73697 1 © 2010 The International Bank for Reconstruction and Development/The World Bank 1818 H Street, NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 11 10 09 08 This volume is a product of the staff of the International Bank for Reconstruction and Development/The World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. 3 Do Our Children Have A Chance? The 2010 Human Opportunity Report for Latin America and the Caribbean José R. Molinas, Ricardo Paes de Barros Jaime Saavedra, Marcelo Giugale With Louise J. Cord, Carola Pessino, Amer Hasan 4 About the Authors José R. Molinas Vega holds a Ph.D. in economics from the University of Massachusetts, Amherst. He is a Senior Economist with the Poverty Reduction and Gender Group in the Latin America and Caribbean Region at the World Bank. Ricardo Paes de Barros holds a PhD in economics from the University of Chicago. He is currently a researcher and the coordinator for the evaluation of public policies at the Instituto de Pesquisa Econômica Aplicada (IPEA), Brazil. Jaime Saavedra Chanduvi holds a Ph.D. in economics from Columbia University in New York City. He is the Sector Manager of the Poverty Reduction and Equity Department of the World Bank. Marcelo Guigale holds a PhD in economics from the London School of Economics. He is Sector Director of the Poverty Reduction and Economic Management Department in the Latin America and Caribbean Region at the World Bank. Louise Cord holds a PhD in development and economic policy from the Fletcher School of Law and Diplomacy, at Tufts University. She is Sector Manager of the Poverty Reduction and Gender Group in the Latin America and Caribbean Region at the World Bank. Carola Pessino holds a PhD in economics from the University of Chicago. She is a professor of economics at the University Torcuato Di Tella, Argentina. Amer Hasan holds a PhD in public policy from the University of Chicago, Harris School of Public Policy Studies. He is a consultant with the Poverty Reduction and Gender Group in the Latin America and Caribbean Region at the World Bank. Table of contents 5 6 Table of contents Acknowledgments 7 Overview 9 Síntesis (In Spanish / En Español) 19 1. How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index 29 1.1. Key Concepts: Basic Goods and Services, Universality, Equality of Opportunity and Circumstances 31 1.2. Constructing a Measure of Progress Towards Basic Opportunities for All 36 1.3. Empirical Considerations for Constructing the Human Opportunity Index 43 2. The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 49 2.1. Progress in Improving Human Opportunities in LAC—Although Universality Remains a Generation Away 51 2.2. Opportunities for Children to Access Basic Services in the LAC Region 59 2.3 Expanding Human Opportunities in Latin America and the the Caribbean: 1995-2010 61 2.4. Unpacking Changes in the HOI: Scale, Equity and Evolving Circumstances 62 2.5 The Inequality of Opportunity Profile 68 3. Human Opportunities in a Global Context:Benchmarking LAC to Other Regions of the World 71 3.1 Human Opportunity Index for Quality Education 73 3.2 Human Opportunity Index for Housing 85 3.3 Understanding the Long-run Evolution of the HOI 91 3.4 Conclusion 93 4. Human Opportunities at the Sub-national Level in Latin America and the Caribbean 95 4.1. The Sub-national Human Opportunity Indices: Some Stylized Facts 96 4.2. Hetereogeneity in Subregional Human Opportunity Indices 104 4.3. Equalizing Regional Opportunities 113 4.4. Summary and Conclusions 117 Annexes Acknowledgements 7 8 “Do Our Children Have a Chance? The 2010 Human Opportunity Report for Latin America and the Caribbean� is the result of a collaborative effort that brought together a team of professionals from within and outside the World Bank. The report was prepared under the guidance of Louise Cord (World Bank) and Marcelo Giugale (World Bank) by a team led by Jose R. Molinas Vega (World Bank). Team members included Ricardo Paes de Barros (IPEA-Brazil), Jaime Saavedra (World Bank), Carola Pessino (UDTD-Argentina), and Amer Hasan (World Bank). The team received valuable support from Joao Pedro Azevedo (World Bank), Eliana Rubiano (World Bank), Carlos Sandoval (World Bank), Gabriel Facchini (World Bank), Ngoc-Bich Tran (World Bank), Will Durbin (World Bank), Samuel Franco (IPEA-Brazil), Andrezza Rosalem (IPEA-Brazil), and Ramiro Soria (UTDT). The report was improved by three principal reviewers: Francisco H. G. Ferreira (World Bank), Peter Lanjouw (World Bank), and Emiliana Vegas (World Bank). Juliana Pungiluppi (World Bank) worked on the Spanish translation and Chris Humphrey served as the editor. Ane Castro (World Bank), Lucy Bravo (World Bank) and Anne Pillay (World Bank) were instrumental in the production of the final report. While the writing of this report has been a collective effort, the principal authors of the chapters are as follows: • Overview: Marcelo Giugale • Chapter 1: Ricardo Paes de Barros, Jaime Saavedra and Jose R. Molinas. • Chapter 2: Jose R. Molinas • Chapter 3: Jose R. Molinas and Amer Hasan • Chapter 4: Carola Pessino Overview 9 10 Do Our Children Have A Chance? The 2010 Human Opportunity Report for Latin America and the Caribbean Overview Imagine a country where your future did not depend on where you come from, how much your family earns, what color your skin is, or whether you are male or female. Imagine if personal circumstances, those over which you have no control or responsibility, were irrelevant to your opportunities, and to your children’s opportunities. And imagine now a statistical tool that can help governments make that a reality. Welcome to the Human Opportunity Index (HOI). The HOI calculates how personal circumstances (like birthplace, wealth, race or gender) impact a child’s probability of accessing the services that are necessary to succeed in life, like timely education, running water or connection to electricity. It was first published in 2008, applied to Latin America and the Caribbean (LAC). The findings were eye-opening: behind the enormous inequality that characterizes the region’s distribution of development outcomes (income, land ownership and educational attainment, among others), there is an even more worrying inequality of development opportunities. It is not only rewards that are unequal; it is also chances. The problem is not just about equality; it is about equity too. The playing field is uneven from the start. This book reports on the status and evolution of human opportunity in LAC. It builds on the 2008 publication in several directions. First, it uses newly-available data to expand the set of opportunities and personal circumstances under analysis. The 11 data is representative of some 200 million children living in 19 countries over the last 15 years. Second, it compares human opportunity in LAC with that of developed countries, among them the US and France, two very different models of social policy. This allows for illuminating exercises in benchmarking and extrapolation. And third, it looks at human opportunity within countries—across regions, states and cities. This gives us a preliminary glimpse at the geographic dimension of equity, and at the role that different federal structures play. The overall message that emerges is one of cautious hope. LAC is making progress in opening the doors of development to all. But it still has a long way to go. At the current pace, it would take, on average, a generation for the region to achieve universal access to just the basic services that make for human opportunity. Seen from the viewpoint of equity, even our most successful nations lag far behind the developed world. And intra-county regional disparities are large, and barely converging. Fortunately, there is much policy makers can do about it. How Does the HOI Work? In its simplest interpretation, the HOI measures the availability of services that are necessary to progress in life (say, running water), discounted or “penalized� by how unfairly the services are distributed among the population. So, two countries that have identical coverage may have a different HOI if the citizens that lack the service are all female, or black, or poor, or have many siblings or, more generally, share a personal circumstance beyond their control. In other words, the HOI is coverage corrected for equity. In theory, you can increase it by changing people’s circumstances (the “composition effect�), providing more services to all (“scale effect�), or distributing services more fairly (“equalization effect�). 12 The HOI runs from zero to 100; a society that has achieved universal coverage of all services would score at 100. To make comparisons possible across countries and across time, the HOI for LAC presented in this report uses only services and circumstances that are available in all household surveys. Specifically, it looks at access to water, electricity and sanitation, and to school attendance and timely completion of the sixth grade. A rich empirical literature demonstrates that, without those basic services, the chances of a productive life are close to nil. And it focuses on seven personal circumstances: parents’ education, family income, number of siblings, the presence of both parents in the house, gender, gender of household head, and location of residence. In all cases, the unit of focus is the child, defined as an individual between the ages of zero and 16. This isolates away the problem of effort and choice— at that age, children can hardly be responsible for their fate. Of course, in country specific applications of the HOI, data availability may allow for more, or more sophisticated, services and circumstances, like preventive dental check-ups, internet access, ethnic identification, or father’s occupation. Some of that will be shown here, when comparing LAC countries with their developed-world peers. Is Human Opportunity Expanding in LAC? Yes, but slowly and with marked differences across countries. Since 1995, the region’s average HOI has grown at a rate of one percentage point per year. This is clearly insufficient. For example, at its current speed, Central America would take 37 years to achieve universality in basic education and housing. The good news is that all countries have raised their HOI in the last decade and a half, some quite rapidly (the fastest improvement occurred in Mexico). Variations remain wide though, from top- performer Chile (HOI of 95) to Honduras (51). Interestingly, the five countries with the highest HOI—Chile, Uruguay, Mexico, Costa Rica, and Venezuela —have very different development models. 100 40 50 60 70 80 90 Chile Uruguay Mexico Costa Rica Venezuela, R.B de Jamaica Ecuador Source: World Bank 2010 Human Opportunity Report Colombia Brazil LAC Average Dominican Republic Figure 1. The 2010 Human Opportunity Index for LAC Paraguay Panama Peru The 2010 Human Opportunity Index for LAC Guatemala El Salvador Nicaragua Honduras 13 14 Some countries excel at certain services and not at others. For example, Jamaica has the highest educational HOI, but is only mid-table in housing. Even within type of service, issues of quality arise: LAC children have more chances to be enrolled in school than to complete sixth grade on time. Attendance, it seems, is no synonym for learning. Sadly, personal circumstances still matter a lot for Latin American children. Your parents’ level of education will very likely determine yours. And your birthplace is still the most powerful predictor of whether you will have access to basic infrastructure. For all their efforts, LAC governments have, in general, not made much progress improving equity. Only a tenth of the average improvement in HOI is attributable to a fairer allocation of services, that is, to better social targeting of public expenditures. The bulk of the new opportunities opened to the region’s children came from changing circumstances (for instance, migration may have reduced the proportion of rural population). Latin America Versus Rich Countries Using standardized test results from the OECD’s Program for International Student Assessment, and the related demographic data, it is possible to construct a HOI that measures the educational opportunities faced by 15 year-old children around the world. In other words, it is possible to measure how important are those children’s personal circumstances in determining their proficiency in reading, mathematics or science. This sheds an uncomfortable light on LAC. Even the countries with the highest score in the region, Chile and Uruguay, rank well below the worst-performing countries in Europe and North America. Much of the gap is not due to the fact that rich countries just provide more education services, but to the relatively unfair way in which those services are distributed in LAC. If you are a Latin 15 American student, the wealthier your family is, the better your test results. A similar exercise can be performed for housing conditions using census data. Again, LAC has work to do: the opportunity of living in a house with sanitation facilities or free from overcrowding is highly dependent on personal circumstances. In both conditions, only a handful of countries in the region score above the European average. And again, this is due less to larger coverage in Europe than to unfair provision in LAC. Finally, international comparison allows us to peek at how human opportunity could evolve in LAC over the long term. Using a half-century’s worth of relevant data for the US and France, an HOI for housing conditions can be built. It shows a clear pattern: rapid initial growth, followed by a marked slow-down, and virtually stalling right before the point of universal coverage. The lesson is clear: the better you do, the harder it is to make progress. Country, State, City How is human opportunity distributed at the sub-national level? There is enough information to replicate the HOI for some 165 states and cities in LAC, over the past 15 years. The results are telling. First, dispersion is wide among sub-nationals, with Tierra del Fuego at one end (HOI of 96) and the Atlantic region of Nicaragua at the other (29). Second, all capital cities rank higher than the rest of their countries, and that gap is wider the lower the level of the national HOI. Third, convergence appears slow, but lagging geographic areas do improve faster and catch up in providing more opportunities to their local population—a mirror image of the observed evolution of human opportunity among countries. Fourth, the bigger or the less decentralized a country is, the more dispersed 16 its regions’ HOI appear. And fifth, decentralization seems to have been more effective in diminishing regional inequity, but more so in education than in housing. What Can Be Done? LAC remains the most unequal region in the world. The result has been acrimonious political disagreement over the proper role of the state: should it redistribute wealth or protect private property? Where there is no disagreement, however, is over the need to give all Latin Americans the same opportunities, as a matter of social justice or as a call to personal effort. While equality is controversial, equity enjoys support across the political spectrum. While not discussed in the report, the HOI makes it possible to redirect social policy towards equity (where there is consensus) and away from equality (where there is not). How? Many existing social policies and programs are already equity-enhancing. But focusing on equity reveals new points of emphasis along the individual’s life-cycle. Early interventions, from pregnancy monitoring and institutional births to toddlers’ nutrition and neurological development, get a new sense of priority. So do preschool access (such as pre-kindergarten social interaction) and primary school achievement (such as reading standards and critical thinking). The physical security, reproductive education, mentoring, and talent screening in adolescents, all areas that are often overlooked, gain new relevance. A battery of legal and institutional pre-conditions become sine qua non, from birth certificates, voter registration and property titles to the enforcement of anti-discrimination, antitrust, and access-to- information laws. And blanket subsidies that, at the margin, are consumed by those who do not need them (free public college education for the rich, to name one), turn into opportunity- wasting aberrations. If anything else, the quest for equity will lead to a final push in the decade-long process of targeting 17 subsidies, and will spell the end-game for a way of giving out public assistance that was blind to the needs of the recipient—a way that was intrinsically unfair. At the same time, when applied within countries, the HOI is a powerful tool to identify and address regional inequities. Shouldn’t a child-citizen have the same chances in life no matter where in the national territory she is born? Several LAC governments have in recent years implemented mechanisms to equalize service provision across sub-national jurisdictions. Most of those mechanisms are based on regional factors such as poverty levels, efforts at self-taxation, and ownership of natural resources. The question now is whether equal opportunity among children should not be taken into account too. 18 Síntesis 19 20 ¿Qué Oportunidades Tienen Nuestros Hijos? Informe sobre la Oportunidad Humana en América Latina y el Caribe 2010 Síntesis Imagínese un país donde su futuro no dependiera de cuánto ganan sus padres, ni del color de su piel, ni de si Usted es hombre o mujer, ni de dónde nació. Imagínese que sus circunstancias personales, aquellas sobre las que Usted no tiene control ni responsabilidad, fueran irrelevantes para sus oportunidades, y para las de sus hijos. Imagínese ahora una herramienta estadística que ayudara a los gobiernos hacer todo esto realidad. Bienvenido al �ndice de Oportunidades Humanas (IOH). El IOH refleja qué tanto las circunstancias personales (como el lugar de nacimiento, la riqueza familiar, la raza o el género) impactan la probabilidad de que un niño acceda a los servicios necesarios para ser exitoso en la vida, tales como educación oportuna, el agua potable o la conexión eléctrica. El índice se lanzó en el 2008, y su primera aplicación fue sobre los países de América Latina y el Caribe (ALC). Los resultados en ése momento fueron sorprendentes: detrás de la desigualdad que ha por siempre caracterizado la distribución de los resultados del desarrollo en la región (ingreso, acceso a tierra, logros educativos, entre otros), existe una desigualdad aún más preocupante en la distribución de las oportunidades que los niños tienen para desarrollarse. No sólo los logros son desiguales; también lo son las posibilidades de éxito. El problema no es sólo de igualdad; sino también de equidad. El terreno de juego está desnivelado desde el principio. Este libro reporta sobre el estado y la evolución de las Oportunidades Humanas en ALC. Construye sobre el informe del 2008 en varias direcciones. En primer lugar, expande el conjunto 21 de oportunidades y de circunstancias personales al beneficiarse de información estadística recientemente disponible. Los datos son representativos de más de 200 millones de niños y niñas en 19 países durante los últimos 15 años. En segundo lugar, compara las Oportunidades Humanas en ALC con las de países desarrollados, entre ellos Estados Unidos y Francia, dos modelos de política social marcadamente diferentes. Esto permite hacer interesantes paralelismos, así como extrapolaciones de tendencias futuras. Y tercero, se evalúan las Oportunidades Humanas dentro de cada país, analizando regiones, estados y ciudades. Esto nos da una radiografía preliminar de la dimensión geográfica de la equidad, y del rol que juegan las diversas estructuras federativas. El mensaje general que surge del análisis es uno de cauta esperanza. América Latina y el Caribe han avanzado en abrirles a todos la puerta al desarrollo. Pero todavía tienen mucho camino por recorrer. Al ritmo actual tomará, en promedio, una generación para que la región logre universalizar los servicios básicos que se requieren para realizarse en la vida. Desde la perspectiva de equidad, aún nuestros países más avanzados están muy lejos del mundo desarrollado. Las disparidades al interior de los países son también amplias, y apenas convergen. Por fortuna, es mucho lo que nuestros gobiernos pueden hacer. ¿Cómo funciona el IOH? En su interpretación más simple, el IOH mide la tasa de disponibilidad de los servicios que son necesarios para progresar en la vida (como por ejemplo, el agua potable), descontando o “penalizando� la tasa por cuán injusta es la distribución de esos servicios entre la población. Así, dos países que tienen la misma cobertura pueden tener distintos IOH si los ciudadanos que no tienen el servicio son todas mujeres, o todos indígenas, o todos tienen un alto número de hermanos o, en términos generales, comparten una circunstancia personal que no esté bajo su control. En otras palabras, el IOH es la cobertura corregida por la equidad. En teoría, el índice puede aumentar cambiando las circunstancias de las personas (efecto de composición), aumentando el servicio a todos (efecto de escala), o distribuyendo el servicio de una forma más justa (efecto de equidad). 22 El IOH va de cero a 100; una sociedad que ha alcanzado cobertura universal de todos los servicios básicos tiene un puntaje de 100. Para poder hacer comparaciones a través de países y a través del tiempo, el IOH que se presenta para la región en este informe incluye solo los servicios y circunstancias disponibles en todas las encuestas de hogares. Específicamente, se observa el acceso a agua, electricidad y saneamiento, la asistencia escolar y la terminación a tiempo del sexto grado educativo. La literatura empírica demuestra de manera contundente que sin acceso a estos servicios básicos, las posibilidades de llevar una vida productiva son casi inexistentes. Al mismo tiempo, este informe se concentra en siete circunstancias personales: educación de los padres, ingreso familiar, número de hermanos, género del niño, presencia de los padres, género del jefe de hogar, y lugar de residencia. En todos los casos, la unidad de análisis es el niño, definido como un individuo entre 0 y 16 años. Esto permite aislar dudas sobre decisiones y esfuerzo personal–a esa edad, un niño difícilmente pueda considerarse responsable por su destino. Por supuesto que cuando se aplica el IOH a un país específico, la disponibilidad de datos permite incluir un grupo de servicios y circunstancias más amplio, o más sofisticado, como chequeos dentales preventivos, acceso a internet, identificación étnica y ocupación del padre. Esto se ve en el informe, cuando se compara la región con países más desarrollados. ¿Ha mejorado el nivel de oportunidad humana en ALC? Sí, pero con lentas y con marcadas diferencias entre países. Desde 1995, el promedio del IOH regional ha crecido a una tasa de un (1) punto porcentual al año. Esto es claramente insuficiente. A este ritmo por ejemplo, a los países de Centro América les tomará 37 años lograr la universalización de servicios básicos en educación y vivienda. La buena noticia es que todos los países de ALC aumentaron su IOH en los últimos 15 años, algunos bastante rápido (el mejoramiento más acelerado ocurrió en México). Hay mucha variación entre países: 23 desde Chile, con el mejor desempeño (IOH de 95) a Honduras (51). Resulta interesante que los 5 países con el IOH más alto (Chile, Uruguay, México, Costa Rica, y Venezuela) tienen modelos de desarrollo muy diferentes. Figura 1. El �ndice de Oportunidades Humanas para ALC 2010 The 2010 Human Opportunity Index for LAC 100 90 80 70 60 50 40 Peru Brazil Mexico Nicaragua Chile Uruguay Paraguay Jamaica Honduras Dominican Republic Colombia Guatemala LAC Average Costa Rica Ecuador Panama Venezuela, R.B de El Salvador Fuente: Banco Mundial Informe Oportunidad Humana 2010 24 A algunos países les va muy bien en unos servicios pero no en otros; Jamaica por ejemplo, tiene el IOH más alto en educación pero está en una posición promedio en vivienda. Dentro de un determinado servicio, pueden verse temas de calidad: los niños de ALC tienen más posibilidad de estar matriculados en una escuela que de terminar el sexto grado a tiempo. Al parecer, la asistencia no garantiza el aprendizaje. Tristemente, las circunstancias personales todavía importan mucho para los niños de América Latina. El nivel de educación de sus padres muy probablemente determinará el suyo. El lugar donde Usted nació sigue siendo el principal predictor de su acceso a infraestructura básica. A pesar de sus esfuerzos durante la última década, los gobiernos de ALC, en general, no han logrado mejorar significativamente la equidad. Sólo una décima parte del avance promedio del IOH en la región se puede atribuir a una asignación más justa de los servicios, esto es, está relacionado a mejoras en los sistemas de focalización del gasto público social. La gran mayoría de las nuevas oportunidades han surgido principalmente gracias a cambios en las circunstancias personales (por ejemplo, la migración ha reducido la proporción de población rural). América Latina y los países desarrollados Utilizando los resultados de las pruebas estandarizadas del Utilizando los resultados de las pruebas estandarizadas del Programa Internacional de Evaluación del Estudiante de la OECD (PISA, por sus siglas en inglés), y la información demográfica relevante, se puede construir un IOH educativo para los niños y niñas de 15 años en el mundo. En otras palabras, se puede medir qué tan importante son las circunstancias personales de esos niños en sus habilidades de lectura, matemáticas y ciencia. Esta comparación ilumina una verdad incómoda para ALC. Aún los países con mejores puntajes, Chile y Uruguay, se encuentran muy por debajo de los países con los peores puntajes en Europa y en América del Norte. Esta brecha no se debe únicamente a que los países avanzados proveen 25 más servicios educativos, sino también a la forma relativamente injusta en que se distribuyen estos servicios en ALC. En el contexto latinoamericano, cuanta más rica sea la familia, mejores resultados en las pruebas de sus hijos. Un ejercicio similar se puede realizar para los servicios de vivienda, utilizando datos censales. Aquí también ALC tiene mucho por hacer: la oportunidad de vivir en una casa con servicios sanitarios, o libre de hacinamiento, es altamente dependiente de las circunstancias personales. En ambas condiciones sólo un grupo muy reducido de países de ALC logra un puntaje por encima del promedio europeo. De nuevo, esto se debe más a la injusta distribución en ALC que a una mayor cobertura en Europa. Finalmente, las comparaciones internacionales nos permiten analizar cómo podría serla evolución de las oportunidades en ALC en el largo plazo. Utilizando datos de casi medio siglo para Estados Unidos y Francia, se puede construir un IOH para servicios de vivienda. Se observa un patrón claro: crecimiento rápido al inicio, seguido de una marcada desaceleración, y un virtual detenimiento justo antes del punto de cobertura universal. La lección es clara: cuánto más equitativo ya se es (más alto el IOH), más difícil es progresar. A la inversa, los países que hoy se encuentran más atrasados, son los que más rápido pueden avanzar. País, Estado, Ciudad ¿Cómo está distribuida la oportunidad humana a nivel sub- nacional? Existe suficiente información para replicar el IOH en 165 estados y ciudades de ALC, en los últimos 15 años. Los resultados hablan por sí mismos. Primero, la dispersión entre regiones es amplia: desde Tierra del Fuego con un índice de 96 en un extremo, a la costa Atlántica de Nicaragua con un puntaje de 29 en el otro. Segundo, todas las capitales tienen mejores puntajes que el resto de sus respectivos países; la diferencia es aún más grande cuanto más bajo sea el nivel del IOH a nivel nacional. 26 Tercero, hay signos de convergencia, pero es lenta. Las regiones que están más atrasadas tienden a mejorar más rápido y a ponerse al día en términos de proveer oportunidades a sus poblaciones–un espejo de la evolución que se observa al comparar la evolución de la oportunidad humana entre países. Cuarto, entre más grande sea el país o cuanto menos descentralizado esté, mayor la dispersión del IOH entre sus regiones. Y quinto, el proceso de descentralización parece haber sido efectivo para disminuir la inequidad regional, pero más en educación que en vivienda. ¿Qué se puede hacer? ALC sigue siendo la región más desigual del mundo. Esto ha resultado en un agrio e inconcluso debate sobre el rol del Estado: ¿debe redistribuir riqueza o debe proteger la propiedad privada? En lo que no existe desacuerdo es en la necesidad de proveer a todos los latinoamericanos las mismas oportunidades, como una cuestión de justicia social, o como un llamado al esfuerzo personal. Mientras que la igualdad es controversial, la equidad cuenta con apoyo unánime a lo largo del espectro político. Aunque no es materia directa de este informe, el IOH hace posible re-dirigir la política social hacia la equidad (donde hay consenso) y no tanto hacia la igualdad (donde no lo hay). Pero, ¿cómo? Muchas de las políticas y programas sociales ya existentes aumentan la equidad. No obstante, cuando el objetivo es la equidad, se iluminan nuevas áreas de énfasis para la acción pública en cada momento del ciclo de vida del individuo. Intervenciones tempranas, como el monitoreo al embarazo, nacimientos asistidos profesionalmente, nutrición infantil y desarrollo neurológico, toman un nuevo carácter prioritario. Lo mismo pasa con el acceso a educación pre-escolar (interacción social temprana) y con los logros de los estudiantes de escuela primaria (como estándares de lectura y de pensamiento crítico). �reas que suelen descuidarse en los adolescentes, como la seguridad física, la educación reproductiva, los programas de mentores y la identificación de talentos, retoman nueva relevancia. Una batería de servicios legales e institucionales se 27 convierte en requisito sine qua non para los adultos desde los registros de nacimiento e identificación, tarjetas de votante, y títulos de propiedad, hasta el cumplimiento de leyes contra la discriminación y el monopolio y de acceso a la información. La lucha por equidad puede también convertirse en el último empuje al proceso de mejoramiento en la focalización de los subsidios, un proceso que ya lleva más de una década en la región. Dicho de otro modo, permitiría eliminar subsidios que, en el margen, son consumidos por quienes no los necesitan (educación universitaria pública para los ricos, por nombrar uno). Se terminaría así para siempre con la vieja práctica latinoamericana de dar asistencia sin tener en cuenta si el beneficiario la necesita o no–un método intrínsecamente injusto. Al mismo tiempo, cuando se miden las oportunidades dentro de los países, el IOH resulta un instrumento poderoso para identificar inequidades entre regiones/ciudades/estados. ¿No deberían todos los niños de un mismo país, como ciudadanos, tener las mismas posibilidades de éxito en la vida sin importar en que esquina del territorio nacional nacieron? Muchos países de ALC han puesto en marcha mecanismos para igualar la provisión de servicios a través de las jurisdicciones sub-nacionales. Estos mecanismos están basados en factores locales como el nivel de pobreza, el esfuerzo tributario propio, y/o la presencia de recursos naturales. La pregunta ahora es si la igualdad de oportunidades entre los niños no debería también ser un factor para tener en cuenta. 28 1 29 Chapter How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index 30 Chapter 1 How Far Are We From Ensuring Opportunities for How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index All? The Human Opportunity Index Universal access to key goods and services never random, and in many cases is not such as clean water, basic education, egalitarian. An equitable development health services, minimum nutrition and process should seek to ensure that the citizenship rights is a crucial step towards opportunity for children to access these justice and fairness. Expanding access to key goods and services is not correlated these goods and services has long been a with circumstances that are beyond central issue in the analysis of economic their control, such as gender, parental development and in public policy background or ethnicity. The Human discussions, including the Millennium Opportunity Index (HOI), first presented Development Goals initiative. The by Barros et al. (2009), combines both chance people have to pursue the life of coverage rates and equity in a single their choosing involves the opportunity measure. The HOI considers a) how far to access key goods and services, which a country is from the goal of providing constitute human capital investments universal access to a set of goods and that expand each individual’s abilities and services to all, and b) the degree to options. The goal of providing universal which each child in the country has an access to key goods and services is often equal opportunity to access those good included in national development plans, and services. national constitutions, and international Equality of opportunity requires that agreements such as the Universal access to key goods and services not be Declaration of Human Rights. This related to variables we call circumstances. chapter presents a method to measure Circumstances are personal, family a society’s progress as it moves toward or community characteristics that a attaining universal access. child has no control over, and that, for At first glance, one might think that ethical reasons, society wants to be simply measuring coverage rates completely unrelated to a child’s access suffices. But this has a fundamental to basic opportunities. For instance, shortcoming. As a country develops, most societies would agree that the opportunity to access key goods opportunities should not be assigned and services is only partial; they are based on gender, ethnicity, nationality, scarce and can be allocated in many parental background or religion. Instead, different ways. The allocation of goods opportunities should be allocated non- and services within the population is systematically and not be detrimental 31 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index to any particular social group. The HOI advance, although they may or may not measures the coverage rate, and then ultimately achieve this advancement. adjusts it according to how equitably In some cases, having access to one goods and services have been allocated specific good or service is not enough. among circumstance groups. For example, the opportunity to learn This chapter discusses what requires a bundle of goods and services— characterizes basic goods and services, access to a good school might not be and the implications of allocating enough; a child also needs adequate them equitably. We also present the nutrition to have the opportunity to conceptual underpinnings of the HOI. learn. It is a synthetic measure of how far a The HOI focuses on goods and services society is from universal access to a good that constitute investments by people or service, and how equitably access is in themselves—those that improve a distributed across circumstance groups. person’s ability to expand her future We briefly outline the HOI’s properties, production possibility frontier. These and present decompositions illustrating investments have a major impact on how progress can be made by expanding what a person can be or do, affecting average coverage and/or more equitably both market and non-market outcomes. distributing opportunities of access. In this broad sense, investing in these Lastly, we outline a methodology to goods and services increases one’s operationalize these concepts in 19 human capital. Latin American and Caribbean countries Our attention is limited to private goods to assess progress during the last decade and services that expand people’s in universalizing basic opportunities chances of living a better life. They for children. The empirical results are are private in the traditional economic presented in the following chapter. sense of being excludable. As long as the provision of these goods and 1.1. Key Concepts: Basic Goods services entails a cost and there are finite and Services, Universality, Equality resources (i.e., a budget constraint), of Opportunity and Circumstances allocative decisions are required. Given the paramount importance of allocative Having opportunities means that people decisions to economic and social policy, can pursue the life of their choosing. A this study focuses precisely on access to critical aspect of this is having access goods or services that expand chances, to key goods and services that are and not on other dimensions of policy fundamentally important for a person to that might also play that role.1 lead a dignified life in modern society. Access gives a person the opportunity to 32 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index Basic goods and services and not collective. For instance, to set a goal universality of universal access to adequate nutrition does not necessarily imply that everyone Societies may decide that the universal is entitled to receive a monthly food access to selected goods and services basket from the government. Societies should be a major social goal. Goals of may use multiple mechanisms in order this sort are often elucidated in national to achieve universality. Universal development plans and sometimes access to primary education may be national constitutions, and are also ensured through a system of free public laid out in the Universal Declaration of schools, through a privately-managed Human Rights. Whenever a national but publicly-funded system, through consensus exists that some goods and a public school system that recovers services should be enjoyed by everyone, costs from wealthier families only, or we refer to them as basic. Even though through private schools with partial or the set of basic goods and services full scholarships depending on family may vary with the socioeconomic and resources. cultural context, the top priorities seem to be quite similar among all societies. Assessing progress towards To be considered basic, goods and opportunity for all: limitation of the services also need to be affordable— coverage rate as a measure otherwise universal access would not be economically feasible. If universal access to basic goods and A societal goal of universal access services is to be considered a major does not necessarily imply either how development goal, then it is critical universality is to be accomplished or to develop adequate measures of the who is responsible. Even if universal progress towards its accomplishment. access to a basic good or service is Traditionally, the coverage rate—the defined as a social right, it does not proportion of the population with access automatically mean that the public to a given opportunity—has been used sector is responsible for provision or to measure progress. It certainly seems financing. In the extreme, a society may natural to measure progress by the set a goal of universal access even when distance of the coverage rate to its ideal the responsibility is entirely individual, 100 percent. However, measures of 1 It should be noted that increasing opportunities sometimes does not require access to goods and services. For instance, to the extent that international migration represents a chance for progress, the right to migrate may be an opportunity in itself. Migration may have private costs, but if they are substantially outweighed by the benefits, it will be the lack of rights that will deter migration and consequent advancement. Many civil rights represent chances to progress and hence opportunities, without necessarily being associated with the access to any key goods or services. We do not dwell here on these types of opportunities. 33 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index progress should be sensitive to allocation. is only 50 percent. A single aggregated When there are sufficient resources to coverage rate is not enough to track provide something to everybody, there progress toward the ideal of opportunity will be no allocation dilemma. However, to all since it is insensitive to the fairness when available resources only allow for of allocation. providing key goods and services to some, the decision of who enjoys access Equality of opportunity, depends on allocation. In this situation, circumstances and incidence analysis measures of progress towards the ideal of opportunity for all should privilege This report, in the tradition of the World egalitarian allocation. Development Report 2006 “Equity Consider, for instance, two societies and Development� and of Barros et al (I and II) made of two ethnic groups (2009), adopts a notion of fairness that (A and B) of equal population size. is related to equality of opportunities. To Suppose that at the current time, there the extent that basic goods and service are enough resources to give access are scarce and indivisible, some people to a specific service only to half of the will have access to them and others will population. Hence, in both societies the not. According to the principle of equality average coverage rate is 50 percent. of opportunity, everybody should have Suppose, however, that in Society I the the same chance of accessing them, service is allocated to the ethnic group regardless of their circumstances. In the A and none to group B—the coverage example of two societies presented above, rates are 100 percent for group A and 0 incidence analysis—which breaks down percent for group B. On the other hand, coverage by different socioeconomic in Society II both ethnic groups equally and demographic groups—uncover share the limited available services, and differences in coverage rates for each as a consequence the coverage rate ethnic group. For equality of opportunity is 50 percent in both groups. Hence, to prevail, all group-specific coverage even though both societies have the rates must be the same. same average coverage rate, they Circumstances, as used here, are personal, differ remarkably in the allocation of family or community characteristics over their scarce services. In principle, the which an individual has no direct control. allocation rules of Society II are more For ethical reasons, society wants these egalitarian. As a consequence, any valid to be completely unrelated (directly measure ought to indicate that Society or indirectly) to one’s access to basic II is closer than Society I to the ideal of opportunities. Boys and girls should all equitably allocating goods and services, have the same opportunities to access even if the total coverage rate in both good quality education and adequate 34 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index nutrition, irrespective of the education for children to obtain access to basic of their parents, their ethnicity or their nutrition support programs. Thus, even place of birth. That is, when basic though society ideally prefers that family opportunities are limited, they should income not be related to children’s be allocated non-systematically and in access to basic food, it may use income a way not detrimental to any particular transfers as an instrument to reduce group. malnutrition on a transitional basis. The ethical ideal of equal opportunity Similarly, in the long run societies want is intimately related to equal treatment, all children to have access to adequate lack of discrimination, citizenship and nutrition and health care independent personal development independent of of their mothers’ education. However, socioeconomic origin. What exactly since a mother’s education has a critical determines which characteristics are role in providing more opportunities to considered a circumstance is more get adequate nutrition and health care, complex. One either provides an social policies are in many cases designed exhaustive list of all circumstances, or to strengthen this externality. 2 a general rule for identifying whether a Incidence analysis is an improvement characteristic is a circumstance or not. over a single aggregated coverage rate Any set of circumstances as used here is for measuring progress towards leveling subjective or at least relative. Ultimately, the playing field, since it can be sensitive each society chooses their own set of to inequality of opportunity. Incidence circumstances that it believes should not analysis substitutes one coverage rate interfere with access to basic goods and with many—one for each circumstance services. group. In the spirit of incidence analysis, In some cases, circumstances may have a one could say that for equality of role as policy instruments in the provision opportunity to prevail, all group-specific of goods and services, because they are coverage rates must be the same. an efficient mechanism for expanding Incidence analysis, however, is not access. For example, despite the fact enough to measure progress towards that a child should have access to basic opportunity for all, since it does not nutrition regardless of their parent’s provide a synthetic scalar measure of income, social policy analysis might how far a society is from both equality of consider family income a valid instrument opportunity and universal coverage. To 2 Can changing the distribution of circumstances be a valid policy? In the case of circumstances like gender, religion, ethnicity or nationality, society has no interest in changing their distribution to reach universal coverage. But society may, for example want to eliminate the influence of parental income on a child’s education, to reduce the intergenerational transmission of poverty. One strategy to do that is through educational policies. But another could be to implement policies to reduce income inequality. A problem of this strategy is that equalizing opportunities through reducing inequalities in the distribution of circumstances is often impractical or might take too long. 35 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index track hundreds of coverage rates would children, there is a strong argument for be too cumbersome to be useful to both universal coverage and defining equality policymakers and other key stakeholders of opportunity in terms of access and use. in society.3 For this study, we assume that as long The scalar measure of equal opportunity as the focus of analysis is children, progress towards universal access to then access and utilization should be access basic goods and services, first considered the same. A child may have presented in Barros et al (2009), takes access to a school reasonably close to into consideration both (a) average her home, but may not attend school coverage and (b) if available goods or because her parents do not value services are allocated equitably. A scalar education or because the school is of measure of progress towards universality very low quality. In this case, we treat that combines these two features can the child as having no access to school. be called an equality of opportunity- If this is a basic service, society must sensitive coverage rate. ensure that the child uses the service, which might entail not only having a Access, utilization, quality and school nearby, but also maintaining outcomes schools at a level of quality sufficient to convince parents that it pays to send their When measuring the access to specific children to school, educating parents on goods and services, one must be very the benefits (economic and otherwise) careful in defining what access means. of education, or enforcing attendance. Does access to schooling mean having Hence, we consider that coverage should a school nearby? Or having a good be measured as a student enrolling and school nearby? Or attending school? Or attending a formal school. Another having all the conditions needed to have consideration is quality. Basic goods and a productive educational experience? Or service are usually not homogeneous, achieving learning outcomes? One could and quality might vary tremendously. If, easily imagine a situation in which a school for example, clean water is a basic good or a health clinic exists in a community, but or service, it is important to empirically few actually take advantage of it. To the assess what modes of provision provide extent that opportunity is just the chance a minimum threshold of quality. of accessing key goods and services for An alternative view considers coverage 3 When consider even a small number of circumstances, the number of relevant circumstance groups in a particular society can be very large. For example, consider the case of a society with only six relevant circumstances: gender (male-female), race (white-black), location (urban-rural), parental education (less than primary-primary-secondary-tertiary), and per-capita family income (classified in quintiles). In this particular case, with only six circumstances and a very parsimonious breakdown of each, we will have 160 circumstance groups. 36 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index to extend only to those who benefit from (a) the average coverage of a good or the use and access of a basic good and service, and (b) if it is allocated according service above a minimum threshold. It is to an equality of opportunity principle.5 effective access to services of quality that Such scalar measures are fundamental produce a minimum level of outcome. for measuring progress towards the For instance, the best practical measure universal provision of basic goods and of effective access to quality education services. Such a summary measure could could be the proportion of children of a also be essential for improving targeting given age with learning proficiency above of neglected groups and for improving a minimum level. In this view, access is the effectiveness of a social policy aimed just a means to reach minimum levels at universal access to basic goods and of certain outcomes that ought to be services. compulsory.4 It would not be sufficient The literature provides many measures to ensure universal access to schools of of equality of opportunity, such as those quality and to guarantee that all families presented in Bourguignon et al. (2007), have the conditions they need to fully Checchi and Peragine (2005), Barros et take advantage of this opportunity. al (2008), Lefranc et al (2006), among We do not pursue this alternative view others. The main contribution of this due to the lack of comparable outcome study is not only measuring equality of indicators for all countries, as discussed opportunity, but also how to incorporate below. equality of opportunity concerns when evaluating coverage. As such, the HOI 1.2. Constructing a Measure assesses the whole empirical distribution of Progress Towards Basic of the provision of opportunities to access Opportunities for All a specific good or service. It encompasses both the average coverage rate of a basic In this section we introduce and good or service and a relative measure evaluate the properties of the Human of equality of opportunity. Opportunity Index (HOI), a synthetic scalar measure for monitoring both 4 According to this line of reasoning, universal access to opportunities is just an instrument to ensure minimum outcomes for all. What should be evaluated is not the universal access in itself, but its consequence on ensuring minimum merit outcomes. Accordingly, whenever available the best option would be a direct measure of these outcomes (percentage of infants surviving, or the percentage of eight year-olds who are literate). Measures based on outcomes are to a large extent at odds with the notion of opportunity as just a chance to progress. In principle, equality of opportunity should not necessarily lead to equality of outcomes or even to a minimum basic outcome for all. However, some of these outcomes, like a minimum learning threshold, may be considered opportunities, as they proxy access to a minimum bundle of goods and services, that, according to current technology, can produce a minimum learning standard. 5 Strictly speaking, this implies that we will calculate a measure that consists of the average coverage rate of a basic good or service (say, access to water). This will be adjusted by the degree by which access to this service (water) is allocated according to a principle of equality of opportunity. So, in this second step we are concerned with the equality of opportunity of having access to water. 37 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index Constructing the Human Opportunity discounting a penalty for inequality of Index opportunity, P, from the overall coverage rate, C, so that Any equality of opportunity-sensitive coverage rate must take into account HOI = C – P both the overall coverage and the differential coverage rates of the several The penalty is chosen such that it is circumstance groups that make the zero if all circumstance group-specific whole population. The construction coverage rates are equal and is positive of an equality-sensitive coverage rate and increasing as differences in coverage amounts to aggregating circumstance- among circumstance groups increase. specific rates in a scalar measure This penalty makes the HOI sensitive that, at the same time, increases with to equality as well as overall coverage. overall coverage and decreases with Intuitively, P is larger the larger the the differences in coverage among dispersion of group-specific coverage circumstance groups. One could rates. Only when the penalty is zero and imagine a number of alternative ways of average coverage is universal does the constructing an equality of opportunity- HOI reach the maximum value of one sensitive coverage rate having these (see Box 1 for computation details). two properties. The HOI is based on Box 1: Computing the Penalty for Inequality of Opportunity Computing P requires identifying all circumstance groups with coverage rates below the average rate; we refer to them as the opportunity-vulnerable groups. For each opportunity-vulnerable group, k, Mk is the number of people with access to a good or service needed for its coverage rate to equal the average rate, while Mk is the number of people in group k with access. Mk- Mk is then the opportunity gap for the vulnerable group k. The penalty is the sum of the opportunity gaps of all vulnerable groups (called the overall opportunity gap) divided by the total population (N): Intuitively, P can be interpreted as the percentage of people whose access would have to be reassigned to people of the groups with below-average coverage rates to achieve equality of opportunity. If all groups had exactly the same coverage rate, that penalty would be zero, and no reassignment would be needed. As coverage approaches universality for all groups, that reassignment becomes smaller. 38 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index Below we present a graphical explanation specific coverage rate is also 59 percent, of the computation and interpretation meaning this is a situation of equality of of the HOI. The explanation uses data opportunity (Figure 1.1). The average on access to safe water for 10 year- coverage rate line represents the equal old children in a fictitious country (a opportunity line. Even though access is detailed numerical example can be not related to circumstances, the playing found in Annex 1). In the first example, field is not level since 41 percent of the the overall average coverage rate is 59 children do not have access to safe water percent, and each circumstance group while 59 percent do. Figure 1.1 Percentage of 10 year olds with Access to Safe Water - Equal Opportunity Allocation - 100 90 80 Percentage of 10 year olds 70 60 59% Average coverage rate 50 40 30 Equal Opportunity Allocation (Not related to circumstance groups) 20 10 0 0 10 20 30 40 50 60 70 80 90 100 Percentile (circumstance group) Source: Simulations for a fictitious country. In the second situation, 59 percent no equality of opportunity (Figure of children still have access to safe 1.2).6 Those circumstance groups with water and 41 percent do not, but now coverage rates below the overall average the allocation is related to children’s rate are called “opportunity vulnerable� circumstances, and as such there is groups. 6 The horizontal axis depicts circumstance groups ordered according to the group-specific probability of access to water. 39 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index Figure 1.2 Percentage of 10 year olds with Access to Safe Water: - Unequal Opportunity Allocation - 100 90 80 Percentage of 10 year olds 70 60 Average coverage rate 50 Access to water available and allocated based on 40 children's circumstances = Average coverage rate 30 59% 20 Opportunity Vulnerable 10 0 0 10 20 30 40 50 60 70 80 90 100 Percentile (circumstance group) Source: Simulations for a fictitious country. To calculate the HOI for the second percent of access to water was allocated situation, the penalty refers to access to inequitably. The HOI is equal to the safe water that was allocated in violation average coverage rate (59 percent) of the equal opportunity principle minus the penalty for inequality of (Figure 1.3). Every allocation of access opportunity (10 percent): 49 percent. to water to circumstances groups above In other words, the HOI can be thought the overall average is a violation of the of as the weighted average of the equality of opportunity principle, since circumstance group-specific coverage access to safe water is not independent rates for all groups with below-average of circumstances. In this example, 10 coverage. 40 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index Figure 1.3 Penalty for Inequality of Opportunity and the HOI Access to Safe Water 100 90 80 Penalty (P) = Percentage of 10 year olds 70 10% Opportunities that were improperly allocated 60 Average coverage rate 50 Access to water available and allocated based on 40 children's circumstances = Average coverage rate 30 49% 20 Opportunity Vulnerable 10 0 0 10 20 30 40 50 60 70 80 90 100 Percentile (circumstance group) Source: Simulations for a fictitious country. Properties inequality may increase sharply, no coverage rate for any circumstance This section discusses three important group would decline. Hence, there is no properties of the HOI. First, it is defined reason for the overall score to worsen. as an equality-sensitive coverage rate. The HOI is indeed Pareto consistent. As such, its value falls as inequality in Whenever no one loses access and at the allocation of a given fixed number least someone gains access, the index will of opportunities increases. In this case always improve, regardless of whether the opportunity gap may increase that person belongs to an opportunity- (it will never decrease), leading to a vulnerable group or not. corresponding increase in the penalty. Third, when the coverage rates of Second, this equality-sensitive measure is all circumstance groups increase Pareto consistent. In principle, sensitivity proportionally, the HOI will increase by to equality should never be so large that the same proportion. It can be easily the index would decline when no one established that in this case both the loses access but someone that previously coverage and the penalty would also had no access gains access. Even though increase by the same percentage, as 41 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index would the index. In the case of an as a “Dissimilarity Index� to measure equal increase in percentage points dissimilar coverage rates across groups for all group-specific coverage rates, defined by circumstances. This index the index would also increase by the stands for the fraction of people who same percentage points. In this case would need to have a good or service all differences in coverage rates and reassigned as a percentage of all people the penalty would remain unchanged, who have access to the good or service. while the overall coverage—and hence Accordingly, (1-D) would stand for the the index—would increase by the same percentage of available opportunities percentage points. that were properly allocated. It can be Thus, when (i) inequality declines and shown that overall coverage remains constant, or (ii) overall coverage increases while HOI = C - P = C * (1- D) inequality remains constant, the HOI will always improve. So it is in fact a Hence, the HOI can be seen as the average valid inequality-sensitive coverage rate. coverage rate, discounted by one minus Lastly, since the HOI is equal to the the inequality index, D. An alternative difference between the overall coverage interpretation of the Dissimilarity Index rate and the penalty, it is always equal to is that it is proportional to the difference or lower than the coverage rate. Since between group-specific coverage rates the coverage rate is lower than one (i.e., and average coverage rates. The larger under 100 percent), so too is the index. the difference, the larger is D. If all group-specific coverage rates are equal The HOI and the Dissimilarity Index to the average, D=0, and the HOI is equal to the overall average coverage Using a penalty allows us to define rate (C). an equality-sensitive coverage rate without actually measuring the level Decomposing changes in the HOI: of inequality of opportunities. But a composition and coverage effects measure of relative inequality in the allocation of the opportunities, D, The HOI is determined by group-specific could be easily obtained by dividing the coverage rates and their corresponding penalty, P, by the overall coverage rate, population shares (the distribution of C. This measure might be constructed circumstances).7 As a result, the HOI 7 The overall coverage rate C is given by where denotes the population share of circumstance-group and its specific coverage rate. It can be shown that, groups with specific coverage rates below (above) average are over (under) weighted relative to their population share. The HOI can be expressed as where denotes the set of all vulnerable circumstance-groups and, is the population share of non- vulnerable groups. The extent to which specific coverage rates are over- or under-weighted to obtain the HOI depends only on the share of the population in vulnerable groups (circumstance groups with specific coverage rates below average). 42 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index can only change when at least one of opportunity remains large, changes in these two features changes. Hence, the HOI could still come from changes in any change in the index can be traced the distribution of circumstances, known either to changes in the distribution of as the composition effect. Most of the circumstances (composition effect) or to composition effect reflects structural changes in at least some group-specific demographic changes, overall economic coverage rates (coverage effect). development and increased investments The coverage effect can be further in education. In certain cases, reducing decomposed into changes due to changes the share in the population of certain in equality of opportunity (equalization groups could be, at least temporarily, an effect) and changes due to average effective instrument to progress towards coverage rates (scale effect). Below we universal coverage. For instance, if discuss the intuition behind each effect. malnutrition rates among children A numerical example and a box with the from income poor families are hard to algebra of the decompositions are found reduce, an alternative policy could be to in Annexes 2 and 3. decrease the proportion of children in poor families through income transfers. The composition effect The coverage effect: scale and Even though any change in the HOI equalization can always be decomposed into composition and coverage effects, these Progress in coverage can be achieved in two components do not have the same two very distinct ways. One would be importance. The HOI measures progress to increase all group-specific coverage towards the goal of opportunities for all. rates proportionally. In this case, the What matters is how far group-specific degree of equality of opportunity would coverage rates are from the ideal of 100 remain unchanged and the HOI would percent. The distribution of circumstances increase exclusively due to a change in is only used to weight the remaining the average coverage rate. We call this gaps. If equality of opportunity prevails type of change a scale effect. and all group-specific coverage rates On the other hand, progress could be are equal, changes in the distribution achieved by increasing coverage rates of circumstances will have no effect on among vulnerable groups, compensated the HOI. And once all group-specific by a concomitant decrease in coverage coverage rates reach 100 percent, the rates among non-vulnerable groups goal will be reached irrespective of the that would hold the overall coverage distribution of circumstance. rate unchanged. In this case, since the Nevertheless, while inequality of overall rate remains unchanged, the HOI 43 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index increases only due to the decline in the or preferences of the child or parents. degree of inequality of opportunity. We The assumption is that as long as society call this type of change an equalization agrees on universalizing an opportunity, effect. it must ensure utilization by children, All changes in coverage can be expressed independently of the preferences of the as a combination of a scale and an child or her family. equalization effect. Hence, in principle In principle, a set of the most basic goods the coverage effect can always be and services for children is quite large, further decomposed into a scale and an covering a wide range of what is needed equalization effect. for children to develop themselves and pursue a life of their choice. In order to   make cross-country comparisons, we 1.3. Empirical Considerations need comparable information on basic for Constructing the Human goods and services for all countries Opportunity Index considered. The challenge stems from different survey terminologies and The HOI is constructed in three steps. sometimes different national standards First, we must select a specific basic regarding adequate levels of service. good or service to focus on, and define For instance, access to safe water minimum standards to fully characterize must have the same meaning and be access. Second, we must choose a set measured similarly in all countries. To of relevant circumstances. Third, based ensure comparability across countries on microdata from household surveys and over time, a set of five indicators we compute the coverage rate and the was chosen to represent the dimensions penalty for the specific basic good or of education and housing (see Annex 4 service at hand. for more details on the indicators). a) Education Dimension. To Basic Goods and Services Considered capture the effective opportunity to and Minimum Standards quality education we use completion of sixth grade at the proper age (13 The HOI focuses on access to key goods years old). Children completing and services by children 16 years of age sixth grade on time are more likely and under. Independent of the intrinsic to have had access to schools of value of measuring access to key goods reasonable quality that ensure and services by children, focusing on minimum learning and consequently this age range obviates the need to can avoid unnecessary grade make any distinction between access repetition. Some education systems and utilization related to effort, attitudes in the region, however, adopt 44 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index automatic promotion while others supply and inadequate sanitation do not, leading to potential inter- and hygiene (WHO 2002). country comparability problems. To balance the potential comparability (ii) Access to electricity. Electricity due to automatic promotions, we enhances the quality of life in a include school attendance of children number of ways. First, it can promote 10 to 14 years old as an additional a healthier lifestyle, for instance by indicator. improving air quality as electric stoves replace indoor biomass cook stoves. b) Housing Dimension. To This is particularly helpful for young evaluate the opportunity to an children and mothers. Replacing enhanced quality of life we use the kerosene lamps with electricity access to basic housing services: (i) has also been shown to reduce safe water, (ii) adequate sanitation; eye irritation, coughing, and nasal and (iii) electricity. problems, and reduce the substantial number of children who die annually (i) Access to safe water and from accidental kerosene poisoning adequate sanitation. Water and (Kaufman et al. 2000). But perhaps sanitation are primary drivers of more importantly, there are other public health. A vast literature opportunities opened by access to finds a strong negative relationship electricity: improved conditions for between children’s mortality rates studying in the evenings; accessing and improved water sources and information and entertainment via sanitation facilities (Abou-Ali radio, television, and the Internet; Hala 2003; Galiani, Gertler, and freeing parents’ time from domestic Schargrodsky 2005; Fuentes, Pfütze, chores so they could potentially spend and Seck 2006; and Rutstein 2000; that time raising their children; and among others). Improved water home and community safety. Studies and sanitation are linked to reduced have documented that children incidence of diarrhea and related spend more time studying after serious long-term consequences such electricity is provided (Gustavsson as malnutrition and opportunistic 2007); electricity also allows access infections (such as pneumonia), to modern educational techniques and physical or mental stunting. using computing, as in rural Peru Moreover, WHO estimates that (Bajak 2007). every year 1.4 million children under the age of five die from diarrheal diseases attributed to unsafe water 45 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index To ensure inter-country comparability we Choosing Circumstances opt to use the simplest possible concept for adequate access to water, sanitation By circumstances we mean personal, and electricity. Most surveys in the region family or community characteristics that do not ask directly about potable (safe) a society believes should play no role in water, but about the location and type determining access to basic opportunities. of the water source and the system used For instance, most societies would agree for distribution. As a result, we consider that opportunities to access key goods as having access all households with and services should not be based on water from the public network inside the gender, ethnicity, nationality, parental dwelling. For sanitation, we consider as background or religion. To the extent that having access all households that have equal opportunity requires independence flush toilets inside the property that are from socioeconomic origin, parents’ connected to a waste removal system. education and family income should also For electricity, we consider access from be treated as circumstances. Location any source adequate. of residence (urban vs. rural) may also be considered a circumstance, to the The basic goods and services used in extent that a society believes that all this study all vary extensively in quality. children should have equal access to It is clear that, for example, access to the same opportunities independently schooling hides a large variance in the of where they live. This wide scope quality of the service, while frequent of circumstances represents a major blackouts, rationing and diminished challenge to any empirical work, due to wattage hinder the benefits a family data limitations. can draw from access to electricity. Data Moreover, to assess the relative access and comparability limitations performance of different countries make it difficult to gauge quality in requires a set of empirically tractable basic goods and services. At this stage, circumstances and basic goods and for comparability purposes, the analysis services. Unfortunately, such a set— is limited to indicators that measure available for all countries—is limited. quantity and not quality. Further analysis Instead, we must use information at the country level should incorporate collected with reasonably similar the quality dimension, both because methods across countries. For instance, quality of services is a critical area of if we wish to use family income as improvement in all countries, and also a circumstance, we must construct because there are large inequalities of compatible income aggregates for all quality of services across different groups countries. However, some important of the population. circumstances, like ethnicity, have 46 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index distinct categories in different contexts, residence (to capture spatial so they are trickier to use in regional disparities) studies. While in Brazil, Colombia and Panama the distinction between African Computing the Human Opportunity and European descendents may be of Index for Access to a Basic Good or major importance, in Guatemala and Service Bolivia the most important distinction maybe between indigenous people and Given a random sample of the population, European descendents. This type of with information on whether child i had variable is therefore difficult to use in a access to a given basic good or service, study that compares a set of countries. and a vector of variables indicating her For this study, to ensure comparability circumstances, we first use a regression across countries and also over time, a set model to estimate the empirical of seven circumstances was chosen: relationship between each circumstance and access to basic service. We then are 1. Parents´ education (to capture able to predict the probability of access socioeconomic origin) to a basic service for each individual 2. Family per capita income (to with a given set of circumstances as capture availability of resources) well as the overall coverage rate. The 3. Number of siblings (to capture next step is the core of the concept of the dependency ratio) inequality of opportunity, as we seek 4. The presence of both parents (to to derive an overall estimate of the capture family structure) extent of the variation in the coverage 5. Gender of the child (to capture rates of individual children in the target one direct form of discrimination) population compared to the average 6. Gender of the household head coverage. The greater the variation, the (to capture one indirect form of higher the inequality of opportunity and discrimination) the smaller the HOI (Box 2). 7. Urban or rural location of 47 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index Box 2: The Six Steps to Build the Human Opportunity Index 1. Estimate a separable logistic model on whether child i had access to a given basic good or service as a function of his or her circumstances. For education, age was also used to predict the probability of completing each grade. The specification was chosen according to the needs of each circumstance: quadratic for years of education, logarithmic for real income, and categorical for age and the other dimensions. In all cases, the functions are linear in the parameters. From the estimation of this logistic regression, obtain coefficient estimates. 2. Given these coefficient estimates, obtain for each child in the sample the predicted probability of access to the basic good or service in consideration, ˆ i based on the predicted relationship, ,and a vector of their circumstances . p 3. Compute the overall coverage rate C, where or some sampling weights. 4. Compute the Dissimilarity Index D 5. Compute the penalty, P = C * 6. Compute the HOI = C – P 48 How Far Are We From Ensuring Opportunities for All? The Human Opportunity Index Constructing an Overall Human is a simple average of the HOI for access Opportunity Index to water, the HOI for sanitation, and the HOI for electricity.8 To generate a single measure of the The next chapter presents the empirical distribution of opportunities for children, results for the HOI for the Latin America we need to construct an overall synthetic and Caribbean region. It tracks the HOI comprising all basic goods and changes children have faced in accessing services under consideration. The overall opportunities to education and housing HOI, in this study, is a simple average services in the region between 1995 and of the HOI of the two dimensions 2010 and decomposes the contributions considered: education and housing. The of the composition and coverage HOI for education is a simple average of effects, and presents a snapshot of how the HOI for completion of sixth grade on circumstances have affected inequality time and the HOI for school attendance of opportunity for the region’s children for children 10-14. The HOI for housing in the last fifteen years. 8 See Annex 5 for a brief discussion on alternatives for the aggregation sequence for building the overall HOI. 2 49 Chapter The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 50 Chapter 2 The State of Human Opportunities for Children in the The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Latin American and Caribbean Region: 1995-2010 Since the mid-1990s, policy makers in longer on average—37 years—while Latin America and the Caribbean have Andean countries are poised to achieve increased spending on basic social services, universality in 18 years. reflecting the increased priority placed Countries in the LAC region show on these services in poverty reduction significant variation in children’s and development strategies. To what opportunities to access basic services. extent have these efforts translated into In Chile, 95 percent of children have improved opportunities for the children an equal opportunity to access basic of Latin America and the Caribbean services, compared to only 51 percent of (LAC) to access basic social services, the children in Honduras. Chile, Uruguay, regardless of their circumstances? This Mexico, Costa Rica, Venezuela, and chapter reviews the Human Opportunity Argentina all have HOI scores above 85, Index (HOI) for children in the region while four Central American countries over the past 15 years and assesses remain below 60. how effective countries have equitably Overall, countries have been more expanded access to the basic education successful in providing equitably and housing services that a child needs allocated opportunities in the area of to be able to lead a life of her choosing. education than in housing. Just over The results reveal slow but steady four-fifths of the region’s children have progress in the region as a whole, but equal access to basic education services, they also underscore that progress has with the greatest challenges relating to not been uniform and that children in the quality of those services (ability to some countries face significantly higher finish sixth grade on time) compared obstacles. Since 1995, opportunities with access alone (school enrollment). for children in the LAC region have Just over two-thirds of all children expanded by 1 percentage point per have equal opportunities to access year. However, it will take a projected 24 basic housing services, with water and years—an entire generation—to achieve sanitation being the most challenging universal provision of basic education for countries to provide. and housing services in the region, based What are the main drivers behind on the recent pace of progress and 2010 inequality of opportunities for children estimates. If current trends continue, in the region? The results suggest Central American countries will take that among the seven circumstances 51 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 considered, parental education, income vis non-vulnerable groups. Improving and location are the most important in the targeting policies of basic services determining inequality of opportunity. to children in vulnerable circumstance Parental education has the largest groups could lead to a significant rise in effect on inequality of opportunity the HOI. for education, suggesting important This chapter is organized as follows. constraints for inter-generational Section 2.1 characterizes the expansion of mobility. Whether or not a child lives the HOI across the 19 LAC countries and in a rural or urban area, and to a lesser discusses data sources and methodology. extent per capita family income, are the Section 2.2 discusses the current state of most important circumstances affecting the HOI in the LAC region, while Section equality of opportunity for housing. 2.3 focuses on the evolution of the HOI More than half (about 55 percent) of over time and Section 2.4 reviews the the improvement in the LAC HOI in drivers of this evolution. Section 2.5 the last 15 years reflects changes in the outlines the forces behind inequality of seven circumstances of children that opportunity, as an input to policymakers are tracked by the study. That is, more to better target policies in favor of than half of the change in the HOI is excluded circumstance groups. due to the fact that fewer children are in disadvantaged circumstance groups, 2.1. Progress in Improving Human for example because parental education Opportunities in LAC—Although improved, per capita family income Universality Remains a Generation increased, or families migrated from Away rural to urban areas. By contrast, less than half (45 percent) Human opportunity in LAC has expanded of the improvements in the HOI are markedly over the last 15 years. The HOI explained by changes in the likelihood for the 18 countries surveyed grew by an that children with a given set of average of 1 percent per year between circumstance (i.e., residence in rural 1995 and 2010, reflecting improvements areas, with illiterate parents, with four in the overall coverage rate and equity siblings, etc.) will be able to access basic of access, as well as fewer children in services. Of that 45 percent, most of the disadvantaged circumstance groups change reflects improved coverage rates (Table 2.1).1 The HOIs are estimated using for all children, while only 10 percent data from 37 household surveys for 19 arises from a reduction in inequality of LAC countries over a period of more than a opportunity—the relative expansion of decade (circa 1995 and circa 2008) (Annex access to basic services for children in Table A2.1). Together, the surveys represent vulnerable circumstance groups vis-à- more than 200 million children ages 0-16. 52 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 The expansion of coverage rates played exception was Honduras, Guatemala a larger role in improving the HOI, and Nicaragua, where the penalty either but the penalty for unequal access to increased or remained constant as little human opportunities also declined, with progress was achieved in improving its negative impact on the overall HOI access among opportunity-vulnerable decreasing from 11 percent in circa groups relative to non-opportunity- 1995 to 7 percent in circa 2008. The vulnerable groups. 1 Bolivia is an exception, for which only one year of data was used. Consequently estimates of rates of change or of extrapolated values and future projections were not possible for Bolivia. In addition, the varying dates of surveys used for the HOI pose serious comparison challenges. To reduce this comparability problem, we use the two point estimates to extrapolate forward and obtain an estimate for 2010 overall HOI, given the recent level and pace of change of the overall HOI for each particular country in the sample. Similarly, we interpolate all countries to a common base year, 1995. This adjustment permits us to assess countries at a similar point in time. 53 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Table 2.1 Human Opportunity Index, Coverage Rate and Penalties, Circa 1995 and 2008 Circa 1995 and 2008 Human Opportunity Index, Coverage Rate and Penalties, HOI HOI Annual Coverage Rate Coverage Rate Penalty Penalty Country (Circa 1995) (Circa 2010) Change (Circa 1995) (Circa 2010) (Circa 1995) (Circa 2010) Argentina 86.1 88.3 0.22 89 91 3 2 (1998) (2008) Bolivia 69 77 (2007) Brazil 57 76 1.44 66 80 9 5 (1995) (2008) Chile 83 92 0.86 88 94 5 2 (1996) (2006) Colombia 67 79 1.06 77 85 10 6 (1997) (2008) Costa Rica 77 88 0.73 82 91 5 2 (1994) (2009) Dominican Republic 64 73 1.11 71 78 8 6 (2000) (2008) Ecuador 60 76 1.45 68 82 8 6 (1995) (2006) El Salvador 44 53 0.99 54 61 10 8 (1998) (2007) Guatemala 43 51 1.36 51 59 8 8 (2000) (2006) Honduras 42 48 0.83 50 57 8 9 (1999) (2006) Jamaica 79 81 0.29 82 84 3 2 (1996) (2002) Mexico 65 86 1.74 73 90 8 4 (1996) (2008) Nicaragua 35 46 1.61 43 56 8 9 (1998) (2005) Panama 66 69 0.49 75 76 9 8 (1997) (2003) Paraguay 61 71 1.14 65 76 4 5 (1999) (2008) Peru 55 69 1.39 65 76 10 8 (1998) (2008) Uruguay 89 90 0.71 92 93 3 2 (2006) (2008) Venezuela, R.B de 82 87 0.45 86 90 4 3 (1995) (2005) LAC Average 64 73 0.99 71 79 7 5 Source: Author's calculations based on household surveys (Actual survey years in parenthesis) 54 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 A Generation to Universalize Basic all children with a level playing field.2 Services in LAC Central America and the Caribbean will take longer than the regional average— Based on the recent rates of progress 37 and 30 years, respectively—while the and assuming linear expansion, the Southern Cone and Andean nations are region is projected, on average, to take projected to take on average 18 years. 24 years starting in 2010 to universalize Mexico and Brazil will require six and 15 the basic services contained in the years respectively to achieve the goal of overall HOI (Table 2.2), thus providing universal access to basic services. 2 Assuming a linear expansion, we estimate the year of achieving universal coverage by: (100-current HOI)/annual rate of growth. The results are essentially the same if we assume that “universality� is achieved with a coverage rate of 98 percent. With this latter assumption it will take 22 years instead of 24 to universalize the basic services contained in the HOI. The linearity assumption can be also seen as optimistic. If we consider that there is a slowdown in the pace of progress as the HOI approaches universality, as the evidence suggests (see Chapters 3 and 4), it will take much longer to universalize the set of basic services considered. 55 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Table 2.2 Estimated 2010 Overall Human Opportunity Index and Simulated Arrival Date by Sub-region Estimated 2010 Overall Human Opportunity Index and Simulated Arrival Date by Sub-region Simulated Years to Simulated Country Estimated HOI 2010 Rate arrival Arrival Date Andean Countries 81 1.1 18 2028 Bolivia Colombia 81 1.1 Ecuador 82 1.4 Peru 71 1.4 Venezuela 89 0.5 Brazil 79 1.4 15 2025 Caribbean Countries 79 0.7 30 2040 Dominican Republic 75 1.1 Jamaica 84 0.3 Central America 63 1.0 37 2047 Costa Rica 89 0.7 El Salvador 56 1.0 Guatemala 57 1.4 Honduras 51 0.8 Nicaragua 54 1.6 Panama 73 0.5 Mexico 90 1.7 6 2016 Southern Cone 87 0,7 18 2028 Argentina 89 0.2 Chile 95 0.9 Paraguay 73 1.1 Uruguay 92 0.7 LAC Average 77 1.0 24 2034 Source: Author's calculations based on household surveys 56 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 The region will take, on average, almost Southern Cone countries will require a generation—22 years—to universalize 38 years to reach an education HOI of basic education services, based on 100. Looking at the two basic services the 2010 levels and recent growth contained in the education component rates (Table 2.3). Mexico is expected of the HOI—completing sixth grade on to universalize access to education time and attending school for 10-14 opportunities within the next decade, year-olds—suggests that the greatest and the Andean countries will follow challenges are in completing sixth grade by 2023. Central American countries on time, especially in Central American and Brazil will take longer than the countries. LAC average—27 years—while the 57 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Table 2.3 Estimated 2010 Human Opportunity Index for Education and Simulated Arrival Date by Sub-region Estimated 2010 Human Opportunity Index for Education and Simulated Arrival Date by Sub-region 2010 Estimates of HOI for Simulated years to Simulated Country Sixth grade on time School attendance Education Rate arrival Arrival Date Andean Countries 79 93 86 1.0 13 2023 Bolivia Colombia 74 94 84 Ecuador 85 88 87 Peru 79 96 87 Venezuela 79 96 87 Brazil 38 99 68 1.2 27 2037 Caribbean Countries 77 96 87 0.6 22 2032 Dominican Republic 57 96 77 Jamaica 97 96 96 Central American 52 91 71 1.1 27 2037 Costa Rica 67 96 82 El Salvador 47 92 70 Guatemala 30 85 57 Honduras 52 87 70 Nicaragua 41 91 66 Panama 74 93 84 Mexico 90 94 92 1.1 7 2017 Southern Cone 77 95 86 0.4 38 2048 Argentina 82 97 89 Chile 85 99 92 Paraguay 58 92 75 Uruguay 81 94 88 LAC Average 68 94 81 0.9 22 2032 Source: Author's calculations based on household surveys 58 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 TThe LAC region will take, on average, Southern Cone countries are posed to approximately 25 years to universalize achieve universal access by 2022, and the access to the three basic services included Andean countries by 2032. In contrast, it in the housing HOI—access to water, will take the Central American countries sanitation and electricity—based on 2010 almost two generations (48 years) to levels and recent growth rates (Table provide full coverage to all children to 2.4). Brazil and Mexico are expected to basic opportunities in housing, and 36 universalize access to housing services years for the Caribbean countries. within the next decade, while the Table 2.4 Estimated 2010 Human Opportunity Index for Housing and Simulated Arrival Date by Sub-region Estimated 2010 Human Opportunity Index for Housing and Simulated Arrival Date by Sub-region 2010 Estimated HOIs for Simulated years Simulated Country Water Electricity Sanitation Housing Rate to arrival Arrival Date Andean Countries 71 88 66 75 1.1 22 2032 Bolivia Colombia 71 94 69 78 Ecuador 82 94 53 76 Peru 43 67 58 56 Venezuela 87 99 85 90 Brazil 86 99 81 89 1.7 7 2017 Caribbean Countries 43 98 74 72 0.8 36 2046 Dominican Republic 71 98 50 73 Jamaica 16 98 98 71 Central America 51 72 42 55 0.9 48 2058 Costa Rica 95 99 94 96 El Salvador 18 88 19 42 Guatemala 68 73 26 56 Honduras 22 54 21 32 Nicaragua 16 55 58 43 Panama 85 64 33 60 Mexico 88 100 75 88 2.4 5 2015 Southern Cone 89 99 76 88 1.0 11 2021 Argentina 98 100 66 88 Chile 98 100 94 97 Paraguay 69 96 48 71 Uruguay 93 99 97 96 LAC Average 67 88 62 72 1.1 25 2035 Source: Author's calculations based on household surveys 59 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 2.2. Opportunities for Children to Mexico, Costa Rica, Venezuela, and Access Basic Services in the LAC Argentina lead the region in moving Region towards universal access of basic services for their children. For each of these The distribution of human opportunities countries, the estimated 2010 overall estimated for 2010 is highly varied across HOI is higher than 85, meaning more the region. The playing field is almost than 85 percent of the services required level for children in Chile, where 95 for universal coverage are available and percent of basic housing and education allocated equitably. Four countries from services are available and equitably Central American are at the bottom of allocated, whereas in Honduras just the ranking, with HOIs lower than 60: over half (51 percent) of the services Honduras, Nicaragua, El Salvador, and are available and distributed equitably Guatemala. among children. Chile, Uruguay, Figure 2.1: The 2010 Human Opportunity Index for LAC. Chile Uruguay Mexico Costa Rica Venezuela Argentina Jamaica Ecuador Colombia Brazil Dominican Republic Paraguay Panama Peru Guatemala El Salvador Nicaragua Honduras 40 50 60 70 80 90 100 HOI 2010 Source: Author’s calculations based on household surveys 60 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Children in LAC are more likely to have Guatemala in 2010. The leaders of the higher levels of equitably allocated HOI for completing sixth grade on time services in education than housing: the are Jamaica, Mexico, Ecuador and Chile, HOI for education in the region is 81 each with an HOI at or higher than 85. compared to 72 for housing. Moreover, By contrast, El Salvador, Nicaragua, there is a wider range in accessing Guatemala and Brazil have HOIs lower services equitably for housing than for than 50 for this indicator. education across countries in LAC. The housing HOI presents higher The education HOI ranges from a high dispersion than the overall HOI, of 96 for Jamaica to a low of 57 for underscoring the uneven rates of Guatemala, suggesting that children face progress in expanding opportunities almost a level playing field in accessing for quality housing in LAC. The education in Jamaica while only slightly housing HOI is nearly universal in Chile more than half of education services are (97), while it dips to as low as 32 in available and equitably distributed in Honduras, indicating that only one third Guatemala. Eleven out of the 19 countries of the housing services are available analyzed have an education HOI higher and equitably allocated. Four countries than 80. Jamaica, Mexico, and Chile in the region have achieved coverage each have an estimated education HOI rates in housing services adjusted for higher than 90 for 2010. equality of opportunity at or above 90 Comparing the two indicators in the in 2010: Chile, Uruguay, Costa Rica, and education HOI shows that countries Venezuela. Only seven countries out of in the region face more challenges in the 19 considered have a housing HOI equitably ensuring that children complete higher than 80. sixth grade on time than ensuring that Disaggregating the housing HOI reveals all children regardless of circumstances that countries in LAC have been more aged 10 to 14 attend school. While successful in providing children with the average HOI for school enrollment equitable access to electricity than in is 94, the average HOI for completing delivering equitable opportunities for sixth grade on time is only 68. children to live in homes with clean water Similarly, the dispersion of the HOI for and sanitation. While the average HOI completing sixth grade on time is much for electricity is 88, the regional averages higher than with school enrollment. for water and sanitation opportunities While the HOI for completing sixth grade are substantially lower, at 67 and 62, on time ranges from a high HOI of 97 for respectively. Thus, at least one-third Jamaica to a low of 30 for Guatemala, of the region’s children do not have the HOI for school enrollment ranges equitable access to water and sanitation from 99 for Brazil and Chile to 85 for opportunities. 61 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Opportunities to access water, sanitation 2.3 Expanding Human and electricity in the region are also more Opportunities in Latin America and widely dispersed compared to educational the Caribbean: 1995-2010 opportunities. The sanitation HOI ranges from a high of 97 in Uruguay to a low Over the last 15 years, LAC countries of 19 in El Salvador, while the water HOI have expanded children’s opportunities ranges from 98 in Argentina and Chile for basic education and housing services. to 16 for Nicaragua. This means that The overall HOI grew at an average pace in El Salvador and Nicaragua less than of 1 percentage point per year since the one out of five children have an equal mid-1990s (Table 2.5). Overall sanitation opportunity to live in homes with access and sixth grade on time indicators saw to both clean water and sanitation. For the largest gains, with annual HOI electricity, Chile is estimated to have growth rates of 1.3 points each. School achieved universal provision by 2010 enrollment and electricity, both of which while Nicaragua has an electricity HOI have higher levels of HOIs, had smaller of 55, indicating that just over half of growth rates of 0.5 and 1 points per the children have equitably distributed year, respectively. access to electricity. Mexico showed the highest rate of Argentina, Costa Rica, and Chile lead in improvement in the overall HOI, at 1.7 the provision of opportunities to access points per year, compared to a low of water, with HOIs higher than 90. By 0.2 points per year in Argentina. The contrast, Jamaica, Nicaragua, and El five indicators that comprise the overall Salvador have HOIs lower than 20 for HOI show more variation over time, this service. In sanitation, the leaders are with expansion rates as high as 4.5 Uruguay, Chile, Jamaica, and Costa Rica, points annually for the sanitation HOI in with HOIs higher than 90, compared to Nicaragua and 4 points for the water HOI HOIs below 50 for Honduras, Paraguay, in Ecuador and Mexico. The fastest rate Panama, Guatemala, and El Salvador. of expansion of the HOI for electricity In electricity, 12 countries out the 19 was 1.9 points annually in El Salvador. In considered display an HOI higher than education, the fastest expansion of the 90, and no country has electricity HOI HOI for completing sixth grade on time lower than 55. was 2.2 points annually in Peru, and 2.3 points annually for the school enrollment HOI in Mexico. 62 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Table 2.5 Growth Rates by Indicators, Dimensions, and Overall Human Opportunity Index Growth Rates by Indicators, Dimensions, and Overall Human Opportunity Index Sixth grade on School Overall Country time Attendance Education Water Electricity Sanitation Housing HOI Argentina -0.17 -0.02 -0.10 0.32 0.10 1.21 0.54 0.22 Brazil 1.53 0.81 1.17 2.02 1.24 1.86 1.70 1.44 Chile 0.81 0.11 0.46 1.07 0.70 2.01 1.26 0.86 Colombia 1.82 0.61 1.21 0.24 0.63 1.83 0.90 1.06 Costa Rica 0.61 0.75 0.68 0.24 0.51 1.59 0.78 0.73 Dominican Republic 1.87 -0.06 0.91 0.97 1.59 1.41 1.32 1.11 Ecuador 1.35 0.62 0.98 3.98 0.91 0.84 1.91 1.45 El Salvador 1.60 0.92 1.26 0.02 1.92 0.23 0.73 0.99 Guatemala 1.31 1.11 1.21 1.35 1.62 1.55 1.51 1.36 Honduras 1.73 1.30 1.52 0.81 0.57 -0.93 0.15 0.83 Jamaica 0.52 0.10 0.31 -0.86 1.75 -0.09 0.27 0.29 Mexico 1.66 0.60 1.13 4.08 0.74 2.24 2.35 1.74 Nicaragua 1.48 1.24 1.36 0.42 0.73 4.46 1.87 1.61 Panama 0.48 0.32 0.40 0.62 0.79 0.32 0.58 0.49 Paraguay 1.21 0.12 0.67 2.25 1.24 1.13 1.56 1.11 Peru 2.24 0.30 1.27 0.50 1.67 2.36 1.51 1.39 Uruguay 1.40 -0.43 0.48 2.15 0.35 0.33 0.94 0.71 Venezuela 1.13 0.25 0.69 0.07 0.05 0.52 0.21 0.45 LAC Average 1.25 0.5 0.87 1.12 0.95 1.27 1.12 0.99 Source: Author's calculations based on household surveys 2.4. Unpacking Changes in the the circumstance groups. Since the HOI: Scale, Equity and Evolving HOI is completely determined by the Circumstances specific coverage rates and population shares, as discussed in Chapter 1, the Understanding what is behind the HOI can only change when at least one changes in the HOI is important for policy of these elements changes. We refer to makers interested in leveling the playing changes in the HOI due to changes in field for children and ensuring that they the distribution of circumstances as the are equipped to pursue a life of their composition effect. Changes in the HOI choosing. The sources of expansion of associated with changes in the group- the HOI can be classified into two main specific coverage rates are referred to as groups: (i) changes in circumstance the coverage effect. group-specific coverage rates and (ii) changes in population shares among 63 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Both Composition and Coverage gains in group-specific coverage rates Effects Drove Changes in the HOI for housing and education services (Table 2.6). Changes in the average Slightly more than half (55 percent) of circumstances of children in LAC (i.e., the improvement in the HOI can be the composition effect) dominate the explained by changes in the average expansion of all the HOIs considered, circumstances of children in LAC, such as except school enrollment where both increased residence in urban areas and effects are equally important and higher parental education and income electricity where coverage effects are levels. The remaining 45 percent of the stronger. observed expansion in the HOI reflects Table 2.6 Share of Composition Effect in Total Change of Human Opportunity Indices Share of Composition Effect in Total Change of Human Opportunity Indices Sixth grade on School Overall time Attendance Education Water Electricity Sanitation Housing HOI (In Percentage Points per year) Composition Effect 0.7 0.2 0.5 0.7 0.3 1.0 0.6 0.5 Total Change 1.3 0.5 0.9 1.1 1.0 1.3 1.1 1,0 (In Percent) Share of Composition Effect 54 50 53 63 33 75 57 55 Share of Coverage Effect 46 50 47 37 67 25 43 45 Source: Author's calculations based on household surveys The growing access to education education), overall economic growth opportunities and the increased equality (higher parental income) and/or of those opportunities reflects mainly the growing use of income transfer improvements in the circumstances of the programs.3 On the other hand, strong average child, due to past improvements coverage effects may reflect significant in educational opportunities (parental efforts to improve the overall provision 3 Changes in circumstances would not expand a child’s opportunities in a society with completely equal opportunity, since all circumstance groups would have the same opportunities. However, in the context of significant inequality of opportunity, policies aimed at improving certain circumstances, such as family income or parents’ education, may be instrumental in expanding a child’s access to basic goods and services. 64 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 (scale effect) as well as the equitable effect was more prominent; in two allocation (equalization effect) of countries these two effects contributed education services. equally (Brazil and Guatemala); and The composition and coverage in two countries these effects were effects contribute roughly equally to of similar magnitude but in opposite the expansion of the HOI for school directions (Argentina and the Dominican enrollment among countries in the LAC Republic). In El Salvador, circumstance region. In seven countries the coverage groups with lower coverage rates effect had a more prominent role; in increased their population shares, making seven other countries the composition the composition effect negative. Table 2.7 Expansion of Human Opportunity Indices in Education: Contribution of the Composition and Coverage Effects Expansion of Human Opportunity Indices in Education: Contribution of the Composition and Coverage Effects Completing Sixth Grade on Time School attendance (ages 10-14) Total change Composition effect Coverage effect Total change Composition effect Coverage effect Country (p.p) (p.p) (p.p) (p.p) (p.p) (p.p) Argentina -0.2 0.5 -0.7 0.0 0.1 -0.1 Brazil 1.5 1.1 0.5 0.8 0.4 0.4 Chile 0.8 0.6 0.2 0.1 0.1 0.0 Colombia 1.8 0.5 1.3 0.6 0.2 0.4 Costa Rica 0.6 0.6 0.0 0.7 0.3 0.5 Dominican Republic 1.9 1.4 0.4 -0.1 0.2 -0.2 Ecuador 1.3 -0.5 1.8 0.6 0.5 0.2 El Salvador 1.6 -0.4 2.0 0.9 -0.3 1.2 Guatemala 1.3 0.8 0.5 1.1 0.6 0.6 Honduras 1.7 0.6 1.2 1.3 0.2 1.1 Jamaica 0.5 0.2 0.3 0.1 0.0 0.1 Mexico 1.7 1.2 0.4 0.6 0.5 0.1 Nicaragua 1.5 0.5 1.0 1.2 0.4 0.8 Panama 0.5 0.7 -0.2 0.3 0.3 0.0 Paraguay 1.2 1.1 0.1 0.1 0.2 -0.1 Peru 2.2 0.7 1.6 0.3 0.2 0.1 Uruguay 1.4 1.1 0.3 -0.4 0.4 -0.8 Venezuela, R.B de 1.1 1.5 -0.3 0.3 0.2 0.1 LAC Average 1.3 0.7 0.6 0.5 0.2 0.2 Source: Author's calculations based on household surveys 65 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 The composition effect was more HOI in 12 out of 18 countries considered. prominent in the housing HOI (Table Similarly, the composition effect had a 2.8), which is not surprising given the larger contribution than the composition importance of location in determining effect in 11 out of 18 countries in the access to better housing. The composition case of the sanitation HOI, and 12 out of effect had a larger contribution than the 18 countries in the case of the electricity coverage effect on expanding the water HOI. Table 2.8 Expansion of the Human Opportunity Indices in Housing: Contributions by the Composition and Coverage Effect Expansion of the Human Opportunity Indices in Housing: Contributions by the Composition and Coverage Effects Water Sanitation Electricity Composition Coverage effect Composition Coverage Total change Composition Coverage Country Total change (p.p) effect (p.p) (p.p) Total change (p.p) effect (p.p) effect (p.p) (p.p) effect (p.p) effect (p.p) Argentina 0.3 0.4 -0.1 1.2 2.5 -1.3 0.1 0.0 0.0 Brazil 2.0 1.1 0.9 1.9 2.0 -0.2 1.2 1.0 0.2 Chile 1.1 0.8 0.3 2.0 1.4 0.7 0.7 0.5 0.2 Colombia 0.2 0.5 -0.2 1.8 1.4 0.4 0.6 0.4 0.2 Costa Rica 0.2 0.1 0.1 1.6 0.7 0.9 0.5 0.2 0.3 Dominican Republic 1.0 1.2 -0.2 1.4 2.4 -1.0 1.6 1.4 0.2 Ecuador 4.0 -1.9 5.9 0.8 -2.6 3.5 0.9 -3.0 3.9 El Salvador 0.0 -0.9 0.9 0.2 -0.9 1.1 1.9 -0.9 2.9 Guatemala 1.3 0.8 0.5 1.5 1.0 0.6 1.6 1.2 0.4 Honduras 0.8 0.4 0,4 -0.9 0.5 -1.4 0.6 0.5 0.0 Jamaica -0.9 0.2 -1.1 -0.1 0.0 -0.1 1.8 -0.2 2.0 Mexico 4.1 2.6 1.5 2.2 2.5 -0.2 0.7 0.5 0.2 Nicaragua 0.4 0.6 -0.2 4.5 0.4 4.0 0.7 0.8 -0.1 Panama 0.6 0.5 0.1 0.3 0.8 -0.5 0.8 1.0 -0.2 Paraguay 2.5 1.6 0.9 1.1 1.6 -0.5 1.2 0.7 0.6 Peru 0.5 0.7 -0.2 2.4 0.7 1.6 1.7 0.8 0.9 Uruguay 2.1 3.1 -0.9 0.3 1.0 -0.7 0.4 0.5 -0.1 Venezuela, R.B de 0.1 1.0 -1.0 0.5 1.8 -1.3 0.0 0.2 -0.1 LAC Average 1.1 0.7 0.4 1.3 1.0 0.3 1.0 0.3 0.6 Source: Author's calculations based on household surveys The Equalization and Scale Effects whereas the equalization effect captures improved coverage rates specifically The coverage effect—the contribution of for circumstance groups with below- changes in the coverage rates of different average coverage rates vis-à-vis groups circumstance groups—can be further with above-average coverage rates. decomposed into the equalization and The equalization effect is at the heart of scale effects. The scale effect captures equality of opportunities. A society that the impact of proportional change in wants to level the playing field will focus coverage rates for all circumstance groups, on expanding opportunities mainly for 66 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 the vulnerable circumstance groups, and and 18 percent in electricity). Equality the equalization effect is a clear indicator of opportunity could accelerate more of progress toward this goal. quickly if services were better targeted Only 10 percent of overall HOI change to under-served circumstance groups in was due to increased equality of the region. opportunity—the equalization effect— The expansion of the education HOI due in the sample of 18 LAC countries to the coverage effect is on the whole considered during the period covered.4 dominated by the scale effect (Table 2.9). That is, improved targeting of basic That is, improved coverage rates came services to children in vulnerable mainly by increasing education service circumstance groups only accounted for provision for the entire population, not about 10 percent of overall improvement. necessarily more to those who were About 13 percent of the change of the previously under-served. However, there HOI for education is due to increased are eight cases out of 36 (18 countries equality of opportunity (13 percent in for each of the two basic opportunities completing sixth grade on time and 15 considered) where the equalization percent on school attendance for ages effect is the same size as the scale effect, 10-14). Only about 8 percent of the and two cases where the equalization change of the HOI for housing is due effect is a bit bigger than the scale effect to increased equality of opportunity (4 (Guatemala and Paraguay for completion percent in water, 4 percent in sanitation, of sixth grade on time). 4 For more details see Annex Tables A2.2 to A2.4. 67 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Table 2.9 Coverage Effect in Education Human Opportunity Indices: Contributions by the Equalization and Scale Effects Coverage Effect in Education Human Opportunity Indices: Contributions by the Equalization and Scale Effects Completing Sixth Grade on Time School attendance (ages 10-14) Total coverage Equalization effect Scale effect Total coverage Equalization effect Scale effect Country effect (p.p) (p.p) (p.p) effect (p.p) (p.p) (p.p) Argentina -0.7 -0.1 -0.6 -0.1 0.0 -0.1 Brazil 0.5 0.2 0.3 0.4 0.1 0.3 Chile 0.2 0.1 0.1 0.0 0.0 0.0 Colombia 1.3 0.5 0.8 0.4 0.1 0.3 Costa Rica 0.0 0.0 0.0 0.5 0.1 0.4 Dominican Republic 0.4 0.2 0.3 -0.2 0.0 -0.2 Ecuador 1.8 0.4 1.4 0.2 0.1 0.1 El Salvador 2.0 0.4 1.6 1.2 0.2 1.0 Guatemala 0.5 0.3 0.2 0.6 0.1 0.5 Honduras 1.2 0.2 0.9 1.1 0.2 0.8 Jamaica 0.3 0.0 0.3 0.1 0.0 0.1 Mexico 0.4 0.1 0.3 0.1 0.0 0.0 Nicaragua 1.0 0.1 0.9 0.8 0.2 0.6 Panama -0.2 0.0 -0.2 0.0 0.0 0.0 Paraguay 0.1 0.1 0.0 -0.1 0.0 -0.1 Peru 1.6 0.4 1.2 0.1 0.1 0.1 Uruguay 0.3 0.1 0.3 -0.8 -0.2 -0.6 Venezuela, R.B de -0.3 0.0 -0.3 0.1 0.0 0.0 LAC Average 0.6 0.2 0.4 0.2 0.1 0.2 Source: Author's calculations based on household surveys Regarding changes of the housing HOI access to electricity) the equalization driven by the coverage effect, the scale effect dominates: Panama and Paraguay effect also is generally dominant (Table in the case of water and Mexico in 2.10). This indicates that in housing as the case of sanitation. In the case of well, progress has mainly been achieved Mexico, the overall coverage effect has through greater overall coverage rates, been contractive. That is, changes in rather than improved targeting to reach the coverage specific rates, especially children in under-served circumstance reduced coverage among the vulnerable groups. However, in three out of 53 groups, have contributed to reducing cases (18 countries for access to water the HOI for sanitation in Mexico. and sanitation, and 17 countries for 68 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 Table 2.10 Coverage Effect in Housing Human Opportunity Indices: Contributions by the Equalization and Scale Effects Coverage Effect in Housing Human Opportunity Indices: Contributions by the Equalization and Scale Effects Water Sanitation Electricity Total coverage Equalization Scale effect Total coverage Equalization Scale effect Total coverage Equalization Scale effect Country effect (p.p) effect (p.p) (p.p) effect (p.p) effect (p.p) (p.p) effect (p.p) effect (p.p) (p.p) Argentina -0.1 0.0 0,0 -1.3 -0.3 -1.0 0.0 0.0 0.0 Brazil 0.9 0.3 0.5 -0.2 0.0 -0.2 0.2 0.1 0.1 Chile 0.3 0.1 0.2 0.7 0.2 0.4 0.2 0.1 0.1 Colombia -0.2 0.1 -0.3 0.4 0.1 0.3 0.2 0.1 0.1 Costa Rica 0.1 0.0 0.1 0.9 0.3 0.6 0.3 0.1 0.2 Dominican Republic -0,2 0.0 -0.2 -1.0 -0.3 -0.7 0.2 0.1 0.2 Ecuador 5.9 0.1 5.7 3.5 0.3 3.2 3.9 1.0 2.8 El Salvador 0.9 0.0 0.8 1.1 0.1 1.0 2.9 0.9 2.0 Guatemala 0.5 0.1 0.4 0.6 0.2 0.3 0.4 0.2 0.2 Honduras 0.4 0.2 0.2 -1.4 -0.6 -0.8 0.0 0.0 0.0 Jamaica -1.1 -0.4 -0.7 -0.1 0.0 -0.1 2.0 0.3 1.7 Mexico 1.5 0.5 1.0 -0.2 -0.2 0.0 0.2 0.1 0.1 Nicaragua -0.2 -0.1 -0.1 4.0 0.7 3.4 -0.1 -0.1 0.1 Panama 0.1 0.2 -0.1 -0.5 -0.1 -0.4 -0.2 0.0 -0.2 Paraguay 0.9 -0.6 1.6 -0.5 -0,2 -0.3 0.6 0.2 0.4 Peru -0.2 0.0 -0.2 1.6 0.6 1.0 0.9 0.4 0.5 Uruguay -0.9 -0.4 -0.5 -0.7 -0.2 -0.5 -0.1 -0.1 -0.1 Venezuela, R.B de -1.0 -0.2 -0.7 -1.3 -0.4 -0.9 -0.1 0.0 -0.1 LAC Average 0.4 0.0 0.4 0.3 0.0 0.3 0.6 0.2 0.5 Source: Author's calculations based on household surveys In summary, only rarely has the To level the playing field for all children, equalization effect played a more policy makers need to know the equality prominent role within the coverage of opportunity profile for a given society effect. The default situation seems to be to design effective public policies for a dominance of the scale effect. In only accelerating the equitable expansion four cases (out of 89 considered) does of human opportunities. This section the equalization effect dominate and analyzes the main circumstances have a positive effect. This indicates that affecting equality of opportunity for Latin American countries could make access to a basic service, and the relative far more effective use of their resources effect on this opportunity of a specific to provide basic opportunities to their circumstance—such as gender, where children by improving the targeting of a child lives, or their parent’s income— social service provision to those most in compared to other circumstances. need. The equality of opportunity measure D—the methodology for which is 2.5 The Inequality of Opportunity explained in Chapter 1—is a synthetic Profile measure that aggregates the differences in coverage among all groups arising 69 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 from a defined set of circumstances.5 In equality of opportunity. addition, it is also possible to measure the A specific D-index can be computed equality of opportunity associated with for each of the seven circumstances for only one specific circumstance. For policy each of the basic goods and services design, it may be important to analyze considered, averaged across the LAC how each circumstance contributes region (Table 2.11; country results are to overall inequality of opportunity. reported in Annex Tables A.2.5–A2.9). Moreover, a constant level of overall These numbers represent the proportion equality of opportunity over time may of the available basic good or service that hide important changes. For example, would have to be redistributed among equality of opportunity in education children for equality of opportunity to resulting from urban or rural location prevail, if only one circumstance was may be increasing, while inequality of considered. For example, for access to opportunity in education resulting from water, the average D-indices calculated differences in a parent’s education may for each circumstance range from 0.3 be declining. percent for gender of the child to 7 To compute the synthetic D-index, percent for area of residence. Hence, all circumstances are considered if the only circumstance considered is simultaneously. An equality of area of residence, 7 percent of available opportunity profile can also be defined water connections need to be reallocated by computing a specific D-index for to eliminate the differences in access to each circumstance (gender, parent’s water across different groups. When education, and so forth), and then considering a child’s gender, only 0.3 comparing them to identify which specific percent of available water connections circumstances elicit larger inequality in a need to be reallocated in LAC to eliminate given basic good or service. There are the differences in access to water. complementary ways of reporting an The inequality of opportunity profile for equality of opportunity profile for the education shows that in LAC, parental LAC region: (i) a profile based on the education and income continue to average D-indices in the region, and (ii) a influence whether or not a child has fair profile based on the number of countries access to education opportunities. In where one specific circumstance is more short, parental characteristics affect the important in characterizing existing ability of a child to improve her situation 5 The equality of opportunity measure, D, is used to estimate the penalty that discounts the overall coverage rate. As discussed in Chapter 1, the penalty is the product of the inequality of opportunity measure and the overall coverage rate (P=C*D). 70 The State of Human Opportunities for Children in the Latin American and Caribbean Region: 1995-2010 over time and achieve inter-generational A2.5). mobility. For completing sixth grade on For school enrollment for children aged time, the most important circumstance 10-14, the inequality of opportunity in LAC countries is parental education, profile is also driven mainly by parental and to a lesser extent the gender of education. The profile based on number the child and number of siblings. A of countries shows that parental complementary profile, based on the education dominates the rankings in 15 number of countries where one specific countries. The presence of two parents D-index dominates, confirms these in a child’s household tops the rankings findings. Parental education dominates in two countries, while per capita family the rankings in 17 out 18 countries; income dominates in one country (Annex while the number of siblings dominates Table A2.6). the rankings in one country (Annex Table Table 2.11 D-Index by Circumstance and Opportunity, Circa 2010 D-Index by Circumstance and Opportunity, Circa 2010 Circumstances Sixth Grade on Time School Attendance Water Sanitation Electricity Parent's education 5.2 1.2 4.3 7.5 1.6 Gender 2.4 0.3 0.4 0.2 0.1 Gender of Household Head 1.0 0.3 1.2 1.5 0.5 Per Capita Income 2.2 0.4 5.8 8.8 2.1 Urban or Rural 2.0 0.4 10.8 13.6 4.2 Presence of Parents 1.0 0.4 1.5 1.7 0.2 Number of Siblings 2.5 0.2 1.2 1.6 0.4 Source: Author's calculations based on household surveys For access to water, sanitation, and location circumstance dominates the electricity, the inequality of opportunity rankings in 13 countries (out of 18) in the profile is driven mainly by where a child case of water, 12 countries in sanitation, lives (rural vs. urban residence), and to and 16 countries (out of 17, Argentina lesser extent by per capita family income. In excluded) in the case of electricity (Annex the profile based on number of countries, Tables A2.7-A2.9). 3 71 Chapter Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World 72 Chapter 3 Human Opportunities in a Global Context: Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Benchmarking LAC to Other Regions of the World The previous chapter showed that the in housing indicators, with HOIs for Latin America and the Caribbean (LAC) access to sanitation and freedom from region has made progress in improving overcrowding exceeding the European equality of opportunities for children to average in two and three LAC countries, access basic goods and services, but that respectively. As with education, much there still is an important challenge to of the housing HOI gap is attributed to achieve universality. To build consensus inequality of opportunity levels often around the agenda ahead, in this chapter twice as high in LAC than European and we compare Human Opportunity Indices North American countries. (HOIs) in educational achievement In the case of two countries (United States and housing between LAC and other and France), we exploit the availability countries in North America and Europe. of a long time series to draw some How much progress does LAC need to lessons on the evolution of the HOI over make to reach OECD countries minimum multiple decades that could be relevant standards? What are the main sources of to LAC. Analyzing the expansion of the the differences between LAC and OECD HOI for housing opportunities in the US countries? Are the observed differences and France from the 1960s indicates that similar across education and housing HOI levels are initially low, experience dimensions? fast growth rates, and then slow down The chapter finds that despite gains as the HOI reaches a high level. in recent years, the gap between LAC The chapter is organized as follows. and OECD countries in Europe and Section 3.1 describes the data North America remains large in both underlying the construction of the education and housing HOIs. Education HOI for educational achievement, and quality outcomes were notably worse presents results. Section 3.2 describes for LAC, with even the best country the data used in constructing HOIs for scores falling below the worst scores access to sanitation and freedom from of Europe, Canada and the US. Much overcrowding, and presents results. of the education HOI gap is attributed Section 3.3 selects the longest time to inequality of opportunity levels horizon available for two countries—the often two or three times higher in LAC US and France—and describes how the than European and North American HOI evolves over long time periods. countries. LAC fared somewhat better Section 3.4 concludes. 73 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World 3.1 Human Opportunity Index for survey of the knowledge and skills of Quality Education 15 year-olds, originally created by the governments of OECD countries (Box Educational achievement measured 3.1). Students are tested in reading, by test scores better captures the mathematics and science. Reading true chances of children to meet the literacy is measured in terms of students’ challenges of the future than educational abilities to use written information in attainment measured by years of situations that they encounter in their schooling or even completion on time, lives. Mathematical literacy measures because of considerable heterogeneity students’ abilities to pose, solve and in the quality of education across interpret mathematical problems schools. To some extent, educational in a variety of situations involving achievement is a good proxy for the quantitative, spatial, probabilistic or quality of schools. Therefore, to estimate other mathematical concepts. Scientific the HOIs for quality education, we use literacy measures students’ abilities to data on educational achievement. identify, explain and apply scientific The OECD Programme for International knowledge and knowledge about science Student Assessment (PISA) is an in a variety of complex life situations. internationally comparable dataset on key Students are placed in different proficiency competencies of 15 year-old students in groups according to the difficulty of reading, mathematics, and science. PISA tasks that they can complete. There assesses the degree to which students are six proficiency groups for reading, near the end of compulsory education and seven groups for mathematics and can extrapolate from what they have science.1 Proficiency level 2 is usually learned and apply their knowledge both considered the level that requires the in school and non-school contexts, thus basic tasks students need to apply the giving an indication of how well they subject area in real life contexts. About have gained the skills and knowledge 80 percent of students in OECD countries needed for full participation in society are at level 2 or above. Longitudinal (OECD 2007). follow-up studies in Australia, Canada and Denmark find that the minority of Estimating the HOI for Quality students classified either at level 1 or Education below are very likely to face difficulties using reading materials to fulfill their The HOI for quality education is built goals and to acquire knowledge (OECD using PISA data. PISA is a triennial 2007:46). The average for students in 1 The groups are below level 1, and levels 1 to 5 for reading, and levels 1 to 6 for science and mathematics. 74 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World OECD countries is level 3. occupation, and household asset items For the HOI for quality education, that allows us to generate a wealth we focus on whether or not a student index (see Annex Table A.3.1 for more achieves a minimum score to place details). herself at proficiency level 2. Hence, We compute the HOI for quality the coverage rate of quality education education by estimating a logistic model used in the HOI is the proportion of on whether student i had achieved at students that took the test and achieve least proficiency level 2 as a function of a score that place the student at least at her or his circumstances. Based on the proficiency level 2. predicted probabilities, we compute the PISA surveys contain information on a coverage rate, the dissimilarity index, common set of six circumstances: gender the penalty for inequality of opportunity of the child, father’s and mother’s and the HOI, following the methodology education, school location, father’s described in Chapter 1. Box 3.1: The PISA Data The first PISA survey was conducted in 2000, and in 2006 the third survey included 30 OECD and 27 non-OECD countries. PISA surveys are administered in countries that together make up close to 90 percent of the world economy.2 The samples of students are nationally representative of the populations of 15 year-olds attending schools in grade 7 and above (Table 3.1). In 2006, the samples used were representative of 20 million 15 year-olds. More than 400,000 students in 57 countries took a two-hour comparable test for PISA 2006. Students also completed a questionnaire about themselves, and their principals completed a questionnaire about their schools. The samples do not cover drop outs and students attending grades below 7. While the enrollment of 15 year-olds is generally universal in Europe, it is not so in LAC. With the exception of Portugal, all European countries considered have enrollment rates of 98 percent and above for 15 year-olds. By contrast, the proportion of 15 year-olds enrolled in the school system is as low as 61 and 63 percent in Colombia and Mexico.3 To the extent that 15 year-olds not enrolled in schools would not be able to achieve minimum expected knowledge, a low proportion of enrollment may over- estimate a national measure of the educational achievement of 15 year-olds. 2 In 2009, PISA was administered in 30 OECD countries and 37 non-OECD countries/economies. Hong Kong-China is included. 3 The proportion of enrolled 15 year-olds that are below 7th grade (not targeted population) is negligible in all countries, except France. 75 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Table 3.1 Sampling and Coverage Rates (PISA 2006) Sampling and Coverage Rates (PISA 2006) All 15- Enrolled Target year-olds 15 year- Ratio population Ratio Coverage Country (a) olds (b) (b)/(a) (c) (c)/(b) Participants Index Latin America Argentina 662,686 579,222 0.87 579,222 1.00 4,339 0.99 Brazil 3,390,471 2,374,044 0.70 2,357,355 0.99 9,295 0.99 Chile 299,426 255,459 0.85 255,393 1.00 5,235 0.99 Colombia 897,477 543,630 0.61 543,630 1.00 4.478 0.99 Mexico 2,200,916 1,383,364 0.63 1,383,364 1.00 30,971 1.00 Uruguay 52,119 40,815 0.78 40,815 1.00 4,839 1.00 Europe France 809,375 809,375 1.00 777,194 0.96 4,716 0.91 Germany 951,535 1,062,920 1.12 1,062,920 1.00 4,891 0.99 Italy 578,131 639,971 1.11 639,971 1.00 21,773 0.98 Norway 61,708 61,449 1.00 61,373 1.00 4,692 0.96 Portugal 115,426 100,816 0.87 100,816 1.00 5,109 0.98 Spain 439,415 436,885 0.99 436,885 1.00 19,604 0.96 Sweden 129,734 127,036 0.98 127,036 1.00 4,443 0.96 United Kingdom 779,076 767,248 0.98 767,248 1.00 13,252 0.97 North America Canada 426,967 428,876 1.00 424,238 0.99 22,646 0.93 USA 4,192,939 4,192,939 1.00 4,192,939 1.00 5,611 0.96 Source: PISA 2006 76 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World The Human Opportunity Index for reading proficiency ranges from a high Reading Literacy of 89 for Canada to a low of 35 for Argentina. Among the six Latin American The Human Opportunity Indices (HOI) countries included in PISA 2006, Chile for reading literacy in LAC countries performs best, with a HOI for reading of are consistently lower than in European 58. However, even Chile is considerably countries and Canada, according to below the lowest performing European findings from 16 countries from the country in the sample, Italy, with an HOI 2006 PISA (Table 3.2).4 The HOI for of 69. 4 Data for reading in the USA PISA 2006 is not available due to an error in the printed test booklet that distorted the results. 77 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Table 3.2 Human Opportunity Index for Reading 2006 Human Opportunity Index for Reading, 2006 Coverage D-index Penalty HOI PISA 2006 2006 2006 (percent) LATIN AMERICAN COUNTRIES Argentina 44 22 10 35 Brazil 45 20 9 36 Colombia 46 17 8 38 Mexico 53 18 10 44 Uruguay 56 15 9 47 Chile 66 12 8 58 EUROPEAN COUNTRIES Italy 75 8 6 69 Spain 76 8 6 70 Portugal 77 9 7 70 France 82 7 6 76 Norway 82 7 5 76 Germany 85 7 6 80 The U.K 87 5 4 82 Sweden 88 5 4 83 NORTH AMERICAN COUNTRIES Canada 91 3 3 89 USA nd nd nd nd Source: Estimates produced based on PISA data, 2000 - 2006. Canada and France don't include school location as exogenous variable 78 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World The HOI discounts the coverage rate A profile of inequality of opportunity with a penalty that is proportional to the can be defined by the relative size of degree of inequality of opportunity in the each D-index to a specific circumstance allocation of existing basic services. The (parental income, gender, etc.).5 inequality of opportunity for acquiring These specific D-indices represent the adequate reading in LAC countries, as percentage of the available opportunity measured by the D-indices, is about twice for adequate reading ability that would the magnitude observed in European and have to be reallocated among children North American countries (Table 3.2). To for equality of opportunity to prevail, if reduce inequality of opportunity more only one circumstance were considered. effectively, policy makers need to know For instance, if we only considered the inequality of opportunity profile for gender of child, roughly 13 percent of a given society to design effective public available opportunities for accessing policies. adequate reading in Argentina would To build this profile, we report the specific need to be reallocated, compared to 3.8 D-indices that inform us about the percent of available opportunities if we inequality of opportunity associated with considered only father’s education. each specific circumstance (Table 3.3). 5 It is important to remember that the specific D-indices do not add up to the overall D-index. That is, this exercise does not have additive properties. 79 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Table 3.3 Profile of Inequality of Opportunity: Specific D-Indices for Proficiency at Level 2 in Reading (PISA 2006) D-Index for Proficiency at level 2 in reading, PISA 2006 (percent) Socio-Economic Country Gender School location Father education Mother education Status Father ocuppation Overall D-Index LATIN AMERICAN COUNTRIES Argentina 12.9 5.4 3.8 4.6 13.3 9.2 22.2 Brazil 9.3 4.0 5.0 4.3 11.4 7.8 20.2 Chile 3.8 3.7 2.0 3.4 5.3 3.4 12.2 Colombia 3.3 6.3 3.5 4.2 9.6 4.6 17.1 Mexico 7.7 8.9 3.1 4.2 5.1 3.4 18.0 Uruguay 8.6 2.8 2.4 5.0 7.2 4.9 15.3 EUROPE France 2.3 0.5 1.1 3.8 3.1 7.3 Germany 2.2 1.0 1.1 1.8 2.5 2.2 6.6 Italy 4.8 2.9 1.6 1.6 4.2 2.3 8.0 Norway 3.9 0.4 0.6 0.9 3.6 3.3 6.6 Portugal 2.8 1.3 1.5 2.6 4.5 3.8 9.1 Spain 3.4 1.0 1.1 2.1 3.7 3.5 8.1 Sweden 2.4 0.5 1.1 0.5 2.2 2.1 4.7 The U.K. 1.9 0.7 0.9 0.4 2.5 2.9 5.0 NORTH AMERICA Canada 1.6 0.3 0.2 1.7 1.0 2.9 Average 4.7 3.0 1.9 2.5 5.4 3.8 10.9 Source: Author's calculations based on PISA data Overall, wealth status is the most America: only 1.7 percent of the available important circumstance associated with opportunities for accessing quality inequality of opportunity for reading in reading would need to be reallocated Canada, Latin America, and European in Canada for equality of opportunity countries. It is the most important to prevail if the only circumstance considered was wealth status, compared circumstance in four out of six LAC to 13 percent in Argentina. Gender of countries considered, in four out of the child is the second most important eight European countries considered, circumstance, and school location is and in Canada. However, the weight of also important in some LAC countries, the circumstance is much higher in Latin notably Mexico and Colombia. 80 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World The Human Opportunity Index for Among the six Latin American countries Mathematical Literacy included in PISA 2006, Uruguay is the best-performing country, with an HOI The HOIs for mathematical literacy in for mathematics of 47. However, even LAC countries are substantially lower Uruguay is considerably below the lowest than in European and North American performing European country in our countries (Table 3.4). The HOI for sample, Italy (63), as well as compared mathematics ranges from a high of 90 to the US (69), the low performer in for Canada to a low of 18 for Argentina. North America. 81 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Table 3.4 Human Opportunity Index for Mathematics (PISA 2006) Human Opportunity Index for Mathematics, 2006 Coverage D-index Penalty HOI Math 2006 2006 2006 2006 LATIN AMERICAN COUNTRIES Brazil 27 32 9 18 Colombia 28 27 8 20 Argentina 37 25 9 28 Mexico 43 20 9 34 Chile 45 23 11 35 Uruguay 56 15 9 47 EUROPEAN COUNTRIES Italy 69 9 6 63 Portugal 71 10 7 64 Spain 77 7 6 72 France 82 7 6 76 Norway 81 6 4 77 Sweden 84 5 5 80 Germany 85 6 5 80 The U.K 86 5 5 81 NORTH AMERICAN COUNTRIES Canada 92 2 2 90 USA 75 8 6 69 Source: Estimates produced based on PISA data, 2000 - 2006. Canada and France don't include school location as exogenous variable 82 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Wealth status is the circumstance most most important circumstance in five out strongly associated with inequality of six LAC countries considered and in of opportunity for accessing quality seven out of eight European countries education in mathematics in Latin considered. Father’s occupation is the America and Europe (Table 3.5). It is the second most important circumstance. Table 3.5 Profile of Inequality of Opportunity: Specific D-Indices for Proficiency at level 2 in Mathematics (PISA 2006) D-Index for Proficiency at level 2 in mathematics, PISA 2006 (percent) Mother Socio-Economic Father Country Gender School location Father education education Status ocuppation Overall D-Index LATIN AMERICAN COUNTRIES Argentina 3.2 7.6 6.0 5.6 17.3 8.5 25.5 Brazil 6.8 5.4 8.2 9.2 19.7 11.1 31.7 Chile 7.2 4.9 4.8 8.7 9.8 8.4 23.2 Colombia 11.4 6.2 5.6 6.4 16.5 10.8 27.4 Mexico 3.6 10.8 2.7 7.3 6.1 4.2 20.4 Uruguay 1.7 3.5 3.2 6.0 7.8 4.5 15.4 EUROPE France 0.5 1.4 1.8 4.0 3.2 7.4 Germany 1.1 0.9 0.9 1.6 3.2 2.5 6.2 Italy 2.2 2.8 1.3 1.6 5.4 3.2 8.8 Norway 0.6 0.7 0.9 0.9 3.6 3.3 5.5 Portugal 1.5 1.8 3.2 2.7 5.4 5.3 10.2 Spain 0.4 0.5 0.7 2.1 4.0 3.5 7.2 Sweden 0.4 0.4 1.7 0.8 2.4 3.4 5.4 The U.K. 0.8 0.9 1.4 0.6 3.1 2.7 5.3 NORTH AMERICA USA 0.9 2.9 0.4 1,2 3.3 5.0 8.4 Canada 0.3 0.2 0.2 1.4 1.3 2.1 Average 2.7 3.5 2.7 3.6 7.1 5.0 13.1 Source: Author's calculations based on PISA data 83 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World The Human Opportunity Index for six Latin American countries included in Science Literacy PISA 2006, Chile is the best-performing country with a HOI for science of 53. The HOIs for science literacy in LAC However, even Chile is considerably countries are also substantially lower below the lowest performing European than in European and North American countries in our sample, Italy and countries (Table 3.6). Scores range from Portugal, with HOIs of 71. 90 for Canada to 30 for Brazil. Among the Table 3.6 Human Opportunity Index for Science (PISA 2006) Human Opportunity Index for Science, 2006 Coverage D-index Penalty PISA 2006 2006 2006 2006 (percent) LATIN AMERICAN COUNTRIES Brazil 40 24 9 30 Colombia 40 18 7 33 Argentina 46 22 10 36 Mexico 49 19 9 40 Uruguay 60 14 8 51 Chile 62 14 9 53 EUROPEAN COUNTRIES Italy 76 7 6 71 Portugal 78 8 6 71 France 83 7 6 77 Spain 82 6 5 78 Norway 83 5 4 78 Sweden 86 4 4 83 Germany 88 5 4 84 The U.K 88 4 4 84 NORTH AMERICAN COUNTRIES USA 79 8 6 73 Canada 92 2 2 90 Source: Estimates produced based on PISA data, 2000 - 2006. Canada and France don't include school location as exogenous variable 84 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Wealth status, again, is the circumstance important circumstance in five out of most strongly associated with six LAC countries considered. Father’s inequality of opportunity for accessing occupation is the second most important quality education in science in Latin circumstance. America (Table 3.7). It is the most Table 3.7 Profile of Inequality of Opportunity: Specific D-Indices for Proficiency at level 2 in Science (PISA 2006) D-Index for Proficiency at level 2 in science, PISA 2006 (percent) Socio-Economic Country Gender School location Father education Mother education Status Father ocuppation Overall D-Index LATIN AMERICAN COUNTRIES Argentina 3.3 6.3 4.6 6.5 13.9 8.1 22.0 Brazil 2.6 5.3 4.5 7.2 14.8 9.3 23.7 Chile 2.8 3.0 3.4 4.5 5.5 3.6 13.8 Colombia 4.1 4.3 2.7 4.1 10.6 6.9 18.0 Mexico 1.9 10.5 3.0 5.0 5.8 4.1 19.0 Uruguay 2.3 2.3 3.9 4.2 6.5 4.1 13.8 EUROPE France 0.0 0.9 1.4 3.7 3.1 7.1 Germany 0.4 0.9 0.7 1.3 1.9 1.9 5.0 Italy 0.4 2.4 1.6 1.4 4.5 2.5 7.3 Norway 0.5 1.0 0.5 1.0 3.4 2.9 5.1 Portugal 0.2 0.7 2.0 2.5 4.0 4.3 8.0 Spain 0.2 0.6 0.7 1.8 2.6 2.6 5.7 Sweden 0.3 0.3 1.3 0.9 1.7 2.8 4.2 The U.K. 0.0 0.5 1.0 0.5 2.4 2.4 4.5 NORTH AMERICA USA 0.8 2.9 0.5 0.9 3.0 4.3 7.6 Canada 0.2 0.4 0.3 1.3 1.1 2.2 Average 1.3 2.9 2.1 2.9 5.6 4.2 11.0 Source: Author's calculations based on PISA data 85 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World 3.2 Human Opportunity Index for to monitor the proportion of households Housing living in overcrowded conditions (US). Due to the growing awareness of In this section we examine how LAC its importance, this study compares countries compare to other regions freedom from severe overcrowding around the world on access to sanitation among different countries. and the degree of overcrowding within the home. While the importance of The Data access to sanitation has been underscored in Chapter 1, the importance of For non-LAC countries, we use census overcrowded housing bears discussion micro-samples from the Integrated here. Public Use Microdata Series (IPUMS) Evidence on the negative impact of International databases. The IPUMS overcrowding based on clear causal data contains information on access to relationships has been compiled in a sanitation services. Data also include the number of studies around the world. This gender of the child, the gender of the ranges from evidence on the link between head of the household, urban or rural mental health and overcrowding in residence, number of siblings, whether Thailand (Fuller et al., 1993) to evidence or not the child lives with both parents, that relationships between parents and the completed education level of and children suffer in overcrowded the head of the household. Since total settings in the US (Gove et al., 1979). household income was not comparable Studies have examined the link between across samples, this circumstance overcrowding and the likelihood of being variable was excluded.6 Because these exposed to unhygienic conditions and are samples of the census data, all the causal link between overcrowding calculations were weighted (see Annex and educational attainment and progress Table A.3.3 for more details). For LAC (Coggan et al., 1993; Currie and countries, we use the harmonized Yelowitz, 2000; and Goux and Maurin, household surveys from the SEDLAC 2005). This confluence of evidence has database described in Chapter 2. led some countries to develop statutory Severe overcrowding is generally overcrowding standards (UK) and considered to exist when there are more others to develop targeting indicators than 1.5 people per room on average 6 When data were available, a series of variables for assets owned by the household were included. The annex reports results both with and without information on asset ownership as a measure of wealth. 86 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World (although the US considers overcrowding 3.1). Only two LAC countries, Venezuela any level above one person per room). (81) and Chile (74), report HOI levels By combining IPUMS data on number of above the European average. When rooms with information on the number the definition is broadened to include of people in the household, we derive septic tanks, five countries are above the number of people per room to use in the European average (Figure 3.2). The the overcrowding HOI. remaining countries also improve when considering septic systems, but remain The Human Opportunity Index for below the average of the European Sanitation countries.7 Much of the sanitation HOI gap is attributed to inequality of When only access to a public sanitation opportunity levels approximately twice connection is considered, the majority of as high in LAC than European and North LAC countries we analyzed were below American countries when considering the average for all available European septic systems (Tables A3.5 A-B). countries of approximately 60 (Figure 7 Annex Table 3.4A reports the results from the most recent round of household surveys analyzed in Chapter 2. Annex Table 3.4B complements these results using data from censuses available in the IPUMS database. 87 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Figure 3.1: HOI Sanitation Access to Sanitation (Public connection only) Spain 92 USA* 70 France** 69 Greece 58 Portugal 45 Hungary 44 Romania 21 South Africa 19 NON - LAC Kyrgyz Republic 8 Vietnam 5 Kenya 2 Venezuela 81 Chile 74 Argentina 47 Uruguay 38 Mexico 37 Brazil 36 Peru 29 LAC Ecuador 28 Bolivia 20 El Salvador 19 Guatemala 16 Panama 14 Costa Rica 14 Dominican Republic 13 Honduras 12 Jamaica 8 LAC EUROPE Nicaragua 8 Paraguay 3 0 20 40 60 80 100 HOI (%) Sources: LAC: CEDLAS Data, Non-LAC: IPUMS Census Data * For USA, the most recent year is 1990 ** For France, the most recent year is 1982 88 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Figure 3.2: HOI Sanitation (Public + Septic Tank) Access to Sanitation (Public and private connections) USA* 99 Portugal 95 Hungary 73 Greece 58 Romania 31 South Africa 21 Kyrgyz Republic 8 Vietnam 5 NON - LAC Kenya 2 Costa Rica 93 Chile 86 Venezuela 84 Argentina 77 Uruguay 77 Peru 52 Ecuador 51 Dominican Republic 49 LAC Brazil 45 Mexico 43 Jamaica 38 Panama 32 El Salvador 30 Bolivia 29 Honduras 23 Paraguay 22 LAC EUROPE Guatemala 21 Nicaragua 11 0 20 40 60 80 100 120 HOI (%) Sources: LAC: CEDLAS Data, Non-LAC: IPUMS Census Data * For USA,the most recent year is 1990 89 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Human Opportunity Index for HOI—92 percent of the opportunities Freedom from Severe Overcrowding for access to overcrowded homes needed for universality are available and The mean value of the freedom from equitably distributed, compared to only severe overcrowding HOI in European 30 percent in Peru. Much of the freedom countries is 60. Only three LAC countries from severe overcrowding HOI gap is are above this value—Brazil (87), Chile attributed to inequality of opportunity (84) and Costa Rica (78). The remaining levels more than twice as high in LAC countries are five or more percentage than European and North American points below the mean (Figure 3.3). countries (Tables A3.6 A-B) In Spain—the highest scorer in this 90 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Figure 3.3. HOI Freedom from Severe Overcrowding Freedom from Severe Overcrowding ( > 1.5 people per room) Spain 92 Austria 91 France 90 USA 88 Portugal 80 Italy 72 Greece 66 Slovenia 58 South Africa 46 Hungary 44 Romania 43 Belarus NON - LAC 35 Kenya 28 Kyrgyz Republic 27 Ghana 23 Armenia 18 Rwanda 17 Jordan 17 Mongolia 6 Brazil 87 Chile 84 Costa Rica 78 LAC Ecuador 56 Mexico 51 Uruguay 50 Dominican Republic 42 Honduras 42 Venezuela 33 Argentina 32 Peru 30 Panama 24 Paraguay 24 Jamaica 18 El Salvador 17 Bolivia 13 Nicaragua 8 EUROPE LAC Guatemala 7 0 20 40 60 80 100 HOI (%) - LAC: IPUMS Census Data Sources: LAC: CEDLAS Data, Non 91 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World 3.3 Understanding the Long-run The long-run evolution of the HOI Evolution of the HOI for Severe Overcrowding The IPUMS-International database Over the past 45 years, the US’s HOI for contains information on multiple rounds freedom from severe overcrowding has of census data for several countries. The improved by roughly 15 points (Figure longest two time series available and 3.4). The HOI improved rapidly in the analyzable were from the US and France. first part of the period analyzed, and then This section exploits that time series by growth slowed as the HOI approached exploring the long-run evolution of the universality (Annex Table A.3.8). This housing and sanitation HOI in these two might be in line with a notion that the last countries. For the US, data is available unit of a good or service is more costly from each decennial census from 1960 to provide than the first.8 The freedom to 2000, and then in 2005 data from the from overcrowding HOI in France began American Community Survey is used. from a much lower base than in the US, In the case of France, the data are less but improved much more quickly over consistently available though they span a shorter period of time—from about an equally long time period. 55 in 1968 to about 87 in 1999. Similar to the US, improvements slowed as the HOI reached higher levels. 8 This slowing down of the HOI growth rate as approach to universality would imply in most cases that the actual arrival time to universality would be slower than what a linear growth assumption would suggests. In light with this finding, the arrival time to universality in LAC discussed in Chapter 2 should be consider as “optimist� projections of the LAC expected trend. 92 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World Figure 3.4: HOI Overcrowding and Sanitation Overcrowding and Sanitation USA 1960 -2005 France 1968 - 1999 95 85 75 65 55 45 35 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 USA: No Severe Overcrowding (>1.5 people per room) USA: Access to Sanitation, public connection only (Children 0 - 16) France: No Severe Overcrowding (> 1.5 people per room) France: Access to Sanitation, public connection only (Children 0 - 16) USA: Access to Sanitation, public and private connections (Children 0 -16) Note: We report all available data in this figure. Data are not available to compute access to septic tanks for France (see annex for details). The long-run evolution of the HOI but no data were available in France on for Adequate Sanitation septic systems to make a comparison. As with overcrowding, both countries The HOI of access to adequate sanitation showed faster growth rates when the (public system only) shows much more HOI was low, with progress slowing similar values in US and France than as the HOI increases. For instance, the overcrowding index, although for the USA while in 1960 the HOI for improvement in the US stagnates after sanitation (public system + septic tanks) 1970 while France shows steady growth. grew roughly 2 points per year between However the French time series for this 1960 and 1970, but then only 0.3 points HOI is much shorter, making direct per year in 1970-1980 and 0.1 points in comparisons difficult. Including septic 1980-1990 (see Annex Table A.3.7 for systems into the calculation brings the US more details). HOI up to near universality after 1970, 93 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World 3.4 Conclusion The results on housing cover two areas relevant for policy makers: access to The analysis in this chapter illustrates sanitation and freedom from severe that LAC countries still need to make overcrowding. Only two of 18 LAC significant progress to achieve the countries analyzed were above the levels of equal opportunity most OECD average for European countries in access countries have achieved in the provision to sanitation via a public connection, of basic services. This is particularly and only five were above the European true regarding the quality of education, average when the definition included where LAC countries all score below septic tanks. The average value of the even the lowest-achieving countries in freedom from overcrowding HOI was Europe and North America. Housing 60 among European countries. Only HOIs are also below European averages, three LAC countries—Costa Rica, Chile but in this case some LAC countries have and Brazil—are above this value, with reached and even exceeded the median the remaining countries five or more European HOI scores. points below. Despite gains in educational outcomes, Analyzing the expansion of the HOI for the gap between the education HOI housing opportunities in the US and in LAC and Europe and North America France from the 1960s indicates that HOI remains large. All LAC countries included levels are initially low, experience fast in this study have a lower HOI than any growth rates, and then slow down as the of the countries analyzed in Europe HOI reaches a high level. This suggests and North America on opportunities that LAC countries may follow a similar for accessing quality education. The pattern—achieving strong gains in the countries with the highest HOIs for equitable provision of basic services reading, mathematics and science in LAC earlier in the development process, and are considerably behind the countries then slowing down as countries come with the lowest performance in Europe close to universal provision. The evidence and North America. Much of the overall from Chapter 2 on trends in LAC over education HOI gap is attributed to time coincides with this finding in certain inequality of opportunity levels often respects. two or three times higher in LAC than This chapter underscores that the European and North American countries. current generation of children in Latin Wealth status is the circumstance most America have fewer opportunities of strongly associated with inequality of accessing key goods and services than opportunity in reading, mathematics their counterparts in Europe and North and science in LAC countries. America, and that existing opportunities 94 Human Opportunities in a Global Context: Benchmarking LAC to Other Regions of the World are distributed less equitably. Endowed reach their potential as adults. This with these more limited opportunities, suggests important barriers remain in the literature suggests that it will be all regions analyzed, but particularly in more difficult for these children to enjoy Latin America, to ensure that the next psychological and physical good health generation of children faces a level as children and to be motivated and playing field in obtaining opportunities equipped to pursue their interests and needed to develop themselves. 4 95 Chapter Human Opportunities at the Sub-national Level in Latin America and the Caribbean 96 Chapter 4 Human Opportunities at the Sub-national Level in Human Opportunities at the Sub-national Level in Latin America and the Caribbean Latin America and the Caribbean Spatial inequalities—both in terms of educational opportunities compared to outcomes, such as income and poverty, opportunities to access quality housing. and in access to critical opportunities— This chapter is organized as follows. have received considerable attention Section 4.1 describes the main findings in the literature. Not only do spatial regarding sub-regional HOIs across inequalities remain marked, but they countries. It considers stylized facts for have the potential to create significant housing and education opportunities political tensions and they present an across the top and bottom-ranked sub- actionable policy challenge. national regions. Section 4.2 analyzes This chapter explores inequalities in the different measures of dispersion of sub- HOI at the sub-national level (region, national HOIs and how dispersion may province, department or state) in Latin relate to different national and regional America and the Caribbean (LAC). policies, most notably the degree of The chapter also analyzes whether the political and fiscal decentralization. dispersion of sub-national HOIs grows or Section 4.3 discusses possible ways to decreases with the level of the HOI and equalize opportunities across regions, the structure of government (degree of supporting the World Development decentralization), and outlines possible Report 2009 message that opportunities fiscal instruments which can be used should be equalized by leveling the to accelerate equality of opportunities playing field, rather than by seeking across sub-national regions. uniformity in outcomes. Section 4.4 Not surprisingly, the chapter discovers offers concluding remarks. significant heterogeneity in sub-national opportunities across time, countries, 4.1. The Sub-national Human and the specific opportunity dimensions Opportunity Indices: Some analyzed. It presents some stylized facts Stylized Facts on overall equality of opportunities at the sub-national level as well as on access A series of sub-national Human to housing and education opportunities. Opportunity Indices (SN HOIs) is The analysis shows convergence estimated using data from 30 household between states with high and low initial surveys for 15 LAC countries over a HOI levels, albeit at a rather slow pace. period of more than a decade (1995- This convergence appears more in 2009). The analysis employs the same 97 Human Opportunities at the Sub-national Level in Latin America and the Caribbean basic services, circumstances and overall the lowest national HOI scores. All the methodology as the national indices regions of El Salvador, Nicaragua and in Chapter 2. Together, the surveys Guatemala are included in the 50 lowest- represent roughly 165 sub-national ranked SN HOIs, except for their capital governments from 15 LAC countries cities (Annex Table A4.2). Uniquely, Peru (Annex Table A4.1 and A4.2).1 has regions both in the top 50 and bottom The analysis finds significant 50 provinces: one in the top and three in heterogeneity in children’s access the bottom. More than 60 percent of to basic services across sub-regions the sub-regions of Honduras, Panama, or sub-national governments in the Ecuador and Brazil are in the middle countries considered, with a much wider ranking, and the rest in the bottom. range than between overall national- They seem to be in the “lower-middle level HOIs.2 The spread extends from class� in terms of provincial distribution a high of 96 for Tierra del Fuego in among this group of countries. On the Argentina to lows of 29 in the Atlantic other hand, Colombia and Costa Rica Region, Nicaragua and 31 in Morazán, also have more than 60 percent of their El Salvador. SNs in the middle with the rest in the top, The ranking of sub-national regions is hence are more “upper-middle class.� roughly similar to the national ranking Colombia and Brazil, two of the most by HOI. Chile and Uruguay are the unequal countries in the region in terms top countries in terms of the national of income inequality and regional income HOI. Of the 165 provinces considered, disparities, do not appear so unequal the first 50 provinces include 18 from in terms of the spatial distribution of Uruguay, 16 from Argentina, 11 from opportunities.4 Chile, three from Colombia, one from Opportunities to access basic goods and Peru and one from Costa Rica. All except services are systematically higher in the one of Uruguay’s provinces (95 percent) capital cities. Comparing the overall are among the top 50, Chile has 85 HOI by capital city and the rest of the percent (all except two), Argentina 70 country, Bogotá, Santiago, Buenos Aires percent, while Colombia has only 30 and Montevideo are the top-ranked, percent (three out of nine provinces).3 and all capital cities provide more Similarly, the lowest-ranked SN HOIs opportunities than other areas of their tend to come from the countries with countries (Figure 4.1).5 1 The heterogeneity in the data is corrected to the extent possible in next sections. A sensitivity analysis of the results was done by grouping all the countries by region and not by province. Although there are some minor changes that will be noted in next sections, most conclusions remain unchanged. 98 Human Opportunities at the Sub-national Level in Latin America and the Caribbean Figure 4.1: Overall HOI Circa 2008 by Main City and Rest of the Country Bogota DC - COL Metropolitana de San - CHL Montevideo - URY City of Buenos Aires - ARG Rest of Chile Central - CRI Lima metropolitana - PER Rest of Uruguay Rest of Argentina Asunción - PRY Distrito federal - BRA Rest of Costa Rica Region -Country Región Distrito Naci - DOM Sierra*- ECU Panamá - PAN Rest of Colombia Tegucigalpa (Dist.Nac) - HND Rest of Brazil Metropolitana - GTM Managua - NIC San Salvador - SLV Rest of Peru Rest of Paraguay Rest of Ecuador Rest of Dominican Republic Capital Cities in Yellow Rest of Panama Rest of Honduras Rest of El Salvador Rest of Guatemala Rest of Nicaragua 30 40 50 60 70 80 90 100 *It includes Quito, the capital city HOI 2008 2 In this chapter the terms regional, sub-national, provincial or state government are used interchangeably unless specified, for example between provinces and municipalities, which are clearly two different government levels. Some of the countries analyzed in Chapter 2 did not have representative data at any regional or sub-national level and hence were dropped from this analysis (see Annex Table A4.1). This is the case of Bolivia, Jamaica, Mexico and Venezuela. Among the 15 remaining countries, seven—Costa Rica, Dominican Republic, Ecuador, Guatemala, Honduras, Nicaragua and Peru—could not be analyzed by province or department since the data was not representative at that level of analysis, and hence regional data was divided into economic or natural regions by the respective national statistics institutes. The rest of the countries were divided across provinces or departments. However, in some cases, like Paraguay, the departments included may not cover all the national territory and population, since their sample size was not sufficiently high to estimate the HOI. The standard error of the sub-national HOIs is much larger than that of the national HOIs, since their sample size is obviously much smaller. In this respect, the ranking among sub-national HOIs should be analyzed with care. 3 This analysis is not strict, not only because of heterogeneity in the definition of sub-regions but also in that countries have different numbers of SNs, whatever the definition used. Nonetheless, the comparisons are revealing. 4 According to the latest data available in World Development Indicators 2009, the Gini coefficient for per capita income was 55 for Brazil in 2007 and 58.5 for Colombia in 2006. ECLAC (2008) classifies Brazil and Colombia within the “high inequality� countries in LAC together with Bolivia, Honduras and Guatemala. It also notes that while Brazil income inequality decreased in the 2000s, the opposite occurred in Colombia. With respect to regional inequality, Herrán (2005) found that in 2000 in Brazil, 24 percent of income inequality was explained by regional differences, while 76 percent was due to household-specific characteristics. Colombia, in spite of perceptions of high regional disparities in income (measured by GNP), is not highly unequal as measured by the HOI, but rather moderate compared internationally. However, it does not show convergence among its regions nor policies to diminish regional inequalities. 5 Comparisons between the capital city and the rest of the country reflect in part the wide differences between urban and rural HOIs, but also the high disparities in LAC between main capital cities—usually mega cities, modern, but with important agglomeration of inhabitants, migrants, and slums, and economic activity centralized around the metropolitan area—and other “provincial� cities. 99 Human Opportunities at the Sub-national Level in Latin America and the Caribbean Among the lower-ranked countries, their respective countries, but also the the gap is greater between the capital higher the HOI, the lower the gap in cities and the rest of the nation (Figure opportunities between capital and the 4.2). Not only are the HOIs of the rest of the country. capital cities higher than the rest of Figure 4.2: HOI Circa 2008 by Main City and Rest of the Country 100 CHI 90 URU ARG HOI for Rest of the Coun try 80 CRICA COL BRA 70 ECU PER PAR DOM PAN HON 60 50 ELS GUA NIC 40 40 50 60 70 80 90 100 HOI for Capital Citie Note: The fact that capital city HOIs are below the 45 degree line indicates that they have higher HOI scores than the rest of the country. Disaggregating SN HOIs by education between capital cities and the rest of and housing opportunities shows that the country is greater for housing than access to opportunities is more uneven for education. The relative ranking of for housing than for education (Figures capital cities does not perfectly match 4.3 and 4.4). It also reveals that the gap the overall national HOI ranking. The 100 Human Opportunities at the Sub-national Level in Latin America and the Caribbean capitals of Brazil, Colombia have the outside of Lima score in the top third of highest rankings for housing, while the the HOI for education), it has low HOIs capitals of Colombia and Peru have the for housing outside of Lima (Figures 4.3 highest ranking for education, despite and 4.4). In Brazil, HOIs are lower in the fact that these are not the top education, but the gap between the countries in the national HOIs.6 capital city and the rest of the country The dispersion of SN HOI scores for is greater for housing: the HOI for education and housing do not necessarily education in the capital is 65, and in the follow similar patterns. While Peru has rest of the country it is 64; while the relatively higher HOIs for education housing HOI is 99 for Brasilia and 82 for across sub-national regions (the areas the rest of Brazil. Figure 4.3: HOI in Housing Circa 2008 for Capital City and the Rest of the Country Distrito federal- BRA Bogota DC-COL Central - CRI Metropolitana de San - CHL Montevideo- URY City of Buenos Aires - ARG Rest of Uruguay Rest of Costa Rica Rest of Chile Rest of Argentina Región Distrito Naci - DOM Asunción - PRY Lima metropolitana - PER Region - Country Rest of Brazil Sierra*- ECU Metropolitana-GTM Panamá- PAN Rest of Colombia San Salvador -SLV Tegucigalpa (Dist.Nac) - HND Managua-NIC Rest of Paraguay Rest of Dominican Republic Rest of Ecuador Rest of Peru Rest of Panama Capital Cities in Yellow Rest of Honduras Rest of Guatemala Rest of El Salvado Rest of Nicaragua 20 30 40 50 60 70 80 90 100 *It includes Quito, the capital city HOI 2008 6 Brasilia might rank top because of its relatively “new� status as capital. Nonetheless, the same ranking holds if Rio de Janeiro or São Paulo are used (with housing HOI of 95 and 97 percent respectively). 101 Human Opportunities at the Sub-national Level in Latin America and the Caribbean Figure 4.4: HOI in Education Circa 2008 for the Capital City and the Rest of the Country Lima metropolitana- PER Bogota DC- COL Rest of Chil e Metropolitana de San -CHL Montevideo - URY City of Buenos Aires - ARG Rest of Argentina Panamá - PAN Rest of Uruguay Sierra*- ECU Rest of Peru Central - CRI Rest of Colombia Region- Country Rest of Ecuador Tegucigalpa (Dist.Nac) - HND Asunción - PRY Rest of Panama Región Distrito Naci - DOM Rest of Costa Rica Managua - NIC Rest of Dominican Republic Rest of Paraguay San Salvador - SLV Rest of Honduras Distrito federal- BRA Metropolitana - GTM Rest of Brazil Capital Cities in Yellow Rest of El Salvador Rest of Nicaragua Rest of Guatemala 20 30 40 50 60 70 80 90 100 *It includes Quito, the capital cit HOI 200 There is some evidence of a decreasing 13 years considered, seven out of the ten gap over time between capital cities and greatest improvement in the HOIs were 8 experienced by the areas outside the y areas of the country for access other to basic services, as well as an overall capitals of Panama, Nicaragua, Paraguay, convergence in levels among the sub- Guatemala, Peru Brazil, and Ecuador. national regions. Comparing annual Among the ten lowest growth areas, overall HOI growth from the mid-1990s seven are capital cities—Buenos Aires, to 2008 between capital cities and the Panamá, Asunción, Bogotá, Santiago, areas outside the capital city suggests San José and San Salvador.7 Plotting the that in general the highest increases in overall HOI growth rate against the HOI in the overall HOIs were attained outside levels indicates convergence of opportunities the capital cities, and in particular in across sub-national regions (Figure 4.6). areas that had lower services in the mid- The results reveal that the higher the HOI, 1990s (Figure 4.5). In the approximately the lower the rate of change.8 102 Human Opportunities at the Sub-national Level in Latin America and the Caribbean Figure 4.5: Annual Growth of HOI Between Circa 1995 and 2008: Capital Cities vs. Rest of Country 2,50 2.10 Capital Cities in Yellow 2,00 1.86 1.77 1.52 1.54 1,50 1.41 1.36 1.31 1.24 1.26 1.14 1.15 1.06 1.07 Rate of Growth 1,00 0.91 0.94 0.84 0.80 0.69 0.58 0.49 0,50 0.43 0.37 0.38 0.26 0.12 -0.32 - 0.11 -0.08 0,00 -0.56 Rest of Honduras City of Buenos Aires - ARG Rest of Colombia Montevideo - URY Rest of El Salvador Rest of Costa Rica Rest of Ecuador Rest of Guatemala Panamá - PAN Rest of Panama Rest of Uruguay Metropolitana de San - CHL Rest of Argentina Región Distrito Naci - DOM Distrito federal - BRA Rest of Dominican Republic Rest of Peru Metropolitana - GTM Rest of Paraguay Asunción - PRY Central - CRI Rest of Chile Rest of Brazil Managua - NIC Rest of Nicaragua Bogota DC - COL San Salvador - SLV Lima metropolitana - PER Tegucigalpa (Dist.Nac) - HND Sierra* - ECU -0,50 -1,00 Region - Country *It includes Quito, the capital city 7 The decrease in the annual growth in services in some of the cities should be interpreted with care. The sample size and some problems with the comparability of the data, especially in the case of Paraguay and Argentina, might be driving the negative result. Most likely the rate of growth in these capitals should be close to zero. 8 A simple OLS regression of the 165 SN HOIs rate of growth and the HOIs estimates a coefficient of -.015 and a t-statistic of -4.67, showing the coefficient to be statistically significant in explaining that regions with more limited opportunities grow on average at faster rates than higher ranked regions. 103 Human Opportunities at the Sub-national Level in Latin America and the Caribbean Figure 4.6: Annual Growth of HOI Between Circa 1995 and 2008 100 90 80 70 HOI 1998 60 50 40 30 20 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 Annual HOI Growth Further evidence for convergence can regions with a higher HOI in 2008 (Figure be found in graphing the HOI circa 1995 4.7). The time convergence seems more against the HOI circa 2008: the trend pronounced in the case of education line is flatter than 45 percent, suggesting than in housing (Annex Figures A4.2 that the rate of change was smaller for and A4.3) 104 Human Opportunities at the Sub-national Level in Latin America and the Caribbean Figure 4.7: Overall HOI Circa 1995 and 2008 Overall HOI : circa 1995 and 2008 100 90 80 70 60 HOI for 2008 50 40 30 20 10 0 0 10 20 30 40 50 60 70 80 90 100 HOI for 1995 4.2. Hetereogeneity in Sub- Many economists interested in regional regional Human Opportunity differences tend to examine variation Indices in regional poverty outcomes, focusing mainly on income inequality (see Shankar In this section we discuss two issues: (i) and Shah, 2003, and Von Braun and the relationship between national and Grote, 2002). While it is valid to analyze sub-national HOIs and (ii) whether or income inequality across regions of a not more decentralized countries are given country, there is less consensus on more effective in equalizing sub-national what type of “just� redistributive policy HOIs. would follow from the analysis. Income (or geographic GDP) equality depends 1) Relationship between national and on factors outside the control of policy sub-national Human Opportunity makers, and not only on equality Indices of opportunity but also on personal 105 Human Opportunities at the Sub-national Level in Latin America and the Caribbean responsibility and random shocks (luck).9 sample between the two periods are Much more consensus exists on creating Peru, Honduras and Nicaragua with redistributive justice by equalizing dispersion values of 15 to 21, followed opportunities—leveling the playing by Guatemala, Paraguay, El Salvador, field—especially for children. Colombia and Dominican Republic with dispersion values between 10 and 15. No single statistical measure can capture Countries with the lowest dispersion are, the myriad dimensions of regional as expected, Chile, Uruguay, Argentina disparities. The last section showed and Costa Rica. One exception is Peru, that while disparities tend to be larger with the highest dispersion circa 1995 in regions with lower HOIs, these same and yet an HOI just below the mean. regions tend to grow faster than those SN HOIs show a degree of convergence, with higher HOIs. The standard deviation as countries expand and equalize human of overall, educational and housing opportunities as they develop. Countries HOIs is a useful summary statistic that with low HOIs tend to have a larger reflects regional differences in access to spread of SN HOIs, while countries with key goods and services. The standard a higher national HOI have a lower sub- deviation is weighted by the population national spread. The dispersion of SN share of each location to avoid biasing HOIs tends to be higher, the lower the the indicator toward small and very national-level index (Figure 4.8). In this unequal regions, considering that some sense, achieving more coverage with countries have more regions or sub- more equality over time tends to close national governments than others. regional disparities in opportunities. When countries have near-universal In most countries the dispersion in SN coverage of access to basic services, HOIs decreased between circa 1995 and such as Argentina, Uruguay or Chile, 2008, except in Nicaragua, Honduras, they also have relatively low dispersion Guatemala, and Ecuador, where it in their sub-national HOIs. Countries increased even though the overall HOI with lower HOIs show a wider disparity for those countries improved during the across sub-national HOIs. The countries period (Figure 4.9). with the highest dispersion in the 9 Even if policy makers could control regional outcomes, the WDR 2009 asserts that it is not efficient to equalize geographical GDP . Instead, people in different regions can be empowered to move if desired to regions with more dynamic economic activity. 106 Human Opportunities at the Sub-national Level in Latin America and the Caribbean Figure 4.8: Dispersion of HOI Circa 1995 and 2008 Circa 199 25 per 20 nic hon par pan els col Dispersion 15 gua dom bra cos chi 10 ecu arg 5 uru 0 30 40 50 60 70 80 90 100 HOI 199 Circa 2008 25 20 hon nic per Dispersion 15 gua els par pan dombracol 10 ecu cos chi 5 arg uru 0 30 40 50 60 70 80 90 100 HOI 2008 Circa 199 and 2008 25 per 20 hon hon nic per nic par pan els col Dispersion 15 gua gua dom par els bra pan dom cos chi bracol 10 ecu ecu cos chi arg 5 arg uru uru 0 30 40 50 60 70 80 90 100 HOI 1998-2008 107 Human Opportunities at the Sub-national Level in Latin America and the Caribbean Figure 4.9. Regional Dispersion decreased in most countries Chile Uruguay Costa Rica Colombia Brazil 2010 Ecuador Dominican Republic 1995 Paraguay Peru Panama El Salvador Guatemala Honduras Nicaragua Ordered by HOI 0.0 5.0 10.0 15.0 20.0 2) Do federal countries equalize some countries have more disparities HOIs more than unitary countries? than others, controlling for the level of their national HOI. The analysis It is clear from the previous sub-section focuses in particular on decentralization that Latin American countries have had as a possible explanatory factor. some convergence in the dispersion Because several countries which are of their SN HOIs, that regions with not formally federal have decentralized the lowest levels of access to services expenditure as much or in some cases tend to improve faster than the more more than federal countries, the analysis favorable regions, and that countries considers both political and expenditure with the highest (lowest) national HOIs decentralization. have the lowest (highest) spread in their The literature on fiscal federalism has SN HOIs.10 In this section, we inquire developed some interesting hypotheses whether other factors help explain why on whether federal countries tend to 10 The literature on geographical GDP convergence (see Barro and Sala-I-Martin, 1991) distinguishes two types of convergence in growth: sigma convergence and beta convergence. A falling dispersion of real per capita income across a group of economies over time signifies sigma convergence, while a negative partial correlation between growth in income over time and its initial level signifies beta convergence. Hence, in the case of the HOI there is both sigma convergence— since there is a reduction in the dispersion of SN HOIs—and also beta convergence that produces faster growth in regions with lower coverage, corrected for inequality opportunities. Some authors also showed that beta convergence is necessary for sigma convergence. 108 Human Opportunities at the Sub-national Level in Latin America and the Caribbean have less or more regional disparities argue that a central government can than more unitary systems. Proponents more easily redistribute resources from of decentralization maintain that the more to less advantaged regions, and has potential threat of disunion in federal an interest in providing public services states generates an incentive to reduce and policies equally across the country.12 regional disparities, and hence federal Some authors (see Burki et al, 1999) also countries are expected to have less claim that decentralizing expenditure can disparity.11 As well, local governments— shift power from national to local elites being closer to the public—will have and might act as an incentive corruption better information and thus will be able at the local level and limit improvements to ensure better provision of public in local services provision. services such as education and housing. In the past few decades, Latin America Decentralization supporters also argue has undergone a major decentralization that centralized policy making often process, involving institutional, fiscal favors particular regions or cities and and political reforms. The process has burdens all regions with uniform policies differed from country to country, not and public services too unresponsive only in the intensity and depth of the to local needs and conditions, thus reforms, but also in their outcomes increasing spatial inequality. (Table 4.1). There are only two formally On the other hand, flexibility in federal countries in the sample of LAC choosing policy instruments is curtailed countries considered: Argentina and by the division of powers in a federation. Brazil. That said, the rest of the countries Central governments in unitary states under study have elections at least at the are relatively unconstrained in their municipal level, offering some elements choice of appropriate policy instruments, of decentralized political systems. and hence might be more efficient in With respect to fiscal federalism, diminishing inequalities. Further, the the experience is more varied. advantage of local information could Some countries have a high level of be outweighed by economies of scale decentralization on the expenditure side, and positive externalities with large- such as Argentina, Brazil and Colombia, scale service provision by a central while other countries such as Chile, government. Critics of decentralization Costa Rica and El Salvador are much 11 Decentralization can also increase regional disparities between jurisdictions. Prud’homme (1994, 1996) argues that decentralized redistribution is likely to lead to different treatment of similar individuals in a country. For example, jurisdictions with higher per capita income would be able to provide higher levels of public services than those with lower per capita income. Residents in wealthier jurisdictions could even be levied at lower tax rates for higher levels of public services than those in poor jurisdictions. 12 See Faguet and Shami, 2008, for an excellent exposition of advantages and disadvantages of decentralization in promoting regional equality. 109 Human Opportunities at the Sub-national Level in Latin America and the Caribbean more centralized. The process of fiscal data. As a result, we use sub-national decentralization on the revenue side is expenditure data to measure the extent not as deep as on the expenditure side. of fiscal decentralization.13 Expenditure Except for Argentina and Brazil, most data, being continuous in nature, LAC countries do not raise significant more accurately reflects the fact that amounts of sub-national taxes. Also, data decentralization is not an “all-or-nothing� on sub-national revenues is much less phenomenon, and that countries exist accurate than sub-national expenditure along a continuum of reform. Table 4.1 Federalism in Selected Countries of Latin America Federalism in Selected Countries of Latin America Subnational Government Spending/Total Country Formally Dummy Formal Government Spending 1995 2004 Argentina Federal 1 49.3 49.3 Bolivia Unitary 0 26.7 25.1 Brazil Federal 1 45.6 47.0 Chile Unitary 0 13.6 12.8 Colombia Unitary 0 39 44.7 Costa Rica Unitary 0 2.3 3.1 Dominican Republic Unitary 0 3.2 3.2 Ecuador Unitary 0 7.5 17.5 El Salvador Unitary 0 6 5.1 Guatemala Unitary 0 10.3 10.3 Honduras Unitary 0 12.3 12.3 Nicaragua Unitary 0 5.2 5.0 Panama Unitary 0 2.9 1.0 Paraguay Unitary 0 6.2 4.0 Peru Unitary 0 10.5 19.0 Uruguay Unitary 0 14.2 13.7 Venezuela, R.B de Federal 1 19.6 19.6 13 We use the decentralization data gathered by the IDB through a series of interviews, and used by Stein (1999) and updated by Daughters and Harper (2007), so we can relate this index to measures of dispersion in 1995 and 2008 to test our hypothesis. 110 Human Opportunities at the Sub-national Level in Latin America and the Caribbean The analysis on the relationship between prescriptions, they can improve our political and fiscal federalism and understanding of how phenomena regional inequality of opportunity is such as decentralization work in general preliminary, suggesting possible areas for terms, and are useful when interpreted further research. To test the hypothesis in this light.14 as to whether decentralization increases The econometric results suggest or decreases spatial dispersion in the lower levels of regional dispersion HOI, we run linear regressions with the are correlated with higher HOI levels, variability measure as the dependent greater expenditure decentralization, variable, and as independent variables and smaller land areas (Table 4.2). While the log of the national overall HOI, the the data is limited and further analysis formal index of federalism or alternatively warranted, the results also indicate that the log of expenditure measure of federal systems are associated with lower decentralization, and the log size of the levels of regional dispersion. This is true country (in square km). using either a specification based on A caveat worth mentioning is that government expenditure or one based decentralization is complex and most on a political definition, although the countries have diverse institutions and latter is significant only at the 10 percent arrangements, and the use of quantitative level (columns 1 and 2, Table 4.2). As methods with a small number of control expected, the higher the HOI index, variables runs the risk of simplification. the lower the dispersion: the coefficient This explains the heavy reliance on case is negative and statistically significant studies and qualitative methods in the at the 1 percent level. The log of area decentralization literature. However, has a positive and statistically significant cross-national analysis can provide coefficient at the 5 percent level in insights into how concepts and theory both specifications of columns (1) and translate into practice, and it can indicate (2), meaning the larger the land size of general trends that might otherwise be the country, the more likely is that the masked. While cross-national studies dispersion in basic goods and services do not provide country-specific policy is also larger. Hence, it appears more 14 Another problem in a cross section without a long time series is that country-specific unobserved variables can bias and confound interpretation of coefficients. Control variables in part mitigate these effects. 15 The log of population was also used as a control variable alternative to the size of the country. Its coefficient was positive as expected, but not statistically significant. The coefficient on the political dummy for federalism is negative and statistically significant at the 5 percent level, meaning that federal countries have a lower dispersion in the level of overall opportunities corrected for inequalities. However, only two countries in the sample are federal (the sample for regions or sub-national areas in Venezuela, the other federal country, was not representative for estimating the HOIs and had to be dropped). Hence this dummy is only for Argentina and Brazil, and thus could be confounding the effect of just federalism (though size in the sample is already controlled by the log of surface area). 111 Human Opportunities at the Sub-national Level in Latin America and the Caribbean difficult for larger countries to equalize Rica, Ecuador, Panama, Paraguay and basic goods and services, for same level Uruguay retain centralized educational of development and decentralization. systems). However, even in LAC A country’s land area is less significantly countries that have decentralized primary associated with the education HOI than education, the central government with the housing HOI, suggesting that retains some functions at the national it is less difficult to bring educational level (Burki et al, 1999). opportunities to remote areas than In the case of housing infrastructure, housing opportunities. The size of the responsibilities are shared among country is positively associated with national, provincial, state and also inequality only in the case of housing, and with state or private enterprises, often appears less relevant in the distribution creating accountability and responsibility of educational opportunities across problems in the provision of water, locations. Expenditure decentralization sanitation and electricity. As noted appears not to exert significant influence by Foster (2005), the decentralization on the spread of housing opportunities of water and sanitation services to across locations, but it does seem to small local governments in LAC led to exert a negative and significant effect a loss of economies of scale in service on the education HOI, significant at delivery in countries such as Argentina, the 5 percent level.16 Here too political Colombia, and Peru, entailing a sudden federalism seems to be negatively fragmentation of the industry into correlated with the extent of regional hundreds of small municipal providers. inequality in accessing housing and In a number of cases, decentralization education opportunities, although again preceded subsequent water sector these results are tentative and call for reform by a number of years. This lack further research. Fiscal decentralization of synchronicity between structural seems associated with lower regional and regulatory reform was unfortunate dispersion of HOIs, and especially of because it meant that regulatory reform educational HOIs. Most countries in Latin had to be superimposed on an industry America have decentralized schooling structure that was often far from optimal systems for at least a significant share in an economic sense. Hence, it might of education responsibilities (only Costa be that well-coordinated and managed 16 There is a small number of observations (N=30) to perform tests only asymptotically valid to analyze further the statistical properties of the OLS model. However, robust and clustered standard errors were estimated for the model, resulting in similar standard errors, albeit smaller in the case of the robust option. None of the conclusions significantly changed. 17 More research needs to be done to ascertain the reasons for the lack of significance of the coefficient on decentralization for housing. Some authors such as Foster (2005) and Zannetta (2004) find that the problem is “inadequate� decentralization without controls from the central government and a clear assignment of specific functions between local and central governments, more than decentralization per se, that explains the policy failure. 112 Human Opportunities at the Sub-national Level in Latin America and the Caribbean expenditure decentralization even in than in the case of sanitation, water the case of housing might help equalize or electricity. At a minimum it is clear housing opportunities.17 that equalizing housing and education This analysis suggests that fiscal opportunities will require different decentralization may have been more policies in each sector, although there effective in diminishing regional are some common policy principles, as inequalities in the case of education discussed in next section. Table 4.2 Table 4.2 OLS Estimates Subnational Dispersion in HOIs OLS Estimates Subnational Dispersion in HOIs (Using Weighted Standard Deviation) All Oportunities Housing Opportunities Education Opportunities (1) (2) (3) (4) (5) (6) Federal -6.5044** -7.6218** -3.7787* (-2.61) (-2.09) (-1.84) Subspend -1.9492* -1.7374 -1.7979** (-1.85) (-1.13) (-2.25) lnArea 1.3765** 1.4244** 1.9282** 1.7281* 0.4801 0.8306 (2.54) (2.07) (2.47) (1.74) (1.08) (1.62) lnHOIj -14.5857*** -15.0744*** -13.1582*** -13.7144*** -10.3580*** -10.5387*** (-5.93) (-5.77) (-5.53) (-5.47) (-3.94) (-4.11) Constant 64.2908*** 69.6409*** 58.1717*** 64.4961*** 48.7070*** 51.1109*** (6.51) (6.77) (6.08) (6.79) (4.51) (4.83) N 30 30 30 30 30 30 r2 0.6235 0.5799 0.5908 0.5445 0.4165 0.4481 F 14,3544 11,9619 12,5105 10,3595 6,1859 7,0376 t statistics in parentheses * p<0.10 ** p<0.05 *** p<0.01 Note: Columns 1 and 2, which show the estimates for the overall HOI constructed from the average of the five basic housing and education opportunities, only differ in the control used to proxy federalism is based on a political definition or empirical government expenditure 113 Human Opportunities at the Sub-national Level in Latin America and the Caribbean We conclude that significant regional to more unequal regions. The problem inequalities persist in many LAC is to find the best and most efficient countries. By most measures of regional approach in LAC, given concerns over inequality, countries with lower access public sector accountability and in light to basic goods and services tend to have of the predominance of hybrid political more regional dispersion, but also tend to structures (neither clearly federal nor experience higher rates of improvement unitary) where functions tend to be in the index, demonstrating a growing superimposed and not always clearly regional convergence in the provision of defined. basic goods and services. Some evidence Since decentralization seems irreversible indicates that more fiscally decentralized in LAC countries, a possible way to help countries achieve less HOI disparity, at accelerate the reduction in regional least in terms of education.18 But large disparities could be to design an equal dispersions still need to be addressed to opportunity policy coordinated by the level the playing field for children born central government and implemented at in different locations. the sub-national level. Local governments in all LAC countries rely at least partially 4.3. Equalizing Regional on transfers from central governments, Opportunities so this policy could take the form of an “equal opportunity grant� transfer from The WDR 2009 “Reshaping Economic central to local governments to mitigate Geography� calls for countries not inequality in accessing opportunities. equalize basic opportunities across This would involve distributing, on a provinces/states, rather than trying per capita basis, resources to provinces to equalize outcomes such as GDP or that attain specific goals in access poverty rates. Hence, the objective is to education, health and sanitation. not only reaching universal coverage The HOI is an ideal index to measure of basic opportunities, but also to performance under conditional grants ensure that until universal coverage is aimed at leveling the playing field in achieved, the distribution of access to accessing opportunities, because it is basic services should be equalized across straightforward to estimate given widely regions. Some policy measures can assist available household data in LAC and has countries to achieve greater fiscal equity clear interpretations. and redistribute government expenditure In the 1980s and 1990s, many countries 18 Regressions were also run using macro-regions for the countries that in Annex Table A4.1 were subdivided by provinces, and the same regions for the rest. Although the size of the dispersion changes in several cases, the results of the regression remain basically unaltered, even increasing the significance of the coefficients of some variables, such sub-national expenditure. 114 Human Opportunities at the Sub-national Level in Latin America and the Caribbean in Latin America introduced fiscal cannot work for most intergovernmental grants to sub-national governments systems. They maintain that there may to support decentralization. Some of be a limited role for a “reward� system these programs were conditional on of grants, in which those who perform inputs (classroom construction, teacher best get the most. Given that most salaries). However, input conditionality local governments depend on secure undermines budgetary autonomy (pre-committed) grant funding to carry and flexibility without providing any out many activities, and where many assurance of achieving results.19 Hence, grant programs are intended to meet the proposal is that the grant should “needs� rather than to reward those be conditioned on attaining equitably who have already succeeded in doing distributed access to basic services. so, performance-based budgeting may That is, the grant would be more be challenging to implement. concerned with outputs rather than with Hence, while output and performance- inputs. A growing number of countries based grants are clearly superior incentives are implementing transfer programs for local governments to help provide conditioned on attaining particular opportunities for accessing key goods opportunities across states or regions: and services, “penalties� can be hard the Australian National Schools Specific to sustain in the face of pre-committed Purpose Payments; the Canadian Health funds designed to provide access basic Transfers (CHT) program; the Brazilian services. Some authors propose a form Unified Heath System (SUS) and the of conditional grant, without curtailing FUNDEF program for primary schooling; pre-committed funds but with public and Chile’s grants to municipal exposure of performance. In Australia, governments for water and sewer access the National School Program is part of for the poor (see Shah, 2010). an overall reporting and accountability Some have suggested that the move in framework, and states must provide recent years to output-based budgeting performance data to the Commonwealth in place of input-based budgeting and to the general public. A continuous should be mirrored by a shift towards and independent dissemination of the performance-based grants, particularly estimation of the HOI across sub-regions for capital grants (Steffensen and Larsen, could be fundamental in improving 2005). In contrast, Smart and Bird performance and accountability without (2009) claim performance-based grants hurting citizens. 19 Hanushek (2003) reviews available evidence on the effects of schooling inputs and outcomes, and did not find any evidence that spending more on teachers or schools improves student outcomes. 115 Human Opportunities at the Sub-national Level in Latin America and the Caribbean In contrast, the United States enacted slowly than whites, hence widening the a highly conditional national education black-white education achievement gap. program in 2001 through the No Child If the HOI for educational achievement Left Behind Act. This act transfers were estimated based on the standardized funds from the federal government and achievement test (as in Chapter 3), and requires states to establish goals for all if grants are conditional on these indices, students and for groups of students some of the problems associated with characterized by race, ethnicity, poverty, inequality in the increased achievement disability, and limited English proficiency, scores could be avoided. and requires schools to make annual Along these lines, the WDR 2009 progress in meeting these goals. If stressed the need for “collecting and goals are not met as determined by test disseminating credible information on standards, the school and the state are service entitlements to increase the subject to sanctions.20 If the performance accountability of service providers and of local governments and schools does improve outcomes� at the sub-national not improve, transfers can be made level. Collecting HOIs periodically can directly to the individual, making sure serve this purpose, and they are an her opportunities are enhanced. optimal index to target and condition While No Child Left Behind has been federal grants, at minimum by publicizing controversial, Hanushek and Raymond performance. Until now, the only data (2005) find that despite flaws the act available has been on poverty rates, has a positive impact on achievement. income per capita, and other outcome However, the impact holds just for measures. While useful, these cannot states attaching consequences to help policies that focus on improving performance.21 This is an important equitable access to basic services, such finding to take into account for countries that children in poorer regions have designing an equal opportunity grant. equal access to opportunities that are The authors also found that although the critical to allow them to develop their grant increased average performance, potential. the performance of blacks increased more Based on the stylized facts about 20 A school failing to make adequate progress for three consecutive years must initiate a performance improvement plan and also give students the option to move to other public schools. A fourth year of failure requires restructuring and supplemental education services. If a school fails to make progress in the fifth year, it must implement restructuring, including changes in staff and management or converting into a semi-private (charter) school. The district must provide transportation to the new school. The state must permit low-income students attending persistently failing schools to use special funds to obtain supplemental educational services from the public or private providers selected by the students and their parents. 21 In the authors’ words, “States that simply provide better information through report cards without attaching consequences to performance do not get significantly larger impacts over no accountability.� 116 Human Opportunities at the Sub-national Level in Latin America and the Caribbean the evolution of HOIs for different be lower for those locations, and opportunities and across regions, the hence the need to equalize with available literature on grant design and even higher expenditure per capita. expenditure decentralization, and the • For ease of monitoring and institutional and political context in most implementation, the grants should LAC countries, several principles should be be divided by key good or service. considered for equal opportunity grants: Canada enacted in 1996 a joint • The grant can be distributed conditional grant (CHST) for health, to different levels of government education and social minimum depending on the specific institutional standards of coverage and insurance structure of the country—unitary, by province, but later subdivided it federal or the myriad of intermediate to better monitor compliance and options. Functions and responsibilities better calculate amounts for each of each level of government in key good or service. equalizing opportunities should • The grant is not meant to replace be assigned clearly. In the case of an equalization transfer. As noted housing, and also to some extent in by Vaillancourt and Bird (2004), education, responsibilities are often the central aim of an equalization shared among national, provincial, transfer is to enable sub-national state as well as private enterprises. governments with different abilities to This mixed approach to service raise revenues to provide comparable delivery can create problems of levels of services. Since no country is accountability and quality, possibly completely uniform, a fundamental explaining the lack of better results in characteristic of a decentralized state equalizing basic service opportunities is that sub-national governments across regions. have different fiscal capacities and • The expenditure “need� for each are unable to provide the same opportunity to access a key good or level of public services at the same service should be calculated. The cost tax rates. Equalization transfers are is likely to vary inversely with the mostly unconditional and thus permit specific coverage rate and directly regional differences. The equal with inequality of opportunity: in opportunity grant can complement more remote locations the cost equalization transfers by conditioning per capita is expected to be higher the grant on actually attaining and it is also expected that other national target goals for service circumstances such as family equitable service provision. In fact, education or family income would the central government in Canada has 117 Human Opportunities at the Sub-national Level in Latin America and the Caribbean two main transfers, the unconditional assurance that nationally important equalization transfer and the CHST, goods and services are adequately a major conditional transfer to provided once they have been guarantee horizontal equity. Australia decentralized. An important part has also a conditional grant called of any decentralization program is Special Purpose Payments (SPP) and thus improved national evaluation a formal equalization program that capacity. is mostly unconditional. About half • Above all, a widely publicized of transfers are through the SPP to commitment to leveling the playing achieve national policy objectives, field across all regions of the country mainly for education and health care. is essential. All individuals should be • This type of grant would be aware of this and should periodically better as a performance-based receive data on the progress of grant rather than one conditioned different local governments in on inputs. The grant should be achieving the agreed goals. When conditional on gradually attaining public commitment or enthusiasm equality of opportunity across wanes and conditionality becomes locations and within each location. lax, then the grant ends up becoming It should reward good performers an unconditional grant without any and penalize bad performers, result or improvement (see Zanetta, without punishing inhabitants. 2004 for an example). The government should be liable, not the individual. To achieve this, 4.4. Summary and Conclusions each country should evaluate the advantages and disadvantages This chapter estimates a series of sub- of conditionality ranging from a national Human Opportunity Indices minimum of publicizing results up (SN HOIs) using data from 30 household to enacting penalties of varying surveys for 15 Latin American and degrees. Caribbean (LAC) countries over a period • Whatever conditionality is em- of more than a decade (1995-2009). ployed, the systematic collection, Together, the surveys represent more analysis, and reporting of information than 160 sub-regions. Using the same is important to verify compliance opportunities for accessing key services, with stated goals and to assist future circumstances and overall methodology decisions (Bird, 2000). Unless central as in Chapter 2 for building the national agencies monitor and evaluate indices, the chapter seeks to uncover local performance, there can be no some basic stylized facts on regional 118 Human Opportunities at the Sub-national Level in Latin America and the Caribbean inequality of opportunities and outlines as Argentina, Chile, Colombia and a possible instrument to redress this Costa Rica). A strong negative and inequality. The main findings are: statistically significant relationship • The range for SN HOIs is was found between the rate of much higher than the range of growth of the 165 SN HOIs and national HOIs: from 96 for Tierra the initial level of the SN HOI. This del Fuego, Argentina to 29 in the convergence was stronger in the case Atlantic Region, Nicaragua and 31 in of the education HOI, suggesting Morazán, El Salvador. Capital cities that it is more challenging to level the tended to have higher HOIs than playing field in housing compared to the rest of their respective countries. education. Moreover, the higher the HOI, the • Sub-regional differences in narrower the gap between access to accessing key services tend to opportunities in capital cities and the decrease as the overall level of the rest of the country. HOI increases. In most countries • Overall, disparities in accessing the dispersion in the sub-national basic services within most LAC HOIs decreased between circa 1995 countries reflect differences in and 2008. Moreover, measuring housing SN HOIs more than inequality in SN HOIs among sub- differences in education SN HOIs. regions through the weighted Housing SN HOIs have a higher standard deviation, countries with range than education SN HOIs, and the highest HOI tended to have the the differences between capital cities lowest dispersion. Since the average and the rest of the country are much education HOI is higher than the more pronounced for housing than average housing HOI, there was less for education. All capital cities ranked dispersion in the education HOI than better than the rest of the country in the housing HOI. for housing, but this was not always • Higher HOIs, expenditure the case for education. decentralization, and small country • Access to basic services appears size are associated with less sub- to be converging over time within regional HOI dispersion. While the regions of a country. The highest data was not conclusive, it suggests increases in the HOIs were attained that political decentralization may mostly by sub-regions with the also be correlated with more regional lowest initial HOI, and the lowest equality. growth was attained by the capitals To help level the playing field, the (mostly of high HOI countries, such chapter offers some guiding principles 119 Human Opportunities at the Sub-national Level in Latin America and the Caribbean to design a performance-based receive equal opportunities to access “equal opportunity� grant. This basic services, thereby accelerating grant would promote government convergence towards universal accountability and would help coverage. ensure that children in each province 120 121 ANNEXES Chapter 1 122 ANNEXES CHAPTER 1 Annex 1: A Numerical Example of the service in that group (Table A1.1c). Computing the HOI Third, the penalty is obtained by dividing the sum of the opportunity To help explain the computation of the gaps of all vulnerable groups (called the HOI, we use the example presented overall opportunity gap) by the total in Tables A1.1a-1i (below), in which population. In our example the penalty the overall population is divided in 16 would be equal to 10 percent, since the circumstance groups, defined by gender, sum of all opportunity gaps equals 160 race and location. We assume that all (Table A1.1c) and the overall population groups are the same size: 100 persons, is 1600. The overall coverage rate of 25 leading to a total population of 1600. percent minus the penalty due unequal Table A1.1a presents the number of allocation of 10 percent leads to an HOI people in each circumstance group that of 15 percent (C-P = O; 25-10 = 15). have access to a specific basic service In this example, only 90 individuals (for example, clean water, electricity from opportunity-vulnerable groups are or vaccinations). Overall, 400 people covered, out of 250 that should have have access to the service. Since the been covered had equality of opportunity total population is 1600, the overall prevailed. Hence, 160 people among coverage rate is 25 percent. Under these vulnerable groups should receive equality of opportunity, 25 percent of the service for their coverage rate to each circumstance group should be rise at least to the average. This total covered. Coverage rates, however, vary opportunity gap, 160, equals 10 percent substantially, from 0 percent in some of the total population. groups to 75 percent in others (Table A1.1b). Since services are available for 400 The computation of the penalty is done people, and 160 people in non- in three steps. First, we identify all vulnerable groups receive services in circumstance groups with coverage rate excess of what is needed for equality of below the average rate (25 percent, in opportunity to prevail, only 240 people our example); we refer to them as the received services that were allocated opportunity-vulnerable groups. There equitably. Hence, as a proportion of the are 10 of these groups in our example, total population, the number of people marked in red in Table A1.1b. Second, we who receive the service according to the compute the gap between the number principle of equality of opportunities is of people in each vulnerable group that 15 percent (240/1600). This is what the should have access to the service for it to HOI measures. reach the average coverage rate, and the actual number of people with access to 123 Table A1.1 Table A1.1a: Distribution of a service Table A1.1d: Distribution of the population Table A1.1g: Improperly allocated services North South North South North South Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Whites 75 60 45 20 Whites 100 100 100 100 Whites 50 35 20 0 Men Blacks 50 50 30 20 Men Blacks 100 100 100 100 Men Blacks 25 25 5 0 Whites 15 10 5 0 Whites 100 100 100 100 Whites 0 0 0 0 Women Blacks 15 5 0 0 Women Blacks 100 100 100 100 Women Blacks 0 0 0 0 Table A1.1b: Group-Specific Coverage Rates Table A1.1e: Opportunity gaps as a proportion of T able A1.1h: Improperly allocated services as a the total population proportion of the total population North South North South North South Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Whites 75% 60% 45% 20% Whites 0% 0% 0% 0% Whites 3% 2% 1% 0% Men Blacks 50% 50% 30% 20% Men Blacks 0% 0% 0% 0% Men Blacks 2% 2% 0% 0% Whites 15% 10% 5% 0% Whites 1% 1% 1% 2% Whites 0% 0% 0% 0% Women Blacks 15% 5% 0% 0% Women Blacks 1% 1% 2% 2% Women Blacks 0% 0% 0% 0% Table A1.1c: Opportunity Gaps (for vulnerable Table A1.1f: Opportunity gaps as a proportion of Table A1.1i: Improperly allocated services as a groups) the population covered proportion of the population covered North South North South North South Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Whites 0 0 0 5 Whites 0% 0% 0% 1% Whites 13% 9% 5% 0% Men Blacks 0 0 0 5 Men Blacks 0% 0% 0% 1% Men Blacks 6% 6% 1% 0% Whites 10 15 20 25 Whites 3% 4% 5% 6% Whites 0% 0% 0% 0% Women Blacks 10 20 25 25 Women Blacks 3% 5% 6% 6% Women Blacks 0% 0% 0% 0% 124 ANNEXES CHAPTER 1 Annex 2: A Numerical Illustration Using the same procedure and the of the Decomposition of the HOI information for population B in the last row of Table A2.4c, we obtain an HOI The HOI can be easily computed from of 37 percent for population B. One can the distribution of the population by think of population A and B as being circumstance groups (Table A2.1a) and the same country in two different time specific coverage rates (Table A2.1b) periods. Population B has an HOI 14.5 through four simple steps (Table A2.1c). percentage points higher than population First, we compute the overall coverage A. This difference is due to disparities in rate as the weighted average of all both the distribution of the populations group-specific coverage rates (35 percent among circumstance groups and the in population A). Second, we identify pattern of their specific coverage rates. groups with specific coverage rates below In population B, the north region has a average (marked in red). Third, we obtain 20 percentage-point higher population the penalty (12.5 percent in population A) share. In terms of coverage rates, those by weighting the difference between the among the vulnerable groups tend to be overall coverage and each group-specific higher in population B, whereas those coverage rate among vulnerable groups: among the non-vulnerable groups are (35-15)*0.25+(35-5)*0.25=12.5. higher in population A. Coverage rates Finally, we obtain the HOI (22.5 percent are on average higher and less unequal in population A) by subtracting the in population B than in population A. penalty from the overall coverage. 125 Table A1.2 Table A2.1a: Distribution of the populatio Table A2.1b: Group-specific coverage rate Table A2.1c: Computing the HOI for A for population A population A 100% North South 35% North South Overall coverage rate 35.0% Men 25% 25% Men 75% 45% Penalty 12.5% Women 25% 25% Women 15% 5% HOI 22.5% Table A2.2a: Distribution of the populatio Table A2.2b: Group-specific coverage rate Table A2.2c: Computing the HOI for B for population A population B with the coverage rates from population A 100% North South 40% North South Overall coverage rate 40.0% Men 40% 10% Men 75% 45% Penalty 14.5% Women 30% 20% Women 15% 5% HOI 25.5% Table A2.3c: Computing the HOI for Table A2.3b: Group-specific coverage rate Table A2.3a: Distribution of the populatio population B with the structure of coverage rates from (structure from population A, average leve B population A and average level from population B) from population B 100% North South 45% North South Overall coverage rate 45.0% Men 40% 10% Men 84% 51% Penalty 16.3% Women 30% 20% Women 17% 6% HOI 28.7% Table A2.4a: Distribution of the populatio Table A2.4b: Group-specific coverage rate Table A2.4c: Computing the HOI for B for population B population B 100% North South 45% North South Overall coverage rate 45.0% Men 40% 10% Men 65% 40% Penalty 8.0% Women 30% 20% Women 40% 15% HOI 37.0% 126 ANNEXES CHAPTER 1 To isolate the effect on the HOI of preserved while the average coverage differences in the distribution of rate is adjusted. To reach the average population among circumstance groups, coverage rate of population B, all group- we estimate the HOI for a combined specific coverage rates from population situation: coverage rates are still those A are multiplied by the ratio between the of population A, but the distribution overall coverage for population B and A of the population among circumstance (45/40=1.125).1 groups is now that of population B. The As a consequence, the HOIs in Tables estimated HOI for this hybrid situation is A2.3c and A2.4c share the same 25.5 percent. distribution of the population among Since this hybrid population shares with circumstance groups and the same population A the same group-specific overall coverage rate (45 percent). They coverage rate, their difference in HOI (3.0 only differ with respect to the inequality p.p.) is entirely due to their differences of their group-specific coverage rates. in the distribution of the population Since the overall coverage rate is the among circumstance groups. This is the same in both cases, the difference in composition effect. On the other hand, inequality is captured by corresponding since the hybrid population shares with differences in the size of the penalty: population B the same distribution of 16.3 percent in Table A2.3c and 8 the population among circumstance percent in Table A2.4c. The difference groups, their difference in HOI (11.5 between these two penalties, 8.3 p.p., is p.p.) is entirely due to their differences a measure of the equalization effect. It is in their specific coverage rates. Hence, the contribution of the greater equality this difference is the coverage effect. of opportunities in population B to the Hence, the total difference in HOI difference in HOI between the two (14.5 p.p.) between populations A and populations. B is decomposed into the composition On the other hand, Tables A2.2c and effect of 3 p.p. and 11.5 p.p. due to the A2.3c share the same distribution of coverage effect. the population and the same inequality The coverage effect can be decomposed among group-specific coverage rates further (see Tables A2.3a-c). The as measured by the ratio between the group-specific coverage rates in Table penalty and the overall coverage rate. A2.3b are constructed to reach the This ratio is equal to 0.3625 in the two average level in population B, holding cases. Hence, the difference between the inequality level of population A. HOIs in these tables (3.2 p.p.) is entirely This is accomplished by proportionally due to their corresponding differences in increasing all group-specific rates from overall coverage rates. It represents the population A. Inequality would be scale effect. 1 Notice that the relevant average for population A is not the original average but that using the population weights from population B (see Table A2.2c). 127 Annex 3: The Algebra of Decomposing the Human Opportunity Index 128 129 Annex 4: Definitions of Indicators Indicator Scope Definition Housing Children 0 to 10 years This variable takes the value of one if the household has access to running water within the Adequate access to water dwelling. Thus access includes public network connections and all water pumped into the old dwelling even if it is not from the public network. Children 0 to 10 years This variable takes the value of one if the dwelling has access to electricity from any source. Thus Access to electricity old sources can range from the electrical grid system to solar panels. This variable takes the value of one if the dwelling has access to a flush toilet (either inside the Children 0 to 10 years dwelling or inside the property) that is connected to any mechanism whereby household waste Adequate access to sanitation old is allowed to flow away from the dwelling. Education Children 10 to 14 This is the measured as children aged 10-14 attending to school, independent of grade. This School attendance rate years old variable measures the gross attendance rate. This is measured by computing the probability of having ended sixth grade on time for all children Probability of completing Children 12 to 16 ages 12 to 16. In most countries of the region, this means having completed primary education. sixth grade on time years old Given that on average children start school at the age of 7, by age 13, students that have survived in the system without repetition should have completed six years of basic education. Source: The World Bank and Universidad Nacional de La Plata (CEDLAS) Socioeconomic Database for Latin America and the Caribbean. 130 ANNEXES CHAPTER 1 Annex 5: Choosing the one needs to know the degree of Aggregation Sequence complementarity or substitutability between goods and services as well as To create an overall HOI, one needs to the distribution of a population’s access aggregate both the different dimensions opportunities.Ultimately,aggregationtakes of each person’s opportunities (that into consideration the complementarities is, the indicators used to proxy a basic and substitutability among the many opportunity) and also aggregate the dimensions. opportunities of different persons to It may be useful to consider the extreme obtain a societal aggregate. Ideally, one case in which the access opportunity is first should aggregate the opportunities either 0 percent or 100 percent. Suppose of each person first, to fully consider I1 indicates whether everyone has (I1=1) the interdependence among the or not (I1=0) proper access to sanitation dimensions. and I2 indicates whether everyone has This interdependence among the (I2=1) or not (I2=0) adequate access to dimensions has two features worth electricity. Three alternative aggregation noting. First, the dimensions could be strategies are possible in this example. complements or substitutes. For instance, The most demanding alternative would should attendance and progression consider that one has an opportunity in school be considered complements when he/she has access to both sanitation or substitutes? Second, the access to and electricity. In this case, the aggregated different goods and services could be opportunity index would be: I=I1xI2 and concomitant or alternative. For instance, I=1 if and only if I1=1 and I2=1. This an overall 50 percent coverage rate is the intersection approach. In the other for water and electricity could still lead extreme, the least demanding alternative to very distinct distributions. In one would consider the two opportunities extreme, it may be the case that those as substitutes. In this case having one of having access to water are the same them is enough. Access to a second would people having access to electricity. In not lead to any significant improvement. this case, the accesses to these two In this case, the aggregated opportunity services are perfectly correlated. On index would be: I=1-(1-I1)*(1-I2) and the other extreme, it may be that either I=1 if and only if I1=1 or I2=1. This is one has access to water or to electricity, the union approach. An intermediate never to both. In this case, access to alternative is to relax the requirement that these two services is inversely related. both are essential and, instead of going to If the two services are substitutes, an the extreme that just one is enough, one inverse relation is preferable. If they are may consider each one as an independent complements, a positive association is advantage. In this case, the aggregated preferable. opportunity index would be: I=I1+I2. Hence, to properly evaluate opportunities This is the counting approach. 131 ANNEXES CHAPTER 1 To properly take the distribution of each dimension (like the opportunity access opportunity and degree of of access to water) across people, and complementarity/substitutability into later aggregate dimensions. Certainly, consideration, the many dimensions of we could aggregate first a proxy of the opportunities faced by each person must five indicators consider in this study, and be aggregated before aggregating among we plan to do this exercise in the very people. Once the aggregation across near future. However, one of the main persons of each dimension is conducted, advantages of the HOI is its flexibility how positively or negatively correlated to track progress in a variety of basic are the accesses to the many relevant key goods and services. In country specific goods and services is lost. works we can combine information from A comprehensive consideration of the different sources such as health surveys, inter-relation among dimensions also living standard measurements surveys, demands that information on the many student assessment test scores, among dimensions be available for each person. others. For keeping this flexibility, we To properly consider inter-dependence, would like to have as our base procedure for each person in the analysis one needs a simple methodology that will allow us simultaneous information on his/her to combine information from different access to all key goods and services being sources in building an overall HOI. considered. Since many opportunities are If the indicators are continuous, a age-specific, like attending school at age variety of other alternative aggregation 10 and completing 6th grade at age 13, procedures would be available. These a longitudinal survey or a survey with alternatives would include weighted considerable amount of retrospective versions of the three approaches information would be required. introduced above as well as completely Given the lack of such information different function forms. For instance, an among Latin American countries and expression mimicking a constant elasticity for simplicity, we opted for reversing the of substitution function could be used. order of aggregation. We first aggregate 132 133 ANNEXES Chapter 2 134 Table A2.1: Surveys Used to calculate the HOI Description of the surveys used to calculate the HOI Country Circa 1995 Circa 2010 Coverage Survey Argentina 1998 2008 Urban Encuesta Permanente de Hogares Bolivia 2007 National Encuesta Continua de Hogares Brazil 1995 2008 National Pesquisa Nacional por Amostra de Domicilios Chile 1996 2006 National Encuesta de Caracterización Socioecononómica Nacional Colombia 1997 2008 National Encuesta de Calidad de Vida Costa Rica 1994 2009 National Encuesta de Hogares de Propósitos Múltiples Dominican Republic 2000 2008 National Encuesta Nacional de Fuerza de Trabajo Ecuador 1995 2006 National Encuesta de Condiciones de Vida El Salvador 1998 2007 National Encuesta de Hogares de Propósitos Múltiples Guatemala 2000 2006 National Encuesta Nacional sobre Condiciones de Vida Honduras 1999 2006 National Encuesta Permanente de Hogares de Propósitos Múltiples Jamaica 1996 2002 National Jamaica Survey of Living Conditions Mexico 1996 2008 National Encuesta Nacional de Ingresos y Gastos de los Hogares Nicaragua 1998 2005 National Encuesta Nacional de Hogares sobre Medición de Nivel de Vida Panama 1997 2003 National Encuesta de Niveles de Vida Paraguay 1999 2008 National Encuesta Permanente de Hogares Peru 1998 2008 National Encuesta Nacional de Hogares Uruguay 2006 2008 National Encuesta Nacional de Hogares Ampliada Venezuela, R.B de 1995 2005 National Encuesta de Hogares por Muestreo Source: The World Bank and Universidad de La Plata (CEDLAS) Socioeconomic Database for Latin America and the Caribbean. 135 Table A2.2: Overall Human Opportunity Index and Decomposition in Latin American and Caribbean Countries (1995 and 2010) Overall Human Opportunity Index and Decomposition in Latin American and Caribbean Countries (1995 and 2010) Opportunity index (%) Decomposition (p.p.) Decomposition (%) Extrapolation (%) If the Considering Coverage Effect Coverage Effect Country Circa population of the inequality Annual rate of Composition Composition Circa 2010 Equalization Equalization 1995 2010 1995 2010 lived in of 2010 and change (p.p.) Effect Total Scale Effect Effect Total Scale Effect 1995 the average Effect Effect Argentina 86.1 90.6 89.4 88.3 0.2 0.45 -0.23 -0.12 -0.11 204 -104 -55 -49 85.4 88.8 Brazil 57.0 70.7 72.8 75.7 1.4 1.05 0.39 0.16 0.23 73 27 11 16 57.0 78.6 Chile 83.0 89.0 90.1 91.6 0.9 0.60 0.26 0.11 0.16 70 30 12 18 82.2 95.1 Colombia 67.1 73.3 75.4 78.7 1.1 0.56 0.50 0.20 0.30 53 47 19 28 65.0 80.9 Costa Rica 77.3 83.2 84.5 88.2 0.7 0.39 0.34 0.09 0.25 54 46 12 34 78.0 88.9 Dominican Republic 63.8 73.6 73.6 72.7 1.1 1.23 -0.12 0.00 -0.12 110 -10 0 -11 58.2 74.9 Ecuador 59.8 46.1 50.0 75.7 1.4 -1.25 2.70 0.36 2.34 -86 186 25 161 59.8 81.5 El Salvador 43.9 38.3 41.3 52.8 1.0 -0.62 1.61 0.33 1.28 -62 162 33 129 40.9 55.8 Guatemala 42.9 48.0 49.2 51.1 1.4 0.85 0.51 0.19 0.32 62 38 14 24 36.1 56.5 Honduras 41.7 44.8 45.1 47.6 0.8 0.44 0.39 0.04 0.36 53 47 4 43 38.4 50.9 Jamaica 79.5 79.7 79.7 81.2 0.3 0.04 0.25 0.00 0.25 15 85 -1 86 79.2 83.5 Mexico 65.4 81.9 83.2 86.3 1.7 1.37 0.37 0.11 0.25 79 21 7 15 63.6 89.7 Nicaragua 35.0 38,6 39.8 46.3 1.6 0.52 1.10 0.17 0.92 32 68 11 57 30.1 54.3 Panama 65.7 69.5 69.6 68.6 0.5 0.63 -0.15 0.02 -0.17 130 -30 4 -34 64.7 72.0 Paraguay 60.7 69.5 68.6 71.0 1.1 0.97 0.17 -0.10 0.27 85 15 -9 24 56.1 73.3 Peru 54.8 60.5 63.4 68.7 1.4 0.57 0.82 0.28 0.54 41 59 20 39 50.7 71.5 Uruguay 89.0 91.3 91.0 90.5 0.7 1.12 -0.41 -0.15 -0.26 157 -57 -21 -36 81.2 91.9 Venezuela, R.B de 82.1 91.3 90.2 86.6 0.5 0.92 -0.47 -0.11 -0.35 204 -104 -25 -79 82.1 88.9 LAC Average 64.2 68.9 69.8 73.4 1.0 0.5 0.4 0.1 0.3 55 45 10 35 61.7 76.5 Source: Author's calculations based on household surveys 136 Table A2.3: Human Opportunity Index in Education and Decomposition Latin American and Caribbean Countries: National, 1995 and 2010 Opportunity index in Education and Decomposition Latin American and Caribbean countries: National, 1995 and 2010 Opportunity index (%) Decomposition (p.p.) Decomposition (%) Extrapolation (%) Countries Considering the Annual rate Coverage Effect Coverage Effect Circa If the population of Circa Composition Composition inequality of 2010 of change Equalization Equalization 1995 2010 1995 2010 lived in 1995 2010 Effect Total Scale Effect Effect Total Scale Effect and the average (p.p.) Effect Effect Argentina 90.6 93.8 93.2 89.7 -0.1 0.32 -0.41 -0.06 -0.35 -328 428 66 362 90.9 89.5 Brazil 50.9 60.3 62.6 66.1 1.2 0.72 0.45 0.18 0.27 61 39 15 23 50.9 68.4 Chile 85.6 89.0 89.5 90.2 0.5 0.34 0.12 0.05 0.07 73 27 11 16 85.1 92.1 Colombia 67.9 71.9 75.1 81.3 1.2 0.36 0.85 0.30 0.56 30 70 24 46 65.5 83.7 Costa Rica 70.8 77.5 78.3 81.0 0.7 0.45 0.23 0.05 0.18 66 34 8 26 71.5 81.6 Dominican Republic 67.7 74.1 74.7 74.9 0.9 0.80 0.11 0.08 0.03 88 12 9 3 63.1 76.7 Ecuador 71.8 71.7 74.3 82.7 1.0 -0.01 1.00 0.23 0.76 -1 101 24 77 71.8 86.6 El Salvador 54.6 51.5 54.3 65.9 1.3 -0.34 1.60 0.31 1.29 -27 127 25 103 50.8 69.7 Guatemala 45.2 49.2 50.5 52.4 1.2 0.68 0.54 0.21 0.33 56 44 17 27 39.1 57.3 Honduras 52.9 55.7 57.4 63.5 1.5 0.40 1.11 0.23 0.88 27 73 15 58 46.9 69.6 Jamaica 92.1 92.6 92.9 94.0 0.3 0.09 0.22 0.04 0.18 29 71 12 59 91.8 96.5 Mexico 76.0 86.7 87.7 89.6 1.1 0.88 0.24 0.08 0.16 79 21 7 14 74.9 91.8 Nicaragua 49.5 52.5 53.9 59.0 1.4 0.43 0.93 0.19 0.74 31 69 14 55 45.5 65.8 Panama 78.3 81.3 81.3 80.7 0.4 0.49 -0.09 0.00 -0.09 123 -23 1 -24 77.6 83.5 Paraguay 68.1 74.0 74.3 74.1 0.6 0.59 0.01 0.03 -0.02 98 2 5 -3 66.3 75.3 Peru 71.9 76.1 78.3 84.6 1.3 0.42 0.85 0.22 0.63 33 67 17 49 68.1 87.1 Uruguay 85.6 87.1 87.0 86.6 0.5 0.72 -0.24 -0.06 -0.18 150 -50 -13 -37 80.3 87.6 Venezuela, R.B de 77.1 85.5 85.4 84.0 0.7 0.84 -0.15 -0.01 -0.13 121 -21 -2 -20 77.1 87.5 LAC Average 69.8 73.9 75.0 77.8 0.9 0.5 0.4 0.1 0.3 53 47 13 34 67.6 80.6 Source: Author's calculations based on household surveys 137 Table A2.4: Human Opportunity Index in Housing Conditions and Decomposition in Latin American and Caribbean Countries: National, 1995 and 2010 Opportunity index in Housing Conditions and Decomposition in Latin American and Caribbean countries: National, 1995 and 2010 Opportunity index (%) Decomposition (p.p.) Decomposition (%) Extrapolation (%) Considering the Countries If the population Coverage Effect Coverage Effect inequality of 2010 Annual rate of Composition Composition Circa 1995 of 2010 lived in Circa 2010 1995 2010 and the average change (p.p.) Effect Effect 1995 Equalization Scale Equalization Scale probability of 1995 Total Total Effect Effect Effect Effect Argentina 81.5 87.4 85.7 86.9 0.5 0.59 -0.05 -0.18 0.13 109 -9 -33 24 79.9 88.0 Brazil 63.2 81.1 83.0 85.3 1.7 1.38 0.32 0.14 0.18 81 19 8 10 63.2 88.7 Chile 80.5 89.1 90.7 93.1 1.3 0.86 0.40 0.16 0.24 68 32 13 19 79.2 98.1 Colombia 66,3 74.6 75.7 76.2 0.9 0.76 0.14 0.10 0.04 84 16 11 5 64.5 78.0 Costa Rica 83.8 88.9 90.7 95.5 0.8 0.34 0.44 0.12 0.32 43 57 16 41 84.6 96.3 Dominican Republic 59.9 73.1 72.5 70.4 1.3 1.66 -0.34 -0.07 -0.26 126 -26 -6 -20 53.3 73.1 Ecuador 47.7 20.4 25.7 68.8 1.9 -2.48 4.40 0.48 3.91 -130 230 25 205 47.7 76.4 El Salvador 33.1 25.1 28.2 39.7 0.7 -0.90 1.62 0.35 1.27 -124 224 49 175 31.0 41.8 Guatemala 40.7 46.8 47.8 49.7 1.5 1.02 0.49 0.17 0.32 68 32 11 21 33.2 55.8 Honduras 30.5 33.9 32.8 31.6 0.2 0.48 -0.33 -0.16 -0.17 320 -220 -105 -114 29.9 32.2 Jamaica 66.8 66.8 66.5 68.4 0.3 0.00 0.27 -0.05 0.32 -1 101 -17 118 66.5 70.5 Mexico 54.7 77.1 78.8 83.0 2.4 1.86 0.49 0.15 0.35 79 21 6 15 52.4 87.7 Nicaragua 20.4 24,7 25.8 33.5 1.9 0.61 1.26 0.16 1.10 33 67 8 59 14.8 42.9 Panama 53.0 57.6 57.9 56.4 0.6 0.78 -0.20 0.04 -0.24 135 -35 6 -41 51.8 60.5 Paraguay 53.3 64.9 66.8 67.4 1.4 1.16 0.25 0.19 0.06 83 17 13 4 49.1 70.2 Peru 37.8 45.0 48.4 52.9 1.5 0.72 0.79 0.34 0.44 48 52 23 29 33.3 55.9 Uruguay 92.4 95.5 95.0 94.3 0.9 1.52 -0.57 -0.23 -0.34 161 -61 -25 -36 82.0 96.2 Venezuela, R.B de 87.1 97.1 95.0 89.2 0.2 1.00 -0.79 -0.22 -0.58 475 -375 -103 -272 87.1 90.3 LAC Average 58.5 63.8 64.8 69.0 1.1 0.6 0.5 0.1 0.4 92.0 8.0 -5.4 13.4 55.7 72.4 Source: Author's calculations based on household surveys 138 Table A2.5: D-Index for Completing Sixth Grade on Time, by Circumstance, Circa 2008 (percent) D-Index for Completing Sixth Grade on Time, by Circumstance, circa 2008 (percent) Gender of Househol Country d Parent's education Gender Head Per Capita Income Urban or Rural Presence of Parents Number of Siblings Overall D-Index Argentina 1.1 0.8 0.2 0.5 0.2 0.7 3.9 Bolivia 3.0 0.5 1.8 0.8 0.5 2.2 1.6 6.4 Brazil 9.0 7.6 0.3 7.2 1.6 2.2 2.6 19.1 Chile 0.7 0.3 0.4 0.3 0.0 0.1 0.6 3.0 Colombia 4.8 3.2 0.5 1.0 2.9 1.8 1.9 9.9 Costa Rica 3.5 1.7 1.0 0.5 0,6 1.3 2.6 7.7 Dominican Republic 7.7 5.2 0.6 3.6 0.6 0.3 2.6 13.5 Ecuador 3.0 0.3 0.2 1.4 0.1 0.1 1.3 5.9 El Salvador 8.3 3.0 2.1 3.6 3.7 0.3 3.9 15.9 Guatemala 18.7 2.6 7.7 4.9 8.6 5.7 6.5 26.6 Honduras 8.2 3.2 1.0 3.0 3.1 1.1 2.8 15.5 Jamaica 0.0 0.2 0.0 0.2 0.0 0.3 0.5 0.8 Mexico 1.6 0.8 0.5 0.3 0.3 0.5 0.8 3.4 Nicaragua 13.1 7.3 1.8 6.3 10.6 0.1 7.7 24.2 Panama 3.5 2.4 0.5 2.1 1.0 0.3 2.8 8.7 Paraguay 5.4 3.4 0.2 2.6 0.7 1.5 3.3 10.4 Peru 3.3 0.5 0.4 1.6 0.7 0.4 1.4 7.3 Uruguay 1.8 1.1 0.1 0.8 0.2 0.5 1.4 5.6 Venezuela, R.B de 2.9 1.9 0.0 0.6 0.3 0.3 2.0 6.1 LAC Average 5.2 2.4 1.0 2.2 2.0 1.0 2.5 10.2 Source: Author's calculations based on household surveys Table A2.6: D-Index for School Enrollment Ages 10-14, by Circumstance, Circa 2008 (percent) D-Index for Completing Sixth Grade on Time, by Circumstance, circa 2008 (percent) Gender of Household Country Parent's education Gender Head Per Capita Income Urban or Rural Presence of Parents Number of Siblings Overall D-Index Argentina 1.1 0.8 0.2 0.5 0.2 0.7 3.9 Bolivia 3.0 0.5 1.8 0.8 0.5 2.2 1.6 6.4 Brazil 9.0 7.6 0.3 7.2 1.6 2.2 2.6 19.1 Chile 0.7 0.3 0.4 0.3 0.0 0.1 0.6 3.0 Colombia 4.8 3.2 0.5 1.0 2.9 1.8 1.9 9.9 Costa Rica 3.5 1.7 1.0 0.5 0,6 1.3 2.6 7.7 Dominican Republic 7.7 5.2 0.6 3.6 0.6 0.3 2.6 13.5 Ecuador 3.0 0.3 0.2 1.4 0.1 0.1 1.3 5.9 El Salvador 8.3 3.0 2.1 3.6 3.7 0.3 3.9 15.9 Guatemala 18.7 2.6 7.7 4.9 8.6 5.7 6.5 26.6 Honduras 8.2 3.2 1.0 3.0 3.1 1.1 2.8 15.5 Jamaica 0.0 0.2 0.0 0.2 0.0 0.3 0.5 0.8 Mexico 1.6 0.8 0.5 0.3 0.3 0.5 0.8 3.4 Nicaragua 13.1 7.3 1.8 6.3 10.6 0.1 7.7 24.2 Panama 3.5 2.4 0.5 2.1 1.0 0.3 2.8 8.7 Paraguay 5.4 3.4 0.2 2.6 0.7 1.5 3.3 10.4 Peru 3.3 0.5 0.4 1.6 0.7 0.4 1.4 7.3 Uruguay 1.8 1.1 0.1 0.8 0.2 0.5 1.4 5.6 Venezuela, R.B de 2.9 1.9 0.0 0.6 0.3 0.3 2.0 6.1 LAC Average 5.2 2.4 1.0 2.2 2.0 1.0 2.5 10.2 Source: Author's calculations based on household surveys 139 Table A2.7: D-Index for Access to Adequate Sanitation, by Circumstance, Circa 2008 (percent) D-Index for Access to Adequate Sanitation, by Circumstance, circa 2008 (percent) Gender of Household Country Parent's education Gender Head Per Capita Income Urban or Rural Presence of Parents Number of Siblings Overall D-Index Argentina 6.6 0.8 0.5 6.6 1.0 1.1 11.7 Bolivia 15.3 0.6 4.0 13.0 29.0 1.5 3.9 33.0 Brazil 2.0 0.0 0.4 2.5 4.5 0.0 0.3 8.7 Chile 1.8 0.2 0.1 2.6 4.0 0.4 0.3 5.2 Colombia 6.4 0.2 1.0 3.7 12.7 1.6 2.2 15.6 Costa Rica 0.9 0.0 0.2 1.1 0.6 0.2 0.2 2.6 Dominican Republic 10.7 0.2 2.1 10.0 16.2 2.9 2.0 20.5 Ecuador 7.2 0.4 0.4 11.4 12.9 1.4 3.3 21.1 El Salvador 18.5 0.0 3.3 27.7 33.3 4.7 4.0 40.5 Guatemala 17.7 0.1 10.6 19.6 35.2 4.5 4.6 40.8 Honduras 14.1 0.7 1.3 17.7 23.6 2.1 0.6 34.7 Jamaica 0.1 0.0 0.2 0.1 0.1 0.0 0.0 0.3 Mexico 2.7 0.0 0.7 2.7 7.5 0.1 1.1 13.1 Nicaragua 8.1 0.3 1.6 8.3 14.0 2.0 0.1 21.2 Panama 14.6 0.1 1.1 20.6 20.8 5.7 3.5 31.5 Paraguay 8.6 0.2 0.8 13.3 16.5 1.5 2.0 23.9 Peru 4.4 0.2 0.2 4.7 11.9 0.7 0.4 16.9 Uruguay 0.3 0.0 0.1 0.7 0.1 0.0 0.0 1.2 Venezuela, R.B de 3.0 0.0 0.5 1.5 1.4 1.1 0.6 5.3 LAC Average 7.5 0.2 1.5 8.8 13.6 1.7 1.6 18.3 Source: Author's calculations based on household surveys Table A2.8: D-Index for Access to Adequate Water, by Circumstance, Circa 2008 (percent) D-Index for Access to Adequate Water, by Circumstance, circa 2008 (percent) Gender of Household Country Parent's education Gender Head Per Capita Income Urban or Rural Presence of Parents Number of Siblings Overall D-Index Argentina 0.3 0.0 0.0 0.3 0.0 0.1 0.8 Bolivia 2.0 0.6 0.7 2.0 12.0 1.0 1.4 14.6 Brazil 1.3 0.1 0.3 1.7 3.2 0.1 0.2 7.0 Chile 0.4 0.1 0.1 0.4 2.1 0.1 0.1 2.6 Colombia 3.4 0.5 0.4 1.8 8.3 1.3 2.1 11.2 Costa Rica 0.5 0.1 0.4 0.6 0.5 0.3 0.1 1.3 Dominican Republic 4.2 0.3 0.7 3.5 6.7 0.9 0.3 9.4 Ecuador 2.9 0.3 0.4 5.7 4.6 0.8 0.1 10.2 El Salvador 17.5 1.0 3.1 29.8 34.5 5.8 4.0 41.5 Guatemala 1.7 0.3 1.1 2.8 8.5 0.3 0.1 10.2 Honduras 15.7 0.8 0.9 17.5 16.4 1.5 1.0 33.1 Jamaica 6.2 0.6 8.9 2.5 30.8 5.3 7.2 31.1 Mexico 1.4 0.3 0.1 1.4 3.8 0.6 0.8 6.1 Nicaragua 13.7 3.1 4.3 21.4 36.8 8.7 3.4 42.7 Panama 1.3 0.1 0.9 2.7 2.6 0.3 0.8 6.4 Paraguay 3.7 0.0 0.1 4.7 5.6 0.9 0.8 10.3 Peru 3.7 0.1 0.5 8.6 15.3 1.0 0.6 21.3 Uruguay 0.9 0.0 0.2 2.1 0.8 0.0 0.1 4.1 Venezuela, R.B de 1.6 0.0 0.1 0.7 2.0 0.6 0.0 3.5 LAC Average 4.3 0.4 1.2 5.8 10.8 1.5 1.2 14.1 Source: Author's calculations based on household surveys 140 Table A2.9: D-Index for Access to Electricity, by Circumstance, Circa 2008 (percent) D-Index for Access to Electricity, by Circumstance, circa 2008 (percent) Gender of Household Country Parent's education Gender Head Per Capita Income Urban or Rural Presence of Parents Number of Siblings Overall D-Index Argentina Bolivia 3.2 0.1 0.3 2.0 10.1 0.7 0.3 17.0 Brazil 0.1 0.0 0.0 0.1 0.5 0.0 0.0 1.6 Chile 0.1 0.0 0.0 0.1 0.2 0.0 0.0 0.3 Colombia 0.7 0.1 0.4 0.3 1.7 0.6 0.0 2.5 Costa Rica 0.1 0.0 0.0 0.1 0.1 0.0 0.0 0.4 Dominican Republic 0.3 0.0 0.1 0.1 0.8 0.0 0.0 2.0 Ecuador 0.2 0.1 0.0 0.7 1.8 0.1 0.3 3.6 El Salvador 1.7 0.0 0.2 1.9 2.7 0.2 0.4 6.4 Guatemala 4.4 0.1 1.1 4.9 5.1 0.7 0.3 10.8 Honduras 3.7 0.1 0.4 7.4 12.2 0.3 0.3 21.7 Jamaica 0.7 0.1 1.6 0.4 1.9 0.8 0.5 2.8 Mexico 0.1 0.0 0.0 0.1 0.3 0.0 0.1 0.7 Nicaragua 6.6 0.7 3.2 6.3 16.7 0.0 0.4 22.5 Panama 3.3 0.3 1.0 10.7 10.0 0.2 3.7 19.4 Paraguay 0.4 0.1 0.1 0.6 0.8 0.1 0.0 2.1 Peru 3.5 0.0 0.2 2.2 10.8 0.2 0.8 16.2 Uruguay 0.1 0.0 0.0 0.1 0.3 0.0 0.0 0.7 Venezuela, R.B de 0.1 0.0 0.0 0.1 0.3 0.0 0.0 0.6 LAC Average 1.6 0.1 0.5 2.1 4.2 0.2 0.4 7.3 Source: Author's calculations based on household surveys 141 ANNEXES Chapter 3 142 Table A3.1: Circumstance Variables Used in PISA Analysis Circumstance Variables Used in PISA Analysis Variable Definition Child gender Dummy variable taking on a value of 1 for male School location Dummy variables taking on a value of 1 for each one of the following categories: § A village, hamlet or rural area (fewer than 3,000 people) § A small town (3,000 to about 15,000 people) § A town (15,000 to about 100,000 people) § A city (100,000 to about 1,000,000 people) The dummy for for a large city (with over 1,000,000 people) was excluded to avoid collinearity. Father's education Dummy variables taking on a value of 1 for each one of next categories: § Completed ISCED 3A (Upper Secondary, general) § Completed ISCED 3B, 3C (Upper Secondary, vocational) § Completed ISCED 2 (Lower Secondary) § Completed ISCED 1 (Primary) § Did not complete ISCED 1 (Did not complete primary) The dummy for tertiary education was excluded to avoid collinearity. Mother's education Dummy variables taking on a value of 1 for each one of next categories: § Completed ISCED 3A (Upper Secondary, general) § Completed ISCED 3B, 3C (Upper Secondary, vocational) § Completed ISCED 2 (Lower Secondary) § Completed ISCED 1 (Primary) § Did not complete ISCED 1 (Did not complete primary) The dummy for tertiary education was excluded to avoid collinearity. Wealth index Dummy variables taking on a value of 1 for each one of the deciles of wealth index created using 14 items common to all countries. Wealth index is score of first four principal components. The dummy for decile 1 was excluded to avoid collinearity. (We are currently re-estimating to use only the scores of the first principal component). Father's occupation Dummy variables taking on a value of 1 for each one of the eight categories of ISCO occupation classification on four digit basis. The dummy for Primary Occupations was excluded to avoid collinearity. 143 Table A3.2: Profile of Inequality of Proficiency at Level 2 in Reading Profile of inequality of Proficiency at level 2 in reading: Relative Importance of the Six Circumstances by Country Country Most Important 2 3 4 5 6 Argentina Socio-Economic Status Gender Father ocuppation School location Mother education Father education Brazil Socio-Economic Status Gender Father ocuppation Father education Mother education School location Canada Socio-Economic Status Gender Father ocuppation Father education Mother education Chile Socio-Economic Status Gender School location Mother education Father ocuppation Father education Colombia Socio-Economic Status School location Father ocuppation Mother education Father education Gender England Father ocuppation Socio-Economic Status Gender Father education School location Mother education France Socio-Economic Status Father ocuppation Gender Mother education Father education Germany Socio-Economic Status Gender Father ocuppation Mother education Father education School location Italy Gender Socio-Economic Status School location Father ocuppation Father education Mother education Mexico School location Gender Socio-Economic Status Mother education Father ocuppation Father education Norway Gender Socio-Economic Status Father ocuppation Mother education Father education School location Portugal Socio-Economic Status Father ocuppation Gender Mother education Father education School location Spain Socio-Economic Status Father ocuppation Gender Mother education Father education School location Sweden Gender Socio-Economic Status Father ocuppation Father education Mother education School location Uruguay Gender Socio-Economic Status Mother education Father ocuppation School location Father education USA Source: Author's calculations based on PISA data 144 Table A3.3: IPUMS Samples Analyzed IPUMS Samples analyzed Sanitation Sanitation Overcrowding (Public) (Public + Country Year Census % Households Persons Septic) Europe 1 Austria 2001 10 341,035 803,471 N N Y 2 Armenia 2001 10 81,929 326,560 N N Y 3 Belarus 1999 10 385,508 990,706 N N Y 4 France 1999 5 1,219,323 2,934,758 N N Y 5 Greece 2001 10 367,438 1,028,884 Y N Y 6 Hungary 2001 5 227,252 510,502 Y Y Y 7 Italy 2001 5 1,168,044 2,990,739 N N Y 8 Portugal 2001 5 258,843 517,026 Y Y Y 9 Romania 2002 10 732,016 2,137,967 Y Y Y 10 Slovenia 2002 10 63,637 179,632 N N Y 11 Spain 2001 5 714,473 2,039,274 N Y Y North America 12 United States 2000 5 6,184,438 14,081,466 N N Y East and Central Asia 13 Kyrgyz Republic 1999 10 110,285 476,886 Y Y Y 14 Mongolia 2000 10 55,795 243,725 N N Y 15 Vietnam 1999 3 534,139 2,368,167 N N N Middle East and Africa 16 Ghana 2000 10 397,097 1,894,133 N N Y 17 Jordan 2004 10 97,343 510,646 N N Y 18 Kenya 1999 5 317,106 1,407,547 Y Y Y 19 Rwanda 2002 10 191,719 843,392 Y Y Y 20 South Africa 2001 10 991,543 3,725,655 Y Y Y 21 Uganda 2002 10 529,271 2,497,449 N N N Note: Additional countries will be included for Asia. Canada could not be analyzed because persons are not 145 Table A3.4A: HOI Access to Sanitation in LAC (Public System) Human Opportunity Index: Access to Sanitation (Public System) (Children 0-16) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Argentina 2009 54 13 7 47 54 12 7 47 Bolivia 2007 32 36 11 20 31 36 11 20 Brazil 2008 44 20 9 35 44 19 8 36 Chile 2006 82 11 9 74 82 11 9 74 Costa Rica 2009 22 35 8 14 22 35 8 14 Dominican Republic 2008 19 29 5 13 18 29 5 13 Ecuador 2006 41 32 13 28 41 31 13 28 El Salvador 2007 32 43 14 18 32 42 13 19 Guatemala 2006 29 47 14 15 29 46 13 16 Honduras 2007 24 52 13 12 24 52 12 12 Jamaica 2002 12 46 6 7 15 44 6 8 Mexico 2008 49 25 12 37 49 24 12 37 Nicaragua 2005 15 50 8 7 15 50 8 8 Panama 2003 25 42 10 14 25 42 10 14 Paraguay 2008 5 45 2 3 5 44 2 3 Peru 2008 46 36 17 29 46 36 17 29 Uruguay 2008 46 19 9 37 46 16 7 38 Venezuela 2005 86 6 5 81 85 5 5 81 Source: Author's calculations based on SEDLAC data 146 Table A3.4B: HOI Access to Sanitation in Selected Countries (Public System) Human Opportunity Index: Access to Sanitation (Public System) (Children 0-16) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Kenya 1999 5 67 3 2 5 65 3 2 Vietnam 1999 17 56 9 7 13 58 8 5 Kyrgyz Republic 1999 17 57 10 7 17 51 9 8 South Africa 2001 36 47 17 19 36 47 17 19 Romania 2002 39 50 19 20 39 48 19 21 Hungary 2001 50 14 7 42 50 12 6 44 Portugal 2001 59 23 13 45 59 22 13 45 Greece 2001 64 11 7 57 64 9 6 58 France 1982 70 3 2 69 USA 1990 73 6 5 68 73 4 3 70 Spain 2001 93 1 1 91 93 1 1 92 Source: Author's calculations based on IPUMS data 147 Table A3.5A: HOI Access to Sanitation in LAC (Public System + Septic Tank) Human Opportunity Index: Access to Sanitation (Public System + Septic Tank) (Children 0-16) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Argentina 2009 82 5 4 77 81 5 4 77 Bolivia 2007 44 32 14 30 43 32 14 29 Brazil 2008 56 21 12 44 56 20 11 45 Chile 2006 91 5 5 86 91 5 5 86 Costa Rica 2009 95 2 2 93 95 2 2 93 Dominican Republic 2008 61 20 12 49 61 19 12 49 Ecuador 2006 64 20 13 51 63 19 12 51 El Salvador 2007 44 34 15 29 44 32 14 30 Guatemala 2006 35 40 14 21 35 39 14 21 Honduras 2007 36 37 13 22 35 36 13 23 Jamaica 2002 45 21 10 36 48 21 10 38 Mexico 2008 55 23 12 42 54 21 12 43 Nicaragua 2005 21 48 10 11 21 47 10 11 Panama 2003 46 31 14 31 45 30 13 32 Paraguay 2008 31 29 9 22 30 27 8 22 Peru 2008 63 17 10 53 63 17 10 52 Uruguay 2008 83 9 7 76 82 7 6 77 Venezuela 2005 88 5 4 84 88 5 4 84 Source: Author's calculations based on SEDLAC data 148 Table A3.5B: HOI Access to Sanitation in Selected Countries (Public System + Septic Tank) Access to sanitation (public + septic tank) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Kenya 1999 6 66 4 2 6 64 4 2 Vietnam 1999 17 56 9 7 13 58 8 5 Kyrgyz Republic 1999 17 57 10 7 17 51 9 8 South Africa 2001 38 45 17 21 38 44 17 21 Romania 2002 48 42 20 28 48 36 17 31 Greece 2001 64 11 7 57 64 9 6 58 Hungary 2001 77 8 6 71 77 6 5 73 Portugal 2001 97 1 1 96 97 1 1 95 USA 1990 99 0 0 99 99 0 0 99 Source: Author's calculations based on IPUMS data 149 Table A3.6A: HOI Freedom from Severe Overcrowding in LAC Human Opportunity Index: Freedom from Severe Overcrowding (Children 0-16) (>1.5 People Per Room) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Argentina 2009 45 29 13 32 44 28 12 32 Bolivia 2007 20 34 7 14 19 31 6 13 Brazil 2008 92 5 4 88 92 5 4 87 Chile 2006 88 5 4 84 88 5 4 84 Costa Rica 2009 85 8 7 78 85 8 7 78 Dominican Republic 2008 55 23 13 43 54 23 13 42 Ecuador 2006 67 16 11 57 67 15 10 56 El Salvador 2007 28 40 11 17 28 38 11 17 Guatemala 2006 15 51 7 7 14 49 7 7 Honduras 2007 54 23 13 42 54 23 12 42 Jamaica 2002 27 34 9 18 27 33 9 18 Mexico 2008 63 19 12 51 62 18 11 51 Nicaragua 2005 16 49 8 8 15 47 7 8 Panama 2003 38 36 14 24 37 35 13 24 Paraguay 2008 37 34 12 24 35 32 11 24 Peru 2008 41 24 10 31 40 23 9 30 Uruguay 2008 64 21 13 51 63 20 12 50 Venezuela 2005 46 27 12 33 45 27 12 33 Source: Author's calculations based on SEDLAC data 150 Table A3.6B: HOI Freedom from Severe Overcrowding in Selected Countries Human Opportunity Index: Freedom from Severe Overcrowding (>1.5 People Per Room) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Mongolia 2000 10 41 4 6 Rwanda 2002 26 34 9 17 26 33 9 17 Ghana 2000 21 23 5 16 21 21 4 17 Jordan 2004 28 36 10 18 28 35 10 18 Kenya 1999 30 26 8 22 30 22 7 23 Armenia 2001 33 17 6 27 33 17 6 27 Kyrgyz Republic 1999 35 22 8 28 35 21 7 28 Romania 2002 45 23 10 35 45 23 10 35 South Africa 2001 52 20 10 41 52 17 9 43 Belarus 1999 50 15 8 42 50 13 7 44 Hungary 2001 56 19 11 45 56 18 10 46 Slovenia 2002 65 10 6 58 Greece 2001 75 11 8 67 74 10 7 66 Italy 2001 78 9 7 71 78 9 7 72 Portugal 2001 86 7 6 80 86 7 6 80 USA 2000 91 4 4 87 91 4 4 88 France 1999 93 3 3 90 Austria 2001 94 3 3 91 94 2 2 91 Spain 2001 94 3 3 91 94 3 2 92 Source: Author's calculations based on IPUMS data 151 Table A3.6A: HOI Freedom from Severe Overcrowding in LAC Human Opportunity Index: Freedom from Severe Overcrowding (Children 0-16) (>1.5 People Per Room) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Argentina 2009 45 29 13 32 44 28 12 32 Bolivia 2007 20 34 7 14 19 31 6 13 Brazil 2008 92 5 4 88 92 5 4 87 Chile 2006 88 5 4 84 88 5 4 84 Costa Rica 2009 85 8 7 78 85 8 7 78 Dominican Republic 2008 55 23 13 43 54 23 13 42 Ecuador 2006 67 16 11 57 67 15 10 56 El Salvador 2007 28 40 11 17 28 38 11 17 Guatemala 2006 15 51 7 7 14 49 7 7 Honduras 2007 54 23 13 42 54 23 12 42 Jamaica 2002 27 34 9 18 27 33 9 18 Mexico 2008 63 19 12 51 62 18 11 51 Nicaragua 2005 16 49 8 8 15 47 7 8 Panama 2003 38 36 14 24 37 35 13 24 Paraguay 2008 37 34 12 24 35 32 11 24 Peru 2008 41 24 10 31 40 23 9 30 Uruguay 2008 64 21 13 51 63 20 12 50 Venezuela 2005 46 27 12 33 45 27 12 33 Source: Author's calculations based on SEDLAC data 152 Table A3.7A: HOI Freedom from Overcrowding in LAC Human Opportunity Index: Freedom from Overcrowding (Children 0-16) (>1 Person Per Room) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Argentina 2009 20 44 9 12 20 42 8 11 Bolivia 2007 8 51 4 4 7 47 3 4 Brazil 2008 78 13 10 69 78 13 10 68 Chile 2006 71 14 10 61 70 13 9 61 Costa Rica 2009 63 19 12 51 63 19 12 51 Dominican Republic 2008 28 39 11 17 26 39 10 16 Ecuador 2006 44 31 14 30 43 30 13 30 El Salvador 2007 13 53 7 6 12 50 6 6 Guatemala 2006 6 65 4 2 6 62 4 2 Honduras 2007 30 40 12 18 30 39 12 18 Jamaica 2002 11 54 6 5 11 51 6 5 Mexico 2008 41 32 13 28 40 31 13 28 Nicaragua 2005 5 62 3 2 5 59 3 2 Panama 2003 18 51 9 9 17 49 8 9 Paraguay 2008 17 45 8 9 16 43 7 9 Peru 2008 19 39 7 12 18 37 7 12 Uruguay 2008 37 37 13 23 35 35 12 23 Venezuela 2005 22 43 9 12 21 43 9 12 Source: Author's calculations based on SEDLAC data 153 Table A3.7B: HOI Freedom from Overcrowding in Selected Countries Human Opportunity Index: Freedom from Overcrowding (Children 0-16) (>1 Person Per Room) Assets included as circumstances Assets not included as circumstances Country Year Coverage Dissimilarity Penalty HOI Coverage Dissimilarity Penalty HOI Mongolia 2000 3 53 2 1 3 52 2 1 Jordan 2004 9 57 5 4 9 57 5 4 Rwanda 2002 10 53 5 4 10 53 5 4 Armenia 2001 9 37 3 6 9 37 3 6 Ghana 2000 9 35 3 6 9 34 3 6 Kyrgyz Republic 1999 12 41 5 7 12 40 5 7 Kenya 1999 13 37 5 8 13 34 4 9 Venezuela 2001 20 40 8 12 19 38 7 12 Belarus 1999 17 35 6 11 17 29 5 12 Romania 2002 18 33 6 12 18 33 6 12 Hungary 2001 27 29 8 19 27 28 8 19 South Africa 2001 30 34 10 20 30 30 9 21 Slovenia 2002 33 18 6 27 Greece 2001 40 23 9 31 39 22 9 31 Italy 2001 43 24 10 33 43 23 10 33 Portugal 2001 60 17 10 49 60 16 10 50 France 1999 73 12 9 64 Austria 2001 80 10 8 72 80 8 6 73 USA 2000 81 10 8 73 81 9 7 74 Spain 2001 82 7 6 76 82 7 6 76 Source: Author's calculations based on IPUMS data 154 Table A3.8: Human Opportunity Index, Census: USA [1960 – 2005] Human Opportunity Index: US, 1960 - 2005 (Children 0-16) Census Year Index 1960 1970 1980 1990 2000 ACS 2005 Housing Freedom from severe overcrowding 79 88 93 91 88 95 Freedom from moderate overcrowding 50 60 76 77 74 80 Access to Sanitation. (public system) 39 68 66 70 NA NA Access to Sanitation. (public system + septic tanks) 76 95 97 99 NA NA Source: Estimates produced based on census (IPUMS), 1960 - 2005. Table A3.9: Human Opportunity Index, Annual Change: USA [1960 – 2005] Human Opportunity Index: USA, 1960 - 2005 (Children 0-16) Annual Change Index 1960-1970 1970-1980 1980-1990 1990-2000 2000-2005 1960-1980 1980-2005 1960-2005 Housing Freedom from severe overcrowding 0.8 0.5 -0.2 -0.3 1.4 0.7 0.1 0.3 Freedom from moderate overcrowding 0.9 1.7 0.1 -0.3 1.3 1.3 0.2 0.7 Access to Sanitation. (public system) 2.9 -0.2 0.3 NA NA 1.4 NA NA Access to Sanitation. (public system + septic tanks) 1.9 0.2 0.1 NA NA 1.1 NA NA Source: Estimates produced based on census (IPUMS), 1960 - 2005. 155 Table A3.10: Human Opportunity Index, Census: FRANCE [1962 - 1999] Human Opportunity Index: FRANCE [ 1962 - 1999 ] Census year Index 1968 1975 1982 1990 1999 Housing Freedom from severe overcrowding (> 1.5 per room) 55 n.a. n.a. 87 90 Freedom from overcrowding (>1 per room) 21 n.a. n.a. 59 64 Access to Sanitation (public connection only) 52 61 69 n.a. n.a. Source: Estimates produced based on census (IPUMS), 1962 - 1968 - 1975 - 1982 - 1990 - 1999 Analyses for children from 0 to16 years old Table A3.11. Human Opportunity Index, Annual Change: FRANCE [1962 - 1999] Human Opportunity Index: FRANCE [ 1962 - 1999 ] Annual Change Index 1968-1975 1975-1982 1990-1999 1968-1982 1968-1999 Housing Freedom from severe overcrowding (> 1.5 per room) n.a. n.a. 0.6 n.a. 0.8 Freedom from overcrowding (>1 per room) n.a. n.a. 1 n.a. 1 Access to Sanitation (public connection only) 0.9 0.8 n.a. 0.9 n.a. Source: Estimates produced based on census (IPUMS), 1962 - 1968 - 1975 - 1982 - 1990 - 1999 Analyses for children from 0 to16 years old 156 Figure A3.1: HOI for Reading Human Opportunity Index for Reading 2006 Canada* 89 Sweden 83 The U.K 82 Germany 80 NON-LAC Norway 76 France* 76 Portugal 70 Spain 70 Italy 69 Chile 58 Uruguay 47 LAC México 44 Colombia 38 Brazil 36 LAC NON-LAC 35 0 20 40 60 80 Sources: PISA 2006 HOI (%) 157 Figure A3.2: HOI for Mathematics Human Opportunity Index Canada* 90 The U.K 81 NON - LAC Germany 80 Sweden 80 Norway 77 France* 76 Spain 72 USA 69 Portugal 64 Italy 63 Uruguay 47 LAC Chile 35 México 34 28 Colombia 20 NON-LAC LAC Brazil 18 Sources: PISA 2006 158 Figure A3.3: HOI for Science Human Opportunity Index for Science 2006 Canada* 90 The U.K 84 Germany 84 Sweden 83 NON - LAC Norway 78 Spain 78 France* 77 USA 73 Portugal 71 Italy 71 Chile 53 LAC Uruguay 51 México 40 36 Colombia 33 LAC NON-LAC Brazil 30 Sources: PISA 2006 HOI (%) 159 ANNEXES Chapter 4 160 Table A4.1. Countries, Provinces or Sub-Regions and Years Countries, Provinces or Sub-Regions and Years Country Survey years Provinces Sub-Regions Argentina 1998 2008 23 Brazil 1995 2008 27 Chile 1996 2006 13 Colombia 1997 2008 9 Costa Rica 1994 2009 6 Dominican Republic 2000 2008 9 Ecuador 1995 2006 3 El Salvador 1998 2007 14 Guatemala 2000 2006 8 Honduras 1999 2007 5 Nicaragua 1998 2005 4 Panama 1997 2003 10(9) Paraguay 1999 2008 16 (9) Peru 1998 2008 7 Uruguay 2006 2008 19 TOTAL 123 42 Source: The World Bank and Universidad de La Plata (CEDLAS) Socioeconomic Database for Latin American and the Caribbean 161 Table A4.2: Regional/Sub-national HOIs 2008 Ranking Ranking Region Region HOI HOI Ranking Ranking Region Region HOI HOI 1 Tierra del Fuego - ARG 96 29 San Luis - ARG 90 2 Bogota DC - COL 95 30 La Pampa - ARG 90 3 Antofagasta - CHI 95 31 Libertador General B - CHI 89 4 Mendoza - ARG 94 32 Colonia - URUG 89 5 Tarapacá - CHI 94 33 Lima metropolitana - PERU 89 6 Atacama - CHI 94 34 Catamarca - ARG 89 7 Metropolitana de San - CHI 94 35 Rio Negro - URUG 89 8 Aysen del General Ca - CHI 93 36 Bio-Bio - CHI 89 9 Salta - ARG 93 37 Treinta y tres - URUG 89 10 La Valleja - URUG 93 38 Salto - URUG 89 11 Santa Cruz - ARG 93 39 Entre Rios - ARG 88 12 Montevideo - URUG 93 40 Orinoquia y Amazonia - COL 88 13 Chubut - ARG 93 41 Tacuarembo - URUG 88 14 Valparaiso - CHI 93 42 Maule - CHI 88 15 Cordoba - ARG 92 43 Canelones - URUG 88 16 Coquimbo - CHI 92 44 Florida - URUG 87 17 City of Buenos Aires - ARG 92 45 Chaco - ARG 87 18 Neuquen - ARG 92 46 La Rioja - ARG 87 19 Soriano - URUG 92 47 Rivera - URUG 87 20 Durazno - URUG 92 48 Maldonado - URUG 87 21 Cerro Largo - URUG 91 49 Tucuman - ARG 87 22 91 50 Santa Fe - ARG 87 23 Flores - URUG 91 51 Bs As - ARG 87 24 Rocha - URUG 90 52 Paysandu - URUG 87 25 Central - CRICA 90 53 San Jose - URUG 86 26 Valle del Cauca - COL 90 54 Los Lagos - CHI 86 27 Magallanes y la Antá - CHI 90 55 Formosa - ARG 86 28 Corrientes - ARG 90 56 Costa urbana - PERU 86 162 Ranking Region HOI Ranking Region HOI 57 Brunca - CRICA 85 85 Espirito Santo - BRA 78 58 Paraná - BRA 85 86 Central - COL 78 59 San Paulo - BRA 85 87 Rio de Janeiro - BRA 77 60 Jujuy - ARG 85 88 Selva urbana - PERU 77 61 San Juan - ARG 85 89 San Pedro Sula - HOND 77 62 Sierra urbana - PERU 85 90 Oriental - COL 76 63 Araucania - CHI 84 91 Amapá - BRA 75 64 Central - PARAG 84 92 Roraima - BRA 75 65 83 93 Región Cibao Central - DOMIN.REP. 75 66 Santa Catarina - BRA 83 94 Herrera - PANAMA 75 67 Asunción - PARAG 83 95 Paraguarí - PARAG 74 68 83 96 Tegucigalpa (Dist.Nac) - HOND 74 69 Distrito federal - BRA 83 97 Chiriquí - PANAMA 73 70 Chorotega - CRICA 82 98 Cordillera - PARAG 72 71 Región Distrito Naci - DOMIN.REP. 82 99 Rio Grande du Norte - BRA 72 72 82 100 Costa - ECU 72 73 Rio Grande du Sul - BRA 82 101 Rondonia - BRA 72 74 Sierra* - ECU 82 103 Sergipe - BRA 72 75 Colón - PANAMA 81 104 Metropolitana - GUAT 72 76 Misiones - ARG 81 105 Los Santos - PANAMA 72 77 Goiás - BRA 81 106 Alto Parana - PARAG 72 78 Panamá - PANAMA 80 107 71 79 Mato Grosso Sul - BRA 80 108 70 80 Huetar Norte - CRICA 80 109 Paraíba - BRA 69 81 Mato Grosso - BRA 79 110 Managua - NICA 69 82 78 111 San Salvador - EL SALV. 69 83 Región Norcentral - DOMIN.REP. 78 112 Ceará - BRA 68 84 Minas Gerais - BRA 78 113 Amazonas - BRA 68 163 Ranking Region HOI Ranking Region HOI 114 Ciudades medianas - HOND 68 142 55 115 Coclé - PANAMA 67 143 Itapúa - PARAG 54 116 Bahia - BRA 67 144 Suroccidente - GUAT 53 117 Región Este - DOMIN.REP. 67 145 Santa Ana - EL SALV. 52 118 Región Noroeste - DOMIN.REP. 67 146 Chalatenango - EL SALV. 51 119 San Andres y Provide - COL 67 147 La Paz - EL SALV. 51 120 Pernambuco - BRA 67 148 San Vicente - EL SALV. 50 121 Región Nordeste - DOMIN.REP. 66 149 Nororiente - GUAT 50 122 66 150 Cuscatlán - EL SALV. 49 123 Para - BRA 65 151 Usulután - EL SALV. 49 124 Región Valdesia - DOMIN.REP. 65 152 Suroriente - GUAT 49 125 Oriente/Amazonia - ECU 64 153 Sonsonate - EL SALV. 48 126 Caaguazú - PARAG 63 154 Selva rural - PERU 47 127 San Pedro - PARAG 63 155 Darién - PANAMA 47 128 Alagoas - BRA 63 156 San Miguel - EL SALV. 47 129 Guairá - PARAG 62 157 Noroccidente - GUAT 44 130 Ciudades pequeñas - HOND 61 158 Cabañas - EL SALV. 44 131 Maranhao - BRA 61 159 Ahuachapán - EL SALV. 44 132 Piaui - BRA 61 160 La Union - EL SALV. 42 133 Costa rural - PERU 60 161 Peten - GUAT 40 134 Región del Valle - DOMIN.REP. 60 162 Central - NICA 40 135 Sierra rural - PERU 59 163 Rural - HOND 38 136 Acre - BRA 58 164 Norte - GUAT 33 137 Veraguas - PANAMA 58 165 Morazán - EL SALV. 33 138 Central - GUAT 58 166 29 139 Bocas del toro - PANAMA 57 140 Región Enriquillo - DOMIN.REP. 56 141 La Libertad - EL SALV. 55 Source: The World Bank and Universidad de La Plata (CEDLAS) Socioeconomic Database Survey years for 164 Figure A4.1: Housing HOI Circa 1995 and 2008 Housing HOI : circa 1995 and 2008 100 90 80 70 60 HOI for 2008 50 40 30 20 10 0 0 10 20 30 40 50 60 70 80 90 100 HOI for 1995 165 Figure A.4.2: Education HOI Circa 1995 and 2008 Education HOI : circa 1995 and 2008 100 90 80 70 60 HOI for 2008 50 40 30 20 10 0 0 10 20 30 40 50 60 70 80 90 100 HOI for 1995 166 Figure A.4.3: Dispersion and Education and Housing HOI 35 per 30 hon per 25 hon nic par els pan 20 nic els gua domcol gua par chi pan bra dom 15 col bra ecu cos 10 ecu arg chi arg cos uru 5 uru 0 20 40 60 80 100 HOI Housing 35 30 25 20 nic 15 perpan col nic par hon hon guagua cos 10 els dompar pan bra per col bra els dom coschiarg 5 ecu uru uru ecu arg chi 0 20 40 60 80 100 Bibliography 167 168 References Abou-Ali, Hala. 2003. “The Effect of Water and Sanitation on Child Mortality in Egypt.� Unpublished. Environmental Economics Unit, Department of Economics, Göteborg University, Sweden. Bajak, F. 2007. “In Peru, a Pint-Size Ticket to Learning: Official Hope 270,000 Laptops for Poor Youngsters Improve Education System.� Washington Post, December 18, A18. Barro, R.J. and X. Sala-i-Martin. 1991. “Convergence across states and regions�, Brookings Papers on Economic Activity, pp.107-82. Barros, Ricardo Paes de, Francisco H. G. Ferreira, José Molinas Vega and Jaime Saavedra Chanduvi. 2009. Measuring Inequality of Opportunities in Latin America and the Caribbean. Washington, DC: Palgrave Macmillan and the World Bank. Barros, Ricardo Paes de, José Molinas Vega and Jaime Saavedra Chanduvi. 2008. “Measuring Inequality of Opportunities for Children.� Unpublished, World Bank, Washington, DC. www.worldbank.org/lacopportunity. Bird, Richard M. 2000. “Transfers and Incentives in Intergovernmental Relations,� in S. Burki and G. Perry, eds., Decentralization and Accountability of the Public Sector (Washington: World Bank). Bourguignon, Francois, F. H. G. Ferreira and M. Menéndez. 2007. “Inequality of Opportunity in Brazil.� Review of Income and Wealth 53 (4): 585-618. Burki, S.J, Perry, G., & Dillinger, W. 1999. Beyond the Center: Decentralizing the State. Washington: The World Bank. Checchi, D., and V. Peragine. 2005. “Regional Disparities and Inequality of Opportunity: The Case of Italy.� IZA Discussion Paper No. 1874, Institute for the Study of Labor, Bonn, Germany. Chombart de Lauwe, 1956 P .H. Chombart de Lauwe, La vie 169 quotidienne des familles ouvrières, Paris, CNRS (1956). Coggan, D., Barker, D.J.P ., Inskip, H., Wield, G. (1993) ‘Housing in early life and later mortality’, Journal of Epidemiology and Community Health, 47, 345-348. Corak, Miles, Lori Curtis and Shelley Phipps, 2010. Economic Mobility, Family Background, and the Well-Being of Children in the United States and Canada. IZA Discussion Paper No. 4814. Currie, Janet and Aaron Yelowitz, 2000 Are public housing projects good for kids? Journal of Public Economics 75 (2000) 99–124 Daughters, R. and L. Harper. 2007. Fiscal and Political Decentralization Reforms. In Lora, E. (ed.). The State of State Reform in Latin America. Stanford, CA. ECLAC. 2008. Panorama social de América Latina 2008. Faguet, Jean-Paul and Shami, Mahvish. 2008. Fiscal policy and spatial inequality in Latin America and beyond. Policy research working paper series, The World Bank, Washington, DC, USA. Foster, Vivien. 2005. “Ten Years of Water Service Reform in Latin America: Toward an Anglo-French Model� Water Supply and Sanitation Sector Board Discussion Paper Series No. 3. Washington, DC: World Bank. Fuentes, Ricardo, Tobias Pfütze, and Papa Seck. 2006. “A Logistic Analysis of Diarrhea Incidence and Access to Water and Sanitation.� Human Development Report Occasional Paper. New York, NY: UNDP . Fuller, T.D., J.N. Edwards, S. Sermsri and S. Vorakitphokatorn, Housing, stress, and physical well-being: evidence from Thailand, Social Science Medicine 36 (11) (1993), pp. 1417–1428. Galiani, Sebastian, Paul J. Gertler, and Ernesto Schargrodsky. 2005. “Water for Life: The Impact of the Privatization of Water Services on Child Mortality.� Journal of Political Economy 113: 83-120. Goux, Dominique and Eric Maurin, 2005. The effect of 170 overcrowded housing on children’s performance at school. Journal of Public Economics, 89:797-819. Gove, W.A., Hughes, M., Galle, O.R. (1979) ‘Overcrowding in the home: an empirical investigation of its possible pathological consequences’, American Sociological Review, 44, 55-80. Gustavsson, M. 2007. “Educational Benefits from Solar Technology: Access to Solar Electric Services and Changes in Children’s Study Routines, Experiences from Eastern Province Zambia.� Energy Policy 35 (2): 122-29. Hanushek, Eric A. 2003. “The failure of input-based schooling policies.� Economic Journal 113, No.485 (February): F64- F98. Hanushek, Eric A., and Margaret E. Raymond. 2005. “Does school accountability lead to improved student performance?� Journal of Policy Analysis and Management 24, No.2 (Spring):297-327. Herran, C. 2005. Reducing Poverty and Inequality in Brazil. Economic and Social Study Series, Inter-American Development Bank. Kaufman, S., R. Duke, R. Hansen, J. Rogers, R. Schwartz, and M. Trexler. 2000. “Rural Electrification with Solar Energy as a Climate Protection Strategy.� Research report Number 9. Washington, DC: Renewable Energy Policy Project. Lefranc, A., N. Pistolesi, and A. Trannoy. 2006. “Inequality of Opportunities vs. Inequality of Outcomes: Are Western Societies All Alike?� Working Paper No. 54 Society for the Study of Economic Inequality, Palma de Mallorca, Spain. Marhsy, Mona, 1999. Social and Psychological Effects of Overcrowding in Palestinian Refugee Camps in the West Bank and Gaza – Literature Review and Preliminary Assessment of the Problem. Available online at http:// www.idrc.ca/uploads/user-S/11244731091marshy.htm. Accessed on March 23, 2010. OECD (Organisation for Economic Co-operation and Development). 2007. PISA 2006: Science Competencies 171 for Tomorrow’s World Executive Summary. OECD 2007 Office of the Deputy Prime Minister London, 2004. The Impact of Overcrowding on Health and Education: A Review of Evidence and Literature. Roemer, John E. 1998. Equality of Opportunity. Cambridge, MA: Harvard University Press. Rutstein, Shea O. 2000. “Factors Associated with Trends in Infant and Child Mortality in Developing Countries during the 1990s.� Bulletin of the World Health Organization 78 (10): 1256-70. Shah, Anwar. 2010 “Sponsoring a Race to the Top: The Case for Results-Based Intergovernmental Finance for Merit Goods� Policy Research Working Paper 5172, The World Bank. Shankar, Raja and Anwar Shah. 2003. “Bridging the Economic Divide within Nations: A Scorecard on the Performance of Regional Development Policies in Reducing Regional Income Disparities,� World Development, 31(8): 1421- 1441. Smart, Michael, and Richard Bird. 2009. “Earmarked grants and accountability in government�, University of Toronto Revised; October 2009 Steffensen, Jesper and Henrik Larsen. 2005. Conceptual Basis for Performance Based Grant Systems and Selected International Experiences, background paper for the National Stakeholder Workshop, 31 May, Kathmandu, Nepal. Stein, Ernesto. 1999: “Fiscal Decentralization and Government Size in Latin America�, Journal of Applied Economics 2, 357-391. UNICEF (United Nations Children’s Fund) 2006. Children and Water, Sanitation and Hygiene: The Evidence. Human Development Report Office Occasional Paper. New York, NY: UNDP . US Department of Housing and Urban Development, 2007. Measuring Overcrowding in Housing. 172 Vaillancourt, Francois & Richard Bird. 2005. “Expenditure-Based Equalization Transfers,� International Tax Program Papers 0512, International Tax Program, Institute for International Business, Joseph L. Rotman School of Management, University of Toronto. Von Braun J. and U. Grote. 2002. “Does Decentralization Serve the Poor?�, in IMF (ed.), Fiscal Decentralization, Routledge Economics, Washington, D.C., pp. 92-119. WHO (World Health Organization). 2008. Safer Water, Better Health: Costs, benefits, and sustainability of interventions to protect and promote health. WHO (World Health Organization). 2002. The World Health Report 2002: Reducing Risks, Promoting Healthy Life. Geneva, Switzerland. World Bank. 2006. World Development Report 2006: Equity and Development. Washington, DC: World Bank. Zanetta, C. 2004. The Influence of the World Bank on National Housing and Urban Policies: A Comparison of Mexico and Argentina during the 1990s. London: Ashgate.