepted expenditure per population without social security coverage depends on the marginality and mortality indexes. The problem is that the distribution of resources is not done through the formula, and it is based upon the staff and infrastructure registry. The Fondo de Aportaciones para la Infraestructura Social (FAIS) should be 2.5 percent of the projected recaudación federal participable in the Ley de Ingresos. FAIS resources are distributed through two funds: FISE, which is distributed to the state governments, and FISM, which is the municipal component. FISM is by far the largest share of FAIS, constituting about 88 percent of it. The share of FISE and FAIS is determined by the Fiscal Coordination Law which establishes a formula for distributing resources from the federal government to the states. The formula uses a measure of Gross State Deficit (GSD, the Masa Carencial Estatal), that aggregates information from indicators for income, education, housing, sanitation, and electricity at the household level. The Fiscal Coordination Law establishes two formulas for allocating funds from the state to the municipal level; each state decides which formula to use. The first formula uses the indicators of the GSD to define the share for each municipality. The law also provides a simpler formula for states that do not have sufficient information. The alternative formula entails four equally weighted components.* The Rural Development program federal funds are allocated to states according to a formula that includes variables such as rural population, agricultural GDP, harvested land, irrigated land, number of production units, and state contribution. 24 Decentralized Service Delivery for the Poor * These components include: the employed population with less than two minimum wages, the illiterate population 15 years or older, population in settlements without sewerage, and population living in settlements without electricity. The volume and distribution of these decentralized resources changed substantially over the past decade, although the total amount did not increase from 2001 to 2002. The subnational governments received transfers equal to about twice as much per capita in real terms in 2004 as in 1992, before the major sectors began to be decentralized (1993) and the last main revision of the participaciones formula was finalized (1994). Figure 2.3 shows the total resources going to states in 2004; Figure 2.4 shows the closest equivalent data for 1992, excluding the federal investment and states' own revenues. The states are ranked by the Consejo Nacional de la Población (CONAPO, National Population Council) index of marginality for 2002, which is the best available statebystate indicator of poverty levels.1 Although there is a lot of inequality among states in the resources transferred to their governments and in the direct spending of the federal government, the inequality in states' public resources per capita is not correlated with poverty levels. Indeed, the relative position of the poorer states (higher indexes of marginality) has improved since 1992, so that, on average, they receive about as much as the national average. At all levels of marginality, the variations depend partly on the size of the state's population.2 1 Unless otherwise specified, all the figures use this ranking. The marginality index, based on the census, which has high statistical coverage for each state, reflects the percent of households lacking access to piped water and sanitation and having dirt floors. It is a proxy for poverty. 2 The resources for the Distrito Federal include direct federal spending for teachers, which is shown as part of aportaciones, in order to be compatible with the representation of other states. Decentralized Service Delivery for the Poor 25 Figure 2.3. States' Public Resources per Capita, 2004 12,000 Own Revenue Participaciones Aportaciones Convenios PAFEF 10,000 8,000 at capi r pe 6,000 MxP 0042 4,000 2,000 0 IS R O E P B O Y N O R X L N S IH H T BC A BC NL DF CH GRO JAL OAX VE SLP HG PU M TA CH YUC ZAC GT SI NA QRO DG MO ME CO SO AGS NA CA MI TLAX QROO CH TAMPS CO Source: SHCP and World Bank staff calculations. Figure 2.4. States' Public Resources per Capita, 1992 12,000 Own revenues Participaciones Aportaciones 10,000 8,000 at pi car pe Px 6,000 M 2004 4,000 2,000 0 sapai acax case albe oglad ocsab e al a a ra echp cán ca tiray sa oc ocix soler hu oa aolan ainro Ch Oa cataZ orerre Pu ístooP Gu isuLnaS Hi catánuY zurcar Ve Ta hua óneLo limoC onoS laredeF amC ngorauD chi axlT oatuaj Na lisaJ Mé estneila Si M anuG étaroreuQ ooRantainuQ lipuamaT Mo sc ilauhaoC Chi guaA evuN ruSainro lifaC lifaC jaaB jaaB itortsiD egarevAlan tioaN Source: SHCP and World Bank staff calculations. 26 Decentralized Service Delivery for the Poor Education is the biggest program analyzed in this report. In 2005, the three levels of government spent 5.7 percent of GDP on education, and close to 72 percent of this was allocated to basic education. In 1990, education expenditures were just 3.7 percent of GDP. Health expenditures for the population without social security coverage were close to 0.9 percent of GDP. This figure more than doubled over the past decade, and in the last two years the Ministry of Health increased its expenditures almost 47 percent. Figure 2.5 shows that FAIS resources have remained relatively small and stable at 0.35 percent of GDP, on average. Expenditures on the Rural Development Program, which is the public infrastructure component of Alianza, have been relatively small. Figure 2.5. Expenditure Trends in Education, Health, FAIS and Alianza 6% Education Health FAIS Alianza 5% 4% GDP of e 3% ar sh As 2% 1% 0% 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Note: Expenditures in health and education are for the whole sector and by the three levels of government. FAIS was created in 1998, so the resources before this year were from the PRONASOL program. Expenditures in Alianza are only for the Rural Development Program and by the state and federal government. Source: SHCP. III. EDUCATION Education expenditures have been growing in real terms at a rate of 3.4 percent a year on average. Figure 2.6 shows that the three levels of government increased education expenditures from 2004 Mx$10,600 per student in 1992 to Mx$15,500 per Decentralized Service Delivery for the Poor 27 student in 2004.3 During the 1994 crisis, there was a substantial drop in spending, and the share financed by states has been increasing over time, while the municipal share has remained very small. The bulk of education spending is budgeted and distributed "inertially," based on the stock of physical capital and employees belonging to the federal system before decentralization. This allocation does not consider the demand of clients, the productivity of suppliers, or poverty (see Figure 2.7; World Bank 2004a, 2002; Merino 2003; OECD 2005b).4 Also, this allocation does not satisfy the compensatory objectives established by the National Agreement for the Modernization of Basic Education (Esquivel, 1999). Despite greater transparency in the allocation of resources through FAEB, marked differences among states persist in the proportion of federally financed transfers and the state's own resources spent on basic education. Figure 2.6 shows state resources as a proportion of total public spending on basic education. On average, states contribute 22.5 percent of total public spending on basic education apart from the funds obtained from federal transfer through FAEB, FAM, and FAETA. However, differences among states in this regard remain huge. While states like Baja California and Estado de México contribute more than 40 percent of basic education expenditures, several other states contribute less than 10 percent. 3 The denominator is the number of students attending public schools only. 4 Figure 2.7 shows the major federal expenditures in basic education. Most of the states received on average almost the same amount of resources per student. Some states are well below the average such as the Estado de México, possibly due to the size of the federal education system before decentralization. Resources are not correlated with poverty. 28 Decentralized Service Delivery for the Poor Figure 2.6. Education Expenditures, 19902004 16,000 Federal State Municipal 15,000 14,000 ent 13,000 stud per 12,000 MXP 4 20011,000 10,000 9,000 8,000 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Source: SHCP. These differences are mainly the result of the number of schools sustained by the state and federal governments before 1992, when the National Agreement for the Modernization of Basic Education was signed. As pointed out previously, the expansion of federal schools occurred at a different pace throughout the country, largely as a result of historical decisions by state and federal authorities. By basing federal transfers after decentralization on the registry of staff and schools transferred to state government in 1992, the distribution of federal funds perpetuates a distribution of federal funds that is not related to the number of students, the cost of the education provided, the relative level of development of the states, educational lags, or the efficiency and quality of educational services, particularly with regard to the expansion of education services sustained by the federal or state governments. Decentralized Service Delivery for the Poor 29 Figure 2.7. Principal Federal Expenditures on Basic Education, 2004 20,000 FAEB FAETA FAM SEP Convenios 18,000 16,000 14,000 t 12,000 denu St r 10,000 pe MxP 8,000 6,000 4,000 2,000 0 IS R E P C C Y N O R X L N L S IH H T BC A BC NL DF CH GRO JA OAX VE SLP HGO PU M TAB CH YU ZA GTO SI NA QRO DG MO ME CO SO AGS NA CA MI TLAX QROO CH TAMPS CO Note: States are arranged with the highest poverty rates on the left and lowest on the right, according to the CONAPO index of marginality. Source: SHCP and SEP. Taking into account the total amount of public spending per student reveals considerable differences among states in the amount of total public spending per student on basic education. Expenditure per student in Baja California and Baja California Sur is, for example, considerably (60 percent) above the national average (Figure 2.8).5 5 Future research includes more analysis of spending per student at the state level. 30 Decentralized Service Delivery for the Poor Figure 2.8. State Expenditure from Own Resources on Basic Education as a Percentage of Total Public Expenditure on Basic Education, Average 19982001 45.0 40.0 35.0 30.0 25.0 % 20.0 15.0 10.0 5.0 0.0 niar ocix s lifoaC Mé óneLo hua aron co nátac albe la ngoa sapai al s as ca tiray soler a ro síot ehce nte ruS oglad acax huaihC So lisaJ aolan ocsab ia ooRa Si oatuaj Yu Pu ilauhaoC urD Ta onicaN zurcar ax Ve Ch Tl nácaohc caeta Na Zac Mo limoC rreeuG Posi mp Hi Oa jaaB uevN anuG io Mi lipuamaT ed Lun Ca oaréteruQ lifornaC taninuQ omrP Sa aliecsaugA jaaB Source: World Bank estimates based on data from SEP and SHCP. Figure 2.9. Public Expenditure per Student on Basic Education, 2002 BCS Baja California Durango Nayarit Coahuila Colima Sonora Campeche Tamaulipas Distrito Federal Nuevo León Zacatecas Chihuahua Tlaxcala Sinaloa Quintana Roo State Michoacán Federal Yucatán Tabasco Guerrero National Average Veracruz Chiapas Aguascalientes Hidalgo San Luis Potosí Morelos Oaxaca Jalisco Querétaro México Guanajuato Puebla 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 MxP per student Source: World Bank staff estimates based on data from SEP and SHCP. Decentralized Service Delivery for the Poor 31 There does not seem to be a relationship between expenditure per student and the proportion of federal or state financing (Figure 2.9).6 While expenditure per student in Baja California and Baja California Sur is almost equal and the highest in the country, in Baja California the state government contributes approximately 42 percent of total expenditure, whereas in Baja California Sur the state contributes barely 4 percent. In contrast, while the state's contribution in Baja California and Estado de México is in both cases above 40 percent, the expenditure per student in Baja California is more than twice that in Estado de México. These disparities in state contributions have been discussed since the beginning of the decentralization of basic education. Particularly, states that contribute more resources to education seek greater recognition and compensation by the federal government through higher transfers. However, one of the issues that federal educational authorities face is that an additional budgetary allocation for states with a high participation in the financing of education will not necessarily represent a greater financial effort on education, especially if the additional federal resources replace part of the existing state effort. Furthermore, an additional allocation of federal resources to states with higher state participation in the financing of education does not necessarily have the largest impact on the coverage or the quality of education (see Figure 2.10). Figure 2.10. Math and Reading Test Results and Expenditures per Student, 2001 Lectura Matemáticas 500 435 495 SIN SIN 430 490 TAM 425 HGO HGO 485 SLP CHIH 420 AGS DGO *odie GTO DGO QRO QR COAH OAXSLP TLAX 480 CHIS TLAX 415 NL BC GTO CHIS JAL *odie NAY TAM NAY om 475 OAX NAL BCS COL COL om 410 CHIHNL COAH prej AGSM MVERICH OR SON QRO NAL M OR JAL 470 YUC prej 405 M ICH SON ntauP ZAC BCS 465 PUE ZAC 400 PUE YUCQR BC M EX ntauP GRO 460 TAB 395 CAM CAM M EX 455 390 VER 450 385 GRO TAB 445 380 6 7 8 9 10 11 12 13 14 15 6 7 8 9 10 11 12 13 14 15 Gasto por alumno en 2001 (miles de pesos) Gasto por alumno en 2001 (miles de pesos) Source: SEP. The federal government retains little authority in education and health other than to negotiate wages with the unions. For teachers, after nationallevel negotiations are finished every year, each statelevel section of the union negotiates an additional package of benefits with its state, and it might be that the state that gives the most 6 In Figure 2.9, states are ranked from the one that spends the most per student to the one that spends the least. The proportion paid by the state is not correlated with the amount of total expenditures per student. 32 Decentralized Service Delivery for the Poor concessions sets the national standard, since the union is more coordinated nationwide than the state governments.7 While state accountability to the federal government has some attractions, the Mexican federal government lacks the constitutional authority to demand that accountability, and the practical politics would be difficult if the states could call on local popular support to resist federal encroachment. However, the federal government would be more likely to enlist popular support if it made more effective delegation of authority to the states conditional on statelevel reforms to improve transparency and accountability to the local populace. To do this, the other part of the authority accountability bargaining package could be for the federal government to make increased authority for the states (or perhaps increased transfers not earmarked for wages) conditional on actions like the publication of teacher attendance, monetary allocations, and schoolbyschool data on student performance.8 The last item would require establishment or improvement of standardized testing.9 Another key issue in the discussions of education and health decentralization is how federal resources should be allocated among the various states and within states. Paqueo, LópezAcevedo, and Parandekar (2003) show that there are two approaches to this question for the education sector: "(a) bilateral negotiations between the Federal government and states with little transparency as to the rules; and (b) formula based distribution." However, the first approach can lead to allocation results that appear contrary to policy objectives like equity improvement and greater social inclusion in education. Paqueo, LópezAcevedo, and Parandekar (2003) propose three formulas based on per student allocation10 and find that any of these three approaches would improve not only efficiency but also equity. The box below discusses the international experience with Colombia. Box 2.2. Per Student Allocation of Education Transfers in Colombia Much like Mexico, responsibility for providing basic education in Colombia is decentralized to departments (states) and larger municipalities, which receive 7 The federation also governs expansion of the number of teachers in the federalized system as well as the curriculum. There might be a national standard in terms of benefits accrued by the union at the local level. For example, it might be the case that benefits accrued in Chiapas influence negotiations or results in Nuevo León but this deserves further analysis. 8 The Ministry of Education recently agreed through INEE to do schoolbyschool testing. 9 The state of Nuevo León now makes evaluations of each of its public schools available on the Internet (www.educacion.nl.gob.mx), which illustrates the opening of possibilities and could serve as an example for other states. 10 The three suggested formulas are (1) uniform per student allocation, (2) productivityadjusted allocation, and (3) a multicomponent formula based on the number of students and a supplementary component allocated according to the number of children in the state with special needs (rural, indigenous, or poor). Decentralized Service Delivery for the Poor 33 tied transfers from the national government. Expenditure in education as a percentage of GDP between 1990 and 2004 increased from 2.4 to 5.5, but coverage increased slowly. In order to provide incentives for better performance, law 715 was promulgated in 2001 to increase efficiency in the system and promote quality. There was a transition period of about 2 years between the old and the new formula for the distribution of resources. This process has been gradual and adjusting to meet the targets set up in the law. In 2004, the transfers began to be distributed among territorial entities based on actual enrolment and a per student allocation. Random audits are carried every year to verify the validity of the enrollment numbers reported by the states. Two cases of misreporting were detected in Cartagena and Cali. In response, the Ministry of Education reduced the amount of resources transferred to these states. The law 715 formalized the option to contract educational services with the private sector, based on successful local experiences. The new law has also reformed the teaching career, limiting the teachers' salary and promotion scales and wage dynamics. Under the previous system, teachers would move rapidly up the salary scale, resulting in growing pressure on the budget. With the implementation of the reform, 50,000 new teachers are being appointed under new conditions, representing 20 percent of the total number of teachers currently employed in the public system. Law 715 also guarantees a continuation of the upward trend in spending, which is targeted to increase by 2.5 percent more than GDP growth between 2006 and 2008. Territorial entities that improve efficiency in personnel administration, via higher student/teacher ratios or less expensive teachers, can plow savings back into quality enhancement. This is to be supported by a censusbased evaluation of competencies for grades 5 and 9, which correspond to the end of the primary and lower secondary cycles, in the areas of math, language, natural sciences, and citizen competencies. These are to be conducted nationwide every three years, with the results to be used by schools, municipalities and the Ministry of Education for the design and implementation of plans to improve the quality of education. Law 715 is already working toward improving performance, resulting in 1,085,000 new enrollments in basic and secondary education levels over the past three years. Assessment and investment in quality is expected to result in improved student performance. Source: Volume II. IV. HEALTH Public health expenditures for the population without social security coverage have been increasing in real terms at a rate of 8.4 percent a year on average. Figure 2.11 shows that states and the federal government increased health expenditures in real terms from Mx$1,530.7 per population without social security coverage patient in 1998 to Mx$2,319 per population without social security coverage in 2004. As in the case of 34 Decentralized Service Delivery for the Poor education, the bulk of spending is budgeted and distributed "inertially," based on the stock of physical capital and employees belonging to the federal system before decentralization. Another problem is that the formula is not implemented as it should be as explained in Box 2.1. This leads to inequity because states that are well endowed receive large portions of FASSA funds. Figure 2.11. Health Expenditures, 19982004 3,000 SSA FASSA 2,500 on 2,000 ersp red insunur 1,500 pe xP M 04 201,000 500 0 1998 1999 2000 2001 2002 2003 2004 Source: SHCP. For the health sector, one should not analyze the allocation of FASSA in an isolated way; the states make their own contribution to the sector through Seguro Popular, and the IMSSOportunidades. Figure 2.12 shows the allocation of these resources. With the exception of some states, richer states spend more resources for their population without social security coverage than poor states. This trend, however, is driven by the shares of population without social security coverage in the states, where the richer states have the lowest shares and the poorest states have the highest. Considering all of the resources of the health sector, the federal transfers to the sector through FASSA help to increase equality (see Figure 2.13), in the sense that their inclusion reduces the variance in the allocation of resources per capita. (The figure excludes the high outliers of the Distrito Federal and Baja California Sur.) However, better design of formulas and its effective application could reduce inequality further. Decentralized Service Delivery for the Poor 35 Ramo 12 is a federal item to support the Ministry of Health (SSA), the health component of Oportunidades, and all the federal health funding such Seguro Popular, national health institutes, and federal hospitals. Ramo 12 (excluding the Seguro Popular component) is the main source of inequality. The allocation of resources changes from year to year, but the inequality remains. Seguro Popular initially introduced some geographical inequality because not all states enrolled in the program. However, as coverage of the program progresses it is expected that the inequality in per capita allocations for the population without social security coverage will be reduced. Figure 2.12. Distribution of Federal Health Resources per Capita for the Population without Social Security Coverage, 2004 3,500 Ramo 12 Ramo 33 Seguro Popular State Contribution 3,000 2,500 ons per de 2,000 surninu per 1,500 Px M 04 201,000 500 0 sapai aca ísot albe e op a a e án a echp ecast cal ocix sa a oc hua ruS est Ch errerouG axO veal-lz oglad tiray amil Hi racrueV PosiuLn soler eni ainro Pu ocsab oatuaj aolan ognar Ta acaZ Du Co onorS lisaJ óneLo amC camOed catuY anuG Na Si eagtrAed axlT Mo Mé lipuamaT uahihC ozgaraZ lifaC evuN Sa cán oa chi aroétreuQ ooRantainuQ ianrofliaCajaB uascalgA deal jaaB raledeFotritsiD eragvAdethgie hui W M oaC Note: Ramo 12, FASSA, and state contributions exclude Seguro Popular. Seguro Popular includes federal and state resources. States are arranged with the highest poverty rates on the left and lowest on the right, according to the CONAPO index of marginality. Source: SHCP and World Bank staff calculations. 36 Decentralized Service Delivery for the Poor Figure 2.13. Distribution of Health Sector Expenditures per Capita, 2004 7,000 Ramo 12 Ramo 33 Seguro Popular State Contribution IMSS-Oportunidades IMSS ISSSTE PEMEX 6,000 5,000 at piac 4,000 per Px M 4 3,000 200 2,000 1,000 0 sapai op án as sa ar oc Ch rreroeuG acax evall oglad ísot albe ec otau tiray soler ocix amil Oa uz- Hi acrreV PosiuLn Pu ehcepm ocsab Ta ainrofil óneLov laredeF egaervA Ca camOed catuY aolan cataZ alacxa lisaJ esnteil anajuG Na Si Du Tl Mo Mé Co lipuamaT onoS auhauhi Ch ruSainrofil zaogaraZ Caaj ueN Sa cánao chi agaetrAedoartéeruQ ognar ooRantainuQ Caaj uascagA dealui Ba itortsiD edthgei Ba M ahoC W Note: Ramo 12, FASSA, and state contributions exclude Seguro Popular. Seguro Popular includes the federal and state resources. States are arranged with the highest poverty rates on the left and lowest on the right, according to the CONAPO index of marginality. Source: SHCP and World Bank staff calculations. V. SOCIAL INFRASTRUCTURE FUND (FAIS) Investments in social infrastructure face somewhat different challenges in terms of service delivery. They mostly are and always have been delivered by local governments, in contrast to education and health, where federal programs previously were passed down to the states with personnel, plants, and funding. Also for social infrastructure, the specific services and corresponding outcome measures are not uniquely defined, as they are substantially with education and health. FAIS resources have been increasing in real terms at a rate of 5.3 percent a year. As shown on 0, the federal government increased social infrastructure expenditures in real terms from MxP$165 per capita in 1998 to MxP$223 per capita in 2004. The Fiscal Coordination Law states that the FAIS should be 2.5 percent of the projected recaudación federal participable in the Ley de Ingresos (which is part of the total revenues that can be shared with other levels of government).11 Therefore, FAIS increases over time with total revenues and becomes a cyclical type of revenue for subnational governments. FAIS resources are distributed through two funds: FISE, which is distributed to the state governments, and FISM, which is the municipal component. FISM is by far the largest share of FAIS, comprising about 88 percent of it. The share of FISE and FAIS is 11 Historically, FAIS has constituted less than 2.5 percent of participaciones in some years, which might be due to the difference between estimated and observed. Decentralized Service Delivery for the Poor 37 determined by the Fiscal Coordination Law. In terms of the quality of expenditures, it is not known how the municipalities and states use FAIS resources. The law asserts clearly that the states and the municipalities must report to SEDESOL how they used the FAIS resources. Also, the congress of each state has a comptroller's office (ORFIS), which is independent of the state executive government and monitors how municipalities and the state government use public resources; thus, in theory, these offices have information on how the municipalities and the state spend the FAIS resources. Still, some states do not report to SEDESOL how FAIS resources have been used. Nonetheless, it should be mentioned that it corresponds to SEDESOL to request this information. Figure 2.14. FAIS Expenditures, 19982004 240 230 220 210 ita cap 200 per MxP 190 2004 180 170 160 150 1998 1999 2000 2001 2002 2003 2004 Source: SHCP. FAIS funds are distributed to the states through a formula based on poverty indicators and the states to the municipalities distribute it in a similar way, but the problems are in terms of incentives and in terms on how the FISM is distributed at the state level. The formula rewards the poorest states, but it does not provide the incentives to invest in quality infrastructure. Also, the FISM is not necessarily distributed to the all the municipalities according to their poverty level; for example, a poor municipality in one state could receive less than another municipality with the same degree of poverty in another state if the allocation formula they use to distribute the FAIS resources is different. 38 Decentralized Service Delivery for the Poor Figure 2.15. Distribution of FAIS per Capita, 2004 700 600 500 400 Capita per 300 MxP 200 100 0 IS R P P H Y N R O X L N S IH H T BC A BC NL DF CH GRO OAX VE SL HGO PUE M TAB C O YUC ZAC GTO SI NA QRO DGO MO CO SO JAL AGS NA CA MI TLAX ME QR CH TAMPS CO Note: States are arranged with the highest poverty rates on the left and lowest on the right, according to the CONAPO index of marginality. Source: SHCP. FAIS is the largest program for the provision of social infrastructure (see Figure 2.16), representing about 63 percent of federal expenditures on social infrastructure. Although it is relatively small, about 0.35 percent of GDP and about 10.5 percent of total public investment, FAIS is a major source of revenues for poor municipalities. Decentralized Service Delivery for the Poor 39 Figure 2.16. Federal Expenditures on Social Infrastructure 2% 18% FAIS Education Infrastructure Water supply and sewage Electricity infrastructure 63% 17% Note: This does not include all items considered as social infrastructure in the Fiscal Coordination Law. Source: SHCP. VI. ALIANZA PARA EL CAMPO: RURAL DEVELOPMENT PROGRAM The Rural Development Program expenditures have been increasing in real terms at a rate of 20 percent a year. Participation varies greatly according to state, as does states' own funding. Federal funds are allocated to states according to a formula that includes variables related to agriculture, livestock, and rural development factors in the states as well as to efficiency of operation and state participation.12 The three main Alianza subprograms are decentralized and are largely implemented by the state authorities. State trust funds are responsible for the financial management and, through technical committees, for the approval of farmers' proposals. Since 2003, part of the funding of the Rural Development Program started going to municipal authorities. States generally complain that excessive rules and regulations hinder the implementation and adaptability of the program. Alianza is not a targeted program; only the Rural Development Program is. Figure 2.17 shows the value of public Alianza funding per capita of rural population for decentralized programs. More marginal states tend to receive more funds per capita than less marginal ones. In fact, the correlation coefficient between the ordering of states according to their well being and the average amount per capita of decentralized Alianza funding received in 2001­05 is 0.74--that is, it is positive and large.13 Furthermore, the coefficient increased steadily between 2001 and 2005 from a value of 0.64 in 2001 to a value of 0.78 in 2005. On the whole, according to a targeting criteria based on state marginality, the decentralized funding of Alianza is regressive and increasingly so. 12Details of this formula are provided in Box 2 of the rural development chapter in Volume II of this report. 13Spearman's rank order correlation coefficient. 40 Decentralized Service Delivery for the Poor The Rural Development Program is relatively small in size, but it is comparable with the other programs (see Figure 2.18). It represents about 31 percent of the total expenditures in Alianza for the average since the program started until 2004. Today it is about 40 percent larger than what was invested in Programa de Empleo Temporal (PET). Nevertheless, the latter has been decreasing significantly since 2001. Procampo remains the largest program in agriculture and is about five times larger than the Rural Development Program. Figure 2.17. Decentralized Alianza Funds per Head of Rural Population, Federal and State Contributions, Average 20012005 noitalup 800 700 600 Pol rauRfotinUrep 500 400 300 200 100 MxP 0 Federal State Note: States are arranged in order of decreasing degree of marginality from left to right, according to CONAPO's 2000 marginality index. Source: SAGAR and World Bank staff calculations. Decentralized Service Delivery for the Poor 41 Figure 2.18. Federal Expenditures on Principal Programs in Agriculture, 1996­2004 noita 300 Rural Development Other Alianza Procampo Programa Empleo Temporal pul 250 Po lar 200 Ru of atip 150 Ca r 100 pe P 50 MX 04 20 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 Source: SHCP. VII. TOWARD A NEW FISCAL FEDERALISM THAT WILL IMPROVE THE ACCESS OF THE POOR TO OPPORTUNITIES A workable longterm model for Mexico should be consistent with the existing values and norms as well as with the emerging pressures on the Mexican federation. It has four general principles: tax decentralization, fiscal equity, conditional citizenship transfers, and transitional guarantees. The discussion here mentions only the key features of the model, which is amenable to alteration, and some of the forces driving it.14 Tax Decentralization The first principle is to devolve more taxation authority to the states. In terms of the accountability relationships described in Chapter 1, this approach has several advantages. First of all, transfer dependence obscures the relationship between the sources of funds and the provision of services. For example, state policymakers can deflect the blame for the quality of service provision to the national authorities claiming that funding is insufficient from national transfers. Citizens/clients can then become confused on how to voice concerns over the quality of services. Secondly, power to tax at the state level implies that tax policy making enters the state level political debate. The necessity of paying taxes to the state government provides an incentive for citizens to become more involved in the decision making process for public services and attempt to influence providers: they see that it is the use of "their money" that is under consideration. The problem arising from tax power devolution, if implemented, is that the wellendowed states will have access to more revenues per capita than the poorer states. Thus accommodating one pressure in the federation would create another. What is the best way to address this emerging challenge to horizontal balance? 14 For more detail see Giugale and Webb 2000; World Bank 2004e; and Webb, Gonzalez and Rosenblatt, 2004. 42 Decentralized Service Delivery for the Poor Equalization and Fiscal Equity The second principle or component of the model is an equalization program to offset the horizontal imbalances arising from increased state taxation autonomy. Three implications follow. First, these equalization payments must be unconditional (without central government earmarking), since the tax revenues of the richer states are likewise unconditional. Second, the existence of an equalization program will (or should) mean that all states are in favor of tax decentralization--richer states because they can pocket their per capita revenues in excess of the national average and poorer states because they receive unconditional equalization payments that bring their per capita revenues up to this national standard. Third, this can be a revenueneutral exercise from the federal government's standpoint. The enhanced taxation autonomy of the states as well as the equalization transfers can come out of reductions in the existing participaciones. Although this lowfiscalcapacity compensation is termed an equalization payment, in reality it is more like a social transfer, since it allows lowtaxcapacity states to be put on some socially determined equal footing with hightaxcapacity states. Moreover, this equalization standard is, in principle, quite flexible--for instance, it could be defined in terms of the top 10 states rather than the allstate average. Equalization transfers provide a substantial compensation to the poorer states already, compared to where they would be without it. The next step for Mexico would be to make the transfers more uniform to places with the same degree of poverty (with a formula like that for FAIS) and to decide if the degree of compensation is enough. The international experience has various examples of transfers adjusting for the estimated cost of providing services, some giving more to densely populated areas (Germany) and some giving more to sparsely populated areas (Australia). In the case of Mexico, the main problem seems to attract staff to the remote areas, so that is perhaps where the higher transfers per capita should go. The most important thing is to have a transparent formula for the adjustments, if any, and to review it periodically to check if it is having the desired effects. Conditional Citizenship Transfers With much of the social cohesion of the federation addressed in the context of the taxautonomycumequalization context, the stage is now set for a new approach to conditional transfers, namely that they should be equal per capita across all states. Their conditionality is predicated on the assumption that they are designed to deliver public goods (health, education) equally to all Mexicans. It would be possible to define this aggregate transfer in terms of specific public goods, along the lines of the Ramo 33 funds. Although strict conditionality might apply at the outset, the spirit of this proposal is eventually to define conditions as a set of agreedupon principles relating to minimum acceptable standards for these expenditures. This would allow states some flexibility in Decentralized Service Delivery for the Poor 43 determining the most efficient way to achieve these standards and principles. As will be discussed in the accountability framework in Chapter 4, there are interests, like teacher unions, that may oppose or delay a transition of this sort, since they may be more concerned with resources allocated according to the number of teachers. Transitional Guarantees This final principle concerns a transitional solidarity program for the states and perhaps for the federal government as well. There would be winners and losers in a direct transition to any model, although under reasonable assumptions the steady state could be a winwin solution for all. Appropriate transitional mechanisms are therefore needed to make the change generally desirable and politically acceptable. 44 Decentralized Service Delivery for the Poor CHAPTER 3 PERFORMANCE OF THE SECTORS This chapter examines the performance of the sectors and programs described in Chapter 1 in relation to poverty in Mexico. In all sectors, coverage and targeting of services has improved, but there is little evidence that quality has improved, with a few exceptions. There is evidence that performance of the services varies across the country, and this variation is not linked solely to the states' income per capita. In education, the gap in primary education coverage between the poor and the rich states has narrowed since decentralization. Equity in coverage has improved, perhaps in part due to the cumulative effect of the compensatory programs that encourage the poor to get education and health care. Efforts to increase lowersecondary schooling for the poor still confront significant challenges. Quality of education is low by international standards even after controlling for Mexico's level of income, particularly for the poor, which exacerbates problems of access. The rapid expansion of the telesecundaria system, which offers schooling for the rural poor, is worrisome given the students' low level of achievement. Despite important institutional limitations in the decentralization process, local innovations have helped some states to improve performance. In addition, some federal educational programs are stimulating discussion about how to improve school management and learning achievement. Still, the challenge is to ensure that better quality becomes systemic, to avoid exacerbating regional differences. With respect to health, life expectancy, infant mortality and other outcome indicators have improved in the last decade, due in part to the success of centralized programs, such as the vaccination program. Maternal mortality, however, has not improved significantly in the last 15 years and is closely related to the quality of the health service. There are a lot of data about the overall population, but having several different systems of provision (IMSS, SSA, ISSSTE, IMSSOportunidades, and state health systems) makes it difficult to establish a relationship between the institutional arrangements and their outcomes.1 A longitudinal health analysis based on a sample of six states found, first, that there are important differences in health outcomes and indicators across states, and the differentiation is not necessarily between the poor and the rich states. Second, the evolution of health performance in high to mediumincome states such as Baja California Sur, Jalisco, and Tabasco has not been even nor 1 SSA is working on compiling health indicators by institution. SSA expects to publish its progress on this front at the end of the year. Decentralized Service Delivery for the Poor 45 significantly better compared to the rest of the country. Third, these states seem to be far advanced in the decentralization process, are investing considerable resources in their health systems, and have reallocated resources relatively effectively, into preventive versus curative expenditures. A new federal program--Seguro Popular--distributes funds to states based on the location of patients that choose to affiliate and is helping to increase coverage among the population without social security coverage and to reduce catastrophic health expenditures (for details see Arredondo 2005 and Chapter 4). In social infrastructure, coverage has improved, although it continues to be low for the poor. FAIS is an important financing source for local social infrastructure investment in the poorest municipalities, and it is, in general, well targeted according to the level of marginality. However, the significance of FAIS in the overall availability of municipal financial resources varies by sector and size of municipality. The sectoral allocation of the FAIS investment (water versus roads versus housing upgrading) is not generally correlated with lack of coverage in that sector, although this finding varies by state and sector. A potential explanation is that municipalities have several sources of funds for investment in social infrastructure, not just FAIS, and another explanation is that FAIS institutional arrangements vary from state to state. Productionoriented rural development programs have the potential to improve income opportunities for the poor. However, decentralization of the Rural Development Program, one of the programs of Alianza, has not had a noticeable impact on outcomes. First, the program is not well targeted to the poor. Second, qualitative evaluations show an effect on productivity, but not in the aggregate. Third, the Rural Development Program has been very difficult to operate at the local level because the center imposes many restrictions and rigidities on its implementation. Table 3.1 summarizes the main findings of this chapter in terms of coverage, targeting, quality, and efficiency (mainly allocation of resources related to quality). Annex 3.1 has a table that links available data sources with dimensions of performance. However, little information exists on the performance of the programs at the state and municipal level; collecting this information systematically should be a priority. 46 Decentralized Service Delivery for the Poor Table 3.1. Summary--Performance of the Sectors Sector/ Dimension of performance analysis program Coverage / targeting Quality Allocation of resources Basic Coverage is almost Quality of education is low by The bulk of spending education universal in primary. international standards and is distributed Coverage in lower controlling for income level. inertially: secondary is still low for Increase in spending the poor. per student reflects increases in wages. Test results are not correlated with spending per student. Health for Much of the population, Important differences in The bulk of spending the especially the poor, outcomes and indicators, not is mainly distributed population remains without access to necessarily between the poor inertially. However, without social security coverage and the rich states; also uneven spending per capita for social and faces high outof through time. the population without security pocket expenditures. social security coverage coverage has doubled in the last decade. FAIS--social FAIS is well targeted No data to assess quality of the Basic social infrastructur according to the level of projects funded by FAIS. infrastructure e marginality. Although several World Bank spending per capita reports, such as IPER (2005), has increased in real document the low quality of terms. Controlling for infrastructure services. needs, resources are being allocated to the coverage of basic social services. Desarrollo The program is not well Some perceived effects on Resources are allocated Rural of targeted to the poor. outcomes, such productivity to states according to a Alianza para and wages. However, there are highly complicated el Campo no clear standards for output formula that includes and thus no concrete among other variables, indicators of quality. agricultural GDP, harvested land, irrigated land, number of production units, and state contribution. I. EDUCATION Basic education (primary and lower secondary) is fundamental for poverty alleviation because education is a major determinant of earnings. Lack of education and lowquality education contribute greatly to earnings inequality. In Mexico, close to 70 percent of public school students (92 percent of the total) come from poor families. In Decentralized Service Delivery for the Poor 47 2005, Mexicans over 15 years of age had, on average, more than eight years of education, but the average hides major variations. For example, in 2004 there was a steep income gradient: persons 25­65 years old living in the top 10 percent of households had 12.6 years of education, while those living in the bottom 10 percent had 3.7 years. Nonetheless, over time, average schooling has increased, particularly in the bottom and middle distribution of income (see Table 3.2). Table 3.2. Average Schooling of 2565 YearOlds, by Income Decile, 19842004 Decile 1984 1992 2000 2002 2004 I 2.1 2.2 2.8 3.2 3.7 II 2.3 3.0 3.7 4.2 5.1 III 3.0 3.6 5.0 5.0 5.8 IV 3.0 4.2 5.6 5.6 6.1 V 3.6 4.8 6.2 6.5 6.7 VI 4.1 5.3 6.9 6.8 7.3 VII 5.0 6.0 7.4 7.4 7.8 VIII 6.2 6.8 8.3 8.2 8.6 IX 7.2 8.2 9.5 9.8 10.1 X 8.6 10.8 12.1 12.0 12.6 Average 6.5 8.6 9.3 8.8 8.9 Source: World Bank staff calculations from ENIGH. Also there is important variation in years of schooling across regions. For example, in certain states, more than 60 percent of the population 15 years and above do not have a lowersecondary education (see Figure 3.1). Figure 3.1. Population Over 15 Years Old without a Lower Secondary Education, 2005 Source: World Bank staff estimates based on the Conteo de Población y Vivienda 2005. 48 Decentralized Service Delivery for the Poor Coverage and Targeting Mexico's basic education system has expanded strongly since 1980, as evident in gross enrollment, which expanded from 18 million to 25 million students between 1980 and 2005. Mexico has achieved close to universal primary education and expanded lowersecondary education. In the last decade, progress in basic education has been particularly important in the lessdeveloped regions of the country, and the gaps between richer and poor states are shrinking (see Figure 3.2).2 Figure 3.2. Coverage Ratio of Basic Education, 3 to 15 year olds, 19912004 90 85 eg 80 tanecr 75 Pe 70 65 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Very Low Low Medium High Very High Note: The trends are simple averages of states grouped by their marginality status, as defined by CONAPO 2000. Source: World Bank staff estimates based on CONAPO 2000 and SEP. However, coverage of net enrollment in lower secondary is still low among the poor. In 2004, the lowersecondary net enrollment rate for the extreme poor was around 56 percent compared with 83 percent for the nonpoor (see Table 3.3). 2 For Figure 3.1 and Figure 3.2, the 2000 marginality index among states is used: very low: Distrito Federal, Nuevo León, Baja California, and Coahuila; low: Aguascalientes, Baja California Sur, Chihuahua, Jalisco, Sonora, Tamaulipas, Colima, and Estado de México; medium: Quintana Roo, Morelos, Tlaxcala, Durango, Querétaro, and Sinaloa; high: Nayarit, Guanajuato, Zacatecas, Yucatán, Michoacán, Tabasco, Campeche, Puebla, and San Luis Potosí; very high: Hidalgo, Veracruz, Oaxaca, Guerrero, and Chiapas. Decentralized Service Delivery for the Poor 49 Table 3.3. Enrollment Rates in Basic Education, 2004 Primary (ages 6­11) Lower secondary (ages 12­14) Poverty status Urban Rural National Urban Rural National Extreme poor 94.5 93.8 94.1 59.0 54.0 55.9 Moderate poor 96.1 96.8 96.4 74.1 66.8 71.2 Nonpoor 95.1 95.6 95.2 81.9 84.5 82.8 Total 95.4 95.3 95.4 75.5 68.2 72.3 Note: Poverty calculated using SEDESOL's food and assetbased poverty lines and current income per capita, using trimester income as reported by INEGI. Urban areas are localities with 15,000 inhabitants or more, using INEGI's classification. Source: World Bank staff calculations from ENIGH 2004. Quality Rigorous analysis of the quality of education being provided is relatively recent in Mexico, in part because very little information has been disseminated. Since 1998, researchers have had access to standardized tests conducted by the General Direction for Evaluation in SEP. Since then, a significant amount of data has been collected, and technical capacity for measuring learning achievement has been developed. The National Institute for Education Evaluation (INEE) is currently undertaking a study comparing the results of the National Standards Test (Estándares Nacionales) for 2000 with the results of new tests taken in 2005, which were designed to be comparable. Results are forthcoming. Mexico's participation in the Organization for Economic Cooperation and Development's Program for International Student Assessment (PISA) since 2000 provides a good basis for measuring, benchmarking, and assessing the determinants of the quality of education. PISA provides a measure of reading, mathematical, and scientific literacy for a nationally representative sample, comparable across teens 15 years old in different countries. Students from Mexico and other Latin American countries are among the worst performers. Overall, Mexican students, on average, perform as well as students from other middle and lowincome Latin American countries that participate in PISA, with the exception of Uruguay. Furthermore, there has been little improvement since the first PISA test was administered in 2000. The results of PISA 2003 confirm these poor results.3 Mexico's performance in all three subjects declined, although this may be because enrollments increased during the same period by about 5 percentage points.4 While Mexico has relatively poor performance, these results have very low dispersion, compared with 3 PISA 2005 results are forthcoming in August 2006. 4 According to the ENIGH, enrollment in lower secondary for children 12­14 years old increased from 70 percent in 2002 to 72 percent in 2004 at the national level. In rural areas, enrollment increased from 59 to 68 percent. In upper secondary for children 15­17 years old, enrollment increased from 45 percent in 2002 to 48 percent in 2004. 50 Decentralized Service Delivery for the Poor those of Brazil and Uruguay.5 Figure 3.3 provides the results for math (reading and science show similar patterns). Figure 3.3. Math Scores in PISA and Dispersion across Countries, 2003 600 Hong Kong 550 Finland Korea Canada Netherlands Australia Japan Macao Swit Denmark Belgium Czech New Zealand Iceland France 500 Ireland Austria Sweden Germany Poland NorwayHungary Latvia Spain US ero Portugal Russia Italy 450 Sc Serbia Greece Uruguay Turkey Thailand 400 MEXICO Indonesia Tunisia 350 Brazil 300 240 260 280 300 320 340 360 380 Dispersion (difference between top and bottom 5%) Source: PISA. Relative to other OECD members, Mexico's performance in relation to GDP per capita suggests that it is performing far below its potential (see Figure 3.4). 5 The level of equity in Mexico has been a subject of debate. Although Mexico's enrollment is much lower than that of comparators such as Brazil, with many of the more disadvantaged children not attending school (OECD 2005b), simulations using the change in enrollment between PISA 2000 and 2003 indicate that, while dispersion has increased, Mexico remains in the quadrant occupied by countries with relatively low test score inequality and is not significantly different from Portugal and Spain. Additional analysis based on relative performance by socioeconomic groups shows significantly less variation in Mexico than in Argentina and Brazil (World Bank 2000). Decentralized Service Delivery for the Poor 51 Figure 3.4. Performance in Mathematics in PISA and GDP per Capita, 2003 600 Rep. of Korea Japan New 500 Norwa Latvia Luxembourg France USA 400 Mexico Scores Indonesia tha M 300 Average 200 100 0 0 10000 20000 30000 40000 50000 60000 GDP per capita Source: PISA; EdStats. The high level of equity in results is also seen among Mexican states (see Figure 3.5). While a small group of states (three or four) are relatively high performers and a small group of states (four or five) are relatively poor performers, the vast majority of states in Mexico have broadly similar results. Withinstate dispersion appears to be a major issue in at least three states, but these are not clustered in any one performance group. 52 Decentralized Service Delivery for the Poor Figure 3.5. Means of the Math Test Scores in PISA, by State, 2003 Means of the Math Test Scores in PISA, by State GNP per-capita, 2003 500 450 400 scores Math350 300 250 ca al tri e oc s í a he n al xa rero osl pas urS Oa Chiapas Guer xcaalT ya Llav caet coxié naloa nátac scoil ional ooR Leó Na agoz la abasT dalgoiH tosoP de caaZ ueblaP M Si Yu More olimaC Ja Sonora Arteaga mpec o Feder Luis oic uanajuatoG rangouD acN entesilac o amauliT liforniaaC Zar a Chihuahua Ca Nuev San ritotsiD Igna omedirP étaroreuQ California de Baj Aguas ntanaiuQ a Baj de oahuilaC uzr raceV States Source: PISA. Figure 3.6 shows test scores by type of school system, not controlling for relevant factors. It shows that telesecundarias, one of the fastestgrowing secondary school systems (around 20 percent of total secondary enrollment), are the worstperforming type of school system. Math scores for students in telesecundarias are, on average, 100 points lower than those of students in general high schools, almost 150 points lower in reading, and about 100 points lower in science. Decentralized Service Delivery for the Poor 53 Figure 3.6. Test Results by Type of School, PISA 2003 Reading Math 500 Science 450 400 350 300 250 200 150 100 50 0 y y y ndar ndar ndar lacin l sr y y y ai onai rof egar ch ngi ardn ardn yardn serkr ndar ecoSr echnicalT ecoSr ralen ecoSr ecoS ecoS Pr na ralen Ge ageervA ess Te of ainrT Worke Avel ecoS Ge echnicalT econdarSr ageervA WoroF secueleT Uppe Uppe Uppe Math Reading Science tioaN erwoL erwoL weoL erwoL Source: World Bank, 2005b. Controlling for relevant school and individual characteristics, all other types of schools have a significant learning advantage over telesecundarias (see Figure 3.7). In math, the advantage is only statistically significant for uppersecondary technical and general high schools, technical professional schools, and training for workers programs (bachillerato técnico, bachillerato general, profesional técnico, and capacitación para el trabajo). That is, there is no significant difference between types of lowersecondary school (secundaria general and técnica and secundaria para trabajadores) after controlling for other factors. But for science and reading, telesecundarias have a significant disadvantage relative to all other types (except training for workers). Analysis of the PISA results undertaken by INEE also shows that telesecundarias perform worse than other types of schools: 89 percent of telesecundaria students are not competent in mathematics, compared with 71 percent of students in general secondary schools (INEE 2005). 54 Decentralized Service Delivery for the Poor Figure 3.7. Performance Advantage (percentage) of Different Types of Schools Compared with Telesecundaria 90 80 70 60 50 40 30 20 10 0 Upper Upper Technical Lower Lower Training for Secondary Secondary Professional Secondary Secondary workers General Technical General Technical Reading Math Science Note: Controlling for other significant factors. Source: World Bank, 2005b. Some state, both rich and poor, are introducing local innovations to improve their basic education system, such as Aguascalientes, Colima6, Nuevo León, Distrito Federal, and Quintana Roo. The annex to this chapter shows an inventory of states' institutional arrangement after decentralization. The first four states ranked at the top of the PISA scale, while the last state is very close to the national average. These states are working to improve learning achievement on several fronts, such as improving pedagogy practices, evaluation, and hiring of teachers. In Aguascalientes, initiatives started after decentralization were intended to push decentralization of administration and pedagogy to the regional level. One of the benefits was to increase control over human resources and to gain fiscal resources (about 9 percent annually) to fund local initiatives. These include a program of teacher training that helped 32 percent of state teachers to pass the first national testing in 1996, compared with only 8 percent nationwide, and the incorporation of learning technologies in the schools, student testing, and social participation. More recent initiatives include pushing to incorporate local schools into the national program Enciclomedia (basically interactive textbooks), a locally financed effort to teach English in primary school (in contrast to the national curriculum, which incorporates English only in lowersecondary school), and the application of exams to newly graduated teachers. 6 Colima is the best performer but only ranks 14th in GDP per capita. Decentralized Service Delivery for the Poor 55 In Quintana Roo, the local innovations started in 1999. Improvements since that time have been notable, particularly in comparison with other southern states. The centerpiece of its efforts has been at the school level, where directors have been "empowered." Along with accountability for school performance, directors have received extensive training from national and international experts on management, leadership, team work, and strategic planning. Fiscal resources have been channeled directly to schools, and 80 percent of schools have been incorporated in PEC. A trust fund has been created for handling contributions from parents and the private sector. An extensive system of student testing has been established for all grades and all schools, with the results fed back into local plans for quality improvement and publicized. A statelevel council for social participation has been created with heavy involvement of education specialists and the private sector, which serves as an important counterbalance to resistance from organized labor. Municipal governments have also been enlisted in the effort with agreements to support construction and maintenance of school facilities. Administrative reforms have facilitated this process, notably with programs of professional development for administrative staff and results based budgeting. Preliminary evidence on federal programs such as the AGEs and PEC might continue stimulating discussion about how to implement policies that encourage greater participation of schools in their management. Both the AGEs and PEC are trying to increase school autonomy, local power, and parental participation. According to beneficiaries and a previous analysis, the AGEs have increased parents' participation in schoolrelated activities, improved parentteacher relations--meaning that parents are more likely to be aware of the teachers' role in the schooling of their children--and encouraged more parental involvement in student life; that is, parents are more aware about their children's performance in school and are more likely to insist that their children fulfill academic obligations such as homework. AGE substantially improves learning outcomes, even after controlling for other interventions such as Oportunidades, suggesting that decisions to decentralize education to local levels can have positive impacts on outcomes (World Bank 2006b). PEC reduced dropout rates by 0.41 percentage point, failure rates by 0.88 percentage point, and repetition rates by 1.04 percentage points (World Bank 2006b). Loera (2005) also analyzed the changes in management, teacher practices, social participation, external efficiency, and educational achievement. Variables such as the school director's ability to lead the transformation of the school community and teachers planning their lessons to include alternatives that consider the diversity of their students, exhibited a positive and statistically significant increase over the first three years of the program. The factors that positively influenced outcomes include: improvements to school infrastructure and security, teachers' high expectations that their students will do well, cultural capital in the home, didactic support for teachers, and a participatory learning environment. While there is no similar information for non 56 Decentralized Service Delivery for the Poor PEC schools, this nevertheless shows that PEC schools are moving in the right direction in improving the school climate and parental participation. An analysis of the determinants of school achievement shows how state actions can affect learning outcomes (Table 3.4; Álvarez et al. 2006). Only two variables are significant: (a) using the state evaluation system to provide feedback to schools and to design interventions and (b) conflicts between the unions and the state. These variables are significant: their coefficients are large, and their levels of significance are as high as or higher than other variables, such as educational resources in the home or mother's education. The full evaluation system has the largest impact by far. This is evidence that states can take significant actions to improve their school systems by developing and using a state evaluation system. Thus institutions matter, but the most significant issues are relatively low cost and under the direct control of state authorities. This makes it a particularly useful investment given its large contribution to learning outcomes as well as the fact that it can be a very cheap investment. Table 3.4. State Actions as Determinants of Student Learning Explanatory factor Significant Coefficient State accountability system Evaluation system (testing) No Dissemination of results No Feedback from schools No Complete process and design intervention strategies Yes 13.1 Conflict between state and union Medium Yes 4.6 Low Yes 9.5 Controls for: Parental participation incl. Withinstate decentralization incl. Union influence to appoint teachers incl. Student characteristics incl. Family background incl. Home incentives and inputs incl. Observations 12,311 Source: Álvarez et al. 2006. II. HEALTH Mexico's health sector over the past two decades has achieved significant improvements in the health status of the population, broader access to basic services, and support for important public health measures. Life expectancy increased more than eight years in the last two decades. Infant mortality decreased 62 percent from 1980 to 2004, and immunization among infants has become nearly universal, at 95.4 percent (see Figure 3.8; SSA 2005). Over the last decade, the mortality rate for children under age Decentralized Service Delivery for the Poor 57 five fell around 20 percent, and mortality from diarrhea and pneumonia fell more than 70 percent (World Bank, 2004a). Vaccinepreventable diseases declined drastically, with no cases of polio or diphtheria since 1993. Maternal mortality, however, which is related to the quality of the service, has remained at high levels according to international standards (Table 3.5). The Millennium Development Goal on maternal health proposed to reduce maternal mortality by twothirds from 1990 to 2015. With only a reduction of 30 percent in the last 15 years, however, Mexico will probably miss this goal. Figure 3.8. Life Expectancy and Mortality, 1990­2004 76 50 75 40 cyna 74 30 ctep 73 1,000 r Ex ef 72 20 pe Li71 10 70 69 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 Life Expentancy Infant Mortality Probability to Die Before Age 5 Source: DGIS, SSA 2006. Table 3.5. Maternal Mortality Rates in OECD Countries, Early 2000's (per 100,000 live births) Australia 4.8 Korea 15 Austria 2.6 Luxembourg 18.6 Belgium 8.6 Mexico 65.2 Canada 7.8 Netherlands 9.9 Czech Republic 3.2 New Zealand 8.8 Denmark 13.6 Norway 5.3 Finland 5.4 Poland 5.4 France 6.5 Portugal 2.5 Germany 2.9 Slovak Republic 7.9 Greece 6 Spain 4.2 Hungary 8.3 Sweden 4.4 Iceland 0 Switzerland 6.4 Ireland 11.6 Turkey 49.2 Italy 2.7 United Kingdom 6 Japan 7.3 United States 9.9 Average without Mexico 8.4 Standard deviation without Mexico 8.8 Source: OECD 2005c. 58 Decentralized Service Delivery for the Poor Figure 3.9 and Figure 3.10 illustrate the inequity in the system: the health status of the poor is much lower than the health status of the nonpoor, and the poor have higher and more frequent catastrophic expenditures on health, which are defined as 30 percent of disposable consumption, after spending on food, education and housing. Figure 3.9. Infant Mortality Rates by Municipality, 2000 80 70 000 1, r 60 pe et 50 40 Ra ytilatr 30 20 Mo10 0 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 Net Income per Capita Source: CONAPO 2000. Figure 3.10. Share of Households with Catastrophic Health Expenditures, 1992 2004 5 4 3 2 1 0 1992 1994 1996 1998 2000 2002 2004 Source: Knaul, Arreola, and Méndez, 2005. Decentralized Service Delivery for the Poor 59 Coverage and Targeting Despite substantial progress in basic health indicators, the health system has weak outcomes for the poor with respect to health, financial protection, and responsiveness. Much of the population, especially the poor, remains without access to social security and faces high out of pocket expenditures. In addition, institutional fragmentation reduces the effectiveness of the system (see Figure 3.11). While households with workers engaged in the formal labor market benefit from health insurance through either IMSS or the Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado (ISSSTE), the informal sector--63 percent of the working population according to ENIGH 2004--remains largely without social security coverage. Coverage of the formal insurance system is even more limited in certain regions of the country (see Figure 3.12). Figure 3.11. Population by Institution of Affiliation, 2004 50 45 40 35 sn 30 25 illio M 20 15 10 5 0 s SS E IM orug ra d s pul SSSTI Se ,XE an , de eta R Po A A mets M NE M dainutr St d serits an ni PE D SE Sy eta SS IM Mi SE St Opo SSA Insured Uninsured Note: Some people are affiliated with more than one institution. Source: SSA 2005. 60 Decentralized Service Delivery for the Poor Figure 3.12. Population without Social Security Coverage, 2005 Source: World Bank staff estimates based on the Conteo de Población y Vivienda 2005. Access to basic health care remains sporadic in certain parts of the country. Indicators of health outcome in Chiapas, Oaxaca, and Guerrero continue to lag behind the national average. In fact, these states are at the bottom of the list of states ranked by life expectancy at birth (see Table 3.6) and at the top in terms of maternal mortality. Nevertheless, recent evidence indicates that the gap in life expectancy between the national average and each of these three states is narrowing. Table 3.6. Life Expectancy at Birth, 1990­2005 1990 1995 2000 2005 National 71.2 72.7 74.0 75.4 Chiapas 68.7 70.6 72.3 73.9 Guerrero 69.3 71.0 72.5 74.1 Oaxaca 68.9 70.9 72.5 74.1 Lowesthighest life expectancy (excluding Chiapas, 69.9­ 71.6­ 73.1­ 74.6­ Guerrero, and Oaxaca) 73.6 74.2 75.4 76.6 Source: DGIS, SSA 2006. Perceived Technical Quality Table 3.7 and Table 3.8 show some indicators of quality for different health systems. In some indicators, such as waiting times and perceived quality by the poor, IMSSOportunidades has ranked a bit better than the other systems. IMSSOportunidades is different from the other systems in terms of structure and financing, as discussed in Chapter 4. Decentralized Service Delivery for the Poor 61 Table 3.7. Indicators of Quality, by Health System, 2003 IMSS Indicator SSA IMSS ISSSTE Oportunidades Others Waiting time for a consultation 28.17 15.23 30.54 18.01 33.75 Waiting time for an emergency 17.42 15.47 25.96 17.55 22.05 Percent of users satisfied with waiting time for a 92.1 87.4 90.1 97.4 89.7 consultation Percent of users satisfied with waiting time for an 87.1 87.2 85.5 94.0 87.0 emergency Source: SSA 2005. Table 3.8. Perceived Quality of Care by Health Care Provider, 2000 Percent of individuals responding Average to Program Very good Good Average Poor Very poor very poor IMSS 4.2 76.7 18.5 0.6 ... 19.1 Oportunidades SSA/State Health 4.7 82.4 11.7 1.1 0.1 12.9 Systems Private 10.7 78.4 9.3 1.6 ... 10.9 Source: World Bank 2004d. Some health indicators in Figure 3.13 could eventually be improved as a result of decentralization. Even some poor states such as Oaxaca show better performance than rich states such as Tabasco. Caution should be taken in interpreting these trends. Due to data constraints, we were not able to analyze the link between a change in policy at the state level and a change in service provision or utilization that, in turn, could have an effect on health outcomes. 62 Decentralized Service Delivery for the Poor Figure 3.13. Health Indicators, 19902003 2,500 2,000 ytilatr 1,500 Mo tn 1,000 fa In 500 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 35 30 ytila 25 rt 20 Mo l naret 15 10 Ma 5 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 700 600 yit idbro 500 400 M seteba 300 200 Di 100 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 National Oaxaca Hidalgo Colima Tabasco Baja California Sur Jalisco Decentralized Service Delivery for the Poor 63 yit 400 idbro 350 M e 300 ases 250 Di yrotarips 200 150 100 Re e 50 ut 0 Ac 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 National Oaxaca Hidalgo Colima Tabasco Baja California Sur Jalisco Source: Arredondo, 2005 estimates based on SSA data. There is also variation in other health indicators, such as the number of doctors per population and the number of beds per population, and in efficiency indicators, such as hospital use and rotation (see Figure 3.14 and Volume II for more details). For coverage, measured by the number of doctors per population, two rich states show opposite results: Jalisco has the highest coverage, and Baja California Sur has the lowest. For efficiency, measured by hospital use and rotation (number of beds per new hospital patient), some rich states such as Jalisco7 and Baja California Sur, are below the national average, while two of the poorest states, Hidalgo and Oaxaca, are above the national average. This reflects the much larger portion of the state population without social security coverage and hence constrained to using SSA or State Health Systems' facilities. 7 In 1997, Jalisco integrated the Civil Hospital to the State Health System, thus increasing rapidly the state health resources. 64 Decentralized Service Delivery for the Poor Figure 3.14. Health Indicators for Select States, 19902001 s 4,500 ntatib 4,000 3,500 nhai 3,000 2,500 000 1, r 2,000 pe srotc 1,500 1,000 500 Do 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 3,000 s ntatib 2,500 hani 2,000 000 1,500 1, r 1,000 pe ds 500 Be 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 90% 90% 80% 80% ycn pauc 70% noitat 70% Oc laitps 60% Ro latips 60% Ho Ho 50% 50% 40% 40% 95 97 99 01 03 19 19 19 20 20 9991 0002 0012 0022 0032 National Oaxaca Hidalgo Colima Tabasco Baja California Sur Jalisco Source: Arredondo, 2005 estimates based on SSA data. Decentralized Service Delivery for the Poor 65 Catastrophic health expenditures follow a similar pattern as some of the health indicators presented above. Catastrophic health expenditures are high for Oaxaca but also for nonpoor states such as Jalisco and Tabasco. According to Arredondo (2005), and a summary table in Annex 3.2, Tabasco, Jalisco, and Baja California Sur have been able to advance further in the decentralization process than other states in terms of institutional organization, financial management, and allocation of resources. However, their performance in health is poor in relation to their level of income per capita, to the amount of public resources invested in their health systems, and to the coordination of their health systems. Figure 3.15. Share of Households with Catastrophic Health Expenditures in Selected States, 2003 6 5 4 3 2 1 0 Baja California Sur Colima Hidalgo Jalisco Oaxaca Tabasco Source: Arredondo, 2005. Recent health initiatives, such as Seguro Popular, are providing health coverage to the population without social security coverage in an effort to minimize their vulnerability to catastrophic or impoverishing health expenditures and more generally to ensure equality of basic health opportunities. Seguro Popular now covers about 3.5 million households.8 Seguro Popular targets 90 percent of the households in the poorest two quintiles. In contrast to the results from administrative data, survey results show that, at this early phase, only 60 percent of the households insured by Seguro Popular are in the poorest 40 percent of the population. More important, other programs such as Oportunidades and IMSSOportunidades appear to be better targeted and therefore offer 8 Affiliation to Seguro Popular is carried through two stages. The allocation of federal Seguro Popular funds to the states and the corresponding state contributions are jointly defined by the federal and state governments through an Acuerdo de Coordinación. Secondly, state health ministries set up Regímenes Estatales de Protección Social en Salud (REPSS) which are responsible for the promotion of the program and affiliation of beneficiaries. 66 Decentralized Service Delivery for the Poor opportunities to improve the targeting of Seguro Popular.9 Figure 3.16 and Figure 3.17 compare the distribution of Seguro Popular with the distribution of the population without social security coverage,10 of Oportunidades, and of the (use of the) principal health services for the population without social security coverage (SSA, IMSS Oportunidades, Institutos Nacionales de Salud) and for the insured (IMSS, ISSSTE, PEMEX). All data was obtained from ENIGH on a common methodological basis. On the one hand, Seguro Popular affiliation in 2004 was more progressive (propoor) than both the overall use of (untargeted) SSA services and the distribution of the population without social security coverage--the longterm target population of the program. On the other hand, the observed degree of progressiveness of Seguro Popular falls short of that achieved by both Oportunidades and IMSSOportunidades. Figure 3.16. Distribution of Seguro Popular and Oportunidades Beneficiaries of SSA Health Services and of the Uninsured Population 45% Seguro Popular 40% Uninsured Oportunidades SP + OP 35% Use of health services for the uninsured (SSA) 30% 25% 20% 15% 10% 5% 0% 1 2 3 4 5 6 7 8 9 10 Note: Household deciles ordered by income per capita net of transfers. Source: World Bank staff calculations using the Módulo de Programas Sociales of ENIGH 2004. 9 A recent initiative currently being discussed is a proposal to affiliate IMSSOportunidades' patients with Seguro Popular (see Arredondo 2005). 10 In this analysis, a household is classified as insured if at least one of its members is covered by any of the public social security schemes or by private health insurance and as uninsured otherwise. Decentralized Service Delivery for the Poor 67 Figure 3.17. Concentration Coefficients for Seguro Popular and Other Health Programs, and the Uninsured Population -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 Oportunidades IMSS-Oportunidades Seguro Popular SSA total Uninsured IMSS Pemex SSA - Institutos Nacionales de Salud ISSSTE Progressive (pro-poor) Regressive (pro-rich) Note: In the case of health service providers, reported distributions refer to the use of services. Population is ordered by income per capita net of transfers. Source: For Oportunidades, Seguro Popular, SSA, and the uninsured, authors' calculations using ENIGH 2004 (including Módulo de Programas Sociales). For the rest, Scott (2005), using ENIGH 2002 (including Módulo de Programas Sociales), except IMSSOportunidades and Institutos Nacionales, which were obtained from ENSA 2000. Household health expenditures are lower on average for Seguro Popular beneficiaries than for the rest of the population without social security coverage, in absolute as well as relative terms (except in the second quintile), although the differences are probably too small to be statistically significant in the poorer quintiles.11 Again, there is some evidence that Oportunidades is helping to reduce household health expenditures. Finally, there is evidence that the incidence of catastrophic health expenditures is lower across deciles for Seguro Popular beneficiaries than for the rest of the population without social security coverage. For a complete analysis, see Volume II. 11 Given that coverage of Seguro Popular was still relatively low in 2004, that the sample of beneficiaries in the ENIGH survey was limited, and that use of and spending on health care (especially catastrophic spending) are infrequent events, these results are presented in coarser household partitions by income (quintiles 1 and 2, and 3­5 grouped) and should be interpreted with care, especially when observed differences are small. Catastrophic health expenditures are defined as expenditures representing more than 30 percent of disposable household income, defined as current income net of the food poverty line, which represents the cost of a minimum food basket. This definition is not comparable to the more complex concept used by the SSA. 68 Decentralized Service Delivery for the Poor III. BASIC INFRASTRUCTURE Access to basic services such as water, sanitation, and electricity are important to reduce poverty and improve the wellbeing of families. Despite advances, an important share of the poor still lack access to basic services. In 2004 almost 10 percent of all households in Mexico did not have access to an improved water source, around 7 percent did not have any kind of drainage service, and more than 10 percent lived in a dwelling with a dirt floor (see Table 3.9). However, as shown in Figure 3.18, the share of households with access to running water, sewage, and electricity varies significantly across regions in the country. Households in the southern part of Mexico have low coverage of at least one of the three services (water, sewage, and electricity). Table 3.9. Access to Services, 19922004 National Urban Rural 1992 2004 1992 2004 1992 2004 Yes 92.4 98.7 98.5 99.5 76.2 95.8 Electricity No 7.6 1.3 1.5 0.5 23.8 4.2 Improved water source 80.7 90.1 92.2 94.4 50.3 75.6 Water Nonimproved water 19.3 9.9 7.8 5.6 49.7 24.4 source Slab or finished floor 26.4 38.8 34.1 46.1 5.8 14.3 Floor Cement 55.1 50.9 57.6 48.3 48.4 59.6 Dirt floor 18.5 10.3 8.2 5.7 45.8 26.1 Improved sewage 65.4 91.2 81.6 95.3 22.5 75.7 Sewerage Nonimproved sewage 6.0 1.5 4.0 1.1 11.3 2.9 No sewerage 28.6 7.3 14.4 3.6 66.2 21.5 Source: World Bank staff estimates based on ENIGH. Decentralized Service Delivery for the Poor 69 Figure 3.18. Households with Access to Piped Water, Sewage, and Electricity, 2005 Source: World Bank staff estimates based on the Conteo de Población y Vivienda 2005. Coverage of basic infrastructure is still low for the poor. In 2004 in rural areas, 48 percent of the extreme poor (those living below the food poverty line) lived in houses with a dirt floor, 37 percent did not have access to an improved water source, a third did not have access to sewerage, and 10 percent lacked access to electricity. In 2004 in urban areas, 19 percent of the extreme poor had a dirt floor, 13 percent did not have access to improved water, 12 percent did not have access to sewerage, and 3 percent did not have access to electricity.12 According to the census, in 1990, 21 percent of the population had no access to improved water, while 15 years later only 11 percent had no access. In the case of electricity, coverage increased from 87 to 97 percent in the same period of time. 13 Despite this progress, disparities exist across municipalities. As shown in Table 3.10, 14 percent of the population in the lowest quartile does not have access to electricity and 29 percent does not have access to improved water. Meanwhile, only 0.6 percent of population in the top quartile lacks electricity, and only 2.1 percent has no access to improved water. More than threefourths of the population in the lowest quartile lives in a rural area in contrast to 2.8 percent in the top quartile. 12 According to World Bank staff estimates based on ENIGH. 13 Conteo de Población y Vivienda 2005. 70 Decentralized Service Delivery for the Poor Table 3.10. Access to Services, by Group of Municipalities, 2000 Indicator I (poorest) II III IV (richest) Total Percent of population 14.0 3.2 1.3 0.6 4.8 without electricity Percent of population 28.9 10.0 4.0 2.1 11.2 without improved water Percent of population 26.7 9.3 2.5 1.1 9.9 without sanitation Percent of population living in localities with less than 77.0 36.3 8.0 2.8 31.0 5,000 inhabitants Total population (millions) 24.3 24.3 24.0 24.7 97.4 Number of municipalities 1,553 669 132 56 2,410 Note: Percentages are weighted by population in the municipality. The groups were defined according to the marginality of each municipality in 2000. Source: World Bank staff elaboration based on CONAPO information at municipal level 2000. FISM is the most important funding source for local social infrastructure investment for poor municipalities. In 2003, an average of 58 percent of the total investment carried out by the municipalities was funded by FISM, with an increasing share for the municipalities with higher marginality. The significance of FISM for municipal resources varies by sector and size of municipality. FISM represents 65 percent of the total infrastructure investment done in the poorest municipalities and 35 percent of all their expenditures. Table 3.11 presents FISM allocations--which are determined by the municipalities--per category for an average municipality as well as total funds distributed to the municipalities in 1999, 2001, and 2003. Funds are distributed as follows: 45 percent for water and sanitation, electricity, and roads, 26 percent for municipal urbanization, and 29 percent for other categories such as basic infrastructure in health and education as well as housing. FAIS allocations across different groups of states and municipalities according to their level of marginality do not show significant differences (see Volume II for states' analysis). Decentralized Service Delivery for the Poor 71 Table 3.11. FAIS Allocations, by Sector, 1999, 2001, and 2003 Unweighted shares Weighted shares (by FISM size) Category 1999 2001 2003 1999 2001 2003 Water 15.4 12.0 13.3 12.9 10.8 11.5 Sewage 4.9 5.7 5.4 6.1 Sanitation 8.6 4.6 5.1 8.6 4.5 5.3 Municipal urbanization 21.6 26.8 24.4 24.9 27.9 26.1 Rural electrification 11.6 10.4 10.3 11.6 10.8 10.4 Basic infrastructure in health 5.6 4.5 4.5 4.6 3.6 3.8 Basic infrastructure in education 13.6 12.7 13.0 13.0 11.6 11.0 Housing 3.4 2.6 3.5 3.3 2.4 4.0 Rural roads 11.2 12.0 11.5 10.4 12.6 13.5 Rural productive infrastructure 3.4 3.5 4.0 3.6 3.0 4.1 Education assistantships 2.0 2.3 Others 5.6 4.1 4.6 7.1 5.0 4.1 Number of municipalities 2,305 2,429 2,059 2,305 2,429 2,059 Source: World Bank staff estimates based on Encuestas a Presidentes Municipales 2000, 2002, and 2004. Only in the water and electricity sectors are FAIS allocations correlated with the municipalities' needs, while there is no correlation in sanitation. After evaluating the coefficients in the mean of the sample, differences in the level of electricity coverage are remarkable. Figure 3.19 illustrates these results, using a linear regression. In the case of water, the differences are not so remarkable, despite a negative slope when water coverage increases. In the case of sanitation, differences across municipalities with different levels of sanitation coverage are not significant. Finally, in the case of electricity, resources seem to be well targeted according to the needs of the municipalities.14 When coverage increases, the share of resources for electricity declines (see Volume II for states' analysis). 14 These figures illustrate the correlation between needs and allocations. However, the sectoral report in Volume II includes estimates in which controls for municipal income have been introduced. The results are qualitatively similar. 72 Decentralized Service Delivery for the Poor Figure 3.19. Correlation of FISM Allocations and Basic Infrastructure, 1999, 2001, and 2003 Note: The parameters for the lines were calculated with OLS regression of the shares of FISM allocations on the level of coverage for each sector. Source: World Bank staff estimates based on the Encuestas a Presidentes Municipales and census data. Many other resources are going to the municipalities for investment in social infrastructure. Table 3.12 shows that total investment in infrastructure (excluding telecommunications) was MxP$90.6 billion, equaling 1.4 percentage points of the country's 2003 GDP. The main investment is electricity, where the share of total investment was 47 percent, followed by roads with 35 percent, and water and sanitation with 18 percent. Within these amounts, FAIS participates with 10 percent, while the private sector contributes close to 9 percent. FAIS participation in electricity and roads is not remarkable; however, it is responsible for 29 percent of the investment in the water sector. Decentralized Service Delivery for the Poor 73 Table 3.12. Public Infrastructure Investment, by Sector, Excluding Telecommunications, 2003 MX$ millions Private Public investment Sector investment NonFAIS FAIS Total Water and 4,000 7,872 4,876 16,748 sanitation Electricity -- 39,921 2,341 42,262 Roads 0 24,200 2,277 26,477 Ports and railroads 3,800 1,330 0 5,130 Total 7,800 73,323 9,494 90,617 Note: Only water and sanitation, electricity, and road expenditures were included in the FAIS total, based on Encuesta a Presidentes Municipales. Source: World Bank staff based on World Bank, 2005b. The existence of other funds that target similar objectives may partly explain the lack of correlation between needs and allocations of FAIS. Other institutional arrangements can also help explain this: (a) the formula does not reward performance; (b) political, technical, budgetary (resources transferred in 10 monthly payments), and institutional issues in the design make it difficult for municipalities to plan for larger projects, which require greater sums of capital; (c) accountability is limited; and (d) FAIS uses different planning models at the local level. These issues are explored in more detail in chapter 4. IV. RURAL DEVELOPMENT Agricultural productivity is low in Mexico despite the large public investment in the sector (see Figure 3.20 and Figure 3.21). World Bank (2005a) highlights the need to improve productionoriented rural development programs. 74 Decentralized Service Delivery for the Poor Figure 3.20. Agricultural Labor Productivity in Latin American and Caribbean Countries, 2002 10,000 9,000 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0 Hai Bol Hon Ecu El Sal Gua Per Mex Nic Pan Par R.D. Col Cub Bra Ven Chi C.R Uru Arg. LAC Note: Measured in 1995 U.S. dollars per worker defined as agricultural value added divided by the agricultural labor force and measured in U.S. dollars of 1995. Source: World Bank 2005a. Figure 3.21. Public Expenditure in Agriculture per Agricultural Worker in LAC Countries, Average 19962000 450 400 350 20006 300 199 ,rekr 250 Wo r 200 pe 150 USD 100 50 0 ilehC xico .up l C an ú iab a al Me Re . Rica at guaar azirB ca LA ega Per genti Cos er Ar lomoC rodavl liviaoB Sa atemu Dom Ni El G Av Source: World Bank 2005a. This section analyzes the Alianza program and, in particular, its Rural Development Program to understand the design and operation of rural development programs in relation to decentralization and poverty in Mexico. First, Alianza is the flagship program of SAGARPA; it has been operating since 1996, bridging presidential administrations, and is likely to continue to grow. Second, most program funds have been decentralized to the states, and some (the majority of the Rural Development Decentralized Service Delivery for the Poor 75 Program funding) have been decentralized to the municipalities.15 Third, the Rural Development Program is a targeted program meant to favor small producers and poor rural dwellers in general, with special attention to marginal areas and vulnerable groups (women, youth, the old, indigenous families, and handicapped rural dwellers). The Rural Development Program has, hence, the double advantage of being decentralized to municipalities and targeted to the poor. Fourth, Alianza is a productionoriented program. This gives it the design and operational complexity typical of programs aimed at promoting independent entrepreneurial activity, which mostly deal with private goods. These programs require numerous and committed staff and plenty of expenditures on organization, training, and technical assistance, which makes them difficult to operate and costly in terms of recurrent costs.16 Finally, the Food and Agriculture Organization (FAO) has been evaluating Alianza for a number of years, and these evaluations contain relevant material.17 Coverage and Targeting Alianza is not a targeted program; only the Rural Development Program is targeted. However, the increased participation of the Rural Development Program in Alianza might have improved its targeting (see Table 3.13). Of the beneficiaries, 47 percent live in municipalities considered to have very high or high marginality (where 49 percent of the rural population lives), and a similar proportion of beneficiaries live in municipalities of medium or low marginality (compared to 39 percent of the rural population). This is in spite of the fact that the operation rules of the program indicate that 70 percent of the program funds managed by the states should go to localities of high or very high marginality. The distribution of funds is more regressive than that of beneficiaries because higher proportions go to municipalities of low and very low marginality, and a smaller proportion goes to municipalities of very high and high marginality. 15 The term "decentralized" is used here in the sense in which the term federalizado is commonly used in Mexico: that is, important aspects of program implementation are devolved to the state or municipal level, and subnational governments contribute to funding. 16 More on the importance of this type of program and the requirements and difficulties of implementing them can be found in World Bank (2005a, Rural Poverty Study, chapters 6 and 8). 17 FAO has carried out quantitative surveys of beneficiaries and systematic interviews with relevant actors in various years, in cooperation with the evaluation teams of the different states, and compiled a large database. With the agreement of SAGARPA, FAO agreed to make this database available to us, keeping the anonymity of the interviewees. We are grateful for this. 76 Decentralized Service Delivery for the Poor Table 3.13. Distribution of Rural Population, Beneficiaries, and Funds of the Rural Development Program, by Marginality of Their Area of Residence Marginality level Shares Very high High Medium Low Very low Rural population 13.9 35.5 22.7 16.7 11.2 Beneficiaries 6.7 40.1 23.8 23.3 6.0 Funds 1.7 30.2 23.3 37.1 7.8 Note: Rural population is defined as those living in localities of less than 5,000 inhabitants. Source: FAO, 2004 and World Bank staff estimates based on CONAPO 2000 and Székely and others 2005. FAO's evaluations of Alianza distinguish five types of beneficiaries. The classification is based on a composite index of five variables: education, value of assets, livestock, land, and (an index indicator of) technology.18 The average characteristics of the 2004 Rural Development Program beneficiaries are shown in Table 3.14. FAO considers types I and II as poor producers, although according to the average years of schooling they would seem to be better off than the rural poor identified using the ENIGH surveys. Close to 80 percent of the Rural Development Program beneficiaries belong to types I and II, compared with 56 percent of beneficiaries of the agriculture program and 34 percent of beneficiaries of the livestock program (Figure 3.22). Table 3.14. Characteristics of the Rural Development Program Beneficiaries, by Type, 2004 Type of beneficiary Average characteristics I II III IV V Total Beneficiaries (percent) 37.8 41.9 17.8 2.5 0.0 100.0 Age (years) 45.2 45.3 45.9 53.9 43.0 45.3 Education (years) 4.8 6.3 8.9 14.3 19.0 6.5 Value of assets (MxP$) 1,799 56,557 208,853 662,765 512,000 83,503 Number of equivalent cattle units 5.6 8.3 13.8 28.6 71.0 8.9 Irrigated land equivalent (hectare) 0.8 3.0 11.1 33.1 10.0 4.6 Technological level 0.2 0.3 0.5 0.6 0.8 0.3 Source: FAO, 2004. 18 The FAO evaluation of Alianza for 2004 was based on a sample of beneficiaries in 30 states with two sampling frameworks, one for beneficiaries of Alianza in 2002, which included 15,898 producers, and the other for beneficiaries of Alianza in 2004, which included 12,109 producers. The first sampling framework was used to estimate the impact of the 2002 Alianza program. Both samples included beneficiaries from the agriculture, livestock, and rural development programs. Interviews were carried out in April­June 2005. Decentralized Service Delivery for the Poor 77 Figure 3.22. Distribution of 2004 Alianza Beneficiaries, by Program and Type of Beneficiary 50 seiraicfie 40 30 Ben of egatnecr 20 10 Pe 0 Type I Type II Type III Type IV Type V Agriculture Livestock Rural Development All Alianza Source: FAO, 2004. Figure 3.23 shows the difference in the distribution of beneficiaries and funds. The percentage of funds allocated to beneficiaries of types I and II is smaller compared with the number of beneficiaries in these categories, while the opposite is true for types III and IV. There are hardly any type V beneficiaries of the Rural Development Program. We do not have information on the distribution of beneficiaries, by type, across states, and hence we do not know if targeting (that is, the participation of type I and II beneficiaries) is better in some states than in others. One would expect leakages to be more important in richer than in poorer states. During our field work, we did observe that the issue of leakage of the Rural Development Program benefits to comparatively richer farmers was more worrying to program operators in the wealthier state of Jalisco than in the poorer states of Guerrero and Veracruz, presumably because its incidence was higher in Jalisco. 78 Decentralized Service Delivery for the Poor Figure 3.23. Distribution of the Rural Development Program Funds and Type of Beneficiary, 2004 45 40 seiraicif 35 30 ne 25 Be 20 of er 15 10 Sha 5 0 Type I Type II Type III Type IV Type V Beneficiaries Funds Source: FAO, 2004. The FAO database only contains data on Alianza beneficiaries, without a control group. To overcome this limitation, the distribution of the Rural Development Program beneficiaries was compared with the distribution of the rural population captured in ENHRUM.19 The overall rural population (as surveyed by ENHRUM) was classified, using the FAO classification. The results are shown in Figure 3.24. Figure 3.24. Distribution of the Rural Development Program Beneficiaries and the Rural Population (Families) 80 70 60 egatenecr 50 40 30 Pe 20 10 0 Type I Type II Type III Type IV Type V Rural Population Beneficiaries Funds Source: Calculated on the basis of Alianza's Rural Development Program evaluation report and ENHRUM data. The two distributions are different. The poorest households (those in type I) are underrepresented in the population of the Rural Development Program beneficiaries, 19 The Encuesta Nacional de Hogares Rurales de México (ENHRUM), taken in 2002 in a collaboration between Instituto Nacional de Estadística Geografía e Informática (INEGI) and El Colegio de México, covers some 1,800 rural households and is representative at the national and large regional levels. Decentralized Service Delivery for the Poor 79 while those in types III and IV are overrepresented, in terms of both families and funds. The probability of a family of type I being a beneficiary of the Rural Development Program is five times less than that of a family of type II, 20 times less than that of a family of type III, and more than 50 times less than that of a family of type IV. Perceived Quality of the Program Alianza's funds seem to have had an effect on the economy of its beneficiaries. Figure 3.25 and Figure 3.26 show the relation between Alianza's investment subsidies and gross income, employment, assets, and technology of beneficiaries of the 2002 agriculture, livestock, and rural development programs. Figures relate to the period between when beneficiaries received Alianza's support in the second part of 2002 and when they were interviewed in the first part of 2005. Figures are based on beneficiaries' perceptions (stated during the evaluation interviews). On average, Alianza seems to have helped to expand beneficiaries' gross income by close to 17 percent, much more than employment, which increased only 3 percent.20 This was particularly noticeable in type II and type III beneficiaries and was larger in rural development than in other programs. Employment was largest for type I beneficiaries and smaller for all other beneficiary categories. Figure 3.25. Percentage Increase in Employment and Gross Income of Alianza 2002 Beneficiaries between 2002 and 2005 35 30 25 20 15 10 5 0 I pe II e III e IV e V e e kcot ent zan m Ty Typ Typ Typ Typ culturi lia ves lar Li Ru A l Agr velop Al De Employment Gross Income Source: FAO, 2004. 20 Most of the impact on gross income (76 percent) was due to growth in output, following increases in scale (73 percent) and yields (27 percent). The remaining 24 percent was due to improvement in prices, following marketing and qualityenhancing investments. 80 Decentralized Service Delivery for the Poor Figure 3.26. Percentage Increase in Assets and Technology Level of Alianza 2002 Beneficiaries between 2002 and 2005 100 90 80 70 60 50 40 30 20 10 0 I e II e III IV V e zan m Typ Typ Type Type Typ uretluc kcot ent lia ves lar Li Ru A l Agri velop Al De Assets Technology Source: FAO, 2004. As a consequence of Alianza support, the value of beneficiaries' assets increased 13 percent,21 and the technology index22 increased 21 percent. Alianza's impact on assets and technology was largest for type I beneficiaries, decreasing rapidly as the category of beneficiaries increased, and was larger for the Rural Development Program than for the other programs. There is no information on the aggregate impact of Alianza on national employment and output, but it is expected to be small in view of the limited number of beneficiaries compared with the total number of producers. Thus the number of direct beneficiaries of the three main decentralized programs of Alianza was 165,000 individuals in 2002,23 compared with a rural labor force of 10 million and an agricultural labor force of 6 million. Since the number of direct beneficiaries and the number of workers in the rural labor force have changed little in recent years, we can conclude that, each year, some 1.7 percent of the labor force in rural areas and some 2.7 percent of the persons employed in agriculture receive direct investment support from decentralized Alianza programs. 21 This includes the value of the assets obtained with Alianza support and an additional increase of 43 percent in the value of beneficiaries' assets that they ascribe to the effect of Alianza. 22 In the case of agriculture, the index of technology includes indicators of quality of planting material, use of machinery, availability and type of irrigation, and use of fertilizer. In the case of livestock, the index includes indicators of the quality of stock, type of feeding used, and type of infrastructure. The index moves between 0 and 1. Impact is calculated here as the percentage change in the value of the index between 2002, before Alianza support, and 2005. 23 Calculated dividing the total amount of resources going into the three main decentralized programs of Alianza in 2002, which was Mx$8.3 billion, by the average value per beneficiary of Alianza resources for these programs in that year, which was Mx$50,000 (both figures include the contribution from SAGARPA, state governments, and beneficiaries). Decentralized Service Delivery for the Poor 81 V. A COMMON CHALLENGE FOR EVALUATION AND ACCOUNTABILITY In addition to the individual sectoral conclusions summarized at the beginning of the chapter, all four areas face the common challenge of the lack of good information for purposes of performance evaluation and accountability. In the case of targeted programs, this should start with knowing who gets coverage from each program and who does not. The most efficient way to proceed is not program by program, which has been the tendency thus far in Mexico, but rather to continue improving the registry of beneficiaries of programs and also of sectoral policies. Box 3.1 discusses the relevant experiences in Argentina and Chile. Box 3.1. Implementing a Nationally Coordinated System for Citizen Identification and Exchange of Fiscal and Social Information: Lessons from Argentina and Chile Many governments face serious problems in accessing information that allows for accurate targeting of social programs and the reduction of tax evasion and informality. All of these issues present challenges for Latin American countries. An identification and information exchange system allows governments to jointly confront the problems of informality and social expenditure. Data or personal information can be used for both purposes. In order to achieve economies of scale, the information processed by one department has to be available to other branches as well. Therefore, a nationally coordinated system that allows and facilitates the exchange of data is needed. Two kinds of problems arise when a government does not have an adequate system of identifying and exchanging social and tax information. The first type of error refers to giving social assistance to the nonpoor--that is, giving assistance to those who do not need the most aid. The second type of error is on the tax side--that is, auditing individuals with a good level of tax compliance and not auditing the tax evaders. Aims of the system. The goals can be summarized as identifying social, fiscal, and wealth attributes of individuals and households, a concept that is broader than that of the family. Coordinating a system that exchanges database information ultimately improves the targeting of public expenditure social programs, helps to detect tax evasion, and assists in protecting the right to privacy. Costbenefit analysis. The functioning of this sort of system has allowed Argentina and Chile to obtain important benefits, including less informality and bettertargeted social public expenditure. In Argentina, during the first two years (1999 and 2000) of the Sistema de Identificación Nacional Tributario y Social (the National Tributary and Social Identification System), which cost US$4.1 million, the program produced a return of US$157.0 million (US$52.8 million in the tax area and US$104.2 million in the social area). 82 Decentralized Service Delivery for the Poor Characteristics. A system for identification and information exchange has to include the following characteristics: 1. Dynamic. The system has to be frequently updated. The database should not prioritize or increase the existing physical connections, but it should focus on how to improve the communication between them. 2. Autonomous. Each organism has to have autonomy. 3. Integrality. It should include the whole population and their attributes. To achieve this objective a continuous process of information exchange that comes from sources with information relating to the fiscal, social, and wealth factors is necessary. For instance, information from pension systems, labor taxes, property and automobile registers, income taxes, and others is necessary for improving the allocation of resources, such as for improving targeting and revenue collection. Integrality implies that the system must include data from the private sector, such as information from insurance companies. How to implement the system. A system of such characteristics could be implemented in the following sequence: 1. Introduce the system to the executive organism that will be responsible for its coordination and management. The arrangement has to be created or ratified by law instead of being a norm of lower hierarchy. These two important requirements will help to institutionalize the system. 2. Construct the initial data matrix with the necessary information, including data on the existing social programs, tax information, and laws concerning the exchange of data and privacy of information. 3. Begin phase I of the demonstration by selecting programs that will be incorporated initially to the system, with the aim of showing how the system functions. 4. Develop phase II of the implementation by incorporating the remaining database. 5. Implement phase III by consolidating the system. Decentralized Service Delivery for the Poor 83 ANNEX 3.1. ANALYSIS OF THE PERFORMANCE OF THE SECTORS The table below lays out the performance of the programs and links it to the sources of information available in this report. Dimension of Performance Analysis Sector/ Sources of Coverage / Targeting Quality Local Resource Program Funding Mobilization (Chapter 2) PISA Net enrollment in International Basic Education comparisons FAEB Total Controlling for PEC By poverty status GDP State CONAFE-- Basic and location State, by contribution AGEs Education Schooling by deciles marginality level s to basic FAETA, FAM, Population over 15 By type of school education Convenios (Ramo years of age without Performance of 11) lower secondary by different types of state schools-- telesecundaria National and state trends on life expectancy, infant mortality, probability to die before age 5, Population by maternal mortality institution of Infant mortality by affiliation Health for municipal SSA Uninsured population the marginality FASSA by states State Population Households with IMSS Distribution of Seguro contribution without excessive health Oportunidades Popular, Oportunidades s to Seguro Social expenditures-- Seguro Popular and Uninsured by Popular Security national (Ramo 12) decile Coverage Quality/ attention Concentration indicators by coefficients of health provider providers/ programs Final and intermediate outcomes and catastrophic expenditures for 6 states 84 Decentralized Service Delivery for the Poor Dimension of Performance Analysis Sector/ Sources of Coverage / Targeting Quality Local Resource Program Funding Mobilization (Chapter 2) FAIS allocation by sector Infrastructure investment--FAIS vs. others FAIS--Social FAIS Households with Infrastructur FISE piped water, sewage, e FISM and electricity by state Access to services by municipal marginality Correlations of FISM allocations with coverage Desarrollo Rural Agricultural labor beneficiaries by type productivity in LAC Distribution of rural PEC Public spending by population, Alianza Procampo worker in LAC Desarrollo beneficiaries and PET Increases in State Rural of funds, by marginality Alianza-- employment, Contributio Alianza para Distribution of Alianza Federal income, assets, and n to Alianza el Campo beneficiaries by type Desarrollo technology of and program Rural Alianza beneficiaries Distribution of funds by type and and beneficiaries by program type Decentralized Service Delivery for the Poor 85 ANNEX 3.2. INSTITUTIONAL ARRANGEMENTS AFTER DECENTRALIZATION Table A1. Education--State Resources State System and/or Teachers Paid with State Resources Does it count Percentage of with a state teachers paid State system of basic with state education? resources Aguascalientes no 3 Baja California yes 39 Baja California Sur no 0 Campeche no 10 Chiapas yes 15 Chihuahua yes 40 Coahuila yes 10 Colima yes 10 Distrito Federal no 0 Durango yes 25 Guanajuato yes 30 Guerrero no 20 Hidalgo no 0 Jalisco yes 40 México yes 55 Michoacán no 10 Morelos no 3 Nayarit yes 20 Nuevo León yes 41 Oaxaca no 0 Puebla yes 20 Querétaro no 0 Quintana Roo no 0 San Luis Potosí yes 10 Sinaloa yes 29 Sonora yes 30 Tabasco no 20 Tamaulipas no 10 Tlaxcala yes 10 Veracruz yes 10 Yucatán yes 25 Zacatecas yes 20 Source: Jesús Álvarez, 2006. 86 Decentralized Service Delivery for the Poor Table A2. Education--Level of Conflict Between Authority and Union (1=low; 2=medium; State 3=high) Aguascalientes 1 Baja California 1 Baja California Sur Campeche 2 Chiapas 2 Chihuahua 1 Coahuila 1 Colima 1 Distrito Federal 2 Durango 1 Guanajuato 1 Guerrero 3 Hidalgo 2 Jalisco 1 México 2 Michoacán 3 Morelos 1 Nayarit 2 Nuevo León 1 Oaxaca 3 Puebla 1 Querétaro 1 Quintana Roo 1 San Luis Potosí 1 Sinaloa 1 Sonora 1 Tabasco 2 Tamaulipas 1 Tlaxcala 3 Veracruz 2 Yucatán 1 Zacatecas 2 Source: Jesús Álvarez, 2006. Decentralized Service Delivery for the Poor 87 Table A3. Education--Decentralization: Regional Offices Have you implemented any Administrative administrative measures in order to Pedagogical State Functions increase the control of incidences of Functions Decentralized the personnel? (with or without Decentralized decentralization) Aguascalientes yes yes yes Baja California yes no yes Baja California Sur Campeche yes no no Chiapas no no no Chihuahua no yes no Coahuila yes yes no Colima yes yes yes Distrito Federal yes yes yes Durango yes no no Guanajuato yes yes yes Guerrero yes no no Hidalgo yes no no Jalisco yes yes no México no yes no Michoacán yes no no Morelos yes yes no Nayarit no no no Nuevo León yes yes yes Oaxaca yes no no Puebla yes no no Querétaro yes yes yes Quintana Roo yes yes no San Luis Potosí no no no Sinaloa no yes no Sonora yes no no Tabasco no yes no Tamaulipas no no no Tlaxcala no no no Veracruz no no no Yucatán no yes no Zacatecas no no no Source: Jesús Álvarez, 2006. 88 Decentralized Service Delivery for the Poor Table A4. Education--School Autonomy, Social Participation, and Evaluation School Autonomy and Social Evaluation Participation 1=national; 2=local efforts; 3=inform results; (1=low; 2=medium; 3=high) 4=incipient feedback; 5=design interventions School Participation of State Autonomy Parents in Increase Schools Aguascalientes 3 3 4 Baja California 2 1 4 Baja California Sur Campeche 1 2 3 Chiapas 1 1 3 Chihuahua 1 2 2 Coahuila 1 2 2 Colima 3 3 5 Distrito Federal 3 2 5 Durango 1 2 2 Guanajuato 3 2 4 Guerrero 2 2 1 Hidalgo 1 2 4 Jalisco 1 2 1 México 2 2 1 Michoacán 1 1 1 Morelos 3 2 4 Nayarit 1 1 2 Nuevo León 3 3 4 Oaxaca 1 1 2 Puebla 1 1 1 Querétaro 3 2 3 Quintana Roo 2 2 3 San Luis Potosí 1 2 2 Sinaloa 2 2 2 Sonora 2 2 4 Tabasco 2 3 1 Tamaulipas 2 2 2 Tlaxcala 1 1 1 Veracruz 1 1 2 Yucatán 3 1 2 Zacatecas 1 2 1 Source: Jesús Álvarez, 2006. Decentralized Service Delivery for the Poor 89 Table A5. Health--Governance Strengths in Six Mexican States Baja Strengths California Sur Colima Jalisco Oaxaca Hidalgo Tabasco Identification and use of legal + ++ + + ++ frameworks and reform strategies Participation in local level + + ++ ++ decisionmaking Management and resource reallocation + +++ + ++ +++ capacity at the state level Accountability with state participation of ++ + state and municipal actors Participation of all actors in the + ++ + + ++ lobbying for new sources of funding Greater interaction among the social actors: providers, ++ + + +++ users, governments, and NGOs Transparency in the management of + + +++ + programs and public resources Intermediation mechanisms to support + +++ ++ ++ ++ vulnerable/priority groups Level of agreement and political stability among + + +++ + +++ actors and their interactions High= +++ Medium= ++ Low = + None = Source: Arredondo, 2005. 90 Decentralized Service Delivery for the Poor CHAPTER 4 INSTITUTIONAL DIMENSIONS OF DECENTRALIZED SERVICE DELIVERY In the provision of public services, it is a challenge to get the actual providers to be accountable for quality and efficiency as well as to have the resources and management authority to do the job well. The accountability should be both to the political decision makers who hire them (on behalf of taxpayers) and to the clients for the service. In all the sectors considered here, the decentralization process has primarily been to state and local governments; in some places secondary institutional creations or even parallel subprograms have sought to create more direct involvement of citizens and clients. Getting these institutions to generate accountability has been a challenge everywhere. Although it is hard to prove definitively, anecdotal evidence supports the presumption of this report that more transparency and accountability will lead to better social services and infrastructure for the poor. Part of the problem has been that the federal government has maintained a significant role and not allowed enough local authority--in education, with national union negotiations and rigid curriculum dictates; in health, social infrastructure, and rural development, with large national programs that dominate their respective sectors. The large federal programs dominate the outcomes of the sector, making it difficult to discern the impact of any locally controlled programs and focusing people's attention on the federal government as the one responsible. In the four areas investigated here, the variety of institutional contexts means that the challenges differ, in at least four ways. First, the implementing levels of government differ: states handle basic education and most health care for the open population, while municipalities handle social infrastructure (more than 80 percent of FAIS) and the Rural Development Program of Alianza, on which this report focuses. Second, the importance of the programs within the sectors differs widely. The SEP basic education program decentralized through FAEB provides basic education to the majority of Mexican citizens and almost all of the education accessed by the poor. The population without social security coverage, which overlaps substantially with the poor, has access to three public health programs, considered here, but the insurance systems constitute the financial majority of the health sector and serve half or more of the population. FAIS/FISM is the largest source of funding only in very marginal municipalities and is much less important in large municipalities. Furthermore, in the main areas of investment--water and sanitation, roads--other federal and state Decentralized Service Delivery for the Poor 91 programs are much larger. In the context of the agriculture sector and the vast array of federal programs, the Rural Development Program is small. Third, in two of the sectors--education and health--the service providers are well organized, and these unions effectively represent their interests in the policy making and operations of the sector. In basic education, the unions are especially strong; nonetheless, some, but not all, states have found ways to cooperate with the unions in achieving some reforms. Fourth, in a related dimension, the issues of the sectors and programs enter differently into the wider political process. In education, quality issues occasionally enter the political debate at the national level surrounding the federal budget process and in those states with important state school systems, such as Nuevo León. The teachers' union takes an active role in national party politics. Moreover, at the national level, with Mexico's participation in PISA, it has become a more general topic, with all of the current presidential candidates discussing the issue. Farmers' unions are also politically powerful, although less powerful than the teachers' union, but the Rural Development Program probably does not attract much of their attention because of its small size relative to the rest of agriculture spending. In health, the most political attention at the national level goes to the social security system; at the state level more attention is paid to the programs for the poor, especially when there is some obvious problem, such as a sharp increase in infant or maternal deaths. FAIS and FISM have political implications for local elections and intergovernmental relations, but little political attention is paid to the particular services they seek to deliver. I. ACCOUNTABILITY FRAMEWORKS ADAPTED TO MEXICO In Chapter 1, we discussed the accountability framework development for the 2004 WDR. To adapt this framework for Mexico's decentralized sectors and programs in this study requires some disaggregation, with details varying for each area. Subnational governments (state and municipal governments) are added to the picture, making political parties explicit and, which is important in the case of Mexico, making explicit the role of worker organizations (teachers and health workers unions, for example). These organizations can have relationships with political parties and influence the design of government policy, in addition to their more traditional role of bargaining with managers over working conditions. Thus the long route of accountability is from the provider back to the government and ultimately to the citizen--taxpayer and service client (not generally in the same proportions)--and the short route of accountability is straight from provider to client. The short route should ensure that the provider is responsive within the context of the larger structure dictated by the long route. It helps to ensure that people are playing by the rules. For example, PEC schools have less corruption in that teachers impose fewer extra fees. The short route seems to facilitate communications greatly, 92 Decentralized Service Delivery for the Poor although the long route has some role to play in assuring technical capacity to supervise and in getting accountability to the taxpayer (Fiszbein 2004; World Bank 2004c). Even the customized WDR framework cannot capture all the important institutional dimensions. We must recognize, for instance, the need to develop and retain competent personnel at local levels, the need to consolidate federal programs and streamline their regulations so that local government can readily access them, and the need to create effective linkages among federal programs and between them and the locally run programs. International experience shows the possibility of several ways to achieve adequate accountability relationships; Box 1.1 summarized the experiences of Canada and Germany. In Mexico the law leans toward the more Germanstyle accountability back up to the top, but the Germanic bureaucratic competence and willingness to obey authority seems absent. As we shall see in the sectoral discussions, some state and municipal governments are implementing systems with more Canadian style local accountability. Bringing in subnational governments adds a node along the government provider side of the triangle, but accountability to citizens would be enhanced if their voice were linked more closely to the subnational governments than to the national government--creating a shorter "long route." This is a common argument in favor of decentralized government, but how well it has worked varies across states and sectors in Mexico. Grindle (2005) does an indepth study of 30 Mexican municipalities, based on a random sample of states and municipalities which is useful for understanding the municipal process of governance and budgeting. She explores four hypotheses on how decentralized initiatives may lead to better governance: local political competition, entrepreneurial mayors, better technical conditions (computers, management techniques), and local social mobilization for accountability. She finds that there is wide variation in municipal governance, but a more complex pattern of interactions, especially between the first three factors, illustrating the complexity of "building" both the long and short routes to accountability. Entrepreneurial mayors stand out as most significant, but they are also products of more open political competition. Importantly, local social mobilization is characterized more by resource extraction from local governments (building on old clientelist cultures) than by holding government accountable. Further decentralization is fundamentally about a combination of increasing the transparency of both resource allocation and the intergovernmental division of responsibilities, along with the overall task of deepening democracy and strengthening governmentsociety links. II. EDUCATION For the main program of basic education in Mexico, the critical triangle of two way interaction is the federal government, the teachers union (Sindicato Nacional de Decentralized Service Delivery for the Poor 93 Trabajadores de la Educación, SNTE), and political party. Figure 4.1 shows the basic diagram with the addition of the union and the political party as well as the state governments. The 1992 law that decentralized basic education was negotiated between the federal government and the union, with the states and the teachers themselves as observers and parents and students far out of the loop. That model remains the core of the sector today, setting the annual increase in base salary through negotiations at the top, although the state branches of the union negotiate supplemental agreements that give additional benefits. The citizens have weak influence, mediated through the party, and no direct influence on providers; this is the short route to accountability. Chapter 2 discussed some important movements away from this traditional model. Deriving from Mexico's traditional structure as a corporatist singleparty state, the SNTE has identity, interests, and power that are distinct from the teachers they nominally represent. Rival unions are not tolerated.1 Although the formal link between SNTE and the PRI was legally dropped in the late 1980s, the effective link remained strong and unquestioned through the 1990s, with SNTE negotiating decentralization of the sector and the head of SNTE leading the PRI delegation in Congress. After 2000, President Fox and the SNTE leader cooperated somewhat, outside the party framework. In 2002­04, the SNTE wing of the PRI acted with increasing independence from the rest of the party, and by 2005­06 the SNTE leadership had formed its own party, splitting off many teachers from the PRI.2 How this will play out in the long term is unclear, but the strong political role of the SNTE has added a unique dimension to its position in negotiating salaries and work rules. The SNTE is strong not only because of its labor and political representation but also because of its powerful position within the SEP itself. The undersecretary or director of personnel is a union member who is selected by the union leaders; around 1.5 percent of the salary budget is automatically transferred to the SNTE coffers, with no accountability for the use of these funds. The union has often overwhelmed the states, despite their theoretical legal power to manage the schools. Figure 4.1 represents the dominant relations in the sector, with human resources and wages dictated largely by the channel between the teachers union and the federal government. 1 SNTE and in particular certain local sections often mobilize in relation to noneducation labor issues (in support for other unions), as well as more general political issues that negatively affect service provision because of work stoppages. 2 Despite the danger that an SNTEbased party would perpetuate the high politicization of the sector, it would at least bring it out into the open. The central plank of the new party's platform is a promise to improve education. Having the party take open political responsibility for achieving this objective could improve accountability in the sector. 94 Decentralized Service Delivery for the Poor Figure 4.1. Channels of Influence in the Mexican Education Sector Federal government Teachers Political union party State government Studentsfamilies/ Schools/ Citizensvoters teachers Political influence Taxes/resources Sectoral policy and operational practices (wage negotiations) PEC and AGEs In basic education, some states have their system of schools (separate from the decentralized federal schools), and some do not. While not having a state system seems to be a characteristic of poorly performing Southern states (which also have socioeconomic reasons for poor education outcomes), the role of the state system in the rest of the states is not strongly associated with performance. Box 4.1. Local Institutional Innovations in Education: The Case of Colima This state has made notable achievements in student performance through a combination of traditional practices and strategic innovations. A small, well connected state, with just two major urban centers, Colima has undoubtedly benefited from these characteristics in its efforts to improve educational outcomes. Continuity has been another factor, with the same secretary of education for the past eight years (and four governors). The starting point for Colima's efforts is a well functioning information and planning system that benefits from unique files for each student and modern communications and computation systems. This has allowed for important decentralization to the municipal level with ten units proving administrative functions in personnel management, buildings, equipment and materials, and four regional centers for teacher training, which also provide technical assistance. Decentralized Service Delivery for the Poor 95 Prior, and independent from the PEC program, almost all schools have benefited for a number of years from the promotion of school autonomy, with tools for diagnostics and activity planning. Supporting these local initiatives is a rather well developed system of assessment, which test annually at five different grade levels, in every school. The results are made available for the subsequent school year for each student, grade, school, and municipality and provide for accountability and the targeting of incentives tied to results. They also provide feedback for quality enhancement activities undertaken by the school. Local authorities also attribute their performance to two other innovations: competitive testing for half of all new teachers; and agreement by organized labor not to rotate teachers during the school year (in other states, such rotations start a chain reaction that can result in some classes having 34 different teachers during the year). Source: Volumen II. Some states have taken more control of their formerly federal schools and teaching staff, through negotiation with the local branch of the union, in order to respond to the education preferences of their families/voters. For instance, a number of states now require testing of teacher applicants, instead of hiring only requiring a normal school degree; several use the exams to have competition for the new positions among them: Aguascalientes, Chiapas, Jalisco, Quintana Roo, Tabasco, Veracruz, and Yucatán. The important issue here may not be who does the hiring--after all, it is not clear that state bureaucracies do any better than the federal government--but that the selection system is transparent and competitive. Supervision is another area in which some states have tried to strengthen management. The issue is related to who does the supervising and how they do it--Veracruz has brought in nonunion personnel as supervisors. Some states just ensure that teachers show up and apply the curriculum (schools themselves, with parent participation, might do this better than the state), but some of the more advanced states actually supervise progress in school autonomy, implement improvement plans, and coordinate state support for these plans. About half of the states have successfully merged the formerly federal schools with the states' own pre1992 systems. Seventeen states have decentralized administrative functions to regional or municipal levels, and seven have decentralized pedagogical functions (supervision). Six have gone ever further, allowing schools to define their problems and propose solutions. In these states, assessment systems help to define problems. The chapter on education in Volume II of this series describes the statelevel innovation in five of the states with betterperforming schools: Aguascalientes, Colima, Distrito Federal, Nuevo León, and Quintana Roo. Still, by international standards, those politically elected and responsible state (and federal) authorities have relatively weak effective power to manage the largest area of the public sector. 96 Decentralized Service Delivery for the Poor Box 4.2. Local Institutional Innovations in Education: The Case of Aguascalientes Many of the innovations of this state were subsequently incorporated into national policy and programs. One of its early initiatives was also to further push decentralization of administration and pedagogy to the regional level. One of the benefits of this was to increase control over human resources and to gain fiscal resources to fund local initiatives. These include a program of teacher training that helped 32 percent of state teachers to pass the first national testing in 1996, as compared to only 8 percent nationwide; scholarships for disadvantaged students; the incorporation of learning technologies in the schools; student testing; and social participation. Indeed, its focus on school autonomy and parental participation was the genesis for the national program for school quality (PEC) launched by the federal government in 2001. More recent initiatives have included pushing for the incorporation of its schools into the national program Enciclomedia (basically interactive textbooks) and a locally financed effort to begin teaching English in primary school, in contrast to the national curriculum, where English is only incorporated in lower secondary. As for performance, success was rapid. History test scores improved from 4.1 to 5.3 (scale of 10) between 1995/96 and 1996/97; but math scores declined. Enrollments among 5 to 14 yearolds increased from 90 to 99 percent from 1992 to 1997, moving the state from 10th to 4th overall in the country. Preschool coverage for 5yearolds was universal by 1997, up from 70 percent in 1992, and moving the state to 2nd place overall in the country, up from 17th. Source: Volumen II. Some federal programs--namely Oportunidades, CONAFE, and PEC--have bypassed the traditional state systems (which still finance teacher salaries) and gone directly to the citizens, communities, and schools in order to achieve reforms by forging accountability connections between families/voters and schools/teachers. As is well known, Oportunidades provides a subsidy directly to poor households on condition that the children attend schools. It has succeeded in motivating better attendance among the poor but was not intended to interact directly with the schools, as the other two programs do. PEC is a pilot program that brings federal funds to school communities that agree to meet the program requirements. Three parties need to participate: (a) the administrators of the individual school, who must develop and implement a fiveyear school improvement plan, (b) the state, which must match funds 1:3 with the federal contribution, and (c) the school community, which must participate in the planning process and can make a financial contribution in order to receive a second federal matching grant.3 Parents have a fundamental role in this, even if they do not contribute 3 The community financial contributions come from a variety of sources--parents, municipalities, and the private sector. Decentralized Service Delivery for the Poor 97 financially. It is not clear where the idea originates for a school to volunteer for PEC, but several states enhance the incentives for schools to participate by providing funding for the requisite planning and training. The most fundamental innovation seems to be the provision of a grant conditional on steps to improve the involvement of the states and communities, with an emphasis on accountability to the clients. The extra funds can only be used for nonsalary purposes. Putting the school at the center is the intent of PEC, and funding is just an incentive. For instance, Colima, one of the most successful states, has implemented PEC without receiving extra federal funds, because the state believes in the value of the program. Although PEC works on top of, rather than replacing, much of the traditional structure of basic education--salaries and work rules are negotiated nationally with the union, which still governs teachers in the PEC schools--it apparently does provide enough financial incentive and moral support for schools and communities to improve the functioning of that structure, as noted in Chapter 3. It worked well as a pilot program and has expanded to more than 10 percent of schools. As the program expands further, a key challenge will be to sustain conditionality and focus, yet avoid becoming another routine topup of school funding, irrespective of what happens in the school.4 To this end, PEC should remain a federal program under Ramo 11 (federal money not assigned to the states) and not be commingled with Ramo 33, to avoid diverting funds from the original intent of the program. CONAFE supports development of curricula, didactic materials, and textbooks in marginalized rural communities. While the main program does not alter the traditional institutional structure, in most beneficiary schools a group of community parents and leaders receives a grant that can be spent on an educational purpose selected by the group. Not only do these funds have the short route of accountability, with direct involvement of clients, they may also have a spillover effect by getting the community involved in the school and thus building an informal assessment and accountability relationship for the school's educational activity. Box 4.3. International Experiences on School Autonomy Starting in the United States, the United Kingdom, Australia and Canada, programs to promote school autonomy (or schoolbased management) have been implemented or are being developed in a number of countries, including Hong Kong (China), Indonesia, El Salvador, Nicaragua, Kenya, Kyrgyz Republic, Nepal, and Paraguay. The programs lay along a continuum of the degree to which decision making is devolved to the local level, from limited autonomy (such as Mexico's PEC), to more ambitious programs that allow hiring and firing teachers (such as El Salvador's Educo program), and more complete programs 4 This was, for instance, the fate of the wellfunded Carrera Magisterial, which evidently does not improve the quality of outcomes, despite its high fiscal cost. 98 Decentralized Service Delivery for the Poor that give schools control over substantial resources (such as those in Chicago). They include education systems that promote schoolbased management, school budgets, choice, and private and community management of schools (for example, the Netherlands), as well as proposals to allow parents to create schools (for example, the United Kingdom, which is similar to what already exists in Denmark and the Netherlands). Evaluations of schoolbased management programs offer mixed evidence of impacts. El Salvador's Educo program gives parent associations the responsibility for hiring, monitoring, and dismissing teachers. In addition, the parents are also trained in school management, as well as on how to help their children with school work. Despite rapid expansion of Educo schools, education quality was comparable to traditional schools. In fact, parental participation was considered the principal reason for Educo's success (Jimenez and Sawada, 1999). Nicaragua's Autonomous Schools Program gives schoolsite councils-- comprised of teachers, students, and a voting majority of parents--authority to determine how 100 percent of school resources are allocated and to hire and fire principals, a privilege that few other school councils in Latin America enjoy. Two evaluations found that the number of decisions made at the school level contributed to better test scores (King and Ozler, 1998; Ozler, 2001). Various evaluations of schoolbased management programs in the United States have found evidence of decreased dropout and student suspension rates but no impact on test scores (Collins and Hanson, 1991; Taylor and Bogotch, 1992; South, 1991). Source: Volume II. III. HEALTH The health sector in Mexico is more fragmented than the education sector, with half of the population and 80 percent of public spending going to the public insurance programs (IMSS, ISSTE, PEMEX, and so forth) and an equally large amount of spending coming out of pocket or from private insurance and going to providers in the private sector (some of whom are also employed in the public sector). Even the poor spend substantial amounts on health care, especially in catastrophic situations. Thus the public programs that go mostly to the poor--SSA, IMSSOportunidades, and Seguro Popular--are a small but increasing part of the total health care sector, although they are the main providers for the poor. SSA In terms of financial resources and number of beneficiaries, the Ministry of Health funds the largest health program for the poor, which the states operate. As with education, the national union negotiates salaries and work rules and plays a strong role in dealing with the individual states. Thus the basic pattern of the sector is similar, with decentralization diagramed in Figure 4.2, but the emphasis differs. While labor relations Decentralized Service Delivery for the Poor 99 are important for the SSA programs in the states, they are less dominant than in basic education because the nonsalary aspects of the sector (facilities, medicines) are more important in health than in education; this occurs largely because SSA workers have opportunities for more remunerative employment in the sector (IMSS, private sector) and because the dominant, most politically active union in the health sector is that of the IMSS workers. The union of SSA workers does not play a prominent political role because (i) Seguro Popular--which handles significant resources--negotiations are local and independent; and (ii) the states also hire through the Organismos Públicos Descentralizados which are independent from the union. Also health outcomes are more publicly obvious--infant mortality, for example--than education outcomes and they tend to grab headlines and the attention of state and local politicians. Some states, like Jalisco and Tabasco, have taken a more active role in managing SSA at the state level and put in more of their unearmarked resources (nonRamo 33) as the direct federal contributions have declined. In other states, like Baja California, Colima, Hidalgo, and Oaxaca, the federal role has stayed large, along with some growth in local contributions. Figure 4.2. Channels of Influence in the Mexican Health Sector: SSA Federal government Nurses and doctors union State government Clients/ citizens/ Hospitals/ voters clinics Political influence Taxes/resources Sectoral policy and operational practices (wage negotiations) The institutional arrangements vary across states, such as in the extent to which the former state health system has been merged with the sistema federalizado. Where there were local and state hospitals before decentralization, as in Baja California Sur, Jalisco, and Tabasco, combining them with the federalized facilities had led to institutional strengthening. Where states are contributing substantially from their own discretional resources (not Ramo 33), the state secretaries of health have exercised more control over their systems, including the hiring and assignment of personnel. Generally, however, the national health workers' union negotiates the salaries and many of the 100 Decentralized Service Delivery for the Poor working conditions with the federal government, and most states therefore still have a weak hand in managing health care personnel. Box 4.4. Local Institutional Innovations in Health: Several States Coordination mechanisms among the Federal, State, and Municipal levels to reach public and private/public agreements: There has been some progress in the implementation of the annual monitoring systems for the renewal and adjustment of the Human Resources and infrastructure management agreements according to the State's Health System. These agreements allow for further adaptation in the assistance model and in the reallocation of resources. However, the level of influence from the Federal level on the management of resources and health programs continues to be high on key areas of the National Health Program. For the state and municipal arrangements, Management Agreements have been implemented by sanitary region at the municipal level, adjusted according to demand of health programs and to the existing infrastructure. Regarding the private/public partnership, progress has been achieved in further decentralized states such as Jalisco. States are taking more active participation in the creation and implementation of coordination instruments for further decentralization. An example of this are the Health State Service Agreements, which make it possible to coordinate and implement health activities financed by the public sector at the state level, with the health services being delivered by the private sector. According to Arredondo (2005), the Seguro Popular is generating management and coordination agreements with the private sector through payment per event system. In addition to the above, Grant Agreements are being negotiated with the private sector for public health services in the most marginalized areas, which lack public sector infrastructure. Institutional/intergovernmental agreements and coresponsibility of health financing among the federal, state, and municipal levels: The decentralization agreements regarding health financing have allowed for important progress in some states regarding financial coresponsibility at the state level and, in some cases, even at the municipal level. There are indeed states where the local government is responsible for 4050 percent of the expenditures on health, and though the municipal government's contribution is not nearly as high (12 percent), there is evidence of its financial responsibility. No evidence of this was present before the decentralization process began. Furthermore, the culture of financial co responsibility created with decentralization has made it possible, in most states, to implement programs such as the Seguro Popular with financial contributions from the state and federal governments, as well as from the households. And although there is still a long way to go, this culture has also allowed for progress in the financing and accountability indicators, for higher transparency in the allocation and use of resources for health, and especially for finding alternative financing sources at the local level. Source: Volume II and Arredondo, 2005. Decentralized Service Delivery for the Poor 101 Seguro Popular Seguro Popular is a new program, started in 20015 to help the population without social security coverage, especially the poor, face catastrophic health situations and to move toward a funding scheme based on demand--number of beneficiaries--rather than on the supply of medical personnel and their distribution as was done in the early 1990s. The objectives of the program are to create a culture of prepayment for health services, to create demandside incentives, and to promote client power visàvis providers. So far, in the first phase there has been only a little change from the standard SSA model. The state receives more money if it registers more beneficiaries, so any increase in funding tends to follow the statetostate location of beneficiaries, and states have an incentive to sign up more beneficiaries. They have to check if the registrant is already in one of the regular insurance schemes (for example, IMSS) but they do not have an incentive to check if the family is already registered in another state or really qualifies for the reduced (or zero) fee.6 Families are supposed to renew their registration and eligibility each year. Once a beneficiary is registered, neither the choice of a clinic or hospital nor the patient's satisfaction (or dissatisfaction) with treatment has any effect on the financing that goes to the provider. So the clients have little real power or ability to influence providers, as they would with a scheme that reimbursed clinics, hospitals, and doctors on the basis of the number of patients and treatments (see Figure 4.3); providers have incentives to improve their services since they face the risk that clients will not renew their registration. The law calls for Seguro Popular to shift to capitation basis by 2010--a major structural change in the health financing model--which should give clients/patients more influence over their providers. 5 Seguro Popular began as a federal program in 2001. In 2004, it ceased to be a program because it became part of the new public insurance arrangements established by the Health Social Protection System. 6 The states are claiming that 90 percent of registrants are in the bottom income quintile (the extreme poor), which means that it costs nothing for them to enroll and the states do not have to demonstrate any collection of payment from the beneficiaries. This would be a remarkable focus of poverty targeting--much better than Oportunidades. A more objective source of information, the ENIGH, shows a much looser targeting--37 percent of Seguro Popular beneficiaries are in the bottom quintile; the program is still redistributive in favor of the poor. It seems that the federal government needs to devote some effort to monitoring the states' compliance with the rules for fee collection and cost sharing. 102 Decentralized Service Delivery for the Poor Figure 4.3. Channels of Influence in the Mexican Health Sector: SSA with Seguro Popular Federal government Nurses and doctors unions State government (2010) Clients/ citizens/ Hospitals/ voters clinics Political influence Taxes/resources Seguro Popular Sectoral policy and operational practices (wage negotiations) The Consejo Nacional de Protección Social en Salud (a board of directors with members from the federal cabinet and state secretariats of health) and the Comisión Nacional de Protección Social en Salud (a devolved agency of the federal Ministry of Health) manage the program at the national level, with the latter making coordination contracts with the state secretaries of health and supervising implementation of the program. The commission inspects and certifies annually whether the state facilities are qualified for participation in the program. The seriousness of this review is not clear yet, as the reports are not publicly available at the moment. In any case, it constitutes at least a potentially important channel of accountability via the long route to the government and more directly to the clients if the reports are published. Seguro Popular does not operate a health care system, with separate staff and separate facilities; rather it provides supplemental funding for SSA and recently IMSS Oportunidades facilities or for medications. Any additional staff hired with Seguro Popular funding have been hired on a temporary contract to respond rapidly to the increased demand and thus are not subject to the usual rules of personnel assignment, negotiated with the union.7 This opens a potentially important margin on which the state secretariats and the facility administrators could improve efficiency and quality. 7 The pay scales are set locally, probably in line with the unionnegotiated rates, but there are no data on this. Decentralized Service Delivery for the Poor 103 IMSSOportunidades With IMSSOportunidades the federal government has, since the 1980s and before decentralization of SSA to the states, provided health services in marginalized rural and urban communities, using facilities and personnel arranged by IMSS but not part of the mainstream IMSS system. Neither the state nor municipal governments play a significant role, in contrast to the other decentralized programs discussed in this report, but management is deconcentrated to regional directorates (each typically covering more than one state), and the program requires "community committees" to provide some (lowskill) labor inputs for provision of local services (see Figure 4.4). "Community participation for the selfcare [autocuidado] of health" is written in the objectives of IMSSOportunidades, and just by their proximity, these citizen groups have the potential to be a conduit of accountability, but there is no formal specification of such a role. Evidence from a few states--Baja California Sur, Colima, Hidalgo, Jalisco, Oaxaca, and Tabasco--suggest that these community connections do improve transparency (see Volume II). Figure 4.4. Channels of Influence in the Mexican Health Sector: IMSS Oportunidades Federal Government ­ IMSS National Health Workers Union ­ Regional SNTSS Directorates Clients/ Citizens/ IMSS Voters Community Hospitals and Volunteers Clinics Political Influence Taxes/Resources Sectoral Policy and Operational Practices (wage negotiations) As of February 2006, there has been no formal connection between SSA providers and financing and the IMSSOportunidades system, except that the State Health Committees had oversight of both. There is an official proposal to change the rules so that users of IMSSOportunidades would pass over to Seguro Popular. Since Seguro Popular is not actually a provider, presumably this proposal would mean passing IMSS Oportunidades funding to the states' SSA facilities, which would pick up the service 104 Decentralized Service Delivery for the Poor obligation and some of the money going to IMSSOportunidades. It is not clear whether all the IMSSOportunidades funding (the part from IMSS) would go over to SSA or what would happen to the IMSSOportunidades staff and facilities. An alternative way to merge Seguro Popular with IMSSOportunidades would be just to give Seguro Popular enrollees the option of taking their money and service needs to an IMSSOportunidades facility or to a state SSA facility. The procedures for linking Seguro Popular with IMSS Oportunidades need to be clarified. The Mexican health sector, which this report examines only in part, is sharply fragmented, with several distinct vertically integrated providerfunder organizations and with geographic fragmentation across states. This creates problems not only for coordinating management from the government's side but also, from the clients' viewpoint, for figuring out where to get service and whom to hold accountable. The same problem arises in other areas, especially agriculture and rural development, where programs have proliferated. The fragmentation increases administrative and transaction costs and duplicates facilities, without giving any benefit of choice to citizens, except to go to the private sector. The high turnover in the labor market and internal migration generate many potentially confusing outcomes. For example, a campesino might live in Oaxaca and obtain a living part of the year working on his own two or three hectares, but then migrate the rest of the year to Sinaloa, where he works as an agricultural day laborer on someone else's farm. He then gets health services part of the year (in Oaxaca) through IMSSOportunidades and part of the year (in Sinaloa) through IMSS. But another campesino from the contiguous state of Guerrero with the same migration pattern receives health services in Guerrero from the state SSA program (because IMSS Oportunidades does not operate there) and from the IMSS in Sinaloa. An urban trabajadora from Tabasco who migrates temporarily to a maquiladora plant in Chihuahua gets health services from IMSS while in the north, but from the SSA in Tabasco if and when she returns home. It is difficult under this setup for the campesinos of Oaxaca and Guerrero and for the trabajadora of Tabasco, all of whom are likely to be poor, to understand which government entity is responsible for their health services and more difficult still to hold anybody accountable for possible shortcomings. From a macro perspective, the fragmentation of the health sector results in many gaps and overlaps in the assignment of responsibility and financing and sometimes a lack of correspondence between the two. In the days of a single party, the PRI could mediate these differences and bring coordination. The complexity of the sector has multiplied over the last decade, while the ability of a party to mediate has diminished, although coordination tends to be better where the same party is aligned in various levels of government. The current funding arrangements for public health services provision in Mexico are unique in their broad reliance on funding based on inputs and capacity--with virtually no use of financing as a tool to motivate or reward performance. Integration between the funder and providers blocks the utilization of such "active purchasing". Decentralized Service Delivery for the Poor 105 The integration between funder and provider, extending as it does, not only to hospitals but also to primary care providers, implies that the barrier to improving incentives applies very broadly. Integration not only precludes paying one's own providers for performance, it also precludes the introduction of any competition (from providers in another public network, or from private providers). Since this integration creates such a barrier to the introduction of active purchasing--it is not surprising that so many analyses agree on the need for separation of funder from provider in the Mexican health system. Box 4.5 describes the rationale for and international experience with such arrangements. Box 4.5. Separating Funders from Providers in Health: The International Experience Analyses of the Mexican health system have concluded that the organizational arrangements for publicly funded services contribute substantially to their unsatisfactory performance (OECD, 2005c). Analysts consistently suggest that some sort of separation of provider from funder is needed to promote efficiency and responsiveness on the part of health care organizations, both in the SSA, as well as the various integrated social security organizations. Likewise this separation is seen as important to allow "crosspurchasing" ­ as a means to create choice of provider (choice for purchasers; choice for patients) and reduce duplication of capacity. The shift to active purchasing. While relatively common 20 years ago, funding of health service providers based on capacity or inputs has diminished throughout the world. An international review of wellperforming health systems reveals few instances of passive funding of providers based on capacity (e.g. historical budgeting). A number of recent publications have documented and sought to explain this trend (WHO, 2000; Figueras, Robison, et al., 2003; Preker, 2005). The separation of purchaser and provider has five main objectives. · Services may be improved by linking plans and priorities to resource allocation, for instance, shifting resources to more costeffective interventions and across care boundaries (such as from inpatient to outpatient care). Purchasing is thus an alternative way to effect some measures traditionally pursued via planning; · Population health needs and consumer expectations are to be met by building them into purchasing decisions; · Providers' performance is to be improved by giving purchasers levers, such as financial incentives or monitoring tools that can be used to increase provider responsiveness and efficiency; · Management can be decentralized and decisionmaking devolved by allowing providers to focus on efficiently producing the services determined by the purchaser; 106 Decentralized Service Delivery for the Poor · The separation of functions can introduce competition or contestability among providers and thereby use market mechanisms to increase efficiency. Another key aim of the separation of purchaser and provider functions within publicly operated health systems is to reduce the administrative rigidities generated by hierarchically structured commandandcontrol models. Often this is approached by substituting a new set of contractual relationships with hierarchical control mechanisms. Experience with introducing active purchasing. Countries that have introduced some form of purchasing within the public sector include Austria in the1980s (Fidler, Haslinger, et al., 2006), Sweden beginning in several county councils in 1990 (Harrison and Calltrop, 2000), Finland in 1993 (OECD, 2005a), New Zealand in 1993 (Scott, McKenzie, et al., 2003; Ashton, Cumming, et al., 2004), and the United Kingdom in 1991 (Ham 2003). Southern Europe also has several examples. The Basque country and Catalonia in Spain have adopted a system of purchasing. In Italy, purchasing relationships are utilized with teaching hospitals. In Portugal, special agencies with the responsibility to contract with health care providers were established in every regional health administration in 1998. While all have faced implementation difficulties (Ham 2003), by and large results are positive (Ham and Hawkings, 2003). In the 90s most eastern European countries, which previously had integrated funding and provision in the public sector, established social health insurance systems. This change separated the financing (insurance and funding) from providers (either autonomous hospitals or independent primary care providers). Countries which have done this include Armenia, the Czech Republic, Estonia, Georgia, Hungary, Latvia, Poland, Romania, the Russian Federation, Slovakia, and Slovenia. Although the introduction of purchasing reforms is most often discussed for use within integrated "National Health System style" systems, it also affects social health insurance systems that have maintained a separation between funding and provision since their inception. Despite this organizational separation, until recently, purchasing was a passive exercise that involved the reimbursement of expenses to providers with only some financial incentives and overall budget ceilings to ensure cost containment. Contracts did not focus on price or efficiency, nor were they understood to be contestable. For instance, in social health insurance countries such as Germany or the Netherlands, sickness funds traditionally had the legal obligation to enter into uniform and collective contract with each physician established in their working area. In recent years, countries such as Austria, Germany, Israel, and the Netherlands have sought to transform insurers from being relatively passive payers to become more discriminating and prudent purchasers. These countries have progressively introduced more selective forms of purchasing according to performance criteria. A triggering factor has been the introduction of insurance competition in many of these countries, allowing individual citizens to choose among statutory insurers and purchasers. This reform suggests that introducing market incentives for insurers would lead to better administration of collection, Decentralized Service Delivery for the Poor 107 more innovative practices, and more costeffective purchasing. In the Netherlands, from 1992, sickness funds and private health insurers were allowed to negotiate lower fees with providers than officially approved fees; from 1994 the sickness funds had the option to contract selectively with physicians and pharmacists. This implies a radical change from previous practices in which each sickness fund had the legal obligation to contract with all qualified providers in the working area that wanted a contract. Special case of primary care. In most middle and upper income countries, primary care is organized and delivered by independent providers who are contracted, and paid by a public or social funding body. In most if not all cases, the payment consists mainly of an annual fee based on patients selecting and registering with the provider. The payment usually combines a capitation component (covering a basic package of services) plus feeforservice payments for some services that tend to be underprovided when grouped in the package. A small number of countries use primarily feeforservice reimbursement (Canada; New Zealand; United States). It has always been the case that passive (salaried) payment for primary care has been relatively little used in OECD cases (e.g. Scandinavia; Portugal; Greece). In the past few years, even these few countries have implement reforms toward more active purchasing. For example, Finland has implemented a payment system that includes 40 percent of payment based on patient choice (registration) and utilization (fee for service). Likewise, Sweden introduced contracting with primary care providers in a number of counties--with the majority of physician payment coming from capitation and feeforservice payments. Evaluations have found that by and large the results are positive. For example, in Sweden, Sorensen and Grytten found that the shift to purchasing generated an increase in productivity of 23 percent in the shortrun, and 40 percent in the long run (Sorensen and Grytten, 2003). As noted above, eastern European countries have implemented separation of funder from provider throughout their system, including primary care. An evaluation in Estonia found the initial results are favorable (Koppel et al., 2003). IV. SOCIAL INFRASTRUCTURE The most important source of funding for local provision of social infrastructure is FAIS in Ramo 33, as noted in Chapter 2. FAIS is a good candidate for efficient decentralization in that, by and large, the externalities associated with the projects it funds are fully internalized at the municipal level, and is a good example of a shorter "longroute." Moreover, if information is available to all--readily available with respect to investment in water and roads, for example--direct participation by beneficiaries should improve the selection and execution of projects. But there is no guarantee that this will be the case or that decentralization will be propoor, although the law mandates it and despite the targeting formulas. Several elements may play a critical role for this situation, such as political issues associated with statemunicipal and withinmunicipal 108 Decentralized Service Delivery for the Poor relations, underrepresentation of the poorer communities in the decisionmaking processes and lack of capacity, and information at local level. Four characteristics make FAIS substantially different from FAEB or FASSA. First, FAIS was not preceded by any decentralizationtype acuerdo with states or municipalities. PRONASOL was created by a presidential decision in 1988; later most of its funds were distributed with an explicit formula to subnational governments by another presidential decision in 1996. It was then converted into FAIS by Congress in 1998 with the reforms to the Fiscal Coordination Law, in a process with no intervention by any national trade union and hardly any participation by state governors. Second, FAIS resources are not tied to a wage bill, as is the case with practically all of FAEB and most of FASSA. These two characteristics create a different political process, with substantially fewer constraints and thus deeper decentralization. Third, the Fiscal Coordination Law required FAIS resources to be targeted explicitly to poverty alleviation. And, fourth, it required most of the funds to be transferred to municipalities, not states.8 Moreover, FAIS funds are not intended for services aimed at the general population (like basic education or health), but for physical investments in a broad range of areas intended to help the poor, thus giving states and municipalities considerable flexibility in their spending decisions.9 The extent to which local governments can and do utilize these resources for their locally determined priorities varies between municipalities and will depend also on the availability of other funds for infrastructure projects. Federal agencies reassert their central authority in some cases where the local governments feel compelled to, or decide to use their FAIS resources to pay the local counterpart share of federal projects, such as in water projects directed from CONAGUA. In other cases, however, the municipalities use FAIS resources to meet their own identified priorities, for which federal or state funding might not be available. Theory and international experience with decentralization suggests that having a mix of these two outcomes is efficient. There are some activities with both national (or at least regional) externalities and clear local benefits, and cofinancing arrangements for these can effectively combine the two types of interest. In addition, municipalities with low taxes bases (approximated by the marginality index in the FAIS formula) need to have some resources for their choice of projects with purely local benefits. 8 The Fiscal Coordination Law mandates that FAIS be budgeted at 2.5 percent of the recuadación federal participable approved by Congress in the Ley de Ingresos, regardless of the latter's fluctuations during the year or its endofyear value. Of the total FAIS budget, 2.2 percentage points are distributed to municipalities, and 0.3 percent is distributed to the states. 9 Funds transferred to municipalities can be used for drinking water, sewerage, urban infrastructure (sidewalks, market plazas), electrification in rural areas and poor neighborhoods, basic health and education infrastructure, housing improvement (cement floors, painting walls), rural roads, and productive rural infrastructure (irrigation). Funds transferred to states are for works and activities with regional or intermunicipal impact (Fiscal Coordination Law, Article 33). Decentralized Service Delivery for the Poor 109 FISM can also support capacity building for municipalities, but preliminary data suggests local governments are not taking advantage of this option. Decentralization in the absence of adequate local capacity can bring little change in results, although this does not mean that it cannot be done at the same time. Experience shows that local capacities expand best as decentralized systems mature, even though sequencing remains difficult. As much as 2 percent of FISM funds can be allocated by municipalities to support institutional capacity building programs. Only for this expenditure are municipalities required to agree on an institutional development plan with the state and federal authorities before spending the money. This requirement probably could explain why municipalities are not allocating the funds for this purpose, particularly given that they can allocate the funds to the remaining categories if not used for capacity building. There is no data on expenditures done in this category, and hence it is difficult to know which are the main issues addressed through capacity building. Available information for the state of Veracruz suggests that the share of this category is very low: 0.64 percent of FISM allocated to institutional development. This figure decreases to 0.25 percent if only the 30 poorest municipalities are considered. Upper levels of government are in a better position to foster capacity at local levels by providing incentives for municipalities to match demanddriven capacity growth with the provision of supplyside assistance and financing. States play usually this role building partnerships with municipal governments or cofinancing institutional development programs. Although not necessarily FISMspecific, this can have an impact in FISM projects. Overall, 74 percent of municipalities report having received training from state governments, compared to 34 percent from the federal government. Compared with the decentralization of education and health, the decentralization of social infrastructure with FAIS is more targeted to the poor, at least to the municipalities with a high proportion of poor persons, particularly in poorer states. This potential benefit to the poor is enhanced by a distinctive feature of FAIS, namely that the law mandates municipal and state governments to (a) inform the inhabitants of the respective jurisdictions regarding the scope, costs, and potential beneficiaries of each individual project financed by FAIS, (b) promote community participation in the ex ante choice of projects, in their execution, and in their ex post evaluation, and (c) inform the population at the end of the fiscal year of the results obtained. The Fiscal Coordination Law, however, does not specify how these three objectives should be reached, how their achievement (or not) will be monitored, or the type of sanctions to be applied in case there is no compliance with these provisions. The extent to which FAIS funds contribute to poverty alleviation depends on two further considerations. The first derives from the interaction between state and municipal governments. The federal budget legally cannot disburse funds directly to the municipalities, so it must transfer the FISM resources first from the central government to state budgets and subsequently from state to municipal budgets under 110 Decentralized Service Delivery for the Poor the conditions and terms mandated by state legislatures as well as by the federal rules.10 Thus in addition to federal rules for the use of FAIS resources, state laws and operational rules are critical, generating a potentially large variance across states in poverty alleviation outcomes. Some states simply transfer FISM funds to municipalities and agree with them on certain planning and implementing requirements, while some impose substantial constraints on their municipalities before funds are disbursed, as elaborated below. Some states also intervene with positive fiscal incentives (typically using FISE resources) to reward municipalities whose FISM investments meet certain criteria or that meet targets in coverage of social infrastructure. Box 4.6 compares the experiences in Puebla and Veracruz. Box 4.6. Comparing Institutional Arrangements: The Case of Puebla and Veracruz Veracruz and Puebla differ in their degree of intervention in the municipal FISM programs. Both states have issued guidelines to regulate planning, implementation, reporting and supervision of FISMfunded projects. An important difference is that Puebla, intervenes more, by conditioning FISM disbursements to a clearance procedure from state sector normative agency. For example, a municipality in Veracruz has to share the technical project files for education, health, water and roads projects to the state agencies, for their revision, but this does not affect the monthly disbursements they receive from FISM funds. In contrast, Puebla requests the clearance of project files ex ante to the disbursement of the funds. Puebla, like other states, imposes rules for the process of planning and allocation of projects and sets some criteria for selection and sequencing of investments, but only in Puebla these rules are really enforceable since funds are conditioned to the state clearances. (Some of these make good sense, like requiring that water coverage reach a certain percent before funds could go to general urban improvement and insisting that water utility work be completed in a neighborhood before it could have a road (re)paving project). During 1998 and 1999 Puebla also had an incentive program for municipalities, providing counterpart funding for priority infrastructure investments funded by FISM. Did the disbursement condition in Puebla have an effect? From the data gathered by a survey of municipal presidents in some years, there is no evidence of better allocation efficiency of municipalities in Puebla for FISM investments. For 1999, when Puebla's fiscal incentive program was in effect, there is a strong correlation of allocation with needs and a clear difference from Veracruz, which had no such program. In later years, when the incentive program ended, there is 10 When FAIS began, there was an important controversy between the federal government and some states as to how far a state's legislation could either alter the Fiscal Coordination Law distribution formulas to municipalities or impose some conditionality. The controversy did not reach the courts and was settled in a compromise: Fiscal Coordination Law formulas would be observed, and the states would be allowed to impose conditions on their municipalities. Decentralized Service Delivery for the Poor 111 no significant difference in the degree of correlation between needs and allocation. Further investigation in other states should attempt to verify or refine this pattern before drawing strong policy conclusions. A review of the institutional arrangements for FAIS in ten states found that state government set general operational guidelines for FISM, and probably nearly all states follow the practice in some way (details are in Volume II.) Eight out of ten have introduced operational rules or manuals and, in some cases, imposed restrictions or requirements to clear projects with the state's sector agency. All of these guidelines require inclusion of the voice of the clients and communities in planning and selecting FISM projects (see Figure 4.5). The main channel is through the Consejos de Planeación Municipal (COPLADEMUN), which are municipal councils that usually involve community representatives from each locality; representatives from the municipal government; and in some cases from the state sector agencies. Details vary in each state, and it is not clear if these mechanisms allow for efficient representation of the population or to what extent they represent the poor. In general, the list of works to be financed by FISM is agreed through the council and disclosed by the municipal government before the start of the calendar year. The existence of operational guidelines facilitates the use of these resources. These state guidelines are desirable if they seek better implementation of the law. However, some of these operational guidelines contradict or go beyond the federal law, for example by allowing spending on scholarships, or setting conditions on the transfers to the municipalities, like in the case of Puebla. In some cases, to improve the sequencing of investments, state regulation has introduced caps to certain expenditures in order to prioritize particular infrastructure works or localities. For example, in Guerrero, municipalities must spend at least 10 percent of the total in rural areas. In Puebla, municipalities cannot spend resources in urban areas until a certain target in sanitation coverage (including rural areas) has been met and cannot pave roads in a community until the water and sanitation work under the roads has been done. In Aguascalientes, FISM investments should be allocated between the cabecera and the localities according to the population distribution and in Zacatecas, the percentage of community financing varies for works of a more "public" nature, such as parks and plazas. These local conditions, in turn, may be fully justified in economic terms. Thus state laws, as in Puebla, may require municipalities to have a community participation process and some minimal costbenefit analysis in order to improve the quality of their projects prior to the disbursement of funds. States may also try to solve a coordination problem between contiguous municipalities, say, with regard to a health clinic (so that perhaps only one is built with joint funds, instead of two separate and less equipped facilities), although there is little evidence of FISM funding for intermunicipal projects. Alternatively, states may want to reach some economies of scale in, say, a rural road (so that, with joint funds, a longer one is constructed rather than two unconnected 112 Decentralized Service Delivery for the Poor segments). Or, finally, they may want to induce a single municipality into developing a larger project, say, sewage facilities, providing incentives to do so with pari passu state funds. While some of these interventions can improve quality of FISM expenditures, they may have a cost in moving away from the decentralization model supported by the 1998 reform and could create opportunity for states to exercise political control by obstructing projects in municipalities from different political parties or groups. Also, this new layer of regulation has further muddied the rules of the game regarding the allowed expenditure categories, the general objectives of infrastructure financing (not all infrastructure investments have a poverty alleviation focus), and the reporting mechanisms, to the extent that some state regulation allows expenditure for categories that are not permitted by the federal law. The second consideration for the povertyreduction effect of FAIS is internal to the municipalities, since the Fiscal Coordination Law does not specify the distribution of funds inside each municipality. While there are approximately 2,500 municipalities in Mexico, there are almost 200,000 localities inside these municipalities, and the majority of them are poor and small (with between 500 and 2,500 inhabitants). Funds may effectively reach a poor municipality, but the incentives, preferences, and relationships among the municipal president, the council, and the other powersthatbe will determine how they are distributed between the county capital and remaining localities, as well as the extent to which projects respond to the needs of the poor, since not all those who live in a poor municipality are necessarily poor. Once the funds reach the municipality there is little or no guidance for the municipal authorities to identify those projects that will better improve the living conditions of the poor; this situation is more acute in poorer municipalities, which in general lack technical capacities. Adequate information can support the decision making process when selecting the works to be financed and can help to counteract pressure from political groups; the federal government is actually making efforts to provide such information. Also critical is the capacity of the municipal authorities to execute and supervise projects, in particular in poorer municipalities. Decentralized Service Delivery for the Poor 113 Figure 4.5. Channels of Influence in Mexico's Social Infrastructure: FAISFISM Federal government Political State party government Clients/ Municipality = citizens/ provider voters Political influence Taxes/resources Sectoral policy and operational practices Federal law provides only general requirements for planning and implementing works financed via FAIS: municipal governments should promote participation of beneficiary communities for programming, implementing, monitoring, and evaluation, for which states and municipalities use their existing participatory planning mechanisms (if any). Thus, the degree to which this community participation is achieved depends more on the adequacy of state and municipal systems than on provision at the federal level. Federal regulation also requires states and municipalities to disclose publicly the cost of infrastructure works; the amount allocated from FAIS to each one, and the sites, main beneficiaries, and general characteristics of the works financed. Additionally, by the end of each year, they must inform the population of the results achieved; although no methodology for ex post evaluation is defined. These requirements have not been systematically enforced. However, in 2004, SEDESOL issued general criteria for reporting the use of FAIS resources, together with an information system to collect this information. As most states have established information systems for FISM budgeting and control, the statelevel information on FISM spending is better than what they report to the federal government through SEDESOL. V. RURAL DEVELOPMENT The Rural Development Program is the only part of Alianza targeted to small and poor farmers, and the only one decentralized to municipalities. Since 2002, an increasing proportion of the Rural Development Program funds have been decentralized to 114 Decentralized Service Delivery for the Poor municipalities. Two other components of Alianza--crop agriculture and livestock--have been mostly decentralized to the state level. As a program to promote productive investments and competitiveness, Alianza's Rural Development Program has different objectives and institutional needs than the social programs. Social programs aim for 100 percent coverage--something for everyone--which is well suited to local politics. However, somethingforeveryone incentives with a productionoriented program like Alianza often lead to dilution, fragmentation, and underfunding of the profitable investments that would do the most to improve competitiveness. SAGARPA promoted municipal decentralization in order to involve municipalities and strengthen the Consejos Municipales de Desarrollo Rural Sustentable (CMDRS)--the municipal rural development councils. Not all states entered municipal decentralization in 2002, but by 2005 all state governments had decentralized at least some of the Alianza funds for rural development to their municipalities. By signing convenios, the state and municipal authorities (and possibly also the SAGARPA delegate in the state) formalize municipal decentralization and state their respective obligations. Municipal governments are supposed to contribute counterpart funds, but this is not compulsory. When they do, they often do it with FISM resources. The national government provides around 80 percent of the government money and writes the rules of the program. The rules include the formula for distributing half of the funding among municipalities, with the other half distributed according to rules set independently in each state. Each state in consultation with SAGARPA decides how much of its total Alianza funding to devote to the Rural Development Program.11 There are two modes of municipal decentralization. In both modes, the resources allocated to a municipality by the decentralization formula set the budget constraint for the approval of funding proposals presented by producers in the municipality. With the first mode, (a) funds are transferred to a separate municipal account, and (b) the entire cycle of reception, screening, and selection of requests, as well as disbursement of funds, is carried out at the municipal level. In the second mode, no separate account is opened. Proposals are received, preliminarily screened, and prioritized in the municipality, but the final selection is performed at the state level by the technical committee of Alianza's fideicomiso, which also disburses the funds. The first mode is restricted to municipalities satisfying certain conditions: (a) the existence and functioning of a CMDRS, (b) the existence of a municipal rural development plan, validated by the CMDRS, (c) the existence of a rural development department within the municipal administration, and (d) the existence of a programmed budget for the current 11 The amount going to each of the decentralized programs--crop agriculture, livestock, and rural development--is decided by the state in discussion with SAGARPA and stated in the technical annex signed each year by SAGARPA and the state government. Decentralized Service Delivery for the Poor 115 year. The effectiveness of this mode depends heavily on the strength of CMDRS and the municipal administration. CMDRS are part of the complex architecture for economic governance and the execution of public policy in rural areas set up by the Law of Sustainable Rural Development (LDRS). This architecture has a sectoral or productbased axis, consisting of product chains and their governance organizations (the Comités de Sistema Producto), and a territorialbased axis formed by a threelevel system of rural development councils: the municipal, district, and state rural development councils (see Figure 4.6). At the top of the system are the intersecretarial committee chaired by SAGARPA (the InterSecretarial Commission for Sustainable Rural Development, CIDRS) and the consultative nationallevel rural development council, the Consejo Mexicano para el Desarrollo Rural Sustentable, which includes national producer organizations, academic and research entities, productchain representatives, and the private sector, alongside the federal ministries included in CIDRS. CMDRS consist of (a) the municipal president (usually the chair) and other representatives from the municipality, usually the person responsible for the municipal rural development office, (b) representatives of state government offices and of the federal offices related to rural development operating in the municipality, and (c) representatives of social and private associations active in agriculture and rural development in the municipality. They are generally assisted by a technical coordinator (normally an agronomist or veterinarian) appointed by SAGARPA, by the state government, or by both, and are generally paid with funds from Alianza's PROFEMOR program. From the operational rules of Alianza and CMDRS statutes, they are supposed to (a) be a locus for rural development planning in the municipality, (b) facilitate the operation of state or federal agricultural rural development programs in the municipality, (c) run demanddriven programs such as the Rural Development Program, receiving, prioritizing, and approving funding proposals from producers and disbursing the funds, (d) be a locus for the local coordination of different rural programs and the local concurrence of funds, and (e) represent local rural interests. Doing these functions well would make the CMDRS a formidable instrument of decentralized rural economic governance. 116 Decentralized Service Delivery for the Poor Figure 4.6. Channels of Influence in Mexico's Rural Development: Alianza's Rural Development Program Federal government -- SAGARPA Farmers State associations government Municipalities/ CMDRS Clients/ citizens/ Private service voters providers Political influence Taxes/resources Technical assistance Sectoral policy and operational practices Unfortunately, most CMDRS cannot discharge their functions in an effective manner, although local situations vary considerably. First, they do not have the technical capacity to plan effectively a medium or longterm rural development agenda for the municipality. Mostly, they act as passive recipients of requests from local producers and as disseminators to their communities of government norms and instructions. Second, most councils are dominated by the municipal president, technical coordinators, or members representing SAGARPA or the state's rural administration. Thus they act more as branches of government (municipal, state, or federal) than as representatives of rural society. Also, in many councils, the majority of civilsociety participants come from ejidos or small communities, and they rotate frequently, usually every year. This makes it very difficult for them to understand the purpose and workings of the councils, assimilate the basic rules of the programs, and participate effectively. Finally, the LDRS is perceived by other government agencies as pertaining to SAGARPA and state rural administrations, in the same way as other laws are seen as pertaining to other ministries--for example, the Social Development Law is seen as pertaining to the Secretaría de Desarrollo Social (SEDESOL). This is despite the multisectoral language of the laws and the existence of various intersecretarial councils. As a result, the CMDRS rarely can count on the participation of agencies other than SAGARPA and the states' rural development agencies, and hence cannot coordinate Decentralized Service Delivery for the Poor 117 programs or support integrated rural development. In practice, they usually implement one program only, Alianza's Rural Development Program. Since the municipal president, or his representative, is typically the strongest force in CMDRS, implementing the Rural Development Program depends on the capacity of the municipal authorities, who are in a weak position to carry out rural development activities. Service provision and the development of urban centers are the focus of municipal activity, not rural development, which is not included among the many municipal functions listed in Article 115 of the Mexican Constitution. Also, the Mexican system of threeyear municipal administrations with no reelection is inimical to the longterm perspective required in rural development. The numerous additional activities required of municipalities by the new normative instruments12 have been established with little regard for the administrative restrictions, procedures, and routines of municipal governments.13 In order to access the money from the new programs, municipal governments have adapted the new policy to their own routines instead of modifying their behavior to fit the new policy. As a consequence of these weaknesses, CMDRS find it difficult to prioritize investment demands and to countervail the bias of programs in favor of those better informed. They also show a spontaneous tendency to choose atomized rather than clustered investments, which limits the ability of the Rural Development Program to improve productivity--the core objective of Alianza. In spite of their shortcomings, CMDRS must be considered a step forward in establishing privatepublic coordination platforms and in encouraging the municipal governments' involvement in rural development issues, including the design of a municipal rural agenda. They also serve as forums for discussion of municipal rural issues among interested parties and for the dissemination of information regarding the government's provision of rural support services. The most important trade unions affecting rural development policy are not those of providers but those of clients (farmers or campesinos). Farmers unions apply pressure on both national and subnational governments, lobbying for privileged access to public resources. They have traditionally enjoyed large political power, as they used 12 These instruments are the Ley de Desarrollo Rural Sustentable, the Programa Especial Concurrente, the operational rules of Alianza, the convenios signed with the states, the statutes of the district councils, and the statutes of the municipal councils. 13 Merino and Macedo (2005 p. 25) identify 10 new activities to be carried out by municipalities in connection with the new normative instruments. To carry out these activities municipalities must implement five organizational tasks: "(1) establish a collective body of citizens and federal, state, and municipal authorities to partake in the rural development strategy; (2) have a plan and a budget with resources exclusively devoted to the rural development strategy; (3) appoint a formal entity to take charge of the administrative demands deriving from the implementation of the rural strategy; (4) identify producers' organizations and promote their participation in the rural development strategy; and (5) establish specific procedures to handle the requests from producers with an intensive use of information and registry systems." 118 Decentralized Service Delivery for the Poor to be critical to ensuring political support for PRI from rural dwellers and helped to stabilize political conditions in rural areas, keeping conflicts under control. Today, rural unions are connected not only to PRI but also to other political parties. Producers' unions make their influence be felt on government directly through negotiations and pressure instruments and indirectly through their capacity to get representatives of their corporate interests elected to Congress. Although the rural programs tend increasingly to make beneficiaries compete for resources with demanddriven systems directly administered by government or by joint committees, some rural programs, like many of the centrally managed programs of Alianza, are operated by SAGARPA through farmers unions. Even in the Rural Development Program, a limited amount of the funds are not decentralized, but instead are operated directly by SAGARPA. These funds are set aside for farmers unions, which implement the programs.14 Service providers are hired under shortterm contracts by the federal or subnational governments or by producers themselves (with funds provided by the programs). Their pay usually comes from program resources. In some cases, they may be employed on a more or less regular basis by farmer organizations, possibly unions, or associative enterprises. They provide technical assistance to producers for the preparation and implementation of investment proposals, and they are the main interface between government officials and beneficiaries of the program. Service providers are also the main local source of information about program opportunities and norms. Individually, they sometimes take advantage of their position as intermediaries between government and producers and of their superior information, but they are weak and disorganized as a group. Unlike teachers or health workers, they are not unionized and have no entitlement to secure employment or social security benefits. The position of service providers is thus extremely weak, and they have few economic or moral incentives to do a good job. Nor do they receive technical support, systematic and effective training, or support to operate as networks. These "streetlevel" operators are probably the weakest link in the implementation chain. Rationing mechanisms--global and individual--are important in demand driven programs, like the Rural Development Program, where demand usually exceeds the supply of funds. Global rationing operates on a geographic basis, with ceilings for states and municipalities, and on the basis of beneficiary classes. For the part of the program directly implemented by the states, Alianza's operational rules require at least 70 percent of program funds to go to marginal regions of high or very high marginality indexes, 20 percent to priority groups (youth, women, indigenous groups, elderly producers, and handicapped rural people), 35 percent to investments on priority production chains with high social inclusion, another 35 percent to projects proposed by CMDRS, and 25 percent to lowincome producers in nonmarginal regions.15 The rules 14 Part of the money of decentralized Alianza programs is set aside for national execution and is mostly given to farmers union for execution. 15 Percentages add up to more than 100 because of overlapping among the categories. Decentralized Service Delivery for the Poor 119 are different in the case of municipal implementation. In this case, the funds available to the state are allocated among its municipalities in accordance with an objective formula. According to this formula, fifty percent of the funds in all states must be allocated according to municipal marginality and rural population criterion, while the other fifty percent are allocated according to criterion set by the state government. Individual rationing takes place within geographic or producer class ceilings and is ideally done according to the quality of the proposals, based on established criteria and evaluation procedures. In practice, however, priority usually goes to the first proposals duly completed and with all the attachments required when the window opens. Hence, a substantial advantage goes to applicants with good information about the timing of window opening and access to help for preparing all the required documents. Delay in the disbursement of funds is a serious problem for the Rural Development Program, partly because the federal government does not publish the Alianza rules until March, April, or even May for a number of reasons, and partly because the processes involved in implementation are so lengthy. Even if the rules undergo only minor adjustments, operators cannot start their work before the new rules are known, especially since the forms to be used are generally part of the rules. Then the states have lengthy processes to receive and evaluate proposals. While the Rural Development Program is small relative to total government spending for rural and agricultural development, it yields valuable lessons for decentralized programs in any sector that focuses mainly on productive investment. The use of publicprivate coordination platforms at the municipal, district and state levels generally has created some of the conditions necessary for a better discussion of investment planning and local rural agendas (see Box 4.7 for differences in state performance). There is the problem, however, that the agriculture and rural development investments supported by the Rural Development Program are not organized around clear economic strategies and are too fragmented to have significant economic impact. For program investments to reach a critical mass and have a strong impact, strategic investment programs to strengthen rural productive activities should be designed at the regional level. These programs should be based on the potential offered by territorial assets and on the identification of the main development axes around which the investment support provided by the program should cluster. The municipal space is generally too small in Mexico, and municipal administrations are too weak and focused on other activities for municipal councils to be effective economic governance entities capable of designing and implementing the required strategic investment programs. State governments normally divide their territories in regions, which are more appropriate spaces to design strategic investment programs. The district councils created by the Law of Sustainable Rural Development mostly coincide with these regions. These councils could take the lead in the participatory design and implementation of programs for rural investment clustering, and generally act as economic governance entities for the rural areas of the regions. However, in order to be able to assume this role, the councils would need to be reinforced with a sufficient cadre 120 Decentralized Service Delivery for the Poor of professional staff, and they would also need the authority to decide on the allocation of the investment resources of the program. Box 4.7. State Experiences in the Implementation of Alianza Despite the centrally issued operational norms, there are differences in the way Alianza decentralized programs operate in the states. These differences depend on the priorities of the states and the negotiation of those priorities with SAGARPA. The allocation of funds among the three decentralized programs (crop agriculture, livestock, and rural development) varies, reflecting the relative importance of commercial agriculture, livestock activities, and the small farm sector in the states. Thus, for instance, Guerrero allocates 73 percent of decentralized program resources to the rural development program, Oaxaca 66 percent, and Veracruz 57 percent, whereas Baja California and Nuevo León allocate only 16 percent, Sonora 21 percent, and Sinaloa 24 percent. The degree of poverty targeting also differs because of differences in share of the rural development program, the only targeted one in Alianza, and because of targeting differences within the rural development program itself. For instance, in 2004 only 0.6 percent of funds from the rural development program in Jalisco went to municipalities of very high marginality and 42 percent to municipalities of low and very low marginality, whereas in Guerrero 64 percent of the funds went to highly marginal municipalities and none to municipalities of low and very low marginality. Differences in personal targeting are also important. Although we do not have comparable figures, the Bank's field work and the evaluation documents of Alianza for the states show that leakage of benefits of the rural development program to relatively welloff producers are much higher in Jalisco than in Guerrero or Veracruz. Another difference is in the decentralization to municipalities of the rural development program and in the operation of the sustainable rural development councils. Guerrero has more than average municipal decentralization because it started in 2002 and all municipalities operate under the first of the two systems explained before. Veracruz is an intermediate case since municipalización started in 2004 and only some municipalities operate under system one. Finally, municipalización only started in 2005 in Jalisco, with all municipalities operating under system 2. Municipal councils exist wherever there have been decentralization of the rural development program to municipalities, but district councils may or may not exist and when they exist their role may be more or less important. Thus, for instance, in Guerrero and Veracruz district councils do not seem to play any role and their existence is little noticeable. In Jalisco, instead, they are more active, meeting frequently, often once per month. Source: Volume II. Looking at the whole array of agricultural and rural development programs, one can see that there is inefficiency coming from the segmentation and overlapping of federal programs, and also from these programs not being tailored to the varying needs Decentralized Service Delivery for the Poor 121 of different states and regions (World Bank 2005, chapters 5 and 6). Under the present system, state governments cannot properly design the rural agenda for their states and have no opportunity or incentive to internalize the global budget constraint because (i) the great majority of funding comes from federal Secretarías attached to specific federal programs; (ii) most programs are still not decentralized to any degree; and (iii) program objectives, eligibility criteria and operation norms are fully decided at the central level. Also, as mentioned above, the rural municipalities lack the scale and capacity to utilize efficiently the wide array of theoretically available programs. One way to improve the efficiency of agricultural and inefficiencies in rural development spending, particularly for productive programs, would be to concentrate the decisionmaking at the state level. If they received a formulabased block grant for rural development, then they would have the opportunity to plan and implement program packages appropriate to their locales. They would also internalize the political and economic costs and benefits of rural development actions and be accountable for the successes, and failures. Several European countries, like Spain, Italy, and Germany, follow this type of decentralization model. The equivalent to the Mexican states--the autonomías, regioni, or laender--are the pivotal entities of rural development authority and policy making. An important difference, of course, is that for Mexico there is not a European Union with a strong agricultural and rural development policy to take up part of the policy and funding authority. To achieve this type of devolution in Mexico would require three main things: 1. Transferring to the states most of the funds of rural development programs of federal Secretarías that do not have specific multistate externalities. This should be done in the form of block grants that combine the funds from the different programs now in place. The national government could still reserve for itself some funds to carry out a few selected rural development programs considered to be strategic and require nationallevel implementation. There should be an understandable formula to allocate funding to the states, and full transparency about how much each state gets. For purposes of poverty reduction, at least some of the funding should go according to rural marginality, although this corresponds more specifically to Oportunidades, not the productive programs. 2. Dismantling federal rural development programs and letting state governments replace them with their own rural development programs. The municipal rural development program would also be absorbed by the states, and each state would have the option to devolve some things to the district or municipal level, as appropriate to local circumstances. 122 Decentralized Service Delivery for the Poor 3. Transferring to state governments most of the presently federal assets, staff, and other operational resources that the states would need to design and implement their own rural development agendas. Clearly, a transition period would be necessary to carry out such changes, and this could involve going first in some states as pilots. The whole transition might be done within a sexenio. VI. CONCLUSIONS In education and health, the challenge for most of the programs (especially as weighted by fiscal importance) is to develop effective mechanisms to assure accountability for performance both by the short route from providers to clients and by the long route to governments. The most promising developments within the core basic education program are that some states have decentralized management to the region or even school level in a few cases. More progress has been made in programs outside the mainstream, particularly in PEC and, to some extent, IMSS. The PEC approach comes from statelevel experiences and in some ways is designed to promote this throughout the country; that is, the intention is not to have PEC in every school, but to provide experiences to the states (and parents/schools) in using this model. In health, Seguro Popular is a step towards having funding follow the client, rather than the historical distribution to providers, but it has not yet achieved equal funding per client across geographic areas or health care systems. To date it has improved inequities in federal allocation per beneficiaries across states. Further equalization will follow once fully implemented by 2010. IMSSOportunidades has mechanisms that could improve accountability to local communities and clients, but they have not been systematically utilized. Perhaps if Seguro Popular becomes more of a mobile financing vehicle, as intended in the law (by 2010), it could link with IMSSOportunidades to achieve more accountability to clients. For the municipal programs, FAIS, and Alianza's Rural Development Program, there are mechanisms for local participation at least down to the municipal level and perhaps to the community level with FISM. One challenge is to assure that these governments have the capacity, protection against corruption, and continuity to carry out good programs and that the local populace gets sufficient information. Even the most capable municipalities face some unavoidable structural constraints, however. First, in marginal areas, municipalities tend to be too small to undertake some types of investment efficiently. Second, confusion arises from the large number of programs, mostly federal or joint federalstate, in the same sectors, such as water and agricultural support, where the municipalities also have responsibility for service delivery. Third, the somethingforeveryone attitude in local politics undermines the need to choose competitive investments clustered around development axes. Effective devolution to state governments of rural development programs, in particular productionoriented ones, could be the best option to increase program efficiency. It would help overcome Decentralized Service Delivery for the Poor 123 the segmentation and overlapping of programs, would facilitate adjusting programs to varying state needs, would assist focusing on implementation issues, and would make it easier to adopt a territorial approach to rural development. Effective devolution would require transferring to state governments as block grants the resources currently allocated to federal rural development programs, along with the authority to design and implement their own programs. In this way, state governments would have a central role in driving the rural agenda in their states and would be more likely to internalize the benefits and costs associated with it. In terms of promoting efficiency, compared to the current system which does little, the states could be technically more efficient that the municipal level. The incentives for efficiency would need to come through the political system, where the state officials would be closer to clients than federal officials. The federal government could promote statelevel accountability and efficiency by publicizing the amounts of funds going to the states for rural development and by requiring participatory planning and expost evaluation processes. Responsibility and authority would go together. Looking more broadly at the institutional dimension of decentralization in Mexico over the past 15 years, one can see that it has been integrally linked with the increased competitiveness of the democratic process. As this process continues and deepens, it seems unlikely that the decentralization would be reversed in the near future, although there is scope for important adjustments at the margins. While the appropriate adjustments vary across sectors, a common theme is the need to increase both the transparency of resource allocation and the clarity of the intergovernmental division of responsibilities. A crucial and as yet underdeveloped role for the federal government in most areas is to establish service performance standards and to provide incentives for states and municipalities to inform their citizens about the extent to which they have achieved. International experience indicates that a national government can perform such standardsetting and monitoring roles more efficiently and with less conflict of interest if the national government is not itself doing the detailed design and micromanaging the programs. 124 Decentralized Service Delivery for the Poor ANNEX BACKGROUND NOTES ON EXPENDITURE DECENTRALIZATION AND POVERTY ALLEVIATION IN MEXICO I. INTRODUCTION These notes provide background information for a larger study of the impact of expenditure decentralization on poverty alleviation in Mexico. They identify some of the main problems and issues that need to be addressed to ensure that the expected positive impact on the poor is fully realized. The approach is eclectic, mixing historical and analytical observations. The notes need to be complemented with detailed studies that incorporate uptodate information about programs and budgets in the different areas of interest. Further, more empirical evidence of the effects observed to date in each sector is needed before offering policy recommendations. II. BACKGROUND From December 1978, when the Fiscal Coordination Law was promulgated by president Lopez Portillo, up to December 1997, when major reforms to this Law were promulgated by president Zedillo, Mexico experienced almost twenty years of a fairly stable institutional understanding for revenue collection and expenditure sharing between the federal and state and municipal governments. The cornerstone of this understanding was the National System for Fiscal Coordination (Sistema Nacional de Coordinación Fiscal) created by the Fiscal Coordination Law. Under this arrangement, states and municipalities would forego the collection of an agreed set of federal taxes, levies, and contributions in their own jurisdictions, in exchange for a share of total revenues obtained from these sources (Recaudación Federal Participable). Although the Fiscal Coordination Law established that joining the National System was voluntary, in practice all states signed the corresponding agreements with the Ministry of Finance (Secretaría de Hacienda y Crédito Público, SHCP), establishing a single national unified system.1 1 Since the agreements were signed, no state has left the System, although occasionally there have been threats to do so. These threats are not very credible, however, because it would be extremely difficult for states to setup their own tax collection mechanisms for what are currently federal taxes, aside from the political costs that it could imply for a state government. There are also unanswered questions if a state leaves the System: Would it have to pay for a share of the costs of the public debt? Would it profit from Decentralized Service Delivery for the Poor 125 Revenues shared through these agreements, labeled "participaciones," were the main source of funding for states and municipalities, since local taxes or contributions played (and still play today) a minimal role as subnational sources of income. Participaciones by Law "belong" to states and municipalities. Therefore, they cannot be earmarked for any purpose, neither by the Fiscal Coordination Law nor, evidently, by the Federal Government. Moreover, there is no federal supervision or audit as to their use. These tasks are performed by subnational authorities (state legislatures and state auditing offices); in fact, after participaciones are disbursed from the federal budget to states and municipalities, legally they are no longer federal resources. Each individual state or municipality's share in the Recaudación Federal Participable is determined by various formulas contained in the Fiscal Coordination Law. These formulas are fully transparent, and allow almost no discretion to federal authorities (in particular, to SHCP, the ministry in charge of administering the Fiscal Coordination Law) to alter the amounts due to subnational governments.2 The little discretionary power in the hands of SHCP derives mainly from its ability to vary the calendar when participaciones are disbursed. In particular, through "advance payments" (socalled adelantos) SHCP can aid a particular state facing a transitory difficult situation. But the margin for doing this is, on the whole, small. By and large, the Sistema Nacional de Coordinación Fiscal worked reasonably well over these years as a mechanism to collect taxes and transfer resources to states and municipalities (without implying, evidently, that the tax structure or the revenuesharing formulas were the most appropriate). Participaciones were not the only resources received by subnational governments from the federal budget during the 19791997 period. These were complemented with resources originally allocated by Congress to the budgets of federal ministries and agencies, but that could be transferred to subnational governments through "convenios." These are adhoc agreements signed between federal ministries/agencies and states and municipalities, by which the latter are entrusted to fully carry out tasks, or help in the execution of tasks, that are in principle a federal responsibility. Convenios share or delegate this responsibility, and transfer resources for that specific task. And though these resources are spent by a state or municipal ministry or agency, they are still federal, and are subject to supervision, audit, and sanction by federal authorities (the audit office of the federal ministry/agency in charge of oversight is the Ministry of the Comptroller and Congress's own audit office). Convenios are a very flexible mechanism by which, in principle, federal, state and municipal governments can work together to address specific needs. They can and are signed by various federal ministries/agencies in many areas: for roadbuilding, public positive oil shocks? The benefitcost ratio of the current System has been positive for states, and it is becoming more so as the discussion below points out. 2 Since some of the formulas depend on a state's or municipal's share of population in the national aggregate, disputes may arise at times, particularly since this information is updated only every five years. This and other technical issues are ignored in what follows. 126 Decentralized Service Delivery for the Poor security, justice administration, agricultural development, among others; as well as in areas associated with social protection and poverty alleviation like education, health, housing, and water sanitation. The flexibility of convenios gives the Federal Government a strong tool to interact with state and municipal governments, particularly in tasks in which the latter have a clear comparative advantage. They are also a powerful instrument to nudge sub national governments to follow and implement federal programs and policies, even though these policies and programs may not be contained in a Law (a new program for food distribution in poor communities, a subsidy for agricultural producers, etc.). This flexibility, however, comes at a cost: the rules for signing convenios are vague, so that--in contrast to participaciones--there is insufficient transparency and, one could argue, substantial discretionary power for the Federal Government who, at the end of the day, "owns" these resources (as opposed to participaciones, "owned" by subnational governments). Although convenios are signed continuously, there is no legal obligation for either party to do so. A particular federal ministry or agency might sign a convenio one year with a given state, but not the next. Yearly, the amount transferred might change from one year to the next with no obvious explanation; or the tasks delegated or required from the state government could be modified; or the conditions for the convenio to operate could be altered. The causes behind this instability, in turn, are many. For example, a federal administration comes to an end, and the new one is pursuing a different approach in, say, health policy, with the conditions imposed by the Ministry of Health (Secretaría de Salud, SSA) on state health authorities changing importantly. Or, Congress may substantially reduce the highway construction budget from one year to the next, thus the Transport Ministry (Secretaría de Comunicaciones y Transportes, SCT) signs convenios for a much smaller amount for caminos rurales with states, perhaps leaving some caminos unfinished. Another example might be a fall in aggregate tax revenues in one year, forcing SHCP to cut the federal budget. This in turn impedes the Ministry of Education (Secretaría de Educación Pública, SEP) from signing the convenios with states for building, say, technical secondary schools. Alternatively, political considerations could at times filter into decisionmaking, favoring a particular party/governor. Clearly, the flexibility of convenios was a positive element from the point of view of updating and adapting policies and budgetary allocations to changing needs and circumstances, but the process could also lend itself to rentseeking or influence peddling principally by states and, sometimes, by powerful or important municipalities (e.g., Monterrey or Acapulco).3 3 In this context of insufficiently defined budget constraints, rentseeking could also result from the activities of private groups. Thus, agricultural producers could, in the face of an unexpected drop in the world price of a particular commodity, or as a result of a larger than expected harvest, for instance, pressure the Ministry of Agriculture (Secretaría de Agricultura, Ganadería, Desarrollo Rural, Pesca y Alimentación, Decentralized Service Delivery for the Poor 127 Convenios are signed by executing ministries and agencies in the Federal Government, without the direct intervention of SHCP. With few exceptions, SHCP has no exante knowledge of how many resources will be transferred, when, by what ministry/agency, and to which state. As a result, various ministries/agencies have both normative and budgetary responsibilities. Negotiations between federal ministries/agencies and state governments are at least yearly, and take place on many fronts (convenios for education, agriculture, health, schoolbreakfast programs, water, social development, financial "support", etc.). In principle, attempts are made to unify these negotiations into a single instrument (the Convenios Únicos de Desarrollo later relabeled as Convenios de Desarrollo Social), but the coordination problem within the Federal Executive is too difficult to solve. De facto, each federal ministry/agency engages in bilateral negotiations with each governor. It is important to note that SHCP also had the ability to sign convenios with states to transfer resources in addition to participaciones, when states ran into "serious" financial difficulties. This possibility gave SHCP a strong hand with which to influence states' policies. However, this practice created substantial problems of its own, since the discretionary nature of these transfers, political considerations aside, implied the usual incentive problems associated with noncredible defaults and soft budget constraints. Resources transferred by SHCP to states (and sometime to municipalities) were in principle for debtrelief and restructuring. But given the fungible nature of funds, they could end up paying for a state highway, an urban development project, a university, a hospital, or whatever project for which a state had borrowed previously.4 In some cases, states are required to contribute with their own resources to program funding (socalled paripassus) with the purpose of expanding coverage faster, or modifying incentives to states by making them bare at least part of the cost of a given policy. And, although this may be a sensible policy in many circumstances, at times it could also impede full program execution (because states do not have or were unwilling SAGARPA) for "emergency" resources, resulting in an addendum to the convenios previously signed by SAGARPA with the states experiencing the shock, or new convenios. 4 From 1995 to 1998 SHCP tried to use these convenios to restructure states' debts, which had been affected by the 1995 economic crisis. In 1999 no resources were incorporated in the federal budget for this purpose, in an attempt to give credibility to the Federal Government's intention to show states a harder budget constraint. This attempt worked partially. In the federal budget for 2000 Congress included a new program to transfer nonearmarked resources to states, aside from participaciones and the aportaciones of Ramo 33. There were three important differences, however: first, the statedistribution of funds was determined by Congress and not by SHCP, as was the case up to 1998, thus making it more transparent while reducing SHCP's negotiating power with states; second, the amounts were fixed and could not be increased during the year (which was not the case with the previous convenios); and third, there was no conditionality of fiscal adjustment or reporting fiscal information to SHCP, as was the case with the 199598 agreements. Moreover, there is no transparency in the formula which has been changing every year. This program continues today in the federal budget, the Programa de Apoyo para el Fortalecimiento Financiero de los Estados, PAFEF, and its overall size is the object of yearly negotiations in Congress, as it is not contained in any Law, but incorporated in the yearly budget decree. PAFEF generates important considerations for fiscal decentralization, and some additional remarks are offered below. 128 Decentralized Service Delivery for the Poor to commit the required resources).5 In other cases, timing issues would also affect program execution. Given the yearly nature of the budget, and the fact that ministries/agencies could transfer resources late in the year, states were impeded from spending the full amount agreed to (because of calendar manipulation by SHCP, prolonged negotiations between a particular governor and a given federal ministry/agency, or simple bureaucratic inefficiency). Lastly, it is important to point out that, under these circumstances, it was difficult for communities to fully understand how many resources would be spent on any given year by which level of government and for what purpose. A school breakfast program might be notably expanded, but communities may not know whether this is the municipal president's policy, the governor's policy, or the President's policy. The hospital might need repairs and new medical equipment, but it is not clear who is responsible for doing so; the same is true with a road that was severely damaged after the rainy season. With little transparency as to the source of funding, and substantial overlap in federal, state and municipal responsibilities (particularly in social programs), ordinary citizens find it hard to make a clear association between the taxes they pay, the subsidies they receive, and the level of government accountable for success or failure of key programs or policies. The same observation holds, mutatis mutandis, for programs and policies for poverty alleviation. III. MOTIVATION FOR CHANGE The institutional arrangement described above was subject to progressive pressures overtime. Population growth, migration, increased complexity of the services provided and a maturing democracy, among many factors, would make the decentralization formula [participaciones + convenios] progressively insufficient and inefficient.6 Direct federal management of very large and geographically disperse servicedelivery organizations became increasingly difficult, and outcomes, at least in terms of quality, were deemed below expectations. Overtime, a consensus emerged centered on the need to involve state and municipal governments more directly in the day to day administration and provision of basic services, particularly in areas like education and health. 5 A related problem arises when federal ministries fail to notify state governments in a timely manner of the resources that they plan to spend in that state. If, as is often the case, state governments commit fully their own budgets at the beginning of the fiscal year, the lack of timely information may impede them from contributing with the pari passu requirements, even if they are willing to do so. 6 For example, provision of health services is gradually changing from simple vaccinations and treatment of contagious diseases to coverage of degenerative diseases like diabetes, hypertension, and cancer, requiring more complex equipment, more specialized personnel, and procurement of a larger number of medicines. All this is to be delivered in an increasing number of locations. In education constitutional reforms expanded basic education from six years of primary school to three additional years of secondary school. Decentralized Service Delivery for the Poor 129 A first step in this direction was the 1992 Acuerdo Nacional para la Modernización de la Educación Básica, signed by the Federal Government, the national teachers' union and state governors. (Note that this was a voluntary agreement among parties, not a Law. Initially Congress did not participate, although in 1993 the main elements of the Acuerdo were incorporated in the Ley General de Educación). A second step was the 1996 Acuerdo Nacional para la Descentralización de los Servicios de Salud (involving a similar set of actors). These two agreements in principle transferred to states (not municipalities) the physical infrastructure and day to day operational responsibilities in the provision of health and basic education services, with the Federal Government specializing in a more normative role. As part of these acuerdos, SEP and SSA transferred the financial resources previously earmarked in their budgets for these tasks to states. This implies that, at least initially, there were no additional resources; states would in principle receive, through a convenio, the same resources that these two ministries were spending previously. Thus, the preacuerdo distributions between states were maintained, although they were not derived from systematic equity considerations. This situation has generated substantial friction among those states that consider that they are not being treated equitably by the Federation. In parallel with the technical and administrative considerations that called for a sharper division of labor among the various levels of government in the provision of social services, Mexico's maturing democracy introduced an equally important set of considerations. The decentralization formula [participaciones + convenios] was politically sustainable in a context where all (until 1989), or almost all, governors were members of the same party as the President, and when the President's party had a majority in both houses of Congress (up to 1997). Disputes over resources between federal ministries/agencies and state and municipal governments were in principle settled on technical terms. The President would always be the last resort arbiter in case a particular governor and a federal ministry did not settle on a mutually satisfactory understanding. As more governors were elected from parties other than the PRI, this dispute settlement mechanism became more openly controversial (e.g., the dispute between the PAN governor of Baja California, on the one hand, and SEP and SHCP, on the other, over the state's basic education budget). At the same time, a larger number of nonPRI legislators in Congress and an increasingly free press also questioned the President's large discretionary powers over budgetary allocations, including the distribution of resources to states and municipalities. In some cases, questions were raised about the potential use of some social programs for electoral purposes (e.g., the criteria used by the Ministry of Social Development, SEDESOL, in the Programa Nacional de Solidaridad, 130 Decentralized Service Delivery for the Poor PRONASOL). By mid1990's more and more voices called for "clearer budgetary rules," including those concerning the transfer of federal resources to states and municipalities.7 The midterm federal elections of July 1997 mark an important turning point; for the first time in more than six decades no political party obtained a majority in the lower house of Congress. It is in the lower house that the spending side of the federal budget (Presupuesto de Egresos) is approved (as opposed to the Ley de Ingresos--the revenue side, which also requires approval by the Senate). Two factors came together at this time. On the one hand, the experience gained from the 1992 and 1996 agreements to decentralize basic education and health, and the 1996 precedent for distributing twothirds of PRONASOL's budget through an explicit formula, aimed to improve quality in the delivery of basic education, health, and social infrastructure, respectively. On the other hand, there were calls by nonPRI governors and a majority of nonPRI deputies in the lower house of Congress, for more transparency and reduced discretion in the budgetary relationships between the federal and state and municipal governments.8 These two factors generated the conditions for a substantive change to the almost twenty year old [participaciones + convenios] decentralization formula. Consequently, at the end of 1997, president Zedillo proposed a reform of the Fiscal Coordination Law to Congress to create a new mechanism to transfer resources to states and municipalities, and to add a new Ramo (number 33) to the federal budget. This proposal was approved by Congress in December of that same year, adding a new chapter to the Fiscal Coordination Law and introducing the novel category, "aportaciones" which had five specific funds. Correspondingly, these funds integrated their budgetary allocations into the newly created Ramo 33 (labeled Aportaciones Federales para Estados y Municipios) of the 1998 federal budget. IV. PURPOSE AND NATURE OF APORTACIONES AND RAMO 33 Aportaciones are, from a legal point of view, an intermediate figure between participaciones and convenios. On the one hand, these resources in principle "belong" to the Federal Government, which allows targeting to specific uses, regardless of states and municipal governments' preferences. Furthermore, the Federal Government retains a last resort capability to supervise and audit their use. On the other hand, state and municipal governments have the legal right to receive these resources from the federal 7 Perhaps partly in response to these observations, in 1996, president Zedillo unilaterally announced to Congress that as of fiscal year 1997 SEDESOL would distribute 65% of the resources in Ramo 26 belonging to PRONASOL to municipalities by means of an explicit poverty based formula. 8 Congress also negotiated other changes to the federal budget, concerning the spending rules in case there were unexpected additional revenues, and the allocation of a lineitem to the Federal Executive not subject to auditing by Congress, among others. Put differently, decentralization of resources was at this point part of a broader concern associated with the distribution of power between Congress and the Executive over control of federal resources. Decentralized Service Delivery for the Poor 131 budget in the dates and amounts specified in the Fiscal Coordination Law as long as they use them for the stated purpose, independently of the will of the Federal Government.9 In principle, aportaciones are a vehicle to: · transfer federal resources to states and municipalities for a specific purpose on the basis of clear and transparent formulas, independently of whether these formulas are the most appropriate ones for the given purpose; · delineate more clearly federal and state responsibilities, allowing the Federal Government to, in principle, concentrate on normative issues, while state and municipal governments focus on operational ones; · provide state and municipal governments with more stability, certainty and information as to the dates and magnitude of the resources they will receive from the federal budget in order to improve their own budgetary planning;10 · induce faster and in principle better program execution by eliminating one layer of bureaucratic control (the federal one), and promote efficiency as savings from programs that were not spent in the fiscal year could be transferred to the next; · reduce the discretionary power of the Federal Executive to negotiate convenios with state governors; · promote, in some cases, community participation in the decisions concerning the use of these funds; and, · shift the principal responsibility with regards to supervision and auditing from the federal to the state level. The "mixed" status of aportaciones is reflected in the legal provisions concerning supervision and auditing. In particular, the Fiscal Coordination Law contains a four step procedure for this purpose: (i) the federal Ministry of the Comptroller supervises that SHCP budgets each fund in Ramo 33 in accordance with the formulas contained in the LFC, and that the corresponding resources are delivered at the specified dates to states or municipalities; (ii) state or municipal level authorities, depending on the nature of each fund, supervise that the resources are spent properly; (iii) the auditing 9 An important consideration generated by aportaciones has to do with sharing macroeconomic risks between the three levels of government. If there is an unexpected negative shock during the fiscal year, SHCP cannot adjust aportaciones downwards, and all of the budget reduction must be borne by federal programs (assuming a constant deficit). This works conversely for a positive shock. The situation here differs from that of participaciones, where variations in the Recaudación Federal Participable (in any direction and coming from any source) are passed on to states. Further discussion is required to determine whether this is an adequate risk distribution, but the issue is not addressed here. 10 To contribute to this aim, as of 1998, the federal budget included the state breakdown of resources for all funds in Ramo 33. Equally important, it also made explicit the distribution of resources budgeted for wage increases during the year for all government employees, including those in health and basic education covered by Ramo 33 resources. This reduced, but did not eliminate, uncertainty over a major component of states' expenditure over the course of the fiscal year, as the discussion below points out. Further, the Fiscal Coordination Law also stated the date at which some of Ramo 33 funds would be disbursed. 132 Decentralized Service Delivery for the Poor authorities of each state legislature then supervise that the executing ministries and agencies of states and municipalities applied the funds to the purposes stated in the Fiscal Coordination Law; and (iv) finally, the auditing authority of the Federal Congress in turn supervises that the Ministry of the Comptroller and SHCP each complied with the requirements of the Fiscal Coordination Law. To facilitate this fourstep procedure, state and municipal supervising authorities are required to notify the Ministry of the Comptroller when they observe any deviations; similarly, states' legislatures auditing authorities are required to notify Congress's auditing authority when resources are not applied to the purposes stated in the Fiscal Coordination Law. With the creation of aportaciones, states had new budgetary responsibilities: in their own use of Ramo 33 resources, in the mechanics for transferring funds to their municipios and, very importantly, in matters of auditing and supervision. To fulfill them properly, state legislatures had to pass new legislation. Thus, the legal framework of the Fiscal Coordination Law for auditing and supervision is complemented by, in principle, 32 additional laws (31 states plus the Distrito Federal). This results in a large potential variance in the nature and depth of the auditing and supervision process across states (as well as in the conditions imposed by states to transfer resources to their municipios, as elaborated upon below). The result of the 1997 reform to the Fiscal Coordination Law was a threepronged [participaciones + aportaciones + convenios] decentralization formula (although, as explained below, other channels have been created since then). Three central results of this reform are: (i) the Federal Congress (Executive) has more (less) power than in the past; (ii) uncertainty has been reduced but not eliminated and transparency has increased; and (iii) the overlap between national and subnational responsibilities in some areas has been reduced though, again, not eliminated. It is important to emphasize that the reforms to the Fiscal Coordination Law did not substitute aportaciones for convenios. The reforms added the former, but the latter remained as a channel to transfer resources to states and municipalities, in addition to aportaciones. The relative magnitude of each is different, as the resources currently decentralized by aportaciones are substantially more significant than those operated through convenios. However, the persistence of convenios leaves open a parallel decentralization channel, with all the advantages and disadvantages mentioned before. Furthermore, in some sectors, federal transfers to states are still wholly operated through convenios. Indeed, for ministries like SAGARPA or the Ministry of the Environment (Secretaría de Medio Ambiente y Recursos Naturales, SEMARNAT) the mechanisms for transferring resources to states and municipalities are not very different in the recently approved federal budget for 2006, than what they were in 1996. The same could be said for the National Water Agency (Comisión Nacional del Agua, CNA).11 11 As a result of this, states still do not know the full amount of federal resources that they will receive in a given year from the federal budget. Decentralized Service Delivery for the Poor 133 Aportaciones have helped to increase transparency and signal a "harder" budget constraint in some sectors, not all. It is noteworthy that, since the addition of two more aportaciones funds in reforms approved in December 1998 to the Fiscal Coordination Law, there have been no further legal initiatives to advance Mexico's decentralization process. Instead, aside from continuing with the convenios as a transfer mechanism between the Federal Executive and state and municipal governments, a new conveniotype mechanism has de facto been created. This mechanism, the PAFEF has been voted into the budget decree by the House of Representatives over the last seven years (20002006). The novelty is that this is an agreement between Congress and state governments, without any formal intervention by the Federal Executive.12 It is clear from the above description that expenditure decentralization in Mexico has changed importantly over the last decade. Decentralization now occurs through many channels. Part of it takes place as a result of a legal mandate derived from the Fiscal Coordination Law (participaciones and aportaciones), with the corresponding attributes of transparency and stability. Part of it continues to take place through old channels (convenios), with the corresponding attributes of flexibilitycumdiscretion. And part of it takes place through new channels (PAFEF and sharing of positive oil shocks) as a result of a yearly negotiation in which Congress plays a key role, with the corresponding attributes of uncertainty, and potentially asymmetric risk distribution. It is not very clear, however, that expenditure decentralization has advanced as part of an overall institutional arrangement that: assigns resources to the level of government best suited to perform a given task; provides incentives to all levels of government to enhance tax revenues; distributes macroeconomic risks appropriately; or ensures that key social services are delivered efficiently to all, including those living in poverty. Aportaciones and Ramo 33 must therefore be seen in a broader framework, since they are only part of the decentralization process. Even if all the objectives posed when Ramo 33 was created were fully achieved (which probably is not yet the case), states would continue to face budgetary uncertainty as to the transfers that they would receive from the federal budget. Further, to the extent that the federal and state governments can still sign convenios, clear budget constraints have yet to be established. Moreover, to the extent that state governments can still extract additional resources from the federal budget through yearly negotiations with Congress, incentives may not be fully aligned 12 There have been other changes that also affect the distribution of federal resources between the three levels of government. Perhaps the most important one has to do with the formulas contained in the budget decree for sharing nonbudgeted revenues derived from a world oil price higher than the one assumed in the Federal Revenue Law, with states now receiving some of the additional benefits. In the end, the persistence of PAFEF and the sharing of unexpected oil revenues is just the budgetary reflection of a continuing change in the balance of power between the Federal Executive and Congress and, indirectly, between the federal and state governments (given the more active participation of the latter in budget discussions in Congress). 134 Decentralized Service Delivery for the Poor in the direction of tax efficiency. In addition, to the extent that the Federal Government continues to introduce new programs in basic education or health, for example, a clear division of tasks between the three levels of government has yet to be achieved, and a clear message to citizens as to who is responsible for what is yet to be delivered. Lastly, to the extent that national trade unions continue to operate in their current form, it will be difficult to make providers of services fully accountable to subnational levels of government; this has evident implications for monitoring and incentives (as the discussion below will point out). Aportaciones and Ramo 33 have probably helped to advance Mexico's decentralization process, but they are not the substitute for an overall strategy for fiscal federalism, tradeunion modernization, or poverty alleviation. V. PRINCIPAL APORTACIONES FUNDS Congress initially approved five different funds in Ramo 33 in December of 1997. As commented, a year later, two more funds were added via an additional reform to the Fiscal Coordination Law. Of these seven funds, five are explicitly associated with social programs, with some transferring resources to states, and some to municipalities.13 The main characteristics of the five funds associated with social programs are discussed next; although, it should be stated at the outset that not all of these funds are directly associated with poverty alleviation. Fondo de Aportaciones para la Educación Básica, FAEB As mentioned, in 1992 the Federal Government signed an Acuerdo with the national teachers' union and state governments to transfer direct responsibility for basic education to the latter, with resources directly transferred to each state by SEP. Five years later, FAEB served as a mechanism to formalize the budgetary aspects of this Acuerdo without, as expanded below, immediately changing the relationship between the teachers' union and the Federal Government, or having an important quantitative effect on the federal resources available to states for basic education.14 Given the budgetary and political constraints under which FAEB was born, the Fiscal Coordination Law stated that federal resources decentralized to states would be based on an inertial formula derived mainly from the number of teachers and schools in each state (Registro Común de Escuelas y de Plantilla de Personal). The formula allows for a yearly update based on the number of schools built or teachers hired in each state 13 The other two funds are for public security (to states), and for financial strengthening (to municipalities). The fact that Ramo 33 contains resources for such diverse purposes helps to reinforce the idea that it is, in essence, just a legal vehicle to transfer (decentralize?) federal resources to subnational governments ear marked for a particular purpose. The specific details of each fund, therefore, are essential to determine to what extent effective decentralization is achieved. 14 Federal resources for basic education increased importantly after the signing of the 1992 Acuerdo, particularly for teachers' salaries during 1993/94. FAEB did not bring any additional increases after 1998, although it reflected the inertia derived from the salary provisions of the Acuerdo, particularly the carrera magisterial. Decentralized Service Delivery for the Poor 135 during the previous year, and salary and other pay raises granted during the year. The formula, however, does not contain any elements associated with states' performance, the number of actually enrolled or schoolaged (potentially enrolled) students, or incentives for quality improvements in the delivery of basic educational services.15 Further, there is nothing in FAEB directly associated with poverty alleviation. (Of course, to the extent that the poor rely almost completely on public education, they should in principle also benefit from this change; however, the point here is only that FAEB was not specifically targeted to them, nor that any special mechanisms were incorporated into it for these purposes.) FAEB mirrors the principles established in the Ley General de Educación in the sense that basic education is a shared responsibility between the Federal and states' governments. This implies that formally there is no role assigned to municipalities, nor are any FAEB resources shared with this level of government. This does not preclude states from, in turn, decentralizing some resources further, or making arrangements with their municipios to participate in the delivery of basic educational services. The Fiscal Coordination Law does not prohibit them, although it does not promote this practice either. If further decentralization occurs this is a statelevel decision, and it takes place at the pace and with the scope that the state governor and legislature determine. Though FAEB brings more certainty to states, it does not do so fully. This is because resources for pay raises and related salary measures (particularly end of year bonuses and carrera magisterial) are initially (i.e., at the beginning of the fiscal year) kept by SEP in Ramo 25. These are budgeted for the purposes of covering the results of a single centralized negotiation between the Federal Government (formally represented by SEP, but sometimes with a behindthescenes participation from SHCP) and the teachers' union. Once the outcome of this negotiation is known (usually in midMay), resources are then transferred from Ramo 25 to FAEB in Ramo 33 and distributed to states on the basis of the national registry of teachers and schools. Pay raises are thus in principle covered by the Federal Government. But since states had some teachers on their own budgets prior to the 1992 Acuerdo and the creation of FAEB, they must extend any pay increases to these teachers with their own resources, meaning by and large their participaciones. As a result, de facto basic education is paid for by a mix of participaciones and aportaciones, with the proportions varying from state to state in accordance with the share of "maestros federalizados" (i.e., covered by FAEB) to the total number of teachers in each state (which varies widely). Furthermore, in some states there may be a further round of negotiations between the teachers' union and the state government. These negotiations may result in additional pay raises or other implicit wage increases, which must also be covered with state resources (i.e., 15 Moreover, since the formula reproduces the pre1992 Acuerdo distributions, it is not equitable among states. 136 Decentralized Service Delivery for the Poor participaciones). Thus, at the beginning of the fiscal year, states know how many resources they will receive initially from the federal budget for basic education, but they do not know what will be its endyear cost, nor the resources that they will have to contribute from their own budgets. Uncertainty over the outcome of the first negotiation (i.e., between the Federal Government and the national teachers' union) is not complete, however. Starting in 1998 resources budgeted for teachers' pay increases have been made explicit in the federal budget, so that state governments have an approximate, though not exact, knowledge of what the results of the centralized bargaining will be. It is difficult to make an overall assessment of the above; but two points can be made. First, it is clear that decentralization is not complete and that states do not have full control over their budgets for basic education; as it stands, the largest component of costs personnel remuneration is in large measure not state determined.16 And second, the importance of this budgetary uncertainty varies significantly from state to state (since, as mentioned, neither the 1992 Acuerdo nor FAEB corrected the historical imbalances in the distribution of resources between states). The de facto division of basic education financing by states, partly with their participaciones and partly with their aportaciones, along with the double negotiations between federal (first) and state (second) authorities over teachers' salaries, has two corollaries. One is that this process may not be fully known to the citizens of a state. The context described makes it difficult for parents to make a clear association between the taxes they pay, and the quantity and quality of the educational services that their children receive. A second corollary is that states' attitudes and incentives with regards to the teachers' union in matters related to remunerations may vary; states with no or proportionately very few "maestros no federalizados" may not offer strong resistance to large pay raises. Indeed, there could be some elements of free riding or opportunistic behavior in that the costs of politically attractive pay raises is borne by the Federation. Conversely, in states where a large share of the costs of basic education is covered by their own budget, attitudes may be different. These two corollaries underscore that FAEB is an intermediate point in a decentralization process, but clearly not its end point. This is particularly true if the objectives of decentralization are to improve accountability, quality, and performance. FAEB brings benefits in terms of transparency and stability; it is most likely better than the convenio arrangement that preceded it. Nevertheless, FAEB does not address a fundamental issue: how to motivate teachers to improve quality and be accountable to local authorities in the context of a national (and powerful) teachers' union. 16 Note also that states do not fully internalize the costs of their teachers even if they cover their social security contributions, given that their pension plan with ISSSTE (the social security institute for government workers) is not fully funded by the state. ISSSTE's deficit is completely covered by the Federal Government. Decentralized Service Delivery for the Poor 137 FAEB also does not have any explicit mechanisms to promote community participation or parental involvement in basic education, although in principle it should facilitate both. These mechanisms are in the 1992 Acuerdo and in the Ley General de Educación. It may be useful to observe, nonetheless, that decentralizing resources to states should not be interpreted as releasing the Federal Government from its responsibilities in basic education. Indeed, decentralization of operations to states may require a strengthening of the Federal Government's involvement in tasks that are complementary to this decentralization including: establishing mechanisms to promote effective parental participation and community involvement; supervising results; and disseminating performance comparisons across states (amongst other possible tasks). Auditing and supervision should not be thought of as only involving financial resources; this is a task which, under FAEB, mostly concerns states. It should also encompass impact and performance results, and SEP could ensure the inclusion of such accountability measures according to the 1992 Acuerdo and the Ley General de Educación. In a decentralization context it is essential that this task is performed by the central government. Communities and parents should, at the least, have ready access to credible information about what is happening in their schools (and in other schools). This allows them to effectively participate in their children's education, even if they are hardly empowered to do anything else, given their almost complete lack of control over spending decisions, or the hiring and firing of personnel. As stated, FAEB does not explicitly have any redistributive objective, though it should according to the General Education Law. From the perspective of poverty alleviation, however, it is essential to understand that the overall institutional arrangement for funding, administering, and operating basic education goes beyond FAEB. First, there are additional resources for basic education in Ramo 33 (in the Fondo de Aportaciones Múltiples, see below). And second, and more importantly from the point of view of the educational needs of the poor, there are other resources for basic education in the budget of SEP in Ramo 11. Through these federal programs, the Federal Government maintains an explicit responsibility in enhancing poor peoples' basic education. This responsibility is in practice reflected in programs to subsidize the demand for education, basically through the educational component of Oportunidades, and programs to improve the supply of educational services (quality and quantity), basically through CONAFE. As is illustrated above, delivering highquality basic education to the poor requires significant coordination between the federal and state educational authorities in order to yield desired results. It is in principle up to state authorities to ensure that teachers lecture from Monday to Friday in the schools where poor parents send their children, and that they correct their children's homework. But federal authorities should also, beyond setting the curriculum and drafting the national textbooks in Mexico City, assume other responsibilities. For example, it should ensure that poor children, particularly in remote rural areas but also in the urban poverty belts, attend school instead of working on the field or begging in the street; it should channel 138 Decentralized Service Delivery for the Poor additional resources to improve the educational infrastructure in very poor communities; and it should provide systematic impact and operational evaluations to judge how the poor are faring visàvis the rest of the population. In the Mexican context, decentralization is not a panacea for improving basic education for the poor. It may be part of the answer if it delivers what it is supposed to, but it must be complemented by strong Federal participation. However, decentralization may also hurt the quality of basic education that the poor receive. This is particularly true in the absence of the right incentives to state and municipal authorities, to the teachers' union, and to individual teachers. And it should be clear that FAEB, by itself, does not provide these incentives. Lastly, FAEB does not include the Distrito Federal. This is because in 1992 this jurisdiction had no independent legal status from the Mexican Federation (the Mayor was appointed by the President). Since then, however, major constitutional reforms have given substantial autonomy to the Distrito Federal.17 Decentralization of basic education to the Distrito Federal and its inclusion in FAEB is pending until the agreement is signed which would bring this into fruition.18 Therefore, those who live in the Distrito Federal still depend directly on the Federal Government for the provision of basic education, and have not seen any effects (positive or negative) of the 1992 Acuerdo or FAEB. It should be noted, however, that there has not been strong political pressure by the Distrito Federal to be included in FAEB. One can speculate whether this is because they perceive few benefits from this inclusion, or whether there are other motivations. Fondo de Aportaciones para los Servicios de Salud, FASSA As with basic education, FASSA builds on a decentralization agreement between the states and the Federal Government, this one signed in 1996. This Acuerdo, among other things, set the administrative basis for the transfer of operational responsibilities in health by inducing states to form specialized units for this purpose (Organismos Públicos Descentralizados, OPDs), that manage FASSA funds together with state resources for health delivery in their jurisdictions. FASSA transfers resources to states, not municipalities, thus leaving the statemunicipality relationship in health services delivery to statelevel legislation. The formula used by FASSA to distribute resources to states is very similar to the one used by FAEB, the only difference being the addition of a sofar quantitatively small component of resources not tied to personnel expenses that try to explicitly compensate 17 Though note that while there are now elections for a Jefe de Gobierno (no longer a Mayor), and there is a local legislature (the Asamblea General), the Distrito Federal is legally not akin to a state. 18 Since the constitutional reforms came into effect, the Distrito Federal has been governed by a party different from the one to which either presidents Zedillo or Fox belong. This, together with the sheer administrative and political complexity of any measure of this nature in the nation's capital, may explain the delay. Decentralized Service Delivery for the Poor 139 states with belowmean health indicators. Despite FASSA, a national union of health workers remains, and the Federal Government is its main counterpart for wage and salary negotiations. This generates a similar set of issues, mutatis mutandis, regarding budget uncertainty; mix of aportaciones and participaciones in funding; accountability; and incentives for community participation to improve the quality of services. From the point of view of poverty, and again parallel to the case of basic education, the Federal Government runs programs to improve health delivery to the poor, notably the health component of Oportunidades, the IMSSOportunidades Program, and (at least until very recently) the Programa de Ampliación de Cobertura, PAC. These compensatory programs are run directly by federal ministries or agencies with federal resources. Thus, strong coordination and cooperation between federal and state authorities, along with strong federal supervision, is required to obtain quality health services. The rules and regulations to promote this cooperation, however, are not in the Fiscal Coordination Law. As with FAEB, FASSA is just a budgetary vehicle that may facilitate, but does not ensure, an effective decentralization process, particularly from the point of view of poor households. It would be wrong, however, to think that decentralization in health faces the same challenges and problems as with basic education. This is not the case, and in fact the budgetary arrangements and institutional setup is even more complex. To understand this, it is essential to point out that in Mexico the responsibility of the State to deliver health services derives from two separate constitutional mandates, and therefore operates through two separate channels. On the one hand, health protection is a constitutional right of all Mexicans regardless of their work status. On the other hand, health protection is also a constitutional right derived from social security legislation, given that social security in Mexico encompasses not only pensions but also health. And to deliver health services to beneficiaries of social security (formal sector workers and their families), Mexico has two large federallyrun institutions that directly provide these services, IMSS and ISSSTE, working independently of the states' OPDs. Moreover, IMSS delivers health services in some rural communities to families that are not beneficiaries of social security through a special program named IMSSOportunidades.19 And while this program is funded with fiscal revenues and not social security contributions, like the federal resources that are transferred to states through FASSA, the funds are spent by IMSS, hiring personnel belonging to the IMSS national union, with no interference by states. Thus, there is a threepronged segmentation of public health delivery: social security institutions for workers in the formal sector (IMSS and ISSSTE), federal 19 IMSSOportunidades operates in 17 out of the 31 states, and covers approximately two million families, or around 10 million people. As a result of a truncated decentralization effort during the administration of president de la Madrid, it is present in some very important states from the point of view of poverty alleviation (Chiapas, Oaxaca, Veracruz, Michoacán, Yucatán and San Luis Potosí, to name a few), but not all (notably Guerrero, Guanajuato and Estado de México). 140 Decentralized Service Delivery for the Poor programs run by social security institutions for non social security beneficiaries (IMSS Oportunidades), and state level organizations (OPDs) also for non social security beneficiaries. In turn, public health delivery is funded by social security contributions (IMSS and ISSSTE), federal resources for federal programs run by social security institutions (IMSSOportunidades), federal resources for aportaciones through FASSA for states OPDs, and state level participaciones to complement OPD funding. It would also be wrong, moreover, to think that from the point of view of poverty alleviation, health delivery by social security institutions is not relevant (particularly by IMSS). This is because access to social security is defined by work status and not by income level. In particular, any person with a bosstoworker relationship (relación obreropatronal) is, according to Mexico's labor legislation, obligated to belong to IMSS. And there is a large (and growing) number of poor workers in both the rural and urban areas that belong to this category and are in IMSS, despite largescale evasion. Construction workers in urban areas and agricultural workers (jornaleros agrícolas) in rural areas are two important examples. Thus, for better or worse, the impact of health decentralization on the poor through FASSA is more limited than the impact of basic education through FAEB. There are a large number of poor people whose health services were unaffected either by the 1996 Acuerdo or FASSA. This is because even though they are poor, they have a formal sector job and they receive services from IMSS, or because they live in a state where IMSSOportunidades operates. The above segmentation should also be seen in a dynamic context, given that high turnover in the labor market and internal migration generate a large set of outcomes. A poor campesino might live in Oaxaca and obtain a living part of the year working on his own two or three hectares, but then migrate the rest of the year to Sinaloa to work as a jornalero agrícola under a relación obreropatronal. He then gets health services part of the year (in Oaxaca) through IMSSOportunidades, and part of the year (in Sinaloa) through IMSS. But another campesino from the contiguous state of Guerrero with the same migration pattern will receive health services during part of the year in Guerrero from his local OPD (because IMSSOportunidades does not operate there), although for the rest of the year, while in Sinaloa, he will also get health services from IMSS. A poor urban trabajadora from Tabasco who migrates temporarily to a maquiladora plant in Chihuahua will get health services from IMSS while in the north, but from the Tabasco OPD if and when she returns home. It is difficult, under this set up, for the campesinos of Oaxaca and Guerrero and for the trabajadora of Tabasco, all probably poor, to understand what level of government is responsible for the quality of the health services that they receive. It is more difficult still to hold anybody accountable for possible shortcomings. These problems were not addressed in the 1996 Acuerdo, or in the Fiscal Coordination Law when FASSA was created. FASSA was an improvement over the previous situation, and despite the shortcomings that it shares with FAEB, it most likely has helped states to deliver better services. There are, nonetheless, deeper issues that Decentralized Service Delivery for the Poor 141 must be discussed to ensure that Mexico has an efficient and stable institutional setup to deliver health services to the poor. Decentralization is part of this discussion, but only one component. Complex incentive problems, derived from the link between health and social security and from the link between social security and labor markets, need to be solved. The issues are conceptually more complex than in the case of basic education, because the financing of social security has a direct effect on the labor market, which is not the situation in the case of basic education. And politically, they are at least as complicated as in the case of basic education, because there are two large national unions involved in the delivery of public health services.20 These issues need to be sorted out, independently of the improvements that can be made in the daytoday operation of FASSA, to improve quality health care for the poor. Fondo de Aportaciones para Infraestructura Social, FAIS The Programa Nacional de Solidaridad, PRONASOL, started by president Salinas in 1988, was by the mid1990s, the main budgetary initiative of the Federal Government directed towards social infrastructure projects for the poor. It was also the predecessor of FAIS. PRONASOL emphasized strong community participation. For this purpose, SEDESOL established direct contact with poor communities, through local Comités de Solidaridad, to agree on priorities, before transferring the agreedupon resources to state or municipal governments via convenios. Thus, a distinctive element of PRONASOL was a direct relationship between the Federal Executive and communities, with subnational governments playing a secondary role. PRONASOL principally financed public goods in the form of physical infrastructure projects including rural roads, water sanitation, urban development, educational and health facilities, and the like. However, some resources were also devoted to programs targeted to developing the human capital of the poor, or towards subsidizing private productive projects.21 In the federal budget a special lineitem was assigned to it (Ramo 26), with very flexible disbursement rules in terms of the types of projects, or the geographical distribution of funds. The absence of explicit rules to distribute resources to states and municipalities was partly remedied in 1996, when president Zedillo unilaterally decided to distribute 65 per cent of Ramo 26 resources to municipalities through a povertybased formula. This generated more transparency, although resources were still spent by SEDESOL. 20 In addition to the political complications, there are also important legal differences. The IMSS union belongs to apartado "A" of article 123 of the Constitution, while the Health Workers Union belongs to "B". These differences affect workers' pension regimes, wage bargaining mechanisms, and health delivery costs. 21 Niños de Solidaridad and Crédito a la Palabra are an example of each. Niños de Solidaridad granted an inkind consumption subsidy (a food basket), and a scholarship to children of poor families who attended primary school. The program was an important precedent of Progresa (now Oportunidades), and its resources were eventually absorbed by it. Crédito a la Palabra gave small credits to poor agricultural producers with no collateral. 142 Decentralized Service Delivery for the Poor Two years later, with the 1998 reforms to the Fiscal Coordination Law and the creation of FAIS, three additional steps were taken to further decentralization. First, transfers were tied to the Recaudación Federal Participable, generating more stability. Second, SEDESOL's participation in the choice and execution of projects was ended; under Ramo 33 these decisions were taken by states and municipalities. And third, a specific calendar for disbursing funds was written into the Fiscal Coordination Law, giving states and municipalities the certainty that budgeted resources would reach them in equal amounts during the first ten months of each fiscal year. Two characteristics make FAIS substantially different from FAEB or FASSA. First, FAIS was not preceded by any decentralizationtype Acuerdo with states or municipalities. PRONASOL was created by a presidential decision in 1988; later a large share of its funds was distributed with an explicit formula to subnational governments by another presidential decision in 1996. It was then converted into FAIS by Congress in 1998 with the reforms to the Fiscal Coordination Law, in a process where there was no intervention by any national trade union, and hardly any participation by state governors. Second, FAIS resources are not tied to a wage bill, as is the case with practically all of FAEB or, to a lesser extent, FASSA. These two characteristics implied a very different political process, with substantially less constraints on the depth of decentralization achieved. First, it permitted an explicit statement in the Fiscal Coordination Law targeting these resources directly towards poverty alleviation. And secondly, it allowed for transferring most of the funds to municipalities, not states.22 The Fiscal Coordination Law indicates, moreover, that FAIS funds are not intended for services aimed at the general population (like basic education or health), but for physical investments on a broad range of items directed towards the poor, thus giving states and municipalities considerable flexibility in their spending decisions.23 Thus, without diminishing the importance of basic education or health, the possibilities that decentralization will benefit the poor proportionately more than the rest of the population is greater in the case of FAIS than in FAEB or FASSA. These possibilities are, in principle, enhanced by another distinctive feature of FAIS visàvis FAEB or FASSA. In this case, the Fiscal Coordination Law mandates municipal and state governments to: (i) inform the inhabitants of the respective jurisdictions of the 22 The Fiscal Coordination Law mandates that FAIS be budgeted at 2.5 per cent of the Recaudación Federal Participable approved by Congress in the Ley de Ingresos, regardless of the latter's fluctuations during the year, or its endofyear value. Of the total FAIS budget, 2.197 per cent is distributed to municipalities, and 0.303 per cent to states. 23 Funds transferred to municipalities can be used for "agua potable, alcantarillado, drenaje y letrinas, urbanización municipal, electrificación rural y de colonias pobres, infraestructura básica de salud, infraestructura básica educativa, mejoramiento de vivienda, caminos rurales e infraestructura productiva rural". Those transferred to states are for "obras y acciones de alcance o ámbito regional o intermunicipal" (Fiscal Coordination Law, Article 33). Decentralized Service Delivery for the Poor 143 scope, costs and potential beneficiaries of each individual project financed by FAIS; (ii) promote community participation in the ex ante choice of projects, in their execution, and in their expost evaluation; and (iii) inform the population at the end of the fiscal year as to the results obtained.24 The extent to which FAIS funds effectively contribute to poverty alleviation depends importantly on two considerations, notwithstanding the above said. The first derives from the interaction between state and municipal governments. The Federation legally cannot have direct budgetary relationships with municipios; in particular, it cannot disburse funds directly to them. FAIS resources must be transferred from the federal budget to states' budgets, and subsequently from states to municipal budgets under the conditions and terms mandated by state legislatures.25 Thus, in addition to federal preferences on the use of FAIS resources as contained in the Fiscal Coordination Law, statelaws matter importantly, generating a potentially large variance across states in poverty alleviation outcomes. Some states simply transfer FAIS funds to municipalities with few or no constraints; but some impose substantive conditions on their municipios before funds are disbursed. These conditions, in turn, may be fully justified in economic terms. Thus, state laws may give governors the right to request municipalities to have a community participation process and some minimal costbenefit analysis in order to improve the quality of their projects, prior to fund disbursement. States may also try to solve a coordination problem between contiguous municipios, say, with regards to a health clinic (so that with joint funds perhaps only one is built, which may be better equipped). Alternatively, states might want to achieve economies of scale in building, say, a rural road (so that with joint funds a longer one is constructed and not two unconnected segments). Or, finally, the state might want to induce a single municipio into developing a larger project, say sewage facilities, providing incentives to do so with pari passu state funds. But governors, however, can potentially also impose conditions on presidentes municipales that de facto translate into obstacles for disbursing funds for other reasons. For example, the governor and the municipal president could be from two different political parties. Or, they might even belong to the same party, but they may lean differently in the selection of potential candidates for, say, the upcoming election for state governor. 24 The Fiscal Coordination Law, however, does not specify how these three objectives should be reached, how their achievement (or lack thereof), will be monitored, nor the type of sanctions to be applied in case there is no compliance with these provisions. 25 When FAIS began there was an important controversy between the Federal Government and some states as to how far state's legislation could either alter the distribution formulas to municipios contained in the Fiscal Coordination Law, or impose some conditionality. The controversy did not reach the courts, and was settled by a compromise where LFC formulas would be observed, along with states' rights to legislate conditions on their municipios. 144 Decentralized Service Delivery for the Poor The second consideration is internal to the municipio, as there is nothing in the Fiscal Coordination Law concerning the distribution of funds inside each municipality. This is an important consideration if one recalls that while there are approximately 2,500 municipalities in Mexico, there are almost 200,000 localities inside these municipalities, the majority of them poor and small (with between 500 to 2,500 inhabitants). Funds may effectively reach a poor municipality, but distribution between the cabecera municipal and remaining localities will depend importantly on the incentives, preferences and relationships between the presidente municipal, the cabildo and the other powersthatbe. There is also the question concerning the extent projects respond to the needs of the poor, not all those who live in a poor municipio are necessarily poor. Effectiveness also depends on the capabilities of municipal authorities in terms of project design, execution, and supervision. These issues are not easily solved by federal legislation, although they may matter importantly for the effectiveness in the use of federal resources. The issues, in fact, are inherent to the purpose of decentralization: in the end, the outcomes are very much up to local authorities. FAIS is a good candidate for efficient decentralization: by and large the externalities associated with the type of projects that it funds are fully internalized at the municipal level. Moreover, if information is available to all, direct participation by beneficiaries should lead to better project selection and execution. But there is no guarantee that this will be the case, or that decentralization will be propoor, despite the targeting formulas. Political issues associated with statemunicipal and withinmunicipal relations also play a critical role. These observations may help identify an important role for the Federal Government in FAIS. While SEDESOL clearly no longer controls the process by which funds are allocated or spent, it can still influence this process in the direction of efficient and povertyoriented outcomes by providing clear, timely, and abundant information. This could complement the information that, under the Fiscal Coordination Law, states, and municipios should provide to their constituencies. It can also perform a useful role comparing outcomes across municipalities, both within a state and across states. It can help identify what type of state legislation is successful in insuring a better use of FAIS funds, or what other legal, administrative or political considerations lead to better outcomes in some situations visàvis others. The role suggested here for the Federal Government is, mutatis mutandis, similar to the one mentioned before in the case of FAEB. Decentralization changed the role of the Federal Government in the provision of basic education but, at least for some years, it did not make it redundant. In Mexico's very varied landscape of states and municipalities, particularly with regards to the functioning of local democracy, the same can be said for basic social infrastructure. Fondo de Aportaciones Multiples, FAM FAM serves as a vehicle to transfer resources to states for two tasks previously carried out by two separate federal agencies: the various food and schoolbreakfast Decentralized Service Delivery for the Poor 145 programs run by the Sistema Nacional para el Desarrollo Integral de la Familia, DIF, and the school construction program run by the Comité Administrador del Programa Federal de Construcción de Escuelas, CAPFCE. Like FAIS, resources are budgeted as a share of the Recaudación Federal Participable approved in the Revenue Law. In another parallel to FAIS, resources are not tied to a wage bill. Yet, in contrast to FAIS, resources are not distributed among states with an explicit formula, nor is there a fixed disbursement calendar. Rather, resources are byandlarge distributed following an inertial formula based on the preRamo 33 spending patterns of DIF and CAPFCE in each state, in a manner similar, though not identical, to FAEB or FASSA.26 Further, the Fiscal Coordination Law states that resources intended for food programs and school breakfasts (i.e., the DIF component), should be specifically targeted on the poor, although no such targeting is expressly pointed out for the school construction funds (i.e., the CAPFCE component). FAM is thus a mixed fund in three senses: it has two distinct purposes; its logic is partly based on FAEB and FASSA, and partly on FAIS; and it is partially intended for poverty alleviation. Decentralizing CAPFCE began before the 1997 reforms to the Fiscal Coordination Law. The reforms only formalized a process, and helped to signal the detachment of the Federal Government from that task. However, the Fiscal Coordination Law also allows these funds to be used for higher education infrastructure (i.e., universities); hence, this component of FAM should not necessarily be considered as the "bricks, blackboards and desks" complement to FAEB. Decisions as to the division of funds between basic and higher education belong to the states. Decentralizing DIF followed a different logic. At the time Ramo 33 was created, the Federal Government was also revamping its food subsidy programs. A new integrated foodhealtheducation program, Progresa (now Oportunidades), started in July of 1997, replacing generalized food subsidies and other isolated food programs run by various federal ministries and agencies. In this context, transferring resources to states previously used by DIF for food and schoolbreakfast programs would allow the Federal Government to focus on a single national program, with states complementing Progresa with locally designed and operated food programs in poor communities with DIF funds via FAM .27 Fondo de Aportaciones para la Educación Técnica y de Adultos, FAETA FAETA transfers funds to states to cover two different programs previously under direct SEP control: adult education programs run by the Instituto Nacional de Educación para Adultos, INEA; and technological education delivered through 26 The Fiscal Coordination Law states that the state distribution of FAM, fixed at 0.814 per cent of the Recaudación Federal Participable approved in the Revenue Law, shall be specified annually in the budget decree, with no further criteria. In practice allocations are approximately carried out on an inertial basis. 27 These programs could take the form of schoolbreakfasts, communitybased kitchens, or some other formulas. 146 Decentralized Service Delivery for the Poor CONALEP. Resources are used mostly for personnel services, and hence the mechanics of FAETA are very similar to FAEB. Nonetheless, the Fiscal Coordination Law explicitly mandates that the resources for the adult education component of FAETA be distributed between states using formulas based on the relative performance in literacy, basic education, and labor training indicators.28 Thus, there is a partial focus on poverty alleviation, as with FAM. We close this section with a general remark on decentralization, Ramo 33 and poverty. In the mid1990's, and in parallel to the political and administrative motivations mentioned for the creation of Ramo 33, the Federal Government was also redesigning its poverty alleviation strategy. Programs, projects, and budgets were organized under a threepronged approach: (i) investments in human capital, (ii) development of physical infrastructure, and (iii) promotion of income opportunities. From the perspective of decentralization, the strategy gave the Federal Government the leading role in component (i), state and municipal governments the leading role in component (ii), with federal and state governments jointly responsible for component (iii). A set of federal programs would subsidize the demand (Progresa) and supply (CONAFE, IMSSOportunidades, PAC) of human capital formation. And a set of state and municipal programs (FAIS, the CAPFCE component of FAM) would subsidize physical capital investments.29 Thus, one should be able to identify an underlying logic linking decentralization with poverty alleviation. However, this is only partly so. Inertia, design shortcomings, administrative expediency, political constraints, and bureaucratic resistances have also affected the division of functions between levels of government, and mixed the purposes of various funds. States receive federal resources to subsidize food programs (the DIF component of FAM); compensatory programs for basic education are federally run, but are state run for adult education (the INEA component of FAETA). FAIS resources can be used for school construction, but the same is true for the CAPFCE component of FAM; health clinics can be built with FAIS but also with FASSA. Sometimes formulas are used to distribute resources, but this is not always the case. Sometimes the formulas are in the Fiscal Coordination Law, although on occasions they are left to the discretion of federal ministries. Responsibilities and functions are not fully delineated, independently of operational issues that produce low quality services. There certainly have been advances in decentralization to the benefit of the poor, but it is probably fair 28 These formulas are not contained in the Fiscal Coordination Law. Rather, the Fiscal Coordination Law mandates SEP to publish them in the Federal Register, but gives little indication as to what elements should be considered, or how often they should be updated. 29 Income generating opportunities for the poor would come from rural development and temporary employment programs (together with economic growth). These were to be implemented by different levels of government, depending on the nature of each program. However, as mentioned above, SAGARPA and SEMARNAP continue to decentralize their resources through convenios, so the discussion of the third component is not pursued here. Decentralized Service Delivery for the Poor 147 to say that the design of a coherent institutional framework for decentralization and poverty alleviation is very much work in progress. VI. FINAL CONSIDERATIONS In less than a decade Mexico's expenditure decentralization changed substantially. There is a large difference between the 1997 [participaciones + convenios] and the 2006 [participaciones + aportaciones + convenios + PAFEF + a share in positive oil shocks] formulas. Undoubtedly, the expenditure powers of the Federal Government have been reduced to the benefit of subnational governments. Additionally, many more authorities now participate actively in the budgeting, spending and auditing of federal resources. Today, Congress, governors, state legislatures, municipal presidents and cabildos all have a greater responsibility to taxpayers and citizens in general, as do the auditing authorities at the federal and state level. There is more balance and more diversity. Expenditure decentralization has been a process fueled by two parallel trends. On the one hand, there is a broad desire for increased efficiency in the use of public resources, with poverty alleviation being only one of many objectives. On the other hand, there is a desire to reduce the President's control over the federal budget. This process has taken place in the context of Mexico's evolving democracy, and in turn has affected and contributed to define the shape of this democracy. Expenditure decentralization thus matters on equity and efficiency grounds, but also on political grounds. A successful decentralization process would yield improvements in both these areas; an unsuccessful one would be harmful. The decentralization process experienced over the last decade has been conditioned by three key factors. First, there is the continued presence of the national trade unions in education and health. Second, there is an absence of clear and stable definitions regarding the responsibilities of each level of government in the provision of services. And third, the decentralization process has been primarily centered on the expenditure side of the federal budget. These three factors--very much a reflection of Mexico's current political context--have shaped and molded the nature of the outcomes, both at the macro and microeconomic levels. They have also introduced important distortions along the way. Until the political process comes to terms with these three factors, they will continue to do so. From a macroeconomic perspective, while, to date, decentralization has been consistent with overall stability, important issues need to be addressed to ensure continuity in this respect. Such central issues include the appropriate distribution of macroeconomic risks derived from unexpected shocks; the incentives of subnational governments to achieve tax efficiency in their jurisdictions; and the moral hazard of debt contracting by state governments when defaults are not credible. More generally, incentives have yet to be aligned at the subnational level, changing the relative rates of 148 Decentralized Service Delivery for the Poor return to rentseeking from the federal budget (either in the annual discussions with Congress or in their yearround negotiations with federal ministries), visàvis own efforts at tax collection. Decentralization should not be a continuing raid on the federal budget by state and municipal governments. Who is going to finance the large public infrastructure projects that are required to foster the country's competitiveness, particularly given the increasing liabilities of the public pension system? And how will stability be preserved if the world price of oil drops? From a microeconomic perspective there are also important challenges ahead. First, the problems associated with national unions and decentralized delivery of services need to be solved. This is, of course, easier said than done, and the fact that many years have elapsed since the respective basic education and health Acuerdos were signed without significant changes signals the large difficulties in doing so. Nonetheless, it is essential to do so if quality, performance, and accountability are to be enhanced. And second, the roles and responsibilities of each level of government need to be clearly defined. After the 1997 reforms to the Fiscal Coordination Law, the Federal Government initiated new programs, or continued with old ones, in areas that clearly are subnational responsibilities. This generates duplication of administrative costs, and sends a confusing signal to citizens as to the level of government responsible for given services. It can also generate perverse incentives to subnational governments, inducing them to shirk from some of their responsibilities and shift costs to the Federal Government while reaping political benefits. This, like the trade union issue, remains difficult. Mexico has a long tradition of assuming and expecting that problems should be solved by the President. Conversely, presidents have at times also sought legitimacy and popularity by intervening in areas of public policy that in principle belong to sub national governments. Old habits die hard, but program proliferation and duplication may not be the most efficient and effective way of using resources, holding subnational governments to their responsibilities, or inducing them to progressively develop the administrative and technical skills required for these tasks.30 These considerations matter for the poor in Mexico. Another episode of macroeconomic instability would have large negative affects on them, as always. But even if that is not the case, the poor desperately need a growing economy to create jobs and raise incomes faster than what has been observed in the past few years. Since large public infrastructure projects play a key role in promoting competitiveness and growth, it is clear that over the medium term the poor do not benefit from a division of the federal budget that progressively nullifies the Federal Government's possibilities to carry out such investments. Similar effects are observed when one considers the microeconomic distortions that currently characterize Mexico's decentralization, as the 30 The foregoing should not be interpreted in the sense of a complete disengagement by federal authorities. As argued in the case of FAEB and FAIS, for example, there are still important tasks for SEP and SEDESOL in these areas. The point is to find the right balance between constructive collaboration and freeriding. Decentralized Service Delivery for the Poor 149 poor are most likely the ones that suffer the most from lowquality basic education or health care. The remarks made above should be put into perspective, however. The process of decentralization has advanced over the last years. Compared to the 1997 situation, there is more transparency, increased budgetary stability and certainty for state and municipal governments, less discretion at the federal level and, as remarked, greater balance and diversity. These are important achievements for a country that was very centralized less than a decade ago. Moreover, at least for some states and municipios, decentralization should have helped to enhance the quality of services and augment the social infrastructure, which includes benefits for poor communities.31 The point, nonetheless, is to recognize that the process is far from complete, and that it needs to continue, perhaps with more speed and depth than that observed over the last few years and, hopefully, with a coherent vision of where it is headed. While this occurs, there are immediate tasks for federal ministries and agencies, that can be carried out within the existing legal framework: increasing performance monitoring and supervision; divulging information; fostering the adoption of state or municipal best practices; comparing and disseminating results among states and municipios; containing federal program proliferation; paying more attention to incentives; and inducing states to legislate laws that require more transparency and accountability in their own auditing procedures, among others. There are, however, more difficult design and political issues that in today's Mexico cannot be solved by the Federal Government alone: (i) a political understanding for a taxcumspending federalism that is sustainable and efficient from a macro and a microeconomic point of view, respectively; (ii) a clearer definition of responsibilities in key areas, like health; and (iii) a political agreement on the nature and role of national trade unions. In furthering a decentralization process that can improve services for all, and particularly for the poor, Mexican democracy faces a daunting challenge indeed. 31 The results of the sector studies mentioned at the beginning of these notes will hopefully provide empirical evidence to assess this. 150 Decentralized Service Delivery for the Poor REFERENCES Álvarez, José de Jesús (2005). Background Paper on Basic Education. ­ Washington, D.C. The World Bank. ­ Mimeo. ______, Vicente García Moreno, and Harry Anthony Patrinos (2006). Institutional Effects as Determinants of Learning Outcomes : Exploring State Variations in Mexico. ­ Washington, D.C. : The World Bank. ­ Mimeo. Andres, Luis Alberto (2005). Background Paper on FAIS : Social Infrastructure. ­ Washington, D.C. : The World Bank. ­ Mimeo. Arredondo, Armando (2005). Background Paper on Health Services for the Poor : The Mexican Case. ­ Mexico : The World Bank. ­ Mimeo. Arroyo Alejandre, Jesús y Antonio Sánchez Bernal (2003). "Políticas municipales para la promoción del desarrollo económico regional". ­ En: Políticas públicas municipales : una agenda en construcción / ed. Enrique cabrero Mendoza. ­ México : Miguel Angel Porrúa : CIDE, 2003. "Articulo 115 (1993). Los estados adoptaran, para su régimen interior, la forma de gobierno republicano, representativo, popular, teniendo como base su división territorial y de su organización política y administrativa el municipio libre". ­ En: Constitución Política de los Estados Unidos Mexicanos : Titulo Quinto de los Estados de la Federación y del Distrito Federal (Reformado mediante decreto publicado en el Diario Oficial de la Federación el 25 de Octubre de 1993). "Articulo 26 (1983). Capitulo IV. Plan y Programas". En: Ley de Planeación. Publicada en el Diario Oficial de la Federación el 5 de Enero de 1983. Ashton, Toni., J. Cumming, J. McLean (2004). "Contracting for health services in a public health system: the New Zealand experience." ­ In: Health Policy 69(1): 2131. Bardhan, Pranab and Dilip Mookherjee (2005). Decentralization, Corruption, and Government Accountability: an Overview. ­ For Handbook of Economic Corruption / ed. By Susan RoseAckerman, Edward Elgar. ­ Revised June 20, 2005. ­ 53 p. Caballero, José María (2005). Background Paper on Rural Development Programs. ­ México : The World Bank. ­ Mimeo. Cabrero Mendoza, Enrique (2003). "Descentralización de la Política Social y Nuevo Papel de los Gobiernos Locales. ¿Políticas Multinivel o Múltiples Políticas?: el Decentralized Service Delivery for the Poor 151 Caso de México.--En: Descentralización y Políticas Sociales en América Latina : dilemas políticos y estrategis de gestión pública/ ed. Jacint Jordana, Ricard Gomá. ­ Barcelona : CIDOB. Collins, R.A. and M.K. Hanson (1991). Summative Evaluation Report: SchoolBased Management/Shared DecisionMaking Project 198788 through 198990. ­ Office of Educational Accountability. ­ Miami, Florida : Dade County Public Schools. Colombo, F. and H. Oxley (2005). OECD Reviews of Health Systems :Mexico. ­ Paris : OECD. ­ 158 p. Díaz Cayeros, Alberto and Charles E. McLure (2000). "Tax Assignment". ­ In: Achievements and Challenges of Fiscal Decentralization : lessons from Mexico / ed. Marcelo M. Giugale and Steven B. Webb. ­ Washington, D.C. : The World Bank. ­ pp. 177199. ______ and Sergio Silva (2004). Descentralización a Escala Municipal en México : la Inversión en Infraestructura Social. ­ México : CEPAL. ­ 73 p. ­ (Serie Estudios y Perspectivas ; 15). EdStats : Education Statistics Database. ­ Washington, D.C. : The World Bank. Esquivel, Gerardo (1999). Educación y Desarrollo Regional : Una Evaluación Inicial de las Asignaciones del Fondo de Aportaciones para la Educación Básica y Normal. ­ Presentado en el Seminario El Federalismo Desde los Estados y Municipios / Colegio Nacional de Economistas y el ITAM (Noviembre, 1988 : México). FAO (2004). Evaluación de la Alianza para el Campo 2004. ­ México : FAO. Fidler, Armin H. [et al.] (2005). Incorporation of Public Hospitals : A Silver Bullet Against Overcapacity, Managerial Bottlenecks and Resouce Contraints? Case Studies from Austria and Estonia. ­Washington D.C. : The World Bank. ­ 15 p. Figueras, Josep, Ray Robinson, and Elke Jakubowski (ed.) (2005). Purchasing to Improve Health System Performance By ­ Europe : Open University Press. ­ (European Observatory on Health Systems and Policies). ­ 298 p. Fiszbein, Ariel (2004). Citizens, Politicians, and Providers : the Latin American Experience with Service Delivery Reform. ­ Washington, D.C. : The World Bank. ­ Mimeo. "Fórmula de Distribución de Recursos a las Entidades Financieras para los Programas Federalizados de Alianza para el Campo 2005" (2005).--En: Diario Oficial de la Federación (Enero 31, 2005). ­ México : Secretaría de Gobernación. 152 Decentralized Service Delivery for the Poor Giugale, Marcelo M. and Steven B. Webb (eds) (2000). Achievements and Challenges of Fiscal Decentralization : lessons from Mexico ­ Washington, D.C. : The World Bank. Grindle, Merilee (2004). "Interests, Institutions, and Reformers : The Politics of Education Decentralization in Mexico". In: Crucial Needs, Weak Incentives : Social Sector Reform, Democratization, and Globalization in Latin America / ed. Robert R. Kaufman and Joan M. Nelson. ­ Baltimore, M.D : Johns Hopkins Press. ­ 542 p. ______ (2005) Study of Mexican Municipal Development. Mimeo. Ham, Chris. (2003). "Betwixt and Between: Autonomization and Centralization of UK Hospitals". ­ In: Innovations in Health Service Delivery : The Corporatization of Public Hospitals / ed. Alexander S. Preker, April Harding. ­ Washington, D.C. : The World Bank. ­ pp. 265304. ­ (Health, Nutrition, and Population Series). Ham, Chris. and Loraine Hawkins (2003). "Implementing Organizational Reforms to Hospitals in the Public Sector". ­ In: Innovations in Health Service Delivery : The Corporatization of Public Hospitals / ed. Alexander S. Preker, April Harding. ­ Washington, D.C. : The World Bank.-- pp. 79104. ­ (Health, Nutrition, and Population Series). Harrison, M. I. and J. Calltorp (2000). "The reorientation of marketoriented reforms in Swedish healthcare". ­ In: Health Policy 50(3): 21940. Instituto Nacional de Estadística, Geografía e Informática (INEGI) (2002). Encuesta Nacional de Hogares Rurales de México (ENHRUM). ­ México : INEGI : El Colegio de México. Instituto Nacional para la Evaluación de la Educación (INEE) (2005). La Telesecundaria Mexicana: Desarrollo y Problemática Actual. ­ Mexico: Instituto Nacional para la Evaluación de la Educación. Jimenez, E. and Y. Sawada (1999). "Do CommunityManaged Schools Work? An Evaluation of El Salvador's EDUCO Program." In: The World Bank Economic Review 13 (3): 41541. Kerrigan, George (2001). Public Spending for the Agriculture Sector and Development of Rural Areas : LAC, Trends, and Challenges. ­ Santiago de Chile : FAO. King, E. and B. Ozler (1998). What's Decentralization got to do with Learning? The Case of Nicaragua's School Autonomy Reform. ­ Working Paper on Impact Evaluation of Education Reforms. ­ Washington, DC: World Bank. Decentralized Service Delivery for the Poor 153 Knaul, Felicia, Héctor ArreolaOrnelas, and Oscar Méndez (2005). "Protección Financiera en Salud : México, 1992 a 2004". ­ En: Salud Pública de México 47 (6):430439. Koppel, A. [et al.] (2003). "Evaluation of primary health care reform in Estonia." ­ In: Social Science & Medicine Jun:56(12): 24616. "Ley de Desarrollo Rural Sustentable" (2001). ­ En: Diario Oficial de la Federación (Diciembre 7, 2001). ­ México : Secretaría de Gobernación. Loera Varela, A. (2005). Cambios en las Escuelas que Participan en el PEC 20012004. Evaluación Cualitativa del Programa Escuelas de Calidad. Reportes Nacionales Fase 3. ­ México. Mimeo. LópezCalva, Luis [et al.] (2005). Poniendo al Ingreso de los Hogares en el Mapa de Mexico. ­ México. Mimeo. M.I.T. and the Bank of the Northeast (2000). Draft The Economic Wars Between the States. ­ (Presented at the OECD/State Government of Ceará International Meeting on Foreign Direct Investment in Brazil, December 12, 2002 Fortaleza.). Merino, Mauricio and Ignacio Macedo (2005). La Política Autista : Crítica a la Red de Implementación Municipal de la Ley de Desarrollo Rural Sustentable. ­ México : CIDE. ­ (Serie Documento de Trabajo ; 168). Merino, Gustavo (2003). "Descentralización del Sistema de Salud en el Contexto del Federalismo". ­ En: Caleidoscopio de la Salud. ­ Mexico : Fundación Mexicana para la Salud. ­ pp. 195207. Milanovic, Branco (2002). Worlds Apart : InterNational and World Inequality 19502000. ­ Washington, D.C. : The World Bank. ­ 120 p. Moreno, Carlos (2005). Decentralization, Electoral Competition, and Local Government Performance in Mexico. ­ Austin, Tex. : The University of Texas at Austin, LBJ School of Public Affairs. Organization for Economic Cooperation and Development (OECD) (2005a). Health Systems Review : Finland. ­ Paris, OECD. _____ (2005b). Mexico Economic Surveys. Paris : OECD. _____ (2005c). Review of Health Systems : Mexico. ­ Paris, OECD. Ozler, B. (2001). Decentralization and Student Achievement: The Case of Nicaragua's School Autonomy Reform. ­ Working Paper on Impact Evaluation of Education Reforms. ­ Washington, DC: World Bank. 154 Decentralized Service Delivery for the Poor Paqueo, Vicente, Gladys LópezAcevedo and Suhas Parandekar (2003). On the Use of Transparent Formulas to Allocate Federal Education Transfers. ­ Washington, D.C. : The World Bank. ­ ( Policy Research Working Paper ; 3173). ­ 14 p. Parker, Susan and Carla Penderzini (2002) "Diferencias de Género en la Educación en México". En Elizabeth G. Katz y María C. Correia (Coord.) La Economía de Género en México. México: Banco Mundial y Nacional Fianciera. Preker, A. (2005). Spending Wisely : Buying Health Services for the Poor . ­ Washington, D.C. ­ The World Bank. Presupuesto de Egresos de la Federación para el Ejercicio Fiscal 2005. ­ México : Cámara de Diputados del H. Congreso de la Unión. ­ 262 p. Programa de las Naciones Unidas para el Desarrollo (PNUD) (2004). Informe sobre el Desarrollo Humano 2004 : la Libertad Cultural en el Mundo Diverso de Hoy. ­ Mexico : PNUD. ­ 385 p. Qian, Yingyi and Barry R. Weingast (1997). Federalism as a Commitment to Reserving Market Incentives. ­ In: Journal of Economic Perspectives. 11(4):8392. Reglas de Operación de la Alianza para el Campo : con modificaciones http://www.sagarpa.gob.mx/sdr/progs2003/alianza2003.htm Scott, Graham, Lynne McKenzie and James Webster (2003). "Maladjustments in the Corporatization Model : Hospital Reform in New Zealand". In: Innovations in Health Service Delivery : The Corporatization of Public Hospitals / ed. Alexander S. Preker, April Harding. ­ Washington, D.C. : The World Bank. ­ pp. 305344. ­ (Health, Nutrition, and Population Series). Scott, John (2005). Background Paper : Seguro Popular Incidence Analysis. ­ México : The World Bank. ­ Mimeo. Secretaría de Agricultura, Ganadería, Desarrollo Rural, Pesca y Alimentación (SAGARPA) (2002). Programa Especial Concurrente para el Desarrollo Rural Sustentable : 20022006. ­ México : SAGARPA. Secretaria de Salud (SSA) (2005). Salud México 2004 : Información para la Rendición de Cuentas. ­ México : SSA. ­ 229 p. Serven, Luis and Cesar Calderón (2004). "The Effects of Infrastructure Development on Growth and Income". ­ In: Econometric Society 2004 Latin American Meetings ; 173. ­ 43 p. Decentralized Service Delivery for the Poor 155 Sorensen, R. J. and J. Grytten (2003). Service Production and Contract Choice in Primary Physician Services.. ­ In: Health Policy 66(1): 7393. South, O. (1991). From Compliance to Continuous Improvement­Leading and Managing Schoolbased Management. ­ Monroe County, Florida. Székely Pardo, Miguel [et al.] (2005). Poniendo a la Pobreza de Ingresos y a la Desigualdad en el Mapa de México. Mimeo. Taylor, D.L. and I.E. Bogotch (1992). Schoollevel effects of teachers' participation in decisionmaking. ­ Baton Rouge : Louisiana State University. Mimeo. Tendler, Judith (1987). What Ever Happened to Poverty Alleviation? ­ A Report Prepared for the MidDecade Review of the Ford Foundation's Programs on Livelihood, Employment and Income Generation. (March). ______ (2000). The Economic Wars Between the States. ­ Presented at the OECD/State Government of Ceará Meeting on Foreign Direct Investment and Regional Development, Fortaleza, 12 December 2002. http://www.oecd.org/dataoecd/44/17/2489851.pdf Vidal, Cecilia [et al.] (2006). Cobertura Efectiva del Sistema de Salud en México: 20002003. ­ México : SSA, Dirección General de Información en Salud : Instituto Nacional de Salud Pública : Harvard Initiative for Global Health. Webb, Steven, Christian González and David Rosenblatt (2004). Background Paper : Mexico Fiscal Overview. ­ Washington, D.C. : The World Bank.­ Mimeo. ______ and Juan Antonio Zapata ed. (2006).Experiencias Internacionales en Relaciones Fiscales Intergubernamentales : Alemania, Australia, Brasil, Canadá y España ­ Washington, D.C. : The World Bank : Innovación México :USAID : ITESM. World Bank (1999). Beyond the Center : Decentralization the State / Shahid Javed Burki... {et al.} ­ Washington, D.C. : The World Bank. ­ (World Bank Latin American and Caribbean Studies. Viewpoints). ______ (2000) Mexico: Transforming Schools into Effective and Efficient Learning Centers. ­ Report 20593ME. ­ Washington, D.C. : The World Bank. ______ (2001a) Mexico: A Comprehensive Development Agenda for the New Era. Washington, D.C. : The World Bank. ______ (2001b) Mexico: Advancing Educational Equity and Productivity in the Context of Decentralization. Washington, D.C. : The World Bank. 156 Decentralized Service Delivery for the Poor ______ (2002) Mexico : StateLevel Public Expinditure Review : the case of VeracruzLlave. ­ Washington, D. C. : The World Bank. ­ Report No. 25162ME). ______ (2003). Southern States Development Strategy. ­ México : The World Bank. ­Mimeo. ______ (2004a). Mexico Public Expenditure Review. ­ Washington, D.C. : The World Bank. ­ Vol. II. Main Report. ­ (Report No. 27894ME) ______ (2004b). Poverty in Mexico : An Assessment of Conditions, Trends and Government Strategy. ­ México : The World Bank. ­ lx, 270 p. ______ (2004c). World Development Report : Making Services Work for the Poor. ­ Washington, D.C. : The World Bank : Oxford University Press. ­ xv, 271 p. ______ (2004d).Universal Health Insurance in Mexico : In Search of Alternatives. ­ Washington, D.C. : The World Bank. ­ Mimeo. ______ (2004e).State Fiscal Overview. ­ Washington, D.C. : The World Bank. ­ Mimeo. ______ (2005a). "A Study of Rural Poverty in Mexico".--In: Income Generation and Social Protection for the Poor. ­ México : The World Bank. pp.141416. ______ (2005b) Mexico : Determinants of Learning Policy Note. ­ Washington, D.C. : The World Bank. ­ (Report No. 31842MX) ______ (2005b). Mexico : Infrastructure Public Expenditure Review (IPER). ­ Washington, D.C. : The World Bank. 196 p. ­ (Report No. 33483MX. ______ (2005c). World Development Indicators. ­ Washington, D.C. : The World Bank. ­ 440 p. ______ (2005d). Income Generation and Social Protection for the Poor. ­ México : The World Bank. ______ (2006a). Background Notes on Expenditure Decentralization and Poverty Alleviation in México. ­ México : The World Bank. ­ Mimeo. ______ (2006b) Quality of Education. ­ Washington, D.C. : The World Bank. ­ (Forthcoming) World Health Organization (WHO) (2000). The World Health Report 2000. Health Systems : Improving Performance. ­ Geneva : WHO. Decentralized Service Delivery for the Poor 157 Decentralized Service Delivery for the Poor: Reports Finished printing August, 2006 in IMPRIME TUS IDEAS, S.A. de C.V., Horacio 340, Polanco, CP. 11560, México, D.F. imprimetusideas@prodigy.net.mx Edition 200 copies