Building an Equitable Society in Colombia Building an Equitable Society in Colombia © 2021, International Bank for Rights Reserved Rights and Permissions Concept and Cover Design Reconstruction and Development / This volume is a product of the The International Bank for Carlos Reyes, Reyes Work Studio World Bank staff of the International Bank for Reconstruction and Development/ Reconstruction and Development The World Bank encourages the Interior Design and Typesetting 1818 H Street N.W. / The World Bank. The findings, dissemination of its work and Carlos Reyes, Reyes Work Studio Washington D.C. 20433, interpretations, and conclusions will normally grant permission to United States of America expressed in this volume do not reproduce portions of this work Telephone: (202) 473-1000 necessarily reflect the views of the promptly, provided the sources are Executive Directors of the World Bank acknowledged. Internet: www.worldbank.org or the governments they represent. In Spanish: www.bancomundial.org The World Bank does not guarantee Email: feedback@worldbank.org the accuracy of the data included in this publication. Team Members 3 BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO M BI A | and Acknowledgements This report was prepared by a core team co-led by María Eugenia Sebastián Higuera Pedraza, Pedro Cerdan-Infante, Leonardo Dávalos and Paolo Dudine under the guidance of Jorge Araujo Canon Rubiano, Vanessa Alexandra Velasco Bernal, Diana Marcela (Program Manager, ELCMU) and Ximena Del Carpio (Program Rubiano Vargas, and Luis Miguel Triveno Chan Jan. Finally, Chapter Manager, ELCPV), Donato De Rosa and Gabriel Demombynes 6 was authored by Julian Lee, Hugo Rojas-Romagosa, Juan José (Program Leaders), and overall direction of Ulrich Zachau (Country Miranda, Hasan Dudu, Viviana Perego, Santiago de la Cadena, and Director for Colombia). Chapter 1’s analysis of inequality trends and Joaquin Urrego; the team is thankful to the Dirección de Ambiente drivers was led by Jose Cuesta, María Eugenia Dávalos, Julieth Pico, y Desarrollo Sostenible of the National Planning Department for with contributions from Irene Clavijo; the full chapter was prepared co-leading this chapter. Desiree Gonzalez provided administrative by Paolo Dudine and María Eugenia Dávalos with inputs from all and logistics support throughout the preparation of this work; chapters. Chapter 2 was authored by Pedro Cerdan-Infantes, Gabriel Maria Clara Ucros and Jairo Bedoya guided the dissemination and Demombynes, Santiago de la Cadena, Mateo Prada, Manuela Villar- communication strategy; Carlos Reyes and Patricia Carley provided Uribe, Jeremy Veillard. Chapter 3 was authored by Hernan Winkler design and editing services, respectively. and benefitted from coordination and interactions with the World The team is grateful for the active dialogue with the Ministry of Bank team leading the Jobs Diagnostics for Colombia. Chapter 4 Finance and the National Planning Department in the preparation was authored by Paolo Dudine, Julieth Carolina Pico Mejia, Andres of this work, and the comments and suggestions received. The David Pinchao Rosero, and Maryan Raquel Porras Barrera, with team is also grateful for the contributions from the peer reviewers: contributions from Cristina Savescu. Chapter 5 was authored by Samuel Freije-Rodriguez (Lead Economist); Maurizio Bussolo (Lead Diana C. Tello Medina, Nancy Lozano Gracia, Ivonne A. Moreno Economist); Marcel Ionescu-Heroiu (Senior Urban Development Horta, Juan C. Duque, Fernando Carriazo Osorio, Juan P. Ospina, Specialist); Marco Hernandez (Lead Economist, EAWM2); and Andrea Mauricio Quiñones, Gustavo A. García, and Jorge E. Patiño. It included Liverani (Program Leader). contributions from German Freire, Steven Schwart, Paula Andrea Rossiasco, Kelly Yelitza Montoya Munoz, Vladimir Tafur Hernandez, Table of Contents 4 BUI LD I N G A N E Q U I TABLE S OCI ETY I N COLOM BI A | This report is designed to be read on screens. List of Acronyms 5 Chapter 5. Territorial Inequalities in Colombia 53 Some pages may not print with a legible font size on a standard A4. Executive Summary 6 5.1. Introduction 55 Chapter 1. Overview of the Challenge 8 5.2. Diagnostics: Regional and Intra-Urban Inequalities 56 1.1 Introduction 9 Regional inequality: high and persistent along different 1.2. The Drivers of Inequalities 11 scales and dimensions 57 Inequalities in human capital: uneven access Leveraging the potential of cities to reduce rural poverty 58 to education and health 11 Cities with high inequality and spatial segregation 59 Inequalities in physical and financial assets 12 Territorial inequality, access to opportunities, Inequalities in access to good jobs 13 and lagging infrastructure 60 Large territorial gaps in opportunities across the country 14 More efficient spending to help reduce intra-urban inequality: the strata system 61 Fiscal policy and inequality 15 5.3. Policy Options 62 Shocks affect the most vulnerable 16 Institutions that unify: strengthening subnational governments Inadequate social assistance programs 17 to respond to the needs of vulnerable territories 62 1.3. Policy Options to Tackle Inequalities in Colombia 18 Infrastructure that connects: improving accessibility Chapter 2. Human Development and Equity in Colombia 20 to bring opportunities closer to territories 62 2.1 Introduction 22 Interventions that target: well-aimed investments 2.2. Diagnostics: Gaps in Human Capital for more equal and inclusive cities 62 and Inequality in Colombia 23 Conclusion 63 Education: skills differences and inequality in human capital 24 Chapter 6. The Effects of Climate Change on Equity The health care system 25 in Colombia 65 The social protection system: inequalities in opportunities 6.1. Introduction 67 for resilience 26 6.2. Diagnostics: Protecting Vulnerable Populations Impact of the COVID-19 pandemic on human capital 27 against Shocks 68 2.3. Policy Options 28 Summary of the methodology 68 Healthy citizens: high-quality primary health care for all 28 Climate change disproportionately affecting the poor 69 Skilled citizens: accelerating learning and ensuring Climate change and economic output 70 complete education trajectories 28 Climate change to widen existing inequalities 71 Resilient citizens: an integrated and dynamic social registry Colombia’s limited capacity to help vulnerable groups and to respond effectively to shocks 28 the potential of climate-smart agriculture 72 Cross-cutting areas: improving the quality of service delivery 28 Social protection programs insufficiently adaptive Conclusion 29 to climate shocks 73 Chapter 3. Making the Colombian Labor Market More Inclusive 31 Insufficiently developed carbon pricing 74 3.1. Introduction 33 6.3. Policy Options 75 3.2. Diagnostics: Addressing Inequality in the Labor Market 34 Invest in climate-smart agricultural value chains that Low growth in labor productivity 34 combine inclusion with climate mitigation and adaptation 75 Unequal access to new technologies 35 Strengthen the adaptiveness of the social protection system to ensure the resilience of the poorest to climate shocks 75 COVID-19 and the acceleration of labor market trends 36 Expand carbon pricing to combine equity 3.3. Policy Options 37 and climate mitigation. 75 Minimizing distortionary labor market policies 37 Conclusion 76 Adapting to the future of work: global value chains Annex 1. Inequalities in Subjective Well-Being and technology trends 38 in Colombia across Individuals and Space 79 Conclusions 39 Annex 2. The Disaggregated Human Capital Index Chapter 4. Public Finance and Equity in Colombia 41 for Colombia 80 4.1. Introduction 43 Annex 3. Labor markets estimations 81 4.2. Diagnostics: Utilizing Fiscal Policy to Reduce Inequality 44 Annex 4. Intra-Urban and Accessibility Inequalities Reliance on transfers over progressive taxes 44 and Territorial Poverty 82 Personal income taxes: high deductions 45 A4.1. Intra-Urban Inequalities 82 VAT exemptions benefit the rich 46 A4.2. Accessibility 82 Transfers and subsidies 47 A4.3. Interactions between Poverty, Income, and Inequality 83 Pensions and labor income inequalities 48 A4.4. Urban-Rural Channels 84 Bringing it all together: the overall redistribution A4.5. Robustness checks with a new combination of territories 85 of fiscal policy in Colombia 49 Annex 5. MANAGE CGE Model For Colombia 86 4.3. Policy Options 50 A5.1. Characteristics and Implementation 86 Conclusion 51 A5.2. Technical Specifications 86 References 88 List of Figures, Tables, and Boxes 101 List of Acronyms 5 BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO M BI A | ALMP Active Labor Market Program GEIH Gran Encuesta Integrada de Hogares (Large Integrated Household Survey) PISA Programme for International Student Assessment ART Alternate Response Technologies GHG Greenhouse Gas PIT Personal Income Tax BEPS Beneficios Económicos Periódicos (Periodic Economic Benefits) GSSSH General System for Social Security in Health PNACC Plan Nacional de Adaptación al Cambio Climático (National Plan for Adap- CAE Centro de Atención Empresarial (Business Service Center) GVC Global Value Chains tation to Climate Change) CAL Computer-Assisted Learning HCP Human Capital Project PND Departamento Nacional de Planeación (National Planning Department) CDE Constant Differences of Elasticity HCI Human Capital Index POTs Planes de Ordenamiento Territorial (Territorial Development Plans) CES Constant Elasticity of Substitution ICBF Instituto Colombiano de Bienestar Familiar (Colombian Institute of Family PPP Public-Private Partnership CET Constant Elasticity of Transformation Welfare) PTA Programa Todos a Aprender (Everyone to Learn Program) CGE Computable General Equilibrium ICM Índice de Ciudades Modernas (Modern Cities Index) QRIS Quality Rating and Improvement System CIT Corporate Income Tax ICT Information and Communications Technology SAM Social Accounting Matrix C-MPI Colombian Multidimensional Poverty Index LAC Latin America and the Caribbean Sisbén Sistema de Identificación de Potenciales Beneficiarios de Programas So- CONPES Coordinación Nacional para la Planeación de la Educación Superior (Na- LFP Labor Force Participation ciales (Identification System for Potential Beneficiaries of Social Programs) tional Council for Economic and Social Policy of Colombia) MADR Ministry of Agriculture and Rural Development SLS Stage Least Squares CSA Climate-Smart Agriculture MADS Ministry of Environment and Sustainable Development SWB Subjective Well-Being DANE Departamento Administrativo Nacional de Estadística (National Adminis- MCI Modern Cities Index (Índice de Ciudades Modernas) UGRD Unidad de Gestión de Riesgos y Desastres (National Unit for Disaster Risk trative Department of Statistics) MDB Multilateral Development Bank Management) DNP Departamento Nacional de Planeación (National Planning Department) MDM Medición de Desempeño Municipal (Municipal Performance Measure) UHCI Utilization-Adjusted Human Capital Index DPS Departamento para la Prosperidad Social (Department of Social Prosperi- MPI Multidimensional Poverty Index UN-GGIM United Nations Committee of Experts on Global Geospatial Information ty) Management NAMA Nationally Appropriate Mitigation Action DRM Disaster Risk Management UPRA Unidad de Planificación Rural Agropecuaria (Rural Agricultural Planning NARP Negros, Afro-descendientes, Raizales y Palenqueros EATs Esquemas Asociativos Territoriales (Territorial Associative Schemes) Unit) NDC Nationally Determined Contribution ECD Early Childhood Development USGS U.S. Geological Survey NEET Not in Education, Employment, or Training ENCV Encuesta Nacional de Calidad de Vida (National Quality of Life Survey) UVT Unidad de Valor Tributario (Tax Value Unit) NPD National Planning Department (Departamento Nacional de Planeación ENPH Encuesta Nacional de Presupuestos de los Hogares (National Household VAT Value Added Tax OECD Organisation for Economic Co-operation and Development Budget Survey) VUE Ventanilla Única Empresarial (One-Stop Shop) OLS Ordinary Least Squares FDI Foreign Direct Investment WDI World Development Institute OSS One-Stop Shop FELA Framework for Effective Land Administration WDR World Development Report PDET Programa de Desarrollo con Enfoque Territorial (Program with a Territo- FINAGRO Fondo para el Financiamiento del Sector Agropecuario (Financing Fund for WHO World Health Organization rial Focus) the Agriculture Sector) PGOT Política General de Ordenamiento Territorial (General Policy on Territorial G2P Government to Person Planning) GDIM Global Database on Intergenerational Mobility PHC Primary Health Care Building an 6 Scroll down page BUIL D ING AN EQ UITABL E S O CIETY IN CO LO M BIA | E x ecutive S ummary Equitable Society in Colombia Executive Summary Colombia’s high level of inequality is a core constraint to economic growth and social prog- ress. The country has one of the highest levels of income inequality in the world, the second highest among 18 countries in Latin America and the Caribbean (LAC), and the highest among all OECD countries. The disparities in income across adults grow from gaps that open early in life in opportunities for high-quality childhood development, education, and health care ser- vices. Inequality in access to good jobs further amplifies these gaps, making Colombia among the countries where inequalities are the most persistent across generations. Longstanding in- equality across regions overlaps with the large gaps in welfare between Afro-descendants and indigenous Colombians and the rest of the population. The COVID-19 pandemic has further amplified disparities and threatens to have prolonged negative effects, but this is just one of many potential extreme shocks, including climate change–related disruptions, that could substantially widen the inequality gaps. Current tax and transfer policies at best have only a modest positive impact on these imbalances, so there is clearly ample potential to improve the redistributive role of fiscal policy in Colombia. Policy reforms across many areas could help to chart a more equitable future for the country. Key Findings HIGHLIGHT 1 Colombia is one of the most unequal countries in the world, INCOME GINI COEFFICIENT, circa 2018 even more so after the COVID-19 shock, and one with very low Brazil social mobility. Colombia St. Lucia Panama Income inequality in Colombia is the highest among all OECD countries and the second high- Guatemala Costa Rica est among 18 LAC countries.1 The Gini coefficient of household income (after paying taxes and Honduras Nicaragua receiving transfers), a standard measure of inequality, reached 0.53 in 2019. For comparison, Ecuador Paraguay the Gini coefficient of the most equal country in the OECD, the Slovak Republic, was 0.24. The Mexico richest 10 percent of Colombians earned more than 11 times the income of the poorest 10 per- Chile Argentina (urban) cent. Again, for comparison, in the Slovak Republic, the richest 10 percent earned three times as Dominican Republic Bolivia much as the poorest 10 percent. The COVID-19 economic shock has increased inequality further, Peru pushing the Gini coefficient up to 0.54 in 2020 and pulling 3.6 million more people into poverty. Uruguay United States El Salvador Large inequalities also exist between different population groups. A woman in Colombia is 1.7 United Kingdom Lithuania times more likely to be unemployed than a man.2 An indigenous Colombian receives on average Latvia Israel two years less schooling than other Colombians, and an Afro-Colombian is twice as likely to live Korea Italy in a slum. Two-thirds of the children of Venezuelan migrants are not enrolled in school compared Spain to less than one-tenth of non-migrants. Australia Luxembourg Portugal Strikingly, inequality in Colombia extends beyond the material aspects of one’s livelihood. Greece Estonia Less-educated Colombians, rural residents, and those who are unemployed or poor are much Canada less likely to consider themselves to be happy. France Switzerland Ireland Inequalities also persist across generations. Children in Colombia face very different prospects Hungary Germany in life because of the circumstances they are born into: a child from a low-income parent is Netherlands Poland likely to earn less than a child from a high-income parent. Among a group of 75 countries, the Austria transfer of income gap from one generation to the next in Colombia is the most entrenched Sweden Finland (Narayan et al. 2018).3 Denmark Norway Belgium Reducing inequalities is not only an objective in itself on moral grounds, but it also makes Iceland Slovenia good economic sense. Tackling inequities can lead to a better prepared, greater skilled, and Czech Republic Slovak Republic more productive labor force, stronger and more sustainable economic growth, and tighter social cohesion. For example, closing gender gaps in labor force participation and education 0 0,1 0,2 0,3 0,4 0,5 0,6 would increase Colombia’s GDP per capita by an estimated 14 percent by 2050 (Devadas and Source: Authors’ calculations, based on Gran Encuesta Integrada de Hogares (GEIH)/Large Inte- Kim 2020). A more equal society would mean better lives for all. grated Survey of Households (left), OECD income inequality database, and World Bank Equity Lab ( https://www.worldbank.org/en/topic/poverty/lac-equity-lab1/overview). Note: Latest year available. For most OECD countries data are from 2017 and 2018. For most LAC countries data are from 2019. Data for Guatemala and Nicaragua are from 2014. FIGURE1.3. Differences in Poverty Headcount by Group, GAPS IN TOTAL percentage POVERTY points, RATES 2019 BETWEEN GROUPS, %, 2019 2 Richest vs poorest UN EN VI RO N M EN T F R O N TI ERS 202 0— 21 R EP O RT | SE C TION department 41 Indigenous - None 27 Migrant - non-migrant 25 Rural - Urban 15 Afro-descendant 10 - None Household head woman - Man 4 0 10 20 30 40 50 Source: Authors’ calculations, based on GEIH 2019. Note: (i) The Afro-descendant (or NARP) group includes Black, Afro-Colombian, Raizal, and Palenquero populations. The Afro-descendant and indigenous groups’ rates are compared to those of people with no ethnic identification (group “none”). (ii) The category woman-man makes references to the sex of the household head. (iii) The comparison among departments shows the differences between the one with the largest and small est indicator. FIGURE 2 Departm HIGHLIGHT 2: Inequalities start early in life with gaps Am in education and health care. La G In Colombia, inequalities affect individuals early in their lives in a way that has consequences ACTUAL LEARNING OUTCOMES DIFFER GREATLY Pu San Andrés on human capital accumulation and thus the opportunities available when it comes time to ACROSS ETHNIC GROUPS. Valle de enter the labor market or otherwise earn an income. Learning Gaps Across Ethnic Groups (years of education adjusted by learning) C First, learning opportunities are not the same for all children in Colombia. Afro-Colombians FIGURE 2.5. Learning Gaps Across Ethnic Groups and indigenous populations lose the equivalent of 4.7 and 4.5 years of education, respectively, 11,1 11,0 Mag 10,7 C when adjusting years of schooling by actual learning outcomes, figures that exceed what oth- A er groups lose by a full year or more. Gaps between schooling and learning are largely linked Norte de Sa to differences in teacher quality and how teachers are assigned to schools. An Moreover, despite a substantial expansion in health insurance in recent years, there are large C differences in access to high-quality health care. This contributes to disparities in outcomes: 7,6 the poorest children have stunting rates that are three times larger than those of the richest 6,3 6,2 R children. Two factors affect the quality of health service provision: (i) current formulas to fi- Sa nance health care providers do not account for the risk profile of the typical patient, offering Cundin very little incentive to extend differentiated care to patients with different risk factors; and (ii) No ethnic group Afrocolombian Indigenous information on the quality of service provision is limited and not sufficiently used to better plan and deliver services at the local level. Source: World Bank analysis of DANE (2018b) and SABER 11 Test Scores (national stan- Source: Wo Promoting human capital accumulation from early childhood requires simplifying the admin- dardized assessment for 11th graders 2019). (2018a), and istrative procedures for citizens to access early childhood development (ECD) services, intro- ducing a core curriculum for basic competencies across the education system, and providing pedagogical support to teachers on the core curriculum guidelines. At the same time, it also requires strengthening linkages between basic and tertiary education and ensuring the quality and relevance of the curriculum. On health, the service delivery model should be transformed into a system of primary health care that is adapted to local needs and accreditation and finan- cial incentives should be provided to health insurers. HIGHLIGHT 3: Disparities in access to good jobs amplify the inequalities in human capital. THE QUALITY OF JOBS IN TERMS Just 40 percent of working Colombians hold formal sector jobs, one of the lowest rates in the OF FORMALITY AND EARNINGS IS LAC region. Stringent labor regulations and high minimum wages discourage the creation of ONE OF THE LOWEST IN LAC. FIGURE 3.4a. The Quality of Jobs in Terms of Formality and formal sector jobs, leaving most Colombians working in the informal sector. The jobs of the Circa 2018 Earnings in LAC, circa 2018 future may also be out of reach for many owing to the slow adoption of new technologies 80 among disadvantaged groups, a barrier that the COVID-19 crisis has aggravated. In fact, with 70 a ranking of 109 out of 141 countries, Colombia has one of the largest disparities in the world Costa Rica Chile Uruguay in the use of technology across socioeconomic groups: 73 percent of people with incomes in 60 Argentina formality (percentage points) Panama the top 60 percent use the internet compared to only 53 percent among those in the bottom Mexico Brazil 50 El Salvador 40 percent. 4 FIGURE 1.19. Unmet Basic Needs in 2005 and 2018: Indigenous Colombia Dominican Rep. and Afro-Descendent Populations Peru Paraguay 40 Beyond the overall agenda of job creation, policies to promote more inclusive labor markets Ecuador A. Indigenous population 30 Honduras need to target interventions to the groups that are traditionally excluded and that have suf- Bolivia fered further setbacks in economic participation as a result of the COVID-19 shock. Policies to 20 reduce labor market distortions 100 that affect these groups include making social security con- 10 tributions proportional to the hours worked and capping the growth of the minimum wage to inflation until it reaches a level80that is friendlier to job creation. Closing the gaps between 0 0 1 2 3 4 5 6 7 groups also calls for removing barriers to equitable access to economic opportunities. On gen- Unmet Basic Needs 2018 Hourly earnings (current US$) der, for instance, there is a pending agenda to address barriers in labor regulations that affect 60 women and to increase access and quality to childcare centers so that more women can par- Source: Author’s estimates, based on data from SEDLAC. Note: Data for Colombia are from 2019. Informality defined using the productive definition. ticipate in the labor market. 40 20 HIGHLIGHT 4: 0 0 1.19. Unmet FIGURE 20 Basic 40 2005 and 2018: Needs in 60 80 Indigenous 100 Territorial inequalities are also high, leaving many people and Afro-Descendent Populations Unmet Basic Needs 2005 disconnected from critical services and opportunities. A. Indigenous population B. Afro-descendant population The gap between the richest and poorest region in Colombia is more than double that of other TERRITORIES WITH HIGH SHARES OECD countries.5 Spatial disparities overlap with population groups defined by ethnicity: the 100 OF INDIGENOUS POPULATION ARE 100 ALSO THOSE WITH HIGH LEVELS municipalities with high concentrations of indigenous Colombians have persistently high lev- OF VULNERABILITY. els of unmet basic needs, and Afro-Colombians predominantly live in the urban areas where, 80 Unmet basic needs in 2005 between 2005 and 2018, unmet basic needs remained higher than in other cities. 80 and 2018; size of the circle 2018 showing the share of indige- 2018 Inadequate housing and infrastructure are the main sources of inequality within cities. Out of 60 nous population per munici- Needs 60 Needs the 5.1 million households with housing vulnerability6 in Colombia, almost 4.4 million are in pality Basic urban areas. Inequality has worsened in the territories that have been most affected by the Basic 40 armed conflict, which has intensified disparities in access to productive factors, in particular 40 Unmet Unmet rural land. Colombia is among the top five most unequal countries in the world in terms of land concentration (Cuesta and Pico 2020b): 81 percent of private land is concentrated in the 20 20 top 1 percent of farms, the highest among 15 countries in the region and significantly higher than the regional average of 52 percent (Guereña 2017). 0 00 20 40 60 80 100 0 20 40 60 80 100 Reducing territorial inequality requires policies that strengthen the technical capacity and fis- Unmet Basic Needs 2005 Unmet Basic Needs 2005 cal performance of subnational governments, particularly among those that are lagging and need more support. Expanding connectivity from residential sections of peri-urban areas and B. Afro-descendant population Source: DANE (2005) and (2018a). smaller municipalities to the tertiary and secondary road network and reinforcing housing Notes: DANE (2005) and (2018a), with the size of the circle showing the share of indigenous populations, respectively, in each municipality in Colombia; with circles in grey having a programs can also increase access to opportunities and reduce inequalities. Policies need to 100 percent higher than 0 percent but lower than 0.01 percent, in orange those with a percent better target those population groups that have been historically segregated (e.g., Afro-de- higher than 0.01 percent and lower than 1 percent, and in red those cities with a percent higher than 1 percent. Cities with 0 percent are not in the graph. scendants, indigenous populations, and more recently, migrants). 80 Unmet Basic Needs 2018 60 n to Total Collection of Personal Income FIGURE 1.23. Income Level at which Individuals Pay the PIT me (percent) (% of median income and % of GDP per capita) 40 THE FISCAL SYSTEM HAS A LOW 500 HIGHLIGHT 5 REDISTRIBUTIVE IMPACT; FOR 20 EXAMPLE, THE INCOME LEVEL AT 450 Taxes and transfers do little to address the evident inequalities. WHICH A PERSON STARTS PAYING 400 TAXES IS VERY HIGH. Relative 0 to other countries in the OECD and LAC, in Colombia taxes and transfers do little 350 0 20 40 60 80 100 Income level at which individ- to reduce income inequality. Because deductions and tax thresholds in the personal income uals pay the PIT (% of medi- 300 Unmet Basic Needs 2005 tax (PIT) are very high, individuals start paying it only if their income is very high, about four an income and % of GDP per Percent 250 capita) times the median income. This deprives the state of resources that could be redistributed to Source: DANE (2005) and (2018a). 200 DANE (2005)Also, the poorest. Notes: value and (2018a), withadded tax the size of the(VAT) exemptions circle showing and the share of zero rates, which are meant to make indigenous populations, respectively, in each municipality in Colombia; with circles in grey having a 150 VAT less regressive, end up granting large tax discounts to high-income individuals: more than percent higher than 0 percent but lower than 0.01 percent, in orange those with a percent 100 thanpercent) half (57 higher 0.01 percentofandthe tax lower expenditures than 1 percent, and in on benefit VAT cities red those with athe top three deciles of the income dis- percent higher than 1 percent. Cities with 0 percent are not in the graph. tribution. Moreover, cash transfer programs and subsidies to gas, water, and electricity suffer 50 from large leakages to high-income households. It is estimated that based on their socioeco- 0 4 5 6 7 8 9 10 nomic profile, over 65 percent of households receiving subsidies should either be receiving a Denmark Canada Korea Ireland Israel Czech Republic Netherlands Poland Italy Iceland Mexico Japan Slovenia Luxembourg Switzerland Latvia Spain Belgium Australia United States Germany Norway Austria Slovak Republic France Estonia Finland Sweden Chile Colombia iles of income (market and pension) smaller subsidy or no subsidy at all. Finally, the public pension system generates implicit (and fairly generous) subsides that accrue mostly to recipients of high pensions. Estimated Reported Policies to increase the redistributive impact of the fiscal system include (i) extending the PIT In percent of median income In percent of GDP per capita to the top two deciles of the income distribution in the short run, aiming to extend it to the top FIGURE Source: 4.22. OECD Relative Revenue and Database and Absolute Development of Progressivity World Bank World Familias en Indicators. half of the income distribution over the long run as income increases and poverty is signifi- Acción (percent) cantly reduced; (ii) reducing the list of goods that are VAT exempt, which could be done grad- FAMILIAS EN ACCION HELPS THE 25 25 ually by levying a low rate initially that increases over time to give time to production chains POOR, BUT A LARGE SHARE OF (especially for those goods that are excluded from VAT) to adapt their prices; and (iii) better THE PROGRAM REACHES THE NON-POOR. 20 20 focusing transfers and reducing leakages in utility subsidies. Moreover, creating a dynamic, Percent of total spending Relevance and destination reliable, and integrated single social registry can inform the design and implementation of Percent of income of Familias en Accion trans- 15 15 more effective social expenditure programs. fers, by decile of income 10 10 5 5 0 0 1 2 3 4 5 6 7 8 9 10 Deciles of income (market and pension) Relative Absolute (r.a.) Source: Nuñez et al. (2020). HIGHLIGHT 6: Both internal and external shocks hinder progress toward equality. SHOCKS DERAIL THE PATH TOWARD REDUCING The COVID-19 shock increased poverty 6.8 percentage points in 2020, and 3.6 million more POVERTY AND INEQUALITY; THE CONSEQUENCES people became poor, particularly in urban areas. It also caused extreme poverty to increase OF CLIMATE CHANGE WILL AFFECT MORE INFORMAL WORKERS AND WOMEN. by 5.5 percentage points, leaving 2.6 million more people unable to cover basic food needs. Wage effects in 2050 (percent) with respect The pandemic has also worsened inequities in human capital: it is expected to increase the ensionable Age by Pension Status and baseline, to the 1.26. FIGURE high-impact Aggregate scenario Resources Destined to Pension Benefits rate of learning poverty7 among 10-year-old children from 53 to 60 percent if schools maintain ecile, 2019 by Income Decile, 2017 a hybrid in-person learning program through 2021, or 63 percent if distance learning contin- l Urban Rural Urban Rural Urban Rural Urban ues for the entire year. 2q 3q 3q 4q 4q 5q 5q Skilled Male formal Skilled Female formal However, COVID-19 is just one of the extreme shocks that can worsen inequalities. Climate Skilled Male informal shocks are another. Estimates indicate that households in the lower two quintiles of the in- Skilled Female informal come distribution would suffer income losses from climate change shocks that are on average Unskilled Male formal Unskilled Female formal between 1.5 and 1.6 times higher (as a percent of the pre-shock income) than those suffered Unskilled Male informal by the top quintile. Rural households are estimated to suffer from income losses that are on Unskilled Female informal average between 1.8 and 1.9 times higher than urban households. All Female Yet social assistance programs are not designed to flexibly protect households against shocks, All Male especially climate related, and the consequent depletion of assets. The lack of a social registry All Unskilled with dynamic and accurate information on households and equipped with climate risk assess- All Skilled ment tools limits these programs’ ability to adapt to new circumstances. All Formal All Informal Policy options related to climate mitigation and adaptation policies that are aimed at reduc- -0,08 -0,07 -0,06 -0,05 -0,04 -0,03 -0,02 -0,01 0 0,01 ing the impact of climate shocks on the most vulnerable include (i) strengthening the National mpact scenario Lower-impact scenario Agriculture Extension System by mainstreaming mitigation and adaptation criteria in the De- sed on GEIH 2019. Source: Authors’ calculations, based on the Encuesta Nacional de Presupuestos de los Hogares partmental Extension Plans and building the capacity of agricultural extension service provid- (ENPH) 2016–2017. ers; (ii) consolidating the Sistema de Identificación de Potenciales Beneficiarios de Programas Sociales (Sisbén) (the Identification System for Potential Beneficiaries of Social Programs), social assistance administrative registries, and other key databases into a dynamic, reliable, and integrated single social registry equipped with climate risk assessment tools; and (iii) strengthening carbon pricing by expanding the carbon tax and introducing an emissions trad- ing system. A more detailed set of policy options for this and other challenges is presented in table ES.1, and also presented in each chapter accompanied by relevant international experiences for each policy. 7 TABLE ES.1. Policy Options for a More Equitable B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Ex ecutive S ummary Society in Colombia Observed Equity Gap Drivers of the Equity Gap Policy Options Timing for Implementation Policy Options for More Equitable Access to Human Capital Quality of health care Hospital-centric model of care delivery is ineffec- • Transform service delivery model into a PHC-based model of care adapted to local needs. Medium/long term services tive in meeting the needs of the most vulnerable • Introduce financial incentives in the capitation payment for health insurers and health care populations with chronic conditions who require providers to improve quality care and achieve better health outcomes for the most vulnera- comprehensive care and better integration of ble populations. health and social care • Introduce an accreditation system for health insurers requiring them, among other measures, to develop population health management strategies for the most vulnerable populations. • Strengthen local governance by reporting health inequality data at the municipal level on a regular basis and requiring municipalities to develop plans to reduce health inequalities and assess progress on a yearly basis. Inequality in access to The limited coverage of ICBF ECD services in areas • Simplify the administrative procedures to access ECD services that generally act as entry Immediate/ ICBF’s ECD services where private childcare is not available or afford- barriers, particularly in rural areas. This implies: (1) reducing the number of documents re- short term (for simplifying able to the poorest families means that many quired by verifying information directly with other public agencies; and (2) integrating the the administrative proce- children are placed into low human capital ac- ICBF’s information systems. dures to access ECD ser- cumulation trajectories, with negative effects on • Increase the supply of ICBF services in rural areas. vices) productivity over the lifecycle • Define more effective quality-assessment strategies for ECD services. • Improve the monitoring and evaluation structure of ECD service delivery by: simplifying the Medium/long term (all oth- In rural areas, the fragmentation of ECD service current formats based on input, output, process, and results indicators; and building infor- er measures) modalities may be causing quality gaps in ICBF mation systems to ensure timely, high-quality data collection and processing at the local services that also affect the human capital accu- level, useful to generating alerts to detect problems with service provision. mulation of children over their lifecycle Inequalities in learning The curricular autonomy at the school level, • Introduce core curriculum guidelines for basic competencies. Immediate/ structural inequalities in teacher quality, and in- • Link the Programa Todos a Aprender (which provides pedagogical support to teachers) to short term equities in allocation of financing across schools/ the core curriculum guidelines. regions/municipalities create large inequalities in • Introduce/expand programs that teach students at the right level (focus on developing basic the quality of pedagogical practices math and reading skills based on students’ needs rather than age or grade through, for ex- ample, student tutoring such as Brujula, Aprendamos Todos a Leer). Insufficient access to services in rural areas re- • Complement in-person learning with Adaptive CAL where connectivity is available. mains a concern in post secondary education Inequalities in early Lack of relevance and quality of upper secondary • Introduce a competency-based curriculum in upper secondary school, with socio-occupa- Medium/long term dropout education, combined with low levels of learning tional skills. when entering upper secondary education, gener- • Increase the supply of upper secondary education in rural areas. ates early dropout for disadvantaged students • Strengthen linkages with tertiary education and improve the relevance of the Articulación con la Media program by diversifying the supply of programs offered from SENA (Servicio Nacional de Aprendizaje, the National Training Service) and diversifying providers of the pro- gram, with participation from the local private sector. Persistent inequalities in Inequitable financing of public tertiary education • Financing should move toward a per student allocation of service provision transfers that Full implementation in the access to quality tertiary ed- institutions generates substantial differences in reflects the operational cost of providing a good quality tertiary education and also equity medium term, but there ucation the quality of service provision concerns. are steps that can be taken • Any resources for public tertiary education institutions that are not covering operational in the immediate term to High reliance on private provision and student costs should be pooled to avoid fragmentation and linked to results and institutional im- set up the systems. loans results in high repayment burdens that lim- provement plans. it the demand for student loans for low-income • Income-contingent loan schemes should be implemented in which repayment depends on families income upon graduation. Unequal access to social The Sisbén lacks dynamism because of its census • Create a dynamic, reliable, and integrated single social registry by: (1) integrating Sisbén Immediate/ programs sweep structure, causing data to become outdat- with both administrative data sources and nominalized social programs registries to keep short term for the social ed very quickly the information updated; (2) implementing a co-responsibility clause in the social programs registry consolidation registry to ensure that citizens update their information on Sisbén. The lack of quality of Sisbén data usually explains • Introduce user-centered design into current and future social programs, particularly those the persistently high exclusion and inclusion er- implemented by the DPS, by: (1) removing unnecessary documentation and implementing Medium/long term for the rors from flagship social programs an electronic household folder that stores the basic documentation required by the multiple adjustments of social pro- programs, avoiding the need for citizens to make multiple requests; (2) reactivating family grams Furthermore, even when social programs are pro- support programs, such as UNIDOS, as the initial step for assessing the needs of families and vided by the same public agency, most of them defining personalized integrated poverty reduction intervention plans (i.e., Ruta de la Su- still have separate and independent delivery peración de la Pobreza). chains, which translates into entry barriers and administrative burdens for citizens. Particularly for rural areas, where electronic and G2P pay- ments are difficult, people may need to travel sev- eral times to claim their benefits or get a specific document for registry Policy Options for More Equitable Labor Markets Regulation-driven barriers High transaction costs of formalization • Reduce formalization costs, including by: Medium/long term to access formal sector jobs • Reducing the number of days it takes to register a firm among the poor and • Reducing the costs of firm registration (“registro mercantil”) vulnerable • Expanding the “Ventanilla Unica” (VUEs) across the country • Developing “Formulario Unico” to facilitate the registration of employees High wage labor costs • Limit minimum wage growth to the inflation rate for some years (until the minimum wage Medium to long term reaches a level friendlier to job creation) and establish an hourly minimum wage (so that part-time employment is not penalized). • Create a unit of independent experts to advise the Comisión Permanente de Concertación de Políticas Salariales y Laborales regarding decisions on the minimum wage through analy- ses of economic and social impacts. • Consider reforms to allow for the minimum wage to be adjusted according to the economic context (for example, taking into account productivity and labor market indicators) and not automatically adjusted to inflation as in the current legal setting. High non-wage labor costs • Improve the link between mandatory contributions and benefits. OECD (2019) highlights Medium/long term two strategies to achieve this goal. First, since part of the contributions to health care and family compensation funds is used to finance benefits for which formal workers are not eli- gible, it is important to broaden the sources of funding in order to reduce formal sector em- ployee contributions. Second, some of the recreational and commercial activities funded by the family compensation funds (which are not typically paid from employer contributions in advanced countries) could be made optional for employers. Labor market barriers faced Gender-biased laws and regulations • Make social security contributions proportional to the hours worked (instead of on a full- Medium/ by specific demographic time or weekly basis), which would promote formalization among female workers. Long term groups • Establish and monitor quality and safety standards within childcare centers. • Equalize the age at which men and women can retire with full pension benefits across sec- tors (as in Slovenia). • Introduce legislation prohibiting gender-based discrimination in access to credit (as in Uz- bekistan, Bahrain, Jordan, Maldives). Low spending on ALMPs that work8 • Increase spending on ALMPs by carefully evaluating existing programs in Colombia, as well Medium/ as past experiences from other countries. These should be cost efficient, while minimizing Long term distortions and displacement effects on non-participants. Weak targeting of labor market interventions to- • The Servicio Nacional de Aprendizaje (SENA) and the Servicio Publico de Empleo (SPE) Medium/long term ward minorities could tailor their services better to the specific needs of the NARP, migrant, and displaced populations. Policy Options for a More Redistributive Fiscal System PIT raises little revenue and Very high exemptions and zero-rate bracket re- • Expand the coverage of the PIT to the top two deciles of the income distribution in the short Immediate/short run is not progressive enough duce the base of the PIT to individuals whose run (and aim to expand it to individuals earning the median income over the long run as income belongs to about the top 7 percent of the income increases and poverty is significantly reduced), by reducing allowable deductions income distribution (including by making deductions a flat amount, not proportional to income) and eliminating the zero-rate bracket. VAT exemptions and Exemptions and reduced rates involve goods that • Keep zero rates only for a few goods that are primarily consumed by low-income individuals. Immediate/short run reduced rates generate are consumed by everyone, but in greater share For all other goods, use a 5, 12, and 19 percent VAT rate, depending on the share of total con- a loss of revenues, the by high-income individuals sumption contributed to by high-income individuals. At the same time, increase the amount majority of which benefit and coverage of the VAT devolution program to compensate households with income up to high-income individuals the median income. Transfers and subsidies do For cash transfers, leakages reflect inclusion er- • Improve the targeting system of subsidies by complementing the strata system with infor- Immediate/short run reach the vulnerable, but a rors in the Sisbén scoring system. For subsidies, mation from Sisbén (see concrete proposals in next section). large share of the spending leakages reflect targeting based on outdated leaks to high-income indi- proxies for household income viduals or individuals for whom the benefit is mini- mal relative to their income The pension system suffers Low participation and coverage reflect a system • Increase participation and coverage, possibly linking the pension more directly to the actual Immediate/short run from low participation and that requires a long contribution period in a amount of contributions over the work life of the individual, to make the system actuarially low coverage and as a re- context of high informality and low contribution fair. Reforms should also aim at preserving the financial sustainability of the system, i.e., sult, it faces a deficit, all of density. Also, the replacement rate generates a ensuring that the present value of future payment is matched by the present value of future which effectively generate system that is actuarially generous. The treat- contributions. transfers from general taxa- ment of savings of those individuals who will not tion to high-pension individ- receive a pension generates implicit transfers to uals current retirees Policy Options for More Equitable Opportunities across the Territory Territorial inequalities per- Better coordination between existing plans and • Strengthen subnational governments by (i) enhancing information systems and data man- Immediate/ sist at different scales and tools is needed to enhance, simplify, and stream- agement to feed territorial planning and decision making; (ii) building technical strength short term dimensions line current initiatives to reduce territorial in- in weak subnational governments to plan, design, and manage key infrastructure and ser- equalities vices; and (iii) translating legal reforms into concrete instruments that can guide subnational governments in the implementation of territorial laws/decrees. For example, support from the national government is needed to implement the PGOT. Existing indices could be used to identify those municipalities that need more support (e.g., Índice Municipal de Riesgo de Desastres Ajustado por Capacidades), focusing first on urban cores due to their potential to benefit surrounding territories. Urban-rural inequalities The links between rural and urban areas (e.g., • Strengthening local fiscal performance is key to reducing inequality in the territories. Local Medium/ are high infrastructure, transport, and socioeconomic finances can be improved by (i) enhancing municipal government capacities in managing a long term interaction) are weak, as are the potential mech- multipurpose cadaster and property tax collection, and (ii) deepening the operational link- anisms that can support the reduction of inequal- ages between the cadaster and fiscal management. Since Law 14 (1983), the government ity in territories has had the initiative to strengthen local finances by establishing a national system to link the cadaster and property tax; however, updating the cadaster has been a difficult task. Through current catastro multiproposito initiatives, the objective is to register the rights, restrictions, and responsibilities of land and its owners to enable the introduction of a prop- erty tax. • Reinforce subnational government performance by bringing capacities where they are Immediate/ lacking through agreements or associations between municipalities, departments, and/or short term regions to address a common challenge. Colombia has different alternatives for this, such as the Pactos Territoriales or the EATs, depending on local needs. For transport, support the nascent initiatives of regional transport authorities (ARTs), under consideration by the Cali/ Valle and Bogotá/Cundinamarca governments. • Expand connectivity from residential parts of peri-urban areas and smaller municipalities to Medium/long term the tertiary and secondary road network to improve overall access to jobs, markets, and ser- vices, which tend to concentrate in larger urban areas. The Ministry of Transport has helped departments with their planning and project structuring capacities (Plan Vial Departamen- tal). Similarly, this program could be leveraged to support municipalities in the definition of competencies and instruments needed to manage the tertiary road network for access to fa- cilities. Intra-urban socioeconomic Cities have grown rapidly in the past decades, • Reinforce national housing programs with financial mechanisms to support municipalities Medium/long term inequalities, particularly in but infrastructure has not grown proportionately, in their efforts to increase the amount of available housing, aligning incentives between the the housing and education leaving many behind national government and the cities. This response would vary from city to city, going from dimensions land acquisition to provision of needed infrastructure and services that can better integrate households into the urban fabric to enable inclusive and productive urban areas. Financial mechanisms could be in the form of block grants for municipalities to invest locally and bet- ter complement and guarantee inclusive, well-connected, sustainable housing. The program should have clear eligibility parameters for participating urban areas, and the rules should specify the characteristics that cities must have to be eligible. Colombia is well advanced on identifying its local challenges and priorities (POTs modernos), and financial incentives from the central government can help operationalize them, while better supporting existing and new housing. Poor accessibility at the • Carry out a baseline accessibility analysis and identify the changes in access from improved Immediate/ intra-urban level transport services and infrastructure, which can help inform the national policy for improv- short term ing regional and urban mobility that the government approved in April 2020. Segregation in cities, with • Expand intra-urban spatial data and analysis to properly identify poverty traps and tackle Medium/long term the periphery less equipped urban segregation. The use of local indicators of the spatial association of socioeconomic vulnerability, as the ones used in this diagnostic, can help identify those areas most in need. A prioritization strategy (hoja de ruta) to improve conditions in such areas can help cities guide urban upgrading interventions. • Better target groups of the population that have been historically segregated (e.g., Afro-de- Immediate/ scendants, indigenous populations, migrants). This can be done through 4 steps: short term (i) track progress in human capital accumulation, social inclusion, and anti-discrimination measures; (ii) design policies and programs that meet their specific demands and needs, which helps to reverse the processes of structural discrimination; (iii) integrate protection of cultural rights of ethnic minorities when designing investments; (iv) strengthen their voice and participation in decision-making spaces by supporting the technical, financial, and organizational capacities of indigenous peoples and Afro-descen- dants through their representative associations (e.g., community driven-development). Inefficiency of expenditure Inefficient allocation of subsidies • Restructure the targeting methodology of the subsidies for Servicios publicos domiciliarios Immediate/ meant to reduce intra-urban by integrating household socioeconomic characteristics (beyond dwelling conditions) from short term inequality the Sisbén database, which can increase efficiency in the use of resources and help to better identify those in need. Current methodology of stratification can be complemented by data from Sisbén to better target the most vulnerable (see last column for an example). Policy Options for Addressing the Effects of Climate Change Increased vulnerability to Insufficient investments in resilient CSA • Strengthen the National Agriculture Extension System by mainstreaming mitigation and Medium/long term wage losses due to climate adaptation criteria in the Departmental Extension Plans and building the capacity of agricul- impacts, especially among tural extension service providers. poor rural households and • Strengthen the National System of Agriculture Innovation by financing research on climate Medium/long term informal and female workers adaptation and mitigation. • Pilot and scale up green incentive packages to support mitigation and adaptation (e.g., sub- Medium/long term sidized credit lines; payments for environmental services; tax credits; expanded agriculture insurance, etc.) • Generate guidelines and instruments to establish a framework for climate finance in the ag- Immediate/Short term riculture sector (e.g., green taxonomy for private investment) to promote private financing for sustainable practices. • Support existing initiatives with joint potential for inclusion and mitigation benefits, such as Medium/long term the NAMAs and zero-deforestation agreements, by co-financing the establishment of mea- surement, reporting, and verification systems of GHG emissions or co-funding the invest- ments required. • Encourage MADR, in coordination with DNP and MADS, to define sectoral targets for climate Immediate/Short term co-benefits generated by public sector investments in the agriculture sector and to develop guidance and instruments for monitoring and reporting on such targets. Lack of rapid response ca- Social protection systems are not sufficiently • Finalize, approve, and implement a Plan Integral de Gestión del Cambio Climático (PIGCC) Immediate/Short term pacity to climate effects on adaptive that (i) includes both mitigation and adaptation measures; and (ii) embeds specific ques- vulnerable segments of the tions into the Sisbén and other social program registries to assess the exposure of house- population holds to climate change risks, such as climate-smart cash transfers, or climate-related insurance schemes. • Consolidate Sisbén, social assistance administrative registries, and other key databases into Medium/long term a dynamic, reliable, and integrated single social registry, ensuring that it is equipped with climate risk assessment tools, such as: (i) climate projections and forecasts of impacts on different geographies/ communities; (ii) identification variables for households to identify their potential risks related to climate change; and (iii) early warning systems to identify po- tential climate change risks for households in a timely manner. • Reform social assistance programs so that they are operationally and financially prepared to rapidly and flexibly respond to climate-related events by: (i) expanding coverage and gener- Medium/long term osity, (ii) introducing effective economic inclusion mechanisms to prevent households from falling back into poverty, and (iii) promoting insurance schemes (private, public, or commu- nity-based) for generating climate-resilient livelihoods. Unequal impact of second- Rudimentary carbon pricing infrastructure • Strengthen carbon pricing by: Immediate/Short term ary effects of GHG emissions in place expanding the carbon tax and and insufficient resources to introducing an emissions trading system. fund adaptation measures Medium/long term Endnotes 1 This is according to 2019 data from the World Bank Equity Lab. 6 The vulnerability of households is defined by: i) the number of households living 2 Based on 2021 March unemployment data. in overcrowded housing units; and (b) the number of households living in housing 3 See Narayan et al. (2018) and the World Bank’s 2018 Global Database on In- units with a qualitative housing deficit (National Statistics Institute and Ministry of tergenerational Mobility, https://www.worldbank.org/en/topic/poverty/brief/ Housing, City and Territory). Municipalities with the highest quantitative and qual- what-is-the-global-database-on-intergenerational-mobility-gdim. Intergenerational itative housing deficit areas are located in the Caribbean and the Central regions; income mobility (intergenerational persistence in earnings or relative intergenera- however, when looking at the connection of households to public services, the allo- tional income mobility) is measured as the regression coefficient of a child’s earn- cation of vulnerable households is somewhat different. ings on the parents’ earnings. A higher coefficient indicates greater persistence and 7 Learning poor refers to children who can only read and understand a basic text. thus lower mobility. 8 OECD (2019), chapter 2. 4 Based on Gallup and World Development Institute (WDI) data. 5 This is based on the ratio of regional GDP per capita for the richest and poorest re- gions in 2018. 8 BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LOMBI A CHAPTER 1. Overview of the Challenge 1.1 9 B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Overview of the challenge Introduction Argen Dominic U Unit FIGURE 1.1. Gini Coefficient for Colombia (2008–2020)... Income inequality in Colombia is very high. Colombia’s income inequality in 2019 was the FIGURE 1.1. Gini Coefficient for Colombia (2008–2020) highest among all OECD countries and among most countries in the Latin America and Ca- 0,6 ribbean (LAC) region. Moreover, inequality in Colombia has been increasing since 2018 and 0,55 was exacerbated further by the COVID-19 shock (figure 1.1). At the pace of decline observed between 2008 and 2019, just before the pandemic broke out, it would take Colombia at least 0,5 three and a half decades to reach the average inequality level of OECD countries.1 0,45 Beyond income inequality, Colombia scores poorly in many other dimensions of in- 0,4 equality. Figure 1.2 considers different indicators of individual well-being and welfare. For 0,35 each indicator, the figure shows where Colombia stands in terms of inequality relative to 0,3 the best performers in OECD and LAC (distance to the frontier). In this international compar- Cze Slov ison, Colombia scores toward the bottom in many indicators. 0,25 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Source: A FIGURE 1.2. Colombia’s Distance to the Frontier from LAC and OECD Countries across Inequality Dimensions ...and Compared to OECD compared to and LAC OECD and LAC Countries (circa2018) Countries (circa 2018) Integrate 9 This figure is the number of years it will take for Colombia to go from 0.526 (Colombia’s Equity La Gini in 2019) to 0.3175 (the average OECD Gini), keeping the reduction speed constant to Note: La Brazil the average from 2008 to 2019, i.e., 0.003727 Gini points per year. most LAC Colombia St. Lucia Gini Panama Guatemala Costa Rica Honduras Fiscal system Nicaragua redistributive impact Ecuador Paraguay Mexico Chile Argentina (urban) Dominican Republic Bolivia Peru Uruguay United States El Salvador United Kingdom Lithuania Latvia Israel Korea Poverty gaps Italy by ethnicity Spain FIGURE 1.1. Gini Coefficient for Colombia (2008–2020)... Afro-descendants* Australia Luxembourg Inter-generational Portugal 0,6 Greece income mobility Estonia Canada 0,55 France Switzerland Ireland 0,5 Poverty gaps by Hungary ethnicity Indigenous* Germany Netherlands 0,45 Poland Inter-generational Austria educational mobility 0,4 Sweden Finland Denmark 0,35 Norway Belgium 0,3 Gender gaps: Iceland Slovenia education & labor Czech Republic Slovak Republic 0,25 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 0 0,1 0,2 0,3 0,4 0,5 0,6 Richest - Poorest 20%: Rural-urban: PISA PISA reading scores gap reading scores gap Source: Authors’ calculations, based on Gran Encuesta Integrada de Hogares (GEIH)/Large Integrated Survey of Households (left), OECD income inequality database, and World Bank This figure is the number of years it will take for Colombia to go from 0.526 (Colombia’s 9 Equity Lab ( https://www.worldbank.org/en/topic/poverty/lac-equity-lab1/overview). Source: Data on inequality of opportunities are from the 2018 Global Database on Intergenerational Mobility (GDIM). Data on the redistributive impact of the fiscal system and the Gini coef- Gini in 2019) to 0.3175 (the average OECD Gini), keeping the reduction speed constant to Note: Latest year available. For most OECD countries data are from 2017 and 2018. For ficient are from the OECD Income Distribution Database, CEQ Institute, and Colombia Commitment Source: Data on inequality of opportunities are from the 2018 Global Database on Intergenerational Mobility to Equity Analysis (GDIM). Data on (latest year available, the redistributive 2017). circafiscal Data and on differences in poverty rates the average from 2008 to 2019, i.e., impact of the 0.003727 system Gini points per year.the Gini coefficient are most LAC countries data are from 2019. Data for Guatemala and Nicaragua are from 2014. between from groups the OECD are from Income the World Distribution Bank Equity Database, Lab andand CEQ Institute, consider onlyCommitment Colombia Latin American countries to Equity with Analysis available (latest year data for this available, dimension circa of on 2017). Data ethnicity. Data differences inon PISA scores poverty are based rates between on au- groups thors’ are fromcalculation usingEquity the World Bank the PISA Lab2018 dataset (see and consider OECD only Latin 2020). countries with available data for this dimension of ethnicity. Data on PISA scores are based on authors’ calculation using the American PISA 2018 dataset (see OECD 2020). Inequalities are particularly acute between specific population groups in detriment FIGURE 1.3. Differences in Poverty Headcount by Group, to women, people living in rural areas, indigenous and Afro-descendant groups, and percentage points, 2019 migrants. Poverty rates are significantly higher among rural, migrant, indigenous, and Af- Richest vs poorest ro-descendant households (figure 1.3).2 A Colombian born in Chocó is five times more likely department 41,41 to be born into poverty than one in Bogotá. Similarly, a woman in Colombia is 1.7 times Indigenous - None 27,34 more likely to be unemployed than a man. An indigenous Colombian attains on average over Migrant - 24,97 two years less schooling than a non-indigenous one. Two-thirds of the children of Venezue- non-migrant lan migrants (nearly 250,000 children, ages 5–18) are not enrolled in school compared to 8 Rural - Urban 15,18 percent of other children. Afro-Colombians are twice as likely to live in slums compared to Afro-descendant 10,14 non-Afro-descendants. When group characteristics overlap, exclusions from opportunities - None reinforce one another. For example, the income gap between a woman and the average Co- Household head woman - Man 3,77 lombian man is greater if the woman is indigenous or if she lives in a rural area, everything 0 10 20 30 40 50 else being equal. Source: Authors’ calculations, based on GEIH 2019. Inequalities also endure across generations, and Colombia has one of the highest per- Note: (i) The Afro-descendant (or NARP) group includes Black, Afro-Colombian, Raizal, and Palenquero populations. The Afro-descendant and indigenous groups’ rates are compared sistence of inequality rates between one generation and the next. The level of education of to those of people with no ethnic identification (group “none”). (ii) The category wom- the parents has a strong positive influence on the educational attainment of children.3 This is an-man makes references to the sex of the household head. (iii) The comparison among departments shows the differences between the one with the largest and smallest indicator. common in many countries. However, among 146 countries included in the World Bank’s Glob- al Database on Intergenerational Mobility (GDIM), Colombia places 122nd in the persistence of education across generations and ranks similarly poorly in educational mobility (figure 1.4). FIGURE 1.4. Percent of Children Born in Families with Parents FIGURE 1 This is the case despite progress in educational attainment and even if there has been an in- in the First Two Quartiles of Educational Attainment who Level th crease in the share of children who are better educated than their parents (figure 1.5). In Colom- Manage to Reach the Highest Quartile (relative mobility) bia, if a person is born to parents in the bottom half of the educational attainment ladder, her 30% 0,9 chance of reaching the top 25 percent of educational achievement is only around 10 percent (it 0,8 is around 15 percent in the median developing country) (Narayan et al. 2018).4 25% 0,7 Even more salient is the persistence of earnings from one generation to the next. (figure 20% 0,6 COL BRA CRI 1.6).5 Among the 75 countries for which data on the intergenerational persistence of earnings ARG VEN ECU PER DOM 0,5 are available, Colombia ranks highest. Concretely, if a parent earns double the amount earned 15% MEX URY PRY HND NIC BOL 0,4 by another parent, his or her child will on average earn more than double than the child of CHL PAN 10% 0,3 the lower-income parent (Narayan et al. 2018). An analysis of 48 countries with available data shows that for Colombia, the contribution of parental education to the children’s income is 0,2 5% higher than for countries with similar income levels; in addition, around three-quarters of the 0,1 intergenerational persistence in income is explained by the effect of all other parental char- 0% 0 1 acteristics, other than education, on the income of the children (e.g., better connectivity and infrastructure, access to jobs, or availability of credit). The largest impact of this channel is common across countries and averages 80 percent of the total persistence. Source: 2018 GDIM. Source: 201 Moreover, inequality in Colombia extends beyond the material aspects of livelihood. Despite being on average a “happy”FIGURE thereof 1.4. Percent country, is Children Born in Families large heterogeneity with among Parents Colombi - FIGURE 1.5. Percent of Children with a Higher Education ans about subjective perceptions the inof First Two6Quartiles well-being. Colombia of Educational ranks 37th out Attainment who in of 156 countries Level than their Parents (absolute mobility) Manage to Reach the Highest Quartile (relative mobility) terms of subjective well-being (SWB). Yet, it ranks 16th in terms of disparities (measured as the of 156 countries (figure 1.8). As with other inequalities, there standard deviation of SWB) out 30% 0,9 are considerable individual and spatial differences in SWB. Individuals with lower education, 0,8 25% or poor evaluate their well-being at significantly lower living in rural areas, and unemployed 0,7 levels than the average individual, and there are similarly large differences between regions in 20% BRA 0,6 reported levels of well-being (see Annex 1 for further CRI ARG details). COL VEN ECU PER DOM 0,5 15% MEX URY PRY HND NIC BOL 0,4 CHL PAN 10% 0,3 0,2 5% 0,1 0% 0 1940 1950 1960 1970 1980 Promedio Padre Máximo Madre Source: 2018 GDIM. Source: 2018 GDIM. FIGURE 1.6. Intergenerational Income Persistence 1,2 COL ECU GTM 1 PAN BOL 0,8 PER BRA 0,6 CHL 0,4 0,2 0 Source: 2018 GDIM. Note: 75 countries are included in the analysis. The relative intergenerational income mo- bility should be read as follows. If a Colombian earns double the amount another man earns, the first man’s son may be expected to make, as an adult, more than double than the son of the lower-income man (as the coefficient for Colombia is larger than 1). FIGURE 1.7. Gini coefficient (right) and Share of People Inequalities FIGURE are a concern 1.8. Well-Being for Colombians Score (0-10) . Around and Standard four out of five Colombians believe Deviation Reporting Unfair Income Distribution (left) that of the distribution Subjective of income is unfair or severely unfair7 (figure 1.7), and this perception Well-Being has increased even in periods of slight declines in the Gini coefficient. At the policy level, 100 0,6 the deep inequalities that exist in Colombia have not gone unnoticed and have taken center 90 stage in recent years. Mitigating inequality has ranked high on the government’s agenda as 0,55 80 laid out in the National Development Plan (2018–2022). Tackling inequities is currently even 70 0,5 more important to the goal of reducing poverty in Colombia, where over a decade of prog- 60 ress has been erased by the COVID-19 crisis. 0,45 50 Reducing inequalities is not only an objective in itself on moral grounds, but it also makes 0,4 40 economic sense. Confronting inequities is important to reducing poverty, boosting growth, 30 0,35 and promoting social cohesion (World Bank 2016a). Lifting the educational attainment and 20 health of the disadvantaged improves overall human capital and the growth potential of the 0,3 10 economy. Addressing market imperfections or inequitable institutions (or policies) promotes 0 0,25 a more efficient use of the talent pool within the population. As an illustration, the loss of 2001 2002 2007 2009 2010 2011 2013 2015 2016 2017 2018 GDP per capita from gender gaps in labor force participation and education in Colombia for Injusta-Muy injusta (izquierda) Gini (derecha) the next three decades is estimated to reach 12.6 percent by 2050 (Devadas and Kim 2020). Finally, closing inequality of opportunities between population groups promotes stability and Source: Authors, with data from Latinobarometer and the National Administrative Source: Burger, Hendriks, and Ianchovichina (2020). Department of Statistics/ Departamento Administrativo Nacional de Estadística (DANE). social Notes: cohesion. Includes In other 56 countries. words, For both reducing indicators inequality the average and of the 2010–18 promoting period is economic growth can be taken from Gallup World Poll data. complementary goals. FIGURE 1.8. Well-Being Score (0-10) and Standard Deviation of Subjective Well-Being 8 DK NL FICH IS NO SE NZ CA IL AU AT CR Average subjective well-being (0-10) US PR 7 LU BE IE GB AE OM MX DE BR PA FR CZ CL QA SG ARUY ES TH KWMT SA SR TT Colombia GT ITSK 6 PLJP SI KR UZ EC CY BZ SV NI VE MYKZLT BO BH MU PE EE MD PY KO LVBY HR RU LY JM TM TR RO HK DZ PK ME VN ID HU PT GR JO PH HN BT KGTJ CN BA NGRS DO 5 AZ MN MA SO LB MK AL LA TN PS ZA DJ GH IQ IR BD UA NP CM ZM MZ MR BG AM SZ NA CG GMSL CD ETMM GE IN KE GA AO CI SN SD LKML NE EG ZW KH BF UG LS TD HT GN BJ 4 KMMG MW LR YE BW AF TG TZ SY RW BI CF SS 3 1.25 1.5 1.75 2 2.25 2.5 2.75 3 Subjective well-being inequality (standard deviation) Source: Burger, Hendriks, and Ianchovichina (2020). Notes: Includes 56 countries. For both indicators the average of the 2010–18 period is taken from Gallup World Poll data. 10 REPORT: Building an Equitable Society in Colombia BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Overview of the challenge This report analyzes Colombia’s standing as an equitable society. An equitable society is defined as one in which individuals have equal opportunities to pursue a life of their choos- ing, regardless of the circumstances they were born into, and they are not subject to poverty (World Bank 2005). This report documents and describes the main sources of inequality in Colombia, identifies the drivers of those disparities, and proposes a menu of policy options to address the causes of inequality and to promote a more equitable society. To explore the factors that lead to disparities between groups, the report relies on the assets-based framework (figure 1.9) (Attanasio and Székely 1999; Carter and Barrett 2006; Bussolo and Lopez-Calva 2014). The framework is centered on the elements that determine the capacity of an individual or a household to generate market income, going from the assets they possess and accumulate to the uses and returns they obtain from those assets. It also describes how shocks and policies can impact that market income. The report is not meant to be comprehensive across all dimensions through which inequities may materialize, but rather focuses on those that, through initial exploration, have been identified as key drivers of inequality in Colombia. FIGURE 1.9. Assessing Equity in Colombia through the Assets-Based Framework FIGURE 1.9. Assessing Equity in Colombia through the Assets-Based Framework Risks and shocks Stock of assets Market income can be influenced or impacted by the risk environment Individuals have different endowments of assets. This includes human capital (incorporat- and various shocks. Natural disasters and the effects of climate change ing education and health, as well as access to certain services such as water and sanitation), can affect, for instance, households’ physical assets, but they can also physical assets such as land and property, financial assets, natural capital such as water and lead to disinvestment in human capital. Economic crises, such as the forests, and social capital encompassing the norms and behaviors that drive relations in COVID-19 shock, can impact workers’ labor market opportunities and society. The stock of assets is determined by the medium- to long-term opportunities for earnings, as well as the receipt of private transfers. To the extent that asset accumulation. they lead to a depletion of assets, shocks could have permanent conse- quences on households’ ability to generate market income. ks oc s sh an sk d Ri Inten sity Private transfers of Intensity of use of assets u These include transfers from family and Individuals might differ in the way that they se friends inside the country or outside in the can access markets to obtain a yield from of form of remittances. these assets (for example, some individuals ass are more likely, at equal asset, to predomi- ets nantly access informal types of jobs). The intensity of use refers to how assets are used Stock to produce income, and particularly relevant is the participation in labor markets. of assets ansfers ts tr e e s t Returns to assets s Policies and institutions a a v Pri to Even with equal access, the yield on the Policies and institutions can impact the rn same asset may differ (for example, women u ability of different groups to generate market t are likely to receive lower wages than a man, Re income and also have a redistributive role, even for the same type of job). The returns particularly through fiscal policies. These can that individuals receive for these assets, Policies & include, for example, the role of labor regula- including earnings in the labor markets, from dividends, and from other sources, institutions tions; differences in quality and access in service delivery, including education and affect their income-generating capacity. health; the structure of taxes; and social transfers. Source: Authors, with data from Latinobarometer and the National Administrative Department of Statistics/ Departamento Administrativo Nacional de Estadística (DANE). The report is structured around the main components of the assets-based framework. Specifically: CHAPTER 2 CHAPTER 4 • Focuses on the main assets that individuals, particularly • Explores the redistributive properties of the fiscal sys- the poorer ones, can count on to generate income: educa- tem in Colombia, looking in detail at the extent to which tion and health, or what is known as their human capital. direct and indirect taxes, social spending, and subsidies It also describes how gaps in access to social assistance reduce inequality. can affect the accumulation (or erosion) of these assets CHAPTER 5 and households’ overall resilience to shocks. • Focuses on the drivers of differences in well-being across CHAPTER 3 territories, that is, between urban and rural areas and • Explores the barriers that affect the utilization of and within cities, as a crucial source of inequalities between returns to human capital in the labor market. Although groups in the country. there are a host of different factors driving labor market CHAPTER 6 exclusion, this chapter focuses on the two that are most critical, given the seismic changes experienced by labor • Assesses the long-term impacts of climate change on in- markets across the world: barriers to labor mobility from equality, through its effects on sectoral productivity, la- low-productivity to more dynamic sectors and the un- bor productivity, and the hydroelectric energy supply, an equal access to new technologies that limit the poor’s important energy source in Colombia. access to the jobs of the future, which has been exacer- bated by the COVID-19 shock. Drawing from chapters 2 through 6, the remainder of this overview summarizes the main findings about the drivers of inequalities in Colombia and the key policy options. 1.2. 11 BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Overview of the challenge The Drivers of Inequalities Inequalities in human capital: uneven access to education and health Although income disparities are acute in Colombia, inequality in human capital (educa- tion and health) pre-dates labor market entry and other work opportunities and hence, income inequality. Inequalities in the accumulation of human capital (especially education) or of other assets (e.g., financial and social capital) determine differences in the quality, type, and level of assets that individuals can bring to the labor market and thus the resulting return that individuals can derive from their human capital. Human capital gaps are evidenced by the large disparities that exist in the Human Capital Index (HCI) between income, geographic, and ethnic groups. The HCI is a synthetic index that measures the potential for a young person to reach her full income potential in life. Co- lombia’s HCI is 0.53 for the poorest fifth of the population, indicating that Colombian children TABLE 1.1. Subnational Human Capital Index in this group will reach slightly over half of their lifetime income potential as compared to 0.73 and Components for Colombia for the wealthiest fifth (table 1.1). Gaps in human capital by geography are also substantial, with a strong correlation between poverty levels and the subnational HCI. Income Probability of Sur- Learning-Adjusted Child Stunt- Human Capi- Quintile vival to Age 5 (%) Years of Education ing Rate (%) tal Index The issue is that learning opportunities are not the same for all FIGURE 1.11. Years of Schooling and Learning Gaps across Most childrendifferences in Colombia. thefactors inTwo HCI between stand out:groups—by ethnicity, (i) actual learning (as region, or income group—are Departments Poorest 97 6.3 15 0.53 opposed due to schooling) to gaps differs Yet, greatly in education. across this is notethnic groups the case and for gender disparities. Although Colom- 2 Vaupés 3,6 6,4 2 98 6.9 10 0.58 across space; and (ii) repetition and early dropout are still too high bia shows a high level of gender equity in the components captured byVichada the HCI, women are 3,9 6,5 in rural areas among the poorest segments of the population and MEN T FR ON T IERS 2 0 2 0 —2 1 R EPORT | S EC T I ON 3 99 7.4 9 0.62 less withinlikely some toethnic fully use their groups. stock For of human instance, capital. Thisyears learning-adjusted can be seen in theGuainía Utilization-Adjusted 4,5 7,6 Chocó 5,6 10,3 4 98 7.9 8 0.66 Human Capital of schooling Index are much (UHCI), lower thanwhich adjusts average the HCI for ethnic for the groups, but fraction of the working-age popu- Amazonas 5,7 10,2 the gap is much larger for Afro-Colombians and indigenous people lation that is employed, making it possible to capture the underutilization of human capital in La Guajira 6,4 11,1 Richest 99 8.8 5 0.73 (figure 1.10). Spatial disparities in learning-adjusted years of edu- the labor market. Colombian women have a much lower cation are very large, ranging from 3.6 in Vaupés to 9.0 in Bogotá UHCI (0.33) than men Guaviare (0.48), reflect- 6,4 10,6 Sources: World Bank staff analysis of using data from Ministerio de Salud (2015), DANE Cauca 6,4 10,6 (2018b), and OECD (2020). ing their lower (figure 1.11). level of labor force participation. Putumayo 6,8 10,7 San Andrés y Prov. 6,8 11,7 Valle del Cauca 6,9 10,3 FIGURE The 1.10. issue Learning is that Gaps learning Across Ethnic opportunities Groups are not the same for all FIGURE 1.11. Years of Schooling and Learning Nariño 6,9 Gaps across 10,7 children in Colombia. Two factors stand out: (i) actual learning (as Departments Caldas 7,1 10,8 opposed to schooling) 11,1 11,0 across ethnic groups and differs greatly Córdoba 7,2 12,0 2 10,7 Vaupés 3,6 6,4 7,2 across space; and (ii) repetition and early dropout are still too high Cesar 11,6 Vichada 3,9 6,5 7,3 in rural areas among the poorest segments of the population and T I ON Bolivar 12,3 Guainía CON 4,5 7,6 within some ethnic groups. For instance, learning-adjusted years Magdalena 7,3 12,6 IR Chocó 5,6 10,3 11,6 VE of schooling are much lower than average for ethnic groups, but Caquetá 7,4 | S UN EN Amazonas Sucre 5,7 7,5 10,2 12,4 the gap is much larger for Afro-Colombians and indigenous people UN EN V IR ON MEN T FR ON T IERS 2 0 2 0 —2 1 R EPORT La Guajira Atlántico 6,4 7,6 11,1 11,7 (figure 1.10). Spatial disparities in learning-adjusted years of edu- Guaviare Arauca 6,4 7,6 10,6 11,6 cation are very large, ranging from 3.6 in Vaupés to 9.0 in Bogotá Cauca 6,4 7,7 10,6 11,6 Norte de Santander (figure 1.11). Putumayo 6,8 7,7 10,7 11,8 Antioquia San Andrés y Prov. Meta 6,8 7,8 11,7 11,8 7,6 Valle del Cauca Casanare 6,9 7,9 10,3 11,7 FIGURE 1.10. Learning Gaps Across Ethnic Groups Nariño 6,9 7,9 10,7 11,8 6,3 Huila 6,2 Caldas Tolima 7,1 8,0 10,8 12,5 11,1 11,0 Córdoba Quindío 7,2 8,3 12,012,5 10,7 Cesar 7,2 8,4 11,6 Boyacá 12,2 Bolivar Risaralda 7,3 8,5 12,3 12,7 Magdalena Santander 7,3 8,6 12,6 12,2 No ethnic group Afrocolombian Indigenous Caquetá Cundinamarca 7,4 8,6 11,6 12,3 Sucre Bogotá 7,5 9,0 12,4 12,5 Atlántico 7,6 11,7 Source: World Bank analysis of DANE (2018b) and SABER 11 Test Scores (national standardized Arauca Learning-adjusted Years of Schooling 7,6 11,6 assessment for 11th graders 2019). Norte de Santander 7,7 11,6 Expected Years of Schooling Antioquia 7,7 11,8 Meta 7,8 11,8 The issue is that learning opportunities are 7,6not the same for all children in Colombia. Casanare Source: World 7,9 Bank analysis of the Sistema Integrado de Matrícula 11,7 DANE (SIMAT) (2018), 6,3 Huila Saber 9 (2017). (2018a), and Pruebas 7,9 11,8 Two factors stand out: (i) actual learning (as opposed to schooling) differs 6,2 greatly across Tolima 8,0 12,5 ethnic groups and across space; and (ii) repetition and early dropout are still too high in Quindío 8,3 12,5 rural areas among the poorest segments of the population and within some ethnic groups. Boyacá 8,4 12,2 Risaralda 8,5 12,7 For instance, learning-adjusted years of schooling are much lower than average for ethnic Santander 8,6 12,2 groups, but the gap is much larger for Afro-Colombians No ethnic group people (figure 1.10). and indigenous Indigenous Afrocolombian Cundinamarca 8,6 12,3 Spatial disparities in learning-adjusted years of education are very large, ranging from 3.6 Bogotá 9,0 12,5 in Vaupés to 9.0 in Bogotá (figure 1.11). Source: World Bank analysis of DANE (2018b) and SABER 11 Test Scores (national standardized assessment for 11th graders 2019). Learning-adjusted Years of Schooling Expected Years of Schooling Although there are multiple reasons for low learning outcomes, teacher quality and their allocation to schools play a key role. The main factors that explain the low quality of educa- Source: World Bank analysis of the Sistema Integrado de Matrícula (SIMAT) (2018), DANE tion are: lower preparation at entry (students who enter the teaching profession tend to have (2018a), and Pruebas Saber 9 (2017). significantly lower exam scores than the average upper secondary graduate), the low quality of teacher training, selection bias in teacher-to-school assignments (whereby the best teach- ers choose to teach in schools with the best students), the lack of a national curriculum, in- sufficient investment in the quality of education, and an unbalanced field for funding schools (schools with poorer outcomes systematically receive less funding). Similarly, and despite a substantial expansion in health care coverage in recent years, there are large differences in access to health care of good quality. Regarding access, al- though on average in Colombia, more than 90 percent of the population receives a consultation within 48 hours, there are large variations in waiting times for general medicine consultations across departments. Two factors explain this: (i) costs and distance remain the main barriers to access for lower income individuals; and (ii) hesitation about the perceived service quality remains the major deterrent to the use of health services across the entire population. Differ- ences in access to quality primary health care (PHC) services, especially prenatal care, creates differences in the effective service coverage for the entire population. Regarding the quality of health service provision, two factors stand out: 1. Current formulas to finance health care providers do not account for the risk profile of the typical patient, providing service providers with very little incentive to offer differen- tiated care to patients with different risk factors. 2. Although the country measures health outcomes effectively and often, the information on the quality of service provision is limited and not sufficiently used to better plan and deliver services at the local level. There are standards for service provision, but compli- ance is often limited. The COVID-19 pandemic has worsened inequity in human capital in Colombia, primarily through three channels: 1) lower income across the board has reduced the resources available for human capital investments, 2) medical services not related to COVID-19, including routine care like basic vaccinations and prenatal visits, have decreased, and 3) learning has declined as a result of school closures. In particular, the equivalent of 1.2–1.7 years of learning have been lost. This is between 10 and 25 percent of the pre-COVID level of learning. 12 Inequalities in physical and financial assets BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | Overview of the challenge Inequalities in access to land are severe, particularly among rural households. Indeed, FIGURE 1.12. Use of Financial Services the distribution of productive assets is even more unequal than that of income, as suggested 100 by measures of the land and real estate Gini coefficients, with estimated values of 0.89 and 90 0.68, respectively, compared to 0.52 for income.8 Although unequal possession of land assets 80 may be less problematic in the urban context (where human capital accumulation is arguably 70 % of Firms 60 more relevant as a determinant of access to employment opportunities), the high concen- 50 tration of land remains a cause for concern in rural settings, where the poorest departments 40 30 have the highest levels of land concentration. Between 2000 and 2015, the Gini coefficient 20 for land distribution increased in 67 percent of the departments in rural areas (21 out of 31 10 0 departments), while this proportion in urban areas was 20 percent (Núñez, Parra, and Pira- Small (5-19) Medium (20-99) Large (100+) quive 2017). Indeed, in rural areas, the increase in land concentration has been taking place in With checking/savings account With bank loan the most isolated areas and where resource extraction has historically been high (Ibañez and Muñoz 2011). Source: World Bank, “Enterprise Surveys 2017, Colombia,” https://microdata.world- bank.org/index.php/catalog/3388. Inequalities in financial assets can also limit economic opportunities, particularly en- trepreneurship. The availability of credit is key to promoting entrepreneurship since it per- FIGURE 1.13. Savings Rate by Region mits projects to be funded that otherwise would be constrained by each firm’s limited pool 25 of funds. Indicators of the use of financial services at the private firm level suggest there is 20 unequal access to credit, depending on firm size. Only one out of two small firms (those em- ploying fewer than 20 people) have access to a bank loan compared to 80 percent of medium 15 and large firms (figure 1.12). Similarly, although a quarter of small firms cite access to credit as % 10 a major constraint to conducting business, only 12 percent of bigger firms do.9 These smaller firms overrepresent the less well-off in society: around 75 percent of the households in the 5 first three deciles have a household member working in a microenterprise, while just 40 per- 0 cent of the households in decile 10 do. Inequalities in financial assets (particularly in savings) 2010 2013 2016 are also observed at the individual level by gender, region, and socioeconomic levels (see, Urban Rural for example, figure 1.13). Furthermore, the richest people in the country are capital rentiers Source: “Encuesta Longitudinal Colombiana de la Universidad de Los Andes, ELCA” and not workers who depend on their labor income. They accumulate over two-thirds of the (Rounds 2010, 2013 y 2016). voluntary pensions, around half of the balances and investments in certificates of deposit, around 30 percent of credit from banks, and over half of investments in funds.10 Overall, 70 percent of total cadastral and financial wealth is concentrated among the richest 10 percent of the population, and at 0.718 in 2015, the Gini coefficient of financial wealth was significant- ly higher than that of labor income.11 13 Inequalities in access to good jobs BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Overview of the challenge In addition to differences in the human capital with which they enter the labor mar- FIGURE 1.14. Labor Market Gaps between the Vulnerable and the Average Worker ket, Colombians also face large disparities in access to quality jobs. Colombia has one (with respect to comparison group in 2019, percentage points) of the lowest levels of labor formality among LAC countries. Accordingly, Colombia has the Women second-lowest average real hourly earnings in the region; even in countries with lower levels 10 of GDP per capita, such as Paraguay and Peru, workers receive a higher hourly pay than in 0 Colombia. Although disparities in labor force participation and employment rates between Venezuelan -10 Youth groups are not very large, gaps in access to formal jobs and in earning levels are significant -20 (figure 1.14). The latter is particularly true for Venezuelan immigrants, the indigenous, the -30 low skilled, the rural population, and the poor. Considerable differences in labor incomes are -40 observed between rich and poor people, and between the Afro-descendant and indigenous -50 populations and those that do not self-recognize as part of a minority. Differences persist even -60 when controlling for educational attainment and sector of employment, with lower earnings NARP -70 Low skilled particularly for migrants, rural workers, and women (figure 1.15) and lower returns on their -80 -90 education.12 But low-quality jobs—high informality and low earnings affect vulnerable work- ers the most in Colombia—are often a symptom of low productivity levels. Although the factors driving labor market exclusion are complex and diverse, this report focuses on two that are critical, given recent mega trends in labor markets around the world. The first is related to the existence of distortions that limit the reallocation of workers Indigenous Rural from traditional sectors with low productivity to more dynamic ones. The second is the slow adoption of new technologies among disadvantaged groups and the role of the COVID-19 pan- demic in accelerating those trends. Quintile 2 Quintile 1 The weak reallocation of labor toward high-productivity sectors is consistent with the existence of distortions in the Colombian labor market. Although labor productivity has LFP Employment Formality (legal) Hourly wages grown almost 25 percent over the past 20 years, it was not enough to reduce the gap with developed countries. The weak labor productivity growth experienced since 2008 has been Source: Authors’ estimates based on data from SEDLAC. exclusively driven by within-sector productivity growth instead of the efficient reallocation of Notes: The NARP group includes Black, Afro-Colombian, Raizal, and Palenquero populations. Sample includes individuals aged 15–64 years (except for youth, 15–24 years). The labor force participation, employment, and formality (legal) gap is the difference between the labor force workers across sectors. Without accelerating labor productivity, it will be challenging to in- participation, employment, and formality rates of the corresponding group and that of the comparison group, as a percentage of the latter. crease wages and the overall quality of jobs in Colombia. Yet, stringent labor regulations limit The earnings gap is the difference between the earnings of the corresponding group and that of the comparison group, as a percentage of the latter. Low skilled are those with eight years of education or less. The bottom chart shows absolute changes in gaps over time, and negative labor market fluidity and weaken the creation of formal jobs, especially for vulnerable work- numbers indicate a reduction in the gap. The comparison group for each category is shown in parentheses: a) women (men); b) youth (25–40 years old); c) low skilled (more than 13 years of education); d) rural (urban); e) quintile 1 (quintile 5); f) quintile 2 (quintile 5); g) indigenous ers. High minimum wages and high non-wage labor costs can hurt the creation of job oppor- (non-indigenous and non-NARP); h) NARP (non-indigenous and non-NARP); i) Venezuelan (Colombians and non-Venezuelan migrants). tunities in the formal sector, particularly for low-skilled workers. Almost 20 percent of firms in Colombia claim that labor regulations are a major obstacle to creating new jobs, higher than in most developing countries (figure 1.16). In addition, the minimum wage in Colombia—rel- FIGURE 1.15. Hourly Earnings Gap between Groups, 2019 ative to actual median income levels—is among the highest in the region and higher than (% differences in earnings) ts an among OECD economies, yet minimum wages do not seem to be achieving the goal of raising d en us c wages for the majority of vulnerable workers in Colombia (figure 1.17). Moreover, relative to no s es nt en ge -d ra l om ra ro di ig earnings in the informal sector, Colombia has one of the highest levels of non-wage labor Ru Af In W M 0 costs among LAC countries, second only to Uruguay, mostly owing to mandatory social secu- -0,05 rity contributions. As shown in figure 1.16, these are positively correlated with firms’ reports of the burden of labor regulations. In sum, even though Colombia is labor abundant and has -0,1 low wages, regulations substantially increase the costs of hiring formally, particularly when -0,15 focusing on vulnerable workers with low earnings in the informal sector. -0,2 Unequal access to new technologies fosters inequality in the labor market. The internet low- -0,25 ers information barriers and allows workers to more easily connect to jobs. It also fosters the -0,3 creation of new occupations and more flexible work arrangements and can facilitate access to Source: Authors, based on Mincer equations and the GEIH 2019. markets and finance. Yet globally, Colombia has one the largest disparities in technology use Notes: These results are interpreted as percentage changes in hourly earnings for each across socioeconomic groups. Among 141 countries included in the 2019 Gallup survey, Colom- group, derived from Mincer regressions of log wage with dummy variables for women (vs men), rural (vs urban), migrant (vs non-migrant), and indigenous and Afro-descendants bia ranks 109th in terms of the size of the gap in internet use between the richest top 60 percent (vs non-ethnic identification). and the bottom 40. Moreover, before the pandemic, telecommuting was not only rare but also widely unknown in the private sector. In 2017, a survey indicated that 42 percent of firms in Co- lombia did not know what telecommuting meant, and only 9.5 percent had a telecommuting program for staff (DANE 2017). FIGURE 1.16. Share of Firms Citing Labor Regulations as a FIGURE 1.17. Distribution of Wage Earnings in Colombia Major Constraint (%) and Mandatory Social Security A Earnings distribution by formality status, 2019 B Workers with earnings lower than the minimum wage, by household per Contribution Rate capita income quintiles 50 .002 100 96 97 SSC contribution rate, % 40 90 COL 80 30 .0015 71 72 70 60 percentage 20 .001 49 51 50 10 40 33 34 30 0 .0005 20 0 10 20 30 40 50 12 12 10 Labor regulations as a major constraint 0 0 0 1000 2000 3000 4000 5000 Q1 Q2 Q3 Q4 Q5 Source: Authors’ elaboration, based on data from Colombia’s Social Security Administra- tion for 2019. (https://www.ssa.gov/policy/docs/progdesc/index.html), and the World 2019 monthly earnings in 1,000 current pesos 2008 2019 Bank Enterprise Survey (latest available year). Note: The horizontal axis shows the share of firms identifying labor regulations as a major Informal salaried Formal salaried Informal self−employed constraint to doing business. The vertical axis shows the mandatory social security contri- bution rate (both for employees and employers). Source: Authors’ estimates, based on GEIH 2008–2019. Note: a) Solid lines show the distribution of earnings for informal and formal workers according to the legal definition. The vertical dashed line shows the minimum wage. b) Each bar shows the share of workers with earnings lower than the minimum wage. In contrast to cross-country trends, jobs that are intensive in the tasks that complement new technologies are shrinking in Colombia, while jobs that are intensive in tasks that can be automated are increasing. Moreover, the profile of workers holding the jobs of the future includes urban and highly educated people working in large firms or in the public sec- tor. In contrast, low-skilled rural workers in the informal economy are lagging behind. For example, although 57 percent of workers with 13 or more years of education have a job that is highly intensive in analytical tasks, only 13 percent of workers with less than eight years of education do. Gaps in access to technology-intensive occupations are highly linked to gaps in educational attainment, and tasks embedded in jobs account for 30 percent of earnings inequality in Colombia. The labor market trends that have been taking place over the past two decades experi- enced a sudden acceleration during COVID-19, particularly in the use of technologies to access work opportunities. The inability to work from home due to the characteristics of the job or the lack of connectivity may exacerbate the exclusion of vulnerable groups. Although more than a third of workers with high levels of education have jobs amenable to telework- ing, only one out of 10 workers with low levels of education is in the same situation. Rural and informal workers and those in small firms also have a lower incidence of working-from-home- friendly jobs. Minority groups, such as indigenous groups, Afro-descendants, and Venezuelan migrants, also have jobs that are less amenable to teleworking. Not surprisingly, teleworking jobs are disproportionately held by workers in the richest income quintiles. Last but not least, social norms and discrimination magnify inequalities and certain groups’ access to opportunities. Roughly 5 percent of Colombian girls marry before they turn 18 years old, which substantially limits their agency or their ability to make choices to achieve desired outcomes (World Bank 2012). Gender violence is also a grave concern in Colombia; al- most one in two married women and more than one in four women in a union reported having experienced emotional and physical violence at some point in their relationship, perpetuat- ed by their partners. Gender norms that dictate a disproportionate responsibility for women in household tasks and care work can also limit access to opportunities (World Bank 2019a). Family responsibilities are stated by economically inactive women as the main reason for not looking for a job, and this is the case to a greater extent in rural areas: in 2017, 28.6 percent of women compared to only 1.1 percent of men identified family care as an impediment to work (World Bank 2019a). The LGBT+ community is a minority strongly discriminated against in Co- lombia, with one out of every four of its members reporting discrimination at the workplace and over half suffering mental health disorders, such as anxiety or suicidal thoughts (Choi et al. 2019). Another salient minority in Colombia (given their vulnerability) is that of migrants, in particular those of Venezuelan origin. As of early 2020, more than 50 percent of Colombian citizens disagreed with the government’s policy of receiving migrants, and up to 70 percent re- ported having a negative opinion about Venezuelan migrants residing in the country.13 In sum, certain social norms that put some groups at a disadvantage, as well as discrimination target- ed at certain groups or vulnerable minorities, exacerbate inequalities, and if these attitudes are not addressed, they could impede policy efforts to promote equity. U N EN VIRO 14 Large territorial gaps in opportunities across the country BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Overview of the challenge FIGURE 1.18. Gaps in Education between Municipalities and Across countries, well-planned territorial development that integrates strong capacities 1.18. Gaps FIGURERegio in Education ns: Avera ge Testbetween Municipalities Scores by Municipalityand at the subnational level and clear financial mechanisms for needed investments leads Regions: Average Test Scores by Municipality to lower poverty 2 and inequality. Better access to public goods and services (e.g., educa- tion, health services, connective infrastructure) across territories can enable better quality of FIGURE 1.18. Gaps in Education between Municipalities and UN ENVIRO NMENT FRONT IERS 2020— 21 REPOR T | SEC TION life, higher levels of productivity, and wider access to job opportunities, all of which have an Regions: Average Test Scores by Municipality impact on the various elements that enhance (or undermine) the capacity of an individual or household to generate income. Territorial inequalities in Colombia are high, in particular in access to infrastructure, and the same regions tend to be vulnerable and far from opportunity across multiple dimen- sions. Regional inequalities within Colombia are twice as high as those in other OECD coun- tries. These high inequalities persist at different spatial scales (region, urban-rural, intra-urban) and various dimensions related to economic and social issues and accessibility. Quality of housing, measured by connection to services and overcrowding, represents the higher gap Source: Saber 11, average by municipality and region. between regions. In particular, having no natural gas, no sewerage, and no piped water are the three socioeconomic characteristics along which regions differ the most. When consider- ing levels of vulnerability14 across regions, the most acute sources of regional inequality are related to education and access to the internet. Yet, though unequal access to services is an essential aspect of territorial inequality, it is not, as explained above, only about access but also about quality. There are important quality gaps in education between municipalities and regions (figure 1.18). Spatial disparities overlap with population groups defined by ethnicity. Municipalities in Colombia that concentrate the highest number of indigenous people are also experiencing higher levels of unmet basic needs (figure 1.19). Similarly, Afro-descendants are mostly con- centrated in urban areas where unmet basic needs have decreased less between 2005 and 2018 compared to other cities. FIGURE 1.19. Unmet Basic Needs in 2005 and 2018: Indigenous and Afro-Descendent Populations A. Indigenous population Source: Saber 11, average by municipality and region. 100 Source: Saber 11, average by municipality and region. 80 Cities in Colombia are reducing multidimensional poverty in their urban-rural territo- Source: Saber 11, average by municipality and region. ries, but they are less effective at reducing monetary inequality. When one looks at the non-monetary dimensions of poverty, it is possible to see that in Colombia (as in other coun- Unmet Basic Needs 2018 60 tries), cities help reduce inequality in the areas surrounding them. Cities with more developed local government institutions, higher fiscal and managerial performance, and tighter connec- tivity can foster territorial development in their rural hinterland. Yet, this is not happening for income inequality. A key ingredient for this to occur is the strength of the economic, insti- 40 tutional, and fiscal capabilities of municipalities. These results in Colombia can be linked to worldwide evidence that cities offer better socioeconomic conditions than rural territories (reducing multidimensional poverty), but that living in a city is expensive (having a less im- 20 mediate impact on the reduction of monetary poverty). Moreover, improvements in fiscal and municipal government performance as well as connective infrastructure are associated with lower multidimensional and monetary poverty in urban-rural territories. 0 Territorial inequalities persist even within city boundaries, as cities in Colombia are grow- 0 20 40 60 80 100 ing without the needed infrastructure, leaving many disconnected from opportunity. The Unmet Basic Needs 2005 main contributors to intra-urban inequality are housing and education, and many urban dwell- ers remain far away from basic services (13 percent of cities have the closest clinic/hospital at a B. Afro-descendant population distance of more than 15 kilometers). More specifically, quality of housing, when measured by connectivity to services and overcrowding, is poor, and the illiteracy and school absence rates are producing greater disparities in education. When analyzing capital cities around Colombia with 100 2018 Census data, it is found that people located in blocks close to the city center need to travel shorter distances to get to their desired urban facilities. This is what occurs when cities expand without the needed pairing between infrastructure and services: most facilities remain in the center of the city. Inhabitants located on the periphery, especially displaced populations due to 80 conflict within Colombia or in neighboring Venezuela, are usually the most disadvantaged. For example, in Bogotá, 52 percent of the victims of conflict are concentrated in only five of the city’s 20 localities, and within them, they are still located in the most peripheral and depressed areas. It is not only the periphery that reveals a spatial concentration of socioeconomic vulnerabilities, Unmet Basic Needs 2018 60 however; Colombian capital cities show this as well. Residential segregation is an undesirable characteristic, as it has been linked to several dimensions of inequality, including labor, educa- tion, health, and housing. 40 Territorial inequalities in Colombia are closely linked to the dynamics of the armed con- flict that has affected the country for decades. The growth of armed conflict beginning in the mid-1990s rapidly increased population displacement to urban areas, eventually leading 20 to more than 6 million internally displaced persons, the second-largest number of displaced persons in the world after Syria. At the same time, the Colombian countryside experienced a massive exodus, causing a serious decline in its productivity and accelerating institutional weakening. As such, in recent decades, inequality has intensified in the territories most affect- 0 ed by the country’s internal conflict, which has led to unequal access to productive factors, in 0 20 40 60 80 100 particular land (Gáfaro, Ibañez, and Zarruk 2012). These predominantly rural and peripheral Unmet Basic Needs 2005 areas have had a weak government presence, characterized by high levels of informality in property ownership and outdated cadastral information, limiting the safeguard of land prop- Notes: DANE (2005) and (2018a), with the size of the circle showing the share of indigenous and Afro-descendant populations, respective- ly, in each municipality in Colombia; with circles in grey having a percent higher than 0 percent but lower than 0.01 percent, in orange erty rights and facilitating their illegal violent appropriation. This dynamic has been further those with a percent higher than 0.01 percent and lower than 1 percent, and in red those cities with a percent higher than 1 percent. Cities with 0 percent are not in the graph. fueled by the rise of illegal economic activities that sought control of territories for the pro- duction and trafficking of illicit goods. These characteristics have prolonged a vicious cycle in which there are disputes over territorial control, high rates of forced displacement, aban- donment or stripping of movable or immovable property, and the subsequent loss of access to productive economic and social capital by the rural population. In sum, decades of forced displacement and land dispossession in rural areas have increased the concentration of pro- duction factors, intensifying the social, economic, and political exclusion of the poorest pop- ulations and exacerbating inequalities over time. 15 Fiscal policy and inequality B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Overview of the challenge The fiscal system, one of the key instruments in influencing the redistribution of income and assets, has only a small impact on reducing income inequalities in Colombia (figure 1.20). This reflects the combination of a tax system that is not sufficiently progressive and social programs that are limited in size and suffer from substantial leakages. Colombia raises fewer taxes in total, and fewer progressive taxes, like the personal income tax (PIT), than do other countries. Colombia raises about 3 percent of GDP fewer taxes than the average across LAC countries, and it raises less than two-thirds of the taxes, in percent of GDP, than the average across OECD countries. In terms of composition, Colombia relies less on the PIT than countries in LAC or the OECD. With fewer resources collected, there is less that can be redistributed or spent on programs to protect the most vulnerable (figure 1.21). FIGURE 1.20. Effects of the Fiscal Policy Measured by the FIGURE 1.21. Level and Composition of Taxes, average Reduction in the Gini 2014–18 (percent of GDP) 0,7 40 0,6 35 30 0,5 Percent of GDP 25 0,4 20 0,3 15 0,2 10 0,1 5 0,0 0 Colombia LAC OECD FRA DEU GRC PRT CZE ITA ROU JPN SVK LTU CAN LVA ARG ECU MEX VEN GTM IRL POL NZL CHL BOL GBR ISR KOR TUR PER Gini before taxes and transfers Gini after taxes and transfers VAT* PIT CIT Green Wealth Other SSC Source: OECD, “OECD Income Distribution Database,” https://www.oecd.org/social/in- *VAT and/or other taxes on consumption come-distribution-database.htm; and Nuñez, Olivieri, Parra and (2020). Source: OECD, Revenue Statistics 2020: Tax Revenue Trends in the OECD (Paris: OECD, 2020). Colombia is not exploiting the opportunity to redistribute more through the PIT. The PIT has a very narrow base and reaches only those individuals in the top decile of the income dis- tribution. Indeed, deductions and the zero-rate income bracket push the income above which the PIT kicks in to a high level. Figure 1.22 shows the income level at which a person starts pay- ing income taxes in Colombia and in other countries, expressed both in a percent of GDP per capita and of median income. Although in most advanced economies the PIT reaches even a person whose income is half the median (or half the GDP per capita), in Colombia, individuals start paying the PIT only when their income is about four times the median income (or about 3.5 times the GDP per capita). As a result, almost the entirety of PIT collection is being contrib- uted by the very top decile of the income distribution (figure 1.23). Individuals in deciles nine or eight (whose income is about twice as much the median income and can be considered better off relative to the rest of the population) are not reached. In addition, deductions and the zero-rate bracket reduce the income over which the PIT is calculated. As a result, most individuals who pay the PIT are effectively taxed on less than 10 percent of their income. Also, value added tax (VAT) exemptions and zero rates, which are meant to make VAT less regressive, end up granting large tax discounts to high-income individuals. As in other countries, VAT tends to weigh more on the budget of low-income households. This is because consumption (almost entirely of necessities) eats up almost all of low-income household in- come, whereas high-income households tend to consume only a fraction of their income on these items. To correct for this, certain goods are either exempt from VAT or are subject to it at a lower rate. Yet, exemptions and zero or reduced rates extend to about 60 percent of goods and services, by total transacted value. Although these exemptions and zero rates significant- ly alleviate the tax burden on low-income individuals, because high-income individuals con- sume much more, including of VAT exempted goods, almost 30 percent of the lost revenue from VAT exemptions or reduced rates originates from individuals in the top decile of the in- come distribution. FIGURE 1.22. Contribution to Total Collection of Personal FIGURE 1.23. Income Level at which Individuals Pay the PIT Income Taxes, by decile of income (percent) (% of median income and % of GDP per capita) 100 500 450 80 400 Percent of total collection 350 60 300 Percent 250 40 200 150 20 100 50 0 0 1 2 3 4 5 6 7 8 9 10 Denmark Canada Korea Ireland Israel Czech Republic Netherlands Poland Italy Iceland Mexico Japan Slovenia Luxembourg Switzerland Latvia Spain Belgium Australia United States Germany Norway Austria Slovak Republic France Estonia Finland Sweden Chile Colombia Deciles of income (market and pension) Estimated Reported In percent of median income In percent of GDP per capita Source: Nuñez et al. (2020). Source: OECD Revenue Database and World Bank World Development Indicators. In addition, cash transfer programs and subsidies to gas, water, and electricity suffer FIGURE 1.24. Pension versus Labor Income: Pseudo-Gini from large leakages to high-income households. The main cash transfer programs (such Evolution by Income Source, 2008–2019 as Más Familias en Acción, or Colombia Mayor) and subsidies to utilities (Subsidios a los Servi- cios Domiciliarios) do help low-income households. For example, the conditional cash transfer 0,73 0,75 0,74 0,74 0,74 0,73 0,73 0,72 0,72 0,72 0,73 0,73 program Más Familias en Acción helps households in the bottom decile of the income distri- bution add about 20 percent of their market income. Similarly, electricity subsidies alleviate utility costs for these same households by about 12 percent of their market income. However, because of inclusion errors, a large slice of spending on these programs and subsidies reaches Pseudo Gini individuals and households with incomes above the median, up to 60 percent for electricity 0,51 0,52 0,51 0,50 0,50 0,49 0,48 subsidies. For gas, the value of the subsidy (which reaches households at all income levels) is 0,48 0,47 0,46 0,47 0,48 at most 2 percent of the recipient’s income. Overall, around two-thirds of households receiv- 0,44 0,44 0,44 0,43 0,43 0,43 0,41 0,36 0,42 0,42 0,42 0,42 0,42 0,42 ing the utility subsidies under the current system should either be receiving a smaller subsidy 0,33 or no subsidy at all. 0,28 0,26 0,25 0,22 0,23 Finally, inequalities that a person suffers during his or her working life are amplified 0,20 0,21 0,22 0,20 during retirement. Pension income is even more unequally distributed than labor income 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 (figure 1.24). Of course, pension income is deferred labor income, so inequality in pensions derives, in part, from inequality in labor income. However, the greater inequality of pension Labor Income Pensions Housing+Others income derives from a pension system that, with a snowball effect, exacerbates labor market Capital Transfers inequalities. In practice, as in other countries in Latin America, Colombia’s pension system makes participation in the contributory system very onerous for individuals with low skills Source: Authors’ calculations, based on GEIH 2008–2019. who are more likely to experience long periods of labor informality and whose wage, in gen- eral, tends to be lower. As a result, people with low income are very likely to end up without a pension or with a non-contributory pension, whereas high-income individuals are very likely to end up with a contributory pension (figure 1.25). In addition, the contributory pension sys- tem is fairly generous in actuarial terms. In other words, if workers were to invest only in the financial market, they would need to save more than they contribute to the pension system in order to receive an income of the same level as their pension. Essentially, the state subsi- dizes returns on savings, and those who end up making those savings are more likely to be high-income individuals. Over the work-retirement cycle, these inequalities are accentuated by a privileged tax treatment of pension income, which effectively reduces the amount of in- come taxes that a person who will obtain a pension pays over the course of his life. Ultimately, pension benefits, including the subsidized portion, are received disproportionately more by those in the upper-income quintiles (figure 1.26). FIGURE 1.25. People in Pensionable Age by Pension Status FIGURE 1.26. Aggregate Resources Destined to Pension and by Per Capita Income Decile, 2019 Benefits by Income Decile, 2017 100% 18 000 90% 15 000 80% 70% 12 000 60% 50% 9 000 40% 6 000 60% 60% 59% 61% 59% 30% 56% 53% 54% 46% 20% 3 000 34% 10% 0% 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Colombia Mayor beneficiary Pension as savings Pensioner Pension as government subsidy Without any protection mechanism Colombia Mayor Source: Authors’ calculations, based on GEIH 2019. Source: Authors’ calculations, based on the Encuesta Nacional de Presupuestos de los Ho- gares (ENPH) 2016–2017. 16 Shocks affect the most vulnerable BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Overview of the challenge The shock of the COVID-19 pandemic reversed the gains in poverty reduction that Colom- bia had made in the past 10 years, with a disproportionate impact on women. Although emergency transfers have mitigated around a quarter of the impact, it is expected that 3.6 million additional people fell into poverty in 2020. The impact was particularly severe at the onset of the crisis, when social distancing measures were most intense. By April 2020, 5.5 million jobs had been lost,15 particularly among women: 27 percent of female workers lost their job compared to 18 percent of male workers, and the labor income of female workers de- clined 40 percent in the second quarter compared to 30 percent for male workers. By Decem- ber 2020, 4.2 million jobs had been regained, but this recovery is not closing the gaps between gender. Almost all the jobs lost by men have been recovered, yet women are still far from a full recovery, which is partly linked to their concentration in sectors that are taking more time to recover, such as tourism and services. The COVID-19 crisis could have a long-term impact on inequalities. A phone survey admin- istered in August 202016 showed that on average 32 percent of households faced some food insecurity, measured by households in which one adult had to skip a meal in the previous month because of a lack of resources. Although 88 percent of students continued with at least one learning activity and maintained contact with their teachers, the emerging international literature emphasizes that the consequences of school closures may be enormous and long lasting. 17 Students from the lower quantiles may be more affected by the shock (World Bank 2020b), as all public schools were closed in 2020 while some private schools remained open. Even before the pandemic, 53 percent of 10-year-old children were “learning poor,” meaning that they could read and understand only a basic text. The pandemic is expected to increase the rate of learning poverty to 60 percent if schools can maintain a hybrid in-person program through 2021, or 63 percent if distance learning continues for the entire year. With distance learning alone, the learning poverty rate could reach 70 percent for those in the poorest quin- tile and 72 percent for rural residents. Other assets may be depleted due to the shock; a 2020 survey18 found that one out of three respondents has used his or her savings to pay for food, health care, or other expenses during the quarantine. Additionally, about 64 percent of re- spondents said that their debts had increased during this period. Climate change constitutes yet another collective shock that is predicted to intensify ex- FIGURE 1.27. Household Income Changes in 2050 (percent) FIGURE 1.28. Aggregate Resources Destin isting disparities. Colombia may be at the front end of commitments to stop climate change, with Respect to the Baseline, by rural and urban income Benefits by Income Decile, 2017 but it will not be spared its effects. Although the country’s vulnerability to climate change is quintiles not unusually high compared to that of other countries, it is among the highest among Latin Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban American countries. Climate change has a direct effect on human and physical capital, la- 1q 1q 2q 2q 3q 3q 4q 4q 5q 5q 0 bor, and land productivity. Unexpected and significant changes in climate conditions modify the productivity of agricultural land and trigger or even accelerate the deterioration of phys- ical capital. Climate change affects working conditions and labor productivity, for example, -0,005 through excessive heat and through the proliferation of heat-related health conditions. Climate change will not affect all Colombians equally. Some individuals and households -0,01 are more exposed to the physical and economic effects of climate change and/or have few or no coping mechanisms to deal with these effects. For example, low-income households have little or no savings and are more vulnerable to changes in wages and product prices. Rural -0,015 households are more exposed to changes in agricultural productivity. Changes in wages due to reductions in labor productivity disproportionally affect informal workers. -0,02 -0,08 -0,07 -0,06 -0,05 -0,04 -0,03 -0,02 -0,01 0 0 Poor, rural Colombians will see their wages decrease substantially more than wealthier Higher-impact scenario Lower-impact scenario and urban segments of the population as a consequence of climate change (figures 1.27 and 1.28). Informal workers’ wages are estimated to drop more than those of formal workers, Source: Authors’ calculations, based on GEIH 2019. Source: Authors’ calculations, based on the Encuesta Nacion gares (ENPH) 2016–2017. and women’s more than men’s. All of these effects will widen the inequality gap. Indeed, a FIGURE 1.27. Household Income Changes in 2050 (percent) FIGURE 1.28. Aggregate Resources Destined to Pension conservative estimate is that climate change would lead to an annual GDP loss of between with Respect to the Baseline, by rural and urban income Benefits by Income Decile, 2017 0.48 percent and 0.88 percent by 2050 compared to a baseline case without climate change. quintiles Climate shocks are estimated to affect predominantly poor rural households. Skilled Male formal Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Skilled Female formal 1q 1q 2q 2q 3q 3q 4q 4q 5q 5q Colombia is ill prepared to mitigate and respond to the equity impacts of climate shocks. Skilled Male informal 0 Climate-smart agriculture (CSA) technologies have been proven to generate immediate bene- Skilled Female informal fits for producers. CSA increases productivity and profitability by improving natural resource Unskilled Male formal efficiency. In the medium run, CSA also leads to greater resilience in production systems and Unskilled Female formal -0,005 Unskilled Male informal better adaptation of agricultural value chains to climate change. However, the rate of adop- Unskilled Female informal tion of CSAs among Colombian farmers is low because of socioeconomic factors (e.g., low -0,01 issues) and because funds (as well as extension services incomes and education, land tenure All Female and financial incentives) are insufficient to support producers in the transition to these new All Male technologies. All Unskilled -0,015 All Skilled Although Colombia has been a front runner in putting a price on carbon emissions through All Formal its carbon tax, its carbon pricing regime has so far had only limited impact. Carbon pricing All Informal -0,02 in Colombia is too low and limited in coverage of greenhouse gas–emitting products, passing -0,08 -0,07 -0,06 -0,05 -0,04 -0,03 -0,02 -0,01 0 0,01 up the opportunity to reap the double-dividend Higher-impact from climate scenario Lower-impact mitigation scenario and greater resource mobilization. Source: Authors’ calculations, based on GEIH 2019. Source: Authors’ calculations, based on the Encuesta Nacional de Presupuestos de los Ho- gares (ENPH) 2016–2017. 17 Inadequate social assistance programs BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | Overview of the challenge Social protection programs can have an important role in promoting the accumulation of human capital and shielding it from shocks. In Colombia, however, this function of social programs is not yet fully developed. The social protection system has a dual role: it provides social assistance to the poor and vulnerable and a safety net to protect all citizens from poverty after income or other shocks materialize. Colombian social assistance programs have been successful in providing income support to those in need, but the current structure of the Colombian social protection system is not prepared to help the poor and the non-poor manage risks and cope with crises and shocks. As an example, estimates for 2017 show that 55 percent of Colombians are not able to cope with the negative effects of natural disasters through their assets. Over time, this causes a decrease in aggregate household consumption (well-being), estimated at roughly 3 percent of GDP (Báez, Fuchs, and Rodriguez-Castelán 2017). More recently, although the creation of the unconditional cash transfer Ingreso Solidario as a response to the COVID-19 pandemic was useful to temporarily extend social assistance to the poor and vulnerable households that were not traditionally covered by flagship programs, under normal circumstances, the access of these groups to social assistance is still insufficient and many households remain highly vulnerable to shocks. The lack of a solid social registry, with dynamic and accurate information on households in Colombia, translates into persistent inclusion and (to a lesser extent) exclusion er- rors in the targeting of social programs and limits the adaptiveness of programs to new circumstances. Since 1995, Colombia has been using the Sistema de Identificación de Poten- ciales Beneficiarios de Programas Sociales (Sisbén) as the main tool for targeting 21 social pro- grams at the national level. The coverage of Sisbén is appropriate, and as of March 2020, the most recent census sweep, Sisbén IV, contained information on more than 39.4 million peo- ple, equivalent to approximately 78 percent of Colombia’s population. The issue is that Sisbén information on households is collected over more than one year (for example, information for Sisbén IV was collected between 2017 and 2019) and is updated only on direct request of a household. This generates two problems: 1. Lack of a prompt assessment of the population in need deprives social programs of dyna- mism, making it difficult to decide how to adapt programs quickly to the materialization of new shocks. This is particularly evident when looking at the responsiveness of social protection programs to climate-related shocks. The country has well-defined institutions and policy frameworks for disaster risk management (DRM) (Sistema Nacional de Gestion del Riesgo de Desastres). However, this system does not include a specific role for the social inclusion sector and key stakeholders, such as the National Unit for Disaster Risk Management (Unidad Nacional para la Gestión del Riesgo de Desastres [UNGRD]) and the Department of Social Prosperity (Departamento para la Prosperidad Social [DPS]), which act together only after a disaster occurs and on a case-by-case basis. This limits socioeco- nomic preparedness and resilience to climate shocks. 2. Social programs lose efficiency over time. If they do not target effectively those in need, they start leaking resources to those who are not in need. This is because the information contained in Sisbén loses accuracy over time, resulting in both inclusion and exclusion errors. For example, in the previous sweep (Sisbén III), inclusion errors (that is, the num- ber of non-poor households classified as poor or vulnerable) were as high as 49.9 percent for monetary poverty and 64.8 percent for multidimensional poverty (according to the Multidimensional Poverty Index [MPI]), meaning that depending on the metric used to evaluate needs, between half and two-thirds of individuals that Sisbén III classified as poor were not actually poor. At the same time, exclusion errors (that is, the number of poor households that are not identified as poor) were as high as 18.8 percent (if evaluat- ed using monetary poverty) or 29 percent (if evaluated using the MPI, see CONPES 2016). In other words, up to one-third of people who were poor were not classified as such un- der Sisbén III. 1.3. 18 B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Overview of the challenge Policy Options to Tackle Inequalities in Colombia Given that inequality is a multidimensional phenomenon, the policy toolkit to reduce the imbalances must be equally multidimensional. Well-targeted and monitored policies can close gaps between socioeconomic and population groups toward a more equitable so- ciety. Policies can address gaps in access to assets (especially human capital), promote their accumulation (in particular, among those falling behind), and protect the stock of assets from shocks. Through labor regulations and institutions, policies can promote equitable access to economic opportunities and influence the impact of norms and discrimination on those op- portunities. By closing the digital divide, policies can increase access to services and jobs to those in remote areas. The redistributive role of fiscal policies can be enhanced, achieving also a more efficient use of public resources. The areas of reforms and policy actions are many, but the analysis suggests a set of key policy objectives that are outlined in greater detail in Chapters 2–6. The analysis and menu of policy options (see also table ES.1) indicate that to reduce inequality over time, policies would need to address the following issues: 1. Ensuring equitable opportunities in human capital is now more urgent than ever. This requires actions aimed at providing human capital accumulation from the early stages of life. Specifically, this can be achieved: • On the education front, by intensifying learning for all by “teaching at the right lev- el.” Implementing a prioritized curriculum for accelerated learning in basic competencies can be effective at recovering the learning lost from COVID-19-related school closures and online learning. A new pedagogical approach in upper-secondary education would also be important to ensure complete learning trajectories. Access to tertiary education needs to be more equitable and flexible, at the same time that the vast inequalities be- tween income and population groups that exist in the quality and relevance of tertiary education need to be reduced. The use of results-based financing for tertiary education is an effective and efficient tool in improving its quality. • In the health sector, by raising the quality of PHC. Information and risk management could be used to target health services and tailor health risk management to different patients. Systems could be scaled up to ensure the quality of service, measuring perfor- mance and equity in something close to real time for all PHC providers. The financing of health care services could be based on the risk profile of the patient. A strong national policy on human resources for health could be implemented with a view to: ensuring that the system has the community health workers, health care managers, technicians, nurses, general practitioners, and specialists who can work together in multidisciplinary teams; adapting health care provision to the local context; and promoting community participation. 2. • Overall, by improving the management and financing of education and health. For Labor markets need to become more inclusive. example, in the education sector, there is a need to use data beyond learning assess- This objective can be achieved: ments more effectively. Current management and information systems are too narrow, outdated, and rigid, focus too much on measuring inputs (rather than outputs), and col- • By increasing the incentives to offer formal jobs. This would require interventions on lect information only from school principals. Information could be used more system- two complementary fronts. The first includes lowering the cost of formalization for firms atically to target interventions more quickly and effectively and monitor progress more by making minimum wages consistent with actual earnings and by reducing non-wage often, which could result in more cost-effective interventions. In health, the situation is labor costs. The second includes rethinking employment protection policies and intro- analogous. Using data can facilitate and improve management. Specifically, manage- ducing instruments that protect people from poverty and allow consumption-smoothing. ment practices could be improved across service providers, local governments, and the To these ends, a “flexicurity” approach would help people manage shocks and facilitate central government, and their coordination enhanced. Finally, the adequacy of financing hiring and firing decisions for firms in the private sector. (in level) could be increased and structured to balance equity with performance results to avoid penalizing poor conditions at birth. • By facilitating workers’ adaptation to the future of work. This would require policies to (i) foster technology adoption among poor and rural workers, which would contribute to promoting equity in the labor market while improving their skills; (ii) advance efforts toward a policy framework that promotes telecommuting and “telecommutable” jobs in an equitable fashion; and (iii) introduce programs specifically geared toward vulnerable groups, such as ethnic minorities and displaced populations, and programs expressly focused on integrating displaced populations into the urban labor market. 3. Improving the progressivity of the tax system and refocusing transfers and subsidies would help to mobilize more resources for redistribution and utilize resources more efficiently. This could be done: • By extending the PIT to the top two deciles of the income distribution in the short run, aiming to extend it to the top half of the income distribution over the long run as income increases and poverty declines. This would involve reducing the maximum amount of the permissible tax deduction and raising the marginal tax rate on the first in- come bracket above zero, but to a level (for example, 5 percent) that is still affordable for individuals with an income at the margin between zero taxable income and some taxable income. It would also involve taxing pension income the same as labor income. • By gradually reducing VAT exemptions and increasing VAT rates on those goods that are currently taxed at a rate below the standard (19 percent) rate. Reducing the list 4. of goods that are exempt could be done gradually by levying a low rate first and increas- Attenuating territorial inequalities can complement other poli- ing it over time. An alternative is to set a 5 percent VAT rate on those goods that are con- cies and amplify the overall results. Reducing territorial inequality sumed almost in equal proportions by different income groups, and a 12 percent VAT rate could be done: on those goods that are consumed in greater share by high-income individuals. Measures to compensate individuals for the extra burden of these changes could be financed with • By strengthening subnational governments. This would make it possible to develop part of the additional VAT revenue. differentiated policies geared at responding to the heterogeneous needs of vulnerable ter- ritories. Very concretely, this would require: (i) enhancing information systems and data • By addressing the inequity of pensions. Achieving this objective would require a com- management to feed territorial planning and decision making (e.g., linking land admin- prehensive reform that increases participation in the system, coverage of the elderly, and istration and cadaster, adequate information on disaster threats and risks); (ii) building the actuarial fairness of the system. To preserve the sustainability of the system, mea- technical strength in weak subnational governments; and (iii) translating legal reforms sures could include revising the number of years needed to obtain a pension (to help into concrete instruments that can guide subnational governments in the implementa- improve the prospects for coverage) and revising the replacement rate (so that it reflects tion of territorial laws/decrees. Strengthening the capacity of local government would the effective amount that an individual has saved over the course of his/her working life). also require (i) adapting legal provisions on Esquemas Asociativos Territoriales (EATs) to A combination of these policies and a better focus of cash transfers and subsidies (see below) more effectively communicate existing legislation among subnational governments, and could help mobilize up to 2 percent of GDP, which could be used to increase the level of trans- (ii) providing tools and resources to build/improve subnational governments’ political, fers and subsidies or extend coverage to those eligible individuals who are currently excluded financial, managerial, and technical capabilities. In the medium term, implementation of and still leave resources (about 1 percent of GDP) that could be used for other programs or Programs with a Territorial Approach (Programas de Desarrollo con Enfoque Territorial fiscal objectives. [PDETs]) (see Chapter 5) needs to be accelerated, underpinned by capacity building to overcome significant institutional challenges. • By better equipping municipalities with new sources of financing and improving the way existing resources are being used to respond to the growing demand for ser- vices. This would require (i) enhancing municipal government capacities in managing multipurpose cadaster and property tax collection, and (ii) deepening the operational linkages between the cadaster and fiscal management. • By improving accessibility within cities, and between urban-rural areas in Colom- bia, which is crucial to bringing opportunities closer to territories. Improving access to jobs, services, and opportunities for vulnerable and disadvantaged populations, and particularly those without access to a private vehicle, has proven to have a significant im- pact on reducing intra-urban inequality in Colombia. Improving pedestrian infrastructure with safety and climate-resilient features can enhance connectivity to public transport, for instance. Carrying out baseline accessibility analyses and understanding the changes in access provided by improving transport services and infrastructure can help inform the national policy on improving regional and urban mobility that the government of Colombia approved in April 2020 (CONPES 2020b). For inter-urban accessibility, over the long term, peri-urban areas and smaller municipalities can improve the connectivity of mainly residential areas to the tertiary and secondary road network as a means to im- prove road safety, reduce fuel consumption, maintenance costs, and travel times, and overall increase access to jobs, markets, and services, which tend to concentrate in larger urban areas. 5. • By targeting investments in cities and towns to improve access to services and bridge spatial divides. Tackling the housing challenges requires working beyond housing itself. The time is now to address the effect of climate change on in- It requires improvements in coordinating land use and transport policies to ensure better equality. This can be done: access to services, amenities, and jobs. National housing programs could be coordinat- • By investing more in in climate-smart agricultural value chains that combine in- ed with financial mechanisms that help municipalities provide connective infrastructure clusion with climate mitigation and adaptation. Productive projects involving CSA (e.g., roads) and basic services (e.g., water, education, health). Neighborhoods should technologies and practices and involving vulnerable rural producers would help pro- be made more inclusive, for instance, by creating places to enhance interactions and mote CSA adoption. To bridge the knowledge gap on CSA practices, investments need providing housing solutions that are both affordable and attractive to different groups. to be accompanied by strengthening agricultural extension services and ensuring that Additionally, affordable housing should be made available through inclusive land-use adequate training and technical assistance on climate-smart technologies and practices regulations and suitable social housing systems. Finally, policies would have to focus on reach small and vulnerable producers. A practical step in this sense is to make mitigation vulnerable groups, such as ethnic minorities and displaced populations. and adaptation key pillars of the Departmental Agricultural Extension Plans. Mobilizing the necessary financing would require boosting private participation, which can be done by consolidating incentives for green investments and adaptation. Finally, in addition to continuing or extending existing initiatives, better coordinated and more effective cli- mate action across different government activities can be facilitated by ensuring that climate co-benefits are systematically considered across government investments in the agriculture sector. • By expanding carbon pricing to combine equity and climate mitigation. This might require: (i) expanding the coverage of the carbon tax to new emissions sources, (ii) ad- 6. justing the level of the carbon tax, and/or (iii) introducing an emissions trading system by regulating Law 1931/2018. To boost equity, these measures would need to be combined with reforms that would allow the recycling of revenues for equity-boosting purposes, for Improving the flexibility, targeting, and efficiency of the social example, by reducing the tax wedge, especially on lower wages. protection system requires increasing the system’s flexibility and informative value to identify eligible beneficiaries. This can be done: • By including socioeconomic characteristics in the stratification system, which could improve the effectiveness and efficiency of utility subsidies. For example, using Co- lombia’s MPI for the identification of the eligible beneficiaries of utility subsidies would significantly reduce inclusion errors. This would make it possible to gear resources where they are most needed, including outside of the subsidy area. • By strengthening the adaptiveness of the social protection system to ensure the resil- ience of the poorest to climate shocks. The social protection system should include an institutional and policy framework that not only provides support rapidly to households affected by a crisis, but that also allows assessing and reducing the exposure of house- holds to climate change risks before crises occur. For example, this could be achieved by integrating a Climate Change Vulnerability Index (like in the Dominican Republic) to social registries. The Sisbén IV already includes some innovations that will be key to bet- ter management of climate-related risks for the poor and the vulnerable. However, the “sweep” structure of the Sisbén data and the lack of regular data updates usually limits the current usability of the data for identifying, mitigating, and responding to most of the unexpected shocks that households face. On the institutional front, Colombia could build a policy framework with better linkages and clearer synergies between the DRM and the social inclusion sectors, specific roles and responsibilities for key stakeholders, and an appropriate balance between ex ante and ex post risk management. Finally, Colombia should ensure that current and future social protection programs are well prepared to rapidly and flexibly respond to climate-related disasters, while contributing to bolstering asset accumulation and the resilience of the poor and vulnerable. Many of these policy options resonate with the issues that matter most to Colombians, particularly for the worse off (Burger, Hendriks, and Ianchovichina 2020). On average, is- sues related to economic opportunities (such as economic sufficiency and optimism about the economy), educational attainment, and digital access matter most to Colombians’ SWB. Yet, for those with lower levels of SWB (the “unhappiest”), education, access to a job, and level of income weigh on their perception of well-being. These same issues, along with low access to health and digital services, drive spatial and intra-group differences in SWB. These priori- ties contrast with those of the better off (the “happiest”), for whom higher-order needs, such as civic engagement and housing affordability, emerge as important. Finally, although the set of proposed policy options is expected to benefit the most dis- advantaged groups disproportionately, addressing additional barriers to opportunities that affect specific population groups will require targeted policies and interventions. Preventing and sanctioning discrimination toward women and the LGBT+ population will help close gaps in accessing opportunities. A foundational step is ensuring that the legislation to curtail discriminatory practices is in place. For instance, as detailed in a recent gender assess- ment (World Bank 2019a), reforms could ensure that legislation prohibits prospective employ- ers from inquiring about the family status of women and establishes the principle of equal pay for equal work. School contexts in Colombia are also vulnerable to racial discrimination, including prejudice-based representations in teaching materials or inappropriate classroom interactions (Freire et al. 2018, 76–77). Moreover, to close gender gaps in access to economic opportunities, enhancing effective access to quality childcare can remove barriers for women to participating in the labor market while delivering the benefits of early childhood develop- ment interventions to disadvantaged children. Moreover, continuing to work toward the social and economic integration of Venezuelan migrants while also attending to the needs of the dis- advantaged in host communities can promote equality. The adoption of temporary protection status for Venezuelan migrants is an important step in that direction. To close spatial dispar- ities, efforts should focus first on those vulnerable territories with a higher concentration of ethnic minorities and displaced populations. Lagging regions require territorial development policies that combine interventions that are not necessarily people-focused (e.g., improving basic infrastructure, expanding education and health systems, enhancing connectivity) with efforts that directly protect the cultural rights of ethnic minorities, for example, strengthening land rights and political autonomy, supporting community-driven development, and protect- ing consultation and consent in decision making. 19 Endnotes BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | Overview of the challenge 1 This figure is the number of years it will take for Colombia to go from 0.526 (Colom- 10 2013 estimates from Núñez, Parra, and Piraquive (2017). bia’s Gini in 2019) to 0.3175 (the average OECD Gini), keeping the reduction speed 11 The total wealth of households is made up of both their real estate and their assets constant to the average from 2008 to 2019, i.e., 0.003727 Gini points per year. in the financial and pension markets. The sources of information for its construc- 2 The 2019 income ratio between urban and rural areas was 2.6. It was 3.6 between tion are IGAC (Geographic Institute Agustín Codazzi) and decentralized cadasters the richest and poorest department; 1.1 between men and women; 2.3 between for estimating the concentration of ownership, information from DECEVAL (Central non-ethnic identification and indigenous; 1.5 between non-ethnic identification and Securities Depository) to clarify the distribution of financial assets, and DIAN (Tax Afro-descendants; and 1.5 between non-migrants and migrants. The 2020 shock and Customs Directorate) tax declarations to clarify everything concerning taxpayer affected these gaps, particularly due to the significant increase of poverty in urban assets. Financial wealth is defined at the individual level and contains information areas. on financial assets, credit, bank accounts, and equity (see Núñez, Parra, and Pira- 3 Around 72 percent of that inequality of opportunity is linked to the education of the quive 2017). parents; 14 percent to the ethnic group the person belongs to; and 8 percent to the 12 Results from Mincer equations on log hourly earnings for 2019 and for each group region of birth, among other characteristics. This is based on authors’ calculations separately show that returns on higher education are lower for women than for using the 2019 GEIH. men, for rural areas than for urban areas, for migrants than for non-migrants, and 4 See also the World Bank’s 2018 Global Database on Intergenerational Mobility for ethnic groups than for groups that do not report an ethnic identification. (GDIM) (https://www.worldbank.org/en/topic/poverty/brief/what-is-the-global-da- 13 Observatory for the Project “Migración Venezuela” based on the Survey by Invamer tabase-on-intergenerational-mobility-gdim). Intergenerational income mobility Gallup Poll, February 2020. (intergenerational persistence in earnings or relative intergenerational income mo- 14 To measure vulnerability, this section uses four socioeconomic dimensions (demo- bility) is measured as the regression coefficient of a child’s earnings on the parents’ graphics, quality housing, education, and labor), using data from DANE (2018a). earnings. A higher coefficient indicates greater persistence and thus lower mobility. 15 When comparing the number of workers in April with the number of workers in Feb- 5 World Bank, 2018 GDIM. ruary. 6 Burger, Hendriks, and Ianchovichina (2020). SWB is assessed using Gallup data, and 16 World Bank. “Colombia COVID-19 High-Frequency Survey,” online, specifically the following question: “Please imagine a ladder with steps numbered https://www.worldbank.org/en/programs/lsms/brief/lsms-launches-high-frequen- from zero at the bottom to 10 at the top. The top of the ladder represents the best cy-phone-surveys-on-covid-19. possible life for you and the bottom of the ladder represents the worst possible life 17 As found by Fuchs-Schündeln, Kuhn, and Tertilt (2020), school closures may have for you. On which step of the ladder would you say you personally feel you stand at negative economic effects on the human capital accumulation of children that arise this time? 0 – Worst possible life; 10 – Best possible life.” only over the long term. The authors find that the children affected by the school 7 Using data from the 2018 Latinobarometer (https://www.latinobarometro.org/lat.jsp). closures could suffer long-term average wage losses of -1 percent. 8 Estimates for the land Gini by Unidad de Planificación Rural Agropecuaria (UPRA) for 18 RECOVR is a panel survey that was conducted in two rounds in May and August 2020 2019. Estimates for the real estate Gini are the authors’ own estimates, based on the (with a third round set for November) and reached 1,508 respondents in the first latest available rounds from the Encuesta Nacional de Presupuestos de los Hogares round and 1,013 in the second through Random Digit Dialing. The RECOVR was de- (ENPH), 2016–2017. Estimates for income Gini based on the 2019 GEIH. veloped by IPA, partnered with Colombia’s National Planning Department (Departa- 9 World Bank, “Enterprise Surveys 2017, Colombia,” https://microdata.worldbank. mento Nacional de Planeación, or DNP) and UNICEF. org/index.php/catalog/3388. BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LOMBI A 20 CHAPTER 2. Human Development and Equity in Colombia Summary 21 BU I L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | H uman Development and Equity Improving equity in social and economic outcomes in Colombia requires improving equity in human capital. Many Afro-descendant and indigenous Colombians, rural residents, and poor people lack access to quality services from early in life, preventing them from reaching their full potential and thus perpetuating inequality across generations. Although Colombia has made large strides in recent years in boosting access to services, service quality remains a source of inequality. The poor quality of services to promote human capital is driven by insuf- ficient focus on measuring quality and management, inequitable financing, and inadequate use of information at all levels of government. This chapter proposes three policy priorities: 1. Boost the quality of education, prioritizing basic competencies in the early grades by in- troducing common curriculum guidelines and expanding effective pedagogical support for teachers and school managers. 2. Focus efforts on improving the quality of the primary health care system. 3. Improve the targeting of social programs through a more dynamic and inclusive social registry and better coordination of social programs. Improving the efficiency of spending through better management and use of data is a priority in all three sectors. The challenge posed by the COVID-19 crisis means that more resources will be needed to address existing inequalities. It is imperative that any additional resources are distributed in an equitable way that also incentivizes results. 2.1 22 BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | Human D evelopment and Equity Introduction Differences in human capital are a primary driver of the persistence of inequality over time. As noted in Chapter 1, Colombia has one of the highest levels of persistence of income— and thus income inequality—across generations. The principal channels for this lack of mobil- ity include human capital. A child who grows up in a poor Afro-Colombian family in Chocó has much less chance of success in life than a child of a wealthy, non-minority family in Bogotá, Medellin, or Cali. Despite large strides by Colombia to make access to basic services near uni- versal in recent decades, marginalized populations still have limited access to quality educa- tion, health, and social protection. Consequently, they accumulate little human capital and continue to be stuck in poverty. Figure 2.1 shows a schematic of this relationship. FIGURE 2.1. The Vicious Cycle of Human Capital and Poverty The government has made great progress in expanding access to services over time, but there are still important gaps in access to quality services for ethnic minorities, people in rural areas, and poor populations. Though primary enrollment is near universal, a child growing up in Vaupes, Vichada, or Guainía completes just half the learning-adjusted schooling Limited access (see below) of an average child in Bogotá. Although 98 percent of Colombians have health to quality health and education insurance, differences in access to and use of health care services remain. For example, in in poor places 2015, 32 percent of women in Chocó reported that their most recent birth took place outside a health center, versus 3 percent for Colombia overall (Ministerio de Salud 2015). This chapter identifies the key bottlenecks to more equitable and effective human capital Low levels formation and proposes policy options to address them. The first half presents a diagnostic High of learning, levels of use of health of inequities in human capital, including new estimates of the Human Capital Index (HCI) for poverty services, and resilience to Colombia broken down by gender, ethnic group, department, and income group. The follow- shocks ing section describes policy priorities to close those gaps and ensure that Colombians can be healthy, smart, and resilient. Several key policy objectives have been identified: Minimal 1. Boost learning in basic education and ensure more equitable access to good quality ter- accumulation tiary education. of human capital 2. Improve the quality of primary health care (PHC). 3. Make social programs more integrated, better targeted, and more dynamic. 4. Three cross-cutting approaches apply to all the policy objectives: · Focus on standards and quality assurance. · Ensure adequate and equitable financing. · Improve management through the effective use of data. 2.2 23 BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Human Development and Equity Diagnostics: Gaps in Human Capital and Inequality in Colombia The HCI provides a useful diagnostic tool to assess human capital in Colombia. The index FIGURE 2.2a. The Impact of Current Investments in Children was developed by World Bank researchers, launched in 2018, and updated in 2020.1 Combin- on Future Income Productivity (HCI) ing measures of survival, schooling, and health, it quantifies the level of human capital that a child can expect to attain by age 18 and is designed to capture the impact of human capital on future productivity, earnings, and economic growth. Figure 2.2a presents a conceptual ver- sion of the HCI. Compared to that of major countries in Latin America, Colombia’s overall HCI is in the middle but substantially below the level of Chile and the successful economies of East Asia. Figure 2.2b shows this comparison. Colombia’s HCI value of 0.60 means that an average Colombian child, in the absence of renewed efforts on human capital, will reach 60 percent of Survival School Health Productivity of his or her potential in terms of productivity and lifetime income. future worker, relative to potential The HCI at the national level hides large gaps within Colombia by income, geography, and ethnic group. For the purposes of this report, a new subnational HCI was estimated for FIGURE 2.2b. Human Capital Index for Colombia and Colombia. Table 2.1 shows this HCI and its major components by quintile of income. The over- Comparators all HCI is 0.53 for the poorest fifth of the population, indicating that Colombian children in this 0,80 group will reach just over half of lifetime income potential as compared to 0.73 for the wealth- 0,69 0,71 0,65 0,65 iest fifth. Across income quintiles there are substantial differences in child stunting rates and 0,60 0,60 0,61 0,55 0,56 years of education after adjusting for learning.2 TABLE 2.1. Subnational Human Capital Index and Components for Colombia Income Quintile Probability of Learning-Adjusted Child Stunting Rate Human Capital Index Survival to Age 5 Years of Education Poorest 97% 6.3 15% 0.53 2 98% 6.9 10% 0.58 3 99% 7.4 9% 0.62 il e a a ico ile a m Ko age e p. ag er m az in bi in Re na Ch ex nt Ch m av nco Br er 4 et 98% 7.9 8% 0.66 a, lo ge M av Vi re try h i Co Ar C un Hig Richest LA 99% 8.8 5% 0.73 Sources: World Bank staff analysis, using Ministerio de Salud (2015), DANE (2018b), and OECD (2019) data. co Source: World Bank (2020). FIGURE 2.3. Human Capital Index vs Poverty Rate by Region Gaps in human capital by geography are also substantial in Colombia. Figure 2.3 plots the Bogotá value of the HCI at the department level (relative to the national HCI) against the poverty rate. 0.05 Cundinamarca Santander Boyacá Departments with values above zero on the vertical scale have higher levels of human capital Difference with respect Risaralda Quindio than the average for all of Colombia, and those with negative values have lower values than to the National HCI Tolima Huila 0.00 Antioquia Meta Norte de Santander the average. The graph shows that there is a strong correlation between poverty and HCI; in Caquetá Sucre Atlántico Bolívar Magdalena other words, the greater the incidence of poverty, the lower the HCI. To put HCI values in per- Caldas Cesar Córdoba spective, it is worth noticing that Bogotá’s HCI is similar to that of China and Chile, while at the -0.05 Valle del Cauca Nariño Cauca La Guajira other end of the spectrum, Chocó’s HCI is close to that of Nicaragua and Nepal. Chocó -0.10 Human capital gaps by ethnic group in large part reflect gaps across geography. The Pacíf- 20% 10% 30% Poverty 40%rate 50% 60% 70% ico region is home to half of the country’s Afro-Colombians, while the large majority of indig- Amazonia Caribe Central Eje Cafetero y Antioquia enous people live in the Amazonia, Pacífico, and Caribe regions. These are the areas of the Pacífico Santanderes Seaflower Llanos - Orinoquía country with the highest levels of poverty and the lowest levels of human capital. (See Chapter Source: World Bank staff analysis. See notes for figure 2.2 for sources for HCI estimates. 5 for a focused discussion on the equity challenges faced by indigenous people and Afro-Co- Poverty rates are from the Departamento Administrativo Nacional de Estadistica (DANE), lombians.) Gran Encuesta Integrada de Hogares (GEIH), 2019. Finally, although Colombia shows a high level of gender equity in the components cap- tured by the HCI, women are less likely to fully use their stock of human capital. The over- FIGURE 2.4. Human Capital: Women versus Men all HCI is higher for women (0.61) than men (0.59). This reflects a slightly higher probability of 0,7 survival to age 5, a higher adult survival rate, and a lower childhood stunting rate among girls. 0,59 0,61 Average learning-adjusted years of education are equal for boys and girls. Yet, women do not 0,6 utilize their stock of human capital in the same way. This can be seen in the Utilization-Ad- 0,5 0,48 justed Human Capital Index (UHCI), which adjusts the HCI for labor market underutilization of human capital, based on the fraction of the working-age population that is employed. Co- 0,4 0,33 lombian women have a much lower UHCI (0.33) than men (0.48), reflecting their lower level of 0,3 labor force participation. 0,2 0,1 0 Human Human Utilization- Utilization- Capital Index, Capital Index, Adjusted Adjusted Men Women HCI, Men HCI, Women Source: World Bank (2020). FIGURE 2.6. Years of Schooling and Learn Departments 24 Education: skills differences Vaupés Vichada 3,6 3,9 6,4 6,5 and inequality in human capital Guainía 4,5 7 B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Human D evelopment and Equity Chocó 5,6 Amazonas 5,7 La Guajira 6,4 Guaviare 6,4 Cauca 6,4 Putumayo 6,8 San Andrés y Prov. 6,8 Valle del Cauca 6,9 Nariño 6,9 Caldas 7,1 Córdoba 7,2 Differences in skills are the most critical source of inequality in human capital. The bulk of FIGURE 2.5. Learning Gaps Across Ethnic Groups Cesar 7,2 Bolivar 7, the variation within Colombia in the HCI is due to disparities in the education subcomponent, Magdalena 7, 11,1 11,0 10,7 Caquetá 7, which captures differences in skills. Specifically, boosting the learning-adjusted years of edu- Sucre 7 cation of the poorest quintile to the level of the richest quintile would close 59 percent of the Atlántico 7 Arauca 7 gap in HCI between those two quintiles. Norte de Santander 7 Antioquia Meta The HCI highlights that learning is not the same as schooling. Traditionally, education out- Casanare Huila comes have been measured by how many years of schooling a child receives. What matters, Tolima however, is not only how long a child sits in a classroom, but how much he or she learns while 7,6 Quindío 6,3 6,2 Boyacá at school. To reflect this point, the HCI combines two measures. The first is the number of years Risaralda Santander of schooling a child is expected to complete by her 18th birthday. The second is a measure of Cundinamarca Bogotá learning based on student achievement tests. These two measures are combined to calculate No ethnic group Afrocolombian Indigenous “learning-adjusted years of schooling.” The adjustment reflects the gap between actual learn- Learning-adjusted Years of Sc Expected Years of Schooling ing and what a child would learn in a high-performing school. Source: World Bank analysis of DANE (2018b) and SABER 11 Test Scores (national stan- Source: World Bank analysis of the Sistema Integrado de M Differences in quality generate substantial gaps in learning by ethnic group and across dardized assessment for 11th graders 2019). (2018a), and Pruebas Saber 9 (2017). space. Figure 2.5 shows both expected years of schooling and learning-adjusted years for Af- ro-Colombians, indigenous people, and those who are not a member of an identified eth- FIGURE 2.6. Years of Schooling and Learning Gaps across nic group. Expected years of schooling—reflecting rates of enrollment up through age 18—do Departments not vary substantially by ethnic group. Learning-adjusted years of schooling are much lower Vaupés 3,6 6,4 Vichada 3,9 6,5 than expected for all three groups, but the gap is larger for Afro-Colombians and indigenous Guainía 4,5 7,6 people. For non-minority children, average learning-adjusted years are 7.6 compared to 6.3 Chocó 5,6 10,3 Amazonas 5,7 10,2 for Afro-Colombians and 6.2 for indigenous people. Figure 2.6 presents a similar breakdown La Guajira 6,4 11,1 Guaviare 6,4 10,6 by department. Early dropout is still prevalent in some regions, resulting in very low expect- Cauca 6,4 10,6 ed years of education (students in Vaupes, Vichada and Guanía, for example, complete less San Andrés Putumayo 6,8 10,7 y Prov. 6,8 11,7 than secondary school on average). When adjusting for quality of education, attainment in Valle del Cauca 6,9 10,3 Nariño 6,9 10,7 all regions falls, with large differences across regions. For example, although students in Mag- Caldas 7,1 10,8 dalena, Sucre, or Cordoba complete on average a similar number of years of education as stu- Córdoba 7,2 12,0 FIGURE 2.5. Learning Gaps Across Ethnic Groups Cesar 7,2 11,6 dents in Bogotá or Risaralda, their learning-adjusted years of education are much lower (7.3 in Bolivar 7,3 12,3 Magdalena 7,3 12,6 11,1 Magdalena compared to 9.0 in Bogotá), signaling the significantly lower quality 11,0 10,7 of education Caquetá 7,4 11,6 Sucre 7,5 12,4 in those regions. Atlántico 7,6 11,7 Arauca 7,6 11,6 Norte de Santander 7,7 11,6 Antioquia 7,7 11,8 Meta 7,8 11,8 Casanare 7,9 11,7 Huila 7,9 11,8 Tolima 8,0 12,5 7,6 Quindío 8,3 12,5 6,3 6,2 Boyacá 8,4 12,2 Risaralda 8,5 12,7 Santander 8,6 12,2 Cundinamarca 8,6 12,3 Bogotá 9,0 12,5 No ethnic group Afrocolombian Indigenous Learning-adjusted Years of Schooling Expected Years of Schooling Source: World Bank analysis of DANE (2018b) and SABER 11 Test Scores (national stan- Source: World Bank analysis of the Sistema Integrado de Matrícula (SIMAT) (2018), DANE dardized assessment for 11th graders 2019). (2018a), and Pruebas Saber 9 (2017). A root cause of the gaps in learning is the differences in preparation children have before reaching primary school. Access to the early childhood development (ECD) services provid- ed by the Instituto Colombiano de Bienestar Familiar (ICBF)/Colombian Institute of Family Welfare is still low. In 2020, ICBF provided ECD services to 36 percent of 0–5-year-old children in Colombia (1.7 million children). Coverage varies widely across departments, as figure 2.7 shows. It is almost twice as high in urban areas than in rural areas (40 percent in the former compared to 21.8 percent in the latter), despite the importance of ICBF in rural regions be- cause of the lower supply of private services (92 percent of children who receive services in rural areas do so from ICBF). Those lacking access tend to be poorer, as roughly 60 percent of the families that declared not having access to ICBF ECD services in the Encuesta Nacional de Calidad de Vida (ENCV) (National Quality of Life Survey) in 2018 were poor or extremely poor and 30 percent were vulnerable (DANE 2018b). FIGURE 2.7. Percentage of Children Aged 0–5 from Strata 1, 2, or 3 Enrolled in ICBF Services by Department 90% 79% 80% 70% 67% 63% 59% 60% 52% 50% 47% 46% 45% 44% 43% 42% 42% 41% 41% 41% 41% 39% 37% 40% 35% 35% 31% 31% 30% 28% 28% 27% 27% 26% 26% 25% 25% 23% 20% 20% 20% 10% 0% Bo o tu a da o na Ar s az a At nas An tico ia e Bo la Ca ca a de ima ca nt s Bo a Qu ta Ca dio nt e a Cu Vic ta in da ca An o Gu res Gu e Co ira Na a Ca r r Ri der Gu er a sa a Sa pe cr Sa nar La iar Pu uc Am auc et ld ni ob riñ M may oc qu i go e d liv ld ya au ar Hu le nd ha aj d Ce ai Su ra qu M in an an o de au Ch lle ol n av rd Ca am tio sa lC la sa T V ag n Sa Va rte No Source: Encuesta Nacional de Calidad de Vida 2018. The supply of ECD services provided by ICBF is insufficient to meet existing needs and it is limited by informal entry barriers. Although there is a broad portfolio of programs imple- mented by ICBF, qualitative work undertaken by the World Bank has shown that many families face entry barriers beyond physical access that include excessive documentation and other re- quirements or a lack of clarity on how to access the services. This severely and unnecessarily limits access for populations in rural areas or in poverty. The quality of the programs included in the portfolio also varies, and there is little reliable information about the service quality of the different programs and providers. Moreover, providers do not have adequate incentives to provide quality services. Although ICBF relies on private providers to manage most of its programs, rigorous measures of service quality are not available and are not a determinant in securing a provider contract. Repetition and early dropout are still too high in rural areas, among the poorest segments of the population, and among ethnic minorities. Colombia has the second largest repeti- tion rate among all the countries that participate in the Programme for International Student Assessment (PISA) exam: 41 percent of 15-year-old students have repeated at least one grade. This is due principally to low levels of learning in the early grades, which lead to repetition and early dropout. Shockingly, 49 percent of 10-year-old students cannot read a simple text and answer basic questions about that text correctly (World Bank 2019b). These poor reading outcomes are a severe constraint on students’ ability to absorb the rest of the curriculum. Poor learning outcomes also contribute to inequalities in access to post-basic education. Enrollment falls rapidly after age 16 for rural students and students from poor families. Al- though three-fourths of students in rural areas are still enrolled in formal schooling at age 16, that percentage falls by half by age 18 (38 percent) and another half by age 20 (17 percent). The decline is similar for the poorest quintile. Importantly, too many youth still leave schooling without completing upper secondary education. In rural areas, 45 percent of 18–22-year-old youth leave schooling without completing upper secondary education (only 27 percent in ur- ban areas). The differences by ethnicity are also salient: 45 percent of indigenous youth and 35 percent of Afro-descendants leave education without completing upper secondary school. Although poor learning outcomes are at the core of these differences, they are also driven by other factors. For example, limited physical access to an upper secondary school remains problematic in rural areas (Garcia et al. 2016; Garcia, Maldonado, and Jaramillo 2016). Finan- cial constraints also limit access to tertiary education for disadvantaged groups, since private provision accounts for half of enrollment at that level. Completing upper secondary education FIGURE 2.8. Percentage of 15-Year-Old Students in Private and especially tertiary education greatly increases income and reduces the probability of fall- School By Decile ing into poverty or vulnerability (OECD and World Bank 2012). 70% Differences in the quality of education are in part a result of large inequity in private school enrollment. Among countries participating in the PISA exam, Colombia has one of 43% the highest rates of enrollment in private schools not funded by the government (17 percent).3 28% Seventy percent of 15-year-old students in the top socioeconomic quintile are enrolled in pri- 23% 18% vate school, compared to just 2 percent of those in the bottom quintile (see figure 2.8). Private 8% 10% school students perform far better on standardized tests, reflecting a combination of advan- 2% 3% 3% tages in student background and school quality. The difference between average reading 1 2 3 4 5 6 7 8 9 10 scores of students in public and private schools was 83.5 points in the 2018 PISA Assessment Decile of Socioeconomic Index (OECD 2020; ICFES 2020). This is one of the largest public-private differences among all PISA countries and is equivalent to the overall difference between Colombia and France in PISA Source: World Bank staff analysis of 2018 PISA data (OECD 2020). Note: Deciles shown were calculated using PISA’s Economic, Social, and Cultural Status reading scores. (On average Colombian private school students score just below the OECD Index. average for all students.) Although there are multiple reasons for the low learning outcomes, curricular autonomy at the school level, coupled with low teacher quality and disparities in teacher alloca- tion, are the main drivers of inequalities. Colombia lacks any form of national curriculum and relies on each school to develop its own. Although the Ministry of Education regularly produces materials and guides and sometimes sends textbooks to schools that effectively become the curriculum, the curricular autonomy at the school level places the burden of cur- ricular development on individual teachers and school managers (OECD 2018b). This is not only inefficient but also requires teachers with skills in curriculum development, which most teachers lack since curriculum development is a specialized expertise that is not a core part of pedagogical programs. In addition, the lack of a curriculum complicates the ability to provide systematic support at a large scale for these teachers and to develop concrete, competen- cy-based, teacher trainings programs. Colombia’s teachers are also generally low skilled and unequally distributed. Students who enter pedagogical programs in tertiary education have significantly lower exit exam scores in upper secondary education4 than the average graduate. Their SABER PRO5 results, taken after graduation from tertiary education, are also among the lowest scores among grad- uates (Forero and Saavedra 2019; García et al. 2014). The teacher selection process is compet- itive, and teachers with the highest scores choose their posts. This results in the best teachers selecting the best schools and leaving the most vulnerable schools with poor quality teachers or unable to fill posts with permanent teachers and resorting to temporary contract teach- ers, exacerbating existing inequalities in the quality of education (Forero and Saavedra 2019). Since teachers are responsible for curriculum development, this results in extremely poor FIGURE 2.9. The Two Main Sources of Spending on Quality pedagogical practices in schools with low-quality teachers—the very ones who serve more vulnerable populations. Although there is a need for some curricular flexibility to adapt to Per student psending (COP 2019) 700 000 the different contexts and populations in Colombia, the introduction of a core curriculum for 600 000 basic competencies would improve equity in pedagogical practices, facilitate the implemen- 500 000 tation of pedagogical support and teacher training at scale, and allow for the implementation of successful interventions in low-capacity settings such as scripted lessons (Piper et al. 2018). 400 000 300 000 The financing system provides insufficient investment in the quality of education and 200 000 does not compensate for the large disparities in the existing quality of inputs and teach- 100 000 ers. Figure 2.9 shows the per student spending on quality by the level of learning poverty in the local government. The graph shows the two main sources of quality spending: quality - Lowest Medium Highest transfers included in the Sistema General de Participaciones and those from the local govern- learning poverty learning poverty learning poverty ment’s own resources. It shows that though ETC (Entidades Territoriales Certificadas en Edu- Quality transfer per student Own resource spending per student cación) with high levels of learning poverty do receive marginally larger transfers for quality, the differences in own resources far outweigh the transfers. This results in ETC with high levels Source: Ministry of Finance (accessed at https://www.datos.gov.co/). of learning poverty having fewer resources to improve the quality of service delivery. 25 The health care system BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM B I A | Human D evelopment and Equity Nearly all Colombians today have health insurance. A vast expansion of both the contribu- FIGURE 2.10. Health Insurance Coverage in Colombia tory and subsidized health insurance programs over the last quarter century has meant that 100% nearly all have access to health care. Today, the population is roughly evenly split between the 90% contributory and subsidized programs. 80% Despite broad health access, differences in health outcomes across geography and eth- 70% 60% nic groups are marked. Gaps across space are most vividly captured by differences in life 50% expectancy at birth across departments. These gaps are shown in figure 2.11. “Lost years 40% of potential life” are defined as the difference between life expectancy in each department 30% and Bogotá, which has the longest life expectancy in the country (78.9 years). A Colombian in 20% Caquetá, Chocó, or Casanare will live on average more than eight years less than a Bogotano. 10% The fraction of indigenous children who are stunted—a marker of malnutrition—is almost 0% triple the national average. 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 One factor driving inequity in outcomes across space is the large differences in health Subsidized coverage Contributory coverage Other care personnel. Bogotá surpasses the World Health Organization’s (WHO) recommendation of 4.45 physicians, nurses, and midwives per 1,000 population, but poorer departments like Source: Ministerio de Salud, Serie Departamental Afiliados Contributivo 2005–2016,” Chocó, Vaupés, Caquetá, and La Guajira lag far behind that goal with 1.6 or fewer health work- available at https://www.minsalud.gov.co/sites/rid/Lists/BibliotecaDigital/RIDE/VP/ DOA/serie-departamental-afiliados-contributivo-2005-2016.pdf. ers per 1,000 people (see figure 2.12 below). FIGURE 2.11. Lost Years of Potential Life by Department, FIGURE 2.12. . Combined Number of Physicians and Relative to Bogotá (2015–2020) Nurses per 1,000 Population by Department Caquetá 8,4 Bogota 6,2 Chocó 8,2 Atlantico 4,3 Casanare 8,1 Risaralda 4,2 Grupo Amazonia 7,8 Santander 4,0 Arauca 7,4 Caldas 3,4 Meta 6,5 Valle del Cauca 3,4 Cauca 6,1 Cesar 3,3 Putumayo 5,9 Quindio 3,2 Norete de Santander 5,6 Huila 3,1 Nariño 5,4 Antioquia 2,9 Huila 5,0 Amazonas 2,8 Córdoba 4,7 Casanare 2,7 Tolima 4,7 Bolivar 2,6 Bolívar 4,4 Meta 2,6 Cesar 4,4 Boyaca 2,5 Sucre 4,3 Sucre 2,4 La Guajira Magdalena 2,3 4,2 Norte de Santander 2,3 Cundinamarca 4,1 Narino 2,3 San Andrés y Providencia 4,0 Tolima 2,3 Caldas 3,3 Arauca 2,2 Quindio 3,3 Cauca 2,1 Risaralda 3,2 Cundinamarca 1,9 Atlántico 3,0 Cordoba 1,9 Santander 2,8 Guaviare 1,9 Boyacá 2,7 Putumayo 1,7 Magdalena 2,6 La Guajira 1,6 Antioquia 2,6 Caqueta 1,5 Valle del Cauca 2,2 Vaupes 1,3 Bogotá 0,0 Chocó 1,2 Source: Projections from DANE, “Proyecciones de Población 2005–2020” (Bogotá: Depar- Source: Ministerio de Salud, “Observatorio Nacional de Calidad en Salud,” http://oncali- tamento Nacional de Estadística, 2007), https://www.dane.gov.co/files/investigacio- dadsalud.minsalud.gov.co/Paginas/Inicio.aspx. nes/poblacion/proyepobla06_20/8Tablasvida1985_2020.pdf. Box 2.1. The Colombian Health Care System Colombia is an upper-middle-income country that has continuously invested in the de- velopment and outreach of its health system, currently reaching more than 96 percent of its resident population with a comprehensive health insurance program. The Colombian General System for Social Security in Health (GSSSH) is divided into two interconnected subsystems: (i) the Contributory Regime, involving those who have the capacity to con- tribute a percentage of their income and who do so in solidarity with the financing of the system; and (ii) the Subsidized Regime, involving people from the informal sector or with income below the poverty threshold who do not pay contributions. The insured, regardless of their income, age, sex or geographic location within the coun- try, are entitled to almost all the health services they will require, with the exception of technologies without proven efficacy and cosmetic surgeries. Under the stewardship of the Ministry of Health and Social Protection and the Health Superintendency, private or public health insurers receive a per capita payment for each person affiliated and must provide services to their insured population for all levels of care through a network of public and/or private health care providers. The system was designed with the goal of promoting the equitable distribution of health services among the population and en- couraging the provision of quality services through market competition among insurers and among providers. Market failures stemming from financing, payment, and gover- nance challenges have made the attainment of these goals a continued reason to strive for improvements. Inequalities in access to care remain. Figure 2.13 shows a comparison of the waiting times FIGURE 2.13. Inequalities in Access to Health Care for general medicine specialties between rural and urban areas. Rural residents have to wait 45 twice as long as urban dwellers for general appointments and five times as long—more than a month on average—to see a pediatrician. 40 38,7 35 The poor health outcomes for the most vulnerable Colombians largely reflect the low quality 30 of PHC. Differences in inputs for and access to care are compounded by examples of inequities in 25 the quality of basic services received by the population. Less wealthy Colombians depend largely 20,7 on public health facilities for their care. Figure 2.14 shows health care facility usage for antenatal 20 16,3 14,1 care and childbirth. The large majority of wealthier Colombians access care at private facilities 15 and health centers. For those in the poorest wealth quintile, 85 percent went to public facilities 10 8,0 7,4 6,4 for antenatal care. Seventy-four percent of childbirths among the poor take place at government 5 2,7 facilities, and an additional 10 percent take place at home. Particularly in rural areas, government 0 facilities are typically the only option. This highlights the importance of improving the quality of General Gynecologist General surgery Pediatrics physician PHC in public facilities to address health inequities. Waiting times days (2016) Rural Urban Source: Ministerio de Salud, “Observatorio Nacional de Calidad en Salud,” http://oncali- dadsalud.minsalud.gov.co/Paginas/Inicio.aspx. FIGURE 2.14. Health Facility Usage by Wealth Quintile Place of antenatal care Place of child delivery 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Poorest Poorer Middle Richer Richest Poorest Poorer Middle Richer Richest Wealth Quintile Wealth Quintile Private hospital/clinic EPS health center Government hospital/center/health post Other Home Source: World Bank analysis of Ministerio de Salud (2015). Note: Place of antenatal care is as reported by pregnant women for a current pregnancy, excluding those who do not report any antenatal care. Place of child delivery is for the most recent birth. A recent comprehensive World Bank assessment of PHC in Colombia identified gaps in performance and opportunities for improvement. The results show in a comparative way that the PHC system in the country has achieved high levels of access, but barriers still persist in the quality of services, which affect the effective coverage provided by the system (World Bank 2020b). Key findings from the study are as follows: • Barriers to access to health services related to distance and cost perceived by the Co- lombian population have steadily decreased since 2010. Barriers to perceived quality of services have remained relatively constant. • The level of comprehensiveness of the services provided by health care providers in Co- lombia has also increased steadily, although the supply of services for chronic diseases has remained below other health services and the burden of chronic diseases has in- creased. • Although financial and distance barriers have been reduced and the availability of ser- vices from the country’s health care providers has risen, adherence to clinical practice guidelines has not increased in recent years. • The gains in achieving effective coverage of health services in recent years have not been equitably distributed throughout the territory of the country. 26 The social protection system: inequalities in opportunities for resilience BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | Human Development and Equity A third key factor that explains inequality in human capital is the limited impact of social assistance on poverty reduction and households’ generally low resilience to shocks. The social protection system has a dual role: as a provider of social assistance to the poor and vul- nerable and as a safety net to protect all citizens from present and future poverty. Colombian social assistance programs have been successful in providing income support to those in need while reducing the incidence of monetary poverty in recent years by roughly 5 percentage points, from 40.8 percent in 2012 to 35.7 percent in 2019 (DANE, Encuesta de Calidad de Vida 2019) 6. Although flagship programs such as Familias en Acción, Jóvenes en Acción, and Co- lombia Mayor deliver cash transfers—conditional and unconditional—to 4.65 million house- holds, they still face significant challenges in terms of targeting and implementation and their contribution to household resilience is limited. (The incidence of these programs is discussed further in Chapter 4.) FIGURE 2.15. Distribution of Beneficiaries of Flagship Social The lack of dynamism and accuracy of the information on eligible households translates Assistance Programs by Income Decile into persistent inclusion and exclusion errors in the targeting of social programs. Since 25% 1995, Colombia uses the Sistema de Identificación de Potenciales Beneficiarios de Programas Sociales (Sisbén)7 as the main tool for targeting 21 social programs at the national level. As of 20% March 2020, the most recent census sweep of Sisbén—Sisbén IV—contained information on Share of beneficiaires more than 39.4 million people, equivalent to approximately 78 percent of Colombia’s popu- 15% lation. As Sisbén IV information on households was collected between 2017 and 2019 and is updated only on direct request of a household, it tends to lose accuracy over time. This lack of 10% dynamism is not new to Sisbén IV; in the previous sweep, Sisbén III, the number of non-poor households classified as poor or vulnerable (inclusion error) increased 7 percent for monetary 5% poverty and 12 percent for multidimensional poverty (according to the Multidimensional Pov- erty Index [MPI]), and the number of poor households not identified as poor (exclusion errors) 0% grew 44 percent for monetary poverty and 3 percent for multidimensional poverty (CONPES 1 2 3 4 5 6 7 8 9 10 2016). As consequence, social programs are not able to target effectively and usually have Income decile beneficiaries in all income deciles (figure 2.15). Even with the new Sisbén IV, inclusion errors Jóvenes en Acción Colombia Mayor Familias en Acción in social programs range from 2.1 percent in Familias en Acción to 9.6 percent in Jóvenes en Acción (CONPES 2021). Source: World Bank, based on data from DANE’s Encuesta Nacional de Presupuestos de los Hogares (ENPH) 2016–17. COVID-19 made evident the limited resilience of households against shocks. The pan- demic halted the growth of the middle class and increased overall poverty, reflecting the extreme vulnerability of households to shocks. The Departamento Nacional de Planeación (DNP) (the National Planning Department) estimates an increase in poverty of 6.9 percentages points due to the pandemic, which caused a contraction of both the vulnerable (-6.7 percent) and middle classes (-12.8 percent) (CONPES 2021). Although the creation of the unconditional cash transfer Ingreso Solidario as a response to the pandemic was useful to temporarily extend social assistance to the poor and vulnerable households that were not traditionally covered by flagship programs, under normal circumstances, the access of these groups to social assis- tance is still insufficient, and even the middle class remains highly vulnerable to the crisis. As an example, estimates for 2017 showed that 55 percent of Colombians were not able to cope with the negative effects of natural disasters on their assets, which caused a decrease in house- hold consumption (well-being) of around 3 percent of GDP every year.8  (Further discussion of this issue can be found in Chapter 6.) The low resilience to shocks, such as the one revealed by the COVID-19 pandemic, indicates how the current structure of the Colombian social pro- tection system is not prepared to help the poor and the non-poor alike to manage risks and cope with crises and shocks, making any victory in terms of poverty reduction highly unstable in the long term. 27 Impact of the COVID-19 pandemic on human capital The COVID-19 pandemic has likely had disproportionate health impacts on the poor and BU I L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | H uman Development and Equity ethnic minorities. Studies focused on Bogotá have shown that those living in the poorest two (of six) socioeconomic strata were four times as likely to be infected with COVID as those in the top two strata, and that “falling ill with a serious case of COVID has been over eight times more likely for an individual in the lowest stratum, where the poorer population concentrates, compared to one in the highest.”9 For the country as a whole, the case fatality rate—the fraction of recognized COVID-19 cases that results in death—is 3.2 percent for indigenous Colombians and 3.0 percent for Afro-Colombians compared to 2.6 percent for those who do not identify as a member of an ethnic group. This could reflect more limited access to high-quality health care among Afro-Colombians and indigenous groups.10 However, it is important to recognize that these conclusions are uncertain because only a fraction of COVID-19 infections are recognized FIGURE 2.16. Share of Students at or below the Minimum Reading Level in Grade 5 (Saber 5 – 2017) as cases, and some deaths due to the virus are undercounted. Further research is needed to better understand how the pandemic’s health impacts vary across socioeconomic groups. 80% 70% 72% 70% 65% 68% The pandemic has also affected human capital through indirect channels. These include 63% 60% 60% 1) diminished income as a result of pandemic-related restrictions, which has reduced resourc- 60% 57% 53% es available for human capital investments, 2) drops in non-COVID-19 medical care, including 50% routine services like basic vaccinations and prenatal visits, and 3) the decline in learning as 40% a result of school closures. In all of these areas, hard data are sparse. The discussion here fo- cuses on the third channel—school closures—because this effect has very likely widened the 30% existing gap in learning between rich and poor and will have long-term repercussions. 20% School closures in response to the pandemic have driven a large increase in learning pov- 10% erty, particularly for poorer and rural students. Even before the pandemic, 53 percent of 0% 10-year-olds were “learning poor,” meaning that they could read and understand only a basic National Poorest quintile Rural text. The pandemic is expected to increase the rate of learning poverty to 60 percent if schools Share of students at or below minimum level can maintain a hybrid program11 through 2021 but could reach 63 percent if distance learning Pre COVID-19 continues for the year. With distance learning alone, the learning poverty rate could reach 70 With Alternancia in 2021 With distance education in 2021 percent for those in the poorest quintile and 72 percent for rural (Figure 2.16). The impact of decreased learning is estimated to be US$8,500–$20,000 (2018 purchasing power parity) over Source: World Bank calculations using Pruebas SABER 5, ICFES (2017). the lifetime of the average student, representing 10 and 25 percent of pre-COVID levels. 2.3. 28 BUIL D ING AN EQ UITABL E S O CIETY IN CO LO M BIA | Human D evelopment and E quity Policy Options Ensuring equitable opportunities in human capital is now more urgent than ever. The COVID-19 crisis has exacerbated pre-existing inequities in outcomes and also in opportunities to build the kind of human capital needed to pursue a useful and meaningful life. Without an ambitious strategy to reduce these inequalities, they will continue to increase. More impor- tantly, success in the recovery from the crisis will depend on how effectively and equitably human capital is developed, as the recovery will need healthy and productive citizens who can create or take up new jobs in new economic activities. This chapter has identified three fundamental sources of inequities in human capital formation: (i) inequality in access to good quality PHC services, (ii) the large disparity in the quality of education and learning outcomes, which begins in early childhood, grows in the early grades. and is the main cause of incomplete education trajectories; and (iii) inadequate access to social protection programs. This section proposes some specific actions to address the sources of inequity to ensure that all citizens are healthy, skilled, and resilient. HEALTHY CITIZENS SKILLED CITIZENS RESILIENT CITIZENS Primary Health Care Learning Unified Social Registry HEALTHY CITIZENS SKILLED CITIZENS RESILIENT CITIZENS Healthy citizens: high-quality primary health care for all Given the already very high levels of health insurance coverage, the key to improving equity in health outcomes in Colombia is improving the quality of PHC, particularly in Primary Health Care government Learning facilities where most at the lower end of the economic spectrum get their Unified Social Registry care. The World Bank PHC study identified three key approaches to improving primary care: reducing barriers to quality PHC at the local level, introducing incentives for better quality and health outcomes in payment systems, and empowering citizens and local governance (World Bank 2020b). Ensure equitable access to multidisciplinary care at the local level. Better models of care at the local level make care more accessible to persons and communities through a reduction in the barriers to access and through proactive outreach interventions for patients who are normally underserved owing to distance, social perceptions, cost, or socioeconomic disad- vantage. The Ministry of Health and Social Protection is actively engaged in working with local health departments to promote the implementation of new care models meeting these re- quirements, and such efforts should be encouraged and strengthened. In addition, care mod- els should be adapted to reduce cultural barriers to access for specific populations, such as Afro-descendants and indigenous peoples. At the local level, efforts should also be made by local health departments to reduce barriers to care and proactively reach out to the most vulnerable with the participation and engagement of patients and citizens in the system’s planning and accountability processes. One example includes providing clear information on the rules of access to care and the scope of PHC ser- vices, which helps patients from different social and economic backgrounds to access and use PHC services in a more timely and appropriate manner. This could include developing specific strategies for outreach and accountability for low-income populations, indigenous peoples, or Afro-descendants. In Guatemala, for example, successful strategies were implemented for health advocates, volunteers, and representatives of indigenous communities who carry out accountability and mediation activities between health care providers and patients. Final- ly, mention should be made of the positive experiences Bogotá, Medellín, Barranquilla, and several municipalities in Caldas, among others, where gaps in access and health outcomes between populations from higher and lower socioeconomic levels were reduced through pro- active outreach strategies and better coordination mechanisms among health care providers. Improve payment systems to ensure quality and better health outcomes for the most vulnerable. The design of better incentives and innovation in payment systems is one of the stewardship responsibilities of the Ministry of Health and Social Protection and should incen- tivize the implementation of a PHC-centered model of care with a focus on care integration and coordination for targeted high-burden diseases, such as cancer and cardiovascular diseases. Health insurers should promote better alignment of personalized care plans, adherence to in- tegrated health and clinical practice guidelines, integrated PHC plans, and care networks. The integration processes must take into account the delegation of the service provision function for the population from the public sector to the private sector, which entails the development of integration mechanisms through the formulation of national policies and care guidelines that are based on health outcomes. To achieve this, one option would be to broaden the scope of the government High-Cost Account Program to other chronic diseases, such as chronic obstructive pulmonary disease (COPD), mental illness, other rheumatic diseases, and comorbidities, and index episodic dis- eases (for example, myocardial infarction, stroke, or appendicitis). It would also be useful to increase the emphasis on results-based accountability rather than only on compliance with care procedures. The incentives must also encourage health insurers and care providers to scale up self-management support approaches for patients with chronic diseases. Current payment methods do not account properly for the specific risk profiles of citizens, providing little incentive to health care providers to offer differentiated care to patients with different risk factors. Incentivizing health insurers and providers to produce better patient results and to reduce inequities in health outcomes in turn creates incentives for providers to develop innovative models of integrated care. Paying for care bundles, including for complex popu- lations, should be tested to encourage a holistic approach to care integration and distribute risks more fairly across insurers and providers.  Promote citizen empowerment and local governance for better health outcomes for all. Given the diversity of contexts and needs in Colombia, it is important that local health author- ities strengthen community participation mechanisms in the definition of local health care services. Community participation in design, planning, governance, and service provision is essential so that these activities can adequately meet the needs and expectations of the persons for whom they were designed. Community participation can be implemented with a greater integration of members of the community on the boards of directors of health facilities or in the setting up of patient and family advisory councils; community meetings; feedback systems that value the voices, opinions, and experiences of users. Strong local governance promotes local prioritization of PHC needs based on population health, which can be defined at the departmental, municipal, or community level. Coordination between these various lev- els seeks primarily to reduce the variability in the quality of and access to the services avail- able to the population and ensures that they meet the needs of the most vulnerable. HEALTHY CITIZENS SKILLED CITIZENS RESILIENT CITIZENS Skilled citizens: accelerating learning and ensuring complete education trajectories Reduce inequality in learning outcomes, which begins by reducing inequalities in child de- Primary Health Care Learning Unified Social Registry before formal schooling begins and minimizing inequalities in basic competen- velopment cies in the early grades. This requires increasing equity in access to good quality ECD services, strengthening basic competencies in early grades, and implementing programs to teach at the level of the students’ skills, such as tutoring. In addition, it needs complementary reforms in up- per secondary and tertiary education to facilitate complete education trajectories. Improve equity in access to good quality ECD services. First, the easiest and most cost-ef- fective way to increase equity is by reducing the administrative requirements and streamlin- ing the application procedures for ICBF programs. This implies: (1) reducing the number of documents required for pre-registry, registry, and monitoring by verifying information directly with other public agencies; and (2) integrating the ICBF’s information systems to ensure that families with multiple programs do not have to provide the same documents every year and for each and every program individually. Second, the supply of ICBF services also needs to increase, especially in rural areas. Third, reducing inequities in the quality of services requires better measurement of the quality of services, which right now are not sufficiently rigorous and are mostly self-reported by providers. These service quality measures should be included in the criteria for selecting providers and in their multiyear contracts, which should incorpo- rate performance incentives linked to quality of service provision. Lastly, when planning for the expansion of services, there is a need to evaluate the quality of the portfolio of programs currently implemented by ICBF, especially those provided outside of Centros de Desarrollo Infantil (Child Development Centers). Implement a core and prioritized curriculum for accelerated learning in basic competen- cies (Shah and Choo 2020). Even if the introduction of a national curriculum is difficult due to the long history of school-level curricular autonomy, the COVID-19 crisis and the need to ac- celerate learning after school closures is an opportunity to introduce prioritized core curricu- lum guidelines for basic competencies, leaving the rest of the curriculum flexible. A prioritized curriculum focuses efforts on improving basic competencies, especially language and math, and has the added benefit of facilitating coordination across interventions to support learn- ing. Many countries are planning to implement a version of a prioritized curriculum to recover learning losses from the COVID crisis.12 Chile and Vietnam are examples of this prioritized cur- riculum approach. Provide pedagogical support for teachers linked to a core curriculum in basic competen- cies by institutionalizing and expanding the Programa Todos a Aprender (PTA). Though initially ineffective and costly (Barrera-Osorio et al. 2018), the PTA was reformulated in 2016 and has proven to be successful at improving learning outcomes in rural and low-perform- ing schools through an increased focus on basic competencies in basic education.13 The pro- gram provides tutoring for teachers on pedagogical content, classroom management, and socio-emotional skills and reached over 2 million students in 2020. It covers only basic edu- cation, focusing on fundamental competencies, and was a key tool in the response to school closings by supporting teachers in distance education. The program should be evaluated, in- cluding its pedagogical aspects, but given that the model has shown positive impacts, it could be expanded to include secondary education with a more cost-effective implementation mod- el and could leverage teachers supported by the program to generate teacher networks and share good practices. Accelerate learning equitably by “teaching at the right level” through tutoring programs, the grouping of students, and the use of computer-assisted learning (CAL). Among the successful interventions to teach at the right level (Banerjee et al. 2016) are different mod- els of tutoring (Fryer and Howard 2017), the grouping of students by level, and the use of CAL (Muralidharan, Singh, and Ganimian 2017). All of these interventions will need support for teachers and greater coordination with parents and caregivers so that learning at home complements learning at school. One good example of this kind of initiative already proven to work in Colombia is the Aprendamos Todos a Leer program, initially designed and imple- mented in Manizales by the Luker Foundation (Álvarez Marinelli, Berlinski, and Busso 2019), which has also been adapted to remote learning in very low-capacity settings through written materials and phone calls, as well as the general use of WhatsApp and short message services (texting). However, CAL can only complement these strategies in areas with adequate access to computers and the internet (Baron, Taveras, and Cuevas Zuñiga 2018). Introduce upper secondary reform. Ensuring complete trajectories also requires important changes to upper secondary education. Upper secondary should be the bridge between basic education and tertiary education or the labor market at a time when the lives and brains of stu- dents are changing dramatically. This demands a new pedagogical approach, based on build- ing the competencies students need for the labor market or for tertiary education (EC 2019). In the short term, the learning losses and economic needs resulting from the COVID-19 crisis will push many more students out of the schooling system if no remedial action is taken. Develop- ing early warning systems to identify students on the brink of dropping out is thus crucial in order to intervene before that happens. Programs that combine tutoring with socio-emotional development, such as PODER in Mexico (Avitabile et al. 2019) and the original version of the PODER program, “Becoming a Man” in Chicago (Heller et al. 2015), have proven to significantly reduce drop out rates for high-risk students and those in violent environments.14 Ensure more equitable financing of public tertiary education institutions. The financing system for tertiary education, anchored in Law 30 of 1992, is very unequal. About 45 percent of total transfers from the central government are allocated to three universities that account for 20 percent of enrollment. The regional public universities that are more likely to serve disadvantaged students rely mostly on financing from local governments, which is uneven, and are often forced to charge higher tuition fees. Importantly, almost all financing is based on historical trends or ad hoc allocations (i.e., it is not formula based) and is not associated in any way to performance or results indicators (including enrollment). Financing should move toward a per student allocation of service provision transfers that reflects the operational cost of providing a good quality tertiary education and also equity concerns (i.e., providing more resources for more disadvantaged students who may need more support). In addition, any re- sources that are not covering operational costs should be pooled to avoid fragmentation and linked to results and institutional improvement plans. Institute income-contingent loans. Access to tertiary education needs to be more equitable and flexible while reducing the vast inequalities that exist in quality and relevance. Colombia relies on private provision for almost half of its tertiary education enrollment and has a robust student loan system in the hands of ICETEX. However, the loan conditions, and the resulting high repayment burden (the share of income graduates devote to repaying the loan), severely limit the demand for student loans. The implementation of income-contingent loans, in which students repay the cost of tertiary education after graduation with a progressive contribution that depends on their income, can be effective at removing remaining access barriers for ter- tiary education in an equitable and sustainable manner.15 TIZENS SKILLED CITIZENS RESILIENT CITIZENS Resilient citizens: an integrated and dynamic social registry to respond effectively to shocks The Colombian social protection system could do more to promote household resilience. h Care Learning Unified Social Registry In the context of a growing vulnerable middle class, it is essential to define an institutional and policy framework for social protection that not only focuses on providing continuous support to the poor, but is also capable of protecting the non-poor from present and future poverty. The significant reduction in poverty rates experienced by Colombia in the past decade con- stitutes a clear success of government investments in key social assistance programs, such as Familias en Acción and Jóvenes en Acción. However, many of the recently non-poor households are still highly exposed to the effects of both expected and unexpected shocks to their livelihoods that make them susceptible to falling back into poverty. The low resilience to shocks, such as the one introduced by the COVID-19 pandemic, reflects how the current structure of the Colombian social protection system is not prepared to help the poor and the non-poor manage risks and cope with crises and shocks, making any victory in terms of poverty reduction highly unstable in the long term. Chapter 6 on climate change and equity provides further insights into the challenges faced by Colombia in creating a more adaptive social protection system. The consolidation of a dynamic, reliable, and integrated social registry equipped with specific tools for constantly assessing social risks is the first step toward making citizens more resilient. The capacity of the social protection system to rapidly manage social risks and effectively support households after a shock depends almost entirely on the maturity and uptake of the social registry. On the one hand, the establishment of an integrated social registry that includes both direct program beneficiaries and non-beneficiaries is key to ensur- ing that the government is able to support households that are more exposed to risks by, for example, implementing emergency cash transfers to mitigate the impact of shocks on house- hold income and assets. On the other hand, the dynamism and reliability of the information in the social registry is essential to both defining effective social insurance schemes to protect citizens from shocks and creating more integrated and better-targeted social protection inter- ventions that ensure a successful and sustainable poverty reduction. Although the new Sis- bén IV constitutes a step forward in the consolidation of the social registry, it was not enough to support a rapid government reaction to the recent pandemic. The absence of reliable and updated data on the non-beneficiaries of the traditional social programs meant that it was extremely hard to extend the cash transfer programs horizontally to include the vulnerable middle class during the crisis. Ensuring equitable access of citizens to social programs is essential to unraveling the im- pacts of the social protection system on equity. Over the past three decades, Colombia has successfully created a comprehensive and robust social program portfolio with mature institu- tional arrangements and steady financing and coverage. However, many social programs are still facing significant implementation challenges that tend to create access and quality gaps, both within the beneficiaries and between beneficiaries and non-beneficiaries. In most cases, access gaps tend to disproportionally affect poor rural/ethnic households either by creating barriers to the effective access to services or by not having an implementation arrangement consistent with the demanding contexts in which these families are located. Finally, Colombia’s social protection system needs to promote an integrated and citi- zen-centered social program portfolio for building resilience. User experience in the system should be put at the center of the discussion on how to improve services and benefits in the short term. The fragmentation and disconnection of the many social programs and benefits usually translate into an excessive administrative burden for citizens. Some estimates from DANE’s 2017 National Survey on the Use of Time (Encuesta Nacional de Uso del Tiempo [ENUT]) indicate that a citizen spends an average of 1.5 hours every time he or she needs to collect a subsidy or a transfer, with at least 25 percent spending more than two hours and 10 percent more than four. Women tend to experience this burden more heavily, as 25 percent of them spend more than 2.5 hours collecting subsidies or transfers. Creating a citizen-centered social protection system requires both a more effective deliv- ery of social programs and an integrated approach to what citizens need to exit poverty and build resilience. Closing gaps in equity requires that social programs provide services to citizens in a coordinated way so that they can escape poverty and are prepared to face shocks. Initiatives such as the Ruta para la Superación de la Pobreza (the Route to Overcoming Pov- erty) at the Departamento para la Prosperidad Social (DPS) (Department of Social Prosperity) and the preliminary attempts at creating guaranteed minimum income schemes in Bogotá are consistent with the idea of providing more integrated social services. Moving from different, uncoordinated programs acting on the same households to reduce poverty to well-organized joint intervention plans, consistent with the expectations and characteristics of citizens, is the next step in the social protection system’s ability to achieve greater and sustainable impacts on poverty and equity. Cross-cutting areas: improving the quality of service delivery The challenges are likely to demand more resources, but improving the management and fi- nancing of education, health, and social protection programs is the immediate priority. Al- though the specific challenges in each sector are different, there are commonalities in the priorities across the human capital sectors. First, there is a need to measure the quality of ser- (i) vice provision and establish standards for education, health, and social programs. Second, in a highly decentralized system, this information should be used at all levels of government to Measuring the quality of service provision monitor service quality and to target programs and support more effectively to improve qual- and setting quality standards ity through better management practices. Third, adequate and equitable financing should be Improving quality requires effort from service providers, but there is currently very little provided that also incentivizes the quality of service provision. Lastly, there is a need for im- incentive on the part of education and health providers to make any changes. Outcomes proved coordination across sectors by linking information systems and better integrating ECD are generally measured and reported, but service provision is rarely evaluated. In education, services with education services and better school health programs. for example, Colombia has systematically collected learning outcome measurements since 199016 in multiple grades,17 producing excellent reports for schools. These assessments were halted in 2018. It is more important than ever that these assessments take place regularly, even if it is in a nationally and regionally representative sample of schools and students, in order to target efforts and monitor the results.18 There is a need to systematically measure inputs, processes, and user perceptions about the quality of services. In education, the COVID-19 crisis has shown that education manage- ment and information systems are too narrow, outdated, and rigid. They focus on measur- ing inputs and collect information only from the principal. The technology exists to collect and use information more systematically at a relatively low cost, making it possible to target interventions more quickly and effectively and to monitor progress more often, which can result in much more cost-effective interventions. There are now multiple examples of mea- suring service delivery more successfully, for example, the experience of the Monitor Esco- lar in Bogotá, which collects and automatically analyzes information from school principals, teachers, students, and parents. The information can be collected through different means, and the platform offers great potential to be up-scaled and established as the basis for a new management focused on results. In health, the situation is analogous. Although there are standards for service provision, measurement of compliance with standards is very limited. For example, only 80 percent of pregnant women are screened for HIV and only 3 percent of newborn children are screened for hypothyroidism, indicating clear gaps in adherence to basic standards of quality of care. Unfortunately, there is no information about the adherence to most other clinical guidelines or care pathways, especially for PHC, posing a limitation on the potential analyses of the qual- ity of care received by patients across the system. In addition, since incentives for health care quality are not fully built into payment mechanisms, there is little motivation for health insur- (ii) ers and health care providers to improve the quality of services. In social protection, the quality of the interventions depends on both the pertinence of Improving the management and effective use the implementation arrangements of the social programs and the user experience of cit- of data at all levels of government izens interacting with the system. It is key to ensure that social programs that deliver goods or services to the communities—such as ICBF’s ECD services—have implementation arrange- Local governments play a fundamental role in the provision of services, especially in ed- ments that certify the quality independently of the location or context of the beneficiary com- ucation but also in health and social programs. Schools and public hospitals and clinics munities. Toward that end, it is important to implement reliable, timely, and well-designed are managed and supervised locally. However, the capacity of local governments in terms of monitoring and evaluation systems and adequate tools for capturing user feedback. For cash physical and human resources and management practices varies significantly. That capacity transfer schemes, modernizing and simplifying delivery systems are usually good strategies for tends to be lower in places with poor outcomes, which is likely to increase inequality. reducing the administrative burden imposed on citizens. Implementing government-to-per- son (G2P) payment schemes or promoting one-stop shops for social services are examples of There is a need to improve management practices across service providers, local govern- how to make the social protection system more accessible. ments, and the central government, improving their coordination. Local governments have very little capacity to manage programs and the use of data for diagnosing school needs, and planning based on those needs is very rare. In programs such as the School Feeding Program (PAE), the role of the local government is essential to ensuring equitable access and quality. The lack of reliable sources of information and data analysis tools significantly hampers the ca- pacity of local governments to prevent mismanagement and to identify implementation risks on time. In health, the World Bank assessment of PHC system capacity (World Bank 2020c) showed important variations in capacity for planning health services and implementing new models of care at the territorial level. (iii) Providing adequate, equitable, and results-based financing Financing is often inadequate and inequitable. In addition to developing standards and qual- ity assurance systems and improving management of programs, it is critical to provide financ- ing that is adequate and equitable and that incentivizes results. Currently in education, there is very little investment in quality improvements, and those resources are highly unequally distributed. In health, the capitation transfers provided to health insurers do not sufficiently take into account the risk profile of the patients, incentivizing the quantity instead of quality of service provision even as quality has become the most prominent barrier to accessing care. Another example is the tertiary education financing system, where there is ample consensus about the need for reform.19 Under the current system, the financing of public universities is not associated with any outcome—not even enrollment.20 In addition, the transfers are highly inequitable among institutions. These inequalities should be reduced, but the funding also needs to be more results based. The use of results-based financing for tertiary education is growing as an effective and efficient tool to improve its quality. This can be used to support and incentivize institutions to achieve specific goals, such as quality improvements or the curricular reform flexibility discussed above. (iv) Improving cross-sectoral coordination Better intersectoral coordination is also needed. School health programs need to be strengthened, particularly health screening, substance abuse, and nutrition programs. The School Feeding Program (PAE) is the ideal anchor for nutrition interventions in schools. The National School Health (Encuestas Nacionales de Salud en Escolares [ENSE]) and Youth Tobac- co Use (Encuestas Nacionales de Tabaquismo en Jóvenes [ENTJ]) surveys in 2017 show that too many students have poor health nutrition practices. Almost half of 13–18-year-olds in the survey report having visual problems, yet almost a quarter of students report never having been screened for visual health issues (inside or outside the school). In rural areas, it is almost a third of students. Alcohol consumption is too frequent and starts at early ages: 40 percent of 13–18-year-olds report having consumed alcohol before age 14. Health indicators are consis- tently worse in public schools than in private schools, showing the need to strengthen school health programs in the former. Nutrition practices also need to improve. Nine in 10 students fail to meet the recommended fruit and vegetable consumption frequency and almost three- fourths of students consume sugary drinks daily. Policy Options for More Equitable 29 B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Human D evelopment and Equity Access to Human Capital Observed Drivers of Policy Options Relevant International Experiences Timing for Consideration about and Equity Gap the Equity Gap Implementation Estimates of Fiscal Impact of (Immediate/ Implementation Short Term, or Medium/Long Term) Quality of health Hospital-centric model of care Transform service delivery model into a PHC- Costa Rica has rolled out national multi- Medium/long term Further work is needed to care services delivery is ineffective in meeting based model of care adapted to local needs. disciplinary care teams and empanelment quantify the cost. the needs of the most vulnerable strategies to improve access to high-quality populations with chronic condi- Introduce financial incentives in the capitation PHC (https://improvingphc.org/promis- tions who require comprehen- payment for health insurers and health care ing-practices/costa-rica). sive care and better integration providers to improve quality care and achieve of health and social care. better health outcomes for the most vulnera- Chile has rolled out the systematic collec- ble populations. tion of data on PHC performance, man- dating accreditation at the PHC level, and Introduce an accreditation system for health has strengthened the governance of pri- insurers requiring them, among other mea- mary care at the municipal level (https:// sures, to develop population health man- improvingphc.org/measuring-prima- agement strategies for the most vulnerable ry-health-care-system-performance-us- populations. ing-shared-monitoring-system-chile-0). Strengthen local governance by reporting Peru has rolled out mechanisms to health inequality data at the municipal level strengthen engagement with communities on a regular basis and requiring municipalities at the municipal level to engage in better to develop plans to reduce health inequalities planning of services depending on popula- and assess progress on a yearly basis. tion health needs (https://improvingphc. org/pursuing-universal-health-cover- age-through-local-community-participa- tion-peru-0). Inequality in access The limited coverage of ICBF ECD Simplify the administrative procedures to ac- Chile’s MDSF Gestión Social Local platform Immediate/ Minimal fiscal costs for adjust- to ICBF’s ECD ser- services in areas where private cess ECD services that generally act as entry and some other local strategies, such as the short term (for sim- ing documentation require- vices childcare is not available or af- barriers, particularly in rural areas. This im- ECD monitoring platform in Peñalolén, are plifying the adminis- ments. Up-front fiscal costs fordable to the poorest families plies: (1) reducing the number of documents good examples of user-centered and inte- trative procedures to for the integration of IT sys- means that many children are required by verifying information directly with grated social services delivery. access ECD services) tems. Further work is needed placed into low human capital other public agencies; and (2) integrating the to quantify the up-front cost. accumulation trajectories, with ICBF’s information systems. Countries like the United Kingdom, Chile, Medium/long term (all negative effects on productivity and India include specific child develop- other measures) Up-front fiscal costs for in- over the lifecycle. Increase the supply of ICBF services in rural ment monitoring indicators in their mon- cluding quality assessments areas. itoring and evaluation systems, including into the service provision and In rural areas, the fragmentation standardized tests and age-appropriate for improving the monitoring of ECD service modalities may be Define more effective quality-assessment qualitative assessments. and evaluation structure. Fur- causing quality gaps in ICBF ser- strategies for ECD services. ther work is needed to quanti- vices that also affect the human Most high-income countries have strong fy these costs. capital accumulation of children Improve the monitoring and evaluation struc- ECD monitoring systems in place that com- over their lifecycle. ture of ECD service delivery by: simplifying bine official inspections with other forms Minimal fiscal costs for im- the current formats based on input, output, of monitoring, such as self-assessment and proving data used as input for process, and results indicators; and build- parental surveys. Out of 22 OECD countries, BETTO22. ing information systems to ensure timely, 18 (82 percent) have self-assessments and high-quality data collection and processing at 15 (68 percent) also use parental. For ex- the local level, useful to generating alerts to ample, the quality rating and improvement detect problems with service provision. system (QRIS) in the United States and the Ofsted system in the United Kingdom give families a way to see and compare pro- grams’ quality ratings. (World Bank, 2020)21 Kenya, Argentina, Germany, and Brazil have childcare programs that have, as Colombia has, decentralized ECD service provision and that can illustrate some solutions to challenges related to decentralization, monitoring, procurement, and quality as- surance. Decentralization can positively impact services by facilitating greater sen- sitivity to local needs; however, it can also raise challenges, especially in widening differences in access and quality between regions (World Bank, 2020f) Inequalities The curricular autonomy at the Introduce core curriculum guidelines for basic Several countries are responding to the Immediate/ The additional cost of the in learning school level, structural inequali- competencies. pandemic with evaluations and prioritized short term introduction of core curricu- ties in teacher quality, and ineq- curricula to accelerate learning (Chile, lum guidelines is small if it is uities in allocation of financing Link the Programa Todos a Aprender (which United Kingdom, New Zealand, Canada). articulated with PTA, since the across schools/regions/munici- provides pedagogical support to teachers) to See https://www.mineduc.cl/chile-recu- program already exists and palities create large inequalities the core curriculum guidelines. pera-y-aprende/. covers more than 2 million in the quality of pedagogical students. practices. Introduce/expand programs that teach stu- The United Kingdom has launched a na- dents at the right level (focus on developing tionwide massive student tutoring program Scaling up tutoring programs Insufficient access to services in basic math and reading skills based on stu- (https://nationaltutoring.org.uk/). will have additional costs. rural areas remains a concern in dents’ needs rather than age or grade through, Further work is needed to es- post secondary education. for example, student tutoring such as Brujula, Teach to the Right Level programs ran by timate these costs, but they Aprendamos Todos a Leer). Pratham Foundation in India have shown are expected to be small. strong results on learning outcomes at Complement in-person learning with Adaptive a small cost (https://www.teachingat- CAL where connectivity is available. therightlevel.org/). Inequalities Lack of relevance and quality Introduce a competency-based curriculum in In Mexico, upper secondary education Medium/long term The additional cost of the pro- in early dropout of upper secondary education, upper secondary school, with socio-occupa- reform has improved student enrollment posed reforms will be associ- combined with low levels of tional skills. and graduation rates (http://sems.gob.mx/ ated only with the training of learning when entering upper en_mx/sems/reforma_educativa_ems). school management teams secondary education, generates Increase the supply of upper secondary edu- and the teacher training com- early dropout for disadvantaged cation in rural areas. The European Union is moving to a compe- ponent. students. tency-based framework (https://ec.europa. Strengthen linkages with tertiary education eu/education/policies/school/key-compe- Articulacion con la Media al- and improve the relevance of the Articulación tences-and-basic-skills_en). ready exists. The cost of its con la Media program by diversifying the sup- reform will be small, though ply of programs offered from SENA (Servicio The challenge of introducing competen- the fact that SENA current- Nacional de Aprendizaje, the National Training cy-based curricula is its proper implemen- ly covers most of the cost is Service) and diversifying providers of the pro- tation. It requires intensive school director a challenge if providers are gram, with participation from the local private and teacher training and can be a challenge diversified. It may require a sector. in very low-capacity areas. reallocation of funds to the Ministry of Education. Persistent Inequitable financing of public Financing should move toward a per student Per student financing is a feature of most Full implementation in Changing formulas does not inequalities in tertiary education institutions allocation of service provision transfers that advanced tertiary education financing the medium term, but necessarily imply higher fiscal access to quality generates substantial differences reflects the operational cost of providing a models (including most countries in Eu- there are steps that costs initially. As enrollments tertiary education in the quality of service provi- good quality tertiary education and also equi- rope). can be taken in the im- increase in the medium term, sion. ty concerns. mediate term to set up they will demand more re- Chile has long experience with perfor- the systems. sources. High reliance on private provi- Any resources for public tertiary education mance-based financing in tertiary educa- sion and student loans results institutions that are not covering operational tion through the MECESUP program (http:// Higher up-front investment in high repayment burdens that costs should be pooled to avoid fragmenta- dfi.mineduc.cl/). in exchange for more sus- limit the demand for student tion and linked to results and institutional im- tainability in the short term. loans for low-income families. provement plans. The United Kingdom and Australia have Income-contingent loans re- very advanced income-contingent student sult in less repayment in the Income-contingent loan schemes should be loan schemes that also include public edu- short run (as students have implemented in which repayment depends on cation. Among developing countries, Chile relatively low entry salaries), income upon graduation. and South Africa have small schemes, as but higher repayment over all do Hungary and Thailand, with varying (as students contribute more success. Colombia is very well placed to when their salaries increase). implement this scheme because of its long tradition of student loans and the availabil- ity of robust information systems. Unequal access to The Sisbén lacks dynamism be- Create a dynamic, reliable, and integrated The social registries of Brazil and Chile in Immediate/ Up-front small fiscal costs for social programs cause of its census sweep struc- single social registry by: (1) integrating Sisbén Latin America and Pakistan and Philippines short term for the so- the consolidation of the social ture, causing data to become with both administrative data sources and in other regions are good examples of how cial registry consolida- registry and for the adjust- outdated very quickly. nominalized social programs registries to keep integrated social registries used to channel tion ment of the design and inte- the information updated; (2) implementing a the targeting and registries of multiple so- gration of social programs The lack of quality of Sisbén data co-responsibility clause in the social programs cial programs could help improve the allo- usually explains the persistent- registry to ensure that citizens update their in- cation of benefits. Medium/long term for ly high exclusion and inclusion formation on Sisbén. the adjustments of so- errors from flagship social pro- Perú’s Performance Fund (Fondo de Es- cial programs grams. Introduce user-centered design into current tímulo al Desempeño y Logro de Resultados and future social programs, particularly those Sociales [FED]) is an example of inter-in- Furthermore, even when so- implemented by the DPS, by: (1) removing un- stitutional coordination. FED acts as an cial programs are provided by necessary documentation and implementing incentive mechanism directed at regional the same public agency, most an electronic household folder that stores the governments, providing them with both of them still have separate and basic documentation required by the multiple technical assistance and supplementary independent delivery chains, programs, avoiding the need for citizens to resources to incentivize them to reach pre- which translates into entry bar- make multiple requests; (2) reactivating fam- determined targets related to the delivery riers and administrative burdens ily support programs, such as UNIDOS, as the of an integrated package of services for for citizens. Particularly for rural initial step for assessing the needs of families pregnant women and children under five, areas, where electronic and G2P and defining personalized integrated poverty such as prenatal treatments, micronutrient payments are difficult, people reduction intervention plans (i.e., Ruta de la supplementation, vaccines, growth mon- may need to travel several times Superación de la Pobreza). itoring, and early stimulation and educa- to claim their benefits or get a tion, as well as different community-level specific document for registry. services like childcare or water and sanita- tion, among others. Conclusion This chapter has shown that improving equity in human capital is not only essential to promot- ing equitable outcomes, it is also an urgent priority. The impacts of COVID-19 have deepened pre-existing inequalities in human capital, and the magnitude of the impact demands urgent and bold actions. Improving the quality of education, PHC, and social protection services emerges as a key priority. Improving the efficiency of spending through better management and use of data is also a priority in all three sectors. The challenge posed by the COVID-19 cri- sis means that more resources will be needed to address existing inequalities. It is imperative that any additional resources be distributed in an equitable way that also incentivizes results. 30 Endnotes BU I L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | H uman Development and Equity 1 Extensive information about the Human Capital Index, including visualiza- 11 At the time of the writing of this report, the hybrid scheme in which not all chil- tions, detailed data, and methodology, can be found at https://www.world- dren attend school every day in Colombia is called “Alternancia.” bank.org/en/publication/human-capital. 12 See World Bank 2021b. 2 A fourth subcomponent of the HCI—the adult survival rate—is not available 13 Departamento Nacional de Planeación (2017-2018). for Colombia by income. For this reason, the version of the HCI calculated by 14 Heller et al (2015). income presented here does not incorporate the adult survival rate. The adult 15 For more on this issue, see B. Chapman, T. Higgins, and J. Stiglitz, ”Intro- survival rate is used for the national HCI calculation as well as the calculation duction and Summary,” in Income Contingent Loans: Theory, Practice and of the HCI by department. Prospects, ed. Bruce Chapman, Timothy Higgins, and Joseph E Stiglitz (Bas- 3 OECD, “Education GPS: Colombia,” https://gpseducation.oecd.org/Country- ingstoke and New York: Palgrave Macmillan, 2014), 1–11; and Lozano, Pirog, Profile?primaryCountry=COL&treshold=10&topic=PI. and Cerdan-Infantes, “Equity and Sustainability in Higher Education Financial 4 The exit exam is SABER 11, which is a prerequisite for graduation and for ac- Aid: Analysis of Income Contingent Loans in Colombia” (Lousville: Journal of cess to tertiary education. The exam score is also used by universities in their Student Financial Aid, forthcoming). student acceptance decisions. Thus a high score results in a higher probability 16 Before that, since its creation in 1968, the learning assessments from ICFES of accessing good quality tertiary education. were purely for accessing tertiary education. 5 SABER PRO is a standardized test taken at the end of tertiary education that 17 Colombia is one of a few countries that also measures learning outcomes in measures basic competencies. See https://www.icfes.gov.co/acerca-del-exam- tertiary education through SABER PRO. en-saber-pro. 18 Luna Bazaldua,Diego Armando; Levin,Victoria; Liberman,Julia.2020. Guidance 6 Downloaded from https://www.dane.gov.co/index.php/estadisticas-por-tema/ Note on Using Learning Assessment in the Process of School Reopening (En- salud/calidad-de-vida-ecv/encuesta-nacional-de-calidad-de-vida-ecv-2019 glish). Washington, D.C. : World Bank Group. http://documents.worldbank.org/ 7 In English, the Identification System for Potential Beneficiaries of Social Pro- curated/en/856951606239586214/Guidance-Note-on-Using-Learning-Assess- grams. ment-in-the-Process-of-School-Reopening. This can be done with representa- 8 World Bank (2017). tive samples instead of a census, considering that cost is an important driver in 9 M. Eslava and others, “The Socioeconomic Patterns of COVID outside Advanced the current situation. Economies: the Case of Bogotá,” Documento CEDE 45 (Bogotá: Centro de Es- 19 See World Bank, “Tertiary Education in Colombia” (Washington, DC: World tudios sobre Desarrollo Económico, 2020). See also Rachid Laajaj and others, Bank, 2012). See also World Bank (2020d). “SARS-CoV-2 Spread, Detection, and Dynamics in a Megacity in Latin America,” 20 Law 30 from 1992 establishes transfers to existing universities and fixes a floor Documento CEDE 18 (Bogotá: Centro de Estudios sobre Desarrollo Económico, for increases in these transfers as IPC. The floor, in effect, has become the ceiling. 2021). 21 World Bank. 2020f. 10 These figures are from the Colombia Economic Reactivation dashboard: 22 BETTO is the name of the AI algorithm created by the ICBF, designed to create a https://coronaviruscolombia.gov.co/Covid19/estadisticas-covid-19/reactiva- more transparent and effective selection of service providers for ECD services. cion-economica.html. More information on BETTO is available in: https://www.icbf.gov.co/betto 31 CHAPTER 3. Making the Colombian BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LOMBI A Labor Market More Inclusive Summary 32 BUI L DI N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | Making the Colombian Labor Market More Inclusive Inequities in the labor market are the main driver of overall income inequality in Colombia. The advent of COVID-19 and its associated labor market impacts may widen these gaps fur- ther. Making the labor market more inclusive is critical to improving overall equity. Despite improvements over the past decade, important gaps remain, particularly with regard to ac- cess to formal jobs with decent earnings among women, indigenous populations, Afro-de- scendants, the low skilled, people in rural areas, and migrants. There are three policy objectives for labor markets in Colombia that are critical given the seis- mic changes experienced by labor markets in the country and across the world: making labor markets more efficient, promoting inclusion, and improving access to technology. There are a number of policy recommendations to promote inclusivity in the labor market. These include: i. Improving regulations that discourage the creation of new jobs in the formal sector, par- ticularly for low-skilled workers. These include policies to: a. Reduce formalization costs for firms and the self-employed. b. Create a unit of independent experts to advise the Comisión Permanente de Concerta- ción de Políticas Salariales y Laborales on decisions about the minimum wage, mini- mize increases beyond the inflation rate, and consider reforms to prevent automatic increases that may weaken job creation and efficiency. c. Improvethe link between contributions and benefits in the social insurance system while strengthening the redistributive aspect. ii. Implementing reforms to strengthen the inclusion of vulnerable workers by: a. Expanding active labor market programs following international practices. b. Reducing gender gaps by facilitating the formalization of part-time jobs, improving the supply of childcare, and equalizing the retirement age across genders. the link between the menu of services offered by the Servicio Nacional de c. Improving Aprendizaje and the Servicio Publico de Empleo and the needs of the migrant, minority, and displaced populations. iii. Carrying out reforms that promote a more equal access to digital technologies to open op- portunities to the jobs of the future for more vulnerable workers and to mitigate the impacts of COVID-19 by expanding the internet infrastructure, improving the quality of education, and implementing an equitable and business-friendly teleworking legal framework. 3.1 33 B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | M ak ing the Colombian Labor M ark et M ore Inclusive Introduction Labor market outcomes have improved in Colombia over the past decade. Both labor force participation and employment rates increased 5 percentage points in roughly one de- cade (between 2008 and 2019), and most progress was achieved during the commodity price boom that ended in 2014 (figure 3.1). The share of workers with a formal job also increased. According to the legal criteria, which consider a job informal if it does not provide social insur- ance coverage, formality grew by 7 percentage points. The productive criteria, which consider the unskilled self-employed and those in small firms and domestic services to be informal, also show an increase of about 3 percentage points. Real hourly earnings grew by 13.4 percent at an average yearly rate of 1.1 percent during this period.1 These improvements in labor market outcomes since 2008 have been inclusive for most but not for all, according to some indicators. Changes in earnings and formality across groups are important because the gap in access to jobs with decent earnings and in the for- mal economy is wider than the gap in access to jobs per se. Figure 3.2 shows that though gaps in labor force participation (LFP) and employment rates are not very large, gaps in access to FIGURE 3.1. Labor Market Outcomes in Colombia, 2008–19 formal jobs and in earnings levels are significant. The latter is particularly true for Venezuelan immigrants, the indigenous, the low skilled, those in rural areas, and the poor. When looking 80 72,6 65,0 at trends, improvements in labor market inclusion are mixed. The earnings gap has decreased 70 67,8 significantly for most groups except low-skilled workers and indigenous people (figure 3.3d). 60 60,3 43,9 However, the low skilled continue to receive hourly earnings that are 74 percent lower than 50 39,7 40 high-skilled workers. This skills earnings gap is comparable to that observed in other Latin 40,8 30 33,1 American and Caribbean (LAC) countries, such as Brazil (72 percent) and Honduras (74 per- 20 cent), but wider than that of other LAC countries with a more equitable income distribution, 10 such as Costa Rica (65 percent), Panama (63 percent), and Uruguay (60 percent). Indigenous 0 people received earnings 38 percent lower than non-indigenous and non-NARP2 workers in Labor Force Employment Formality rate Formality rate Participation (productive) (legal) 2019, a gap wider than in 2014 (31 percent). The gap in access to formal jobs has increased or 2008 2019 remained stable for all groups (figure 3.3c). This is because though the share of formal jobs increased for all groups, it increased more rapidly for richer workers with higher levels of ed- Source: Author’s estimates, based on data from the GEIH (Gran Encuesta Integrada de ucation and living in urban areas. In other words, the relative gap with vulnerable workers Hogares/Large Integrated Household Survey) 2008, 2019. Note: Individuals aged 15–64 years. The legal informality criteria consider a job to be infor- expanded. For instance, though the share of workers with a formal job in the poorest quintile mal if it does not provide social insurance coverage. The productive informality criteria consider the unskilled self-employed and those in small firms and domestic services to be increased 3 percentage points from 2008 to 2019, the increase reached 9 percentage points informal. among workers in the richest quintile. FIGURE 3.2 Labor Market Gaps across Different Groups, 2019 Women 10 0 Venezuelan -10 Youth -20 -30 -40 -50 -60 NARP -70 Low skilled -80 -90 Indigenous Rural Quintile 2 Quintile 1 LFP Employment Formality (legal) Hourly wages Source: Author’s estimates, based on data from the Socio-Economic Database for Latin America and the Caribbean. Notes: The LFP, employment, and formality (legal) gap is the difference between the LFP, employment, and formality rates of the corresponding group and that of the comparison group as a percentage of the latter. The earnings gap is the difference between the earnings of the corresponding group and that of the comparison group as a percentage of the latter. The comparison group for each category is shown in parentheses: a) women (men); b) youth (25–40 years old); c) low skilled (more than 13 years of education); d) rural (urban); e) quin- tile 1 (quintile 5); f) quintile 2 (quintile 5); g) indigenous (non-indigenous and non-NARP); h) NARP (non-indigenous and non-NARP); i) Venezuelan (Colombians and non-Venezuelan migrants). The NARP group includes Black, Afro-Colombian, Raizal, and Palenquero populations. Sample includes individuals aged 15–64 years (except for youth, 15–24 years old). Low skilled are those with eight years of education or less. FIGURE 3.3. Labor market Gaps across Different Groups, changes 2008–19 A LFP B Employment 8 8 Percentage points (change in gap) Percentage points (change in gap) 6 6 4 4 2 2 0 0 -2 -2 -4 -4 -6 -6 -8 -8 en h ed l 1 2 ) ) en h ed l 1 2 ) ) ra ra 19 19 19 19 ut ut e e e e om om ill ill Ru Ru til til til til 0 20 0 20 Yo Yo -2 -2 sk sk in in in in 4- 4- W W 14 14 Qu Qu Qu Qu w w 01 01 20 20 Lo Lo (2 (2 s( s( RP RP ou ou NA NA n n ge ge di di In In C Formality (legal) D Hourly wages 8 8 Percentage points (change in gap) Percentage points (change in gap) 6 6 4 4 2 2 0 0 -2 -2 -4 -4 -6 -6 -8 -8 en h ed l 1 2 ) ) en h ed l 1 2 ) ) ra ra 19 19 19 19 ut ut e e e e om om ill ill Ru Ru til til til til 0 20 0 20 Yo Yo -2 -2 sk sk in in in in 4- 4- W W 14 14 Qu Qu Qu Qu w w 01 01 20 20 Lo Lo (2 (2 s( s( RP RP ou ou NA NA n n ge ge di di In In Source: Author’s estimates, based on data from SEDLAC. Notes: Each bar shows absolute changes in gaps over time, with negative (positive) numbers indicating a reduction (increase) in the gap. See figure 3.2 for a description of the demo- graphic groups. The LFP, employment, and formality (legal) gap is the difference between the LFP, employment, and formality rates of the corresponding group and that of the compari- Box 3.1. The World Bank’s Jobs Diagnostic Report for Colombia Colombia’s labor market is highly susceptible to slowdowns in economic activity, even as it is becoming less re- sponsive to upswings. Employment growth has become gradually less responsive to aggregate growth in the economy. Although both employment and wage employment expanded in tandem with economic growth of 5 percent per annum in the 2009–15 period, they hardly grew thereafter, even though the economy continued to grow at 2 percent per annum (figure 1). The halt to employment growth in the post-2015 period implies that fewer jobs were being created for every percentage point in GDP growth (figure 2). This high susceptibility of the labor market to a relatively modest slowdown is surprising. Most likely, however, it reflects structural weaknesses in the Colombian economic model that call for an in- depth BOX examination 3.1. of theJobs The World Bank’s changing nature Diagnostic of thefor Report country’s labor market. Colombia FIGURE 1. GDP, Employment, and Wage Employment Growth FIGURE 2. Change in Employment in Response to a One In Comparison, 2009–19 Percent Increase in GDP 1.0 140 0.8 Percentage points 130 Index (2009=100) 0.6 120 0.4 110 0.2 100 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 1 15 9 1 15 9 02 01 02 01 0 0 -2 -2 -2 -2 -2 -2 09 12 15 09 12 15 GDP employment wage employment 20 20 20 20 20 20 elast. of employment wrt. GDP growth Source: Author’s calculations using GEIH and national accounts data. Source: Author’s calculations using GEIH and national accounts data. A forthcoming World Bank report analyzes the structural weaknesses of the Colombian economic model and how these intertwine with outcomes in the labor market. By applying the World Bank’s Jobs Diagnostics methodology and drawing on both micro- and macro-level data covering the period 2008–19, the report casts light on Colombia’s most pressing jobs challenges, including: • a slowdown in net job creation combined with a high exposure of the labor market to the economic cycle • an excessive concentration of jobs in low-productivity service activities, particularly in urban areas, with few quality jobs available in rural areas • limited employment opportunities for the lesser skilled in the population, implying low earnings, high rates of self-employment, and high levels of informality • fewer opportunities for skilled workers, resulting in rising unemployment rates and longer unemployment spells for the more educated Colombia’s economic boom during the pre-2015 expansionary cycle exacerbated the reliance on services as a driver of employment growth. The steady rise in international commodity prices between 2009 and 2014 attracted investment in the extractive sector, driving a strong expansion in production. Most of the benefits were indirect on ac- count of the income windfall triggered by rising extractive rents. Higher incomes fueled a boom in domestic demand that primarily benefited more labor-intensive activities in the services and construction sectors. This contrasts with the more subdued performance in the agriculture and manufacturing sectors, which suffered a decline in export demand due to the political crisis in neighboring Venezuela and also a loss in competitiveness because of rising costs for (typically non-tradable) activities, such as financial intermediation, transport, or real estate. These developments therefore exac- erbated the reliance on services as a driver of employment growth. Strong increases in the minimum wage pushed unskilled workers into the informal sector. Colombia’s policy inter- ventions in the labor market have traditionally been geared at strengthening workers’ rights and improving labor market intermediation in a predominantly urban context. The minimum wage is arguably Colombia’s most important instru- ment for intervening in the labor market and is set annually. A key challenge comes from the fact that the minimum wage is constitutionally mandated to increase every year by at least the inflation rate, which has left the country with one of the highest minimum wages in the region. It also makes the minimum wage downwardly rigid, which leaves the onus of adjustment during slowdowns on employment rather than on wages. The forthcoming report documents how this has particularly negative effects on unskilled workers who are forced to seek employment opportunities outside the formal sector in less productive self-employment activities. The report also documents how the slowdown in aggregate demand and the strong increases in labor costs be- cause of the rising minimum wage have had important effects on firm performance in the private sector and how this in turn affected hiring dynamics in formal enterprises. Aggregate firm dynamics in Colombia suggest that the main problem for employment growth in the formal sector is the low survival rates of firm start-ups rather than low lev- els of firm entry per se. Over the past decade, rising labor costs seem have been a major constraint in new firms’ ability to grow and in established firms’ ability to survive. Entry and exit dynamics by firms resulted in a shift toward larger, more capitalized firms and a displacement of unskilled labor, whose costs rose particularly steeply on account of the minimum wage. At the sectoral level, it resulted in a contraction of labor-intensive activities. These trends were exacerbated by the post-2015 slowdown in economic activity. Source: World Bank, “Colombia: Jobs Diagnostics” (Washington, DC: World Bank, forthcoming). FIGURE 3.4a. The Quality of Jobs in Terms of Formality and Workers’ FIGURE earnings 3.4b. and Wage vs. formal status Household Income continue to2008–19 Inequality, be relatively poor in Colombia when Earnings in LAC, circa 2018 compared to that of its peers (figure 3.4a). Colombia has one of the lowest levels of formal- ity in the region. Accordingly, it has the second-lowest average real hourly earnings; even in 80 0,51 countries with lower levels of GDP per capita, such 2010 as Paraguay and Peru, workers receive a 0,51 2008 70 higher hourly pay than in Colombia. gap between Colombia and other countries in LAC This2011 2012 Chile 2014 Costa Rica Uruguay 0,5 has shrunk, but it is still significant when comparing 2009 the outcomes of workers with high levels formality (percentage points) Hourly wage inequality (Gini) 60 Argentina 0,5 El Salvador Mexico Brazil Panama of education across countries. For example, 2013 the average earnings (in current U.S. dollars) of 50 0,49 Colombia Dominican Rep. workers in Colombia with at least 13 years of education represents 48 percent of that of their Peru 0,49 40 Paraguay counterparts in Uruguay. When comparing workers with at most eight years of education, the Ecuador 0,48 2015 Honduras fraction is 31 percent. 30 Bolivia 0,49 2016 2019 20 0,47 Inequities in the labor market continue to be significant and are important drivers of 2018 0,47 income inequality overall in Colombia. Given the importance of labor among all income 10 0,46 changes sources, 2017 in earnings inequality are accompanied by changes in the same direction in 0 0,46 inequality of household income per capita (figure 3.4b). In other words, reducing inequality in 0 1 2 3 4 5 6 7 0,5 0,51 0,52 0,53 0,54 0,55 0,56 0,57 the labor market could have significant repercussions on overall equity in Colombia. Hourly earnings (current US$) HH Income inequality (Gini) Source: Author’s estimates, based on data from SEDLAC. The labor Source: market Author’s estimates,has basedbeen on datadeeply impacted by the COVID-19 crisis, widening inequalities. from SEDLAC. Note: Data for Colombia are from 2019. Informality defined using the productive defini- As mentioned Note: in The vertical axis Chapter measures 1, at the Gini the onset coefficient of earnings. of hourly The there the crisis were horizontal axis huge job losses: 5.5 million jobs tion. measures the Gini coefficient of household income per capita. had been lost in April 2020 compared to the previous year. The results from the World Bank High Frequency Surveys (HFS) indicate that countries with higher levels of informality expe- lity of Jobs in Terms of Formality and FIGURE 3.4b. Wage vs. Household Income Inequality, 2008–19 rienced larger job losses during the first months that lockdown measures were implemented a 2018 (figure 3.5a). The shock affected women workers more than men and even as the labor market 0,51 recovered toward the end of 2020, the pace of recovery for females trailed that of men (figure 2010 0,51 2008 3.5b). Chile 2012 Costa Rica 0,5 2014 2011 Uruguay 2009 Hourly wage inequality (Gini) Argentina 0,5 The COVID-19 crisis is accelerating global trends in the labor market that will tend to in- Mexico Brazil Panama 2013 crease inequality in Colombia. In recent decades, countries at very different stages of eco- 0,49 Dominican Rep. 0,49 nomic development have witnessed slow but steady changes in the labor market. The spread Paraguay 0,48 2015 of new technologies—such as the internet, artificial intelligence, industrial robots, and so on— Ecuador onduras 0,49 implied that the demand for workers with the skills that facilitate their adoption increased, Bolivia 2016 0,47 2019 while the demand for workers with the skills that can be automated declined (Acemoglu and 2018 Autor 2011; Kelly et al. 2017; World Bank 2016b). The implementation of lockdown and social 0,47 distancing measures during COVID-19 implied that the gap between both groups of workers 0,46 2017 increased suddenly and dramatically. Low-skilled workers outside sectors deemed essential 0,46 2 3 4 5 6 7 0,5 0,51 0,52 0,53 0,54 0,55 0,56 0,57 experienced massive employment losses, while the high-skilled were more likely to be able Hourly earnings (current US$) HH Income inequality (Gini) to work from home and hence keep their jobs and wage levels (Ashcraft, Fernández-Val, and sed on data from SEDLAC. Source: Author’s estimates, based on data from SEDLAC. Lang 2013; Montenovo et al. 2020). Moreover, some of these changes in work arrangements rom 2019. Informality defined using the productive defini- Note: The vertical axis measures the Gini coefficient of hourly earnings. The horizontal axis may not be temporary but may stay even after the end of the crisis. measures the Gini coefficient of household income per capita. This chapter focuses on two barriers to improving equity in the Colombian labor market. Although factors driving labor market exclusion are complex and diverse, this chapter focuses on two that are critical given recent mega trends in labor markets around the world. The first one is related to the existence of distortions that limit the reallocation of workers from tradi- tional sectors with low productivity levels to dynamic ones. The second is the slow adoption of new technologies among disadvantaged groups and the role of the COVID-19 pandemic in accelerating those trends. FIGURE 3.5a. Wage Workers Pre-Pandemic and Loss of Jobs FIGURE 3.5b. Ratio of Number of Employed in Each Quarter at the Onset of the Pandemic (May 2020) and Each Group, compared to first quarter of 2020 (Q1 2020=100) 75% 120 Bolivia % of workers who loss their job either permanently or 70% 65% 110 temporarily at the onset of the crisis Perú 60% El Salvador 55% Honduras 100 Colombia República Dominicana 50% Ecuador 45% Guatemala 90 Paraguay 40% México Argentina Costa Rica 35% R2 = 0,848 80 30% Chile 25% 70 25% 45% 65% 85% Q1 Q2 Q3 Q4 Q1 Q2 Q3 % of wage workers (2019) 2019 2020 Hombres Mujeres Source: World Bank High Frequency Phone surveys, round 1 (May 2020). Source: Authors, with data from GEIH 2019 and 2020. 3.2. 34 B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | M ak ing the Colombian Labor M ark et M ore Inclusive Diagnostics: Addressing Inequality in the Labor Market Low growth in labor productivity Labor productivity growth in Colombia has been weak and can help explain the sluggish FIGURE 3.6. Labor Productivity (share of high-income performance of the labor market. Although labor productivity has grown almost 25 percent countries’ labor productivity) over the past 20 years, this growth was not enough to reduce the gap with developed coun- 40 tries: Colombia’s labor productivity as a share of that of rich countries has increased only 1 35 percentage point since 2000 (figure 3.6).3 Other countries that started with more similar pro- 30 ductivity levels in 2000 managed to achieve a higher rate of convergence. For example, Panama percentage points 25 and Costa Rica had productivity levels equivalent to 17 and 20 percent of that of high-income economies in 2000 and reached levels equivalent to 28 and 25 percent in 2019, respectively. 20 17 17 17 18 17 17 17 17 17 18 18 18 18 18 18 16 16 15 16 16 Without stronger productivity growth, it will be challenging if not impossible to increase wag- 15 es and the overall quality of jobs in a sustained manner in Colombia. 10 Raising labor productivity levels is a necessary condition for raising the earnings and 5 reducing the informality of workers in Colombia. As described above, high informality and 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 low earnings affect vulnerable workers the most in Colombia. But low-quality jobs—according to these dimensions—are often a symptom of low productivity levels. In fact, they are more Colombia Costa Rica Chile prevalent among sectors with low levels of labor productivity. As seen in figures 3.7a and 3.7c, Uruguay Panama Latin America and the Caribbean wages and informality levels across sectors are highly correlated with the level of labor pro- ductivity. In particular, a 10 percent increase in labor productivity is associated with a 3.7 per- Source: Author’s elaboration, based on data from the International Labour Organization cent increase in wages and a reduction of 2.6 percentage points in informality.4 For example, (ILO). Note: Labor productivity is defined as output per worker (GDP constant 2010 U.S. dollar), the coffee sector has a level of productivity 73 percent lower than the vegetable oil sector. ILO modeled estimates, November 2019. Accordingly, the share of formal jobs and the level of earnings are 90 percentage points and 146 percent higher in the latter. Low-productivity sectors, at the same time, concentrate a disproportionate share of employment, as shown by the size of the circles. The link between job quality and productivity is not only static but also strong when looking at changes over time. Sectors that have experienced the largest improvements in job quality (i.e., higher earn- ings growth and a larger informality decline) since 2008 are those that have also displayed stronger labor productivity growth (figures 3.7b and 3.7d). In particular, a 10 percent increase in labor productivity over time at the sector level is associated with a 1.2 percent increase in wages and a reduction of 0.4 percentage points in informality. The information and communi- cations technology (ICT) and food processing sectors are good illustrations of the importance of productivity growth for the creation of formal jobs with higher earnings. Although the for- mer witnessed a labor productivity growth of about 139 percent during the period, the latter experienced a decline in productivity of 44 percent. Accordingly, earnings (informality) in the ICT sector increased (decreased) by 43 percent (26 percentage points). In contrast, the food processing sector experienced a decline in earnings of about 6 percent and an increase of 4 percent in the share of informal jobs. FIGURE 3.7 Labor Productivity, Wages, and Informality by Sector A Wages vs. Productivity, 2019 B Wages vs. Productivity, 2008-2019 change 10.5 .5 Hourly wages (log), 2008−2019 change 10 Hourly wages (log), 2019 9.5 0 9 8.5 8 −.5 16 18 20 22 24 −1.5 −1 −.5 0 .5 1 Labor productivity (log), 2019 Labor productivity (log), 2008−2019 change C Informality vs. Productivity, 2019 D Informality vs. Productivity, 2008-2019 change 1 .4 Informality, 2008−2019 change .2 .5 Informality, 2019 0 0 −.2 −.5 −.4 16 18 20 22 24 −1.5 −1 −.5 0 .5 1 Labor productivity (log), 2019 Labor productivity (log), 2008−2019 change Source: Author’s estimates, based on data from GEIH 2008–2019 and the National Administrative Department of Statistics/Departamento Administrativo Nacional de Estadística (DANE). Note: Each bubble is proportional to the employment share of each sector in 2019. Labor productivity is computed as value added (from DANE’s estimates for 61 sectors) per worker (where the total number of workers comes from GEIH). Informality is computed using the legal definition. The straight line is estimated using an OLS regression, using sector employ- ment shares as weights. Informality not only leaves workers unprotected from risks but also further undermines the creation of formal jobs due to negative externalities on the formal sector in Colom- bia. The 2017 World Bank enterprise survey for Colombia indicates that when firms are asked to choose the top obstacle for businesses, the practices of informal firms are the most im- portant one, followed by taxes and corruption.5 At the same time, 47 percent of firms claim that such practices are a major constraint, a fraction well above the average of 31 percent for Latin America. These perceptions are consistent when comparing the outcomes of firms that faced different degrees of competition from informal firms. Rozo and Winkler (2019) show that as a result of the escalation of the Colombian armed conflict in the late 1990s and the 2000s, large sudden flows of displaced individuals joined the informal sector in the municipalities of destination. Formal firms operating in sectors that faced more intense informal competition experienced lower production, higher exit, and lower entry rates. The process of structural transformation—where traditional sectors shrink and new and more productive ones expand and absorb workers—is key to strengthening the inclusion of vulnerable workers in the labor market, but it has been very weak in Colombia. The weak labor productivity growth experienced since 2008 has exclusively been driven by with- in-sector productivity growth. In contrast, changes in the allocation of workers across sec- tors did not improve productivity, as labor shares remained stable and tended to move from relatively high- to relatively low-productivity sectors. As seen in figure 3.8, although some low-productivity sectors that absorb large numbers of workers (such as retail and wholesale sales, domestic services, and agricultural activities) have shrunk since 2008, the sectors that absorbed more employment (such as restaurants and hotels, transport, and construction) were not necessarily highly productive. At the same time, highly productive sectors such as ICT have experienced a significant decline in terms of their share of total employment since 2008. These movements account for a decline in labor productivity growth equivalent to 6.4 percent of the total growth experienced since 2008. FIGURE 3.8 Changes in Labor Shares vs. Labor Productivity Levels across Sectors, 2008–19 3 2,5 Restaurants and Hotels 2 Construction supporting activities Construction 1,5 Health and social services Administrative services Professional services Infrastructure Food (meat) 1 Public sector Education (public) Employment share change (p.p), 2008-2019 Transport-related activities Recycling Food (oils) Trasport Sewage Real estate 0,5 Finance Air transport Art & Entertainment Electricity -2,5 -1,5 -0,5 0 0,5 1,5 2,5 3,5 Sugar Oil Vehicle repair Forestry -0,5 Machinery and Gas equipment Coal Chemicals Coffee products Paper -1 ICT Other mineral products Domestic service Education (private) Coffee Commerce Grains Textiles -1,5 Livestock Agriculture -2 Labor productivity relative to the median (log), 2019 Source: Author’s elaboration, based on data from DANE and GEIH 2008–2019. Note: The size of each bubble is proportional to the total employment in the sector in 2019. The limited movement of workers from low- to high-productivity sectors points to dis- tortions in the Colombian labor market. Evidence from matched employer-employee data suggest that labor market flexibility in the U.S. labor market is substantially higher than in Co- lombia (Flórez et al. 2020). In particular, the reallocation of workers across firms through hiring and job separations is substantially lower in Colombia than in the United States. Flórez et al. (2020) argue that this is often a symptom of rigidities caused by economic policies and insti- tutions, such as rigid labor regulations or very high minimum wages. Although this evidence ignores job transitions within the informal sector or between the formal and informal sectors, Morales, Hermida, and Dávalos (2019) find that the informal sector has even less fluidity than the formal one in Colombia. This is because most informal jobs are in self-employment, tran- sitions from the informal to the formal sector are limited, and job-to-job transitions within the informal sector are not very common. Stringent labor regulations limit labor market fluidity and weaken the creation of formal FIGURE 3.9. Minimum Wages in Colombia jobs, especially for vulnerable workers. Almost 20 percent of firms in Colombia claim that A Earnings distribution by formality status, 2019 B Worker labor regulations are a major obstacle, a fraction higher than in most developing countries capita in (figure 3.10). Although it is challenging to show empirically how current labor regulations re- .002 100 strict job creation in Colombia, there is a large body of literature documenting this link in other 90 countries and even with regard to past reforms in Colombia. Studies focused on low-income 80 countries find that more stringent labor regulations are associated with lower formal sector .0015 70 and higher informal sector employment (Nataraj et al. 2014). Cross-country studies also find percentage points 60 that a heavier regulatory burden in the labor market tends to increase informality (Loayza, .001 50 Oviedo, and Servén 2005) and reduce employment among women, youth, and low-skilled 40 workers (Betcherman 2015). Núñez (2005) and Amarante, Arim, and Santamaría (2005) exam- 30 ined the impacts of the labor reform of 2003 in Colombia, which aimed at reducing youth and .0005 20 low-skilled unemployment through a new apprenticeship contract and also increasing flexility by expanding the workday, reducing firing costs, and allowing for more flexible work sched- 10 0 0 ules. Both studies found strong positive impacts of the reform on formal employment. Núñez 0 1000 2000 3000 4000 5000 (2005) found strong impacts on young and low-skilled workers, and Gaviria Uribe (2005) found 2019 monthly earnings in 1,000 current pesos significant effects on formalization among small enterprises. There is also evidence that the Informal salaried Formal salaried Informal self−employed reduction in labor costs driven tax Minimum by the3.9. FIGURE reform of 2012 WagesininColombia Colombia had large and positive employment impacts. Studies A found that it helped create more than 600,000 jobs in the long Earnings distribution by formality status, 2019 B Workers with earnings lower than the minimum wage, by household per Source: Au term (particularly among micro and small firms) and reduced informality (Bernal, Eslava, and capita income quintiles Note: a) S Meléndez 2015; Fernández and Villar 2017; Morales and Medina 2017). according Each bar s .002 100 96 97 High minimum wages can hurt the creation of job opportunities in the formal sector, par- 90 ticularly for low-skilled workers. The minimum wage in Colombia—relative to actual median 80 72 72 income levels—is among the .0015 highest in the region, and higher than among OECD economies 70 percentage points (Messina and Silva 2019). Flórez et al. (2020) argue that this policy could be harming the effi- 60 cient reallocation of workers across .001 jobs in Colombia. This instrument directly affects fewer 50 52 50 than half of workers (40 percent) in the economy, that is, formal workers. In fact, the vast ma- 40 33 34 jority of them do receive earnings just above the minimum wage, which suggests that in fact it 30 .0005 is binding (figure 3.9a). This policy does not reflect the level and the distribution of earnings. 20 13 12 Since 2008, almost all workers in the poorest quintile of the income distribution have received 10 earnings lower than the minimum 0 wage. In contrast, that figure is just 12 percent of workers in 0 0 minimum the richest quintile (figure 3.9b). Thus, 1000 wages2000 do not seem 3000 4000 to be achieving the 5000 goal Q1 Q2 Q3 Q4 Q5 of raising earnings for the majority of vulnerable 2019workers earnings monthlyin in 1,000 Colombia. current High pesos wages minimum 2008 2019 Informaljob could harm job creation or could reallocate to thesalaried creationFormal salaried informal sector (Acar, Informal Bossavie, self−employed and Makovec 2019; Del Carpio et al. 2015; Pérez 2020; Comola and De Mello 2011). Given that by construction, the minimum wage tends to be binding on poorer workers, they would be Source: Author’s estimates, based on GEIH 2008–2019. Note: a) Solid lines show the distribution of earnings for informal and formal workers, more exposed to any negative impacts than richer workers. To foster wage growth, improving according to the legal definition. The vertical dashed line shows the minimum wage. b) labor productivity seems like a more effective strategy. Each bar shows the share of workers with earnings lower than the minimum wage. FIGURE 3.10. Social Security Contribution Rate and Labor Non-wage labor costs are high in Colombia and may introduce distortions in the labor Regulations as Constraints for Firms market. Non-wage labor costs include social security contributions, bonus pay (aguinal- do), annual leave, vacations, severance pay, and firing notices. Relative to earnings levels in the 50 informal sector, Colombia has one of the highest levels of non-wage labor costs in LAC, second only to Uruguay (Alaimo et al. 2017). Most of these non-wage labor costs are due to mandatory SSC contribution rate, % 40 social security contributions, which are very high in Colombia. Among 181 countries, Colom- COL 30 bia ranks 34th in this regard. Moreover, this parameter is positively correlated with the share of firms claiming that labor regulations are a major constraint to growth (figure 3.10). Mandatory 20 social security contributions increase the minimum costs of hiring vulnerable workers dispro- 10 portionately. In other words, even though Colombia is labor abundant and has low wages, regulations substantially increase the costs of hiring formally, particularly when focusing on 0 vulnerable workers with low earnings in the informal sector. Empirical evidence for several 0 10 20 30 40 50 countries—including Colombia—shows that reducing the costs of hiring formally can foster Labor regulations as a major constraint job creation and formality, particularly for low-skilled labor (Kugler and Kugler 2009; Saez, Schoefer, and Seim 2019; Fernández and Villar 2017) Source: Author’s elaboration, based on data from the Social Security Administration for circa 2019 (https://www.ssa.gov/policy/docs/progdesc/index.html) and World Bank Enterprise Survey (latest available year). Labor market rigidities were important before the COVID-19 crisis and will also help shape Note: The horizontal axis shows the share of firms identifying labor regulations as a major the recovery. The previously mentioned association between the pandemic’s impact on job constraint to doing business. The vertical axis shows the mandatory social security contri- bution rate (both for employees and employers). loss and informality was likely driven by several issues, including the fact that: formal jobs were more likely to be protected through government interventions and often involved writ- ten contracts that were costly to terminate, and workers in these jobs tend to be more skilled and thereby more amenable to teleworking. Uncertainty regarding the recovery path, coupled with the high cost and rigidity of hiring formal workers, may hinder job creation and relegate employment growth to the informal sector. For example, evidence for the United States right after the financial crisis of 2008–09 indicate that the states that increased their minimum wag- es experienced a slower recovery in employment and earnings (Clemens and Wither 2019). 35 Unequal access to new technologies BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM B I A | M ak ing the Colombian Labor M ark et M ore Inclusive Technology and the changing nature of jobs Colombia has one of the largest disparities in the world in technology use across socio- economic groups. Among 141 countries, Colombia ranks 109th in terms of the size of the gap in internet use between the richest top 60 percent and the bottom 40 percent of the popula- tion (figure 3.11a). Although 73 percent of people in the top 60 percent use the internet, that figure is only 53 percent among those in the bottom 40 percent. The gap is also sizable across urban and rural areas (figure 3.11b). This holds both for internet use at home and at work. In 2018, while 23.7 percent of people used internet at work in urban areas, that figure was only 4.6 percent in rural areas (DANE 2017). There is also an important rural-urban gap when look- ing at more sophisticated uses of the internet. For example, while about 13 percent of people in urban areas use the internet for e-commerce or e-banking, that fraction is only about 2 percent in rural areas. Gender gaps also exist. Although men and women use the internet at home at roughly the same rates, only 20.5 percent of women use it at work compared to 27.3 percent of men. FIGURE 3.11. Internet Use Gap by Income and Area vs. GDP per capita, 2019 A Gap by income vs. GDP per capita B Gap by urban/rural area vs. GDP per capita 0 10 Internet access gap by income, % Internet access gap by area, % 0 −10 −10 −20 Colombia Colombia −20 −30 −30 −40 −40 7 8 9 10 11 12 7 8 9 10 11 12 GDP per capita PPP (log) GDP per capita PPP (log) Source: Author’s elaboration, based on Gallup and World Development Indicator (WDI) data. Notes: In the vertical axis, each dot shows the difference (in percentage points) in the share of internet users by (a) income (bottom 40 vs. top 60 percent) and (b) area. The horizontal axis shows the logarithm of GDP per capita in purchasing power parity. Access to the internet and other digital technologies can have a host of benefits that many in Colombia are not fully reaping. The internet lowers information barriers, allows workers to better connect with jobs, fosters the creation of new occupations and more flexible work arrangements, and facilitates access to markets and finance. Evidence from developing economies confirms these claims. In Nigeria, mobile broadband coverage reduced extreme FIGURE 3.12. The Task Content of Jobs, 2011–19 poverty by 4.3 percentage points. This was driven by greater labor force participation and em- 0,100 ployment, particularly among women (Bahia et al. 2020). Access to the internet also reduced gender gaps and the prevalence of traditional gender norms in Jordan (Viollaz and Winkler 0,080 2020). Narrowing the large access gaps in Colombia, particularly across areas and income lev- els, could have important equity implications by allowing for more and better jobs among 0,060 disadvantaged groups. Routine Cognitive Index (2011=0.0) 0,040 In contrast to global trends, job growth in Colombia has not been disproportionately con- centrated in occupations that complement technology adoption. Across countries at very Non-routine analytical 0,020 Routine manual different levels of economic development, a clear trend has emerged over recent decades: jobs that are intensive in the tasks that complement new technologies are growing both in Non-routine manual 0,000 number and wage levels, while jobs that are intensive in tasks that can be automated are 2011 2012 2013 2014 2015 2016 2017 2018 2019 Non-routine interpersonal shrinking and their wage levels declining (Lo Bello, Sanchez-Puerta, and Winkler 2019). Jobs -0,020 that complement new technologies include those intensive in analytical and interpersonal tasks, such as managers, teachers, and engineers (table A3.1 in Annex 3 provides a more de- -0,040 tailed description of these tasks). In contrast, jobs that face a greater risk of automation and NR Analytical NR Interp NR Manual R Cognitive R Manual displacement are those intensive in routine tasks, such as machine operators, bookkeepers, and cashiers. In Colombia, the type of jobs that grew the most are those intensive in routine Source: Author’s elaboration, based on data from GEIH 2011–2019. cognitive tasks, that is, one of the groups that face a higher risk of displacement (figure 3.12). Note: Each line shows an index equal to zero in 2011. Magnitudes should be interpreted as the number of standard deviations from the mean value in 2011. NR Analytical: Non-rou- Accordingly, the share of jobs intensive in routine manual tasks also increased. In contrast, the tine analytical; NR Interp: Non-routine interpersonal; NR Manual: Non-routine manual; R jobs that are intensive in non-routine interpersonal tasks that would benefit from increasing Cognitive: Routine cognitive; R Manual: Routine Manual. Table A3.1 in Annex 3 provides a description of these tasks. technology adoption declined. The jobs of the future, intensive in technology use, tend to be held by those who are al- FIGURE 3.13. Gaps in the Share of Workers in Occupations ready better off, while those holding traditional jobs are not only poorer but also face Intensive in the Tasks of the Future worse prospects. Evidence from both developed and developing countries indicates that the an us ed 20 irm el no ill al m adoption of new technologies leads to job destruction and lower earnings among low-skilled zu sk lf rm ge tto RP l al ne ra w fo di Sm NA Bo Ru Lo Ve In In workers but benefits high-skilled ones (Acemoglu and Restrepo 2017; Artuc, Christiaensen, and 0 -5 -1 Winkler 2019). This is because people holding jobs intensive in analytical and interpersonal -4 -3 -4 -3 -3 -10 -8 tasks that facilitate the adoption of new technologies tend to have higher levels of education. -10 percentage points -15 In Colombia, the profile of workers holding the jobs of the future includes urban and highly -16 -15 -15 -20 -17 educated people working in large firms or the public sector. In contrast, low-skilled rural work- -25 -25 ers in the informal economy are lagging. For example, although 57 percent of workers with 13 -30 -28 -28 -35 years of education or more have a job highly intensive in analytical tasks, that fraction is only -40 13 percent among those with less than eight years of education. As seen in figure 3.13, gaps -45 -44 in educational attainment are highly linked to gaps in the access to occupations intensive in -50 technology use. Venezuelan migrants, as well as the indigenous and NARP populations, also Non-routine analytical Non-routine interpersonal have jobs less intensive in technology use. Source: Author’s elaboration, based on data from GEIH 2019. Tasks embedded in jobs account for 30 percent of earnings inequality in Colombia. This is Note: Each bar shows the gaps in the share of workers in occupations in the top quartile of task intensity. Low-skilled: gap between those with less than eight years of education and partly because workers with jobs intensive in analytical tasks have earnings about 30 percent those with 13 years or more; Rural: gap between rural and urban workers; Small firm: gap higher than the rest of the working population. In fact, the distribution of tasks and their as- between workers in small and large firms in the private sector; Informal: gap between informal and formal workers (legal definition); Bottom 20: gap between workers in the sociated earnings premium accounted for more than 30 percent of the total variance of hourly poorest and richest quintile of household income per capita; Indigenous, NARP and Vene- earnings—a proxy for earnings inequality—in 2019 (figure 3.14a and table A3.2 in Annex 3). zuelan: gap between each group and the complement (i.e., non-indigenous, non-NARP, and non-Venezuelan, respectively). This fraction is almost the same as the fraction of inequality that stems from the educational attainment of workers. The role of tasks embedded in jobs becomes even more relevant when analyzing inequality trends. Even though earnings inequality has experienced a significant re- FIGURE 3.14 Decomposition of Hourly Earnings Inequality duction since 2011 (about 3.4 Gini points), changes in the distribution of tasks and their earn- A Decomposition of hourly earnings variance, 2019 B Decom ings premia accounted for an increase in the Gini coefficient of almost 1 point (figure 3.14b and table A3.3). In other words, even though the Colombian labor market has become more 10,0 equitable on the surface, underlying occupational changes related to technology use have Changes in Gini Coefficient (points, 0-100 scale) pushed to make it more unequal. 6,9 14,3 5,0 In a globalized world, technology adoption abroad can still have adverse effects on work- ers with skills that can be automated. For example, job reshoring has raised concerns about 0,0 the traditional role of the tradable sector as an engine of job creation in developing countries 16,1 30,5 (Hallward-Driemeier and Nayyar 2017). In particular, industrial robot adoption and automation in general in rich countries may challenge developing countries’ comparative advantage in labor -5,0 abundance. In other words, it may be more efficient for some companies in rich countries to close down labor-intensive production processes in developing economies and instead rely on capital -10,0 intensive processes at home. The type of jobs most vulnerable to this kind of displacement are those that abound in Colombia, meaning those that are intensive in routine tasks. In fact, there is 32,3 evidence that Colombia is already experiencing some of these negative impacts. The data show -15,0 substantial job destruction and the declining earnings of Colombian workers in sectors that have high levels of automation—measured by robots per worker—in the United States (Kugler et al. 2020). As a result, from 2011 to 2016, U.S. robots destroyed between 63,000 and 100,000 jobs Age and gender Tasks Education Sector Region nd ea back to3.14 in Colombia by reshoring themFIGURE Decomposition of Hourly Earnings Inequality Ag the United States. Furthermore, this process affects vul- nerable workers the most, as the A Decomposition of hourly earnings negative displacement variance, effects 2019 of these robots are greater for B Decomposition of changes in hourly wage Gini, 2011-2019 women, older workers, and those employed in small and medium-sized enterprises. Source: Au 10,0 Note: (a) S full set of r 6,4 Changes in Gini Coefficient (points, 0-100 scale) of changes 6,9 include fir 14,3 5,0 be found i 2,0 1,0 0,2 0,0 -0,2 16,1 30,5 -0,7 -3,4 -5,0 -10,0 32,3 -12,1 -15,0 s n n s ) e r r ed de to sk tic ng tio io c Ta at in n ris ha ca Se ge la uc te lc Lo Age and gender Tasks Education Sector Region p nd Ed ac ex ta ea ar To un ch Ag r( b he Jo Ot Source: Author’s elaboration, based on data from GEIH 2011–2019. Note: (a) Shapley decomposition of the variance of the logarithm of hourly earnings (the full set of results can be found in table A3.2 in Annex 3). (b) RIF-regression decomposition of changes in the Gini coefficient of the logarithm of hourly earnings. Job characteristics include firm size, formal status, and salaried/self-employed job. The full set of results can be found in table A3.3. 36 COVID-19 and the acceleration of labor market trends BUI LD I N G A N E Q UI TA BLE SO CI E TY I N COLOM BI A | M ak ing the Colombian Labor M ark et M ore I nclusive Technology adoption and the associated changes in the labor market that have taken place over several decades experienced a sudden acceleration during COVID-19. Having access to a job intensive in technology use was associated with higher earnings and better benefits before the pandemic. Ever since lockdown measures were implemented around the world, those who had access to such jobs were typically among the few who were still able to keep them. In other words, the labor market returns to technology use were sudden- ly magnified. The divide between jobs that can be done from home and those that cannot will tend to increase inequality everywhere but likely more in Colombia. The share of workers who can work from home in Colombia is about 18 percent, a figure similar to that of other countries at the same level of development (figure 3.15a).6 However, the unequalizing impact of working from home may be more of a concern in Colombia, where inequality is already very high (fig- ure 3.15b). Even before the COVID-19 pandemic, having a job amenable to working from home was associated with a significant earnings premium even if people did not actually work from home. In 2019, those holding jobs whose duties were friendly to working remotely and at the same time had an internet connection at home earned hourly wages 54 percent higher than other workers with similar levels of education, experience, and other characteristics. Accord- ingly, the possibility of working from home and the associated wage premium accounted for about 7.5 of overall wage inequality in 2019 in Colombia (table A3.2, column 6, in Annex 3). Moreover, changes in these job characteristics tended to push wage inequality upwards be- tween 2011 and 2019 by more than one point of the Gini coefficient (table A3.3, column 8).7 FIGURE 3.15 Percentage of Jobs that Can be Done from Home A By level of development B By inequality level 60 60 Jobs that can be done from home, % Jobs that can be done from home, % X LU LUX SWE SWE GBR DNK ISR NOR NOR DNK GBR ISR 40 NLD FIN BEL CHE 40 BEL NLD CHE SVN MLT FIN MLT EST FRA EST FRA SVN ITA AUT IRL IRL PRT AUT DEU ITA CYP DEU CYP PRT LVA LVA ESP MNE HRV ESP HRV MNE RUS POL POL HUN ARE HUN ARG SVK RUS URY URY SVK GRC ARG BGR GRC MNG BRA CHL BGR 20 PSE SRB MUS TUR 20 MNG MUS SRB TUR CHL PAN BRA PHL COL ROU PAN PSE ROU PHL COL BIH MEX EGY DOM MEX EGY DOM LAO BOL THA THA ALB BOL HND SLV LKA ALB LKA GHA SLV CIV HND UGA NPL KHM GTM TJK GHA TJK CIV UGA LBR BFA GMB MMR BGD MMR GMB LBR ZWE RWA ZWE BGD PAK PAK ETH RWA MDG 0 ETH 0 7 8 9 10 11 12 0 20 40 60 80 GDP per capita PPP (log) Inequality ranking Source: Author’s elaboration, based on data from GEIH 2019 and WDI. Note: The share of jobs that can be done from home was calculated following Garrote Sanchez et al. (2020). The inequality ranking was calculated using the average Gini index from Povcal- Net for the 2015–18 period. The inability to work from home due to the characteristics of the job or the lack of con- FIGURE 3.16. Jobs Amenable to Working from Home, nectivity may exacerbate the exclusion of vulnerable groups. Although more than a third by workers’ characteristics of workers with high levels of education have jobs amenable to teleworking, only one out of 10 40 workers with low levels of education is in the same situation (figure 3.16). Rural and informal 35 35 workers and those in small firms also have a lower incidence of working-from-home-friend- 30 28 percentage points ly jobs. Minority groups, such as the indigenous, NARP, and Venezuelan nationals, also have 25 24 jobs less amenable to teleworking. Not surprisingly, these jobs are disproportionately held by 19 21 20 20 18 workers in the richest income quintiles. 15 16 1514 1515 15 14 14 13 13 11 12 11 11 1011 Although jobs’ amenability to working from home are very similar across genders, there 10 9 7 are important hidden inequalities. Most analyses have shown that across countries at vary- 5 ing levels of economic development, the probability that a job could be done at home is sim- 0 ilar for men and women, or even higher for the latter (Garrote Sanchez et al. 2020; Hatayama, Low skilled Middle skilled High skilled Urban Rural Small Large Formal Informal Q1 Q2 Q3 Q4 Q5 Indigenous NARP Venezuelan migrants Total Colombia Guatemala Honduras Bolivia Dominican Republic Mexico Brazil El Salvador Viollaz, and Winkler 2020). This is the case in Colombia, where the probability that a job could be done from home is 2 percentage points higher for women. However, studies that tracked the early impacts of the crisis find that women lost jobs disproportionately (Adams-Prassl et al. 2020; Cueva, Del Carpio, and Winkler 2021). High-frequency phone surveys conducted in 13 countries in LAC by the World Bank indicate that female workers were 44 percent more likely Source: Author’s elaboration, based on data from GEIH 2019. than male workers to lose their jobs at the onset of the COVID-19 crisis. The difference in job Note: The share of jobs that can be done from home was calculated following Garrote San- chez et al. (2020). losses by gender persisted even as labor markets began to recover (Cucagna and Romero 2021). The fact that women are disproportionately concentrated in sectors intensive in face-to- face interactions—and thereby more vulnerable to social distancing measures—and that they are more likely to be the family caregivers helps to explain these findings. In other words, even if there is gender equality in terms of the percentage of jobs that can be done from home, gen- der disparities in telework would persist if other labor market inequalities are not addressed. 3.3. 37 BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | M ak ing the Colombian Labor M ark et M ore I nclusive Policy Options Minimizing distortionary labor market policies Reduce transaction costs. Reducing the costs of business formalization can bring efficiency gains by eliminating un- necessary transaction costs for small firms while at the same time potentially strength- ening job creation in the formal sector. There are several opportunities in this area. First, while the number of days needed to start a business declined substantially with the creation of CAEs (Centros de Atencion Empresarial/Business Service Centers), important gaps remain in lagging cities and regions. Although in Bogotá it takes only one day to register, in cities without a CAE, such as Mitú or Puerto Carreño, it takes 40 and 38 days, respectively (CONPES 2019). Second, the costs to register a business continue to be high relative to the levels of income per capita. On average, it is more than four times than the levels among OECD countries. Third, although the creation of one-stop shops (OSS) or VUEs (Ventanillas Unica Empresarial) was an important step toward the simplification and centralization of the different procedures need- ed to operate in the formal sector, the VUEs are still geographically concentrated and need to be rolled out on a wider scale. Fourth, implementing a Formulario Unico (Single Form) to facil- itate the affiliation of employees to social insurance and BEPS (Beneficios Económicos Periódi- cos/Periodic Economic Benefits) would also be important. Consider reforms to the minimum wage–setting process. Making minimum wages consistent with actual earnings could help improve labor mar- ket outcomes, particularly for vulnerable workers. As discussed above, policies that inflate labor costs—such as high minimum wages and mandatory social contributions—act as a tax on labor, especially unskilled labor. Setting up a clear minimum wage formula that avoids ex- cessive rigidity in the process and is easy for all stakeholders to understand and apply could make the yearly adjustment more transparent and efficient. Preventing large annual adjust- ments beyond the inflation rate for a number of years could also help bring the minimum wage to a level more consistent with actual wages and productivity. It would also be useful to create a unit of independent experts to advise the Comisión Permanente de Concertación de Políticas Salariales y Laborales (Permanent Commission for the Agreement on Salary and Labor Policies) regarding decisions on the minimum wage through analyses of economic and social impacts. Finally, the automatic indexation of minimum wages to the inflation rate es- tablished in the law may have adverse effects on productivity and labor demand. Allowing for adjustments that consider the current economic context, including productivity and labor market indicators (as opposed to automatic adjustments), could better serve the goal of pro- moting job creation and equity. Reduce non-wage labor costs. Accordingly, it is important to improve the link between mandatory contributions and the benefits of the Social Security Administration. OECD (2019) highlights two strategies to achieve this goal. First, since part of the contributions to health care and family compensation funds are used to finance benefits for which formal workers are not eligible, it is important to broaden the sources of funding to reduce formal sector employee contributions. Second, some of the recreational and commercial activities funded by the family compensation funds (which are not typically paid from employer contributions in advanced countries) could be made optional for employers. The evidence reviewed in this chapter highlights that reducing the costs of including unskilled workers in the formal sector can have large payoffs in terms of making the labor market more inclusive. Even though the benefits of labor market reforms along the lines discussed above have been extensively documented, history shows that implementing these policies is a chal- lenging endeavor in countries around the world. Lessons from various settings point to the importance for governments of promoting an evidence-based dialogue among all relevant stakeholders. Khemani (2017) shows that labor market reforms are more likely to be success- ful if they are supported by extensive technical evidence that is credibly non-ideological and non-partisan on how costly the status quo is and how the reform would bring substantial ben- efits. In the case of Colombia, the evidence regarding past reforms is rigorous and abundant. Improving the communication of these findings could help gain reform momentum. Also, ev- idence suggests that a context such as the post-COVID period might help implement reforms. Different papers (Agnello et al. 2015; Duval, Furceri, and Miethe 2018) show that reforms are more likely to happen in times of crisis. For instance, Lora and Olivera (2004) analyzed the wave of reforms in Latin America in the late 1980s and early 1990s and found that crises that are characterized by falls in real incomes and by negative rates of growth facilitate the adop- tion of reforms. More generally, rethinking current employment protection policies in the context of a more dynamic labor market is necessary. Forces disrupting markets and changing the na- ture of work across developed and developing economies are challenging the prevailing set of industrial-era, employment-based risk-sharing policies. A new World Bank report outlines a new set of policy recommendations on moving on from such policies, which often do more harm than good and are an obstacle to jobs rather than a source of effective protection (Pack- ard et al. 2019). A key component of the recommended reforms is that poverty-prevention and redistribution objectives (vertical redistribution) should be pursued transparently with instru- ments financed from broad-based taxes, whereas statutory contributions should be reserved to finance consumption-smoothing instruments with actuarially fair parameters (horizontal redistribution). At the same time, it is important that any contributions be tied to the individual and not to the job, because the goal should be focused on protecting workers instead of certain jobs. With more effective guaranteed poverty-prevention and consumption-smoothing instru- ments in place, a “flexicurity” approach that helps people manage shocks and that facilitates hiring and dismissal decisions for firms in the private sector is a robust policy response to a changing labor market. It would be important to allow for voluntary savings programs aimed at people who cannot satisfy the requirements of the Social Security Administration, such as the self-employed and seasonal workers. Although Colombia already has an instrument to promote voluntary contributions (BEPS), it is important to foster take up. Successful examples draw on insights from behavioral economics, using simple commitment devices or behavioral nudges (for example, New Zealand’s Kiwi Saver retirement-savings scheme and commitment devices in telephone payment platforms in Kenya that have increased savings). Reduce labor market barriers for vulnerable groups. Eliminating barriers to the labor market inclusion of women can bring important pay- offs. First, social security contributions are based on the full-time or weekly minimum wage, which tends to penalize part-time work. Making contributions proportional to the number of hours worked can facilitate the formalization of part-time workers, which is an important source of income for people who need to balance work with family responsibilities. Second, establishing and monitoring quality and safety standards within childcare centers can foster the demand for these services and thereby the employment of women, who are more likely to be the family caregivers. Third, matching the retirement age across genders would contribute to improving gender equality so that men and women are entitled to the same level of bene- fits upon retirement. Addressing the particular needs of groups facing exclusion from the labor market would be a good step. First, since spending on active labor market programs (ALMPs) is low in Co- lombia (OECD 2019), it would be important to evaluate an increase in this regard. However, this should be done by carefully evaluating existing programs in Colombia, as well as past experiences from other countries. The ALMPs should be cost efficient while minimizing the distortions and displacement effects of non-participants. Second, the Servicio Nacional de Aprendizaje (SENA)/National Training Service and the Servicio Publico de Empelo (SPE)/Public Employment Service could tailor their services better to the specific needs of the NARP, mi- grant, and displaced populations, for example, by better understanding these groups’ skills profiles, launching campaigns to promote the use of SENA and SPE services among minority groups, and strengthening their institutional presence in geographic regions with larger mi- nority populations. 38 Adapting to the future of work: global value chains and technology trends BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | M ak ing the Colombian Labor M ark et M ore I nclusive Reducing inequities in the labor market requires not only addressing present challenges, but also getting ready for the challenges of tomorrow. This involves, for example, removing barriers to global value chain (GVC) integration and to the adoption of new technologies by people and firms. Improving Colombia’s integration into GVCs requires policy reform in a number of areas, including changes in tariff schedules to reduce dispersion across industries, enhanced regulation in freight and transport markets, and rules of engagement that make opera- tions by multinational enterprises more transparent and stable. A major impediment to sufficient fiscal space for transfers to the affected workers and households during the COVID-19 crisis has been limited access to foreign exchange as exports have effectively collapsed with the drop in oil prices. A further disincentive for GVC participation has been the multiple tariff and nontariff barriers for import entry. Trade policy can become instrumental as a tool for re- covery in the aftermath of the COVID-19 pandemic. Greater competition in core sectors, such as logistics and freight transportation, can be important in facilitating Colombia’s insertion into GVCs (World Bank 2020e) through the modernization of maritime port and airport infra- structure, which could also be improved by removing the barriers to entry for ancillary service provision. In fact, intra-national freight costs could drop considerably with both more compe- tition and road-network modernization in a way that enables Colombian firms to engage with GVCs operating with just-in-time inventory strategies to cover international markets. Lower transportation costs from factory to port (and vice-versa) are key, as is better transport infra- structure to export and import within GVCs. Further support for entrepreneurship and greater access to finance will also be fundamental for investment in innovative projects to accelerate new firm entry and in technology upgrading by existing firms that can supply components to leading edge GVCs.8 Foreign direct investment (FDI) has a key role to play through both the setting up of subsidiaries and technology spillovers. Therefore, the regulation of FDI should be streamlined in a way that stabilizes net inflows. These policies could promote investment as well as exports as drivers of innovation and technological change. At the same time, there is room for a reduction of non-tariff impediments for multinationals as well as limits on profit repatriation.9 Mitigating the impacts of technology on labor market inequality requires both a demand and supply response. In Colombia, barriers to technology adoption prevent the integration of certain socioeconomic groups into the labor market. However, there is another angle to this argument. In particular, technology adoption can also push inequality upwards by increasing the demand for (already relatively scarce) skilled workers. However, the magnitude of this in- equality impact depends greatly on the possibility of supporting an increase in the supply of skilled workers. If rapid technology adoption is accompanied by an increase in the number of skilled workers, the inequality impacts would likely be smaller. FIGURE 3.17. Fixed Internet Penetration among Municipalities Fostering technology adoption among the poor and in rural areas would contribute to Targeted by the Proyecto Nacional de Fibra Optica promoting equity in the labor market. More competition in the telecommunications sector can contribute to reducing costs and expanding access. Meltzer and Pérez Marulanda (2016) .025 propose several recommendations in this area. First, they point out the importance of em- .02 powering the Superintendencia de Industria y Comercio (SIC)/Superintendency of Industry and Subscriptions/Population Commerce to enforce competition laws and strengthening its capacity to carry out competition .015 analysis and address regulatory overlap with the Comision de Regulacion de Comunicaciones (CRC)/Communications Regulation Commission. Second, campaigns should be carried to ad- .01 dress the resistance of several municipalities and local communities—due to misinformation about health effects—to the installation of antennas needed to expand service. Third, take-up .005 among small and micro firms could be promoted, for example, by using FONTIC (Fondo para las 0 Tecnologias de Informacion y las Comunicaciones) and implementing information campaigns 2010 2011 2012 2013 2014 2015 2016 2017 2018 about the benefits to productivity of using the internet. In addition, the role of the government Rollout on 2013 Rollout on 2014 in addressing the public good nature of digital infrastructure cannot be overemphasized. For Rollout on 2015 Rollout on 2016 example, the Proyecto Nacional de Fibra Optica (National Fiber Optic Project) had the goal of extending the fiber optic network to almost 800 municipalities across the national territory. Source: Viollaz and Winkler (forthcoming). As seen in figure 3.17, all eligible municipalities had rather similar levels of fixed access to the Note: Each line shows the average number of fixed internet access subscriptions relative to population size across municipalities targeted by the Proyecto Nacional de Fibra internet before 2013, that is, before they gained fiber-optic access. However, municipalities Optica. that benefited earlier displayed a much more rapid growth in subscriptions. By 2018, munici- palities that gained access in 2013 had penetration levels about twice as high on average than those who were covered in 2016. However, even among the former, the average penetration rate remains low at about 2 percent of the population, which is driven mostly by small locali- ties. This highlights the fact that improving the supply of technology is not a sufficient condi- tion to foster take up, and demand-side policies are important as well. Improving the skills of the labor force would facilitate the adoption of technology and mitigate the impacts of technological change in the labor market. Evidence for Colombia indicates that improving technical and socio-emotional skills can have significant positive im- pacts on formal sector employment, but that there are important resource constraints among people who consider vocational training (Barrera-Osorio, Kugler, and Silliman 2020). Chapter 2 of this report provides specific policy recommendations on improving the quality of Colom- bia’s human capital. In summary, these policies involve a comprehensive approach through- out the life cycle. In addition to the policy recommendations related to labor supply and demand as well as to technology adoption outlined above, it would be key to implement a policy frame- work that promotes telecommuting and “telecommutable” jobs in an equitable fashion. Before the pandemic, telecommuting was not only rare but also widely unknown in the pri- vate sector. In 2017, a survey indicated that 42 percent of firms in Colombia did not know what telecommuting meant, and only 9.5 percent had a telecommuting program for staff (DANE 2017). Although there are laws and regulations about telework in Colombia, they may need to adapt to the new normal during and after the pandemic.10 However, although promoting the rise of teleworking and “teleworkable” jobs is key for Colombia to adjust to the future of work, in a country with such a large informal sector it would be important to focus on preventing an exacerbation of the existing formal-informal segmentation. As outlined above, labor market regulations in Colombia add significant distortions that tend to raise the cost of labor—par- ticularly low-skilled labor—disproportionately. Any new legal provisions related to telework- ing should be designed in a way that does not further exclude low-skilled workers from the formal sector. This could be achieved by making sure that new regulations do not increase the costs for firms to hire workers remotely as opposed to on the premises. At the same time, it is important to continue the progress in reducing gender gaps in other dimensions so that they do not affect the access to teleworkable occupations, for example, by fostering access to affordable childcare. Policy Options for More 39 BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | M ak ing the Colombian Labor M ark et M ore I nclusive Equitable Labor Markets Observed Drivers of the Policy Options Relevant International Timing for Consideration about and Equity Gap Equity Gap Experiences Implementation Estimates of the Fiscal (Immediate/Short Term, Impact of Implementation OR Medium/ Long Term) Regulation-driven barri- High transaction Reduce formalization costs, including by:11 Mexico has successfully improved the efficiency Medium term The direct fiscal costs of fa- ers to access formal sector costs of formaliza- of opening and expanding businesses by reducing cilitating formalization pro- jobs among the poor and tion Reducing the number of days it takes to the number of days required to register a business cedures are expected to be vulnerable register a firm and by creating OSS in the most populous cities. moderate. In Azerbaijan, the Minister of Taxes made the OSS Reducing the costs of firm registration operational within two months, as specified in a (“registro mercantil”) presidential decree. Fundamental to the project was coordination with other related state insti- Expanding the “Ventanilla Unica” (VUEs) tutions. For this reason, bilateral interministerial across the country agreements setting up a clear framework of mutual responsibilities and cooperation were signed be- Developing “Formulario Unico” to facilitate tween all involved entities. Business registration the registration of employees increased by over 30 percent in the first year of OSS operation.12 High wage labor Limit minimum wage growth to the infla- Several countries have hourly based minimum wag- Medium to long term The direct fiscal costs of ad- costs tion rate for some years (until the mini- es (e.g., Australia, Canada, United States, etc.).13 In justing the minimum wage mum wage reaches a level friendlier to job Korea, the system for adjustment is set up so that are expected to be moderate. creation) and establish an hourly minimum rate level changes are based on technical evidence wage (so that part-time employment is not (such as living costs, effects of minimum wage pol- penalized). icies, minimum wage systems in other countries, implementation visits, etc.) rather than political Create a unit of independent experts to grounds.14 advise the Comisión Permanente de Con- certación de Políticas Salariales y Laborales regarding decisions on the minimum wage through analyses of economic and social impacts. Consider reforms to allow for the mini- mum wage to be adjusted according to the economic context (for example, taking into account productivity and labor market in- dicators) and not automatically adjusted to inflation as in the current legal setting. High non-wage la- Improve the link between mandatory Using general revenues to support social insurance Medium to long term Changes to the social security bor costs contributions and benefits. OECD (2019) systems: in Chile, though the contributory part of system can have large fiscal highlights two strategies to achieve this the Chilean pension system is fully funded by work- and equity implications in goal. First, since part of the contributions ers’ contributions, the government finances the the short, medium, and long to health care and family compensation Basic Solidarity Pension for persons who did not terms. More analysis is need- funds is used to finance benefits for which contribute, and the Pension Solidarity Complement ed to assess the fiscal and formal workers are not eligible, it is im- for persons whose contributions are below a spe- distributional impacts of dif- portant to broaden the sources of funding cific threshold. The government also guarantees ferent scenarios over time. in order to reduce formal sector employee pension benefits if a pension fund is shut down or contributions. Second, some of the recre- becomes insolvent.15 ational and commercial activities funded by the family compensation funds (which are not typically paid from employer con- tributions in advanced countries) could be made optional for employers. Labor market barriers Gender-biased Make social security contributions propor- About hourly based social security contributions, Medium/ More work is needed to quan- faced by specific demo- laws and regula- tional to the hours worked (instead of on in the Netherlands, every hour worked is counted Long term tify these costs and benefits. graphic groups tions a full-time or weekly basis), which would toward social insurance contributions. Austria and promote formalization among female Germany extended coverage to workers who were workers. previously uncovered by social insurance, such as marginal part-time workers.18 Establish and monitor quality and safety standards within childcare centers. About childcare, quality rating and improvement systems (QRIS) are used in several countries, with Equalize the age at which men and women a focus on staff qualifications, group size, curricu- can retire with full pension benefits across lum, and child outcomes. In the United States, QRIS sectors (as in Slovenia).16 are defined at the state level.19 In Mexico, the 2011 Childcare Law (Ley de Guarderias) integrates poli- Introduce legislation prohibiting gen- cies and programs and improves their compatibility der-based discrimination in access to and coherence in the medium and long terms.20 credit (as in Uzbekistan, Bahrain, Jordan, Maldives).17 Low spending on Increase spending on ALMPs by carefully Some countries in LAC added an ALMP component Medium/ More work is needed to quan- ALMPs that work21 evaluating existing programs in Colombia, (such as training) to conditional cash transfer pro- Long term tify these costs and benefits. as well as past experiences from other grams. For example, the food purchase grants com- countries. These should be cost efficient, ponent in the Panamanian Opportunities Network while minimizing distortions and displace- provides transfers in kind to support the agricul- ment effects on non-participants.22 tural training courses included among the families’ conditionalities.23 Weak targeting The Servicio Nacional de Aprendizaje In Spain, the ACCEDER program provides individu- Medium term More work is needed to quan- of labor market (SENA) and the Servicio Publico de Empleo ally tailored employment services to Roma at the tify these costs and benefits. interventions to- (SPE) could tailor their services better to municipal level. Case studies in Greece, France, and ward minorities the specific needs of the NARP, migrant, Hungary show that training and other activities and displaced populations. to promote labor market integration can help mi- cro-enterprises and support Roma entrepreneurs.24 Conclusions Making the Colombian labor market more inclusive will require policies to reduce the distortions associated with certain regulations, to facilitate the inclusion of vulnerable groups, and to prepare both firms and workers for the future of work. Reducing the trans- action costs of formalization, adjusting the minimum wage, and adapting the social security system to a changing world of work would help low-wage workers disproportionately, thereby fostering equity. Policies to facilitate the integration of women, low-skilled workers, and mi- nority populations would help reduce inequality as well. 40 Endnotes BUI LD I N G AN EQ U ITA BLE S O CI ETY I N CO LO MBI A | M ak ing the Colombian Labor M ark et M ore Inclusive 1 This chapter focuses on the equity dimensions of the labor market. A forthcoming 14 Del Carpio and Pabon (2014). More details can be found here: https://www.mini- Jobs Diagnostic report for Colombia contains a general overview of recent trends in mumwage.go.kr/eng/sub04.html. the labor market from the macro, supply, and demand sides (see box 3.1). 15 H. Winkler, E. Ruppert Bulmer, and H. Mote, “Expanding Social Insurance Coverage 2 This acronym refers to various groups of Afro-Colombians (in Spanish, población to Informal Workers” (World Bank: Washington DC, 2017). Negra, Afrocolombiana, Raizal y Palenquera). 16 World Bank (2021). World Bank. 2021. Women, Business and the Law 2021. Wash- 3 It is important to mention that total factor productivity also grew between 2000 and ington, DC: World Bank. doi:10.1596/978-1-4648-1652-9. License: Creative Commons 2019 but has declined since 2014 (see https://fred.stlouisfed.org/series/RTFPNA- Attribution CC BY 3.0 IGO COA632NRUG; and Feenstra, Inklaar, and Timmer 2015). 17 World Bank (2021). World Bank. 2021. Women, Business and the Law 2021. Wash- 4 These figures come from the ordinary least squares (OLS) regression lines plotted in ington, DC: World Bank. doi:10.1596/978-1-4648-1652-9. License: Creative Commons figure 3.6. Attribution CC BY 3.0 IGO 5 See World Bank, “Enterprise Surveys: Colombia,” https://www.enterprisesurveys. 18 C. Behrendt and Q. Nguyen, “Innovative Approaches for Ensuring Universal Social org/en/data/exploreeconomies/2017/colombia#biggest-obstacle. Protection for the Future of Work” (Geneva: International Labour Organization, 6 Using a different methodology, Cárdenas and Montana (2020) also find that about a 2018). fifth of workers in Colombia could work from home. 19 OECD, “Data and Monitoring to Improve Quality in Early Childhood Education and 7 Calculated as the sum of the explained and unexplained contribution of working Care,” in Engaging Young Children: Lessons from Research about Quality in Early from home amenability to changes in the Gini coefficient. Childhood Education and Care (Paris: OECD Publishing, 2018), 95–103. 8 See Neusser and Kugler (1998) for evidence of the influence of access to finance as a 20 M. Mateo Díaz and L. Rodriguez-Chamussy, Cashing in on Education: Women, Child- key determinant of productivity growth. care, and Prosperity in Latin America and the Caribbean, Latin American Develop- 9 Extract from Del Carpio et al. (2020). ment Forum (Washington, DC: World Bank, 2016). 10 Law 1221 from 2008 defines the legal framework for telecommuting, while Decreto 21 OECD (2019), chapter 2. 884 outlines the implementation. 22 Ibid. 11 CONPES (2019). 23 S. Cecchini and A. Madariaga, Conditional Cash Transfer Programmes: The Recent 12 International Finance Corporation, “Reforming Business Registration: A Toolkit for Experience in Latin America and the Caribbean (New York: United Nations, 2011). the Practitioners” (Washington, DC: World Bank, 2013), https://openknowledge. 24 R. Tulumovic, “Potentials for Roma Employment in the Enlargement Region,” Re- worldbank.org/bitstream/handle/10986/17634/840140WP0Box380usiness0Registra- gional Cooperation Council (Belgrade: Roma Integration 2020 Action Team, 2018), tion.pdf?sequence=1&isAllowed=y. https://www.rcc.int/romaintegration2020/files/admin/docs/456df932ca6433b- 13 A. Adema and others, “Minimum Wages Across Countries,” ifo DICE Report 16, no. 78cfb328d31d76035.pdf. 4 (2018): 55–63. See also R. Dickens, “How are Minimum Wages Set?: IZA World of Labor, November 2015.. BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LOMBI A 41 CHAPTER 4. Public Finance and Equity in Colombia Summary 42 B U IL DIN G A N EQ UI TABL E SO CI ETY I N CO LO MBI A | P ublic Finance and Equity in Colombia Fiscal policy is a critical tool in reducing inequality. However, in Colombia, fiscal policy has had little impact on this important objective. First, the level of tax collection is low, and with few resources there is not much that can be redistributed. Second, Colombia collects little revenue from progressive taxes like the personal income tax. This is because deductions are so high that only very few individuals pay personal income taxes. Third, the foregone reve- nue from value added tax (VAT) exemptions and reduced tax rates accrues mostly to high-in- come individuals. Finally, although the targeting of cash transfers and utility subsidies makes it possible to cover the most vulnerable, it also produces inefficient leakages to high-income earners. On top of these factors, the pension system perpetuates the income inequalities ac- cumulated during the working life of individuals. Reforms could therefore focus on correcting these distortions and specifically aim at: 1. expanding the reach of the personal income tax from a fraction of the top decile of the income distribution to the top deciles, up to 20 percent in the short term and 50 percent in the long term; 2. eliminating regressive VAT exemptions and reduced VAT rates ; 3. compensating the most vulnerable for the increase in the amount of VAT that they would end up paying on their consumption; 4. improving the targeting of cash transfer programs and utility subsidies . 4.1. 43 BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | P ublic F inance and Equity in Colombia Introduction Fiscal policy is an important tool in correcting income inequalities. Conceptually, there are three stages at which fiscal policy affects income inequality (Lustig 2018). At the first stage, through direct taxes (on income and assets) and transfers, fiscal policy modifies the income that individuals receive from participating in the markets (market income, the sum of labor income, the yields from fiscal assets, and private transfers). In this first stage, fiscal policy de- termines disposable income, that is, the income that individuals can dispose of after paying direct taxes and receiving direct transfers. At a second stage, through indirect taxes (for exam- ple, value added tax [VAT] and taxes on tobacco and liquor) and consumption subsidies (such as water, gas, electricity, and other fuel), fiscal policy affects how much of their disposable in- come individuals can consume (consumable income). Finally, through the provision of public goods (especially education and health), fiscal policy effectively determines the final income from which individuals can benefit. FIGURE 4.1. Redistribution by Fiscal Policy in Colombia However, in Colombia, fiscal policy redistributes little, especially compared to peer coun- tries, and there are thus ample opportunities to improve fiscal redistribution. Figure 4.1 60 shows the impact of fiscal policy on income inequality along the three stages outlined above. The impact is measured by looking at the reduction in the Gini coefficient of market income, Gini coefficient, percentage points disposable income, consumable income, and final income (Nuñez et al. 2020).1 The figure 55 55,1 shows that fiscal policy reduces inequality only at the last stage, that is, only after spending 51,7 51,5 on health and education is distributed across all individuals. Because it is difficult to mone- 50 tize the individual value of public goods (especially education and health, where the amount that the government spends does not necessarily reflect quality), figure 4.2 compares the way 45,4 that fiscal policy reduces inequality at stages one and two in Colombia and in other countries. 45 The vertical axis measures the reduction in the Gini coefficient from market to disposable income—the redistributive properties of direct taxes and transfers. The horizontal axis mea- sures GDP per capita. When compared to emerging market countries with a similar GDP per 40 Market income Disposable Consumable Final income capita, the redistributive property of fiscal policy in Colombia (the red dot) scores toward the income income bottom. When compared to other countries in Latin America and the Caribbean (LAC), the pic- Confidence interval ture improves, but still, there are countries where fiscal policy redistributes (or at some point was able to redistribute) more than is the case in Colombia (Argentina, Brazil, Costa Rica). Source: Nuñez et al. (2020). When compared to other OECD countries, Colombia’s fiscal policy is definitely far from being redistributive. FIGURE 4.2. Redistributive Properties of Fiscal Policy, by country This chapter will address the following questions: GDP per capita, in PPP current U.S. dollars 1. Which specific factors make Colombia’s public finances less distributive than other coun- 0 20 000 40 000 60 000 80 000 tries? 0 2. How progressive are direct and indirect taxes in Colombia if one accounts for the effect of (from market to disposable income) Percentage points reduction in Gini -5 COL exemptions and deductions? -10 3. What are the redistributive properties of social spending and subsidies? -15 4. How would a different VAT and personal income tax (PIT) system, combined with an ex- pansion of transfers at the bottom of the income distribution, change inequality? -20 The Diagnostics section will start with an aggregate view of revenues and social spending and -25 compare the individual redistributive role of direct taxes and transfers in both Colombia and -30 other countries. Next will come a focus on the progressivity and incidence of PITs by income groups and the same for VAT and VAT exemptions. This will be followed by a look at the distri- OECD LAC Other emerging market countries bution and progressivity of the main cash transfer programs and utility subsidies, with some Source: CEQ Institute, various years, OECD and World Bank. consideration of the equalization properties of the pension system. Finally, the policy section will discuss policy options to improve the redistributive impact of fiscal policy.2 4.2 44 BUI LD I NG A N EQU I TA B L E S OCI ETY I N COLOM B I A | Public F inance and Equity in Colombia Diagnostics: Utilizing Fiscal Policy to Reduce Inequality The system relies more on transfers than on progressive taxes for redistribution FIGURE 4.3. General Government Tax Revenue, average Despite various tax reforms in the past 20 years, Colombia raises fewer taxes than other 2014–18 (percent of GDP) countries in the region. Over the period 2014–18, Colombia’s general government tax revenue 40 (that is, excluding dividends from Ecopetrol and extraction royalties) averaged 19.4 percent of 35 GDP compared to 22.6 percent of GDP (on average) in LAC and 34 percent (on average) in the 30 OECD (figure 4.3). A government that collects fewer taxes has fewer resources to redistribute. Percent of GDP 25 Colombia also collects less from progressive taxes such as the PIT and more from cor- 20 porate taxes than comparator countries. The composition of taxes in Colombia is different 15 from that of other countries. Although Colombia collects as much from VAT as the average 10 country in LAC and the OECD, it collects less from PITs and more from corporate income taxes (CITs) than the average country in LAC and in the OECD. Moreover, since 2000, the VAT and the 5 CIT are the taxes that most contributed to the increase in overall tax collection (figure 4.5a). At 0 Colombia LAC OECD a macroeconomic level, CIT collection corresponds to at least 9.2 percent of corporate profits (in Colombia’s national accounts, wages paid informally are counted along with corporate VAT PIT CIT Green Wealth Other SSC profits), while PIT collection corresponds roughly 3.7 percent of employee compensation (fig- Source: OECD Global Revenue Statistic Database. ure 4.5b). At 32 percent, Colombia has one of the highest CIT rates among OECD countries, which will remain the case even after it is reduced to 30 percent after 2021 (figure 4.5c).3 The wealth tax has undergone various reforms in the past several years. Until 2018, Colombia col- FIGURE 4.4. Composition of General Government Tax lected more from the wealth tax than comparator countries, but the reform of 2019 has signifi- Revenues, average 2014–18 (percent) cantly decreased the contribution from this tax.4 Although green taxes constitute on average 100% a very small fraction of tax collection both in LAC and the OECD, in Colombia collection from 90% these taxes is about half of the average in LAC (0.6 percent of GDP against 1.1 percent of GDP), 80% and a bit more than one-third of the average in the OECD (1.6 percent of GDP). Collection of 70% social security contributions is not entirely comparable, because in Colombia individuals can Percent of Total 60% choose between contributing to a private pension scheme or to the public pension system. 50% The comparison is even starker if one contrasts the contribution of individual taxes with the 40% 30% total. Excluding social security contributions, VAT and CITs are effectively the two pillars of the 20% tax system in Colombia, while among OECD countries, it is VAT and PITs. 10% 0% Colombia LAC OECD FIGURE 4.5a. Increase in revenues and contribution between FIGURE 4.5b. CIT and PIT Collection Relative to Corporate 2000 and 2018 (percent of GDP) Profits and Employee Compensation (percent) VAT* PIT CIT Green Wealth Other SSC 6 12 Source: OECD Global Revenue Statistic Database. 5 10 4 8 3 6 2 4 1 2 0 0 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 PIT CIT VAT Total revenue PIT to compensation of employees CIT to corporate profits 1/ Source: OECD Global Revenue Statistic Database. Source: OECD Global Revenue Statistic Database and DANE. Note: 1. Corporate profits include also mixed income, that is, the estimate of informal wages and profits. FIGURE 4.5c. Corporate Tax Rate in 2020 (percent) 35 30 25 20 15 10 5 0 Ire ry Cz Lit and Re ia Po ic ite lo d ng ia Es m Fi ia Ic d d Re via itz blic Sw nd De den No rk Tu y ey Lu r l m e Au rg Be tria m er e Un S ds d in Ca es re da p. Ze ly Ja d Ge pan Au any M ia Po ico lo l Fr ia ce ae Co uga a xe eec th hil Un S lan an an an bl w Ita h an d ven n l b ite pa Re a do iu a u at an rk rw n ra Ko na la ak Lat ex Isr to m ng nm bo s pu Sw pu rm la Ne C e l nl el al lg St ec hu rt st er a, G Hu Ki Ne ov Sl Source: OECD Corporate Tax Statistics. FIGURE 4.6. General Government Spending, 2017 Spending on social benefits is lower than in the average LAC and OECD country. The defi- (percent of GDP) nition of social benefits adopted for this comparison includes all cash transfers and consump- 45 tion subsidies and all types of pensions. The lower level of social benefits in Colombia partly 40 reflects the coexistence of both a private and a public pension system, especially if compared 35 to the average in OECD countries (figure 4.6). Yet, general government spending on social ben- 30 efits broadly defined is 30 percent less than the average of LAC countries (7.5 against 10.3 Percent of GDP 25 percent of GDP) and less than half the average of OECD countries (16.5 percent of GDP). Subsi- 20 dies are almost the same as in the average LAC country (about 0.4 percent of GDP in 2017, the 15 last comparable data available), but they are half the level of the average OECD country (0.8 10 percent of GDP, see figure 4.6). Notably, because the rest of spending is almost the same in Co- 5 lombia (about 20.2 percent of GDP) as in the average LAC and OECD country (20.4 percent of 0 GDP) and is only slightly below the average in OECD countries (23 percent of GDP), Colombia Colombia LAC OECD effectively devotes a lower share of its spending on social benefits and subsidies (figure 4.7).5 Social benefits Subsidies Rest of spending Reflecting the low collection of PITs, it is mostly through transfers and not taxes that fis- cal policy in Colombia reduces inequalities. Figure 4.8 shows the contribution of the differ- Source: OECD, “Government at a Glance 2019, Country Fact Sheet: Colombia” (Paris: OECD, 2019), https://www.oecd.org/gov/gov-at-a-glance-2019-colombia.pdf; and OECD, ent taxes and transfers to reducing inequality between market and consumable income. VAT, “Government at a Glance: Latin America and the Caribbean 2020” (Paris: OECD, 2020), https://www.oecd-ilibrary.org/sites/13130fbb-en/index.html?itemId=/content/ which is overall a regressive tax, and other indirect taxes increase market income inequality publication/13130fbb-en. slightly by a total of 0.6 of one percentage point. The wealth tax and the PIT, which are sup- posed to be progressive by construction, do reduce inequality but by very little: 0.5 of one FIGURE 4.7. Composition of General Government Spending, percentage point. Other direct taxes (for example, the property tax or the tax on vehicles) do 2017 (percent) a better job at reducing inequality (0.7 of one percentage point). The greatest contribution 100% to reducing income inequality comes from direct cash transfers (conditional or uncondition- 90% al) and consumption subsidies (especially to the consumption of gas, electricity, and water). 80% Combined, these reduce inequality by 2.8 percentage points. To summarize, direct cash trans- 70% fers contribute 2.8 percentage points out of the total 3.4 percentage points of the reduction in Percent of Total 60% the Gini coefficient between market and consumable income. 50% 40% Colombia uses direct taxes and transfers less than other countries to reduce inequality. 30% Figure 4.9 shows how much direct taxes and transfers reduce the Gini coefficient of market 20% income in countries for which these estimates are available. As discussed above, in Colombia, 10% direct taxes and transfers reduce the Gini coefficient by 3.4 percentage points. Other countries 0% show that it is possible to reduce inequality through these means by twice as much (Austria, Colombia LAC OECD Denmark, or Poland) or even nearly four times as much (Belgium). Although these countries Social benefits Subsidies Rest of spending have higher income per capita than Colombia, their success indicates that the tax and transfer system can be designed to achieve greater redistribution (Coady, D’Angelo, and Evans 2019). Source: OECD, “Government at a Glance 2019, Country Fact Sheet: Colombia” (Paris: OECD, 2019), https://www.oecd.org/gov/gov-at-a-glance-2019-colombia.pdf; and OECD, “Government at a Glance: Latin America and the Caribbean 2020” (Paris: OECD, 2020), https://www.oecd-ilibrary.org/sites/13130fbb-en/index.html?itemId=/content/ publication/13130fbb-en. FIGURE 4.8. Contribution of Direct and Indirect Taxes and FIGURE 4.9. Reduction in the Gini Coefficient from Direct Transfers to Reducing Inequality in Colombia Taxes and Transfers (percentage points) 60 MEX ZAF 58 PER Gini coefficient, Percentage points COL 56 BRA GRC 54 CHL ITA 52 NLD 50 GBR ESP 48 IRL PRT 46 HRV FRA 44 LUX SWE 42 AUT POL 40 DNK DEU e T s lth T ire ers tra s s tra e l in rs e s o axe ct axe er om m m VA PI Fi nsfe FIN ea sf Co nsf co co t nc t he ran BEL w n Ta ect ct ti I n . t na ns ke ir ct d 0 2 4 6 8 10 12 14 nd rd in ar re re xe -k ri M di Di Percentage points In he In Ot Ot Source: Nuñez et al. (2020). Source: EUROMOD and CEQ institute (various years). 45 Personal income taxes: high deductions B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Public F inance and Equity in Colombia Owing to generous deductions and a high zero-rate bracket, the PIT has a very narrow base and reaches only those individuals in the top decile of the income distribution. Fig- ure 4.10 shows how much each decile of the income distribution contributes to total PIT col- lection. It uses both estimates from the survey on household budget (DANE, 2018c) as well as administrative information. Effectively, almost the entirety of PIT collection is being contribut- ed by the very top decile of the income distribution. Individuals in deciles nine or eight (whose income is about twice as much as the median income and who can be considered better off than the rest of the population) are not reached. Deductions and the zero-rate income bracket push income above which the PIT kicks in to very high levels (see box 4.1).6 Figure 4.11 shows the income level at which a person starts paying income taxes. To allow for a sensible compar- ison, this income level is expressed in percent of both GDP per capita and median income. Al- though in most advanced economies the PIT reaches even a person whose income is half the median (or half the GDP per capita), in Colombia individuals start paying the PIT only when their income is about four times the median income (or about 3.5 times the GDP per capita). FIGURE 4.10. Contribution to the Total Collection of Personal FIGURE 4.11. Income Level at which Individuals on Average Start Paying the PIT Income Taxes, by decile of income (percent) 120 500 450 100 400 350 Percent of total collection 80 300 Percent 250 60 200 150 40 100 50 20 0 Ca ark Ko da a h Is d th ub l er lic Po nds nd Ic taly M nd Ja ico Lu Slo pan Sw mb nia er rg La nd Sp ia l n ite str m S a rm es rw y ak Au ay pu ia Fr blic Es nce Fi nia ed d Co Ch n lo ile a Ne Rep rae No an re d ali bi Be ai e n Sw an tv Re str Un Au giu itz ou Ge tat na a la a la ex xe ve to m nm a la I l el nl r e 0 I De 1 2 3 4 5 6 7 8 9 10 ec ov Cz Sl Deciles of income (market and pension) Estimated Reported In percent of median income In percent of GDP per capita Source: Nuñez et al. (2020). Source: OECD Revenue Database and World Bank World Development Indicators. Box 4.1. The Personal Income Tax in Colombia The PIT applies to the income that a person receives during the calendar year. In Colombia, it is mandatory to submit a tax return as long as a person meets any of the following requirements: gross equity higher than 4,500 UVT7 within the respective fiscal year, annual gross income higher than 1,400 UVT, credit card purchases or whole purchases during the taxable year that exceed 1,400 UVT, and an accumulated balance on banking accounts, savings, deposits, or financial investments during the taxable year that are higher than 1,400 UVT. In 2020, one UVT corresponded to 35,607 Colombian pesos, the equivalent of about US$10. According to Law 1943 of 2018 and Law 2010 of 2019,8 there are three categories of taxable income (income baskets): 1. General income (includes employment income, non-employment income, and capital income related to financial interests and rental income); 2. Pension income; 3. Dividends. For each type of gross income, the law defines sources or uses of income that can be subtracted from gross income to derive liquid income. The most common example is mandatory contributions to the health and pension systems. These are excluded from general income and hence are not part of taxable income. Deductions and exemptions are further subtracted from liquid income to derive taxable liquid income. Most deductions and exemptions are defined in percent of liquid income up to a threshold. The most common exemptions include: • voluntary contributions deposited into a Colombian pension fund (25 percent of gross income up to 240 UVT per month); • deposits in AFC accounts (savings accounts to purchase a house) opened in local commercial banks (30 percent of gross income up to 3,800 UVT per year); • deductions for economic dependents (10 percent of gross income up to 32 UVT per month); • payments to private health enterprises, interest and loans for the acquisition of a taxpayer’s house, and payments of the financial transaction tax; • the first 1,000 UVT per month of pension income. It should be noted that exempt income and deductions to the general basket cannot exceed 40 percent of the liquid in- come, or 5,040 UVT. The figure below illustrates schematically how taxable income is derived. Marginal tax rates apply to the sum of the taxable general liquid income and taxable pension income. It is worth noting that a zero marginal tax rate applies to the first income bracket, which is akin to having yet an additional deduction (see table below). Progressive rates between 0 and 15 percent apply to dividends, depending on the year in which the distributing company obtained the profits. Derivation of Taxable Liquid Income Brackets in UVT 68 Monthly income (45 degree line) 30 Tax Tax Calculation Rate 64 Liquid income From To 60 Effective tax rate (r.a.) 56 25 0-rate bracket >0 1,090 0% 0 52 Maximum deductions 48 (Taxable Base in UVT- Personal deduction 20 >1,090 1,700 19% 44 1,090UVT) x 19% Deduction for dependents Millions of Pesos 40 Por ciento Deduction for AFC 36 (Taxable Base in UVT - contributions 15 >1,700 4,100 28% 1,700UVT) x 28% + 116 UVT 32 28 24 (Taxable Base in UVT - 4,100 10 >4,100 8,670 33% UVT) x 33% + 788 UVT 20 16 (Taxable Base in UVT - 8,670 12 5 >8,670 18,970 35% UVT) x 35% + 2,296 UVT 8 4 (Taxable Base in UVT - 18,970 >18,970 31,000 37% 0 0 UVT) x 37% + 5,901 UVT 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 (Taxable Base in UVT - 31,000 Monthly wage, millions of pesos >31,000 Onwards 39% UVT) x 39% + 10352 UVT Source: Authors’ calculations, based on Law 2010 of 2019. Source: Law 2010 of 2019. As a result of its narrow base, the PIT is insufficiently progressive. Using the 2020 Personal FIGURE 4.12. Effective PIT Tax Rate on Annual and Life Income Tax Code, figure 4.12 shows the simulated effective rate on labor income (the red line) Income and Distribution of Wages (percent) for different levels of monthly labor income (on the horizontal axis). The simulation assumes 24 16 that the individual can use all available deductions up to the available limit. The figure also 14 shows the mass of individuals who belong to a specific income group (green bars). Deductions 20 and the zero-rate bracket have two effects: 12 16 Percent of all wages Percentage points 1. The effective PIT rate (that is, the ratio of PIT paid over total income) is zero until almost 10 the 93rd percentile of the income distribution. An individual whose labor income is slight- 12 8 ly above what 90 percent of workers earn (that is, almost up to 6 million pesos per month, the equivalent of about US$3,600 per month) would pay zero or very little PIT because 6 8 the entirety of this income would be deducted or would be taxed at a zero rate. 4 2. Because marginal rates apply only to the part of income that exceeds the sum of deduc- 4 2 tions and the zero-rate bracket, even for individuals who actually have to pay PIT, the PIT applies to a small fraction of their income. As a result, for most high-income individuals, 0 0 0 2 4 6 8 10 12 14 16 the effective PIT rate is no more than 10 percent. Using the same simulation as in figure Monthly wage, COP millIon 4.12, figure 4.13 shows the effective rate corresponding to each marginal rate bracket. For Distribution of wages example, at a level of labor income at which the 39 percent marginal tax rate takes effect, Effective tax rate on life income (RHS) the effective PIT rate is only about 26 percent. Notably, the 39 percent marginal tax takes Effective tax rate (RHS) effect at monthly earnings above 112 million pesos, the equivalent of approximatively Source: Authors’ estimates, based on the survey of households’ budget, DANE (2018c) and US$31,000. the tax code. 3. Because deductions (up to a cap) are calculated as a percent of income (instead of as a FIGURE 4.13. Marginal and Effective PIT Rates (percent) fixed amount), the effective marginal tax rate is lower than the statutory one, which not only decreases the amount of taxes paid but also reduces the (perceived) progressivity of 40 increasing marginal rates. 35 Because deductions on pension income are even higher than those on labor income, the effective PIT rate on lifetime income is even lower than that on annual income. Over the 30 life of an individual, pensions can be thought of as deferred labor income. Indeed, to avoid Percentage points double imposition over the life of an individual, in Colombia (as in almost all tax systems) the 25 contributions to pension savings (the part of income that is being deferred) are deducted from 20 income for computing taxable income, but pension income is taxed. However, in Colombia, the maximum amount of deductions from pension income is higher than that for labor income. In 15 other words, income that is saved and later received as a pension is taxed at a different effec- tive tax rate than the labor income that originated the savings, which is against the principle of 10 horizontal equity (according to which two individuals with the same income should be taxed 5 the same). As a result, the effective PIT rate that applies to lifetime labor income is lower than that that applies to annual labor income. The blue line of figure 4.12 shows the effective PIT 0 0 20 40 60 80 100 120 140 rate on lifetime labor income for different levels of labor income. This is computed as the sum Monthly wage (millions of pesos) of all PITs paid on labor income and on pension income over the course of an individual’s life, Effective tax rate Marginal tax rate expressed as a percent of total labor and pension income. The calculation assumes that an individual (i) earns the same income (in real terms) over his/her work life, that (ii) he/she con- Source: Authors’ estimates, based on the household survey and the tax code. tributes to the public pension system for 25 years, and (iii) he/she enjoys a pension for 20 years after retirement. High deductions on pensions reduce the effective PIT rate on lifetime labor income (blue line) to almost two-thirds of the effective PIT rate on annual labor income. This calculation does not assume voluntary contributions to private pension funds. Because these savings are deductible up to 30 percent of a person’s taxable income (that is, these savings are deductible at the marginal PIT rate), the resulting effective tax rate on lifetime income (the blue line of figure 4.12) would actually decrease as income increases, which is very regressive. 46 VAT exemptions benefit the rich B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Public F inance and Equity in Colombia At 19 percent, the standard VAT rate in Colombia is in line with that of other countries, but because of exclusions and zero or reduced rates, this rate applies to only 40 percent of total consumption. The standard VAT rate is among the highest in the region and is close to the average across OECD countries. Also, as in many countries, basic goods (for example, grains, meat, fruit, medicines, etc.) are either exempt or taxed at a zero or reduced (5 percent) rate. As a result of these exemptions and reduced or zero rates, in Colombia, the standard 19 percent rate applies to only about 40 percent of total household consumption. This does not mean that raising the VAT rate across the board to 19 percent would increase collection by 150 percent, however, because an estimated 64 percent of total purchases of goods is done in shops with sales below the VAT registration threshold.9 These shops do not need to register for VAT and hence do not charge VAT to final customers. VAT is still collected on the products that they sell but only at the earlier stages of the production chain. The 5 percent reduced rate ap- plies to about 3.6 percent of household consumption, while the zero rate or exempted goods represent the remaining 56 percent. FIGURE 4.14. VAT Rate across Countries in 2020 (percent) FIGURE 4.15. Share of Total Consumption Affected by Standard and Reduced VAT Rates 25 60 20 50 15 40 Percent Percent 10 30 5 20 10 0 a ay ia ic o ru le a y CD r do ua bi in liv i 0 Pe gu Ch OE ex nt m ua ug Bo ra lo ge M Ec Ur 19 5 0, or exempt Pa Co Ar Standard rate OECD 1st decile OECD 9th decile VAT rates, percent Source: Ernst & Young (2020) and OECD Statistics. Source: Authors’ calculations, based on the household survey. Reduced or zero rates are meant to correct the fact that VAT weighs more on low-income than high-income individuals. In principle, VAT is a neutral tax—it taxes all transacted values of goods and services at the same rate. However, because low-income individuals consume a larger share of their income (mostly for essential goods) than high-income individuals (saving rates tend to increase with income), sale taxes or VAT weigh more on the income of low-income than high-income individuals. The blue bar of figures 4.16 and 4.17 shows this for Colombia. The blue bars on these figures indicate how much of their income individuals spend on consumption and on VAT, respectively (relative progressivity). However, because the volume of consumption increases with income, high-income individuals contribute more than low-income individuals to the total VAT collection (orange bars of figures 4.16 and 4.17, absolute progressivity).10 Part of the regressivity of this tax is attenuated by the fact that in a context of high labor informality such as Colombia’s, the VAT allows the tax net to be broadened by taxing consumption: the income from informal work or informal firms is not taxed when it is generated, but rather when it is consumed. Also, as Bachas, Gadenne, and Jensen (2020) show, in a context where informal points of sale exist, low-income individuals can access de facto VAT-free products, which reduc- es the weight of VAT for them. However, in Colombia, as in many other countries, the bulk of the attenuation is done through VAT exemptions and reduced rates. Indeed, by reducing the tax on essential goods and services (which constitute most of the consumption basket of low-in- come individuals), VAT exemptions and zero or reduced rates decrease the weight that this tax has on lower-income individuals. An additional policy to decrease the regressivity of VAT is the unconditional cash transfer program Devolución del IVA, introduced in 2020 (CONPES 2020a). The value of the transfer is calibrated based on an estimate of the value of VAT that low-income families pay. To be eligible, a person must be in the Sistema de Identificación de Potenciales Beneficiarios de Programas Sociales (Sisbén) and be classified among the poorest recipient of the programs Familias en Acción and Colombia Mayor. FIGURE 4.16. Relative and Absolute Progressivity of FIGURE 4.17. VAT, Relative and Absolute Progressivity Consumption (percent) 500 35 12 40 450 35 30 10 Percent of total consumption 400 Percent of tax collection 30 350 25 Percent of income Percent of income 8 25 300 20 250 6 20 200 15 15 4 150 10 10 100 2 5 5 50 0 0 0 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Income deciles (market and pension) Income decile (market and pension ) Relative Absolute (r.a.) Relative Absolute (r.a.) Source: Authors’ estimates, based on the household survey. Source: Nuñez et al. (2020). However, because the volume of consumption differs across income groups, the lost rev- enue deriving from VAT exemptions and zero or reduced rates ends up disproportionally benefiting high-income individuals. Figure 4.18 shows the distribution of the lost VAT deriv- ing from exemptions and zero and reduced VAT rates by income group. At the household level, reduced rates and VAT-exempted goods greatly alleviate the cost of consumption for poorer households. The blue bars in figure 4.18 show the value of this alleviation in percent of income on average for individuals in different income deciles. However, because richer households consume larger volumes, effectively more than half of the lost VAT revenue originates from the consumption of the three highest deciles of the income distribution. The orange bars in figure 4.18 show how much of the total VAT loss originated from the consumption of individuals in different deciles of the income distribution. To put it differently, more than half (57 percent) of the tax expenditures on VAT benefit the top three deciles of the income distribution. This can also be observed at the level of individual classes of exempted or zero-rate goods. Figure 4.19 shows how much of aggregate spending, by all households and individuals, on a specific group of goods and services that are either VAT exempt or have a zero VAT rate, is spent by persons in different income deciles. For example, although individuals in lower deciles of the income distribution contribute the most to aggregate spending on rice, individuals in the top two deciles contribute to one-third of aggregate spending on bread, meat, dairy products, fruits, and vegetables. For recreation, arts, and museums (all VAT free), the top decile of the income distribution contributes to 60 percent of the aggregate spending. That VAT reduced rates and exemptions are poorly targeted is also found in other countries (Phillips et al. 2018). FIGURE 4.18. Tax Expenditure on VAT Exemptions and Zero or FIGURE 4.19. Absolute Progressivity of Spending on Selected Reduced Rates: Relative and Absolute Progressivity Goods and Services with Zero VAT Rate or Exempted (percent) (percent) 70 35 25 70 60 30 Percent of total lost revenue 60 Percent of total consumption Percent of total consumption 20 50 25 Percent of income 50 40 20 15 40 30 15 30 10 20 10 20 10 5 5 10 0 0 1 2 3 4 5 6 7 8 9 10 0 0 Income deciles (market plus pension) 1 2 3 4 5 6 7 8 9 10 Relative Absolute (r.a.) Decile of income (market plus pension) Source: Authors’ estimates, based on the household survey. Rice Dairy Recreation (r.a.) Bread Fruit and vegetables Meat Transport Source: Authors’ estimates, based on the household survey. Essentially, because VAT exemptions and reduced rates reduce revenue collection, they create a missed opportunity for more redistribution. Specifically: 1. The revenue lost from exemptions and reduced rates is large. According to estimates by the Ministry of Finance, the missed revenue from VAT exemptions and reduced rates is about 4.8 percent of GDP (Tax Expert Commission, 2021). This is high by international standards. Figure 4.20 shows a proxy of the so-called c-efficiency ratio for many coun- tries. This is an indicator of how much VAT is collected relative to what could potentially be collected. The higher the ratio, the more efficient the VAT collection and the least lost to evasion, compliance issues, or reduced rates and exemptions. This proxy is computed by looking at the effective VAT rate on total private consumption (that is, VAT collected over total private consumption), scaled by the statutory base rate (19 percent for Colom- bia). As the figure shows, Colombia is among the countries with the lowest efficiency in VAT collection. 2. If all goods and services were taxed at the standard 19 percent rate, the government would be able to collect enough resources to fully compensate lower-income individuals and still be able to have resources to spare. Figure 4.21 shows estimates of the resources that would remain available after compensating individuals of incomes up to a given decile for the increased VAT burden. The figure assumes no secondary effects on the volume of demand. If all individuals (up to the richest) were compensated, there would be no extra resources left. If only the first decile were compensated, almost all of the increased rev- enue would be left. If compensation, say, was applied to the bottom seven deciles (that is, everyone except the top three deciles for whom VAT benefits amount, on average, to less than 13 percent of their income), then 3.2 percent of GDP revenue would be left to increase spending on other programs. Of course, the problem of such a scheme would mostly be on the administrative side, in particular on ensuring proper targeting. Yet it offers an interesting perspective on the redistributive inefficiency of current VAT exemp- tions and reduced rates in Colombia. FIGURE 4.20. Proxy C-Efficiency Ratio FIGURE 4.21. Resources Available after Compensating Individuals for the Loss of VAT Exemptions and Reduced 1 Rates, up to a given decile of income 8 70 0,8 7 60 6 Percent of income 50 Percent of GDP 0,6 5 40 4 30 Ratio 3 0,4 20 2 1 10 0,2 0 0 1 2 3 4 5 6 7 8 9 10 Decile of income (market plus pension) 0 Value for the individual of VAT exemptions and reduced rate (r.a.) ico Co Sp ly lo ain d rt ia d l Po om Ic land ak thu nd p ia ov lic s ia lg ia Fr ium Au nce Ch ia Cz Net La ile h rla ia Ge ub s rm lic De nla y nm nd ng rk e y Es den Sw Ire nia er d xe orw d bo y sr g w Ko el al a d ng a p d Fi an Sw ar m a ur Ze re itz lan Lu N lan an Ita ite Po mb Re an Au en Be tral r ec he tv a Ki ug Hu a Sl ub Re n st ov Li ela ex Extra resources available for redistribution (in percent of GDP) to a I M Ne Un Sl Source: Authors’ estimates, based on the household survey. Source: OECD Statistics. 47 Transfers and subsidies leak to high-income earners BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Public F inance and Equity in Colombia The main transfer programs in Colombia do provide support for lower-income house- holds and are progressive. If one looks at the support that cash transfer programs, such as Familias en Acción (which reaches poor families with children) and Colombia Mayor (which reaches poor elderly people—see box 4.2 for a description of these and other programs), offer to households in relation to their income, it is clear that they are progressive (blue column of figures 4.22 and 4.23). They provide up to 20 percent additional income to poorer households, and the support phases out fairly quickly. For a household in the third decile, the benefit rep- resents less than 5 percent of household income. However, these programs suffer from leakages, and about 40 percent of their spending reaches households for whom the transfers add little. For example, only about 40 percent of total spending under Familias an Acción reaches households in the first two deciles of the income distribution, and 18 percent reaches households in the third decile (figure 4.22, blue bars). The rest (42 percent) reaches households up to the top decile of the income distribu- tion, households for which the transfer increases their market income by less than 3 percent. A similar pattern can be observed for Colombia Mayor. Of course, these programs (in particular, Familias en Acción) pursue objectives that go beyond income support, specifically to promote school enrollment and provide medical care among children. Yet, reducing their coverage only to families for whom the impact is above a 3 percent relative threshold (that is, 3 percent of income) would reduce coverage only to individuals in the bottom two deciles of the income distribution but double the level of the benefits (or expand coverage to those in the bottom two deciles of the income distribution who currently do not receive the benefits) in a fiscally neutral way. FIGURE 4.22. Relative and Absolute Progressivity of Familias Box 4.2. Main Cash Transfer Programs and Subsidies in Colombia en Acción (percent) 25 25 The Colombian Social Protection System includes cash transfer programs (e.g., Familias en Acción, Jóvenes en Acción, Colombia Mayor), transfers in kind (e.g., nutritional aid un- 20 20 Percent of total spending der the Feeding School Program), utility subsidies, and programs for productive inclu- Percent of income sion (e.g., Iraca). 15 15 Familias en Acción is the largest social program in Colombia, benefiting more than 2.6 10 10 million poor families. It is a cash transfer program that fosters human capital formation and citizen competencies and provides a bi-monthly cash transfer to poor and vulnera- 5 5 ble families with children below 18 years of age, conditional on complying with health and education commitments. Beneficiaries are targeted according to their Sisbén scores 0 0 and other requirements in the case of victims of forced displacement and indigenous 1 2 3 4 5 6 7 8 9 10 Deciles of income (market and pension) people. Geographic conditions determine the cash transfer amount (people in the poor- Relative Absolute (r.a.) est municipalities receive the most), just as the school grade in the program’s education component. The health component grants a transfer to families with children below 6 Source: Nuñez et al. (2020). years old. FIGURE 4.23. Relative and Absolute Progressivity of Colombia Jóvenes en Acción extends social assistance to nearly 296,000 poor and vulnerable youth Mayor (percent) for their human capital formation. It is a conditional cash transfer program targeted at 14 25 young adults between 16 and 24 years who are in face-to-face technical, technological, 12 and vocational training programs and living in poverty or vulnerability. The benefit is 20 Percent of total spending meant to help finance living expenses while in school. Like Familias en Acción, beneficia- 10 Percent of income ries are identified based on their Sisbén score and the availability of tertiary education 15 8 in the municipality where they reside. The cash incentive amount and frequency depend 6 10 on the education institution type and the level of the student’s commitment (enrolling, staying in the program, and maintaining high grades). 4 5 2 Colombia Mayor offers an unconditional cash transfer to vulnerable old adults without a pension or income who are living in destitution or extreme poverty. The program extends 0 0 1 2 3 4 5 6 7 8 9 10 to approximately 1.6 million people and provides a monthly transfer sufficient to cover Deciles of income (market and pension) basic needs. Program eligibility is based on the legal pension age (57 years for women Relative Absolute (r.a.) and 62 for men), not having a pension, living with a family whose income is below the Source: Nuñez et al. (2020). minimum wage, living on the streets, living in specialized nursing homes, the Sisbén score, and the presence of physical or mental disabilities. Some beneficiaries receive an additional amount when the local authorities co-finance the transfers or the beneficia- The coverage of utility subsidies, which are targeted at low-income families, reflects a ry participates as a community mother in the programs of the Instituto Colombiano de large inclusion error. Since 1994 the identification of beneficiaries of utility subsidies has Bienestar Familiar (Colombian Institute of Family Welfare). been based on a classification of the dwellings in six groups (strata, from the word used in Electricity and gas subsidies are granted through a solidarity fund, partially funded by Colombia), based on characteristics of the dwelling (and of the dwellings surrounding it) that users with a higher ability to pay. Households from Sisbén strata 1, 2, and 3 receive an ad are considered to be a good proxy of income (for example, a house with a concrete floor is valorem subsidy (up to 60, 50, and 15 percent, respectively) of an electricity “subsistence considered to indicate lower household income that a house with a tiled floor). Households consumption,” while strata 5 and 6 and commercial establishments pay a 20 percent living in a house classified in the first three (that is, the poorest) strata receive a subsidy for surcharge on their electricity bill. For the gas subsidy, strata 1 and 2 do not pay up to 60 water, gas, and electricity, whereas households living in a house classified in the top two strata or 50 percent of their consumption (as long as it does not exceed a “subsistence con- pay a surcharge that helps finance the subsidy (figure 4.24). While Sisbén tends to correctly sumption” threshold), while households in higher strata pay an additional 20 percent of assign lower-income households to the lower strata, high-income households are incorrectly total fixed and variable components of the utility bill to finance part of the subsidy. assigned to the bottom strata. This is because the strata system is not updated regularly and reflects the characteristics of houses—not the income of the people living in them. As figure Finally, local authorities finance (optionally) water subsidies under some general guide- 4.25 shows, about 91 percent of households in the sixth income decile and above are classified lines. Lower strata households (1, 2, and 3) receive the subsidy, while higher-strata in the lowest three strata. Chapter 5 on territorial inequality discusses in more depth the households (5 and 6) and commercial establishments pay a solidarity contribution. The inclusion and exclusion error under these programs and ways to address them. subsidy cannot exceed 70 percent of water consumption for stratum 1, 40 percent for stratum 2, and 15 percent for stratum 3. Users in stratum 4 or public establishments do not participate in the cross-subsidy scheme. FIGURE 4.24. Distribution of Selected Subsidies, by stratum FIGURE 4.25. Distribution of Strata, by income decile (Percent) (percent) 100 100% 90 80 80% 70 60 60% 50 40% 40 30 20% 20 10 0% 0 1 2 3 4 5 6 7 8 9 10 Electricity Gas Water Decile of income (market plus pension) Strata 1 Strata 2 Strata 3 Strata 1 Strata 2 Strata 3 Strata 4 Strata 5 Strata 6 Source: Nuñez et al. (2020). Source: Authors’ calculations. As a result, about 60 percent of total spending on utility subsidies reaches households for which these subsidies offer little income relief. As figure 4.26 shows, electricity subsidies represent about 12 percent of income for households in the first decile of the income distribu- tion. Yet, only about 9 percent of total spending on electricity subsidies reaches these house- holds. The majority of spending on electricity subsidies reaches households of income deciles five and above. For these households, the electricity subsidy represents at most 2 percent of household income. A similar pattern is observed for subsidies to gas and water. The distortion is even starker for gas subsidies; possibly because poorer families do not have access to gas for cooking or heating, the majority of this subsidy is spent on middle-income families compared to poorer ones. FIGURE 4.26. Relative and Absolute Progressivity of Electricity FIGURE 4.27. Relative and Absolute Progressivity of Gas FIGURE 4.28. Relative and Absolute Progressivity of Water Subsidies Subsidies Subsidies 14 14 14 14 14 14 12 12 12 12 12 12 Percent of total spending Percent of total spending Percent of total spending 10 10 10 10 10 10 Percent of income Percent of income Percent of income 8 8 8 8 8 8 6 6 6 6 6 6 4 4 4 4 4 4 2 2 2 2 2 2 0 0 0 0 0 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Deciles of income (market and pension) Deciles of income (market and pension) Deciles of income (market and pension) Relative Absolute (r.a.) Relative Absolute (r.a.) Relative Absolute (r.a.) Source: Nuñez et al. (2020). Source: Nuñez et al. (2020). Source: Nuñez et al. (2020). 48 Pensions and labor income inequalities BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Public F inance and Equity in Colombia The pension system is characterized by low coverage and low participation, especially FIGURE 4.29. Individuals Eligible for Retirement, Receiving in the lower decile of the income distribution (for more on this issue, see also Chapters a Pension, and Contributing to a Pension 1 and 2). Figure 4.29 shows the number of adults older than retirement age (blue bars) and 1200 100 90 Percent of working age populaiton those who actually receive a pension (orange bars) by income decile. Many of those who would 1000 be eligible for a pension, based on age only, do not receive one, and this is more the case the 80 Thousands of people 70 lower the income. This results from a higher incidence of informality at lower income levels 800 60 (which means lower contribution density at lower income), combined with the requirement 600 50 that a person contribute for 25 years (Bosch et al. 2015). Figure 4.29 also shows the fraction 40 of those working who make pension contributions by income decile (gray bar). Possibly re- 400 30 flecting the two factors mentioned above and individual expectations about the probability of 20 200 receiving a pension in adulthood, fewer lower-income workers contribute than higher-income 10 workers, which results in a higher fraction of high-income individuals retiring and receiving a 0 0 1 2 3 4 5 6 7 8 9 10 pension later in life. Deciles of income (market and pension) As a consequence, public pension payments concentrate on individuals in higher deciles Above retirement age Is contributing (r.a.) Receives a pension of the income distribution. The low participation rate among low-income individuals, and Source: Nuñez et al. (2020). their (consequent) low coverage, has two effects. First, individuals in lower deciles who receive a pension need to complement pension income, possibly by continuing to work. Indeed, the FIGURE 4.30. Relative and Absolute Progressivity of Pension Spending contributory pension that they receive is no more than 25 percent of total income (figure 4.30, blue bars). Second, most spending on public pensions reaches the top decile of the income 9 70 distribution. This, of course, reflects the larger contributions and higher contribution density 8 60 among this group and derives from differences in wages, formality status, and stability of ca- 7 50 Percent of spending Percent of income reer during the work life. In a way, this result does not represent, in itself, an equity problem. 6 5 40 4 30 3 20 2 1 10 0 0 1 2 3 4 5 6 7 8 9 10 Deciles of income (market and pension) Relative Absolute (r.a.) Source: Nuñez et al. (2020). The inequity of the pension system derives from the existence of implicit subsidies. FIGURE 4.31. Internal Rate of Return for Different Wages There are two sources of implicit subsidies: 12 1. Those who contribute to the public pension system but who do not accumulate enough 11,5 weeks of contribution to obtain a pension receive their contributions back, adjusted for 11 Rate of return (percent) inflation but not for the real return that their savings would have generated if invested in 10,5 the financial market. Because in a pay-as-you-go system such as Colombia’s public pen- 10 sion program, current contributions are used to pay current pensions, this is financially equivalent to the government borrowing at a subsidized interest rate to pay current ben- 9,5 efits, where the subsidy is obtained from those affiliates who will not receive a pension. 9 8,5 2. Second, the combination of retirement age, fairly high replacement rate, and low con- tribution rates makes the system actuarially generous. At the individual level, this gen- 8 1 2 4 8 16 32 erosity is progressive, that is, the internal rate of return of contributions and pensions is Ratio to minimum wage at the beginning of career higher for individuals with lower wages (figure 4.31), including for individuals with wage progression over their working life.11 However, the fact that high-income earners (whose Stable ratio of wage-to-minimum wage careers tend to be longer, more stable, and with greater wage progression) are more likely The ratio wage-to-minimum wage doubles by the end of career to accrue a pension than low-income earners tilts the aggregate generosity of the system Source: Authors’ simulations. toward high-income pensions. The implicit subsidies in Colombia’s pension system accrue primarily to recipients of high FIGURE 4.32. Distribution of Transfers to Colpensiones, pensions. The distribution of these subsidies can be proxied by looking at how the transfers to by decile of income Colpensiones (the administration of public pensions) to cover the deficit in the public pension 70 Percent of total transfers to Colpensiones system are distributed by level of income (figure 4.32). Most of the deficit is concentrated at 60 the greatest mass of pension payments, which is on high pensions. However, it is important to stress here that the public pension system is actually designed to offer higher internal returns 50 to lower salaries and shorter careers, and that (i) the deficit results from the combination of 40 low participation and high replacement rates and (ii) the distribution of this deficit results 30 from the fact that individuals with long career paths and high wages are more likely to accrue 20 a public pension. Yet, it is striking that overall, the pension system generates a transfer of re- sources toward higher-income retirees. 10 0 1 2 3 4 5 6 7 8 9 10 Deciles of income (market and pension) Source: Administrative data and the household survey. 49 Bringing it all together: the overall redistribution of fiscal policy in Colombia BUI L DI N G AN EQ UI TABL E SOCI E TY I N CO LO MBI A | P ublic F inance and Equity in Colombia Overall, totaling direct and indirect taxes, cash transfers, and subsidies, Colombia’s fis- cal policy benefits individuals up to the fifth decile of the income distribution. For individ- uals in the lowest two deciles, transfers and subsidies represent a significant support to their income, up to 80 percent for individuals in the first decile and up to 30 percent for those in the second. Indirect taxes are a burden for them, costing roughly 20 and 8 percent of their income, respectively (figure 4.33). The net impact declines rapidly (to around 10 percent of income for individuals in the third decile of the income distribution) and becomes zero for individuals of median income. Individuals of higher income are net contributors to redistribution, but the amount that they provide does not exceed on average 13 percent of their income. The aggregate contribution to total direct and indirect taxes and the distribution of trans- fers and subsidies confirm that direct taxes contribute little, and that there are substantial leakages in transfers and subsidies to individuals in the top three deciles of the income distribution. In particular, high-middle income individuals (specifically those in the eighth and ninth deciles of the income distribution) contribute little aggregate resources through di- rect taxes (figures 4.34 and 4.35). At the same time, the top three deciles receive about 1.3 per- cent of GDP of transfers and subsidies that add little to their income and represent a missed opportunity to give more to the poor and vulnerable. FIGURE 4.33. Impact of Direct and Indirect Taxes, Transfers, FIGURE 4.34. Distribution of Selected Taxes, Transfers, FIGURE 4.35. Concentration Curve and Subsidies on Individual Income (percent) Subsidies, and Spending by Income Decile (percent of GDP) 1,0 100% 7 0,9 6 0,8 Concentration index 75% 0,7 5 Percent of GDP 0,6 4 0,5 50% 3 0,4 2 0,3 25% 0,2 1 0,1 0 0,0 0% s s s s di on di lth xe er xe ie 0 10 20 30 40 50 60 70 80 90 100 ng ng en ti en ea id sf ta ta sp uca bs an sp H ct ct Percentile of market income Su Tr Ed re re Di di In -25% Direct taxes Market income Disposable income 1 2 3 4 5 6 7 8 9 10 Decile 1 Decile 2 Decile 3 Decile 4 Decile 5 Indirect taxes Direct transfers Decile 6 Decile 7 Decile 8 Decile 9 Decile 10 Impuestos directos Direct transfers Indirect taxes Subsidies Net balance Source: Nuñez et al. (2020). Source: Nuñez et al. (2020). Source: Nuñez et al. (2020). 4.3. 50 BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Public F inance and Equity in Colombia Policy Options To improve the redistribution of the fiscal system, policy options could aim primarily at FIGURE 4.36. Effective Tax Rates under Different Option improving the progressivity of the tax system and refocusing transfers and subsidies.12 of Alternative Tax Schedule, and Distribution of Income Specifically, policies could plan to: (percent) 24 40 1. Extend the PIT to the top two deciles of the income distribution in the short run, aim- ing to extend it to the top half of the income distribution over the long run as income 20 increases and poverty is significantly reduced. There are different ways in which the 30 extension can be done in the short term. For illustrative purposes, two extremes and a 16 Percent of all wages Percentage points combination are presented. For each option, the impact on total PIT collection was sim- ulated by replicating different PIT schedules using the framework and 2017 Household 12 20 Survey as outlined in Nuñez et al. (2020). 8 • One possibility is to reduce the maximum amount of permissible tax deduction. How- 10 ever, because the first (zero marginal tax) bracket arrives at 1,090 UVT (Unidad de Valor 4 Tributario, see notes 8 and 10), about 37 million pesos annually, or the equivalent of about US$880 per month, permissible deductions should be reduced to zero so that 0 0 individuals in the 9th decile of the income distribution pay PIT. This would also imply 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 eliminating the declaration threshold, which would increase the effective tax rate for Monthly wage, millons of pesos all income levels and increase PIT collection an estimated 138 percent, that is, from Distribution of gross income about 1.2 to 2.7 percent of GDP. However, this would also create horizontal inequity: Effective tax rate (option 1: no deduction) two individuals with the same income but different contributory capacity (for exam- Effective tax rate (option 2: lower zero rate bracket) Effective tax rate (option 3: lower deduction ple, one with a dependent child and the other without dependents) would have to pay and 5 percent first bracket) the same amount of PIT. Effective tax rate (current PIT schedule) Source: Authors’ estimations. • On the other extreme, a second possibility is to halve the upper bound of the first (zero marginal rate) bracket and leave the rest the same. The zero rate bracket has a pur- FIGURE 4.37. Estimated PIT Collection under Current PIT pose. Because deductions are based on a ratio of income (instead of a deduction of a Schedule, and Different Reform Option (percent of GDP) fixed amount), individuals of low income would have to declare some taxable income. 3,0 The zero rate bracket makes it possible to shield these individuals from paying the PIT. Reducing the width of the first bracket by half would allow an extension of the base of 2,5 the PIT to approximately the same percentile as the elimination of deductions. This would also imply reducing the declaration threshold by the same amount. However, 2,0 by retaining deductions, this reform would contain the increase in the effective tax 1,5 rate, preserve horizontal equality, and lead to an increase in collection of about 115 percent from 1.2 to 2.6 percent of GDP. 1,0 • A third possibility is to implement a combination of the above. This would consist of 0,5 transforming all deductions in fixed amounts (instead of the current percent of income) up to a maximum of 900 UVT. The declaration threshold would also have to be reduced 0,0 Current PIT Option 1: Option 2: Option 3: to this amount. At the same time, a marginal rate of 5 percent would apply to the first schedule no deduction lower zero lower deduction bracket. Relative to the two other options, this would make it possible to reduce the rate bracket and 5 percent first bracket effective tax rate for those individuals who were not effectively taxed before. Although it would imply collecting less on a greater mass of individuals, this option would allow Source: Authors’ estimations. an increase of an additional 64 percent of revenues from 1.2 percent to about 2 percent of GDP. • Over the long run, extending the coverage of the PIT even further down the income dis- tribution would require lowering deductions (to about 450 UVT) and limiting the first bracket and the declaration threshold (to about 545 UVT). 2. Tax pension income at least the same as labor income. Using the same deductions for pension income as in the third option presented above would increase collection by an additional 0.2 percent of GDP. 3. Reduce VAT exemptions. Reducing the list of goods that are exempt from the VAT could be done gradually by levying a low rate initially that would be increased over time to allow production chains (especially for those goods that are exempted) to adapt prices. An alternative is to set a 5 percent VAT rate on goods that are consumed almost in equal proportions by different income groups, and a 12 percent VAT rate on those that are con- sumed in greater share by high-income individuals. This would allow an increase in VAT collection of about 10 percent, the equivalent of about 0.7 percent of GDP. This assumes that there would be no change in the share of goods sold in establishments registered in the VAT system. Reforms to increase registration could increase the gain from the VAT rate changes. Compensating households up to the median income would cost about a quar- ter of the total savings, that is, slightly less than 0.2 percent of GDP. The rest could be used to further strengthen cash transfer programs. 4. Better focus transfers and reduce leakages of utility subsidies. For both transfers and subsidies, improving the targeting can be done so that individuals receive the transfer only if their income is below a certain threshold or only if the transfer contributes more than a given share of their income. • For example, if transfers were received only by individuals with an income smaller than the median, spending on Familias en Acción would decrease by 40 percent and spend- ing on Jovenes en Acción by 58 percent. The total amount that could redistributed to beneficiaries would amount to 0.15 percent of GDP. • Chapter 5 advances concrete proposals on improving the targeting of subsidies on gas, electricity, and water. To put reforms into context, savings were estimated on elimi- nating subsidies for households in which the subsidy amount represents less than 2 percent of their income. This would imply reducing spending on electricity subsidies by 61 percent, on gas by almost 100 percent, and on water by almost 87 percent. The total saving would amount to 0.3 percent of GDP. 5. Address the inequity of pensions. Achieving this objective would require a comprehen- sive reform that increases participation in the system, coverage of the elderly, and the actuarial fairness of the system. Such measures should also preserve the system’s sus- tainability, that is, ensure that the present value of pension payments’ net contribution is close to zero. Reforms would require a revision of both the number of years needed to obtain a pension (to help improve the prospects for coverage) and the replacement rate (so that it reflects the effective amount that an individual has saved over the course of his/her working life). Capping the level of income subject to public pension insurance would also reduce the proportion of pension subsidies accruing to the highest-income individuals. All of these reforms are necessary to guarantee coverage while preserving the sustainability of the system. A combination of the above policies would lead to a reduction in the Gini coefficient be- tween market and disposable income by a further 2 percentage points. Specifically, in- creasing the PIT by a total of 1 percent of GDP (as per the third option above), increasing VAT by 0.5 percent of GDP (net of a program to compensate the most vulnerable), and freeing 0.45 percent of GDP of resourcing by refocusing transfers and subsidies would free up enough re- sources (a total of about 2 percent of GDP) to double spending on current transfers (for ex- ample, under Familias en Acción, Colombia Mayor, Jóvenes en Acción) and on other programs (for example, the unemployment benefit, Programa de Alimentación Escolar, Asistencia a la Primera Infancia). This would still leave another 1 percent of GDP for other programs or fiscal objectives, including improving the quality of education or health care or reducing the fiscal deficit and debt. Policy Options for a More 51 BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | P ublic F inance and Equity in Colombia Redistributive Fiscal System Observed Drivers of Policy Options Relevant International Timing for Estimated Fiscal Impact Equity Gap The Equity Gap Experiences Implementation of Implementation PIT raises little revenue Very high exemptions and ze- Expand the coverage of the PIT to the Flat deductions (personal allowances and Immediate/ Gain (extra revenue) of 0.8 per- and is not progressive ro-rate bracket reduce the base top two deciles of the income distri- allowances for dependents) are present in short run cent of GDP in the short run enough. of the PIT to individuals whose bution in the short run (and aim to Argentina, Chile, Costa Rica, Guatemala, and income belongs to about the expand it to individuals earning the Peru, and among OECD countries, in Germa- top 7 percent of the income median income over the long run as ny, Italy, the United Kingdom, and the United distribution. income increases and poverty is sig- States. Also, in most countries, qualifying ex- nificantly reduced), by penses can be deducted up to a percentage of reducing allowable deductions (in- the expense and up to a fixed amount. cluding by making deductions a flat amount, not proportional to income) Among OECD countries, very few (Luxemburg and eliminating the zero-rate bracket. and Sweden) have both personal deductions and a zero marginal rate bracket. If personal deductions are allowed, the marginal rate on the first bracket is greater than zero. VAT exemptions and re- Exemptions and reduced rates Keep zero rates only for a few goods Zero or reduced VAT rates for food, medicines, Immediate/ Net fiscal gain of 0.5 percent of duced rates generate a involve goods that are con- that are primarily consumed by transport, and other basic goods (or for merit short run GDP (extra revenues of 0.7 and loss of revenues, the ma- sumed by everyone, but in low-income individuals. For all other goods, such as books, newspapers, etc.) are extra cost of 0.2 percent of GDP jority of which benefit greater share by high-income goods, use a 5, 12, and 19 percent common in many countries (Argentina, Ger- for VAT devolution program) high-income individuals. individuals. VAT rate, depending on the share of many, Italy, Mexico, the United Kingdom). Em- total consumption contributed to by pirical evidence shows that VAT exemptions or high-income individuals. At the same reduced rates are poorly targeted (Phillips et time, increase the amount and cover- al. 2018). age of the VAT devolution program to compensate households with income The OECD (Consumption Tax Trends 2020) rec- up to the median income. ommends that exclusions from VAT be limited to a short list of goods and services to avoid negative effects, such as cascading, on distri- bution chains and pricing. Transfers and subsidies For cash transfers, leakages Improve the targeting system of subsi- In 2016, Argentina improved the targeting of Immediate/ Fiscal gain (reduction in spend- do reach the vulnera- reflect inclusion errors in the dies by complementing the strata sys- subsidies to gas and electricity by linking eligi- short run ing) of 0.5 percent of GDP ble, but a large share of Sisbén scoring system. For sub- tem with information from Sisbén (see bility criteria to those of other social programs. the spending leaks to sidies, leakages reflect targeting concrete proposals in next section). high-income individuals bases on outdated proxies for Chile uses means testing for its water subsidy or individuals for whom household income. program. the benefit is minimal relative to their income. The pension system suf- Low participation and coverage Increase participation and coverage, To improve the financial sustainability of their Immediate/ Area of future research fers from low participa- reflect a system that requires a possibly linking the pension more systems, Italy, Latvia, Poland, and Sweden short run tion and low coverage long contribution period in a con- directly to the actual amount of con- radically reformed their public pension system and as a result, it faces a text of high informality and low tributions over the work life of the during the 1990s, shifting from defined benefit deficit, all of which effec- contribution density. Also, the individual, to make the system actu- to notional (non-financial) defined contribu- tively generate transfers replacement rate generates a sys- arially fair. Reforms should also aim at tion. In many countries, coverage has been from general taxation to tem that is actuarially generous. preserving the financial sustainability expanded by introducing a basic non-contrib- high-pension individuals. The treatment of savings of those of the system, i.e., ensuring that the utory pension (for example, in Chile, Ecuador, individuals who will not receive a present value of future payment is and Panama). pension generates implicit trans- matched by the present value of fu- fers to current retirees. ture contributions. Conclusion Improving the distribution of the fiscal system would require taxing more—and more progressively. It would also require spending not just more but more efficiently. The re- sults in terms of a reduction in the Gini coefficient might not be large. Yet, fiscal redistribution is not (and cannot be) the only solution to reducing inequality and poverty. However, fiscal redistribution can complement policies aimed at improving human capital accumulation or producing a more equal participation in markets, and it provides a protection against income shocks. These reforms might require not only a redesign of the system to identify eligible indi- viduals but also a rethinking of eligibility criteria and targeting. 52 Endnotes B U IL DIN G A N EQ UI TABL E SO CI ETY I N CO LO MBI A | P ublic Finance and Equity in Colombia 1 See also Moller (2012). 5 Policies to promote equality of opportunity in access to public services and to im- 2 This chapter will not discuss in depth reforms to the pension system. In principle, prove the quality of spending on public services are discussed in Chapter 2. the design of a pension system should ensure adequate income for occasions when 6 Labor informality contributes little to the fact that the majority of people who pay the individual is out of work, offer people insurance against uncertainty about the PIT are top income earners. This is because the gross income threshold above which duration of life, and ensure the sustainability of the pension system. Given the com- an individual has to file for income taxes (about 3.1 million pesos per month) ex- plexity and multidimensionality of pension reforms and the specific focus of this cludes almost all informal wage earners. chapter on equity, the equity angle of the pension system will be discussed only 7 Law No. 1111 of 27 December 2006 created the Tax Unit Value (Unidad de Valor Trib- briefly. The chapter will also not discuss the impact of fiscal policy on the spatial or utario, or UVT) as a value measurement to adjust the amount of taxes and other group dimensions of inequality. On the tax side, there are no specific provisions for obligations contained in the laws administered by the Colombian tax authority (Di- taxes on individuals that are based on geography, ethnicity, or gender, and the redis- rección de Impuestos y Aduanas Nacionales [DIAN]). The UVT, an index used to de- tributive property of taxes along territorial, ethnic, or gender lines reflects income fine tax brackets, thresholds, and deductions that is updated with inflation, was COP inequalities along these dimensions. On the spending side, issues related to the 35,607 for 2020. spatial distribution of transfers and subsidies are discussed extensively in Chapter 5. 8 Prior to this, Law 1819 of 2016 was the one that applied and classified the income Although there are significant overlaps in poverty along the income, ethnic, and geo- into five types of taxable income: employment income, pensions, capital, non-labor graphic dimensions, this chapter focuses on the redistributive properties of spend- income and dividends, and participations. ing along the income dimension only. 9 The registration threshold is set at 3,500 UVT. For 2020, this corresponds to about 3 It is difficult to compare the effective CIT rate across countries, as this depends on 125 million pesos, or the equivalent of about US$35,000. VAT registration thresholds the treatment of investment goods, taxes on labor, taxes on dividends and reinvest- for small enterprises vary greatly by country (for example, in the European Union, it ed earnings, and so on. An additional (and different) set of considerations should spans from zero in Spain to over €80,000 in France). Colombia’s registration thresh- be used to increase revenues through the taxation of corporate profits or financial old is comparable in value to that of Belgium, Bulgaria, or Hungary. income. Labor income tends to be less elastic to taxes than capital or corporate income. As a result, efficiency concerns and considerations about fostering innova- 10 The results of the analysis would not change if one computes and separates the in- tion, firm development, and research and innovation overlaps with considerations cidence of VAT along the entire production process of goods sold in informal shops about the transfer of resources (and the consequent equity improvements) that (those below the VAT threshold) and those sold in formal shops (above the threshold). reforms to CITs or taxes on financial income and wealth would make possible (IMF 11 The simulation in figure 4.31 considers individuals who contribute for 26 years to the 2017). Finally, the redistribution properties of taxes on corporate income depend on public pension system. The calculation is done for individuals who start their work- how dividends and capital gains are taxed within the PIT. ing life with a wage that is a multiple of the minimum wage. The blue line considers 4 Colombia instituted a wealth tax in 2002 under the name “Contribution to Democ- individuals whose wage will remain at the same starting ratio relative to the mini- racy.” This tax has undergone a sequence of changes. In 2014, a tax on net wealth mum wage, while the orange line considers individuals whose wage relative to the (net financial plus non-financial assets) was established for the period 2015–18, with minimum wage doubles (for example, from two to four times the minimum wage, or increasing marginal rates (topping 1.5 percent for individuals) applied to net wealth from 32 to 64). The calculation assumes a constant 3 percent inflation rate over time. exceeding 1 billion pesos (the equivalent of about US$300,000 on average during 12 The government has tasked a committee (the Comisión de Beneficios Tributario) that period). At end-2018, this tax became a temporary “tax on liquid wealth,” set with reviewing tax benefits in different areas and proposing reforms. The objective is to expire in 2021. It is levied at a rate of 1 percent on the net wealth (financial and to increase revenue and improve the fairness of the system. At the time of this writ- non-financial assets minus liabilities) above 5 billion pesos (the equivalent of about ing, the recommendations of this committee had not yet been released. US$1.5 million) of persons, corporations, and trust funds. At its peak (in 2015), this tax raised 0.7 percent of GDP, but in 2019 collection declined to 0.1 percent. BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LOMBI A 53 CHAPTER 5. in Colombia Territorial Inequalities Summary 54 BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | T erritorial Inequalities in Colombia Territorial inequalities in Colombia are high and likely to widen further. The disparities exist on different spatial scales—regional, urban-rural, intra-urban—and along various dimensions related to economic and social issues and accessibility. This chapter explores the multidimensionality of territorial inequality, examining different dimensions where in- equality emerges and using various spatial scales as a lens through which to assess them. First, cross-country comparisons suggest that regional inequalities in Colombia are two times higher than in other OECD countries. Second, an assessment of urban-rural disparities sug- gests that improvements in fiscal and municipal government performance and connective infrastructure are associated with lower multidimensional and monetary poverty in urban-ru- ral territories. Third, a focus on urban areas indicates that territorial inequalities persist even within city boundaries. The main contributors to intra-urban inequality are housing (the great- est gap) and education (the highest levels of deprivation), with many urban dwellers remain- ing far away from basic services (for 13 percent of cities, the closest clinic/hospital is more than 15 kilometers away). Lastly, this chapter shows that improving the targeting of utility subsidies can enhance the efficiency of expenditures, which is key in a resource-constrained environment. An assessment of the current stratification system used to target these subsi- dies suggests minor exclusion errors, indicating that those in need are receiving assistance, but inclusion errors of almost 70 percent also suggest that resources are being wasted. To reduce territorial inequalities, actions are needed at all spatial scales. • In the short run, strengthening subnational governments can help to effectively prioritize programs and investments in housing and education, where the deepest disparities per- sist. • Strengthening national programs on multipurpose cadaster, tax administration, and land value capture instruments can improve territorial planning, help identify deprived areas, and better channel resources where they are most needed. • Improving accessibility within cities, and between urban and rural areas, is crucial to bringing opportunities to everyone throughout the entire territory of Colombia. • Enhancing the targeting mechanism of the subsidies for servicios públicos domiciliarios (public utility services) may help free up resources that could be channeled to these oth- er priorities. 5.1 55 BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Territorial I nequalities in Colombia Introduction FIGURE 5.1. Inequality among Colombia’s Regions Compared Territorial inequality in Colombia is high, leaving segments of the population far from to Other Countries the opportunities needed to thrive. Regional inequalities in Colombia are more than two 10 times higher than in OECD countries (see figure 5.1). International experience suggests that 9,27 9 disparities in living standards across a country can foster political instability and increase so- 8 cial unrest (Rodriguez-Pose 2018). But although spatial imbalances in economic growth are 7 6,19 inevitable and even desirable for good economic performance, disparities in access to op- 6 5,46 portunities and living standards are not (World Bank 2009). In Colombia, poverty rates are as 5 high as 59.3 percent in Chocó and 50.7 percent in Cauca and as low as 8 percent in Santander 4 3,26 3,3 2,91 and 7.3 percent in Cundinamarca.1 In rural areas, only 74 percent of people have access to 3 2 1,46 1,9 improved water sources, compared to 97 percent in urban areas (UNDP 2015b). Despite over- 1 all reductions in illiteracy rates between 2005 and 2015 (from 8 to 5 percent) and a halving of 0 the regional gap, the illiteracy rate is five times greater in the department of La Guajira than Slovenia Spain OECD Poland Slovakia Chile Colombia Mexico average* in Cundinamarca (DANE 2018d). Also, poor rural departments usually have fewer kilometers of road built than other areas (Gamarra-Vergarra 2007), making it harder to access markets, Source: Authors’ calculations (ratio of regional GDP per capita for richest and poorest institutions, and other services and increasing transport and logistics costs and product pric- regions in 2018); for Colombia, DANE (2018d); for the others, OECD Regional Statistics. Note: *Not all OECD countries are included due to data availability. es. Building links between prosperous and vulnerable territories (where territories can be re- gions, urban-rural areas, or urban blocks) and ensuring that the benefits of economic activity contribute to a good quality of life across all territories can help build social cohesion while leaving no area behind. Well-planned territorial development may lead to lower poverty and inequality through higher levels of productivity, more employment opportunities, better education and health services, and a higher provision of public goods (UN-Habitat 2017). Better territo- rial planning can help effectively reduce financing gaps where needs are greatest. As such, the formulation of territorial development instruments requires an understanding of the com- plexity of territories and their interactions. Clear financial mechanisms and good planning are needed to ensure good prioritization and the efficient use of resources. But a good under- standing of the complexity of the challenges is needed first to better prioritize policy inter- ventions. Box 5.1 explains the World Bank’s World Development Report (WDR) 2009 framework that this chapter will use to prioritize policy interventions to promote territorial development and reduce spatial inequality. This chapter analyzes inequality across spatial scales and dimensions. It contributes to the overall conceptual framework of this report by integrating the territorial lens into the analysis of inequality, with a cross-sectoral and multidimensional approach. Territo- rial inequality is multidimensional, and looking at it through different spatial scales can help identify the main problems and where they are deeper. This chapter focuses on the spatial variation of inequality in terms of assets, the intensity of use of these assets, and their returns. Specifically, to assess territorial inequalities, this chapter focuses on four socioeconomic di- mensions: quality of housing, demographics, education, and the labor market. It addresses such questions such: where are the most pressing inequalities in Colombia and on what as- sets? What are the returns to those assets when urban and rural areas are connected? How can price mechanisms, through efficient subsidy targeting, address intra-urban inequality? The chapter is organized in three sections. The first provides diagnostics that look at in- equality at varying territorial scales, from regional to intra-urban. The second section focuses on policy options. It concludes by providing a summary of key messages and recommenda- tions for the steps ahead. Box 5.1. Framework for Prioritizing Policy Interventions: Institutions that Unify, In- frastructure that Connects, and Interventions that Target Policy prioritization is needed to guide policy makers to effectively use limited financial resources. The World Bank’s World Development Report 2009 puts forward a framework for this that outlines the need for three i’s (3Is) for places to develop: institutions, infra- structure, and interventions: • Territories first need institutions that level the playing field, institutions that ensure access to basic services for everyone, everywhere and that set the stage for market players to establish and promote economic activity and jobs. These policies are spatially blind in distribution across a country and should aim for universal cov- erage. Examples include regulations affecting land, labor, international trade, or social services, such as health, education, water, and sanitation. • As development advances in a place and institutions become stronger, connective infrastructure within cities and across urban-rural areas and regions becomes es- sential. Transport infrastructure can enable access to jobs, services, and opportu- nities for all groups of the population. Examples include roads, railways, airports, harbors, and communication systems that facilitate the movement of goods, peo- ple, and ideas in different cities and regions. • For places where institutions and infrastructure are not a pressing problem, tar- geted interventions should be considered to tackle challenges in specific sectors or areas of the territory, such as slums, or fiscal incentives for a productive sector (e.g., manufacturing firms). Ultimately, places need institutions that unite, infrastructure that connects, and interven- tions that target to enable territorial development across the board. Hence, depending on the development level of the territory, different challenges will be more pressing and different sets of policy instruments should be used. This chapter uses the “3Is” frame- work to propose policy recommendations tailored to each territory’s context. Source: World Bank (2009). 5.2. 56 BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | T erritorial Inequalities in Colombia Diagnostics: Regional and Intra-Urban Inequalities Regional inequalities in Colombia are large, including between urban and rural areas, and persist even within urban boundaries. A thorough review of the literature2 suggests that regional inequalities have been well documented in Colombia. The discussion on region- al inequalities in this diagnostic characterizes the eight regions used in the Plan Nacional de Desarrollo 2018–2022 (National Development Plan) (see figure 5.2)3 in terms of interregional gaps and the socioeconomic dimensions of higher vulnerability. Similar to global trends,4 Co- lombia’s population is concentrated in the cities (OECD and EC 2020). It has an urbanization rate close to 76 percent, with 34.1 million people living in urban areas (DANE 2018a). This work then focuses on urban-rural inequalities and identifies how urban areas, which overall fair bet- ter than rural areas, can better support their surrounding rural territories to reduce inequality. The diagnostic closes with a deeper look into cities, where inequalities persist and mimic na- FIGURE 5.2. Regions in Colombia Used in this Chapter tional disparities. Spatial disparities within cities have been less studied in Colombia, despite international evidence stressing the need to build an evidence base on the causes of spatial segregation and unequal access to opportunities at the intra-urban level (OECD 2018a). This chapter closes this gap by identifying the most pressing dimensions that contribute to in- creasing disparities within Colombia’s cities. It also assesses the targeting methodology of the Departments per region stratification system in Colombia and proposes including socioeconomic characteristics in the methodology to increase the efficiency of resource allocation and to help reduce intra-ur- ban inequality. FIGURE FIGURE 5.2. Regions in Colombia FIGURE 5.2. Regions Used 5.2. Regions in Colombia in this Chapter in Amazonia Used in Colombia Used this Chapter in this Chapter Amazonas, Caqueta, Gua Departments per region Caribe Departments per region Departments per region Atlantico, Magdalena, Bo Amazonia Amazonas, Amazonia Caqueta, Guainia, Guaviare, Amazonas, and Guainia, Caqueta, Putumayo Vaupes Guaviare, Putumayo and Vaupes Central Boyaca, Cundinamarca, Amazonia Amazonas, Caqueta, Guainia, Guaviare, Putumayo and Vaupes Caribe Atlantico, Caribe Magdalena, Bolivar, Cordoba, Atlantico, Magdalena, Sucre, Cesar andBolivar, Cordoba, Sucre, Cesar and La Guajira La Guajira Caribe Atlantico, Magdalena, Bolivar, Cordoba, Sucre, Cesar and La Guajira Central Boyaca, Central Cundinamarca, Tolima, Huila Boyaca, and Cundinamarca, Bogota D.C. Tolima, Huila and Bogota D.C. Central Boyaca, Cundinamarca, Tolima, Huila and Bogota D.C. Eje Cafetero y Antioquia Eje Cafetero Caldas, y Antioquia Quindio, Caldas, Quindio, Risaralda and Antioquia Risaralda and Antioquia Eje Cafetero y Antioquia Llanos y Orinoquia Pacífico Eje Cafetero y Antioquia Llanos y Arauca, Pacífico Cauca, Orinoquia Casanare, Llanos y Orinoquia Caldas, Quindio, Risaralda and Antioquia Arauca, Casanare, Meta and Vichada Meta and Vichada Arauca, Casanare, Meta and Vichada CaucaChocó, Nariño and Valle del Cauca Cauca, Chocó, Nariño and Valle del Caldas, Quindio, Risarald Pacífico Cauca, Chocó, Nariño and Valle del Cauca Santanderes Seaflower Llanos y Orinoquia Santanderes Santander Santanderes Seaflower San Seaflower and Norte de Santander Andres, Providencia and Santa Santander and Norte de Santander Santander and Norte de Santander San Catalina. Arauca, Casanare, Meta Andres, Providencia and Santa Catalina. San Andres, Providencia and Santa Catalina. Source: Source: Regions considered in the Plan Regions Nacional Pacífico considered de Desarrollo in the Plan Nacional de Desarrollo 2018–2022. 2018–2022. Source: Regions considered in the Plan Nacional de Desarrollo 2018–2022. Cauca, Chocó, Nariño an Santanderes Santander and Norte de Seaflower San Andres, Providencia 57 Regional inequality: high and persistent along different scales and dimensions BUIL D ING AN EQ UITABL E S O CIETY IN CO LO M BIA | T erritorial Inequalities in Colombia Colombia has different types of territories, each with its own specific challenges and de- velopment needs. Territorial imbalances have persisted for a plethora of reasons. An extreme- ly uneven topography, accentuated by the three mountainous ranges that cross the country from north to south, naturally isolates certain areas and perpetuates the persistent lack of connectivity between regions. Moreover, historically poor transport infrastructure has result- ed in the exclusion of periphery areas around cities like Bogotá (Hernandez and Titheridge 2016; Hernandez and Dávila 2016). Moreover, natural hazards can affect some territories more than others. For example, 64.7 percent of the 3.2 million people affected by La Niña 2010–2011 (which increased rainfall) were located in rural areas, with 2.5 times as many homes destroyed in rural areas than urban areas.5 Furthermore, Colombia is among the top five most unequal countries in the world in terms of land concentration (see box 5.2) (Cuesta and Pico 2020b), with 81 percent of private land concentrated in just 1 percent of farms (Guereña 2017). Lack of information and limited institutional capacities contribute to perpetuating inequality in access to land and exacerbate insecurity of land tenure for the most vulnerable. Twenty eight percent of the country does not have cadastral records, and where these do exist, they are likely to be outdated.6 Furthermore, a long period of armed conflict has destroyed physical, human, and social capital, especially in rural areas (see box 5.2) (Sánchez-Cuervo and Aide 2013). Box 5.2. The Impact of Armed Conflict on Territorial Inequalities in Colombia Territorial inequality in Colombia is closely linked to the dynamics of the armed conflict that took place for over six decades. Historically, inequality has been accentuated in those territories most affected by the conflict, experiencing unequal access to productive factors, in particular land (Gáfaro, Ibañez, and Zarruk 2012). These predominantly rural and peripheral areas have had a weak government presence, characterized by high levels of informality in property own- ership and outdated cadastral information, limiting the safeguard of land property rights and facilitating their illegal vio- lent appropriation. This dynamic has been further fueled by the rise of illegal economic activities, which required control of certain territories for the production and trafficking of goods. These characteristics have facilitated the persistence of a vicious cycle in which disputes arise over territorial control, followed by high rates of forced displacement, abandon- ment or stripping of movable or immovable property, and the subsequent loss of access to productive economic and social capital by the rural population. This is how decades of forced displacement and land dispossession in rural areas has increased the concentration of production factors, intensifying the social, economic, and political exclusion of the poorest populations (see figure below). Forced Displacement Victims and Land Gini for Colombia The cycle of inequality, violence, displacement, and land dispossession has radically transformed the 800000 country’s demographics, economy, and living con- 0,885 0,885 ditions. The growth of armed conflict in the mid-1990s 700000 0,88 600000 rapidly increased population displacement to urban ar- 0,878 0,875 500000 eas, reaching more than 6 million internally displaced 400000 0,865 persons—the second largest number of displaced per- sons in the world after Syria. The Colombian country- 300000 0,855 side experienced a massive exodus, causing a serious 200000 0,845 decline in rural productivity and accelerating its institu- 100000 0,8403 tional weakening. Forced displacement has caused sub- 0 0,835 1988 1997 2002 2009 2010 stantial declines in household well-being due to the loss Forced displacement Gini of land ownership of assets, destruction of social networks, and precarious economic conditions in the receiving municipality. Dis- Source: For forced displacement, Unidad de Víctimas (https://cifras.unidadvictimas.gov- placement has also impacted poverty and inequality in .co/Reporteador); for land Gini, Instituto Geográfico Agustín Codazzi - Atlas de Distribu- ción de la Propiedad 2012, considering different authors for years 1989, 1997, and 2002. urban areas, creating further challenges for cities (Gáfa- ro, Ibañez, and Zarruk 2012). Forced displacement and armed conflict have created a vicious circle of inequality in rural territories, particularly for those that are even “more rural,” the ones farthest from opportunities. For most rural people in Colombia, land is the only productive factor they will be able to access. Hence, forced displacement and internal conflict have reinforced vulnerability for those affected, prolonging the incidence of inequality (Ibañez and Querubin 2004). In rural areas, when people lack land as an initial endowment, it becomes more difficult for them to access the same economic opportunities and good-quality public services than for someone with land. This limits their possible sources of income, making them less likely to ever escape from their vulnerable conditions, particularly in those “more rural” territories (Ibañez and Quer- ubin 2004). When integrating inequality into human development indicators, the well-being of the population is lower. 7 In addition, poor land tenure in Colombia has widened the gap between rural and urban areas. The patterns of inef- ficient land use, high concentrations of property, and informality that characterize land tenure in Colombia magnify ur- ban-rural inequalities. The rise of violence and illicit economic activities, which initiated free competition for abandoned or stripped-up land, made it more difficult for thousands8 of forced displacement victims to return home. High rural in- formality in land tenure and delays in cadastral updating also made it more difficult to identify and safeguard property rights. As a result, economic interests in the abandoned lands increased, progressively expanding the concentration of productive land into the hands of only a few people. The land dedicated to agriculture continues to expand in a disor- derly way due to the growing demand for legal and illegal mining, illicit economies, deforestation, or large extensions for agribusiness.9 The tenure structure has led to conflicts in land use, low taxation, and the unequal use of incentives designed to boost the sector. Small and medium-sized rural producers cannot develop economies of scale, while large landowners keep land unproductive to capture a greater value from its future sale or for illegal activities. In Colombia, successful public policy needs to consider the armed conflict, the displaced population, and the il- legal economies. Local governments need to be strengthened to address their multidimensional challenges. The national territory is heterogeneous, with complex interagency coordination and cross-sectoral dialogue, resulting in a challenging allocation of resources and implementation processes that can hinder the success of the programas de de- sarrollo con enfoque territorial (PDETs) (development programs with a territorial focus). The targeting of public resources is strongly centralized and driven mainly by national entities. Outdated information and weak local capacities make planning more difficult for rural territories, resulting in the poor use of resources. The PDETs seek to address these chal- lenges and enable a better allocation of spending. Territorial development has been an explicit policy objective over time. One of the most important changes to the Colombian Constitution in 1991 was fiscal and administrative de- centralization. The goal was to reduce regional disparities by conferring more expenditure au- tonomy on subnational governments. Since then, the government has implemented several programs aimed at reducing territorial gaps in the country.10 More recently, the Plan Nacional de Desarrollo 2018–2022 put forward a chapter on Pacto por la descentralización: conectar ter- ritorios, gobiernos y poblaciones, with the clear vision of “empowering regions and connecting rural areas to foster equitable development”(in Spanish, regiones empoderadas y zonas rura- les conectadas para un desarrollo con equidad territorial). The Colombian government has undertaken an ambitious territorial development agen- da that has led to important advances. Through the efforts of the Misión del Sistema de Ciudades,11 the Misión para la Transformación del Campo,12 and more recently, the Misión de Crecimiento Verde13 and the Misión de Decentralización,14 it has carried out comprehensive reviews of recent trends and key constraints in urban and rural areas. These exercises em- phasize the importance of moving beyond the traditional rural-urban classification typically used in planning instruments toward one that recognizes the diversity of places, with the fi- nal objective of bridging the gaps in living standards across regions. These efforts to adopt a territorial approach have led to important advances, such as putting forward the planes de ordenamiento territorial (POTs) modernos (territorial development plans), support for a mul- tipurpose cadaster, instruments for designing programas de desarrollo con enfoque territorial (PDETs), or development programs with a territorial focus,15 and a more recent effort to move the decentralization agenda forward. Furthermore, in response to the COVID-19 pandemic, the government instituted a pack- age of comprehensive emergency response measures to save lives, protect the poor and vulnerable, and sustain firms. Key components of the response program include ensuring access to infrastructure services for the poor and vulnerable through targeted and tempo- rary electricity and water tariff deferrals, mandated reconnection of previously disconnected households to water and electricity services, and temporary suspension of tolls on intercity roads to facilitate the supply of critical goods. On March 21, 2020, the government created the Emergency Mitigation Fund (Fondo de Mitigación de Emergencia [FOME]) to finance additional spending associated with the emergency response. The size of the FOME, COP 21.4 trillion, represents approximately 2 percent of GDP. Colombia is well positioned to reduce territorial inequality by better targeting invest- ments to overcome its infrastructure development gaps. Despite recent progress, the in- frastructure financing gap in Colombia remains large. The G20 Investment Hub estimates a financing gap of US$339 billion through 2037 that includes financing needs in transport, en- ergy, water, and telecoms infrastructure. This amount represents an additional 3–4 percent of GDP spending in infrastructure annually relative to current levels.16 And although infrastruc- ture and financing gaps are often assessed at the national level, their implications are local, with wide variations in existing infrastructure and needs across the national territory. Prior to the COVID-19 crisis, Colombia ranked 108th out of 144 countries in terms of infrastructure quality,17 and overall infrastructure investment needs were estimated at US$85 billion,18 sub- stantially higher than historic levels of infrastructure spending. To address these challenges, the country put in place a strong regulatory public-private partnership (PPP) framework un- derpinned by a sound contingent liability management system, which helped build a good track record of attracting private investment. Despite these efforts, regional inequalities persist in Colombia. The most vulnerable region in Colombia in terms of socioeconomic conditions is Amazonia, which is almost twice as vulnerable as the Central region. To measure vulnerability, this section uses four socioeconomic dimensions (demographics, quality housing, education, and labor),19 using data from the Colombian Population Census 2018 (DANE 2018a). After Amazonia, the most vulnerable regions are Caribe and Seaflower (see figure 5.3b). Amazonia is most vulnerable in these five variables in the following order: no internet connection, a non-educated household head, no tertiary education, no natural gas, and demographic dependency.20 The Central re- gion, followed by Eje Cafetero y Antioquia and Santanderes, is the least vulnerable region in Colombia and also the one with the highest concentration of the population (see figure 5.3a). However, it is interesting to see that except for the availability of natural gas for domestic use, both regions (Amazonia and Central) share the same top vulnerabilities, although in different order and at different levels. The top five vulnerabilities for the Central region lie in the high levels of a non-educated household head, the lack of tertiary education, high demographic dependency, no access to internet, and with a tie, high economic dependency and high num- ber of female-headed households. FIGURE 5.3. Extremes in Socioeconomic Vulnerability A Regional Population, 2018 B Regional Vulnerability, 2018 Seaflower Seaflower Caribe Caribe Eje Caf. Santanderes Eje Caf. Santanderes and Antioquia and Antioquia Llanos and Orinoquia Llanos and Orinoquia Central Central Pacifico Pacifico Amazonia Amazonia Population 2.00 mln Av. Vulnerability 4.00 mln 6.00 mln 0.20 8.00 mln 0.25 10.00 mln 0.30 12.00 mln 0.35 Source: DANE (2018a). Note: The socioeconomic variables are standardized to guarantee that observations take values ranging from 0 to 1 and that all of the selected variables move in the same direction, where 0 is “no vulnerability” and 1 is “total vulnerability.” The map on the right shows the average of all socioeconomic variables considered in the analysis by region. FIGURE 5.4. Unmet Basic Needs of the Indigenous and Additionally, Amazonia is the region with the highest share of indigenous people, one Afro-Descendant Populations of the population groups with the greatest vulnerability in Colombia. It is also the most A Indigenous population ethnically and linguistically diverse region in the country. Although Caribe and Pacifico have the highest shares of indigenous people across the country, with 43 percent and 32 percent, respectively, Amazonia holds a higher share as a proportion of its total population, with 20 100 percent of its population identifying themselves as indigenous. In Colombia, as in many coun- tries in Latin America and the Caribbean (LAC), indigenous populations experience in general 80 higher levels of poverty, segregation, and vulnerability.21 Across municipalities, it can be seen that those that have the highest number of indigenous people also experience higher levels of Unmet Basic Needs 2018 60 unmet basic needs (figure 5.4a). In addition to the low coverage of basic services, indigenous people in Amazonia have been caught in the crossfire of the armed conflict, which has result- ed in physical violence, forced displacement, and growing encroachment on their land, which 40 is vital to their economic well-being and cultural survival. And although Amazonia holds over 120 indigenous resguardos,22 these territories continue to be affected by deforestation, the 20 expansion of illegal crops, mining and oil extraction, cattle ranching, and the unplanned con- struction of roads. All of these factors negatively affect biodiversity, hinder local subsistence 0 economies, and restrict indigenous peoples’ right to manage their resources and environment 0 20 40 60 80 100 on their own terms, as national and international legal rights frameworks affirm. Unmet Basic Needs 2005 Pacífico, on the other hand, is the region where half of Colombia’s Afro-descendants live, B Afro-descendant population representing 22 percent of its population. The main vulnerabilities in the Pacífico region involve education, with three out of the five most vulnerable socioeconomic characteristics 100 related to educational deficiencies. A non-educated household head, no tertiary education, no internet, demographic dependency, and no high school23 are the five most vulnerable so- cioeconomic characteristics in that region, in that order. This vulnerability has strong historic 80 roots. The Chocó department, which stretches from the Pacific coast to the Atlantic, bordering Unmet Basic Needs 2018 with Panama, was populated by West African slaves in the 16th and 17th centuries following 60 the discovery of gold deposits. After slavery was abolished and the gold rush receded in the mid-19th century, Chocó fell behind on many socioeconomic indicators. Today, over 80 per- 40 cent of the urban population in Chocó lives in poverty. Only 22 percent of rural Afro-descen- dant households have access to water and only 45 percent have electricity (in contrast to the national average of 85 and 95 percent, respectively).24 Illiteracy is three times higher than the 20 national average, and the region lacks adequate infrastructure, transportation, and basic ser- vices. Chocó has also been overwhelmingly affected by the armed conflict, which has led to 0 0 20 40 60 80 100 mass displacement, forced labor in mines and on coca plantations, and other human rights Unmet Basic Needs 2005 violations (Friere et al. 2018). Afro-descendants in Colombia are mostly concentrated in urban areas, and similar to Source: Census 2005 and DANE (2018a). Notes: DANE (2005) and (2018a), with the size of the circle showing the share of indigenous municipalities with higher shares of indigenous population, unmet basic needs have de- and Afro-descendant populations, respectively, in each municipality in Colombia; with clined less between 2005 and 2018 in cities with higher shares of this group (see figure circles in grey having a percent higher than 0 percent but lower than 0.01 percent, in orange those with a percent higher than 0.01 percent and lower than 1 percent, and in red 5.4b). National averages might not reflect the situation of lagging regions like Pacifico nor those cities with a percent higher than 1 percent. Cities with 0 percent are not in the graph. capture other forms of exclusion that Afro-descendants experience within urban areas, such as exposure to crime and violence, discrimination in the labor market, or a tendency to reside in slums. There are also notable differences in the quality of services received that averages do not account for. In slums, for example, Afro-descendant households tend to have access to only sporadic and low-quality services, including poorer education and inadequate transpor- tation or public spaces affected by environmental pollution, among others (see box 5.3). Box 5.3. The Exclusion of Afro-Colombians About 5 million Colombians identify as Afro-descendants, slightly over 10 percent of the national population. Af- ro-Colombians are a diverse group, comprising people that identify as black, mulatto, and Afro-descendants, as well as Raizal from San Andres and Providencia and Palenquero from San Basilio (two ethnic groups that have a distinct cultural and linguistic identity). As in the rest of the LAC region, Afro-Colombians face cumulative disadvantages, unequal op- portunities, and a lack of respect and recognition. Although they predominantly live in urban areas (72 percent), where they enjoy better access to services, they are twice as likely to live in slums compared to non-Afro-descendants. They are underrepresented in decision-making positions (in both the private and public sectors) and are disproportionately affected by poverty, which give an ethno-racial face to exclusion in Colombia. Based on a recent World Bank report, about 41 percent of Afro-descendants were poor in 2015 compared to 27 percent among non-Afro-descendants.25 Households headed by Afro-descendants were 6.2 percent more likely to be poor (holding all else constant), and 13 percent more likely if they are located in rural areas like the Pacific coast (Friere et al. 2018). Such poverty rates can be partially attributed to the way in which Afro-Colombians are integrat- ed into the labor market. In fact, in 2018, Afro-descendants had higher levels of unemployment (14.3 percent) than non-Afro-descendants (10.1 percent), and they were slightly more likely to have informal jobs that often lacked legal benefits, such as social security (DANE 2018a). Afro-descendants also experience discrimination biases when looking for work, occupational segregation in their workplaces, and income glass ceilings (limiting their chances of social mobility). In fact, when comparing workers with the same level of education, age, gender, marital status, experience, work sector, and household characteristics but of different ethno-racial origin, Afro-Colombians tended to earn 5 per- cent less than their peers in 2015.26 Afro-descendants in Colombia are less likely to complete education levels, experiencing racial discrimination in schools. When comparing households in similar socioeconomic conditions for 2015, Afro-Colombian children are 2.6 percent less likely to complete primary education and 3.5 percent less likely to complete secondary school.27 Moreover, in 2018, the completion of tertiary education was lower for Afro-descendants (21.3 compared to 27.8 percent) (DANE 2018a). Yet, there are other, more subtle ways of exclusion that limit educational quality and the potential of education to lift Afro-Colombians out of poverty. Afro-descendant children have a higher probability of being below the age-appro- priate grade and face a higher likelihood of dropping out. School contexts in Colombia are also vulnerable to racial dis- crimination, including prejudice-based representations in teaching materials or inappropriate classroom interactions. In the past three decades, Colombia has taken significant steps to recognize Afro-descendants through the inclu- sion of ethno-racial variables in official statistical records and legislative reforms and the creation of institutional spaces for the formulation of policies for ethno-racial equality. Law 70 of 1993, for example, protects the rights of rural comuni- dades negras (black communities) of the Pacific coast, including the right to own their collective territory, maintain ru- ral-based traditional economies, and develop in ways that are appropriate to their cultural identity and social practices. It also mandates the inclusion of Afro-Colombian history and culture in public school curricula and reserves two seats for Afro-Colombian representatives in the National Congress. Nonetheless, this legal and policy framework has been less effective in addressing the chronic absence of opportunities and services that affects rural Afro-descendant com- munities, as well as their high levels of poverty and vulnerability. It has also been less successful in fulfilling the needs of urban Afro-descendants, who demand policies of racial equality that tackle important issues, such as employment discrimination, income disparities, and exposure to crime and violence. Closing the socioeconomic gaps that affect Af- ro-descendants, however, will be critical to meeting Colombia’s territorial inclusion goals. Source: Freire et al. (2018). Vulnerability in one socioeconomic characteristic does not happen by itself, as the same regions tend to be vulnerable and far from opportunities across multiple dimensions. The quality of housing, measured by connection to services and overcrowding, presents the highest gap among regions. No natural gas, no sewerage, and no piped water are the three socioeconomic characteristics along which regions differ the most (see figure 5.5a). Although the national average of people with no natural gas halved between 2005 and 2018, at least four out of five lacked access to this service in the Seaflower and Amazonia regions. In the case of piped water and sewerage, there was a slight increase in coverage across the country between 2005 and 2018, though this improvement made the regional gap wider. For the three variables, Seaflower shows the highest levels of vulnerability across regions, and the Central region the lowest. Resilient, safe, and affordable housing provision has been a long-standing priority in Colombia, as it is critical to improving living standards, reducing vulnerability, and increasing equity while also mitigating disaster risks.28 An estimated 5.1 million households suffer from either a quantitative housing deficit (approximately 1.53 million households lack access to a housing unit) or a qualitative one (approximately 3.57 million households live in substandard housing units).29 Since 2005, Colombia has made progress in nearly halving the quantitative deficit, but the qualitative deficit has not decreased, and approximately 23 per- cent of all Colombian households still live in overcrowded and substandard housing. When considering the levels of vulnerability across regions, the three socioeconomic char- acteristics with higher vulnerability are non-educated household head, no tertiary education, and no internet (see figure 5.5b). Despite a national improvement in non-educated house- hold head and no tertiary education between 2005 and 2018, the regional gap increased for non-educated household head, and still, seven out of 10 people in Colombia do not have any kind of tertiary education.30 Again, the Amazonia region is the most vulnerable along these three variables, with the Central region experiencing the least vulnerability. This clearly shows where the government can provide the most support when trying to reduce regional inequal- ity and vulnerability across regions. FIGURE 5.5. Vulnerability across Dimensions A The three socioeconomic characteristics with the greatest gap between regions No natural gas No sewerage No piped water Seaflower Seaflower Seaflower Caribe Caribe Caribe Santanderes Santanderes Santanderes Eje Caf. Eje Caf. Eje Caf. and Antioquia and Antioquia and Antioquia Central Llanos and Orinoquia Central Llanos and Orinoquia Central Llanos and Orinoquia Pacifico Pacifico Pacifico Amazonia Amazonia Amazonia 0.20 0.20 0.10 0.40 0.40 0.20 0.60 0.60 0.30 0.80 0.80 0.40 1.00 0.50 0.60 B The three socioeconomic characteristics with the highest levels of vulnerability across regions Non−educated head No Bachelor degree No internet Seaflower Seaflower Seaflower Caribe Caribe Caribe Santanderes Santanderes Santanderes Eje Caf. Eje Caf. Eje Caf. and Antioquia and Antioquia and Antioquia Central Llanos and Orinoquia Central Llanos and Orinoquia Central Llanos and Orinoquia Pacifico Pacifico Pacifico Amazonia Amazonia Amazonia 0.70 0.70 0.50 0.75 0.75 0.60 0.80 0.80 0.70 0.80 Source: DANE (2018a). 58 Leveraging the potential of cities to reduce rural poverty B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | T erritorial Inequalities in Colombia When cities are well connected to the surrounding areas, they can help to form and be a part of strongly linked territories with environmental, social, and economic interac- tions that are mutually beneficial for both urban and rural areas. Urban-rural linkages are those interactions between urban and rural areas that straddle beyond administrative boundaries, within a rural-urban continuum.31 OECD members, including Colombia, have in- ternalized the notion of functional territories, where urban cores interact with different levels of rurality through different channels (OECD and EC 2020; RIMISP, IDB, and UNDP 2015). The literature proposes several channels, some more important than others, through which cities may impact territorial development.32 For instance, good physical connections (or a lack of natural barriers) between urban and rural areas can enable productive activities, access to goods and services, and social interaction. The development of highways and road networks and the accessibility provided by natural geographic conditions represent enormous poten- tial for territorial development (CEPAL 2020). Tranport networks can also ensure the integrity of territories, allowing the provision of humanitarian aid in cases of emergency, such as the COVID-19 pandemic. Rural territories can sometimes be physically isolated by poor access to roads or by rough natural barriers, explaining in some cases their levels of poverty (CEPAL 2020). Moreover, the interaction between urban and rural areas requires adequate institution- al capacities and governance arrangements (UN-Habitat 2017). When cities are strengthened with strong local capacities, clear financial mechanisms, and social cohesion, they can better plan and allocate resources to foster territorial development. Cities can play an important role in linking the rural hinterland to better services, bigger markets, and economic opportunities, and a functional definition of territories based on commuting flows can contribute to better understanding these interactions. Urban-ru- ral linkages are becoming a focus of interest among policy makers and researchers (Tacoli 1998; UN-Habitat 2017). Carriazo Osorio, Lozano Gracia, and Tello Medina (forthcoming) use commuting patterns in Colombia to define functional territories around urban areas that are consistent with a characterization of labor markets.33 These functional territories aggregate municipal commuting patterns using census 2005 data, thus defining territories as a geograph- ic, political, and social space between the municipal and departmental levels.34 Analyzing the urban-rural gradient based on the functional characteristics of territories can help understand the dynamics of territorial development (see figure 5.6). FIGURE 5.6. Functional Territories in Colombia Source: Berdegué et al. (2015). In Colombia, the analysis suggests that cities do help reduce multidimensional poverty in their functional territories, and this effect increases with city size. By analyzing the changes in poverty, inequality, and income (expenditures) in the above functional territories between 2005 and 2018, this work finds that cities are reducing multidimensional poverty in their territories of influence (for more on this see Annex 4). This work shows that the increase in multidimensional poverty decreases as the urban core size of the functional territory is larg- er. In a territory with a core of 10,000–50,000 people, the multidimensional poverty change between 2005 and 2018 decreases by 0.019 percentage points, and in the case of territories with the largest core, an increase in one unity of this variable decreases the variation of the multidimensional poverty measure change by 0.04 percentage points. These results suggest that urban centers play a crucial role in poverty reduction when poverty is measured in terms of such dimensions as housing, education, health, or other indicators of quality of life. However, between 2005 and 2018, cities in Colombia did not support a reduction in mon- etary poverty. Colombian cities do help reduce multidimensional poverty in their territories but this is not the case for variables linked to income. These results indicate that though cities are providing better socioeconomic conditions in terms of education, housing, and public ser- vices, among others, they are not reducing monetary poverty.35 This suggests that the larger the urban core, the greater the monetary poverty growth in the territory. This effect increases with the city size: in territories with a core of 10,000–50,000 people, poverty change increases by 0.05 percentage points. In the case of the largest urban core, an increase in one unit of this variable augments the variation of monetary poverty by 0.13 percentage points, thus more than doubling the effect of the smallest territories. These results can be linked to the worldwide notion that cities offer better socioeconom- ic conditions than rural territories (reducing multidimensional poverty), but quality of life is expensive (having a less immediate impact on the reduction of monetary poverty). Between 1993 and 2005, cities in Colombia were contributing to income growth and reducing monetary poverty (Berdegué et al. 2015), but this is not the case anymore. One of the possible explanations is the higher cost of living in cities. Between 2005 and 2018 (intercensal period), the consumer price index increased by 70.35 percent,36 reflecting the cost of living increase that was linked to the growth of cities and the increased demand for products and services. For instance, in the case of Bogotá, rising living costs are associated with high housing prices due to the increase in demand driven by the growth in population.37 Transport costs or costs for other services can be driving the cost of living up when cities grow without the needed infrastructure. Reductions in the cost of living in cities can be achieved in a more generous welfare state through a reduction in poverty and inequality levels more generally (DNP 2014). To guide public policy and investments and enable cities to better extend the benefits of economic success concentrated within their boundaries, it is important to identify the mechanisms by which cities influence their territories, that is, strengthen the linkages between urban and rural areas. In other words, what are some of the urban-rural linkage mechanisms through which urban centers may help improve the well-being of surrounding territories in Colombia (measured by the dynamics of poverty, inequality, and income)? When the right policies are in place to foster strong linkages between urban and rural areas, the eco- nomic success of cities can be further leveraged to reduce poverty in neighboring rural areas and reduce inequality. In Colombia, estimates suggest urban-rural linkages (or channels) are strong in func- tional territories.38 When analyzing the correlation between these channels and the urban core sizes of the territories, it was found that all the analyzed channels have a positive and significant relationship with the largest core size of more than 370,000 inhabitants. In other words, along the rural-urban gradient, metropolitan regions in general present higher levels of municipal and fiscal performance, higher index of modern cities, and important levels of accessibility as measured by the density of roads. The terrain heterogeneity variable suggests that bigger territorial cores are located in hilly and very heterogeneous geographic areas. Cities with highly developed local government institutions and fiscal and managerial performance can foster territorial development in their rural hinterland. Cities of inter- mediate size and above have a crucial role in territorial development through better institu- tional and fiscal performance. These results highlight how important it is for municipalities and territories to strengthen their own economic, institutional, and fiscal capabilities, as these improvements reduce both monetary and multidimensional poverty in their functional areas: an increase of one Municipal Performance Measurement (Medición De Desempeño Municipal, or MDM in Spanish) unit reduces monetary poverty change by 0.002 percentage points and multidimensional poverty change by 0.003 percentage points. The Modern Cities Index (Índice de Ciudades Modernas, ICM in Spanish) reduces monetary and multidimensional poverty as well, meaning that an increase of one ICM unit reduces monetary poverty change by 0.008 of one percentage point and multidimensional poverty change by 0.004 percentage points, and an increase of one unit of the fiscal performance index decreases multidimensional poverty change by 0.002 percentage points. If governmental capacity were to be strengthened, the effects of public policy on inclusive territorial development would be greater. Cities can also provide important services (as captured by the ICM) that may help shape favorable territorial development dynamics. Furthermore, good connectivity in territories has a favorable effect on monetary and multidimensional poverty reduction. Results here suggest that connectivity and public in- vestments in road networks have a strong impact on poverty reduction. Nonetheless, these impacts may be offset by the challenges that terrain heterogeneity may impose on territorial development dynamics, as the parameter estimates for the index of territorial roughness sug- gests.39 The road density index generates a negative and strong effect on both monetary and multidimensional poverty change. An increase of one unit of this variable leads to a reduction in monetary poverty growth of 6.1 percentage points and in multidimensional poverty mea- sure change by 5.9 percentage points. On the contrary and as expected, the roughness index has a positive effect on monetary and multidimensional poverty change: an increase of one unit of this variable is associated with an increase of the variation in monetary poverty by 0.025 percentage points and in multidimensional poverty measure change by 0.016 percent- age points. These results may signal the effect of lack of accessibility in very heterogeneous and hilly terrains. 59 Cities with high inequality and spatial segregation B U I L D I N G A N EQU I TA B L E S OCI ETY I N CO LO M BIA | T erritorial Inequalities in Colombia Overall, cities offer economic opportunities, proximity to other people, and physical access to public goods and services for a better quality of life (OECD 2018a). But cities in Colombia do not offer those benefits to everyone, and when looking at intra-urban conditions, Colombia experiences high inequality. Generally speaking, cities are engines of economic growth and productivity in most if not all countries. Large cities in Latin Ameri- ca account for 81 percent of the region’s population and 76 percent of GDP, while large cities in Colombia contribute 80 percent to national GDP (Galvis-Aponte et al. 2008 and Manyika et al. 2012). However, when calculating a Gini coefficient for socioeconomic characteristics in Colombian urban areas,40 this analysis finds that it ranges from 0.13 to 0.56. To give some context, the Gini (of income) for the least unequal countries in the world41 is around 0.25. Most municipalities in Colombia have Gini coefficient values above 0.4 (848 out of 1,122 municipali- ties that account for 95 percent of the urban population), and only a few have Gini coefficients below 0.2 (28 municipalities). If this is examined from a regional perspective, it can be seen that the center of the country is the most unequal (see figure 5.7). FIGURE 5.7. Intra-Urban Inequality in Colombia 40 30 20 10 Gini coeff.* Gini coeff.* 0.42 0.44 0.20 0.46 0.30 0.48 0.40 0.2 0.3 0.4 0.5 0.50 0.50 Gini coefficient* 0.52 Missing Source: Authors’ calculations, using socioeconomic data from DANE (2018a). Note: The Gini coefficient* was calculated for each one of the socioeconomic variables, using the set of blocks belonging to each region or municipality. The overall Gini is a simple average of all the Gini coefficients. To address intra-urban inequality in Colombia, the first step is to identify its main driv- ers. This work analyzes the socioeconomic characteristics at the block level of 1,102 munici- palities using the Colombian Population Census 2018 (DANE 2018a). The data used cover four dimensions to represent different types of assets, such as human and social capital, and their intensities across urban areas: demographics, education, quality of housing, and labor.42 Vul- nerability indicators were then obtained for each dimension that would allow for the classifi- cation of the blocks into six clusters, where the higher the cluster label number, the lower the level of socioeconomic vulnerability.43 Six categories were picked to mimic the current stratifi- cation classification in Colombia. Annex 4 describes how the population is distributed among these six clusters by region. Education is one of the main drivers of intra-urban inequality in Colombia, with illiteracy and school absence rate having the greater disparities.44 Considering the four dimensions, education has the highest levels of vulnerability, which is also the case for five out of the eight regions in the country, namely, Caribe, Central, Llanos y Orinoquia, Pacífico, and Santande- res.45 Nationwide, out of the four variables considered in this dimension, the two that have higher vulnerability across clusters are (i) the percentage of households with a household head without higher education and (ii) the percentage of people without tertiary education. This is important since vulnerability in both variables is linked to effects on employability and remu- neration and ultimately to linkages with poverty. Moreover, despite the fact that the illiteracy and school absence rates have overall lower levels of vulnerability, there is greater inequality in these variables, pointing to places and population that are still far from basic education. As seen in previous chapters, early childhood and middle education are where the greatest gaps in access are seen at the national level. Quality of housing is the other main driver of intra-urban inequality, with the greatest vulnerability gap between the socioeconomic clusters. In Colombia, quality of housing, measured by connection to services and overcrowding, is the dimension with the greatest gap between clusters one and six (see figure 5.8). Out of the 5.1 million households in a state of vulnerability, almost 4.4 million are in urban areas.46 The relevance increased when think- ing of the 1.8 million Venezuelan migrants who have relocated to Colombia. This means that those blocks in cluster one have a significantly worse quality of housing than those in cluster six. This is the case as well for five out of the eight regions in Colombia, namely, Amazonia, Central, Llanos y Orinoquia, Pacifico, and Santanderes.47 Also, people in cluster one face at least double the vulnerabilities in household conditions than those faced by people in the other clusters, regardless of the size of the city. Nationally, no electricity, no piped water, no garbage collection, and no sewage are the four variables that contribute the most to intra-ur- ban inequality in housing and public service conditions in Colombia—they are the ones with greater inequality out of the seven variables included in this dimension. This indicates that there are still some households across the country (within blocks classified mainly in cluster one) that have close to zero access to these services. FIGURE 5.8. Contributors to Intra-Urban Inequality Demographics Education Housing Labor 1.00 0.75 0.50 0.25 0.00 1 2 3 4 5 6 Source: Authors’ calculations, with data from the DANE (2018a). Note: Variables for each dimension are standardized to guarantee that observations take values ranging from 0 to 1 and all selected variables move in the same direction. The average of the standardized variables is calculated to obtain vulnerability indicators for each dimension. These indicators are then used to classify the blocks into six clusters using the k-medians clustering method. The higher the cluster label number, the lower the level of socioeconomic vulnerability. Details in Annex 4. FIGURE 5.9. Intra-Urban Inequality in Larger As in many countries, bigger, more dynamic cities in Colombia do indeed offer overall versus Smaller Cities better socioeconomic conditions, but they also experience important inequalities. Cities that lead in economic performance, such as Bogotá, Medellín, and Cali, are more unequal than less-dynamic cities like Cúcuta, Ibagué, and Pereira (Duque et al. forthcoming). This pattern 0.5 follows a similar trend to that in OECD countries, where bigger cities are richer but more un- median Gini equal (OECD 2018a). The largest cities in Colombia show lower poverty incidence48 and better socioeconomic conditions compared to smaller cities (see figure 5.9). For instance, the cluster Gini coefficient* 0.4 with the most vulnerable housing conditions (cluster one) in large cities (>350,000) is still less vulnerable than cluster four of small cities (<100,000), with clusters five and six of small cities (the ones with the best housing conditions) having a higher vulnerability than cluster two of 0.3 big cities. In other words, those households with the worst housing conditions in big cities (clusters one and two), have on average fewer deprivations than all the households in small cities. In terms of education, people belonging to cluster six in large cities enjoy the best edu- median AVPI 0.2 cational conditions across the country, while individuals for the most favored clusters in small cities face conditions comparable to clusters two to four in large and medium-sized cities. 0.1 0.2 0.3 However, education is more unequal in bigger cities than in intermediate and smaller cities. Adjusted Multidimensional Poverty Index (AMPI) Most of the population in small, less unequal municipalities in Colombia experience poor socioeconomic conditions. As mentioned, housing conditions in small cities overall experi- Gini coeff.*: Population (millions): 2 4 6 0.5 0.4 0.3 0.2 ence higher deprivations than those in bigger cities. In terms of education, the vulnerability of cluster one is similar regardless of the size of the city; however, clusters five and six of small Source: Authors’ calculations, with demographics and socioeconomic data from DANE cities have on average double the deprivations of clusters five and six of big cities. In terms (2018a). of the labor market, the smaller the city, the higher the level of vulnerability. Small cities are mostly located in the Pacific region, in the south in the Amazonia region, and scattered in the Caribe, Santanderes, and Central regions. Furthermore, intra-urban segregation is found in Colombia’s cities, pointing to exacer- bating inequality and possible poverty traps. Evidence can be found of an intra-urban con- centration of socioeconomic vulnerability (see figure 5.10). Different intra-urban clusters are segregated spatially, indicating that the rich and the poor live separated in space. In divided cities, there are gaps and barriers that produce exclusive spaces and concentrations of dis- advantage (OECD 2018a). Inequality in socioeconomic conditions across different blocks and neighborhoods in Colombian cities can exacerbate existing societal disparities, as the most vulnerable population groups can get cut off from important social networks and opportuni- ties for socioeconomic mobility (Ham et al. 2016). Moreover, those areas with higher socio- economic vulnerability, usually located in the urban borders (suburban areas), tend also to be areas with higher exposure to landslides, flash floods, and floods, where people often build housing that ignores basic construction standards. It has been suggested that more than 50 percent of city growth in Colombia over the past three decades has been of this kind of infor- mal origin, due, among other factors, to the limited ability of the authorities to control illegal urbanization and informality processes, especially in risk areas.49 In 2009, about 65 percent of deficit households were located in high-risk areas.50 FIGURE 5.10. Intra-Urban Concentration of Socioeconomic Vulnerability in Colombian Cities A Housing and education dimensions for the top 5 most populous capital cities High-high High-low Low-high Low-low Non-significant BOGOTÁ, D.C. MEDELLÍN CALI BARRANQUILLA CARTAGENA DE INDIAS LISA Housing LISA Housing LISA Housing LISA Housing LISA Housing 0 2 km 0 2 km 0 2 km 0 2 km 0 2 km BOGOTÁ, D.C. MEDELLÍN CALI BARRANQUILLA CARTAGENA DE INDIAS LISA Education LISA Education LISA Education LISA Education LISA Education 0 2 km 0 2 km 0 2 km 0 2 km 0 2 km Source: DANE (2018a). Note: LISA (local indicator of spatial association) significant clusters of high-high vulnerability (dark red) and low-low vulnerability (dark blue) of socioeconomic characteristics, significance at 5 percent. ... B Housing and education dimensions for the 5 capital cities with the highest Gini High-high High-low Low-high Low-low Non-significant NEIVA TUNJA BUCARAMANGA IBAGUÉ PEREIRA LISA Housing LISA Housing LISA Housing LISA Housing LISA Housing 0 2 km 0 2 km 0 2 km 0 2 km 0 2 km NEIVA TUNJA BUCARAMANGA IBAGUÉ PEREIRA LISA Education LISA Education LISA Education LISA Education LISA Education 0 2 km 0 2 km 0 2 km 0 2 km 0 2 km Source: DANE (2018a). Note: LISA (local indicator of spatial association) significant clusters of high-high vulnerability (dark red) and low-low vulnerability (dark blue) of socioeconomic characteristics, significance at 5 percent. Moreover, there is evidence that when there is vulnerability in one dimension, there is usually vulnerability in several other dimensions. Colombian capital cities show a spatial concentration of socioeconomic vulnerability across all four analyzed dimensions. Other di- mensions of urban life, such as the provision of urban green space and open public spac- es, also show important levels of inequality between different socioeconomic groups (Patino 2020). Urban socioeconomic segregation also happens in European cities, for example, where it is the outcome of a combination of inequality, poverty, and the spatial organization of urban housing markets (Ham et al. 2016). Residential segregation is an undesirable characteristic as it has been linked to several dimensions of inequality, including labor, education, health, and vulnerability (Nijman and Wei 2020). For instance, higher-income urban spaces are asso- ciated with lower vulnerability to natural hazards.51 In Bogotá, there is a relationship between the condition of threats to floods and landslides and the price of land, housing, and rental fees, indicating that renting a home located in a risk area is 6.5 to 21 percent cheaper than renting in a non-risk area.52 Moreover, previous works have highlighted the benefits of having a healthy social mix within neighborhoods regarding crime outcomes (Holin et al. 2003; Pa- tino et al. 2014; Popkin et al. 2004). In Colombia, urban development patterns and housing development and availability have played a central role in deepening such segregation, with the socioeconomic stratification system also reinforcing this state of affairs. 60 Territorial inequality, access to opportunities, and lagging infrastructure BUIL D ING AN EQ UITABL E S O CIETY IN CO LO M BIA | T erritorial Inequalities in Colombia Unequal access to services is an essential aspect of territorial inequality. Accessibility emphasizes the interconnected nature of mobility and land use and provides a framework in which to evaluate the city in terms of its ability to connect its citizens to opportunities.53 When individuals face an unequal distribution of services and economic opportunities with constraints in terms of accessibility, the disparities among social groups can grow deeper. Spatial disconnection to services within a city can have negative effects on different social and economic outcomes (Fay et al. 2005; Geurs and van Wee 2004; Handy and Niemeier 1997; OECD 2018a; UN-Habitat 2013). Better access to services, economic opportunities, and social interactions decrease social inequality for disadvantaged people and increase social welfare through a variety of mechanisms (Ong and Blumenberg 1998), such as better labor market outcomes and lower social segregation.54 It is important to highlight that not all municipali- ties need to have all services; sometimes it is better to invest in neighboring territories to bring the service closer to more people and territories. To measure inequality in access in Colombia, this work calculated accessibility to services55 associated with health, education, and sports at the urban level (details in Annex 4). Residents in most municipalities in Colombia have good access to education facilities, though some capital cities do not. Roughly 80 percent of municipalities in Colombia have a travel distance to preschools, schools, and high schools of less than 1 kilometer (see figure 5.11). Most municipalities in the country have education facilities that are within an accept- able distance for urban spaces of 5 kilometers.56 However, 30 percent of the population in cities has to travel more than 5 kilometers to get to school, and people in some capital cities, such as Medellin, Cali, Cartagena, Santa Marta, Pereira, and Ibagué, also need to travel more than 5 kilometers to access education facilities. FIGURE 5.11. Access to Schools in Colombian Cities 100 75 50 25 0 2 4 6 8 Gini coefficient* 14 12 Saber 11 global score Average distance (m) 10 8 6 Gini coeff.* 4 0 2 5 0 10 ía e l co s er qu o uí s ra 15 re ib oq ale io er on w ia a ífi nt e r lo nt fet Ca nd c az rin nt Ce Missing Pa af y A e ca y O orie Am a Se nt Sa Ej os an Ll Source: Authors’ calculations, using DANE (2018a) for information on the location of people and facilities at the block level. The national street network 2014 was used, which is provided by Procalculo Prosis SAS. See methodology in Annex 4. However, as explained in previous chapters, this does not necessarily mean that every municipality has access to quality education. There are important quality gaps in education between municipalities and regions, with a center-periphery pattern (see figure 5.12). Educa- tional institutions with the best national academic results are concentrated in the capital cities and main urban areas of the Santanderes and Central regions. Bogotá, Paipa, Bucaramanga, Sogamoso, Duitama, Chía, and Floridablanca have average test results that are more than two standard deviations above the national average, with Bogotá having the highest results. In contrast, there are small municipalities in Seaflower, Pacifico, Amazonas, and Caribe whose results are far below the national average. These are municipalities located in rural areas and on the periphery of the country. For instance, in Amazonas, scores from Saber 11, the national standardized assessment for 11th graders, are on average 0.8 standard deviations below the mean of the country, with municipalities like Morichal, San Felipe, and Pana reaching more than two or three standard deviations below the national average. The same happens with several municipalities in Pacífico and Caribe, with results below the national average by more than two standard deviations. FIGURE 5.12. Quality Gaps in Education between Municipalities and Regions in Colombia 300 Saber 11 global score 250 200 150 ía e l qu o uí s o es er ra oq ale rib io er fic Quintiles of Saber 11 on w er nt ia a nt fet cí lo Ca rin nt d az Ce Pa an af y A e ca y O orie Am Se Q1 (lowest scores) nt Ej Sa os Q2 an Ll Q3 Q4 Q5 (highest scores) NA Source: Authors’ calculations, using DANE (2018a) for information on the location of people and facilities at the block level. The national street network 2014 was used, which is provided by Procalculo Prosis SAS. See methodology in Annex 4. Access to health facilities is more unequal in Colombia, with a significant number of mu- nicipalities struggling to access a clinic or a hospital. The median travel distance to health services is 0.84 kilometers, and roughly 56 percent of municipalities present a travel distance of 1 kilometer or less. However, 45 percent of the population in cities has to travel more than 5 kilometers to get to a clinic or a hospital; indeed, 5 kilometers is the international stan- dard distance for urban areas to access health services (see Annex 4). Individuals in big cities, such as Cali, Medellin or Bogotá, need to travel more than 5 kilometers: in Cali and Medellin people must travel on average close to 6 kilometers and Bogotá an average of 12.5 kilometers. Furthermore, this analysis points to 13 percent of municipalities in Colombia (equivalent to more than half a million people) whose populations have to travel more than 15 kilometers to access a clinic or hospital, with people in municipalities like Solita (Caquetá) or Orocue (Casanare) needing to travel more than 100 kilometers. Most municipalities with long distances to health facilities are located in the Pacifico region (see figure 5.13a). Access to sports facilities follows a similar pattern (see figure 5.13b). The median travel dis- tance to these amenities is 0.87 kilometers, and around 55 percent of municipalities require travel of 1 kilometer or less. Fifty percent of the population in cities has to travel more than 5 kilometers to access sports amenities. Several capital cities, such as Medellín, Pereira, Cali, and Santa Marta, require more than 6 kilometers of travel to sports amenities. Moreover, for 11 percent of municipalities, the closest amenity is located between 15 and 203 kilometers away; for Leiva (Nariño), Mapiripan (Meta), Solano (Caquetá), and Cravo Norte (Arauca), the closest facility is at a distance of more than 100 kilometers. Again, the Pacifico region has the highest number of municipalities with larger distances to sports facilities. FIGURE 5.13. Access to Health and Sports Facilities A Distance to health facilities 14 12 Saber 11 global score Average distance (m) 300 10 8 200 6 100 4 2 0 0 0 5 0 30 60 90 120 ía e l ia a ico s er ra re rib uí on qu w 10 nt cíf de oq lo Ca az io Ce Gini coefficient* Pa an af rin nt Am 15 Se nt yA yO Sa o Missing es er al et nt af ie ec or Ej os B Distance to sports facilities an Ll 600 14 12 Saber 11 global score Average distance (m) 10 400 8 6 200 4 2 0 5 0 0 10 0 50 100 150 200 ía e l ia a ico s er ra re rib uí on qu w nt cíf de oq lo Ca 15 az io Gini coefficient* Ce Pa an af rin nt Am Se nt yA yO Missing Sa o es er al et nt af ie ec or Ej os an Ll Source: Authors’ calculations, using DANE (2018a) for information on the location of people and facilities at the block level. The national street network 2014 was used, which is provided by Procalculo Prosis SAS. See methodology in Annex 4. When long distances to facilities are coupled with poor access to transport infrastruc- ture, inequality in large and small cities is intensified. For all facilities, the average distance increases as the size of the city increases. However, there are small municipalities with high travel distances (see figure 5.14), and as mentioned previously, small cities tend also to face worse socioeconomic conditions. In Colombia, as others in other Latin American countries, cities sometimes lack the environment needed to make the most out the benefits that cities can provide. However, Colombia is below the LAC average in terms of average road length in a 100-kilometer radius around cities with at least 1 million inhabitants (Ferreyra and Roberts 2018). Levels or quality of transport infrastructure therefore may not be enough to enable access to services, hindering the quality of life of the population and the productivity of cit- ies. In the United States, for instance, lack of public transport connections between minority neighborhoods and employment centers hinders job opportunities for those residents (OECD 2018a). Transport systems that provide access to jobs and services for everyone lead to im- provements in, for example, maternal health, participation in education, and human devel- opment more widely.57 FIGURE 5.14. Access to Public Services for Small Municipalities Average distance to EDUCATION amenities (km) Average distance to HEALTH amenities (km) Average distance to SPORT amenities (km) Gini Population 1.0 1.0 1.0 0.16 1000 Cumulative proportion population Cumulative proportion population Cumulative proportion population 0.24 2000 0.32 3000 0.8 0.8 0.8 0.40 4000 0.48 5000 0.6 0.6 0.6 0.56 6000 7000 8000 0.4 0.4 0.4 9000 10000 11000 0.2 0.2 0.2 12000 0.0 0.0 0.0 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 Source: For data on demographics, DANE (2018a); authors’ calculations. Note: Size of the cities in terms of population is represented by the size of the circles. From left to right: education, health, and sports. Cities in Colombia are growing without the needed infrastructure, exacerbating terri- torial inequality. Most capital cities present a center-periphery spatial pattern in terms of distance to education, health, and sports facilities. When analyzing capital cities around Co- lombia,58 it can be seen that people living in blocks close to the city center have shorter travel distances to reach their desired urban facilities (see figure 5.15). This is the case when cities start expanding without the needed pairing between infrastructure and services, leaving most facilities in the center of the city. Dense and disorderly development hinders work on access to roads or water, sanitation, sewerage, and drainage (Fay et al. 2017). Additionally, those locat- ed on the periphery, including populations that are displaced due to conflict within Colombia (see box 5.2) or in neighboring Venezuela, are usually the most disadvantaged.59 For example, in Bogotá, 52 percent of the victims of conflict are concentrated in only five of the 20 localities of the city,60 and within them, they are located in the most peripheral and depressed areas. FIGURE 5.15. Cities’ Center-Periphery Patterns in Terms of Access to Facilities Distance to sports (km) Distance to education (km) 10 15 20 25 nd 2 3 4 5 nd 0 2 km 0 2 km Source: Authors’ calculations, using DANE (2018a) for information on the location of people and facilities at the block level. The national street network 2014 was used, which is provided by Procalculo Prosis SAS. See methodology in Annex 4. Note: Distance to sports facilities in Bogotá (left), distance to education facilities in Tunja (right). However, there are some exceptions, as several municipalities have a similar distance to facilities throughout the urban space. For instance, when looking at access to health facilities in Medellin, the intensity of the blocks located at the center (lower distance) is not substantially different from those on the peripheries. This indicates that the proximity to ame- nities of a given type tends to be similar throughout the urban territory. Still, as in most cities, when looking at access to health services in Cali, there is a high intensity at the center, which indicates that the amenities are considerably closer to the center areas than to those locat- ed on the peripheries (figure 5.16a). What is Medellin doing differently? Medellin has actively planned the location of its health facilities, facilitating PPPs in locations where they were most needed.61 Colombian cities can learn from each other how to improve access across the urban space. Nevertheless, a similar distance to amenities across the territory still leaves the issue of quality. For example, when comparing the accessibility maps of the distance and the gravity measure of accessibility (which considers the accessibility from each origin block to the facili- ty/service, which is discounted by the travel impedance from the origin to the destination) to sports facilities in Cali, it can be seen that even if the distance map indicated similar distances throughout the territory, the gravity measure of accessibility map shows more intensity to- ward the center (figure 5.16b). Sport facilities in the center of Cali are either closer to people, or they are greater in quantity, or both. FIGURE 5.16. Spread of Health Facilities in Medellin and Cali A Health facilities are better spread across the urban space in Medellin than in Cali. B Distance and accessibility to sports facilities in Cali Medellín Cali Cali Cali Distance to health (km) Distance to health (km) Distance to sports (km) Access to sports (km) 0 2 4 6 8 10 nd 6 8 10 12 14 18 nd 0 5 10 15 nd 0.1 0.2 0.3 0.4 0.5 0.6 nd 0 2 km 0 2 km 0 2 km 0 2 km Source: Authors’ calculations, using DANE (2018a) for information on the location of people and facilities at the block level. The national street network 2014 was used, which is provided by Procalculo Prosis SAS. See methodology in Annex 4. Accessibility to facilities matter, as is demonstrated particularly in the context of the COVID-19 pandemic, since the poor are experiencing worse consequences from the lock- down and mobility restriction measures. Cities are notorious hotspots of transmission, and to effectively plan interventions, it must be understood how accessible a city is, which indi- cates where physical distancing is easier or harder and the areas that need to be targeted for resource allocation (e.g., increased public hygiene). The unequal effects of the lockdown measures on mobility highlight the trade-offs that the poorest must face in terms of protect- ing their health or their livelihoods. Between the second week of February and the second week of April 2020, high-wealth users were able to reduce their traveled distance almost twice as much as low-wealth users. The maximum distance between a user’s GPS locations on a typical day decreased by 59 percent in Colombia, but the gap in mobility reduction between the richest and poorest deciles was roughly 20 percentage points (Fraiberger et al. 2020). 61 More efficient spending to help reduce intra-urban inequality: BUI LD I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | T erritorial Inequalities in Colombia the strata system Subsidies, such as the ones for servicios públicos domiciliarios, are critical in a time of crisis. In 1994, just after the promulgation of Colombia’s most recent constitution, the coun- try implemented a system of crossed subsidies to fund public utilities (subsidios a servicios públicos domiciliarios) in municipalities, also known as the stratification system.62 According to Law-142, the main principles guiding this instrument are “efficiency, universality, and sol- idarity,” with all inhabitants having the “inalienable right to social security,” according to the constitution. However, when resources are constrained, it becomes crucial to spend them more effi- ciently. In the first semester of 2020, the collection of the municipal property tax as a source of income for municipalities was reduced by 38 percent at the national level. These resources are commonly used by cities to invest in urban development. By the end of 2020, it is projected that public revenues in large and medium-sized cities will have declined by 10 and 16 percent, respectively, due to low collection of such taxes as the industry and commerce (ICA) tax and the property tax. This directly affects the construction of infrastructure in municipalities and the support of urban development across the country. Improving the efficiency of expenditure through the better targeting of subsidies can help reduce intra-urban inequalities without the need for more resources. Recent studies have highlighted the stratification system as a key structural challenge to achieving an effective re- duction of income inequality and poverty (Sepúlveda Rico, López Camacho, and Gallego Ace- vedo 2014; Marcos 2018). Despite its benefits and contribution to the expansion of coverage of basic services, the system has had unintended consequences, such as discrimination and stigmatization (Bogliacino, Jiménez Lozano, and Reyes Galvis 2015) as well as socioeconomic and spatial segregation, that pose a serious challenge to inclusion today. It is important to include socioeconomic characteristics in a new, more effective, and ef- ficient targeting methodology for the stratification system. The results in this subsection support the stream of research that indicates that the inclusion of households’ socioeconom- ic characteristics could contribute to improving efficiency in the use of resources. An analysis has been done for all municipalities in Colombia, identifying multidimensional poor people (using DANE 2018a) and contrasting their characteristics against their respective classification based on the current stratification system. An adaptation of the Colombian Multidimensional Poverty Index (C-MPI)63 was used, since data from the 2018 census do not allow for the replica- tion of all the indicators used in the index. The same methodology, the same dimensions and weights, and 10 out of the 15 indicators in the C-MPI were used. It was also possible to take advantage of the six socioeconomic clusters defined earlier to complement the exploration of potential improvements to the stratification system. Enhancing the stratification system of crossed subsidies with a multidimensional identification of the socioeconomic character- istics of the population would be the appropriate starting point in reducing inequality and achieving financial stability. A comparison of the current stratification system with the adapted MPI categories indi- cates an inclusion error of 66.9 percent.64 This suggests that over 65 percent of households receiving subsidies under the current system should be receiving either a smaller subsidy or no subsidy at all.65 The current stratification system is unable to capture improvements in people’s living conditions and may be affecting individual incentives, preventing optimal de- cisions on social and physical mobility and intensifying spatial segregation. The municipali- ties with higher percentages of inclusion errors are mainly located in the Andean and Central regions, while the municipalities of the Pacifico, Caribe, and Orinoquia regions present lower inclusion errors (see figure 5.17). As always, the national average masks great regional varia- tion in inclusion errors, ranging from 21 percent for San Jacinto in Bolívar (Caribbean region) to 87 percent for Funza and Gachancipá in Cundinamarca (Central region). FIGURE 5.17. Comparison of Current Stratification Categories and the Adjusted Multidimensional Poverty Index, 2018 A 66.9% inclusion error B 0.6% exclusion error 30 40 50 0 60 1 70 2 80 3 Missing Missing Source: Adjusted Multidimensional Poverty Index using DANE (2018a). See Annex 4 for details on methodology. The exclusion error is significantly lower, meaning that the stratification system does succeed in identifying the multidimensional poor in Colombia and effectively supports them. An exclusion error66 was found for 61 municipalities in the country, with percentages ranging from about 1 percent for many of these municipalities to 4 percent in Fusagasugá, Cundinamarca (Central region). The map above shows a dispersed distribution of municipal- ities, locating those with the highest percentages in Antioquia, Valle del Cauca, Santander, and Cundinamarca. The low percentage in this type of error is interpreted as a positive sign because it means most of the population identified as multidimensionally poor is covered by the utilities subsidy. In addition, given the small error, the correction of the wrongly excluded population would not be associated with considerable budget requirements. This pattern of a high inclusion error/low exclusion error is also observed when compar- ing the current stratification system with the six socioeconomic clusters. As mentioned, the clusters are created with four dimensions based on household characteristics and are calculated at the intra-urban level, thus the clusters share the multidimensionality with the adapted MPI as the methodological approach for household identification and classification. Nonetheless, the clusters differ from the adapted MPI because they use a relative measure, while the adapted MPI is an absolute measure that compares the whole country. When com- paring the six socioeconomic clusters with the six categories or strata, there are three scenar- ios for households: (i) the perfect match, where the stratum is equal to the socioeconomic cluster; (ii) a mismatch above, where the strata identify a better level of socioeconomic con- ditions than the clusters; or (iii) a mismatch below, where the strata identify a worse level of socioeconomic conditions than the clusters. The mismatch below is therefore akin to the inclusion error. We find that the “mismatch below” is equal to 87 percent (figure 5.18a), meaning that 87 percent of households are classified with better socioeconomic conditions when using the six clusters than the stratification categories. Municipalities with a high percentage of ”mismatch above” (figure 5.18b) are mainly located in Cundinamarca, Antioquia, and Boya- ca, meaning that in these regions, the stratification categories reflect better conditions than those reflected by the clusters. Conversely, the “mismatch below” does not show a regional pattern but reinforces the important ramifications of households’ socioeconomic conditions in most of the municipalities. FIGURE 5.18. Mismatch between Current Stratification System and Socioeconomic Clusters, 2018 A Mismatch below (87.32%) B Mismatch above (1.49%) 40 50 0 60 10 70 20 80 30 90 40 100 50 Missing Missing Source: Socioeconomic clusters that use data from DANE (2018a). See Annex 4 for details on the methodology. 5.3. 62 BUIL D ING AN EQ UITABL E S O CIETY IN CO LO M BIA | T erritorial Inequalities in Colombia Policy Options Despite the country´s progress on its territorial development agenda, additional efforts are needed. Further actions should focus on coordinating the existing plans and tools of the different agencies in charge of territorial development to enhance, simplify, and streamline current initiatives. There is a strong need to invest in consolidating information systems to design and implement public investment strategies and projects targeted to the needs of each territorial entity. Recognizing that the challenges of territorial development cover a broad spectrum of issues, policy interventions to reduce territorial inequality should be prioritized using the WDR 2009 framework (see box 5.1 above): first, institutions that unify; second, infrastructure that connects; and third, interventions that target. Institutions that unify: strengthening subnational governments to respond to the needs of vulnerable territories Since subnational territories in Colombia play a key role in territorial development, cur- rent national policies and approaches on territorial capacity building (e.g., cadaster im- plementation) could be improved and better coordinated in the immediate/short term. This would involve: (i) enhancing information systems and data management to feed territorial planning and decision making (e.g., linking land administration and cadaster, obtaining ade- quate information on disaster threats and risks); (ii) building technical strength in weak subna- tional governments; and (iii) translating legal reforms into concrete instruments that can guide subnational governments on the implementation of territorial laws/decrees. Even strong mu- nicipalities often lack the capacity for territorial planning. Support from the national govern- ment is needed, for instance, to implement the General Policy on Territorial Planning (Política General de Ordenamiento Territorial [PGOT]) where investments are driven by local needs and challenges. Consulting local communities is key to better identifying the most pressing needs. The rollout of the Misión de Descentralización can underpin this effort and remove the main bottlenecks for subnational governments, enabling them to tackle local challenges. The Netherlands, Australia, Tunisia, and Tanzania have used different instruments to strengthen subnational capacity. In the Netherlands, the central government is responsible for a good spatial planning system across the country, ensuring that subnational governments (provinces and municipalities) have access to spatial information that facilitates urban plan- ning. To ensure enhanced data for spatial planning, the central government supports regional and local authorities through technical assistance and financing alternatives.67 To build ca- pacities at the local level, Australia, Tunisia, and Tanzania have implemented effective initia- tives. Australia allocates funding to regional and local governments according to their “internal composition,” which reflects the needs and capacities to self-finance. Tunisia and Tanzania are improving the performance of local governments through a better allocation of capital grants. Lastly, the Colombian government already has experience in translating legal reforms into con- crete instruments to guide subnational governments in improving territorial planning, for ex- ample, the Kit de Asociatividad from the Departamento Nacional de Planeación (DNP) (National Planning Department). This is something that could be further utilized across initiatives. Better equipped local governments can support reductions in territorial inequality in Colombia. A focus on urban cores can be a first step, as the analysis done in this chapter suggests that improvements in urban cores spill over to surrounding territories. Improved capacity and effectiveness among local governments is a necessary step toward improving equity. Existing indices could be used to identify those municipalities that need more support (e.g., Índice Municipal de Riesgo de Desastres Ajustado por Capacidades). In the short run, gov- ernment performance can be improved by bringing capacities when they are lacking through agreements or associations between municipalities, departments, and/or regions to address a common challenge. Colombia has different alternatives for this, such as the Pactos Territo- riales or the Esquemas Asociativos Territoriales (EATs), depending on local needs. For this to happen, two steps are needed first: (i) to adapt legal provisions on EATs to more effectively communicate existing legislation among subnational governments, and (ii) to provide tools and resources to build/improve subnational governments’ political, financial, managerial, and technical capabilities.68 In the medium term, implementation of PDETs needs to be accel- erated, underpinned by capacity building to overcome significant institutional challenges.69 For instance, Germany and Chile have different types of associations between adminis- trative units to address common challenges. In Germany, the associations of municipalities focus on the provision of their own or delegated local public services and territorial plan- ning.70 Chile has non-permanent mechanisms for horizontal policy coordination across policy fields, such as special commissions to implement specific policies or projects.71 For transport initiatives, the United States has as conditional funding for mass transit infrastructure for the creation of metropolitan transport authorities. Strengthening local fiscal performance is also key when trying to reduce inequality among territories. In Colombia, better availability and use of cadastral information can help take initial steps in this direction. To recover from the crisis brought by COVID-19 and respond to the growing demand for services, municipal governments need to be better equipped to find new sources of financing and improve the way existing resources are being used. Local finances can be improved by (i) enhancing municipal government capacities in managing a multipurpose cadaster and property tax collection, and (ii) deepening the operational linkag- es between a cadaster and fiscal management. With Law 14 (1983), the government took the initiative to strengthen local finances by establishing a national system to link the cadaster and property tax; however, updating the cadaster has been a difficult task. Through the cur- rent catastro multiproposito initiative, the objective is to register the rights, restrictions, and responsibilities of land and its owners to enable the introduction of a property tax. Up-to-date cadastral information can help alleviate inequality in the distribution of land, for example, by improving the tenure security of the most vulnerable. Such improvements are reflected in increases in value that have been estimated to be on average around 30 per- cent. In addition, clear property rights can mitigate the impact of displacement by allowing households to generate income from their land even if they do not go back to their place of or- igin. As an example, in Bogotá between 2010 and 2019, the property update process enabled an increase of 24 percent in total local public assets and an increase in cadastral appraisal of 219 percent.72 Another example is the Australian New Zealand Land Information Council’s (ANZLIC) Foundation Spatial Data Framework, which ensures the exchange and widespread accessibility of spatial data in those two countries. It identifies key geospatial data topics, such as land parcel and property, land cover, and elevation and depth.73 To reiterate, in a resource-constrained environment, finding ways to use resources more efficiently is important to freeing up resources and reaching the most vulnerable. Re- structuring targeting can increase efficiency in the use of resources. This work shows the need to integrate socioeconomic characteristics into the targeting methodology of the subsidies for servicios públicos domiciliarios (public utilities) for an efficient allocation of resources in the short run. Restructuring the system with a multidimensional identification of the population’s socioeconomic characteristics is a good starting point in reducing inequality as it enables in- vestments where they are needed most. Taking into account municipalities’ heterogeneity is key to implementing adjustments to the stratification system. The DNP is already working in this direction, with consultations and ongoing discussions to start thinking about a reform to the current subsidy system. Colombia has already made important steps in this direction. To guarantee the afford- ability of water services and the human right to water, further stressed by the COVID-19 crisis, DNP and DANE have been working to improve stratification systems and assessing whether a demand-based subsidy, such as the Chilean water subsidy model (which uses a means-tested subsidy scheme), could also contribute to reducing targeting errors. For the electricity subsidy, there is no specific international comparator, but several simulations are available in the liter- ature that point to potential fiscal savings while keeping affordability—and hence consump- tion for the poor—constant. For example, when subsidies for electricity are combined with the subsidies given to strata 1–3 (as per current methodology) but are limited to the 30 percent most vulnerable households in the country, as per data from the Sistema de Identificación de Potenciales Beneficiarios de Programas Sociales (Sisbén), simulations showed a reduction in the system’s deficit from 150,000 million pesos per month to less than 10,000, while the poor’s expenditure for electricity remained constant (at 5.6 percent of household income). Infrastructure that connects: improving accessibility to bring opportunities closer to territories Improving access to jobs, services, and opportunities for vulnerable and disadvantaged populations, particularly those without access to a private vehicle, has proven to have a significant impact on reducing intra-urban inequality in Colombia. Improving accessibili- ty should be defined as the key outcome of investing in improved transport infrastructure and services, both at the urban and inter-urban levels. For intra-urban accessibility, improving pedestrian infrastructure with safety and climate-resilient features has significant impacts on enhancing connectivity to public transport, contributing directly to improving the number of jobs, services, and opportunities reachable in a given commute time window. Carrying out a baseline accessibility analysis and understanding the changes in access provided by improving transport services and infrastructure can help inform the national policy on improving regional and urban mobility that the government of Colombia approved in April 2020 (CONPES 2020b). The Consejo Nacional de Política Económica y Social República de Colombia (CONPES)/Na- tional Council for Economic and Social Policy of Colombia, with its objective of promoting sustainable public and non-motorized modes of transport, supports three key work streams as a means of achieving its goal: (i) updating and expanding existing co-financing schemes between the national and territorial governments for transport services and infrastructure; (ii) updating and strengthening the local institutional framework for urban transport invest- ment, regulation, and operations; and (iii) supporting the implementation of policy guidelines and tools to reduce the negative externalities associated with transport (i.e., congestion, road safety–related deaths and injuries, and greenhouse gas and other local pollutant emissions). By way of comparison, in Indonesia, accessibility analysis has made it possible to assess the potential impact of mass transit investments in terms of increases in access to jobs, opportu- nities, and services. Good connectivity to reduce inequality is needed not only within cities but also between urban and rural areas. As the evidence showed, connective infrastructure within territories can enable a reduction in poverty. For inter-urban accessibility in the long run, peri-urban ar- eas and smaller municipalities can improve the connectivity of mainly residential areas to the tertiary and secondary road network as a means to improve road safety, to reduce fuel con- sumption, maintenance costs, and travel times, and to increase access to jobs, markets, and services, which tend to concentrate in larger urban areas. The Ministry of Transport has helped departments with their planning and project structuring capacities (Plan Vial Departamental). Similarly, this program could be leveraged to support municipalities in the definition of com- petencies and instruments to manage the tertiary road network for access to facilities (data collection, road inventories, guidelines to manage the network, identification of funding, etc.). For instance, in Peru, the national government created an institutional framework to bet- ter address territorial connectivity. It formed two units to manage its rural roads program: the national Provias, for the management and administration of projects in the national road network, and the decentralized Provias, for the rehabilitation and maintenance of depart- mental and rural roads, the development of institutional capacity in road infrastructure man- agement, and the articulation of transport infrastructure policy on regional development. The program has rehabilitated more than 15,000 kilometers of roads while enabling private sector participation, and more than 650 micro enterprises for road maintenance have been created, generating jobs and implementing results-based contracting. Interventions that target: well-aimed investments for more equal and inclusive cities As different dimensions of intra-urban inequality are strongly interlinked, making a city more inclusive by addressing housing inequalities requires an integrated housing policy with a territorial and cross-sectoral focus. Tackling the housing challenges requires work- ing beyond actual housing itself to include improvements in coordinated land use and trans- port policies to ensure better access to services, amenities, and jobs (OECD 2018a). As seen in the diagnostic, housing is the main driver of intra-urban inequality in Colombia, but just as inequality has many faces, actions on housing need to be comprehensive and coordinated with other dimensions. To reduce housing inequality, the National Housing Policy in Colom- bia needs to better integrate existing housing programs with the territories. This means re- inforcing national housing programs with financial mechanisms to support municipalities in their efforts to increase the amount of available housing, aligning incentives between the na- tional government and the cities. This response would vary from city to city, going from land acquisition to the provision of needed infrastructure and services, in order to better integrate households into the urban fabric to enable inclusive and productive urban areas. Financial mechanisms can be in the form of block grants for municipalities to invest locally and bet- ter complement and guarantee inclusive, well-connected, sustainable housing. The program should have clear eligibility parameters for participating urban areas, and the rules should specify the characteristics that cities must have to be eligible. The benefits of fostering cities capable of supplying affordable housing, coupled with matching services and connectivity, are well documented and include reductions in transport/commuting costs, increased access to jobs, and improved quality of life. Moreover, housing is one of the main sectors that could con- tribute to the employment recovery post-COVID-19, particularly for vulnerable populations.74 To effectively operationalize the coordination between housing programs and financial mechanisms to support inclusive and productive urban spaces, a clear understanding of local challenges is needed. Without adequate policies to ensure that new (or existing) units are well integrated into jobs and markets, housing investments can lead to stranded assets, abandoned houses, and increased vacancy rates.75 For housing units to be integrated into their territories, a local understanding of land use needs to be in place. This would avoid, for instance, housing units located in non-mitigable high-risk areas. Moreover, having an invento- ry of those families in high-risk areas and those in need of formal affordable housing can help municipalities estimate the demand and thus plan effectively. Supporting local governments for robust POTs modernos to integrate all of these aspects, with guidelines on efficient and sustainable construction, can help reduce housing inequality in Colombian cities. Because smaller and vulnerable municipalities also lack capacity as well as the technical and finan- cial mechanisms required for comprehensive urban planning and infrastructure investments, gaps between leading and lagging areas tend to be perpetuated. Working to strengthen local capacities and make financial mechanisms available across the spectrum of cities is needed to reduce territorial inequality. Moreover, accessibility analysis at the local level such as the one done here can help local governments to identify what is needed where, and how much it would cost, to link needs with the required financial mechanisms. For instance, in 1990 the United States implemented the Investment Partnerships Pro- gram (HOME) to incentivize cities to provide the needed infrastructure for resilient housing. The program provides monetary incentives (block grants) for cities to align their investment decisions in infrastructure with regional plans (or housing plans). The funding re- quirement must be accompanied by a consolidated plan that helps states and districts assess their affordable housing and community development needs, based on local data. A comple- mentary instrument that Mexico created to align urban and housing policy instruments is an index called Índice de Competitividad Municipal en Materia de Vivienda (INCOMUV). The index supports the decision making of the actors involved in the housing sector and evaluates the capacities of municipalities to host housing in a sustainable way. As the second dimension driving intra-urban inequalities in Colombia, it is essential to reduce the gap in early childhood education and high school and to address the quality of education in lagging areas. Chapter 2 provides insights into policies to promote access to learning from early on in the life cycle, targeting areas that are currently performing poorly. For inclusion and social cohesion, cities in Colombia need to tackle spatial segregation. Policy makers can make neighborhoods more inclusive, for instance, by creating places to enhance interactions and providing housing solutions that are both affordable and attractive for different groups (OECD 2018a). Additionally, affordable housing should be made available through inclusive land-use regulations and suitable social housing systems. As shown in the diagnostics, very often the most disadvantaged urban areas in one dimension are also vul- nerable in others, indicating the presence of residential segregation in Colombian cities. The proper identification of poverty traps is a central issue in urban regeneration policies (Duque, Royuela, and Noreña 2013). The use of local indicators of the spatial association of socioeco- nomic vulnerability, as the ones used in this diagnostic, can help identify those areas most in need. A prioritization strategy to improve conditions in such areas can help cities guide urban upgrading interventions. Examples of successful comprehensive interventions already exist in Colombia, such as the proyectos de mejoramiento integral de barrios (neighborhood improve- ment projects), which have shown success in cities like Bogotá, Medellín, and Cali, helping ad- dress multidimensional challenges. Recently, the government initiated efforts to reduce the housing deficit (quantitative and qualitative), particularly for the most vulnerable, through the Politica para la Infraestructura Sostenible y Resiliente, and the Resilient Housing Programs Mejoramiento de Vivienda y Vivienda en Renta para Vulnerables. However, as stressed above, these investments alone will not be enough; structural changes are needed in the sector to make urban spaces more inclusive. To narrow regional gaps, policies would have to focus on vulnerable groups, such as ethnic minorities and displaced populations. A first step in this direction is to recognize that ethnic minorities are highly heterogeneous—culturally, socioeconomically, and regionally. They are affected by multiple layers of exclusion, so no single solution will address all of their needs. Although Colombia has made important strides in counting ethno-racial minorities in census- es and household surveys, there are still key statistical records (in health and education) that lack ethno-racial disaggregation. Only with robust data can the government track progress in human capital accumulation, social inclusion, and anti-discrimination measures. Moreover, cross-sectoral coordinated investments in most lagging territories (e.g., improving basic in- frastructure, expanding education and health systems, enhancing connectivity) need to be coupled with efforts that directly protect the cultural rights of ethnic minorities (e.g., strength- ening land rights and political autonomy, supporting community-driven development). This includes strengthening their voice and participation in decision making. It is important to support the technical, financial, and organizational capacities of indigenous peoples and Af- ro-descendants through their representative associations to elevate their needs. Furthermore, correcting the imbalances of lagging regions requires a more effective application of existing legislation. For instance, Decree 1824 (2020) can expand its coverage of the regularization of land titles for the indigenous population to a broader group of ethnic minorities. Additionally, programs to integrate displaced populations into the urban labor market are needed. Evidence shows that linking displaced households to labor markets, income-gener- ating programs, and property rights in the municipality of origin (Ibáñez and Moya 2006) have an immediate effect on the well-being of these populations, allowing increased self-sufficien- cy over time. Income-generating programs need to provide effective mechanisms that enable these groups to access financial credit and nutritional and health programs, particularly in family emergencies, that currently prevent the use of resources dedicated for consumption. Policy Options for More Equitable 63 B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | T erritorial Inequalities in Colombia Opportunities across the Territory Observed Drivers of the Policy Options Relevant International Experiences Timing for Consideration about and Es- Equity Gap Equity Gap Implementa- timates of the Fiscal Impact tion of Implementation Territorial in- Better coordination Strengthen subnational governments by (i) In the Netherlands, the central government is Immediate/ The estimated fiscal impact equalities persist between existing enhancing information systems and data man- responsible for a good spatial planning system short term of implementation depends at different scales plans and tools is agement to feed territorial planning and de- across the country, ensuring that subnational on the scope of the program and dimensions. needed to enhance, cision making; (ii) building technical strength governments (provinces and municipalities) for each of the three sub-rec- simplify, and stream- in weak subnational governments to plan, have access to spatial information that facili- ommendations. More work line current initiatives design, and manage key infrastructure and tates urban planning. The central government is needed to quantify these to reduce territorial services; and (iii) translating legal reforms into supports regional and local authorities through costs. inequalities. concrete instruments that can guide subna- technical assistance and financing alternatives.76 tional governments in the implementation of To build capacities at the local level, Australia al- territorial laws/decrees. For example, support locates from the national government is needed to funding to regional and local governments ac- implement the PGOT. Existing indices could be cording to their ”internal composition,” which used to identify those municipalities that need reflects needs and capacities to self-finance. Tu- more support (e.g., Índice Municipal de Riesgo nisia and Tanzania are improving performance de Desastres Ajustado por Capacidades), focus- of local governments through a better allocation ing first on urban cores due to their potential of capital grants. Lastly, the Colombian govern- to benefit surrounding territories. ment already has experience in translating legal reforms into concrete instruments to guide subna- tional governments to improve territorial plan- ning (e.g., Kit de asociatividad from the DNP). Urban-rural in- The links between Strengthening local fiscal performance is key The United Nations Committee of Experts on Medium/ As an example, in Bogotá equalities are rural and urban areas to reducing inequality in the territories. Local Global Geospatial Information Management (UN- long term between 2010 and 2019, the high. (e.g., infrastructure, finances can be improved by (i) enhancing GGIM), through the Framework for Effective Land property update process en- transport, and socio- municipal government capacities in manag- Administration (FELA) 2019, highlights the impor- abled an increase of 24% of economic interaction) ing a multipurpose cadaster and property tance of effective land administration to improve total local public assets and are weak, as are the tax collection, and (ii) deepening the oper- the social, economic, financial, and sustainabil- an increase in cadastral ap- potential mechanisms ational linkages between the cadaster and ity aspects of a territory. The Australian New praisals of 219%. that can support the fiscal management. Since Law 14 (1983), the Zealand Land Information Council’s (ANZLIC) reduction of inequali- government has had the initiative to strength- Foundation Spatial Data Framework ensures the ty in territories. en local finances by establishing a national exchange and widespread accessibility of spatial system to link the cadaster and property tax; data in those countries. It identifies key geospa- however, updating the cadaster has been a tial data topics, such as land parcel and property, difficult task. Through current catastro multi- land cover, and elevation and depth.77 proposito initiatives, the objective is to regis- ter the rights, restrictions, and responsibilities of land and its owners to enable the introduc- tion of a property tax. Reinforce subnational government perfor- Germany and Chile have different types of asso- Immediate/ More work is needed to esti- mance by bringing capacities where they are ciations between administrative units to address short term mate the fiscal costs. lacking through agreements or associations common challenges. In Germany, the associa- between municipalities, departments, and/ tions of municipalities focus on the provision of or regions to address a common challenge. their own or delegated local public services and Colombia has different alternatives for this, territorial planning.79 Chile80 has non-permanent such as the Pactos Territoriales or the EATs,78 mechanisms for horizontal policy coordination depending on local needs. For transport, sup- across policy fields, such as special commissions port the nascent initiatives of regional trans- implementing specific policies or projects. port authorities (ARTs), under consideration by the Cali/Valle and Bogotá/Cundinamarca The United States has as conditional funding for governments. mass transit infrastructure for the creation of metropolitan transport authorities. Expand connectivity from residential parts of In Peru, the national government created two Medium/long The CONPES 4010 (Vias para peri-urban areas and smaller municipalities units to manage its rural roads program: the term la legalidad) considers invest- to the tertiary and secondary road network national Provias for the management and ad- ments on nearly 1,000 km to improve overall access to jobs, markets, ministration of projects in the national road of primary road segments in and services, which tend to concentrate in network, and the decentralized Provias for the rural areas. Other programs larger urban areas. The Ministry of Transport rehabilitation and maintenance of departmental have budgeted resources has helped departments with their planning and rural roads, the development of institution- to support secondary road and project structuring capacities (Plan Vial al capacity in road infrastructure management, rehabilitation. The coordina- Departamental). Similarly, this program could and the articulation of transport infrastructure tion of national and regional be leveraged to support municipalities in the policy with regional development. The program investment programs in reha- definition of competencies and instruments has rehabilitated more than 15,000 km while bilitation and results-based needed to manage the tertiary road network enabling private sector participation, and more contracts is not expected to for access to facilities. than 650 micro enterprises for road maintenance have a significant fiscal im- were created, generating jobs and implementing pact. results-based contracting. Intra-urban socio- Cities have grown Reinforce national housing programs with fi- In 1990 the United States implemented the Medium/long The fiscal cost will depend on economic inequal- rapidly in the past de- nancial mechanisms to support municipalities Investment Partnerships Program (HOME). It term the design of the program. ities, particularly cades, but infrastruc- in their efforts to increase the amount of avail- provides monetary incentives (block grants) It could first be limited to in the housing and ture has not grown able housing, aligning incentives between for cities to align their investment decisions in certain municipalities (as a education dimen- proportionately, leav- the national government and the cities. This infrastructure with regional plans (or housing pilot), and indicators could be sions ing many behind. response would vary from city to city, going plans). The funding requirement must be accom- used to track its effectiveness. from land acquisition to provision of needed panied by a consolidated plan that helps states infrastructure and services that can better and districts assess their affordable housing and integrate households into the urban fabric to community development needs, based on local enable inclusive and productive urban areas. data. A complementary instrument that Mexico Financial mechanisms could be in the form created to align urban and housing policy instru- of block grants for municipalities to invest lo- ments is an index called Índice de Competitividad cally and better complement and guarantee Municipal en Materia de Vivienda (INCOMUV). The inclusive, well-connected, sustainable hous- index supports the decision making of the actors ing. The program should have clear eligibility involved in the housing sector and evaluates the parameters for participating urban areas, and capacities of municipalities to host housing in a the rules should specify the characteristics sustainable way. that cities must have to be eligible. Colombia is well advanced on identifying its local chal- lenges and priorities (POTs modernos), and financial incentives from the central govern- ment can help operationalize them, while bet- ter supporting existing and new housing. Poor accessibility Carry out a baseline accessibility analysis and In Indonesia, accessibility analysis allows for an Immediate/ The fiscal cost is minimum as at the intra-urban identify the changes in access from improved assessment of the potential impact of mass tran- short term this relates primarily to ana- level transport services and infrastructure, which sit investments in terms of increases in access to lytical activities that can be can help inform the national policy for im- jobs, opportunities, and services. funded by recurring budgets proving regional and urban mobility that the at the Ministry of Transport government approved in April 2020. or DNP, or with support from multilateral development banks. Segregation in Expand intra-urban spatial data and analysis Cities in Indonesia have been improving their Medium/long Further work is needed to es- cities, with the to properly identify poverty traps and tackle local data to enable effective decision making term timate the fiscal costs. periphery less urban segregation. The use of local indicators and support those in more need. For instance, equipped of the spatial association of socioeconomic the city of Palu uses a suitability index to identify vulnerability, as the ones used in this diag- those areas that are far from optimal for popula- nostic, can help identify those areas most in tion densification. This index was used after the need. A prioritization strategy (hoja de ruta) to 2018 earthquake to attend to the affected popu- improve conditions in such areas can help cit- lation and enable settlement relocation. ies guide urban upgrading interventions. Better target groups of the population that Brazil has focused its efforts on integrating Af- Immediate/ The fiscal impact depends on have been historically segregated (e.g., Af- ro-descendants into higher levels of education. short term the type of policy, program, or ro-descendants, indigenous populations, On a voluntary basis from 2002 and by law from project to be implemented. migrants). This can be done through 4 steps: 2013, universities in Brazil need to meet certain (i) track progress in human capital accumula- quotas to include Afro-descendants. Today, Bra- tion, social inclusion, and anti-discrimination zilian public universities have an Afro-descen- measures; (ii) design policies and programs dant/non- Afro-descendant ratio closer to the that meet their specific demands and needs, population distribution. which helps to reverse the processes of struc- tural discrimination; (iii) integrate protection of cultural rights of ethnic minorities when de- signing investments; and (iv) strengthen their voice and participation in decision-making spaces by supporting the technical, financial, and organizational capacities of indigenous peoples and Afro-descendants through their representative associations (e.g., community driven-development). Inefficiency of ex- Inefficient allocation Restructure the targeting methodology of the To guarantee the affordability of water services Immediate/ Simulation exercises per- penditure meant of subsidies subsidies for Servicios publicos domiciliari- and the human right to water, further stressed short term formed that combine sub- to reduce intra-ur- os by integrating household socioeconomic by the COVID-19 crisis, DNP and DANE have been sidies for electricity with ban inequality characteristics (beyond dwelling conditions) working to improve stratification systems and subsidies given to strata 1, 2, from the Sisbén database, which can increase assessing whether a demand-based subsidy, and 3 (as per current method- efficiency in the use of resources and help to such as the Chilean water subsidy model (which ology), but limited to the 30% better identify those in need. Current method- uses a means-tested subsidy scheme), could most vulnerable households ology of stratification can be complemented help to reduce targeting errors. For the electricity in the country as per Sisbén by data from Sisbén to better target the most subsidy, there is no specific international com- data, showed a reduction in vulnerable (see last column for an example). parator, but several simulations are available fiscal deficit from 150,000 mil- in the literature that point to potential fiscal lion pesos per month to less savings, while keeping affordability—and hence than 10,000, while the poor’s consumption for the poor—constant (see last expenditure for electricity re- column). mained constant (at 5.6% of household income). Conclusion Understanding territorial inequalities in Colombia can help the government to better target investments, particularly in a resource-constrained environment. This work pro- vides evidence on territorial inequalities at different scales and dimensions, from the regional perspective to an intra-urban one. Untangling territorial inequalities by understanding where the main vulnerabilities are, in what sector, and among what groups of the population can help better target resources, particularly in a time of crisis when efficiency of expenditure be- comes even more relevant. Recognizing that the challenges of territorial development cover a broad spectrum of issues, key obstacles can be grouped along four lines: (i) weak capacities at the subnational level; (ii) poor targeting mechanisms of investments; (iii) limited connective infrastructure within urban spaces and across territories; and (iv) inefficiency of expenditures. To reduce territorial inequalities across the country, policy actions and investments need to be prioritized. In the short run, Colombia needs to strengthen subnational govern- ments and coordinate with them to prioritize programs and investments in housing and edu- cation, where the deepest disparities persist across multiple scales. Efforts along these lines will require resources that are scarce in a resource-constrained environment, but this work suggests that there may be a way to transform current efforts to reduce inequality and free up resources in the process. Moreover, carrying out a baseline accessibility analysis can help inform national policy to improve regional and urban mobility, thus closing accessibility gaps. In the long run, improving intra-urban connectivity and access to jobs and opportunities, as well as inter-urban connectivity across urban-rural areas to maximize rural access to markets and opportunities, is key. 64 Endnotes BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Territorial I nequalities in Colombia 1 Moderate Poverty rates for 2016 (World Bank Policy Note on Poverty. Draft January 2018); World Bank staff 39 The roughness index proxies the degree of territorial accessibility given pre-established natural geographic calculations on several rounds of the Gran Encuesta Integrada de Hogares (GEIH) (Large Integrated House- conditions. The index is based on the terrain ruggedness index in Riley, DeGloria, and Elliot (1999) and pro- hold Survey) carried out by the Departamento Administrativo Nacional de Estadistica (DANE) (National vides a summary of the geographic characteristics of the landscape, including the elevation and proximity Administrative Department of Statistics. to flat areas such as the coast. For calculations, the Digital Elevation Model GTOPO30, SRTM by US Geologi- 2 More than 12,000 journals were analyzed (Web of Science) across four spatial scales—regional, territorial, cal Survey (1996) was used, which made it possible to analyze an elevation raster with a resolution close to urban, and rural—and three periods: 2000–08, 2009–14, and 2015–20, and 84,125 articles were categorized. 1 kilometer and a vertical precision of about 30 meters. Intuitively, the index is built as the elevation differ- 3 The regional scale is composed of the eight regions oultined in Colombia’s National Development Plan ence between a reference point and close points surrounding it. This index is aggregated at the territorial (2018–2022): Pacífico, Eje Cafetero y Antioquia, Caribe, Central, Santanderes, Llanos y Orinoquía, Amazonía, level by taking the average of the cell values that intersect the geographic polygon of interest. and Seaflower. 40 The Gini coefficient considers 20 variables along four dimensions: demographics, housing, education, and 4 According to the United Nations (2018), 55 percent of the world’s population lives in urban areas, a share labor (see Annex 4). that is estimated to increase to 68 percent by 2050. 41 World Bank Databank. See https://databank.worldbank.org/home.aspx. 5 PND 2014–2018, “All for a New Country.” 42 For the demographic variables, see footnote 102. 6 Historical reforms aiming to reverse high levels of concentration of land property have had marginal results 43 The variables were standardized to guarantee that observations take values ranging from 0 to 1 and that all due to, among other reasons, a lack of information on land characteristics and its economic potential. To the selected variables move in the same direction. Once the variables are standardized, they were averaged address this, the implementation of the multipurpose cadaster was established as a priority in the peace to obtain vulnerability indicators for each one of the four dimensions. These dimension indicators were agreement. The agreement included consultation processes with vulnerable groups, such as victims of used to classify the blocks into six clusters using the k-medians clustering method. This method calculates forced displacement and women. In addition, the government submitted the implementation of the cadas- the shortest distance between the observation features and the group median and assigns the observation ter to free and informed consultation, following the parameters of the International Labour Organization’s to the closest cluster. Convention 169 on Indigenous and Tribal Peoples, signed by Colombia. 44 Illiteracy rate for people aged 15 years and over, and the school absence rate for people aged 5–14 years old. 7 UNDP, “Rural Colombia, Reasons for Hope,” Human Development Report (New York: United Nations Devel- 45 The dimension of demographics is the most vulnerable in the regions of Amazonia and Eje Cafetero y Antio- opment Programme, 2011), https://www.undp.org/content/dam/colombia/docs/DesarrolloHumano/undp- quia, while for the dimension of housing, it is the Seaflower region. co-ic_indh2011-parte1-2011.pdf. 46 The vulnerability of households is defined by: i) the number of households living in overcrowded housing 8 See World Bank, “Colombia: Land Policy in Transition,” Report 27942-CO (Washington, DC: World Bank, units; and (b) the number of households living in housing units with a qualitative housing deficit (DANE/ 2014), https://openknowledge.worldbank.org/handle/10986/14351?locale-attribute=es. MVCT). Municipalities with the highest quantitative and qualitative housing deficits are located in the Ca- 9 See S. Borras and others, “Land Grabbing in Latin America and the Caribbean Viewed from Broader Interna- ribbean and the Central regions; however, when looking at the connection of households to public services, tional Perspectives” (Rome: Food And Agriculture Organization, 2011). the allocation of vulnerable households is different. 10 Misión del Sistema de Ciudades, Misión para la Transformación del Campo, Misión de Crecimiento Verde, 47 For the regions of Caribe, Eje Cafetero y Antioquia, and Seaflower, the dimension with the biggest gap is and so on. education. 11 In particular, the Misión del Sistema de Ciudades has contributed to territorial development by (i) defin- 48 Measured by the Adapted Multidimensional Poverty Index, explained in Annex 4. ing a new generation of territorial development plans, called POTs modernos, which are expected to have 49 MVCT, “Analysis of Capacities and Environments of the Ministry of Housing, City and Territory” (Bogotá: Min- a greater emphasis on practical measures, be responsive to monitoring and evaluation, include disaster istry of Housing, City and Territory, 2020). risk-management (DRM), and be updated regularly; (ii) exploiting the potential of urban areas by proposing 50 CONPES, “Guidelines for the Consolidation of the Policy of Integral Improvement of MIB Barrios,” Documen- the development of a methodology to define regional visions for economic growth around 18 urban ag- to CONPES 3604 (Bogotá: Consejo Nacional de Política Económica y Social República de Colombia, 2009). glomerations identified through the analysis; and (iii) building the data for analysis by proposing the cre- 51 UNGRD, “Evidencia Empírica sobre la Relación Pobreza-Desastres en Colombia” (Bogotá: Unidad Nacional ation of an urban observatory. As a result of the Misión del Sistema de Ciudades, several documents from para la Gestion del Riesgo de Desastres, 2019). the Consejo Nacional de Política Económica y Social República de Colombia (CONPES)/National Council for 52 Observatorio Dinámicas del Territorio (2014). Economic and Social Policy of Colombia were signed and approved, one for the mission itself, another one 53 Accessibility offers a powerful lens through which to assess how a mobility system is serving an urban area. for POTs modernos, a third for the creation of the multipurpose cadaster, and a fourth for policies to sup- The efficiency of transport systems can be reframed in terms of their ability to connect people to opportuni- port sustainable green growth. ties rather than mobility. It provides a starting point to identify land-use planning options and transport in- 12 Key policy and institutional changes that resulted from the recommendations of this Misión included the vestments that can foster more integrated urban labor markets through the use of counterfactual scenarios. liquidation of INCODER (Instituto Colombiano de Desarrollo Rural) and the subsequent creation of the Na- Although the natural candidate to increase accessibility to employment opportunities has traditionally been tional Land Agency (ANT), the Rural Development Agency (ADR), and the National Agency for Renovation of investments in the transportation networks, some works have shined a light on the importance of urban the Territory (ART) to implement rural development projects. planning and coordinating land uses with transport (Quirós and Mehndiratta 2015; Avner and Lall 2016). 13 Its focus includes charting a way to strengthen competitiveness, support sustainability, and ensure resilient 54 From a labor market perspective, Kain (1968) and Gobillon Selod, and Zenou (2007) state that better job ac- growth. cessibility leads to positive labor market outcomes, such as high labor earnings and employment rates and 14 In parallel, the government, through the Departamento Nacional de Planeación (DNP) (National Planning high-quality jobs. Sanchez (2002) argues that social gaps at the intra-urban level are a spatial problem that Department), is working to articulate an overall General Policy on Territorial Planning (Política General de depends on people’s location, a dependence that can be compensated for by improving accessibility. Ordenamiento Territorial [PGOT]). In 2021, the government will officially launch the Misión de Descentral- 55 Facilities for health include clinics and hospitals; for education, preschools, schools, and high schools; and ización as stated in the National Development Plan (PND 2018–2022), and as per Law 1962 from June 2019 for sports, sports courts, covered sports courts, multiple sports courts, and recreational units. The selection (Ley de Regiones). This Mission will be led by a team of experts and stakeholders, who will define the vision of these facilities was given by the availability of data. and road map of the next generation of decentralization. 56 Five kilometers to service facilities is used across the three types of services as the standard distance in an 15 The results from the Misión para la Transformación del Campo provided key inputs to guide the peace urban context (15 kilometers in rural contexts). See Annex 4 for more details. agreements signed in 2016. Efforts are under way to develop instruments for designing the PDETs. The Mis- 57 World Bank, “Improving Accessibility to Transport for People with Limited Mobility (PLM): a Practical Guid- ión de Crecimiento Verde is still ongoing, and its findings and recommendations are informing the discus- ance Note” (Washington, DC: World Bank, 2013). sion on territorial development. 58 Fifteen capital cities were analyzed: the five capitals that concentrate a high proportion of the national pop- 16 This gap is partially explained by limitations in the liquidity of the national banking system, difficulties in ulation (Bogotá, Medellín, Cali, Barranquilla, and Cartagena); the five capitals that have the highest socio- attracting long-term investments in the infrastructure market, or rigid legislation that discourages financial economic Gini level (Neiva, Tunja, Bucaramanga, Ibagué, and Pereira); and the five capitals with the lowest innovation (World Bank 2018a). socioeconomic Gini (Quibdó, Leticia, Inírida, Mitú, and Puerto Carreño). 17 World Economic Forum, “Executive Opinion Survey: the Voice of the Business Community,” 2014–2015 Glob- 59 World Bank (2018a). The figures from the RAMV (Administrative Registry of Venezuelan Migrants) as of June al Competitiveness Report, https://reports.weforum.org/global-competitiveness-report-2014-2015/intro- 12, 2018, show some changes in the distribution of the migrant population as a percentage of the total de- duction-2/. partment: Norte de Santander (18.6 percent), La Guajira (16.92 percent), Bogotá DC (9.83 percent), Atlántico 18 “Plan Nacional de Desarrollo (National Development Plan (NDP) 2018–2022.” See https://colaboracion.dnp. (9.67 percent), Magdalena (6.94 percent), Arauca (5.9 percent), Bolívar (5.51 percent), Antioquia (4.94 per- gov.co/CDT/Prensa/PND-Pacto-por-Colombia-pacto-por-la-equidad-2018-2022.pdf. cent), and César (4.55 percent). However, this registry considers only irregular migrants. 19 Demographic variables are: (i) female head, (ii) ethnicity, (iii) Afro-descendant, (iv) indigenous, (v) disability, 60 Kennedy, Bosa, Ciudad Bolívar, Suba, and Usme. and (vi) demographic dependency; education variables are (i) non-educated head, (ii) illiteracy, (iii) school 61 See OPPCM, “Alianzas Publico Privadas en Medellin” (Bogotá: Observatorio de Politicas Publicas del Con- absence 5–14 years, and (iv) no tertiary education; housing variables are at the household level: (i) no elec- cejo de Medellin, n.d.), http://oppcm.concejodemedellin.gov.co/sites/oppcm/files/2019-08/alianzas-publi- tricity, (ii) no natural gas, (iii) no garbage collection, (iv) no internet, (v) no piped water, (vi) no sewage, and co-privadas-2016.pdf. (vii) overcrowding; and labor market variables are (i) unemployment, (ii) child labor, (iii) economic depen- 62 The stratification system classifies the population into six categories or strata (estratos in Spanish), based on dency, (iv) unpaid work, and (v) not in education, employment, or training (NEET). the characteristics of the dwellings and their immediate surroundings, to identify who will receive a subsidy 20 In 2005, Amazonia was the region with the highest share of demographic dependency (82 percent). The for water, electricity, and waste management. The first three strata are charged a lower price than the actual indicator of the region decreased to 60 percent in 2018, leaving Caribe as the region with the largest share of cost; the fourth category is charged at cost; and strata five and six are charged at a higher price than the cost households with demographic dependency. to compensate the first three strata. The system is applied country-wide, and cities do not have the autono- 21 World Bank, “Indigenous Latin America in the Twenty-First Century: The First Decade” (Washington, DC: my to modify it. World Bank, 2015), 15. 63 The C-MPI has been part of the government dashboard of official policy monitoring since 2012. The C-MPI is 22 These are constitutionally protected indigenous territories that function as self-governing, autonomous en- calculated based on the Alkire-Foster methodology, and it includes five dimensions equally weighted and tities. Resguardo residents are authorized to devise and implement social, economic, and political policies 15 indicators equally weighted within their corresponding dimension. in these areas, which are considered to be of equal legal status to districts and departmental regulations of 64 The inclusion error is defined as the mismatch that occurs when a household that does not meet the re- the Colombian state. See World Bank, “Indigenous Latin America,” 15. quirements to be subsidized receives the benefit. 23 Despite national improvements in high school achievements, the gap has increased between territories. 65 It is highly unlikely that this error will be self-correcting because the household has no incentive to rectify The national indicator decreased by 20 percentage points, the second largest reduction just behind “no the situation and lose the benefit received (see UN-Habitat 2016). natural gas.” All regions improved, but the gap between them rose from 23 to 26 percentage points between 66 The exclusion error corresponds to the mismatch that occurs when a household that meets the conditions 2005 and 2018. Amazonia continued with the highest level of vulnerability and Seaflower with the lowest. to be subsidized is classified within the high-income strata. 24 World Bank, “LAC Equity Lab” (online), https://www.worldbank.org/en/topic/poverty/lac-equity-lab1/overview. 67 Government of the Netherlands, “Spatial Planning in the Netherlands,” https://www.government.nl/topics/ 25 Living on less than US$5.5 a day, 2011 purchasing power parity. Figures are based on authors’ calcula- spatial-planning-and-infrastructure/spatial-planning-in-the-netherlands. tions using the Socio-Economic Database for Latin America and the Caribbean (SEDLAC) (CEDLAS and the 68 The DNP is already working in this direction through its launch in 2020 of the Kit de Asociatividad, which World Bank). supports municipalities and departments in the creation and implementation of EATs. 26 Authors’ calculations using SEDLAC (CEDLAS and the World Bank), 2015. 69 In addition to the intrinsic challenges of PDET territories, the implementation of PDETs faces significant 27 Ibid. difficulties, including: (i) the complexity, scale, and diversity of development needs in the more than 11,000 28 The construction and regeneration of dwellings has also contributed to total employment during the past communities; (ii) the interdependency between investments inside and outside PDET territories; (iii) the decade, showing a high multiplier effect under a COVID-19 recovery scenario. need to strengthen local governments to materialize the transformation; (iv) the need to strengthen the 29 A housing unit in qualitative deficit is defined as one that presents one or more of the following characteris- participatory processes beyond the identification and formulation of PDETs, including the structuring, im- tics: the unit has foundations and walls that are not made by bricks or concrete materials, has a poor quality plementation, and sustainability of investments; and (v) the spatial and timely coordination of initiatives in of roofs and floors materials, does not have an adequate connection to water and sanitation services, and the territory. has overcrowded conditions. An overcrowded housing unit that suffers from qualitative housing deficit is de- 70 OECD, “The Governance of Land Use - Country Fact Sheet Germany” (Paris: OECD, 2017), www.oecd.org/re- fined as one where there are more than four family members living in a one room housing unit (DANE 2018c). gional/regional-policy/land-use-Germany.pdf. 30 Out of all the socioeconomic variables considered, internet connection, economic dependency, unpaid 71 OECD, “The Governance of Land Use - Country Fact Sheet Chile” (Paris: OECD, 2017), www.oecd.org/re- work, and NEET were not measured in the 2005 Census. gional/regional-policy/land-use-Chile.pdf. 31 Most analyses that explore territorial development dynamics are based on administrative units, such as 72 The United Nations Committee of Experts on Global Geospatial Information Management (UN-GGIM), municipalities or departments. Nonetheless, when using administrative units, “urban” and “rural” defini- through the Framework for Effective Land Administration (FELA), 2019, highlights the importance of effec- tions characterize an otherwise continuous territory as a dichotomous artificial division that has often been tive land administration to improve the social, economic, financial, and sustainability aspects of a territory. guided by a residual vision of rurality (all the space that is not urban). Such a dichotomous categorization 73 ANZLIC, “The Australian and New Zealand Foundation Spatial Data Framework: Making Common Foun- tends to neglect urban-rural linkages that shape interactions between urban areas and their hinterland. dation Spatial Data Ubiquitous Across Australia and New Zealand,” http://anzlic.gov.au/sites/default/files/ Regional development analysis has increasingly recognized that “rural” is a multidimensional concept that files/One_ANZ_Foundation_Spatial_Data_Framework_Booklet.pdf. may be defined based on the functional characteristics of territories. They conclude that “rural” is not a 74 The construction of dwellings has increased its contribution to total employment during the past decade, dichotomous category, but rather a gradient or space continuum with several degrees of “rurality” (Tacoli an industry whose labor force consists mostly of unqualified workers with low levels of education and in- 1998, 2003; Veiga 2003; Schejtman and Berdegué 2004; Chomitz et al. 2005; De Ferranti et al. 2005). come. This is particularly true for self-construction. Formal construction has higher aggregate multiplier 32 These include (i) access to a larger number of services and products, since cities tend to have specialized effects for output, taxes, and salaries; however, self-construction exhibits a higher multiplier for employ- and skilled sectors; (ii) physical and virtual connectivity; (iii) knowledge spillover effects and economic ment (FEDESARROLLO 2020: https://www.fedesarrollo.org.co/sites/default/files/enlosmediosimpreso/por- benefits for firms and workers; (iv) greater public investment in the rural environment, generated by the tafolioco12julio2020.pdf ). Improving housing programs can significantly support the economic recovery. attraction of new activities to the urban hinterland; (v) greater social diversity, considering the professions For example, an impact analysis of a US$1 billion investment in jobs creation highlights the importance of and diverse types of jobs that can be found within the territory; (vi) more human capital, with different ed- taking advantage of the multiplier effects of improving the housing stock: (i) all housing solutions, includ- ucation levels; (vii) more opportunities for women within the territory, reducing the gap between male and ing home improvements at scale, can generate more jobs than any other public infrastructure project with female participation in the labor force; (viii) political competition, since the society’s desire to participate government funds; (ii) building more homes for the middle class or wealthier families, without government can increase; (ix) access to financial services such as credit; and (x) regional planning and investment prior- subsidies, generates the largest number of jobs; but (iii) expanding the homes of the poor and building in itization, with policy makers acknowledging the urban-rural interactions. See Berdegué et al. (2015); Tacoli their own plots with a combination of subsidies and microloans can generate just as many jobs. (1998); Ndabeni (2016); and Villegas Rodríguez (2014). 75 When adequate measures have not been taken to ensure that new dwellings are well integrated into jobs 33 Using 394 territories with the categorization of functional territories in Colombia in Berdegué et al. (2015), and markets, the provision of housing units in other countries (notably Mexico, Chile, and Brazil) has result- based on labor markets. See also Tolbert and Killian (1987) and Elbers, Lanjouw, and Lanjouw (2003). ed in high levels of vacancy. 34 As a robustness check, an alternative configuration of territories was created through the combination of 76 Government of the Netherlands, “Spatial Planning.” the 394 derived subregions identified by Carriazo Osorio and Reyes (2012), with an analysis carried out 77 ANZLIC, “The Australian and New Zealand Foundation Spatial Data Framework.” by Sanchez-Serra (2016) on functional urban areas in Colombia, which defined 51 functional territories 78 As noted above, to further benefit from EATs, two steps are needed first: (i) to adapt legal provisions on through the same commuting methodology. With this combination a new base with 421 functional territo- EATs to more effectively communicate existing legislation to subnational governments, and (ii) to provide ries was created (see Annex 4 for details). tools and resources to build/improve subnational governments’ political, financial, managerial, and tech- 35 When estimating three stage least squares for the monetary poverty change variable, the net effect of the nical capabilities. The DNP is already working in this direction by launching in 2020 the Kit de Asociatividad, urban core size on territorial poverty change is positive. which supports municipalities and departments in the creation and implementation of EATs. 36 Data from DANE, Índice de Precios al Consumidor, Índices – series de empalme (mayo 2021). 79 OECD, “The Governance of Land Use - Germany.” 37 Banco de la República (2014). 80 OECD, “The Governance of Land Use - Chile.” 38 Five channels are explored: road infrastructure, natural geographic conditions, municipal government per- formance, municipal fiscal performance, and the qualitative characteristics of cities (see data details in Annex 4). 65 CHAPTER 6. The Effects of Climate BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LOMBI A Change on Equity in Colombia Summary 66 BU IL DI N G AN EQ U ITAB L E SO CI ETY I N CO LO MBI A | T he Effects of C limate C hange on Equity in Colombia Climate change will not affect all Colombians equally and will compound existing underlying inequalities, further widening the inequality gap. Modeling suggests that poor, rural Colom- bians will see their wages decrease substantially more than wealthier and urban segments of the population. Informal workers will see their wages drop more than formal workers, and women more than men. The decline in rural households’ real income is much larger than that of urban households, while poorer households will be more affected than richer ones. At the sectoral level, climate change is expected to hit fisheries and land transportation hardest, fol- lowed by agriculture and livestock. Agriculture, fisheries, and livestock are the sectors with the largest share of employment at the national level (15 percent) and are already the least productive in Colombia, impacting inequality further. The government can counteract these developments and make vulnerable populations that rely on these sectors more resilient to the effects of climate change by promoting climate-smart agriculture to boost productivity, adopting more dynamic social protection systems that are better suited to respond to climate shocks, and strengthening carbon pricing policies that improve environmental equity while also raising revenues that can be applied toward climate adaptation, job creation, or other government priorities. 6.1. 67 BUI LD I NG AN EQUI TABLE SOCI ETY I N COLOM B I A | T he Effects of C limate C hange on Equity in Colombia Introduction Colombia is among the countries in Latin America most at risk from climate change. The country has a high incidence of rainfall variability and extreme events, with emerging prob- lems related to climate conditions. Natural phenomena, such as floods, droughts, heat stress, landslides, flash floods, and fires, are expected to be aggravated by human activity and the varying climate, affecting the country’s development and generating rapid changes in land use, population displacements from rural to urban areas, and environmental degradation, in addition to other socioeconomic challenges (Campos et al. 2012). The majority of the popula- tion—and the bulk of economic activity—is located in two regions that are particularly sensi- tive to climate change: the Andean region in central Colombia, where water shortages and land instability are already a reality; and the Atlantic coast in the north of the country, where the increase in sea level and floods could affect key human settlements and economic activities (UNDP 2015a). Although uncommon, the Colombian Caribbean coast could also be affected by hurricanes and tropical storms (Ortiz Royero 2012). Studies of the impacts of climate change in Colombia have had important limitations. An understanding of the impact of climate change on natural and human systems, as well as the related risks and vulnerabilities, is an important starting point in designing a response to the climate emergency. Most climate change studies for Colombia focus on the agricultural effects (DNP, BID, and CEPAL 2014; Ramírez-Villegas et al. 2012; Boshell et al. 2018). To date, there is no study that has evaluated the economic effects of the most significant climate change shocks simultaneously. Furthermore, the effects of climate variability on household income and inequality in Colombia are also under-researched. Where economy-wide effects have been evaluated, results have been limited to the aggregate level and have not assessed differential impacts by urban and rural areas or income levels. A deeper appreciation is important to un- derstanding the potential policy options. The aggregate impact of climate change on Colombia’s economy will be unfavorable. By 2050, climate change in Colombia is likely to impact the jobs and livelihoods of at least 3.5 mil- lion Colombians, the employment of 15 percent of the population, and the quality of 60 percent of the land currently suitable for agro-industries, among other ramifications (Ramirez-Villegas et al. 2012). Impacts also include soil degradation and organic matter losses in the Andean hill- sides; likely flooding on the Caribbean and Pacific coasts; niche losses for coffee, fruit, cocoa, and banana crops; changes in the prevalence of pests and diseases; and increases in the vul- nerabilities of non-technically developed smallholders (Ramirez-Villegas et al. 2012). DNP, BID, and CEPAL (2014) estimated that climate shocks to the productivity of some vulnerable eco- nomic activities would reduce the country’s GDP on average by 0.5 percent annually between 2011 and 2100. Although the study considered a limited number of sectors in the economy, it was an initial approach to the relevance of climate change in Colombia’s future economic performance. Climate change is also expected to affect water availability and rainfall patterns that will disrupt hydroelectric power generation. This is of particular importance for Colombia, where 70 percent of power generation comes from this source, and climate shocks will require substantive changes in Colombia’s energy policies (Arango-Aramburo et al. 2019). Climate change mainly affects the poor in Colombia. Considering the agricultural shocks described above, higher raw and processed food prices most heavily impact poorer house- holds for whom these products represent a larger share of total spending. Climate change hinders the fight against poverty through several other channels: reduced land productivity in agriculture, reduced labor productivity (usually unskilled agricultural workers), adverse health effects associated with disease incidence, and a shortage of drinking water, affecting both the income of farmers and food prices. Most of the dwellings affected by climate phenomena are those of the most impoverished groups due to the settlement of the poor population in higher-risk areas (e.g., susceptible to flooding and vulnerable to landslides) with inadequate housing conditions. The Colombian municipalities most affected during the 2010–2011 “La Niña” event were those with the highest unsatisfied basic needs and the lowest institutional capacity (Minambiente 2012). La Niña affected 4 million Colombians, 9 percent of the total population, and caused economic losses of approximately US$7.8 billion involving the de- struction of infrastructure and flooding of agricultural lands (Hoyos et al. 2013). Climate equity encompasses the ability to withstand climate change effects. Equity is at the heart of three core issues for climate change policy: addressing the impacts of climate change, which are felt unequally; determining who is responsible for taking action to limit its effects; and understanding how climate policy intersects with other dimensions of develop- ment, both globally and domestically (Klinsky et al. 2015). Nationally Determined Contribu- tions (NDCs), the core national commitments in the Paris agreement, provide a focal point for embedding equity into that treaty. Colombia submitted an economy-wide NDC, pledging to reduce its greenhouse gas (GHG) emissions by 51 percent by 2030 compared to a busi- ness-as-usual scenario. The NDCs’ adaptation components are based on the country’s existing National Adaptation Plan to Climate Change (PNACC, by its Spanish acronym), which provides guidance for territories and sectors on reaching the country’s goal of covering 100 percent of its national territory with implemented climate change plans.1 The studies discussed above have generally taken a sectoral or partial equilibrium ap- proach. Although very useful in understanding the in-depth mechanisms through which cli- mate change impacts the economy, the studies do not always take into account secondary and tertiary effects that can be very important in terms of both limiting the aggregate effect of shocks (such as when people switch sectors or crops to more climate-resistant activities) and assessing the repercussions of sectoral shocks that have large and potentially amplifying effects elsewhere in the economy (for example, on government finances and its ability to pro- vide essential services). Computable general equilibrium (CGE) models help to better track economy-wide im- pacts. They explicitly map the linkages between sectors and model the adaptive behavior of individuals and firms to changes in external conditions due to climate change (and government policy). As a result, they are widely employed to assess climate change impacts. Since the be- ginning of the 1990s, CGE models have been used extensively to analyze the linkages between economic activity and GHG production (Döll 2009); to measure the impacts of climate change on productivity and economic activity (Roson and Sartori 2016); and to examine the efficacy of climate mitigation policies (Dixon and Jorgenson 2012). In Colombia, Calderón et al. (2016) assess the economy-wide impacts of mitigation policies on GDP and consumption. Romero et al. (2018) use a CGE model linked to a microsimulation model to assess the distributional impacts of a carbon tax in Colombia. They find that all households are negatively affected, but that when carefully designed, reform efforts could have a positive distributional impact. The remainder of this chapter explores the economy-wide effects of different climate change effects using CGE modeling. The modeling traces the whole-economy effects of high- er temperatures and increased weather variability on productivity at the sectoral level and on the productivity of labor and hydroelectric generation, and also explores how these devel- opments impact wages, prices, and ultimately, the welfare of individuals across the income distribution. This chapter is organized as follows. The “Diagnostics” section summarizes the methodol- ogy and the primary results from CGE simulations and assesses Colombia’s ability to protect vulnerable groups and mitigate the modeled impacts, especially in the agricultural and so- cial protection spaces, through mitigating policies and financing. The “Policy Options” sec- tion centers on responding to the three transmission channels analyzed in depth through the modeling work, and builds on this analysis to propose no-regret options in economic, social, and environmental terms for rectifying the inequality impacts of climate change, notably by boosting climate-smart agriculture (CSA), strengthening social protection, and strengthening carbon pricing policies. Details on the technical specifications of the modeling work are pro- vided in Annex 5. 6.2. 68 BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | The Effects of C limate C hange on Equity in Colombia Diagnostics: Protecting Vulnerable Populations against Shocks Summary of the methodology Analyzing the economic impacts of climate change requires the integration of geophys- ical and meteorological scenarios with precise economic shocks. A complex and detailed quantitative analysis of how certain climate dimensions, such as temperature levels and heat variability, affect particular economic variables is a prerequisite to analyzing the macroeco- nomic impacts of specific climate shocks. For instance, analyzing how temperature increas- es can affect agricultural productivity requires linkages between temperature and rainfall patterns with different crop yields, which are usually affected differently by these variables. Similarly, assessing the economic effects of the increased incidence and strength of natural disasters caused by climate change requires complex quantitative analyses on how different meteorological conditions (temperature, rainfall, humidity, atmospheric pressure) change the patterns of particular natural disasters (floods, landslides, hurricanes, droughts) and how in turn, these changes in the likelihood and potency of these natural disasters will affect particu- lar economic activities, workers, and/or households. These large data and analytical require- ments constrain the possibility of conducting exhaustive economic climate assessments. As a result, many studies focus on estimating the economic effects of a single climate shock. In the case of Colombia, Ramirez-Villegas et al. (2012) and Boshell et al. (2018) focus on the agricul- tural effects of climate change, while DNP, BID, and CEPAL (2014) analyze the climate impacts on several specific economic activities. The present analysis uses a state-of-the-art, customized CGE modeling framework devel- oped by the World Bank.2 The model details 75 economic sectors, 68 goods and services that are produced by firms and consumed by households and other firms as intermediate goods, eight different types of worker, and 10 household types.3 Each sector employs the labor of different worker types, capital, and energy. Climate change is modeled through the direct ef- fects that a climatic shock is expected to have on the productivity in specific sectors (mainly agricultural activities), on labor productivity (mainly for workers with outdoor jobs), and on hydroelectric energy generation under different climate change scenarios. The analysis takes three climate change scenarios from the Intergovernmental Panel on Climate Change’s repre- sentative pathways,4 and the economic damages induced (productivity effects, etc.) from the literature (notably DNP, BID, and CEPAL 2014). Faced with these initial direct climate shocks, the model determines the indirect (general equilibrium) effects of these shocks on the rest of the economy. Negative sectoral productivity is captured through variations in crop yields, forestry, fisheries, livestock, and land transportation, and lower labor productivity is captured from higher temperatures (heat stress) and deteriorating human health (higher incidence of heat-related diseases). Finally, hydroelectric generation is captured through constraints to fu- ture increases in power generation due to adverse climate conditions. Although focusing on these three channels inevitably leaves out some of the critical impacts of climate change (e.g., increased frequency of extreme weather events), they were chosen based on their prevalence, importance for the country, and data availability. Indirect effects include changes in the pric- es of goods, in the derived demand of sectors not directly impacted, and in wages and labor across skill levels and sectors. These effects in turn affect the real incomes of each household type, as the real incomes of specific household types depend on changes in their wages, their employment, and the prices of the goods they consume. These changes in real household in- come are then used to derive the impacts on inequality. The CGE model is based on an updated social accounting matrix (SAM). The SAM was built from the 2015 SAM produced by the Colombian National Administrative Department of Statis- tics (Departamento Administrativo Nacional de Estadistica [DANE]), but updated by the World Bank using 2018 Supply and Use Tables from DANE to be consistent with national accounts data for 2018. This updated SAM was combined with household survey data, allowing for worker segmentation into eight types and households into rural and urban locations, each grouped into quintiles based on income level. The analysis considers climate change scenarios and their direct effects on sectoral pro- ductivity, labor productivity, and hydroelectric energy supply.5 The study evaluated the potential changes in sectoral productivity for five sectors that are expected to be the most directly affected by climate change: livestock, fisheries, forestry, agriculture, and land trans- portation. Although climate change will likely also affect other economic activities, this study follows the study by DNP, BID, and CEPAL (2014) and assumes that these sectors face the larg- est potential damages and are also measurable.6 The negative effects of heat stress and higher disease incidence on labor productivity are taken from Roson and Sartori (2016). Finally, the interaction between climate change shocks and hydroelectric power generation is based on estimations by Arango-Aramburo et al. (2019).7 For each scenario, the analysis identifies the effect on Colombia’s real GDP level, the economic sectors that are most affected, how wages would change (by gender, skills, and formality of employment), and which households would be most affected. The estimated repercussions on different workers and households allow an analysis of the effects of climate shocks on inequality. Since the analysis does not consider all possible climate change shocks, the results of the analysis are conservative and should be understood as the lower bound of poten- tial impacts. The study accounts for three specific climate shocks but does not account for other shocks that may have significant additional effects. For instance, because of modeling restrictions and/or data availability, the simulations do not include the economic impacts of the increased frequency and strength of hurricanes and other natural disasters (other than those associated with road damage), rising sea levels, and changes in energy demand (e.g., increased air conditioning needs) or the effects of higher temperatures on tourism activities. 69 Climate change disproportionately affecting the poor BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | T he Effects of C limate C hange on Equity in Colombia Climate change impacts vary between sectors both because some sectors are more or less sensitive to fluctuations in temperature, humidity, and extreme weather conditions and because some are more sensitive to indirect fluctuations in supply and demand. Cli- mate change has links to output and productivity losses for weather-dependent activities. For instance, in the agriculture sector, it can negatively affect crop yields, make certain goods more expensive to produce, and/or require irrigation or other additional costs of production or investments. The productivity of the livestock, fisheries, agriculture, and land transportation sectors is expected to decrease, while forestry’s productivity is expected to increase. Fisheries and land transportation are the most affected sectors, with average reductions of around 2.5 percent. Livestock faces the lowest productivity losses (below 1 percent), and the agriculture sector faces reductions that are on average around 1.5 percent, depending on the climate sce- nario. In general, scenario A2 has the highest impacts, and B2 has the lowest (see figure 6.1). Shocks to the agriculture sector, in particular, can widen inequalities. The agriculture sec- tor generates 3 million jobs, and rural agriculture represents more than 70 percent of food production, which in turn is essential for food security (DANE 2020). However, informality is especially widespread in agriculture and the broader rural sector, where more than 80 percent of the self-employed are informal, as are most salaried workers and all family workers (among whom women predominate) (FAO 2020). This implies that climate shocks will magnify the negative effects for these socioeconomic groups, even more so considering that agriculture accounts for 15 percent of total employment and that 38 percent of male informal unskilled workers work in agriculture. In other words, rural agriculture is and will be increasingly vulner- able, and at the same time will be more affected by climate variability. FIGURE 6.1. Sectoral Productivity for Each Sector by Climate Change Scenario A Climate Scenario A2 B Climate Scenario A1B C Climate Scenario B2 2050 2050 2050 2040 2040 2040 2030 2030 2030 -4,0 -3,0 -2,0 -1,0 0,0 1,0 2,0 3,0 -4,0 -3,0 -2,0 -1,0 0,0 1,0 2,0 3,0 -4,0 -3,0 -2,0 -1,0 0,0 1,0 2,0 3,0 Percent Land transport Fisheries Forestry Livestock Other Agro Source: Authors. 70 Climate change and economic output BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | T he Effects of C limate C hange on Equity in Colombia Under the climate scenarios, Colombia could annually lose between 0.5 and 0.9 percent- age points of GDP by 2050 (see figure 6.2). This finding is in line with previous studies (cf. DNP, BID, and CEPAL 2014). The GDP losses will be permanent and long term. The net present value of the cumulative GDP loss under the high-impact scenario between 2020 and 2050 is estimated at US$38.6 billion, or around 12 percent of Colombia’s 2019 GDP.8 For the low-im- pact scenario, it will represent 7 percent. This modeled reduction in overall economic output is composed of the combined reduc- tions in sector productivity, labor productivity, and hydroelectric power generation. Sec- tor productivity results in real GDP reductions ranging from -0.25 percent to -0.28 percent by 2050 (figure 6.3) compared to a baseline scenario without climate change. Climate impacts on labor productivity are expected to reduce real GDP by -0.1 percent to -0.4 percent by 2050 (fig- ure 6.4). Where climate change reduces hydroelectric power generation, real GDP is reduced by between -0.1 percent and -0.22 percent by 2050 (figure 6.5). FIGURE 6.2. Composite Real GDP Effects in 2050 FIGURE 6.3. Real GDP Changes Due to Sector (percent, with respect to the baseline) Productivity Changes 2050 2050 2040 2040 2030 2030 -0,3 -0,2 -0,1 0,0 -1,00 -0,75 -0,50 -0,25 0,00 % change wrt baseline All 3 scenarios (low-impact cases) Sectoral productivity (A1B) All 3 scenarios (high-impact cases) Sectoral productivity (A2) Sectoral productivity (B2) Source: Authors. Source: Authors. FIGURE 6.4. Real GDP Changes Due to Labor Productivity FIGURE 6.5. Real GDP Changes Due to Changes in Hydroelectric Generation 2050 2050 2040 2040 2030 2030 -0,3 -0,2 -0,1 0,0 -0,25 -0,20 -0,15 -0,10 -0,05 0,00 % change wrt baseline % change wrt baseline Labor productivity (2C) Hydroelectric generation (low) Labor productivity (1C) Hydroelectric generation (high) Source: Authors. Source: Authors. 71 Climate change to widen existing inequalities BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | T he Effects of C limate C hange on Equity in Colombia The overall macroeconomic shocks, though not large, disguise unequal income effects between households. In other words, climate change will hit poor households harder than rich ones and rural households harder than urban ones, widening existing inequali- ties. Households in the lower two income quintiles9 are expected to suffer percentage income decreases that are on average between 1.5 and 1.6 times higher than those in the top income quintile. Rural households are expected to suffer from income losses that are on average be- tween 1.8 and 1.9 times higher than urban households (see figure 6.6). Poor rural households are the most vulnerable to income losses because they rely on incomes from agricultural ac- tivities and wages for informal and unskilled jobs, which are the most severely affected by climate shocks. Informal, unskilled, and female workers will be more affected than formal, skilled, and male workers, further widening inequality. In the high-impact scenario, wages for informal workers would decline 4.2 percent compared to a decline of only 1 percent for formal workers. Unskilled workers would face a decrease in wages that is 1.7 times higher than that of skilled workers (2.5 versus 1.5 percent). Women would face wage decreases that are 1.8 times higher than those of men (2.5 versus 1.4 percent) due to the impact of climate change in the sectors in which they are overrepresented (figure 6.7). The combined effects mean that wages for unskilled male workers, the category with the highest potential losses, are expected to drop by 6.9 percent. These results reflect the fact that the modeled climate shocks are mainly concentrated in agricultural activities and primarily affect the productivity of agricultural workers, who are most likely to hold unskilled and informal jobs. This is also the reason that income decreases from climate change are expected to be stronger among rural households than among urban households. This sug- gests that the main transmission channels for household income shocks come from the labor market, where informal, female, and unskilled workers are the most vulnerable to the climate changes considered in this analysis. Climate shocks will affect poor rural households the most. The main driver of this dynamic is productivity loss in agricultural activities. Indirectly, these households are also affected by the relatively larger wage reductions for unskilled and informal work, since these households are more likely to derive substantial income from these sources. Thus, the policy options be- low are presented to achieve both environmental and social equity for Colombia, particularly in the most impacted sectors. FIGURE 6.6. Household Income Changes in 2050 (percent) FIGURE 6.7. Wage Effects in 2050 (percent) with Respect to with Respect to the Baseline, by rural and urban income the Baseline, High-Impact Scenario quintiles Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Skilled Male formal 1q 1q 2q 2q 3q 3q 4q 4q 5q 5q 0 Skilled Female formal Skilled Male informal Skilled Female informal -0,005 Unskilled Male formal Unskilled Female formal -0,01 Unskilled Male informal Unskilled Female informal -0,015 All Female All Male All Unskilled -0,02 All Skilled All Formal -0,025 All Informal Higher-impact scenario Lower-impact scenario -0,08 -0,07 -0,06 -0,05 -0,04 -0,03 -0,02 -0,01 0 0,01 Source: Authors. Source: Authors. 72 Colombia’s limited capacity to help vulnerable groups and the potential of BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | T he Effects of C limate C hange on Equity in Colombia climate-smart agriculture The modeled inequality impacts of climate change suggest that three areas of policy inter- ventions are key to helping the identified vulnerable groups adapt: 1) increasing resilience for the large proportion of the population engaged in agriculture and livestock production, in particular by boosting no-regret climate-smart approaches, 2) increasing the resilience of the poorest by improving the social protection system’s responsiveness to climate shocks, while also 3) boosting the fiscal basis for adaptation and using mitigating policies in an equity-en- hancing manner. This section diagnoses the current state of these three response options in Colombia. Climate-smart agriculture (CSA) technologies and practices in value chains that include many small producers will improve livelihoods and resilience to climate change and gen- erate important climate mitigation benefits. CSA has been proven to generate immediate benefits for producers in the form of better productivity and profitability arising from improved natural resource efficiency. In the medium run, this also means the greater resilience of pro- duction systems and the adaptation of agricultural value chains to climate change. And in the long run, CSA contributes to climate mitigation by reducing emissions and increasing car- bon sequestration in agriculture. Although certain CSA technologies and practices contribute more obviously to either adaptation or mitigation (for instance, introducing a crop mix more suited to climate change contributes more to climate adaptation, whereas replacing diesel water pumps with electric pumps has clearer mitigation implications), many others deliver multiple benefits. The establishment of silvopastoral livestock management systems, for ex- ample, could reduce climate-related losses of productive assets in milk to zero and in beef by one-third (Ramirez and Perez 2019),10 while sequestering carbon, preventing deforestation, and reducing methane emissions. Other examples include pasture recovery or renewal, which increases carbon sequestration and animal production; agroforestry in coffee, banana, fruit, and cocoa crops, which boosts carbon sequestration; green manure, conservation agriculture, and companion planting, which improve water retention and organic matter in soils, diversify livelihoods, and improve access to organic markets; and the adoption of irrigation systems and water- and energy-saving technologies, greenhouse production, or precision agriculture. Despite the potential of CSA technologies and practices, their rate of adoption among Co- lombian farmers is low (World Bank, CIAT, and CATIE 2014). Challenges to adoption include socioeconomic factors (e.g., low incomes and education, land tenure issues), coupled with insufficient funds to support producers in the transition to CSA through extension services and financial incentives. In addition, the incentives for producers to adopt CSA practices vary depending on the type of benefits that can be obtained through each practice: individual in- centives are greater for changes that produce immediate financial gains or short-run climate resilience than those that promote public goods, such as long-run mitigation. In any case, the systematic inclusion of climate change considerations in policies targeting agriculture, food security, forestry, conservation, or economic development needs strengthening to ensure co- ordinated and effective climate action across government activities. 73 Social protection programs insufficiently adaptive to climate shocks BUI LD I N G A N E Q UI TA BLE SO CI E TY I N CO LO MBI A | T he Effects of C limate C hange on Equity in Colombia Colombia has a robust social protection program portfolio, but it is not sufficiently ad- aptative to prevent or swiftly mitigate the exposure of the poor and vulnerable to cli- mate change risks while building household resilience to shocks. Although the country has well-defined institutions and policy frameworks for both disaster risk management (DRM) and social inclusion, key stakeholders, such as the Unidad de Gestión de Riesgos y Desastres (UGRD)/National Unit for Disaster Risk Management and the Departamento para la Prosperidad Social (DPS)/Department for Social Prosperity, act together only after a disaster occurs and on a case-by-case basis. This way of responding to natural disasters, including hydrometeorolog- ical ones linked to climate, contributes to Colombia’s status as one of the countries with the lowest socioeconomic resilience to climate change, producing alarming impacts on house- hold assets and well-being (see Introduction above). The magnitude of the loss of household consumption reflects the limited capacity of the social protection system to quickly prevent and mitigate the exposure of the poor and the vulnerable to shocks and the lack of a more per- manent and well-structured link between DRM and the social inclusion sectors, which often reduces the ability of public interventions to bolster household resilience. The current institutional DRM framework does not encourage linkages between key stakeholders of the social inclusion sector and is exceedingly focused on ex post risk management. Although Law 1523/2012 created a national DRM system (Sistema Nacional de Gestion del Riesgo de Desastres), to be led by the UGRD, and included a specific budget line for financing DRM activities, it did not include a specific role for the social inclusion sector. As a result, not only were key stakeholders such as DPS not included in the formulation of the current 2015–2021 national DRM policy, but most of the social protection programs also do not include specific provisions to cope with the effects of climate change (World Bank 2018b). However, despite the unclear roles of both sectors in DRM, DPS and UGRD have cooperated many times over the past decade, mostly as a reaction to climate-related disasters. During the unprecedented heavy rains that affected 90 percent of the municipalities in Colombia in 2010 and 2011, DPS and UGRD jointly created an integrated registry of households affected using the UNIDOS program11 and introduced many rapid social response programs to promote in- come generation and mitigate the impacts of the crisis on the employment of the poor. More recently, Hurricane Iota in November 2020, which affected the islands of San Andrés and Prov- idencia, led to another joint intervention, and DPS is currently implementing many initiatives on behalf of the victims. Many other examples over the past few years reflect the importance of establishing permanent institutional arrangements for fostering cooperation between DRM and social inclusion public organizations. 74 Insufficiently developed carbon pricing BU IL DI N G AN EQ U ITAB L E SO CI ETY I N CO LO MBI A | T he Effects of C limate C hange on Equity in Colombia Carbon pricing can help address inequality through direct and indirect mechanisms. It can do so directly in two ways. First, by discouraging fossil fuel combustion, carbon pricing improves air quality, which in turn reduces the negative health effects of air pollution. The latter has been estimated to cost the Colombian economy COP 15.4 trillion a year in Colom- bian cities alone,12 or about 2 percent of GDP (DNP and Fondo Acción 2015). Beyond these general economic effects, reducing air pollution boosts equality, as air pollution tends to dis- proportionally affect the poorer strata of society (see, for example, Mura et al. 2020). Second, by discouraging the overconsumption of private road transport, carbon pricing reduces road accidents which, again, disproportionately affect poorer segments of society (Nantulya and Reich 2002) and congestion, both of which represent a drag on growth and welfare (Pigato 2019).13 In addition to these direct mechanisms, carbon pricing can also boost equity indi- rectly by generating revenues that the government can use for equity-enhancing investments such as those outlined in the previous two policy options on climate-smart agriculture and social protection. Colombia has been a front-runner in putting a price on carbon emissions through its car- bon tax, but its carbon pricing regime has so far had only limited impact. Emissions reduc- tions attributable to the carbon tax have been estimated to be about 11.5 million tCO2e per annum, but only a negligible portion of that figure has derived from a reduction in demand for fossil fuels.14 This is the result of a modest carbon price (COP 15,000 per ton of CO2 at the time of the tax’s introduction in 2016, rising each year by inflation plus 1 percent) and a cov- erage that is currently limited to liquid and certain types of gaseous fossil fuels. Government revenues from the carbon tax averaged COP 436 billion a year between 2017 and 2019 (World Bank 2020a). 6.3. 75 B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | T he Effects of C limate C hange on Equity in Colombia Policy Options Addressing climate change and inequality will require adaptive policy responses to reduce the likely impacts and enable an appropriate response that cannot be mitigated ex ante. The heterogeneous effects of climate change on different households call for a policy agenda that considers equity to be a key factor in the design and development of mechanisms to adapt to and fight the coming changes. This will require policy reforms, financial resources, and mit- igation contributions. This section outlines no-regret options that can achieve the above goals by increasing agricultural productivity while mitigating its emissions, making social protection systems more effective to increase the resilience of those most affected by climate change, and raising revenues from carbon pricing while enhancing environmental equity. Invest in climate-smart agricultural value chains that combine inclusion with climate mitigation and adaptation. Given that poor rural Colombians face the highest vulnerability to climate change, Co- lombia should seek double-wins for productivity gains and mitigation efforts by sup- porting the climate resilience and mitigation potential of highly inclusive agricultural value-chains by promoting CSA. In this context, a value chain is identified as “inclusive” if it employs many small-scale producers, especially those drawn from vulnerable populations, such as women, youth, indigenous peoples, and Afro-descendants, among others. Examples of inclusive value chains include but are not limited to: coffee, where 96.5 percent of producers have landholdings between 3 and 5 hectares (National Federation of Coffee Growers, cf. Sebas- tian et al. 2020); cocoa, which is produced by around 35,000 families, often as an alternative to illicit crops (FINAGRO 2018); cattle ranching, where 80 percent of producers own fewer than 50 animals per farm (FEDEGAN 2018); and panela, the second agroindustry in social importance in the country after coffee, involving more than 350,000 families and generating 287,000 direct jobs (MADR 2019). Opportunities for mainstreaming climate change in inclusive value chains include a spec- trum of actions with an increasing scope of application. These range from the promotion of on-farm CSA adoption to the strengthening of green finance, policies with joint inclusion and environmental dividends, and cross-government incorporation of climate targets into sectoral policies. On-farm promotion of CSA can be achieved through productive projects involving the use of CSA techniques and practices by vulnerable rural producers, including leveraging technology and digitalization as tools for climate-smart decisions by producers and other val- ue-chain actors. To bridge the knowledge gap on CSA practices, investments need to be accom- panied by strengthened agricultural extension services and to ensure that adequate training and technical assistance on climate-smart technologies and practices reach smallholders and vulnerable producers at the grassroots level. A practical step in this sense is to make mitigation and adaptation key pillars of the Departmental Agricultural Extension Plans,15 systematically integrating climate aspects into the diagnostic and planning of these instruments, introduc- ing specific climate-related indicators into their results framework, and building the capacity of agricultural extension service providers. To boost private participation in the development of green value chains and spur the adoption of CSA, it will be paramount to pilot and scale up incentives for green invest- ments and adaptation. Although some of the activities to promote CSA can be financed by the public sector, particularly for the production of public goods (e.g., the strengthening of ag- ricultural extension services), the vast majority will require private investments. This will entail providing subsidies, credit lines, and other programs, including payments for mitigation, as well as boosting the uptake of risk management instruments for vulnerable producers, such as agricultural insurance, by strengthening technical capacity and infrastructure and supporting the design of products that address the specific needs of small farmers.16 In a context where formal credit in rural areas is very low (in 2017, less than 15 percent of individuals aged 15 or older borrowed from a financial institution, and only 10 percent borrowed from any source to start, operate, or expand a farm or business [World Bank FINDEX data17]), improved access to credit would also bear substantial equity benefits in terms of financial inclusion. This could be achieved by deepening the coverage of small farmers and other vulnerable categories through FINAGRO’s green credit schemes.18 A key supporting tool will be the green taxonomy for private investment that the national government and the Colombian financial sector are developing. This will generate guidelines and instruments to promote financing for sustainable develop- ment and to assess environmental and social impacts and costs in credit and investment risk analysis. Other options involve supporting existing pilots on environmental compensation approaches to boost climate adaptation and mitigation in the agriculture sector, and the pi- loting and scaling-up of zero-deforestation certification schemes in prioritized value chains, which would contribute to reducing emissions from land-use changes while improving prod- uct access to domestic and foreign markets. At the same time, existing initiatives with joint potential for inclusion and mitigation benefits will need to be continued and supported. This could be achieved by co-financing the establishment of measurement, reporting, and verification systems for GHG emissions or co-funding the investments required. A clear example is Colombia’s portfolio of Nationally Appropriate Mitigation Actions (NAMAs) for agriculture,19 which set forth productive and tech- nological actions to reduce GHG emissions in the panela, coffee, and bovine sectors. Another policy option with a clear value-chain focus is the mobilization of support for the implemen- tation of zero-deforestation agreements for the palm oil, dairy, beef, and cocoa value chains outlined in the 2018–2022 National Development Plan. All four agreements have the double objective of boosting productivity among small producers and improving market access while aiming to reduce deforestation in Colombia to zero by 2030. Their implementation, however, requires that enabling conditions are put in place. It also requires coordinated actions be- tween companies and other value chain actors, including but not limited to the establishment of robust monitoring and traceability systems (Jaramillo et al. 2020). The further strengthen- ing of national information systems, such as the Information and Communication Network of the Agricultural Sector (AGRONET)20 and the National Vegetable Traceability System,21 as well as continued support to current efforts to develop integrated traceability systems for the live- stock sector, will help create an enabling environment for climate-smart decision making and planning. Similarly, the operationalization of the National System of Agricultural Innovation (Sistema Nacional de Innovación Agropecuaria), whose implementation is scheduled for 2021, by supporting research on climate adaptation and mitigation, will play a vital role in creating a level playing field, ensuring the inclusion of small and vulnerable producers as key benefi- ciaries of extension services. Finally, at a systemic level, better coordinated and more effective climate action across different government activities can be facilitated by ensuring that climate co-benefits are methodically taken into account across government investments in the agriculture sector. This recommendation would strengthen public action in support of CSA, thus facil- itating its adoption by small and vulnerable producers. Coordination would be achieved by defining a target for the climate co-benefits generated by public sector investments from rele- vant actors in the sector (e.g., the Ministry of Agriculture and Rural Development, the Ministry of Environment and Sustainable Development, the National Land Agency, the Agency for the Renovation of the Territory), and developing guidance and instruments for monitoring and reporting such targets across public actors and interventions. Strengthen the adaptiveness of the social protection system to ensure the resilience of the poorest to climate shocks. To build a more adaptive social protection system that prevents or quickly mitigates the exposure of the poor and vulnerable to climate change risks as it builds household resil- ience to shocks, it is essential to define an institutional and policy framework that focus- es on rapidly providing support to households during and after a crisis. It should also be capable of assessing and reducing household exposure to climate change risks on a regular basis before crises occur. As not only the poor suffer the consequences of climate change in Colombia, it is important that the social protection system also focus on the vulnerable mid- dle class that is susceptible to falling back into poverty due to unexpected temporary shocks. To protect both the poor and the vulnerable, the DRM and social protection systems should be able to track the exposure of households to climate change risks, aiming to implement ade- quate programs for permanently reducing risk incidence and exposure. This is usually achieved through the definition of dynamic and integrated risk assessment instruments linked to social registries that make it possible for risk-mitigation social programs to be effectively targeted. In the Dominican Republic, for instance, the Climate Change Vulnerability Index (Índice de Vul- nerabilidad ante Choques Climáticos) is used to target ex ante risk mitigation programs, such as climate change–specific productive inclusion programs or private/community insurance schemes. Having an adequate pre-crisis (ex ante) and post-crisis (ex post) social risk manage- ment strategy is key to protecting both current and future household income and is therefore essential to promoting resilience to poverty and equity. To improve the adaptiveness of the social protection system, Colombia should build a policy framework with better linkages and clearer synergies between the DRM and so- cial inclusion sectors, specific roles and responsibilities for key stakeholders such as the UGRD and the DPS, and an appropriate balance between ex ante and ex post risk manage- ment. The government should focus on creating a more favorable institutional environment with a more balanced focus between ex ante and ex post risk management in social protection interventions, and adequate human and financial resources for supporting at-risk poor and vulnerable households. In 2021, the National Planning Department and DPS will develop a Climate Change Action Plan (Plan Integral de Gestión del Cambio Climático Sectorial) for the social inclusion sector, which is a step in the right direction for mitigating the impacts of cli- mate change on poverty and equity. In addition to a stronger policy and institutional framework, the Colombian social in- clusion sector needs to consolidate a dynamic, reliable, and integrated social registry equipped with specific climate risk assessment tools to strengthen its role in risk man- agement. The capacity of the social protection system to swiftly react to a crisis depends almost entirely on the maturity and uptake of the social registry. The establishment of an in- tegrated social registry that includes both direct program beneficiaries and non-beneficiaries will allow the government to respond more quickly to climate-related crises by identifying and targeting emergency cash transfers to existing or new recipients in response to the onset of climate-related disasters. The Sistema de Identificación de Potenciales Beneficiarios de Pro- gramas Sociales (Sisbén) IV22 has already included some innovations that will be key to better management of climate-related risks for the poor and the vulnerable. These changes include the geo-referencing of household data and the inclusion of a module assessing the exposure of households to natural disasters. However, the “sweep” structure of the Sisbén data and the lack of regular data updates usually limit the current usability of the information for identify- ing, mitigating, and responding to most of the unexpected shocks that households face. This was underlined during the COVID-19 crisis, when the absence of reliable and updated data on the non-beneficiaries of the traditional social programs meant that it was extremely hard to extend horizontally the cash transfer programs to include the vulnerable middle class.23 By adopting a dynamic and reliable social registry, with specific instruments for regularly assess- ing the vulnerability of households to climate change, Colombia will be able to set up early warning systems and other tools to manage risk in a more effective way. Finally, Colombia should ensure that current and future social protection programs, es- pecially social assistance programs, are well prepared to rapidly and flexibly respond to climate-related disasters while helping to bolster asset accumulation and resilience among the poor and vulnerable. An adaptive social protection system must be ready to com- pensate for the immediate negative impacts experienced by poor and vulnerable households while promoting a rapid, sustainable, and resilient recovery. Currently, most of the social pro- grams are not prepared to quickly react during and after a climate-related disaster. They gen- erally lack contingent financial and human resources to invest during the onset of a disaster and usually have to wait for additional funds or emergency financing to arrive before being able to act. In some cases, programs need to reallocate their financial and human resources from their traditional beneficiaries to the households affected by climate change disasters, as the programs are not equitable or efficient in most cases. Furthermore, it is also essential that the country ensure that programs to prevent and mitigate the exposure of households to climate-related risks are in place. Social assistance programs tend to be very effective after a crisis has started; however, household resilience in the long term depends entirely on the accumulated impact of both social assistance and non-social assistance programs. Ensuring that households are prepared to cope with shocks is essential to minimizing the impacts on consumption and well-being caused by climate-linked natural disasters. The diversification of the social programs’ portfolio, with a focus on both social assistance and resilience to cli- mate-related shocks, is key to tackling poverty and bolstering equity. Expand carbon pricing to combine equity and climate mitigation. For Colombia’s carbon pricing regime to contribute to the pursuit of greater equity in a meaningful way, an expansion of the regime is necessary. This could take several shapes that can in fact be combined: (i) an expansion of the coverage of the carbon tax to new emis- sions sources, (ii) an adjustment of the level of the carbon tax, and (iii) the introduction of an emissions trading system, by regulating Law 1931/2018. Any combination of these policy op- tions would need to be calibrated to consider the effects on overall growth, competitiveness, and equity, while also taking into account (i) the important co-benefits carbon pricing can generate for public health and air pollution and reduced congestion and traffic accidents, and (ii) the cost-efficiency gains in meeting Colombia’s emissions reductions goals expressed in its NDC, in which carbon pricing currently plays only a limited role. For carbon pricing to boost equity, an expansion of the regime would need to be combined with reforms that add the recycling of revenues for equity-boosting purposes to the cur- rently prescribed environmental uses. This would require a modification of the tax reform that created the carbon tax, Law 1819/2016, which currently limits revenue use to a series of predefined environmental purposes. A promising additional policy alternative, if combined with an expansion of carbon pricing, would be to use new carbon pricing revenues to reduce taxes on labor. This would have the benefit of reducing the share of the informal economy, with positive effects on employment, output, and growth (Pigato 2019). Another option would be to include cash transfers or progressive tax rebates. As an example, Canada’s federal car- bon price provides tax rebates that result in higher net incomes for most households, with the poorest households benefiting the most. However, this may be more difficult to administer in Colombia, given challenges in the income tax collection and social protection systems. Last- ly, increased carbon pricing revenues could be used to fund public infrastructure and basic services (for example, in public transport to offset the impacts of increased private motorized transportation costs). The government could also use the current carbon pricing regime to enhance equi- ty through existing channels, albeit to a more limited degree. Per Laws 1819/2016 and 1930/2018, 70 percent of the carbon tax revenues currently go to the Fondo Colombia en Paz. Beyond accelerating the disbursement rate of this fund, which has been low to date, the gov- ernment could focus the fund’s investments on labor-intensive activities by prioritizing flows to its Subaccount C. The equity impact of such a prioritization measure would, however, be relatively limited. Policy Options for Addressing 76 BUI LD I NG AN EQUI TABLE SOCI ETY I N COLOM B I A | T he Effects of C limate C hange on Equity in Colombia the Effects of Climate Change Observed Equity Gap Drivers of the Policy Options Relevant International Timing for Consideration about and Equity Gap Experiences Implementation Estimates of the Fiscal Impact of Implementation Increased vulnerabil- Insufficient in- Strengthen the National Agriculture Exten- The Forest Investment Program - Brazil’s Medium term Up-front fiscal cost, with returns ity to wage losses due vestments in sion System by mainstreaming mitigation Low-Carbon Agriculture Plan (FIP-ABC) pilot on investment to be assessed. to climate impacts, resilient CSA and adaptation criteria in the Departmental project in Brazil developed a package of training, This is an area for further work. especially among Extension Plans and building the capacity of technical assistance, and field demonstrations to poor rural house- agricultural extension service providers. promote the adoption of low-carbon agriculture holds and informal technologies proven to be effective in enhancing and female workers Strengthen the National System of Agricul- farm profitability and climate resilience. Between Medium term Up-front fiscal cost, with returns ture Innovation by financing research on cli- 2014 and 2019, FIP-ABC trained 20,025 direct on investment to be assessed. mate adaptation and mitigation. beneficiaries, which resulted in sustainable land This is an area for further work. management practices being adopted on 378,513 Pilot and scale up green incentive packages hectares (estimated reduction of 7.4 million tons Medium term Up-front fiscal cost, with returns to support mitigation and adaptation (e.g., of CO2 equivalent over the next 10 years). Esti- on investment to be assessed. subsidized credit lines; payments for environ- mated agricultural income growth among project This is an area for further work. mental services; tax credits; expanded agri- participants was triple that of a control group culture insurance, etc.) during the same period. Generate guidelines and instruments to es- The National Agricultural Innovation System Short term Up-front fiscal cost, with returns tablish a framework for climate finance in Support Project for Peru supported the National on investment to be assessed. the agriculture sector (e.g., green taxonomy Agricultural Innovation System in providing or This is an area for further work. for private investment) to promote private fi- developing improved agricultural technologies. nancing for sustainable practices. Among other measures, the project financed adaptive research, extensions, and community Support existing initiatives with joint poten- seed enterprise subprojects; funded the estab- Medium term Up-front fiscal cost, with returns tial for inclusion and mitigation benefits, such lishment of an award system to recognize ex- on investment to be assessed. as the NAMAs and zero-deforestation agree- cellence in agricultural innovation; and funded This is an area for further work. ments, by co-financing the establishment training programs for extension providers and of measurement, reporting, and verification individual extension agents on technical produc- systems of GHG emissions or co-funding the tion, climate change resiliency, and soft skill ca- investments required. pacity development. Encourage the Ministry of Agriculture and The ABC Plan of Brazil’s Ministry of Agriculture Short term Minor fiscal cost to develop a Rural Development (MADR), in coordination supported the adoption by rural producers of methodology and tracking system with DNP and the Ministry of Environment 6 low-carbon agricultural technologies proven and Sustainable Development (MADS), to de- effective in reducing GHG emissions, increasing fine sectoral targets for climate co-benefits farm profitability, and enhancing adaptation to generated by public sector investments in the climate change. The main financial instrument of agriculture sector and to develop guidance the ABC Plan was a subsidized credit line for farm- and instruments for monitoring and reporting ers to support the up-front costs of converting on such targets. traditional agricultural practices to low-carbon technologies. Between 2010 and 2020, the credit line entailed a total investment of around US$6.7 billion, and the Ministry of Agriculture estimates that the 6 technologies have been adopted in an area of 50 million hectares. The EU Green Taxonomy is an approach to help companies, project promoters, and bond issuers access green financing, as well as to help identify environmentally friendly activities. Guatemala is providing targeted support to val- ue chains with joint potential for inclusion and climate adaptation/mitigation by financing cli- mate-smart investments in primary and post-har- vest production to enhance climate resilience and improve profitability. The focus is on value chains with a high presence of small-scale farmers, Indig- enous peoples, and women and a high potential for reducing food loss and waste. Lack of rapid re- Social protec- Finalize, approve, and implement a Plan Inte- In Mexico, Prospera’s Rules of operation include Short term/ Medi- Minor fiscal costs to develop the sponse capacity to tion systems are gral de Gestión del Cambio Climático (PIGCC) operational processes for declared emergencies. um term management plan. However, the climate effects on not sufficiently that (i) includes both mitigation and adap- implementation of the measures vulnerable segments adaptive. tation measures; and (ii) embeds specific Brazil’s Federal DRM system includes the Sistema and tools devised in the plan may of the population questions into the Sisbén and other social Único de Assistência Social with well-defined pro- generate additional fiscal costs. program registries to assess the exposure of tocols for post-disaster assistance. This is an area for further work. households to climate change risks, such as climate-smart cash transfers, or climate-relat- The Kenya Hunger Safety Net program can pro- ed insurance schemes. vide exceptional transfers in anticipation of droughts and floods thanks to strong existing de- Consolidate Sisbén, social assistance admin- livery mechanisms and pre-positioned financing. Up-front fiscal costs expected istrative registries, and other key databases from the development of a social into a dynamic, reliable, and integrated single Ethiopia’s Productive Safety Net Programme registry. However, low marginal social registry, ensuring that it is equipped (PSNP) enables the poor to meet their most acute costs are involved in including with climate risk assessment tools, such as: and immediate needs and access extra resources the appropriate climate risk as- (i) climate projections and forecasts of im- in the event of climate-related shocks. sessment tools. This is an area for pacts on different geographies/ communities; further work. (ii) identification variables for households In Indonesia, cash transfers can promote both to identify their potential risks related to cli- adaptation and mitigation by raising household mate change; and (iii) early warning systems income so that people are better able to deal with to identify potential climate change risks for sudden shocks and by helping people to abandon households in a timely manner. livelihoods such as logging, which contribute to climate change. Reform social assistance programs so that Up-front fiscal costs expected they are operationally and financially pre- In Mexico and China, there are cash transfer pro- from the development of a social pared to rapidly and flexibly respond to grams that explicitly encourage beneficiaries to registry. However, low marginal climate-related events by: (i) expanding cov- engage in ecological conservation practices. costs are involved in including erage and generosity, (ii) introducing effective the appropriate climate risk as- economic inclusion mechanisms to prevent sessment tools. This is an area for households from falling back into poverty, further work. and (iii) promoting insurance schemes (pri- vate, public, or community-based) for gener- ating climate-resilient livelihoods. Unequal impact of Rudimentary Strengthen carbon pricing by expanding the There are 30 carbon tax and 31 emissions trading Carbon tax amend- Fiscal benefit. Magnitude de- secondary effects of carbon pricing carbon tax and introducing an emissions systems implemented or being designed that can ments: Short term pends on design choices. This is GHG emissions and infrastructure in trading system. offer lessons for Colombia, including numerous an area for further work. insufficient resources place jurisdictions that employ both pricing instru- Emissions trading to fund adaptation ments. system: Long term measures Conclusion The analysis conducted here suggests that even in a conservative scenario that does not include all possible climate shocks, climate change will exacerbate inequality in Colom- bia in the absence of countervailing policies. The overall negative macroeconomic impacts disguise the unequal income effects between households. Poor rural households are the most vulnerable to income losses because they rely on incomes from agricultural activities and wag- es for informal and unskilled jobs, which are the most severely affected by the climate shocks. Also, informal, unskilled, and female workers will be more affected than formal, skilled, and male workers, respectively, further widening inequality. Colombia can mitigate the unequal distributional effects of climate change. On the one hand, doing so would involve helping vulnerable segments of the population adapt to climate shocks. For many of the most vulnerable who are engaged in agricultural activities, this can be achieved through investments in CSA. This should be complemented through investments in a more responsive social protection system that can increase the resilience of vulnerable popula- tions to climate shocks. On the other hand, mitigating the impact of climate change and raising revenue to fund adaptation efforts can be achieved through an effective carbon pricing regime. Several of the policy options identified are no-regret options that will produce multiple benefits for both climate adaptation and mitigation, while boosting the productivity of primary produc- tion activities and reducing the harmful secondary effects of GHG-emitting activities. 77 Endnotes BU IL DI N G AN EQ U ITAB L E SO CI ETY I N CO LO MBI A | T he Effects of C limate C hange on Equity in Colombia 1 See UNDP, “NDC Support Programme: Colombia,” https://www.ndcs.undp.org/content/ndc-support-pro- 10 Silvopastoral systems increase biomass production and biological diversity, improve water retention and gramme/en/home/our-work/geographic/latin-america-and-caribbean/Colombia.html. reduce soil erosion, and overall provide an improved agroecosystem for animals, reducing heat stress and 2 The analysis used the Mitigation, Adaptation and New Technologies Applied General Equilibrium (MANAGE) the impacts of extreme events on biomass availability. model, the World Bank’s recursive dynamic single-country computable general equilibrium (CGE) model de- 11 The Strategy for Overcoming Extreme Poverty (known as the UNIDOS strategy) is a national, transversal, signed to focus on energy, emissions, and climate change. Annex 5 provides a detailed description of MANAGE. and intersectoral initiative that seeks to ensure that the poorest and most vulnerable households in the 3 The model was calibrated using an updated version of the social accounting matrix for Colombia, with 2018 country can overcome the conditions that keep them in poverty and extreme poverty, consolidating their as the base year. It contains 75 economic sectors, 69 commodities, eight worker types (by skill, gender, and capacities for the development and exercise of their rights. In other words, it is social intervention aiming to formality), capital, seven energy sources (coal, wind, hydroelectric at base and peak levels, gas, oil, and oth- provide support to the extreme poor households in Colombia. ers), and 10 household types (rural and urban by income quintile). See Annex 5 for details. 12 2015 Colombian pesos. 4 Scenarios A1B, A2, and B2. These three scenarios are a subset of the Special Report on Emissions Scenarios’ 13 In all of these cases, carbon pricing should be understood as one tool among many in the government’s reg- climate scenarios from the Intergovernmental Panel on Climate Change (2000). The set of A scenarios refers ulatory arsenal, as command-and-control measures are usually required to complement carbon pricing. to policies that focus on economic growth, while set B places emphasis on sustainable development. The 14 The portion of emissions reductions attributable to the carbon tax itself was only 0.4 million tCO2e for the scenarios with number 1 are those where countries act independently on climate change, while number 2017–2019 period, with the bulk of the reductions achieved (34 million tCO2e) stemming from offset proj- 2 refers to the case of global cooperation. Hence, scenario A2 implies the climate scenario with the largest ects allowed under the carbon tax law (World Bank 2020a). negative effects, as policies focus on economic growth, while B2 is a scenario with a lower climate impact, 15 The Departmental Agricultural Extension Plans (Planes Departamentales de Extensión Agropecuaria) are since polices focus on sustainable development. Scenario A1B is an intermediate scenario in which the ef- four-year planning instruments in which each Colombian department, in coordination with its municipali- fects of economic growth are counterbalanced by new technologies that improve energy efficiency. ties and districts, defines the strategic and operational elements for the provision of agricultural extension 5 As mentioned in the introduction, this is not an exhaustive analysis of all possible climate change shocks services. that Colombia can face. These three particular climate change shocks were chosen after close discussions 16 The uptake of crop insurance is currently very low, with less than 2 percent of the total cultivated area in- with the National Planning Department (Departamento Nacional de Planeación [DNP]) and an evaluation sured in 2019 (AXCO 2020). of data availability and modeling restrictions. In particular, due to time and data limitations, other climate 17 World Bank, “The Global Findex Database 2017,” https://globalfindex.worldbank.org/. shocks where not considered. For instance, changes in tourism and energy demand from higher tempera- 18 FINAGRO is the Fondo para el Financiamiento del Sector Agropecuario (Financing Fund for the Agriculture tures were not included. Also, Colombia’s SAM does not distinguish land as a production factor linking sea level rise to reductions in arable land. Finally, the lack of detailed quantitative linkages between climate Sector). shocks, increased likelihood of natural disasters, and potential economic effects prevented an analysis of 19 The NAMA portfolio currently consists of actions in agriculture, energy, housing, industry, transport, waste, the effects of natural disasters associated with climate change. and forestry. 6 The livestock sector will be damaged by lower pasture and forage growth due to higher temperatures and 20 AGRONET is a platform hosted by the Ministry of Agriculture and Rural Development for the management of lower rainfall, as well as the negative effects of increased heat on animals. Fisheries will be negatively affect- information and knowledge on the agricultural sector. ed due to changes in sea levels, temperatures, and currents, which will make it more difficult to capture the 21 Administered by the Colombian Agriculture Institute (Instituto Colombiano Agropecuario), the National main commercial fish species. The shocks to agriculture include reduced output of maize, potatoes, and rice, Vegetable Traceability System is the pool of national information systems that integrate the various produc- due to higher temperatures and the need for higher irrigation levels. The forestry sector, on the other hand, tive chains for plant species. It serves as a tool for the formulation, implementation, monitoring, and evalu- is the only activity that is expected to benefit from climate change. This is due to the positive combination ation of policies and programs on plant health and safety in the production, mobilization, and commercial- of temperature, humidity, and rainfall for specific timber trees and regions. Finally, the land transportation ization of plant species. sector will be negatively affected by increased rainfall and extreme weather conditions, which are expected 22 Colombia’s Identification System for Potential Beneficiaries of Social Program (Sistema de Identificacion de to increase the vulnerability of the road system to floods and landslides. The strain to the road system and Potenciales Beneficiarios de Programas Sociales), designed and managed by the National Planning Depart- the temporary closure of roads for repairs or rebuilds will increase land transportation costs over time. ment, was first introduced in 1995. It has since been the main targeting instrument for social programs in 7 The methodology focuses on climate damages and not on climate mitigation or adaptation policies. Colombia. 8 Net present value using a 3 percent discount rate. A lower (higher) discount rate results in larger (smaller) 23 The horizontal extension of the social assistance program during the COVID-19 crisis was made using a new present value losses. cash transfer program called Ingreso Solidario, which aimed to distribute unconditional cash transfers to 9 Quintiles are defined separately within urban and rural areas. Considering that urban households represent vulnerable households that were not part of any other social assistance program. roughly 80 percent of the total population, the overall (national) distribution effects are very close to the urban quintile effects. BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LOMBI A 78 Annexes 79 ANNEX 1. Inequalities in Subjective BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Anne xes Well-Being in Colombia across Individuals and Space1 In addition to levels of inequality captured by traditional measures such as the income Gini coefficient, perceptions of well-being, or subjective well-being (SWB), also matter. To explore this for Colombia, the Gallup World Poll from 2010–2018 in 156 countries was used. The SWB measure is based on the question: Please imagine a ladder, with steps numbered from 0 at the bottom to 10 at the top. The top of the ladder represents the best possible life for you and the bottom of the ladder represents the worst possible life for you. On which step of the ladder would you say you personally feel you stand at this time? For Colombia, the average SWB is 6.34 (see Burger, Hendriks, and Ianchovichina 2020). FIGURE A1.1. Subjective Well-Being by Level of Economic Despite being on average a relatively happy country, Colombia stands out due to its high lev- FIGURE A1.1 Subjective Well-Being by Level of Economic Development Development at at the the Country Country Level Level el of inequality in SWB (life satisfaction, happiness, and perceived or experienced welfare). Indeed, the country stands out (along with a number of other countries in the region) with a high average happiness rank for its level of development2 (see figure A1.1), a well-document- ed empirical fact known in the happiness literature as the “Latin American phenomenon” (Rojas 2016). However, despite having relatively high subjective perceptions of quality of life, Colombia and other Latin American countries have high levels of inequality in SWB, measured by the standard deviation in the SWB measure. According to recent work examining the distribution of SWB, the perceived welfare of the av- erage Colombian is mainly influenced by conditions and expectations related to economic op- portunities and education. However, a closer look along the experienced welfare distribution reveals substantial differences in the aspects that matter to those at the bottom and the top of the experienced welfare distribution. For the most satisfied members of Colombian society (the top 20 percent of the perceived welfare distribution), life satisfaction is mostly related to improvements in the standard of living, affordable housing, and civic engagement. In contrast, having an education, a job, sufficient income, economic security, and digital connectivity are much more strongly associated with the well-being of the bottom 20 percent. One distinct aspect of inequality in perceived welfare in Colombia is the high degree of spatial FIGURE A1.2 FIGURE A1.2.Subjective SubjectiveWell-Being Well-Being(Cantril (CantrilLadder, Ladder,0-10) 0-10)in inequality (figure A1.2). SWB in the cities of Bogotá, Caldas, and Quindio is well above that Colombia in Colombiaacross Regions across Regions in European nations like France and Spain. At the same time, SWB levels in the peripheral re- gions of Chocó, Nariño, and Putumayo are below those of poorer Latin American nations like Bolivia and Paraguay. Specifically, there seems to be a large urban-rural divide in Colombia, where the percentage of thriving people (i.e., those with SWB equal to or larger than 7) in pre- dominantly rural areas (39 percent) is considerably below the percentage of thriving people in urban areas (57 percent). This empirical finding contradicts the anecdotal rhetoric that SWB might be higher in the rural areas due to a number of quality-of-life enhancing factors present in the countryside (e.g., better environmental quality, shorter work commutes, lower relative food prices, etc.). However plausible, the evidence suggests that this narrative of higher per- ceived life quality in rural areas is not borne out by the data in Colombia. Part of the spatial differences in SWB can be explained by the high degrees of economic in- equality as well as the considerable differences in economic development across Colombian regions (OECD 2019). As highlighted in the most recent World Inequality Report (Alvaredo et al. 2018), the richest Colombian regions have development levels comparable to those of high-in- come countries like Chile and Uruguay, while the poorest regions are in this regard similar to lower-middle-income countries. However, the fact that the relationship between income and 6.5 - 7 SWB is weaker in Latin America than in other parts of the world implies that other factors may 6 - 6.5 play a role in explaining these spatial differences (Rojas 2019). 5.5 - 6 5 - 5.5 No data The policy areas that matter the most to the least fortunate Colombians explain the majority of spatial differences in perceived welfare between residents in urban and rural areas and be- Source: Burger, Hendriks, and Ianchovichina (2020). tween core and peripheral regions. The data suggest that the less fortunate individuals, who Notes: N=156 countries. Sampling weights and two-letter country codes are used. No con- trol variables. Log GDP per capita at PPP using constant 2017 international dollar derived are in the bottom 20 percent of the SWB distribution and tend to be poorer, put a stronger from the World Development Indicators. For both indicators, the average of the 2010–18 weight on basic needs. Consistent across a number of empirical studies is the finding that period is taken. Non-linear regression line and 95 percent confidence interval shown. No control variables. Happiness inequality is measured using the method proposed by variables related to socioeconomic status, including education, unemployment, income, and Delhey and Kohler (2011). For Colombia, the average SWB is 6.34 and the GDP per capita is income insufficiency, have the strongest association with the SWB of the unhappiest people US$13,594. The standard deviation of SWB is 2.48. in society. The importance of implementing effective redistribution policies is underlined not only by the finding that income and employment status matter for SWB, but also by people’s perceptions of inequality in these objective conditions. Although the percentage of people in Colombia who perceive the income distribution as unfair or highly unfair has remained above 80 percent since 2010, according to Latinobarómetro (2018), income inequality measured with the Gini index has steadily declined during the same period. This is in line with earlier empirical studies that showed that relative income and the subjective experience of income are better predictors of SWB in LAC countries than objective living conditions (Diego-Rosell, Tortora, and Bird 2018; Rojas 2019). Policy actions aimed at closing gaps in the areas that are most relevant to the population at the bottom of the distribution (i.e., education, employment, and economic security) have the potential to increase well-being and reduce inequality in Colombia. This notion is of particular relevance in current times since growth was weak before the COVID-19 pandemic, which has struck the country hard, affecting both the lower and middle classes (see for example, Espinel et al. 2020; Gonzalez-Diaz, Cano, and Pereira-Sanchez 2020; Lustig et al. 2020). Although Colom- bians have a generally positive outlook on life—70 percent of the population have prospects of upward mobility—there is low tolerance for inequality in the country from the above-men- tioned perceptions of unfairness. This in turn suggests that perceptions of income inequality may pose risks to social cohesion and stability, as explained long ago in the tunnel parable by Hirschman and Rothschild (1973). 80 ANNEX 2. The Disaggregated BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Anne xes Human Capital Index for Colombia What is the Human Capital Index? In 2018, the World Bank Group launched the Human Capital Project (HCP), an unprecedented global effort to support human capital development as a core element of increasing countries’ productivity and growth. The main objective of the HCP is to bolster rapid progress toward a world in which all children can achieve their full potential. For that to happen, children need to reach school well-nourished and ready to learn, attain real learning in the classroom, and enter the job market as healthy, skilled, and productive adults (Avitabile et al. 2020). Central to this effort has been the Human Capital Index (HCI), a cross-country metric mea- suring the human capital that a child born today can expect to attain by age 18, relative to a benchmark of complete education and full health. The HCI brings together measures of different dimensions of human capital: health (child survival, stunting, and adult survival rates) and the quantity and quality of schooling (expected years of school and international test scores). The index is designed to highlight how improvements in current education and health outcomes shape the productivity of the next generation of workers.3 HCI components • Component 1: Child Survival. This component reflects the fact that not all children born today will survive until age 5, when the process of human capital accumulation through formal education begins. The HCI quantifies the milestones in the trajectory of • Component 2: Expected Learning-Adjusted Years of School. This component combines information on the quantity human capital accumulation and its effects on worker and quality of education. The quantity of education is measured as the number of years of school that a child can expect productivity. The HCI consists of three components (Avi- to obtain by age 18. The quality of education draws on work at the World Bank to harmonize test scores from major inter- tabile et al. 2020): national student achievement testing programs into a common yardstick of learning. • Component 3: Health. Two proxies for the overall health environment are used to inform this component: (i) adult sur- vival rates, defined as the fraction of 15-year-olds who survive until age 60 and (ii) the rate of stunting for children under age 5. Adult survival rates can also be interpreted as a proxy for the range of non-fatal health outcomes that a child born today would experience as an adult. Children are defined as stunted if their height or age is more than two standard devi- ations below WHO’s Child Growth Standards median. Stunting is broadly accepted as a proxy for the prenatal, infant, and early childhood health environment. Human Capital Index Interpretation The HCI ranges between 0 and 1, taking the value 1 only if the average worker in the country (or department) will achieve both full health (defined as an adult survival rate of 100 percent and complete absence of stunting) and full education potential (de- fined as the completion of high-quality schooling up to the age of 14). Therefore, if a country (or department, in Colombia) scores, for example, 0.60, as in the case of Colombia, this implies that a child born today will be only 60 percent as productive as a future worker as she would if she enjoyed complete education and full health.4 Because of its structure, the HCI can be directly linked to scenarios for future income. A country with an index score of x could in the long run have a GDP per capita 1/x times higher with complete education and full health. For example, a country (or depart- ment) with an index of x=0.6 would have per capita incomes 1/0.6 = 1.66 times higher if it reaches the benchmark of complete education and full health. This increase in GDP is achieved through two channels: the direct effects of higher productivity and the indirect effects of greater investments in physical capital induced by having more productive workers. Disaggregating the Human Capital Index for Colombia Geographical disaggregated measures of the HCI can help governments target their resources most effectively to regions lagging behind in the human capital of the young. Moreover, disaggregation of the HCI components can also provide useful insights into improving the quality of human development services and prioritizing spending depending on the region’s performance in each component. Similarly, disaggregating the HCI by different socioeconomic levels can help countries identify and quantify inequali- ties arising early in life and determine policy priorities for the most disadvantaged. Improving human capital accumulation in the most disadvantaged regions and for those lagging behind can have potentially long-lasting effects on creating opportunities and closing equity gaps (Avitabile et al. 2020). To calculate5 the geographical and socioeconomic disaggregated HCI for Colombia the following sources were used: Component Definition Sources Child survival Child survival or probability of survival to age 5 is calculated by subtracting the un- The Estadísticas Vitales (2018) was the main source used to esti- der-5 mortality rate from 1. The under-5 mortality rate is the probability that a child mate the under-5 mortality rate for the subnational HCI. born in a specified year will die before reaching the age of 5 if subject to current age-specific mortality rates. It is expressed as a rate per 1,000 live births Measures of the under-5 mortality rate by socioeconomic status were used as reported in Ministerio de Salud (2015). Expected learning-adjusted years of school The expected years of school (EYS) that a child starting school at age 4 would attain The SIMAT (2018) and DANE (2018a) were the main sources by her 18th birthday is the sum of age-specific enrollment rates over all ages in this used to estimate the TNER and EYS for the subnational HCI. To age range. calculate the TNER and EYS for the socioeconomic disaggregat- ed HCI, DANE (2018b) was the main data source. The HCI uses the “total net enrollment rate (TNER)” at different levels to calculate the EYS: pre-primary enrollment rates approximate the age-specific enrollment In both cases, to adjust the EYS by the quality of education, the rates for 4 and 5-year-olds; the primary rate approximates for 6–11-year-olds; the results from Pruebas Saber 9 (2017) and Pruebas PISA (2018) lower-secondary rate approximates for 12–14-year-olds; and the upper-secondary were used to calculate the harmonized learning outcomes. approximates for 15–17-year-olds. To adjust the EYS to expected learning-adjusted years of school, the HCI uses the Harmonized Test Scores report from major international and regional student achievement testing programs (PISA, in the Colombian case). See https://datacata- log.worldbank.org/harmonized-test-scores. Health The health component is captured by two proxies: Measures on stunting rates were compiled from the Encuesta Nacional de Situación Nutricional-ENSIN (2015). Fraction of Children Under 5 Not Stunted: is calculated by subtracting stunting rates from 1. Stunting rates are defined as the under 5 whose height-for-age is more For the subnational HCI, mortality rates for 15–60-year-olds than two standard deviations below the WHO Child Growth Standards median. were calculated using the Estadísticas Vitales (2018). No data were available to calculate mortality rates by socioeconomic Adult Survival Rate: is calculated by subtracting the mortality rate for 15–60-year- level. For this reason, the HCI disaggregated by socioeconomic olds from 1. Mortality rate for 15–60-year-olds is defined as the probability that a level uses only the “Fraction of Children Under 5 Not Stunted” 15-year-old in a specified year will die before reaching the age of 60 if subject to cur- measure estimate of the HCI’s health component. rent age-specific mortality rates. It is expressed as a rate per 1,000 alive at 15. 81 ANNEX 3. Labor markets BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Anne xes estimations TABLE A3.1. Task Typology Expected Technology Impact Task Description Examples on Employment and Earnings Non-routine analytical Analyzing data/information Researchers, teachers, (+) managers Thinking creatively Interpreting information for others Non-routine interpersonal Establishing and maintaining Professionals, manag- (+) personal relationships ers, technical staff Guiding, directing and motivat- ing subordinates Coaching/developing others Non-routine manual Operating vehicles, mecha- Cleaners, hairdressers, Employment (+), earnings (-) nized devices, or equipment street vendors Spend time using hands to handle, control, or feel objects, tools, or controls Manual dexterity Spatial orientation Routine cognitive Importance of repeating the Bookkeepers, proof- (-) same tasks readers, clerks Importance of being exact or accurate Structured vs. unstructured work (reverse) Routine manual Pace determined by speed of Machine operators, ca- (-) equipment shiers, typists Controlling machines and pro- cesses Spend time making repetitive motions Source: Based on World Bank (2016b) and Acemoglu and Autor (2011)which elegantly and powerfully operationalizes the supply and demand for skills by assuming two distinct skill groups that perform two different and imperfectly substitutable tasks or produce two imperfectly substitutable goods. Technology is assumed to take a factor-augmenting form, which, by comple- menting either high or low skill workers, can generate skill biased demand shifts. In this paper, we argue that despite its notable successes, the canonical model is largely silent on a num- ber of central empirical developments of the last three decades, including: (1. TABLE A3.2. Decomposition of the Variance of the Logarithm of Hourly Earnings   2008 2019 2011 2019 2014 2019   (1) (2) (3) (4) (5) (6) Male 2.8% 3.2% 3.1% 2.9% 4.7% 2.9% Urban 4.0% 3.7% 2.6% 2.5% 2.0% 2.4% NR Analytical     13.3% 14.4% 13.4% 13.9% NR Interpersonal   2.0% 2.1% 2.2% 2.0% NR Manual     2.5% 2.8% 2.2% 2.8% R cognitive     1.0% 1.2% 1.0% 1.2% R manual     2.0% 2.2% 1.8% 2.1% WFH     8.1% 7.8% 8.5% 7.5% Indigenous         0.2% 1.2% NARP         0.1% 0.1% Venezuelan         0.0% 1.6% Age 6.3% 5.0% 5.1% 3.9% 3.7% 3.7% Education 58.1% 49.8% 38.2% 32.3% 39.6% 31.8% Sector 17.8% 24.2% 11.5% 16.1% 13.6% 15.6% Region 11.0% 14.1% 10.7% 11.8% 7.0% 11.1% TABLE A3.3. Decomposition of Changes in the Gini of Hourly Earnings Source: Author’s (Chapter 3) estimates, using the GEIH 2008–2019 and RIF regression methodology. 82 ANNEX 4. Intra-Urban and BUI LD I NG AN EQUI TABLE S OCI ETY I N COLOM BI A | Anne xes Accessibility Inequalities and Territorial Poverty A4.1. a. Methodology to build the socioeconomic clusters Intra-Urban Inequalities We (the authors of Chapter 5) standardized the variables to guarantee that observations take values ranging from 0 to 1 and that all the selected variables move in the same direction. Once the variables are standardized, we averaged them to obtain vulnerability indicators for each one of the four dimensions. We used these dimension indicators to classify the blocks into six clus- ters using the k-medians clustering method. We decided to use six clusters to mimic the current number of strata at the local level. This method calculates the shortest distance between the observation features and the group median and assigns the observation to the closest cluster. The higher the cluster label number, the lower the level of socioeconomic vulnerability. Source: DANE (2018a). Following some other studies that have applied the Gini measure to dimensions other than in- come, we calculated a pseudo-Gini coefficient for each one of the variables mentioned earlier and calculated the overall Gini as a simple average of all the Gini coefficients, using the set of blocks belonging to each region or municipality. b. Population classified in each cluster Nationally, around half of the population have better living conditions than the rest of the population, but this share changes by region. Overall, 50.8 percent of the population are clustered in categories five and six, and close to 17 percent in categories one and two. Even though this shows that socioeconomic characteristics are good for most households, this does not mean that households in categories five and six do not experience difficulties; this classi- fication is a relative measure, and blocks in cluster six indeed have lower vulnerability than those in cluster one, but the vulnerability across the six clusters can still be high. Moreover, these shares change by region (see table below), as the Seaflower region, for instance, has a more even population distribution between clusters two, three, four, and five. TABLE A4.1. Vulnerability across Regions + <------------- Vulnerability ----------------> - Cluster 1 Cluster 2 Cluster 3 Cluster 4 Cluster 5 Cluster 6 Amazonía 9.9 20.4 14.3 24.3 18.9 12.2 Caribe 7.6 13 26.4 10.1 26.5 16.3 Central 3.6 12.1 10 30.1 24.4 19.8 Eje Cafetero y Antioquia 9.3 17 10.8 24.5 21.4 17.1 Llanos y Orinoquía 8.1 12.8 25.7 11.3 25.9 16.3 Pacífico 8.5 8.8 11.7 28.4 26.6 16 Santanderes 5.2 12.7 11.1 28.6 24.4 18 Seaflower 10.6 21.8 19.4 19.9 20.9 7.5 Note: In purple, the higher shares; in blue, the lower shares. We used two sources of information to measure the degree TABLE A4.2. Amenities at the Block Level A4.2. of accessibility inequality. The 2018 Colombian Population Accessibility Category Amenity Census, with information on the location of people and ame- nities related to health care, education, and sports zones at Schools and High Schools the block level (see table A4.2 below), and the national street Education Preschool network, updated to 2014, which is provided by Procalculo Prosis SAS (www.procalculo.com), to estimate the travel dis- Clinics tance along the street network from each urban block to all Health Hospitals the selected amenities. Two methodologies are used: (i) com- puting the distance (proximity) from each block to centers of Sports court amenities within the urban perimeter,6 analyzing it at different Covered sports court territorial levels,7 and (ii) measuring the accessibility from each Sports block to the amenities,8 using the conventional gravity measure Multiple sports court of accessibility at the intra-urban level.9 Distance is used as op- Recreational unit posed to travel time to standardize accessibility regardless of the mode of transportation.10 Five kilometers to the service fa- Source: DANE (2018a). cilities is used across the three types of services as the standard distance in an urban context (15 kilometers in rural contexts).11 83 A4.3. a. Methodology BUIL D ING AN EQ UITABL E S O CIETY IN CO LO M BIA | Anne x es Interactions between Econometric models Poverty, Income, and Our starting point to model territorial poverty changes is Bourguignon (2003), in which the au- thor theoretically disentangles poverty growth in an income effect and a distributional effect. Inequality According to Bourguignon, poverty can be reduced through improvements in distribution with- out growth, or because there is an improvement in growth but not in distribution. The combined effect of growth and distribution improvements would have the greatest impact on poverty re- duction (Berdegué et al. 2015). We first explored the influence of different city sizes on poverty, income (consumption) growth, and income inequality (Gini) changes by estimating exploratory ordinary least squares (OLS) models for each of the mentioned variables. These models allowed us to identify the correlation between the 2005 and 2018 changes in the levels of poverty, income, and inequality, using ter- ritories as a unit of observation and as they were defined in Berdegué et al (2015). After the exploratory OLS analysis, we estimated a system of three equations for variable chang- es by three stage least squares (3SLS). In order to estimate consistent parameters, we used income (consumption) and inequality changes as explanatory variables that are contemporane- ously correlated with the error term of the three equations. The net effects of different city sizes on poverty changes were estimated from the 3SLS models. Finally, we explored the relationship between the size of territorial urban nuclei and some channels through which we expect the transmission of poverty changes. Simultaneous model: poverty, income, and inequality We explored 3SLS estimations that consist of estimating each of the model’s equations by the method of two SLS (2SLS) to calculate residuals and estimate a variance-covariance matrix to finally apply the feasible generalized least squares to the complete model. The system of equa- tions for the 3SLS model is given by: ∆Ind.Pob2018-2005=γ11+β12 ∆GPC+β13 ∆Gini+γ12 TN+γ1i Si+μi1 ∆Gpc=γ21+β21Gpc2005+β22 Gini2005+γ22 TN+γ2i Si+μi2 ∆Gini=γ31+β32 Gini2005+γ32 TN+γ32 Si+μi3 where ∆Ind.Pob2018-2005 is the percentage change of poverty levels for the 2005–2018 intercen- sal period, ∆GPC2018-2005 is the percentage change of per capita expenditures, ∆Gini2018-2005 is the percentage change in the Gini coefficient for the same intercensal period, Gini2005 is the 2005 Gini coefficient, Gpc2005 is per capita expenditures in 2005, TN a vector for nuclei size of urban-rural territories, S is a vector of territorial socioeconomic characteristics, βik and γik are parameters to be estimated by 3SLS, and μij are random errors for each of the equations in the system (j=1,2,3). Net effects on poverty We used parameter estimates from the 3SLS model to find the net effects of each territorial nu- cleus size on poverty changes as follows: ∆Ind.Pob=β12∆GPC (Pobi)+β13∆Gini(Pobi) b. Data A database was built at the municipal level, according to the political administrative division (Divipola) code defined by the Departamento Administrativo Nacional de Estadística (DANE), with approximately 100 variables, including social, demographic, economic, geographic, fi- nancial, fiscal, and environmental variables, among others, plus the three primary study vari- ables: poverty, income, and inequality. From the 1,121 municipalities currently identified in the country, 110 could not be included in the analysis due to the lack of information at that level of disaggregation. Those were mostly from the Orinoquia and Amazonia regions of Co- lombia, with low population densities. The municipal-level data allowed us to explore differ- ent aggregations at the functional subregion levels, using population size as weights. The data come from official databases of Colombian government agencies, such as DANE, DNP, MADS (Ministerio del Medio Ambiente y Desarrollo Sostenible), Ministry of Finance, and IGAC (Instituto Geográfico Agustín Codazzi), and from RIMISP (Centro Latinoamericano para el Desarrollo Rural), which allowed us to include variables, such as the Municipal Performance Measure (Medición de Desempeño Municipal [MDM]), the Fiscal Performance Index, road den- sity, and the Modern Cities Index (MCI) (Índice de Ciudades Modernas [ICM]), among others. We also analyzed the U.S. Geological Survey’s (1996) Data Elevation Model to produce an index of terrain heterogeneity. These variables allowed us to identify the channels and dynamics of the territories. The database contains the measure of each variable for the years 2005 and 2018 so that changes for the intercensal period could be calculated. The delta of the variables calculated at the mu- nicipal level were useful not only to understand the dynamics of variables for each territory, but also to identify changes for the urban core of each territorial unit. A discussion of the territorial units of analysis is provided in the following subsection. c. Descriptive statistics TABLE A4.3. Descriptive Statistics of Changes in Poverty, Inequality, and Expenditures for Territories of Different Core Size Average Household Monetary Poverty Change GINI (Inequality) Change Expenditure Change 2005– Multidimensional Poverty Variable Name Descriptive Statistics 2005–2018 2005–2018 2018 Measure Change 2005–2018 Average -46.9% -8.7% 14.4% -39.9% TN10 Standard deviation (0.1310) (0.0601) (0.1759) (0.1118) Average -40.0% -5.7% 4.3% -41.7% TN10_50 Standard deviation (0.1477) (0.0657) (0.2046) (0.1177) Average -38.1% -5.9% 1.1% -42.4% TN50_100 Standard deviation (0.1274) (0.0531) (0.2157) (0.1084) Average -42.8% -8.1% 5.2% -51.5% TN100_370 Standard deviation (0.1702) (0.0792) (0.1547) (0.1182) Average -44.9% -6.7% 6.9% -54.6% TNm370 Standard deviation (0.1807) (0.0646) (0.1518) (0.0720) Source: Compilation by authors. d. Ordinary-least squares (OLS) models for poverty, inequality, and income The OLS model for the delta of monetary poverty between TABLE A4.4. Ordinary Least Squares Model for Monetary Poverty 2005 and 2018 consists of 14 variables from the database that allow for an understanding of the behavior of the change in Variable Name Description Monetary Poverty Change 2005– 2018 the intercensal period, so that the greatest impact or effect is 0.0530 generated by the variable delta GINI, given that an increase PorcPobMon2005 Monetary poverty 2005 (% population) (0.0451) of one percentage point of this variable implies an increase in the delta of poverty by 1.67 percentage points, as we can see 1.6782*** DGINI GINI change 2005-2018 in the following table. (0.0578) 0.7070*** GINImun2005 Municipal GINI 2005 (0.1210) 0.0144* TN10_50 Core size between 10 thousand and 50 thousand people (0.0083) 0.0524*** TN50_100 Core size between 50 thousand and 100 thousand people (0.0159) 0.0599*** TN100_370 Core size between 100 thousand and 370 thousand people (0.0165) 0.0734*** TNm370 Core size greater than 370 thousand people (0.0223) 0.0376* Rur2005 % rural population 2005 (0.0201) -0.3031*** PorbOcu2005 % employed population 2005 (0.0679) 0.0007* Prom_PerH2005 Average number of people per household 2005 (0.0004) -0.5221** Muj_Cab2005 % women in urban area 2005 (0.2105) 0.2845*** PobDep2005 % Dependent population (NNA and older adults) 2005 (0.1068) 5.58e-06** Trans2005 Current Income Transfers 2005 (billions of pesos) (2.20e-06) Capital Expenditures - Investment 2005 (billions of pesos -7.61e-08*** GastosInv2005def prices 2018) (1.09e-08) -0.5328*** Constant Constant (0.1270) Observations   394 R-squared 0.8169 Adj R-squared   0.8101 Source: Compilation by authors. The OLS model for the GINI delta between 2005 and 2018 con- TABLE A4.5. Ordinary Least Squares Model for GINI Change sists of 11 variables that allow us to understand the behavior of the change in the intercensal period, so that the greatest Variable Name Description GINI (Inequality) Change 2005–2018 impact or effect is generated by the variable monetary pover- 0.0594*** ty change, given that an increase of one percentage point of PorcPobMon2005 Monetary poverty 2005 (% population) (0.2169) this variable implies an increase in the delta GINI by 0.4 per- centage points. 0.4052*** DPorcPobMon Monetary poverty change 2005 - 2018 (0.0139) -0.2640*** GINImun2005 Municipal GINI 2005 (0.0593) 0.0002 TN10_50 Core size between 10 thousand and 50 thousand people (0.0038) -0.0119* TN50_100 Core size between 50 thousand and 100 thousand people (0.0073) -0.0160** TN100_370 Core size between 100 thousand and 370 thousand people (0.0072) -0.0152 TNm370 Core size greater than 370 thousand people (0.0100) 0.00007** TasHomi2005 Homicide rate (x100k pop) 2005 (0.00003) 0.00007*** TasHurt2005 Theft Rate (x100k inhab) 2005 (0.00002) Royalties per capita municipality 2005 (billions of pesos 6.36e-09* RegPerCap2005def 2005) (3.86e-09) 9.61e-11*** ICA_2005def Municipal income - ICA 2005 (thousands of pesos 2005) (1.65e-11) 0.1866*** Constant Constant (0.0266) Observations 394 R-squared 0.7726 Adj R-squared 0.7661 Source: Compilation by authors. The OLS model for the average household expenditure del- TABLE A4.6. Ordinary Least Squares Model for Average Household Expenditure Change ta between 2005 and 2018 consists of 11 variables that allow us to understand the behavior of the change in the intercen- Average Household Expenditure Change Variable Name Description 2005–2018 sal period, so that the greatest impact or effect is generated by the variable percentage of female heads of households in -0.4209*** DGINI GINI change 2005-2018 2005, given that an increase of one percentage point of this (0.10009) variable implies an increase in the average household expen- 0.4660** diture change by 5.8 percentage points. GINImun2005 Municipal GINI 2005 (0.1871) -7.03e-09* GasHogProm2005 Average household expenditure 2005 (3.89e-09) -0.0490*** TN10_50 Core size between 10 thousand and 50 thousand people (0.0146) -0.0525* TN50_100 Core size between 50 thousand and 100 thousand people (0.0277) -0.0951*** TN100_370 Core size between 100 thousand and 370 thousand people (0.0280) -0.0665* TNm370 Core size greater than 370 thousand people (0.0359) 5.8317*** JHMuj2005 % of female heads of households 2005 (0.4615) Proportion of census households with housing shortages -0.0012*** DefHAb_T2005 2005 (0.0003) % Students at Basic Complete educational level in total 1.0164*** PorcEst_BasicC2005 population 2005 (0.1059) 0.0016** IndDesemFiscal2005 Municipalities Fiscal Performance Indicator 2005 (0.0007) -1.0878*** Constant Constant (0.1355) Observations   394 R-squared 0.6637 Adj R-squared   0.6540 Source: Compilation by authors. e. Three-stage ordinary least squares models for monetary and multidimensional poverty measures, and net effects TABLE A4.7. Three-Stage Least Squares Model for Monetary Poverty and Net Effects Average Household Expenditure Variable Name Description Monetary Poverty Change 2005–2018 GINI (Inequality) Change 2005–2018 Change 2005–2018 Average household expenditure -0.1391*** DGasHogPronH change (0.0333) 1.8630*** DGINI GINI change 2005-2018 (0.1718) -0.0000 GasHogPronH2005 Average household expenditure 2005 (0.0000) 0.6784*** 0.0982 0.4639*** GINImun2005 Municipal GINI 2005 (0.1223) (0.1849) (0.0894) Core size between 10 thousand and 50 -0.0034 -0.0524*** 0.0274*** TN10_50 thousand people (0.00811) (0.0149) (0.0070) Core size between 50 thousand and 0.0311** -0.0433 0.0305** TN50_100 100 thousand people (0.0147) (0.0285) (0.0134) Core size between 100 thousand and 0.0342** -0.0778** 0.0254* TN100_370 370 thousand people (0.0151) (0.0305) (0.0132) Core size greater than 370 thousand 0.0385* -0.0648* 0.0652*** TNm370 people (0.0205) (0.0392) (0.0173) -0.8685*** -1.1001*** -0.2962*** _cons Constant (0.1032) (0.1684) (0.0406) Net effects 0.0584*** Core size between 10 thousand and 50 TN10_50 thousand people (0.0142) 0.0629** Core size between 50 thousand and TN50_100 100 thousand people (0.0264) 0.0582** Core size between 100 thousand and TN100_370 370 thousand people (0.0257) 0.1305*** Core size greater than 370 thousand TNm370 people (0.0348) Source: Compilation by authors. TABLE A4.8. Three-Stage Ordinary Least Squares Model for Multidimensional Poverty Measure and Net Effects Multidimensional Poverty Average Household Expenditure GINI (Inequality) Change Variable Name Description Measure Change 2005–2018 Change 2005–2018 2005–2018 0.1947*** PorcPobMon2005 Monetary poverty 2005 (0.0693) 0.0052*** MPM_2005 Multidimensional poverty measure 2005 (0.0006) -0.0049 DGasHogPronH Average household expenditure change (0.0503) -0.7243*** DGINI GINI change 2005-2018 (0.2510) -0.0000 GasHogProm2005 Average household expenditure 2005 (0.0000) 0.2582 0.4723*** GINImun2005 Municipal GINI 2005 (0.1832) (0.0875) -0.0142 -0.0638*** 0.0267*** TN10_50 Core size between 10 thousand and 50 thousand people (0.0127) (0.0149) (0.0071) -0.0209 -0.0696** 0.0305** TN50_100 Core size between 50 thousand and 100 thousand people (0.0225) (0.0281) (0.0135) -0.0248 -0.1111*** 0.0206 TN100_370 Core size between 100 thousand and 370 thousand people (0.0254) (0.0298) (0.0132) -0.0086 -0.0998*** 0.0565*** TNm370 Core size greater than 370 thousand people (0.0346) (0.0390) (0.0173) -0.7546*** -1.0512*** -0.2955*** _cons Constant (0.1028) (0.1702) (0.0400) Net effects -0.0190** TN10_50 Core size between 10 thousand and 50 thousand people (0.0086) -0.0217* TN50_100 Core size between 50 thousand and 100 thousand people (0.0125) -0.0144 TN100_370 Core size between 100 thousand and 370 thousand people (0.0117) -0.0404** TNm370 Core size greater than 370 thousand people (0.0189) Source: Compilation by authors. 84 A4.4. a. Methodology to explore urban-rural channels B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Anne x es Urban-Rural Channels We first estimate the correlation of each channel as a function of the urban nuclei size by OLS as follows: Canal i=f(core size;β)+εi i=1…5. The following five channels were tested: 1) Municipal Performance Measure, 2) MCI, 3) Fiscal Performance Index, 4) road density, and 5) roughness index. Details on the definition of these channels are provided in the following subsection. We estimate the following system using a three-stage model that considers the influence of channels on poverty changes: ∆Ind.Pob=γ11+β12∆GPC + β13∆Gini +μ1i ∆Gpc=γ21+β2icanalesi +μ2i ∆Gini=γ31+β3icanalesi +μ3i Finally, using the parameter estimates from the 3SLS, we estimate the net effects of channels on poverty level changes as follows: ∆Ind.Pob=β12∆GPC (Canali)+β13∆Gini(Canali) b. Data The data for urban-rural channels are defined as follows. (i) For road infrastructure, we created an accessibility index defined as the density of roads (in kilometers) per territorial hectares, us- ing the Colombian road network provided by the Agustin Codazzi Geographical Institute. This variable signals as well public investment in infrastructure as an important poverty-reduction channel.12 (ii) For natural geographic conditions, we created a roughness index, based on a terrain ruggedness index by Riley, DeGloria, and Elliot (1999). This index provides a summary of the geographical characteristics of the landscape, including the elevation and the proxim- ity to flat areas such as the coast.13 (iii) To measure municipal government performance, we use the 2018 Municipal Performance Measure, produced by Colombia’s DNP, which considers two components: management14 and institutional15 performance of municipalities.16 (iv) To measure local fiscal performance, we use the municipalities Fiscal Performance Index 2019 (Índice de Desempeño Fiscal).17 (v) As a measure of qualitative characteristics of a city, we use the MCI 2019 generated by the Cities System Observatory (Observatorio del Sistema de Ciu- dades [OSC]) of the DNP, which includes six dimensions: governance, participation, and institu- tions; productivity, competitiveness, and complementarity; security; sustainability; science; and equity and social inclusion.18 a. Correlation between channels and the core sizes of the territories TABLE A4.9. Urban-Rural Channels Variable name MDM 2018 ICM IndDesem2018 Vías_Hec_Calc IRI_mean 1.4450 -1.7499** 3.4058*** -0.0007* 0.0540 TN10_50 (0.9284) (0.710) (0.8250) (0.0004) (0.0535) 3.4120* -0.7737 6.8647*** -0.0012 0.2300** TN50_100 (1.7877) (1.3673) (1.5886) (0.0009) (0.1031) 11.8177*** 5.0285*** 8.823141*** 0.0012 0.2723*** TN100_370 (1.6559) (1.2665) (1.4715) (0.0008) (0.0955) 15.6872*** 12.5429*** 11.4664*** 0.0004 0.7565*** TNm370 (2.0687) (1.5822) (1.8383) (0.0010) (0.1194) 47.5840*** 32.0736*** 65.3856*** 0.0085*** 0.1070** Constant (0.7509) (0.5744) (0.6673) (0.0003) (0.0433) Source: Compilation by authors. b. Three-Stage Least Squares Model for Monetary and Multidimensional Poverty, with channels When applying the 3SLS model for the change in monetary poverty with the selected channels (i.e., Municipal Performance Measure (2018), MCI, and the road density and roughness indices, the following effects are observed: TABLE A4.10. Three-Stage Ordinary Least Squares Model for Monetary Poverty with Channels Average Household Monetary Poverty Expenditure Change GINI (Inequality) Variable Name Description Change 2005–2018 2005–2018 Change 2005–2018 1.6320*** DGINI GINI change 2005-2018 (0.2045) Average household ex- -0.3995*** DGasHogPronH penditure 2005 (0.1069) Municipal performance 0.0038*** -0.0003 MDM2018 measure 2018 (0.0011) (0.0003) 0.0020 -0.0044*** ICM Modern Cities Index (0.0017) (0.0005) 9.9649*** -1.2458 Vías_Hec_Calc Road density per hectare (2.4118) (0.7769) -0.0008 0.0151*** IRI_mean Roughness index (0.0166) (0.0057) -0.2822*** -0.2682*** 0.0960*** _cons Constant (0.0097) (0.0544) (0.0169) Net effects Municipal performance -0.0020** MDM2018 measure 2018 (0.0008) -0.0080*** ICM Modern Cities Index (0.0011) -6.0151*** Vías_Hec_Calc Road density per hectare (1.7223) 0.025019** IRI_mean Roughness index (0.0127) Source: Estimations by the authors. When applying the least squares model in three stages for the change in the multidimensional poverty measure change with the selected channels, Municipal Performance Measure, Fiscal Performance Index, MCI, and the road density and roughness index, the following effects are observed: TABLE A4.11. Three-Stage Ordinary Least Squares Model for Multidimensional Poverty Measure with Channels Multidimensional Pov- Average Household erty Measure Change Expenditure Change GINI (Inequality) Variable Name Description 2005–2018 2005–2018 Change 2005–2018 0.9805*** DGINI GINI change 2005-2018 (0.3743) Average household ex- -0.6015*** DGasHogPronH penditure 2005 (0.1930) Municipal performance 0.0051*** -0.0001 MDM2018 measure 2018 (0.0012) (0.0004) Fiscal performance in- 0.0017 -0.0012*** IndDesemFiscal2018 dex 2018 (0.0011) (0.0003) 0.0007 -0.0041*** ICM Modern Cities Index (0.0016) (0.0005) 8.4892*** -0.8554 Vías_Hec_Calc Road density per hectare (2.2405) (0.7115) -0.0034 0.0149*** IRI_mean Roughness index (0.0142) (0.0048) -0.3149*** -0.3957*** 0.1600*** _cons Constant (0.0185) (0.0690) (0.0225) Net effects Municipal performance -0.0032*** MDM2018 measure 2018 (0.0007) Fiscal performance in- -0.0022*** IndDesemFiscal2018 dex 2018 (0.0007) -0.0045*** ICM Modern Cities Index (0.0009) -5.9452*** Vías_Hec_Calc Road density per hectare (1.1735) 0.0167* IRI_mean Roughness index (0.0086) Source: Compilation by authors. 85 A4.5. a. Results of the econometric models for the new combination of B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Anne x es FIGURE A4.1. The New Combination of Functional Subregions Map functional subregions Robustness checks with a new combination of FIGURE A4.1. The New Combination of Functional Subregions Map territories Regions 2018 Population 1 118 - 10 000 10 001 - 50 000 50 001 - 100 000 100 001 - 370 000 370 001 - 7 412 566 Source: Compilation by authors. Source: Saber 11, average by municipality and region. b. Three-stage ordinary least squares model for monetary and multidimensional poverty TABLE A4.12. Three-Stage Ordinary Least Squares Model for Monetary Poverty Change for New Combination of Functional Subregions Variable name Coef. Std. Err. P>|z| DGasHogPronH -0.1412932 0.0371348 0.000 DGINI 1.942084 0.1808204 0.000 GINImun2005 0.6034705 0.1266495 0.000 TN10_50 -0.0052798 0.0080979 0.514 TN50_100 0.0188878 0.0151132 0.211 TN100_370 0.0262429 0.01511 0.082 TNm370 0.0364084 0.0232371 0.117 DPorcPobMon Rur2005 0.04192 0.0181965 0.021 PobOcu2005 -0.2101808 0.0614952 0.001 Prom_PerH2005 0.0009132 0.0003785 0.016 PobDep2005 0.469375 0.1119319 0.000 Trans2005def 2.35E-06 1.34E-06 0.079 GastosInv2005def -7.20E-08 1.10E-08 0.000 _cons -0.8268804 0.112359 0.000 GasHogProm2005 -4.20E-09 4.59E-09 0.360 GINImun2005 0.1019916 0.1817434 0.575 TN10_50 -0.0507344 0.0149322 0.001 TN50_100 -0.0571573 0.0279794 0.041 TN100_370 -0.0750267 0.0301878 0.013 TNm370 -0.0846813 0.0421793 0.045 DGasHogPronH Rur2005 0.1614133 0.0413936 0.000 JHMuj2005 6.45688 0.4530051 0.000 PobDep2005 0.4427483 0.2083536 0.034 DefHAb_T2005 -0.0014021 0.000356 0.000 PorcEst_BasicC2005 0.7381646 0.1295099 0.000 IndDesemFiscal2005 0.0015376 0.0008299 0.064 _cons -1.140475 0.164121 0.000 GINImun2005 0.4527898 0.0883575 0.000 TN10_50 0.0276315 0.0070781 0.000 TN50_100 0.0400709 0.0130076 0.002 TN100_370 0.0464116 0.0124748 0.000 DGINI TNm370 0.0879032 0.0188069 0.000 TasHomi2005 0.000113 0.0000609 0.064 TasHurt2005 -0.0001933 0.00004 0.000 RegPerCap2005def 1.31E-08 6.95E-09 0.060 _cons -0.2905698 0.0401552 0.000 Source: Compilation by authors. TABLE A4.13. Net Effects of Monetary Poverty Change Monetary Poverty Change Variable Name Description 2005–2018 Core size between 10 thousand 0.060831*** TN10_50 and 50 thousand people (0.0147018) Core size between 50 thousand 0.0858969*** TN50_100 and 100 thousand people (0.0267824) Core size between 100 thou- 0.100736*** TN100_370 sand and 370 thousand people (0.0257309) Core size greater than 370 0.1826803*** TNm370 thousand people (0.0395323) Source: Compilation by authors. TABLE A4.14. Three-Stage Ordinary Least Squares Model for Multidimensional Poverty Measure Change for New Combination of Functional Subregions Variable name Coef. Std. Err. P>|z| PorcPobMon2005 0.2697064 0.069022 0.000 MPM_2005 0.0049838 0.0006628 0.000 DGasHogPronH -0.0361014 0.0504922 0.475 DGINI -1.117356 0.253108 0.000 TN10_50 -0.0101593 0.0129831 0.434 TN50_100 -0.0097971 0.0238512 0.681 DMPM TN100_370 -0.0143739 0.0262639 0.584 TNm370 0.0072488 0.0394629 0.854 PobOcu2005 0.1419963 0.1048679 0.176 Prom_PerH2005 -0.0019269 0.0004542 0.000 PorcEst_BasicC2005 -0.378809 0.0969908 0.000 _cons -0.7863967 0.0995194 0.000 GasHogProm2005 -4.69E-09 4.66E-09 0.314 GINImun2005 0.2759215 0.1792528 0.124 TN10_50 -0.0636668 0.0149731 0.000 TN50_100 -0.0923957 0.0272606 0.001 TN100_370 -0.1129032 0.0295573 0.000 TNm370 -0.1273403 0.042068 0.002 DGasHogPronH JHMuj2005 5.874012 0.4531022 0.000 DefHAb_T2005 -0.001264 0.0003573 0.000 PorcEst_BasicC2005 0.9477952 0.1173046 0.000 IndDesemFiscal2005 0.0017493 0.0008502 0.040 PobDep2005 0.2130081 0.1980317 0.282 _cons -1.071722 0.1659171 0.000 GINImun2005 0.4251256 0.0834724 0.000 TN10_50 0.0271567 0.0070955 0.000 TN50_100 0.04032 0.0130571 0.002 TN100_370 0.0409771 0.0124255 0.001 DGINI TNm370 0.0745104 0.0186473 0.000 TasHurt2005 -0.0001358 0.0000376 0.000 RegPerCap2005def 1.16E-08 6.66E-09 0.082 _cons -0.2750781 0.038228 0.000 Source: Compilation by authors. TABLE A4.15. Net Effects of Multidimensional Poverty Measure Change Multidimensional Poverty Variable Name Description Measure Change 2005–2018 Nucleus size between 10 thou- -0.0280453*** TN10_50 sand and 50 thousand people (0.0105548) Nucleus size between 50 thou- -0.0417162** TN50_100 sand and 100 thousand people (0.0178069) Nucleus size between 100 -0.0417101** TN100_370 thousand and 370 thousand people (0.0174052) Nucleus size greater than 370 -0.0786575*** TNm370 thousand people (0.0276948) Source: Compilation by authors. c. Urban – rural linkages for the new combination of functional subregions TABLE A4.16. Three-Stage Ordinary Least Squares Model for Monetary Poverty Change with Channels Variable name Coef. Std.Err. P>|z| DGINI 1.730325 0.1603427 0.000 DPorcPobMon DGasHogPronH -0.3385594 0.0865094 0.000 _cons -0.2815364 0.0081253 0.000 MDM2018 0.004602 0.0011971 0.000 ICM 0.0060208 0.0016326 0.000 DGasHogPronH Vías_Hec_Calc 2.70214 2.106008 0.199 IRI_mean -0.0061435 0.0238854 0.797 _cons -0.3706122 0.0639549 0.000 MDM2018 0.0000103 0.0003806 0.978 ICM -0.0049619 0.0005223 0.000 DGINI Vías_Hec_Calc 2.773465 0.6802387 0.000 IRI_mean 0.0221069 0.0080336 0.006 _cons 0.0671237 0.0197783 0.001 Source: Compilation by authors. TABLE A4.17. Three-Stage Ordinary Least Squares Multidimensional Poverty Measure Change with Channels Variable name Coef. Std. Err. P>|z| DGINI 0.4741803 0.359013 0.187 DMPM DGasHogPronH -0.8151857 0.1931785 0.000 _cons -0.3333324 0.0184513 0.000 MDM2018 0.0043293 0.0012632 0.001 IndDesemFiscal2018 0.0029086 0.0009306 0.002 ICM 0.0044883 0.0014276 0.002 DGasHogPronH Vías_Hec_Calc 2.315523 1.742028 0.184 IRI_mean -0.0237385 0.0158694 0.135 _cons -0.5014177 0.068832 0.000 MDM2018 0.0003606 0.0004401 0.413 IndDesemFiscal2018 -0.0010463 0.0004322 0.015 ICM -0.0047373 0.0004995 0.000 DGINI Vías_Hec_Calc 2.803938 0.6418676 0.000 IRI_mean 0.0243842 0.0072661 0.001 _cons 0.1136175 0.0265396 0.000 Source: Compilation by authors. TABLE A4.18. Net Effects of Monetary Poverty Change and Multidimensional Poverty Measure Change Monetary poverty change Multidimensional Poverty Variable name Description 2005–2018 Measure change 2005–2018 -0.0015402* -0.0033582*** Municipal performance mea- MDM2018 sure 2018 (0.0008597) (0.0007042) -0.0106241*** -0.0059051*** ICM Modern Cities Index (0.0011817) (0.0008483) 3.884163** -0.5580094 Vías_Hec_Calc Road density per hectare (1.535902) (1.072238) 0.040332** 0.0309138*** IRI_mean Roughness index (0.0180041) (0.01233) -0.0028672*** IndDesemFiscal2018 Fiscal performance index 2018 (0.0009158) Source: Compilation by authors. 86 ANNEX 5. MANAGE CGE Model B U I L D I N G A N EQU I TA B L E S OCI ETY I N COLOM B I A | Anne x es For Colombia The analysis in Chapter 6 employed the Mitigation, Adaptation and New Technologies Applied A5.1. General Equilibrium (MANAGE) model, the World Bank’s recursive, dynamic, and single-coun- Characteristics and try computable general equilibrium (CGE) model designed to focus on energy, emissions, and climate change. MANAGE includes a detailed energy specification that allows for capital, labor, Implementation and energy substitution in production, electricity generation by different energy sources, and a multi-output, multi-input production structure. MANAGE is a dynamic model, using mainly a neo-classical growth specification. Labor growth is determined by working-age population growth, and capital accumulation derives from savings and investment decisions; there is also a wide range of productivity assumptions, including total factor and labor productivity (see below for details). MANAGE captures the direct effects of different climate shocks, but also the general equilibri- um (indirect) effects that stem from the complex economic interrelationships among several economic agents (such as firms, government, and several household types), more than 70 economic activities, and several production factors. Structural changes from external shocks, such as damage effects associated with climate change, have many direct and complex reper- cussions on the economic activity of different sectors and on different workers and household types. Applied general equilibrium models based on the best available data have emerged as effective tools to assess these effects. MANAGE captures the ex ante impact of these shocks on a range of macro-economic indicators, including national accounts (GDP growth, consump- tion, investments, fiscal balance), industry indicators (output, employment), and labor mar- ket effects (wages and employment by worker type). MANAGE goes beyond the effect of climate shocks on macroeconomic indicators by also cap- turing the distributive effects of external shocks on both factor and household income. It can hence inform policy makers on the necessary compensatory policies that can mitigate these climate change effects and target the relevant population segments. This analysis employed the 2015 Social Accounting Matrix (SAM), originally constructed by the Colombian National Administrative Department of Statistics (DANE), and then updated it to 2018 using national accounts data and supply and use tables.19 This updated SAM was combined with household survey data,20 allowing for segmentation of the single household and labor type in the original SAM into 10 different households, and eight worker types were divided into rural and urban households, each aggregated by quintiles based on income level. Segmentation into workers employed three characteristics: sex (male or female), skill level (unskilled or skilled), and formality (formal or informal). The household survey data also provided the consumption characteristics, income sources, and direct transfers for each household type. Finally, the original SAM was also expanded to include seven electricity generation activities and two electricity commodities using the GTAP-Power database (Peters 2016). Of the total of 12 activities in the database, Colombia is using six power sources: hydroelectric (base and peak load), coal (baseload), wind (baseload), gas (peak load), and oil (peak load). The other activities in the database (nuclear, solar, and other power technologies) were therefore aggre- gated into a seventh activity: others (baseload). The two electricity commodities are distribu- tion and electricity use. The analysis used the cross-entropy method developed by Robinson, Cattaneo, and El Said (2001)21 to combine and balance the SAM using all these data sources. The ensuing SAM has 75 activities and 69 commodities, nine production factors (capital plus eight worker types), 13 agents (10 household types, firms, government, and rest of the world), and six taxes (sales, VAT, wealth, production, and subsidies to production and commodities).22 A5.2. ThissectionsummarizesthekeyrelationshipsintheCGEmodelasdescribedinChapter6.Adetailed discussion of the model, including its equations, can be found in van der Mensbrugghe (2017). The Technical Specifications model is developed from the neoclassical structural modeling approach (cf. Dervis et al. 1982).23 The underlying assumptions are mainly those encountered in the standard CGE literature (de Melo and Tarr 1992; Dixon and Jorgenson 2013). a. Production All sectors are assumed to produce under conditions of constant returns to scale and perfect competition, implying that prices are equal to the marginal costs of output. Producers max- imize their profits by minimizing their unit variable cost under the constraint of a multilevel production function (see figure A5.1). At the top level, the output is obtained by combining value added and the intermediate aggregates, following a Leontief production technology. Therefore, any policy shaping a specific sector would affect that sector directly but also indi- rectly using the output of the sector as intermediate consumption. At the second level, the intermediate aggregates are obtainedFIGURE A5.1. FIGUREMANAGE Production A5.1. MANAGE Structure Production Structure by combining all products in fixed proportions (Leontief struc- ture), and total value added is obtained by aggregating the primary factors (capital, labor, land, and natural resources) Output and energy using a nested structure. At each nest, firms make by vintage (XPv) price-sensitive decisions regarding the inputs into produc- op tion. For example, in the first nest, capital and skilled labor are Intermediate Value added combined into a bundle (KS bundle) based on the rental price demand bundle exlc. bundle (VA) of capital relative to the wage for skilled workers. An increase NRG (ND) ov in wages would cause firms to substitute away from skilled on labor toward the capital. Using a nested structure allows the Unskilled labor Land (LAND) K+SK+E Armington bundle (LAB1) bundle (KSE) model to better capture the production decisions made by demand (XAi ) firms by tailoring the degree of sensitivity of each decision to o ul ok relative prices. Capital Labor Energy skilled labor demand (Ldul ) bundle (NRG) In the next nest, this capital and skilled labor bundle is com- bundle (KSK) bined with natural resources (KF bundle). Unlike standard CGE o ks oe models, MANAGE allows for firms to determine the energy in- Skilled labor Armington tensity of production; the KF bundle is combined with ener- Capital (K) bundle (LAB2) demand (XAe ) gy (KEF bundle). In the agriculture sectors, the KEF bundle is o sl combined with land before it is combined with unskilled la- bor. For non-agriculture sectors, the KEF bundle is combined Labor demand (Ldsl ) with unskilled labor. b. Factor markets Factor markets are assumed to be in perfect competition. The labor market is segmented into eight occupation types, categorized by three characteristics: sex, skill, and level of formality. Each type of labor is perfectly mobile across the different sectors of production within a re- gion; in the absence of wage gaps across sectors, this implies a uniform wage across all sectors within a region. The wage is set according to the supply and demand of labor in each segment; flexible wages clear the markets for the two segments. The current version of MANAGE assumes market-clearing wages in the labor market with an upward sloping labor supply schedule and sluggish mobility of labor across sectors. The mod- el also features labor market segmentation (for example, rural versus urban). For capital markets, MANAGE distinguishes between two vintages of capital: old and new. In- dustries in decline release capital, and this is added to the available stock of “new” capital. This new capital is fully mobile across sectors. This allows the model to capture some rigidities in the capital market by assuming that declining sectors will first release the most mobile types of capital. Capital in expanding sectors earns the same rate of return, while capital in declining sectors has a lower rate of return. c. Household income and consumption Each household type supplies one or several of the different labor types and receives wages in return. The amount of labor supplied is exogenous to the model. Households also receive in- come and transfers from other agents, including the profits from asset holdings. Households use their earnings for consumption, savings, and transfers. Household demand is modeled using the constant-differences-in-elasticity (CDE) demand function that is the standard utility function used in other CGE models. MANAGE allows for a different specification of demanded commodities from supplied commodities. A transition matrix approach is used to convert consumer goods to supplied goods and also relies on a nested constant elasticity of substitution (CES) approach. The transition matrix is largely diag- onal, with consumed commodities directly mapped to supplied commodities. Energy demand is bundled into a single commodity and disaggregated by energy type using a CES structure that allows for inter-fuel substitution. Other final demand is handled similarly, though the ag- gregate expenditure function is a CES function rather than the CDE. Goods are evaluated at basic prices with tax wedges. The model incorporates trade and trans- port margins that add an additional wedge between basic prices and end-user prices. The trade and transport margins are differentiated across transport nodes—farm/factory gate to domestic markets and the border (for exports) and from port to end user (for imports). Import demand is modeled using the ubiquitous Armington assumption, i.e., goods with the same nomenclature are differentiated by region of origin (Armington 1969). This allows for im- perfect substitution between domestically produced goods and imported goods. The level of the CES elasticity determines the degree of substitutability across regions of origin. Domestic production is analogously differentiated by region of destination using the constant-elastici- ty-of-transformation (CET) function. The ability of producers to switch between domestic and foreign markets is determined by the level of the CET. The model allows for perfect transfor- mation, in which case the law of one price must hold. Market equilibrium for domestically produced goods sold domestically is assumed through market-clearing prices. By default, the small country assumption is made for export and import prices, and they are thus exogenous, i.e., export levels do not influence the price received by exporters, and import demand does not influence the import prices of cost, in- surance, and freight. d. Macroeconomic closures Macroeconomic closures determine how macro balances are restored after a shock. Specif- ically, these closures stipulate how MANAGE achieves (i) balanced government accounts, (ii) the macro equilibrium of the capital account (i.e., the investment and savings balance), and (iii) the macro equilibrium of the accounts with the rest of the world (i.e., external balance). The closure rules adopted in the model are discussed below. The government budget bal- ance is exogenous across different scenarios, as is the level of government spending (in real terms). Hence, the level of direct taxes is endogenous and adjusts, in response to policies and economic shocks, to cover any changes in revenues in order to keep the fiscal balance at the exogenous level. For the savings-investment balance, MANAGE assumes a savings-driven closure. Aggregate investment, which, together with an exogenous rate of depreciation, determines the next peri- od’s capital stock, is flexible to ensure that the investment cost will be equal to the value of the savings. The volume of available savings is determined by an exogenous level of foreign sav- ings, endogenous government savings, and endogenous household savings. In this context, an increase in government revenue from a new source of tax revenue, for example, would also be reflected in higher public savings and therefore stimulate current investment and growth. External balance ensures that the path of foreign liabilities is sustainable. This is achieved through an adjustment of the real exchange rate, while the current account is fixed by the available quantity of foreign savings. To maintain the current account constant, domestic pric- es adjust to generate appropriate changes in the volumes of imports and exports demanded. The main implication of this closure is that an increase in exports, for example, would generate an appreciation of the real exchange rate, penalizing the competitiveness of the non-mining sector—in other words, a Dutch disease effect. e. Dynamic behavior of the model • Capital accumulation. The capital stock in each period is the sum of depreciated capital from the previous period and new investment. The dynamic path follows the neoclassical growth framework. It employs a Solow-Swan growth mech- • Labor supply. For each type of labor, the maximum stock of labor available in each period anism, implying that the long-run growth rate of grows exogenously based on population projections of the working-age cohort (15–64 the economy is determined by three main factors: years old). capital accumulation, labor supply growth, and in- • Productivity. For the final determinant of growth, MANAGE assumes exogenous technical creases in productivity. The stock of capital is en- progress specific to sector and production factors. Thus, in the simulations, the real GDP dogenous, while the latter two are exogenously growth rate differs from the growth rate under the baseline scenario due to the policy or determined: shock being simulated. Specifically, policies or shocks affect real GDP growth through their effects on the accumulation of labor and capital. f. Caveats to MANAGE’s modeling framework MANAGE’s methodology assumes free and frictionless movement of labor and capital across sectors. This is a simplification since there might be friction that prevents factors from moving out of sectors in decline and into expanding sectors. Although the methodology has limita- tions, it illustrates a country’s potential to adjust and reap the benefits of the changing struc- ture of the economy or to absorb negative shocks, such as the negative impacts of climate change. The frictionless adjustment of labor and capital to external shocks and the immediate change in relative prices that are required to return to the equilibrium implies that the model is not well suited to analyze short-term shocks or business-cycle fluctuations. Like most CGE models, MANAGE does not capture the productivity effect of policy shocks, since technological progress is exogenous. This is another limitation, as recent economic literature emphasizes the potential benefits of trade diversification on productivity. 87 Endnotes BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | A nne x es 1 This analysis draws from Burger, Hendriks, and Ianchovichina (2020). 14 It includes self-generated resources, budget execution, transparency, open government, and effective tax 2 Colombia ranked on average 37th out of 156 countries in terms of SWB (according to the Gallup World Poll), collection. but only 73rd out of 156 countries in terms of GDP per capita (purchasing power parity [PPP]), based on the 15 It includes results measures in education, health, public services, and security and cohabitation. World Development Indicators for the period 2010–18. 16 Each component and its sub-subjects are rated between 1 and 100, with 100 an outstanding result. As the 3 See “The Human Capital Project,” a World Bank database at https://www.worldbank.org/en/publication/ composite index oscillates between 1 and 100, this range is divided into three qualitative classifications: human-capital. low (less than 45), medium (between 45 and 55), and high (more than 55). 4 See https://www.worldbank.org/en/publication/human-capital. 17 According to the document Medición Nuevo Índice de Desempeño Fiscal Territorial (2020), the index is mea- 5 For a detailed guide to calculating each component of the HCI, please refer to World Bank, “Guide to Calcu- sured taking into account the following dimensions: fiscal results (weighted 80 percent) and territorial fiscal lating a Subnational Human Capital Index” (Washington, DC: World Bank, 2019). management (20 percent). Each of them has some sub-items that are measured from 0 to 100. 6 If cities were lacking amenities within their urban perimeters, the proximity is calculated as the distance 18 This multisectoral index is suitable to understanding urban-rural linkages since it encompasses many of the from their centroid to the nearest amenity located outside of their urban area. variables identified in the literature. However, given the high degree of variables aggregation, we interpret 7 To have a unique value by city, we estimated the average travel distance weighted by the population size of the impact of this index on territorial development as the influence of the qualitative characteristics of cit- the city. ies on poverty dynamics. 8 Equal weight is given to all amenities. 19 DANE, “Cuadro de oferta y utilización a precios corrientes” for 2018, which includes transport and trade 9 It considers the level of attractiveness of the destination—the amenity in this case—discounted by the trav- margins. el impedance from a given block to the amenity located at the destination. The travel impedance is calcu- 20 DANE, “Encuesta Nacional de Presupuestos de los Hogares (ENPH),” 2016–2017, https://www.dane.gov. lated using distance-decay functions. We use four distance-decay functions following Reggiani, Bucci, and co/index.php/estadisticas-por-tema/pobreza-y-condiciones-de-vida/encuesta-nacional-de-presupues- Russo (2011) and consider the function that best fits the data for each city. tos-de-los-hogares-enph. 10 According to Colombian regulations, specifically the NTC 4595 (ICONTEC 2020) on the planning and design 21 Version 3.3 of the GAMS code was developed by Sherman Robinson and Scott McDonald in 2006. of school facilities and environments, 15 minutes is considered an adequate time to get to an education 22 In the future, this SAM can be further expanded and improved using more recent data from the DNP. For facility from a given origin. However, the 15 minutes are very relative depending on the mode of transport instance, the original 2015 SAM had five agricultural sectors (coffee, livestock, fisheries, forestry, and agri- used to travel between the origin and the destination. culture), but it could be expanded to 10 activities. Land could be included as an additional production fac- 11 To consider a unique measure for all types of facilities, and due to the lack of known standards for the other tor (it is currently aggregated with capital), and mixed income could be separated more accurately between types of facilities, this work uses the standard distance proposed by Paez et al. (2010) for health amenities: labor types and capital. Finally, the transfer matrix between economic agents could also be updated. 5 kilometers in urban contexts and 15 kilometers in rural contexts. 23 The theoretical framework relies on neoclassical assumptions of constant returns to scale and perfect com- 12 As is discussed in Berdegué et al. (2015). petition, where firms maximize profits to determine output supply and factor demands. 13 We used the Digital Elevation Model GTOPO30, SRTM by US Geological Survey (1996), which made it possi- ble to analyze an elevation raster with a resolution close to 1 kilometer and a vertical precision of about 30 meters. Intuitively, the index is built as the elevation difference between a reference point and close points surrounding it. We aggregate this index by taking the average of the cell values that intersect the geograph- ic polygon of interest. References 88 BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References Acar, Aysenur, Laurent Bossavie, and Mattia Makovec. 2019. “Do Firms Álvarez Marinelli, Horatio, Samuel Berlinski, and Matias Busso. 2019. Exit the Formal Economy after a Minimum Wage Hike? Quasi-Experi- “Remedial Education: Evidence From a Sequence of Experiments in mental Evidence from Turkey.” Policy Research Working Paper 8749, Colombia.” IDB Working Papers Series. Inter-American Development World Bank, Washington, DC. Bank, Washington, DC. Acemoglu, Daron, and David Autor. 2011. “Skills, Tasks and Technol- Amarante, Verónica, Rodrigo Arim, and Mauricio Santamaría. 2005. ogies: Implications for Employment and Earnings.” In Handbook of “Los Efectos de la Reforma Laboral de 2002 en el Mercado Laboral Labor Economics, vol. 4, part B, edited by David Card and Orley Ashen- Colombiano.” Perfil de Coyuntura Económica 6: 67–82. felter, 1043–71. Amsterdam: Elsevier. Arango-Aramburo, Santiago, Sean W.D. Turner, Kathryn Daenzer, Juan Acemoglu, Daron and Pascual Restrepo. 2017. “Robots and Jobs: Evi- P. Ríos-Ocampo, Mohamad I. Hejazi, Tom Kober, Andres C. Álva- dence from US Labor Markets.” Department of Economics Working rez-Espinosa, German D. Romero-Otalora, and Bob van der Zwaan. Paper 17-04, Massachusetts Institute of Technology, Cambridge, MA. 2019. “Climate Impacts on Hydropower in Colombia: a Multi-Model Assessment of Power Sector Adaptation Pathways.” Energy Policy Adams-Prassl, Abigail, Teodora Boneva, Marta Golin, and Christopher 128: 179–88. Rauh. 2020. “Inequality in the Impact of the Coronavirus Shock: Evi- dence from Real Time Surveys.” Journal of Public Economics 189. Armington, Paul. 1969. “A Theory of Demand for Products Distinguished by Place of Production.” International Monetary Fund Staff Papers 16 Adema, Joop, Yvonne Giesing, Anne Schönauer and Tanja Stitteneder. (1): 159–78. 2018. “Minimum Wages Across Countries.” ifo DICE Report, 16, no. 4. Artuc, Erhan, Luc Christiaensen, and Hernan Winkler. 2019. “Does Auto- Agnello, Luca, Vitor Castro, Joao Tovar Jalles, and Ricardo Sousa. 2015. mation in Rich Countries Hurt Developing Ones? Evidence from the “What Determines the Likelihood of Structural Reforms?” European U.S. and Mexico.” Policy Research Working Paper 8741, World Bank, Journal of Political Economy 37: 129–45. Washington, DC. Alaimo, Verónica, Mariano Bosch, Melany Gualavisí, and Juan Miguel Ashcraft, Adam, Ivan Fernández-Val, and Kevin Lang. 2013. “The Conse- Villa. 2017. “Measuring the Cost of Salaried Labor in Latin America quences of Teenage Childbearing: Consistent Estimates When Abor- and the Caribbean.” Technical Note IDB-TN-1291, Inter-American De- tion Makes Miscarriage Non-Random.” The Economic Journal 123 velopment Bank, Washington, DC. (571): 875–905. Alvaredo, Facundo, Lucas Chancel, Thomas Piketty, Emmanuel Saez, Attanasio, Orazio, and Miguel Székely. 1999. “An Asset-Based Approach and Gabriel Zucman. 2018. “World Inequality Report.” World Inequal- to the Analysis of Poverty in Latin America.” Working Paper R-376, Lat- ity Lab, Paris and Berkeley, CA. in American Research Network, Inter-American Development Bank, Washington, DC. 89 Avitabile, Ciro, Janina Cuevas, Raphael de Hoyos, and Julian Jamison. Barrera-Osorio, Felipe, Sandra García, Catherine Rodríguez, Fabio Sán- 2019. “Addressing High School Dropouts with a Scalable Intervention.” chez, and Mateo Arbeláez. 2018. “Concentrating Efforts in Low-Per- BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References Policy Research Working Paper 9085, World Bank, Washington, DC. forming Schools: Impact Estimates from a Quasi-Experimental Design.” Economics of Education Review 66: 73–91. Avitabile, Ciro, Ritika D’Souza, Roberta Gatti, and Emily Weedon Chap- man. 2020. “Insights from Disaggregating the Human Capital Index.” Barrera-Osorio, Felipe, Adriana D. Kugler, and Mikko I. Silliman. 2020. Working Paper 144839, World Bank, Washington, DC. “Hard and Soft Skills in Vocational Training: Experimental Evidence from Colombia.” NBER Working Paper 27548, National Bureau of Eco- Avner, Paolo, and Somick Vinay Lall. 2016. “Matchmaking in Nairobi: nomic Research, Cambridge, MA. The Role of Land Use.” Policy Research Working Paper 7904, World Bank, Washington, DC. Berdegué, Julio A., Fernanco Carriazo, Benjamin Jara, Felix Modrego, and Isidro Soloaga. 2015. “Cities, Territories, and Inclusive Growth: AXCO. 2020. “Non-Life Insurance Market Reports – Colombia.” Axco In- Unraveling Urban-Rural Linkages in Chile, Colombia, and Mexico.” surance Information Services, London. World Development 73: 56–71. Bachas, Pierre, Lucie Gadenne, and Anders Jensen. 2020. “Informality, Bernal, Raquel, Marcela Eslava, and Marcela Meléndez. 2015. “Taxing Consumption Taxes and Redistribution.” NBER Working Paper 27429, Where You Should: Formal Employment and Corporate Income vs. National Bureau of Economic Research, Cambridge, MA. Payroll Taxes in the Colombian 2012 Tax Reform.” Universidad de Los Andes, Bogotá. Báez, Javier E., Alan Fuchs, and Carlos Rodríguez-Castelán. 2017. “Shak- ing Up Economic Progress: Aggregate Shocks in Latin America and Betcherman, Gordon. 2015. “Labor Market Regulations: What Do We the Caribbean.” World Bank, Washington, DC. Know About Their Impacts in Developing Countries?” World Bank Re- search Observer 30 (1): 124–53. Bahia, Kalvin, Pau Castell, Genaro Cruz, Xavier Pedros, Tobias Pfutze, Carlos Rodriguez Castelan, and Hernan Winkler. 2020. “The Welfare Bogliacino, Francesco, Laura Maria Jiménez Lozano, and Daniel Reyes Effects of Mobile Broadband Internet: Evidence from Nigeria.” Policy Galvis. 2015. “Identificar la Incidencia de la Estratificación Socio- Research Working Paper 9230, World Bank, Washington, DC. económica Urbana Sobre la Segregación de los Hogares Bogotanos. (Identifying the Effect of the Socio-Economic Stratification on Urban Banco de la República (Bogotá). 2014. Economía de las grandes ciu- Segregation in Bogotá).” Investigaciones y Productos CID 24, Centro dades en Colombia: seis estudios de caso / Banco de la República, de Investigaciones para el Desarrollo, Universidad Nacional de Co- Gerson Javier Pérez y otros. -- Editor Luis Armando Galvis. -- Bogotá : lombia, Bogotá. Banco de la República. Borras, Saturnino M. Jr., Jennifer Franco, Cristóbal Kay, and Max Spoor. Banerjee, Abhijit, Rukmini Banerji, James Berry, Esther Duflo, Harini 2011. “Land Grabbing in Latin America and the Caribbean Viewed Kannan, Shobhini Mukerji, Marc Shotland, and Michael Walton. 2016. from Broader International Perspectives.” Food and Agriculture Or- “Mainstreaming an Effective Intervention: Evidence from Randomized ganization. Rome: FAO. Evaluations of ‘Teaching at the Right Level’ in India.” NBER Working Paper 22746, National Bureau of Economic Research, Cambridge, MA. Bosch, Mariano, Solange Berstein, Francesca Castellani, María Laura Oliveri, and Juan Miguel Villa. 2015. “Diagnóstico del Sistema Pre- Baron, Juan, Carmen Maura Taveras, and Janina Cuevas Zuñiga. visional Colombiano y Opciones de Reforma.” Nota Técnica 825, In- 2018. “Adaptive Technology to Help Improve Math Learning in the ter-American Development Bank, Washington, DC. Dominican Republic.” Education for Global Development (blog), December 19, https://blogs.worldbank.org/education/adap- tive-technology-help-improve-math-learning-dominican-republic. 90 Boshell, Francisco, Timothy S. Thomas, Vijay Nazareth, and Nicola Ce- Carter, Michael, and Christopher Barrett. 2006. “The Economics of Pov- nacchi. 2018. “Climate Change, Agriculture, and Adaptation Options erty Traps and Persistent Poverty: An Asset-Based Approach.” Journal BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References for Colombia.” IFPRI Discussion Paper 01790, International Food Pol- of Development Studies 42 (2): 178–99. icy Research Institute, Washington, DC. CEPAL (Comisión Económica para América Latina y el Caribe). 2020. Bourguignon, Francois. 2003. “The Growth Elasticity of Poverty Reduc- “Caminos Rurales: Vías Claves para la Producción, la Conectividad y el tion: Explaining Heterogeneity across Countries and Time Periods.” Desarrollo Territorial.” Facilitación, Comercio y Logística Boletín 377 (1). In Inequality and Growth. Theory and Policy Implications, edited by Theo S. Eicher and Stephen J. Turnovsky. Cambridge, MA: MIT Press. Chapman, Bruce, Timothy Higgins and Joseph Stiglitz. 2014. “Introduc- tion and Summary.” In Income Contingency Loans, edited by Bruce Burger, Martijn, Martijn Hendriks, and Elina Ianchovichina. 2020. “Hap- Chapman and Joseph E Stiglitz, 1–11. Palgrave Macmillan UK, 2014. py but Unequal: Differences in Subjective Well-Being Across Individ- uals and Space in Colombia.” Policy Research Working Paper 9554, Choi, Soon Kyu, Shahrzad Divsalar, Jennifer Flórez-Donado, Krystal Kit- World Bank, Washington, DC. tle, Andy Lin, Ilan H. Meyer, and Prince Torres-Salazar. 2019. “Estrés, Salud y Bienestar de las Personas LGBT en Colombia: Resultados de Bussolo, Maurizio, and Luis F. Lopez-Calva. 2014. Shared Prosperity: una Encuesta Nacional.” The Williams Institute, Los Angeles. Paving the Way in Europe and Central Asia. Europe and Central Asia Studies. Washington, DC: World Bank. Chomitz, Kenneth M., Piet Buys, and Timothy S. Thomas. 2005. “Quan- tifying the Rural-Urban Gradient in Latin America and the Caribbean.” Calderón, Sylvia, Andrés Camilo Alvarez, Ana Maria Loboguerrero, San- Policy Research Working Paper 3634, World Bank, Washington, DC. tiago Arango, Katherine Calvin, Tom Kober, Kathryn Daenzer, and Karen Fisher-Vanden. 2016. “Achieving CO2 Reductions in Colombia: Clemens, Jeffrey, and Michael Wither. 2019. “The Minimum Wage and Effects of Carbon Taxes and Abatement Targets.” Energy Economics the Great Recession: Evidence of Effects on the Employment and In- 56: 575–86. come Trajectories of Low-Skilled Workers.” Journal of Public Econom- ics 170 (C): 53–67. Campos, Ana, Niels Holm-Nielsen, Carolina Díaz, Diana M. Rubiano, Carlos Costa, Fernando Ramírez, and Eric Dickson. 2012. “Analysis of Coady, David, Devin D’Angelo, and Brooks Evans. 2019. “Fiscal Redistri- Disaster Risk Management in Colombia: A Contribution to the Cre- bution and Social Welfare.” Working Paper WP/19/51, International ation of Public Policies.” World Bank, Washington, DC. Monetary Fund, Washington, DC. Cárdenas, Jeisson, and Jaime Montana. 2020. “Possible Effects of Coro- Comola, Margherita, and Luiz De Mello. 2011. “How Does Decentralized navirus in the Colombian Labour Market.” Documento de Trabajo Minimum Wage Setting Affect Employment and Informality? The Case WP2-2020-006, Alieanza EFI, Colombia Científica, Bogotá. of Indonesia.” Review of Income and Wealth 57: 79–99. Carriazo Osorio, Fernando, and Mónica Juliana Reyes. 2012. “Territorios CONPES (Consejo Nacional de Política Económica y Social República Funcionales: un Análisis del Gradiente Rural-Urbano para Colombia.” de Colombia/National Council for Economic and Social Policy of Co- Documentos CEDE, Centro de Estudios sobre Desarrollo Económico, lombia). 2009. “Lineamientos para la consolidación de la Política de Bogotá. Mejoramiento Integral de Barrios (MIB).” Documento CONPES 3604, Departamento Nacional De Planeación, Bogotá. Carriazo Osorio, Fernando, Gracia N. Lozano, and Diana C. Tello Medi- na. Forthcoming. “Urban-Rural Linkages in Colombia.” ———. 2016. “Declaración de Importancia Estratégica del Sistema de Identificación de Potenciales Beneficiarios (SISBÉN IV).”  Documento CONPES 3877, Departamento Nacional De Planeación, Bogotá. 91 ———. 2019. “Política de Formalización Empresarial.” Documento CON- ———. 2018b. “Encuesta Nacional de Calidad de Vida.” DANE, Bogotá. PES 3956, Departamento Nacional De Planeación, Bogotá. BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References ———. 2018c. “Encuesta Nacional de Presupuesto de los Hogares.” ———. 2020a. “Estrategia Para la Implementación del Mecanismo de DANE, Bogotá. Compensación del Impuesto a las Ventas (IVA) a Favor de la Población más Pobre y Vulnerable.” Documento CONPES 3986, Departamento ———. 2018d. “Pobreza Multidimensional en Colombia.” DANE, Bogotá. Nacional De Planeación, Bogotá. ———. 2020. “Informe Sobre Cifras de Empleo y Brechas de Género. ———2020b. “Política Nacional de Movilidad Urbana y Region- Cambios en el Empleo en Actividades de Cuidado Remunerado a Raíz al.”  Documento CONPES 3991, Departamento Nacional De Pla- del COVID-19.” DANE, Bogotá. neación, Bogotá. De Ferranti, David, Guillermo E. Perry, William Foster, Daniel Lederman, ———. 2021. “Política para la Reactivación, la Repotenciación y el Cre- and Alberto Valdés. 2005. “Beyond the City. The Rural Contribution cimiento Sostenible e Incluyente: Nuevo Compromiso por el Futuro to Development.” World Bank, Washington, DC. de Colombia.” Documento CONPES 4023, Departamento Nacional De Planeación, Bogotá. Del Carpio, Ximena V., Ha Nguyen, Laura Pabon, and Liang Choon Wang. 2015. “Do Minimum Wages Affect Employment? Evidence from the Cucagna, Emilia, and Javier Romero. 2021. “The Gendered Impacts Manufacturing Sector in Indonesia.” IZA Journal of Labor & Develop- of COVID-19 on Labor Markets in Latin America and the Caribbean.” ment 4 (1): 1–35. Gender Innovation Lab for Latin America and the Caribbean, World Bank, Washington, DC. Del Carpio, Ximena; Pabon, Laura. 2014. Minimum Wage Policy : Les- sons with a Focus on the ASEAN Region. World Bank, Washington, Cuesta, Jose, and Julieth Pico. 2020a. “COVID-19 Affects Everyone but DC. © World Bank. https://openknowledge.worldbank.org/han- Not Equally: The Gendered Poverty Effects of the COVID-19 Pandemic dle/10986/19027 License: CC BY 3.0 IGO in Colombia.” World Bank, Washington, DC. Del Carpio, Ximena V., Jose A. Cuesta, Maurice Kugler, Gustav Hernan- ———. 2020b. “The Equity Effects of Cadasters with an Application to dez, and Gabriel Piraquive. 2020. “Equity Aspects of Jobs and Eco- Colombia.” Poverty and Equity Global Practice, World Bank, Wash- nomic Transformation (JET) in Colombia: What Effects Could Global ington, DC. Unpublished manuscript. Value Chain and Digital Infrastructure Development Policies Have on Poverty and Inequality after COVID-19?” Background paper for this Cueva, Ronald, Ximena V. Del Carpio, and Hernan Winkler. 2021. “The La- report. Unpublished. bor Market Impacts of COVID-19 in Peru: Insights from a Longitudinal Household Survey.” International Monetary Fund, Washington, DC. Delhey, Jan, and Ulrich Kohler. 2011. “Is Happiness Inequality Immune to Income Inequality? New Evidence Through Instrument-Effect-Cor- DANE (Departamento Administrativo Nacional de Estadística/National rected Standard Deviations.” Social Science Research 40 (3): 742–56. Administrative Department of Statistics). 2017. “Indicadores Básicos de Tenencia y Uso de Tecnologías.” DANE, Bogotá. Dervis, K.; de Melo, J.; Robinson, S..1982. General equilibrium models for development policy. A World Bank research publication Washing- ———. 2005. “Censo Nacional de Población y Vivienda 2005.” DANE, ton, D.C.: World Bank Group. Bogotá. Devadas, Sharmila, and Young Eun Kim. 2020 “Exploring the Potential ———. 2018a. “Censo Nacional de Población y Vivienda 2018.” DANE, of Gender Parity to Promote Economic Growth.” Research and Policy Bogotá. Brief 39, World Bank, Washington, DC. 92 De Melo, Jaime; Tarr, David (1992). A General Equilibrium Analysis of Döll, Sebastian. 2009. “Climate Change Impacts in Computable General U.S. Foreign Trade Policy. MIT Press. Equilibrium Models: an Overview.” HWWI Research Papers, Hamburg BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References Institute of International Economics, Hamburg. Dickens, Richard. 2015. “How Are Minimum Wages Set?” IZA World of Labor. D’Souza, Ritika. 2019. “Guide to Calculating a Subnational Human Cap- ital Index.” World Bank, Washington, DC. Diego-Rosell, Pablo, Robert Tortora, and James Bird. 2018. “Interna- tional Determinants of Subjective Well-Being: Living in a Subjectively Duque, Juan C., Vicente Royuela, and Miguel Noreña. 2013. “A Step- Material World.” Journal of Happiness Studies 19 (1): 123–43. wise Procedure to Determine a Suitable Scale for the Spatial Delinea- tion of Urban Slums.” In Defining the Spatial Scale in Modern Regional Dixon, Peter B., and Dale W. Jorgensen. 2012. Handbook of Computable Analysis. Advances in Spatial Science, edited by E. Fernandez and F. General Equilibrium Modeling. Amsterdam: Elsevier. Rubiera Morollón, 237–54. New York: Springer-Verlag. DNP (Departamento Nacional de Planeación)/National Department of Duque, Lola et al. (forthcoming). “A Literature Review on Inequality and Planning). 2014. Misión Sistema de Ciudades. Bogotá: DNP. Space in Colombia.” ———. 2014-2018. Todos por un Nuevo País. Bogotá D.C. Duval, Romain, Davide Furceri, and Jakob Miethe. 2018. “The Needle in the Haystack: What Drives Labor and Product Market Reforms in ———. 2017-2018. Evaluación de Impacto del Programa Todos a Apren- Advanced Countries?” IMF Working Paper WP/18/01, International der. Bogotá D.C. Monetary Fund, Washington, DC. ———. 2018. “Informe de Resultados MDM 2018.” Medición del Desem- EC (European Commission). 2019. Key Competences for Lifelong Learn- peño Municipal, DNP, Bogotá. ing. Directorate-General for Education, Youth, Sport and Culture. Brussels: European Commission. ———. 2019a. “Plan Nacional de Desarrollo 2018–2022. ‘Pacto por Co- lombia. Pacto por la Equidad.’” DNP, Bogotá. Echavarría, Juan José, Iader Giraldo, and Fernando Jaramillo. 2019. “Cadenas Globales de Valor, Crecimiento y Protección Arancelaria en ———. 2019b. “Resultados Índice de Ciudades Modernas 2019.” DNP, Colombia.” Borradores de Economía 1080, Banco de la República Co- Bogotá. lombia, Bogotá. ———. 2020. “Medición Nuevo Índice de Desempeño Fiscal Territorial.” Elbers, Chris, Jean O. Lanjouw, and Peter Lanjouw. 2003. “Micro-Level Subdirección de Descentralización y Fortalecimiento Fiscal, Direc- Estimation of Poverty and Inequality.” Econometrica 71 (1): 355–64. ción de Descentralización y Desarrollo Regional, DNP, Bogotá. Ernst & Young. 2020. “Worldwide VAT, GST and Sales Tax Guide.” DNP and Fondo Acción. 2015. “Valoración Económica de la Degradación Ambiental en Colombia 2015.” DNP, Bogotá. Esguerra, Maria del Pilar, and Sergio Parra Ulloa. 2016. “Colombia, por Fuera las Cadenas Globales de Valor: ¿Causa o Síntoma del Bajo De- DNP, BID (Banco Interamericano de Desarrollo), and CEPAL (Comisión sempeño Exportador?” Borradores de Economía 966, Banco de la Económica para América Latina y el Caribe). 2014. Impactos Económi- República Colombia, Bogotá. cos del Cambio Climático en Colombia – Síntesis. Bogotá: DNP, BID, and CEPAL. Eslava, Marcela, John Haltiwanger, Adriana Kugler, and Maurice Kugler. 2004. “The Effects of Structural Reforms on Productivity and Profit- ability Enhancing Reallocation: Evidence from Colombia.” Journal of Development Economics 75 (2): 333–71. 93 ———. 2010. “Factor Adjustments after Deregulation: Panel Plant Ev- Ferreyra, Maria Marta, and Mark Roberts. 2018. “Raising the Bar for Pro- idence from Colombia.” Review of Economics and Statistics 92 (2): ductive Cities in Latin America and the Caribbean.” Latin American BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References 378–91. and Caribbean Studies, World Bank, Washington, DC. ———. 2013. “Trade Reforms and Market Selection: Evidence from Man- FINAGRO (Fondo para el Financiamiento del Sector Agropecuario). ufacturing Plants in Colombia.” Review of Economic Dynamics 16 (1): 2018. “Inteligencia de Mercado: Cacao.” FINAGRO, Unidad de Gestión 135–58. de Riesgos Agropecuarios, Bogotá. Espinel, Zelde, Roberto Chaskel, Ryan C. Berg, Hermes Jose Florez, Sil- Flórez, Luz Adriana, Leonardo Fabio Morales, Daniel Medina, and José via L. Gaviria, Oscar Bernal, Kim Berg, Carlos Muñoz, Marisa G. Lar- Lobo. 2020. “Labor Flows Across Firm Size, Age, and Economic Sector kin, and James M. Shultz. 2020. “Venezuelan Migrants in Colombia: in Colombia vs. the United States.” Small Business Economics 1–32. COVID-19 and Mental Health.” The Lancet Psychiatry 7 (8): 653–55. Forero, David, and Victor Saavedra. 2019. “Los 10 Pasos para Hacer de FAO (Food and Agriculture Organization). 2020. “Proyecto ‘Sembrando Colombia la Mejor Educada de América Latina.” Fedesarrollo, Bogotá. Capacidades’ Fortelecerá a los Agricultores Familiares Colombianos.” FAO, Rome. Fraiberger, Samuel P., Pablo Astudillo, Lorenzo Candeago, Alex Chunet, Nicholas K. Jones, Maham F. Khan, Bruno Lepri, Nancy L. Gracia, Fawzy, Samiha, Ali I. Osman, John Doran, and David W. Rooney. 2020. Lorenzo Lucchini, Emanuele Massaro, and Aleister Montfort. 2020. “Strategies for Mitigation of Climate Change: a Review.” Environmen- “Uncovering Socioeconomic Gaps in Mobility Reduction During the tal Chemistry Letters 18: 2069–94. COVID-19 Pandemic Using Location Data.” arXiv:2006.15195, Cornell University, Cornell, NY. Fay, Marianne, Danny Leipziger, Quentin Wodon, and Tito Yepes. 2005. “Achieving Child-Health-Related Millennium Development Goals: Freire, German, Carolina Diaz-Bonilla, Orellana Schwartz, Steven Soler, The Role of Infrastructure.” World Development 33 (8): 1267–84. Jorge Lopez, and Flavia Carbonari. 2018. “Afro-Descendants in Latin America: Toward a Framework of Inclusion.” World Bank, Washing- Fay, Marianne., Luis A. Andres, Charles J. E. Fox, Ulf G. Narloch, Stephane ton, DC. Straub, and Michael A. Slawson. 2017. “Rethinking Infrastructure in Latin America and the Caribbean: Spending Better to Achieve More.” Fryer, Roland, and Meghan Howard. 2017. “High-Dosage Tutoring and World Bank, Washington, DC. Reading Achievement: Evidence from New York City.” NBER Working Paper 23792, National Bureau of Economic Research, Cambridge MA. FEDEGAN (Federación Colombiana de Ganaderos ).2018. “Cifras de Ref- erencia del Sector Ganadero Colombiano.” FEDEGAN, Bogotá. Fuchs-Schündeln, Nicola, Moritz Kuhn, and Michèle Tertilt. 2020. “The Short-Run Macro Implications of School and Child-Care Closures.” Feenstra, Robert C., Robert Inklaar, and Marcel P. Timmer. 2015. “The CEPR Discussion Paper 14882, Centre for Economic Policy Research, Next Generation of the Penn World Table.” American Economic Review London. 105 (10): 3150–82. Gáfaro, Margarita, Ana Maria Ibañez, and David Zarruk. 2012. “Equidad Fernández, Christina, and Leonardo Villar. 2017. “The Impact of Low- y Eficiencia Rural en Colombia: una Discusión de Políticas para el Ac- ering the Payroll Tax on Informality in Colombia.” Economía 18 (1): ceso a la Tierra.” Documento CEDE 2012-38, Centro de Estudios sobre 125–55. Desarrollo Económico, Bogotá. 94 Galvis-Aponte, Luis Armando, Adolfo Meisel-Roca, Gerson Javier Pérez- Gonzalez-Diaz, Jairo M., Juan Fernando Cano, and Victor Pereira-San- Valbuena, and Jóse  R. Gamarra-Vergara. 2008. “Geografıa Económica chez. 2020. “Psychosocial Impact of COVID-19-Related Quarantine: BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References y Análisis Espacial en Colombia.” Banco de la República de Colombia, Reflections after the First Case of Suicide in Colombia.” Cadernos de Bogotá. Saúde Pública 36 (6). Gamarra-Vergarra, Jose R. 2007. “Pobreza Rural y Transferencia de Tec- Guereña, Arantxa. 2017. “Radiografía de la Desigualdad. Lo que nos nología en la Costa Caribe.” Documentos de Trabajo Sobre Economía Dice el Ultimo Censo Agropecuario Sobre la Distribución de la Tierra Regional 89, Centro de Estudios Económicos Regionales (CEER) del en Colombia. ” OXFAM International, Bogotá. Banco de la República, Cartagena. Gyeke-Dako, Hallward-Driemeier, Mary, and Gaurav Nayyar. 2017. Trou- Gamboa, Luis, and Erika Londoño. 2015. “Assessing Educational Unfair ble in the Making? The Future of Manufacturing-Led Development. Inequalities at a Regional Level in Colombia.” Lecturas de Economia Washington, DC: World Bank. 83: 97–133. Haltiwanger, John, Adriana Kugler, Maurice Kugler, Alejandro Micco, García, Sandra, Dario Maldonado, and Luis Jaramillo. 2016. “Gradu- and Carmen Pagés. 2004. “Effects of Tariffs and Real Exchange Rates ación de la Educación Media, Asistencia e Inasistencia a la Educación on Job Reallocation: Evidence from Latin America.” Journal of Policy Media.” Documentos de Trabajo 34, Escuela de Gobierno Alberto Lle- Reform 7 (4): 189–208. ras Camargo, Bogotá. Ham, Marteen van, Tiit Tammaru, Elise de Vuijst, and Merle Zwiers. García, Sandra, Dario Maldonado, Marcella Acosta, Nicolas Castro, Da- 2016. “Spatial Segregation and Socio-economic Mobility in European vid Granada, Érika Londoño, and Harold Villalba. 2016. “Característi- Cities.” IZA Discussion Paper 10277, Forschungsinstitut zur Zukunft cas de la Oferta de la Educación Media en Colombia.” Documentos der Arbeit/Institute for the Study of Labor, Bonn. de Trabajo 33, Escuela de Gobierno Alberto Lleras Camargo, Bogotá. Handy, Susan, and Debbie A. Niemeier. 1997. “Measuring Accessibility: García, Sandra, Dario Maldonado, Guillermo Perry, Catherine Rodrí- An Exploration of Issues and Alternatives.” Environment and Planning guez, and Juan E. Saavedra. 2014. “Tras la Excelencia Docente. Cómo A 29 (7): 1175–94. Mejorar la Calidad de la Educación para Todos los Colombianos.” Fundación Compartir, Bogotá. Hatayama, Maho, Mariana Viollaz, and Hernan Winkler. 2020. “Jobs’ Amenability to Working from Home: Evidence from Skills Surveys Garrote Sanchez, Daniel, Nicholas Gomez Parra, Ozden Caglar, Bob Ri- for 53 Countries.” Policy Research Working Paper 9241, World Bank, jkers, Mariana Viollaz, and Hernan Winkler. 2020. “Who on Earth Can Washington, DC. Work from Home?” Policy Research Working Paper 9347, World Bank, Washington, DC. Heller, Sara B., Anuj K. Shah, Jonathan Guryan, Jens Ludwig, Senhil Mullainathan, and Harold A. Pollack. 2015. “Thinking, Fast and Slow? Gaviria Uribe, Alejandro. 2005. “La Reforma Laboral de 2002. ¿Funcionó Some Field Experiments to Reduce Crime and Dropout in Chicago.” o No?” Coyuntura Económica 35 (1): 73–103. NBER Working Paper 21178, National Bureau of Economic Research, Cambridge MA. Geurs, Karst T., and Bert van Wee. 2004. “Accessibility Evaluation of Land-Use and Transport Strategies: Review and Research Directions.” Hernandez, Daniel Oviedo, and Helena Titheridge. 2016. “Mobilities of Journal of Transport Geography 12 (2): 127–40. the Periphery: Informality, Access and Social Exclusion in the Urban Fringe in Colombia.” Journal of Transport Geography 55 (C): 152–64. Gobillon, Laurent, Harris Selod, and Yves Zenou. 2007. “The Mecha- nisms of Spatial Mismatch.” Urban Studies 44 (12): 2401–27. 95 Hernandez, Daniel Oviedo, and Julio D. Dávila. 2016. “Transport, Urban ICONTEC. 2020. Planeamiento y diseño de instalaciones y ambientes Development and the Peripheral Poor in Colombia. Placing Splinter- escolares. NTC. 4595. BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References ing Urbanism in the Context of Transport Networks.” Journal of Trans- port Geography 51: 180–92. ILO (International Labour Organization). 2018. Enfoques innovadores para garantizar la protección social universal para el futuro del traba- Hirschman, Alberto, and Michael Rothschild. 1973. “The Changing Toler- jo. Geneva: ILO. ance for Income Inequality in the Course of Economic Development.” The Quarterly Journal of Economics 87 (4): 544–66. ———. 2020. Teleworking During the COVID-19 Pandemic and Beyond: A Practical Guide. Geneva: ILO. Holin, Mary Joel, Larry Buron, Gretchen Locke, and Alvaro Cortes. 2003. “Interim Assessment of the HOPE VI Program Cross-site Report.” Of- IMF (International Monetary Fund). 2017. Tackling Inequality. World Eco- fice of Policy Development and Research, U.S. Department of Hous- nomic and Financial Surveys, Fiscal Monitor. Washington, DC: IMF. ing and Urban Development, Washington, DC. Jaramillo, Martin, Juan Pablo Castro, Joel Brounen, Maria Goretti Es- Hoyos, Natalia, Jaime Escobar, Juan C. Restrepo, A. M. Arango, and quivel, and Carlos Alberto Pérez. 2020. “Cumpliendo con los Acuer- Juan C. Ortiz. 2013. “Impact of the 2010–2011 La Niña Phenomenon dos Cero-Deforestación en Colombia: Barreras y Oportunidades.” in Colombia, South America: The Human Toll of an Extreme Weather Solidaridad, Tropical Forest Alliance, and Climate Focus, Bogotá. Event.” Applied Geography 39: 16–25. Kain, John F. 1968. “Housing Segregation, Negro Employment, and Ibañez, Ana Maria, and Juan Carlos Muñoz. 2011. “La Persistencia de Metropolitan Decentralization.” The Quarterly Journal of Economics la Concentración de la Tierra en Colombia: ¿Qué Pasó Entre 2000 y 82 (2): 175–97.  2010? ” Notas de Política 9, Universidad de los Andes, Bogotá. Kelly, Tim, Aleksandra Liaplina, Shawn W. Tan, and Hernan Winkler. Ibáñez, Ana Maria, and Andres Moya. 2006. “The Impact of Intra-State 2017. Reaping Digital Dividends: Leveraging the Internet for Develop- Conflict on Economic Welfare and Consumption Smoothing: Empiri- ment in Europe and Central Asia. Washington, DC: World Bank. cal Evidence for the Displaced Population in Colombia.” HiCN Work- ing Paper 23, Households in Conflict Network, Brighton, UK. Khemani, Stuti. 2017. “Political Economy of Reform.” World Bank Policy Research Working Paper 8224, World Bank, Washington, DC. Ibañez, Ana Maria, and Pablo Querubin. 2004. “Acceso a Tierras y Despla- zamiento Forzado en Colombia.” Document CEDE 2004-23, Centro de Klinksy, Sonja, David Waskow, Wendi Bevins, Eliza Northrop, Robert Estudios sobre Desarrollo Económico, Bogotá. Kutter, Laura Weatherer, and Paul Joffe. 2015. Building Climate Eq- uity: Creating a New Approach from the Ground Up. World Resources ICFES (Instituto Colombiano para la Evaluación de la Educación/Co- Institute: Washington, DC. lombian Institute for the Evaluation of Education). 2017.” Resultados pruebas SABER 5 2017”. Kugler, Adriana, and Maurice Kugler. 2009. “Labor Market Effects of Pay- roll Taxes in Developing Countries: Evidence from Colombia.” Eco- ———. 2019.” Resultados pruebas SABER 11 2019”. nomic Development and Cultural Change 57 (2): 335–58. ———. 2020. Informe Nacional de Resultados para Colombia - PISA 2018. Kugler, Adriana, Maurice Kugler, Laura Ripani, and Rodimiro Rodrigo. Bogotá: ICFES. 2020. “U.S. Robots and their Impacts in the Tropics: Evidence from Colombian Labor Markets.” NBER Working Paper 28034, National Bu- reau of Economic Research, Cambridge, MA. 96 Lau, Charlotte, Andy Jarvis, and Julian Ramírez. 2010. “Colombian Ag- Manyika, James, Jaana Remes, Richard Dobbs, Javier Orellana, and Fa- riculture: Adapting to Climate Change.” CIAT Policy Brief 1, Interna- bian Schaer. 2012. “Urban America: US Cities in the Global Economy.” BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References tional Center for Tropical Agriculture, Cali. McKinsey Global Institute. Loayza, Norman V., Ana Maria Ovíedo, and Luis Servén. 2005. “The Im- Marcos, Ana. 2018. “Los Estratos en Colombia: Eres el Lugar en el que pact of Regulation on Growth and Informality: Cross-Country Evi- Vives.” El País, April 22. dence.” World Bank, Washington, DC. Melendez, Marcela, and María José Uribe. 2012. “International Product Lo Bello, Salvatore, Maria Laura Sanchez-Puerta, and Hernan Winkler. Fragmentation and the Insertion of Latin America and the Caribbean 2019. “From Ghana to America: The Skill Content of Jobs and Eco- in Global Production Networks: Colombian Case Studies.” IDB Work- nomic Development.” Policy Research Working Paper 8758, World ing Paper IDB-WP-374, Inter-American Development Bank, Washing- Bank, Washington, DC. ton, DC. Lora, Eduardo, and Mauricio Olivera. 2004. “What Makes Reforms Likely: Meltzer, Joshua Paul, and Camila Pérez Marulanda. 2016. “Digital Co- Political Economy Determinants of Reforms in Latin America.” Jour- lombia: Maximizing the Global Internet and Data for Sustainable and nal of Applied Economics 7 (1): 99–135. Inclusive Growth.” Working Paper 96, Global Economy and Develop- ment, Brookings Institution, Washington, DC. Lozano-Rojas, Felipe, Maureen A. Pirog, and Pedro Cerdan-Infantes. “Equity and Sustainability in Higher Education Financial Aid: Analy- Messina, Julian, and Joana Silva. 2019. “Twenty Years of Wage Inequali- sis of Income Contingent Loans in Colombia.” Journal of Student Fi- ty in Latin America.” IDB Working Paper IDB-WP-1041, Inter-American nancial Aid, forthcoming. Development Bank, Washington, DC. Luna Bazaldua, Diego, Victoria Levin, and Julia Liberman. 2020. “Guid- Minambiente (Ministerio del Ambiente y Desarrollo Sostenible). 2012. ance Note on Using Learning Assessment in the Process of School “Plan Nacional de Adaptación al Cambio Climático (PNACC).” ABC: Reopening.” Washington, D.C.: World Bank Group. Adaptación Bases Conceptuales. Marco Conceptual y Lineamientos, Departamento Nacional de Planeación, Bogotá. Lustig, Nora, 2018. CEQ Handbook. Estimating the Impact of Fiscal Policy on Inequality and Poverty. New Orleans, LA: CEQ Institute at Tulane Ministerio de Salud/Ministry of Health. 2015. “Encuesta Nacional de De- University and Brookings Institution Press. mografía y Salud (ENDS).” Tomo 2. Ministerio de Salud, Bogotá. Lustig, Nora, Valentina Martinez Pabon, Federico Sanz, and Stephen D. Ministerio de Vivienda, Ciudad y Territorio/Ministry of Housing, City and Younger. 2020. “The Impact of COVID-19 Lockdowns and Expanded Territory. 2020. “Análisis de capacidades y ambientes del Ministerio Social Assistance on Inequality, Poverty and Mobility in Argentina, de Vivienda, Ciudad y Territorio.” Bogotá: MVCT. Brazil, Colombia and Mexico.” Working Paper 92, Commitment to Eq- uity Institute, Tulane University, New Orleans. Molina, Oscar D., and Christian Bernhofer. 2019. “Projected Climate Changes in Four Different Regions in Colombia.” Environmental Sys- Ministerio de Agricultura y Desarrollo Rural (MADR) (2019), Cadena tems Research 8 (1). Agroindustrial de la panela. Dirección de Cadenas Agrícolas y Fore- stales. Diciembre 2019. Link: https://sioc.minagricultura.gov.co/Pan- Moller, Lars Christian. 2012. “Fiscal Policy in Colombia. Tapping its Po- ela/Documentos/2019-12-30%20Cifras%20Sectoriales.pdf tential for a More Equitable Society.” Policy Research Working Paper 6092, World Bank, Washington DC. 97 Montenovo Laura, Xuan Jiang, Felipe Lozano Rojas, Ian M. Schmutte, Ndabeni, Lindile L. 2016. “An Analysis of Rural-Urban Linkages and their Kosali I. Simon, Bruce A. Weinberg, and Coady Wing, 2020. “Determi- Implications for Policies that Sustain Development in a Space Con- BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References nants of Disparities in COVID-19 Job Losses.” NBER Working Paper Se- tinuum.” Global Monitoring Report, 1–59. ries 27132, National Bureau of Economic Research, Cambridge, MA. Neusser, Klaus, and Maurice Kugler. 1998. “Manufacturing Growth and Morales, Leonardo F., Didier Hermida, and Eleonora Dávalos. 2019. “In- Financial Development: Evidence from OECD Countries.” The Review teractions between Formal and Informal Labor Dynamics: Revealing of Economics and Statistics 80 (4): 638–46. Job Flows from Household Surveys.” Borradores de Economía 1090, Banco de la República Colombia, Bogotá. Nijman, J., & Wei, Y. D. 2020. Urban inequalities in the 21st century econo- my. Applied Geography, 117(April), 102188. https://doi.org/10.1016/j. Morales, Leonardo F., and Carlos Medina. 2017. “Assessing the Effect of apgeog.2020.102188 Payroll Taxes on Formal Employment: the Case of the 2012 Tax Re- form in Colombia.” Economía 18 (1): 75–124. Núñez, Jairo. 2005. “Éxitos y Fracasos de la Reforma Laboral en Colom- bia.” Documento Cede 2005-43, Centro de Estudios sobre Desarrollo Mura, Ivan, Juan F. Franco, Laura Bernal, Nicolas Melo, Juan J. Díaz, and Económico, Bogotá. Raha Akhavan-Tabatabaei. 2020. “A Decade of Air Quality in Bogotá: A Descriptive Analysis.” Frontiers in Environmental Science 8. Nuñez, Jairo, Sergio Olivieri, Julieth Parra, and Julieth Pico. 2020. “The Distributive Impact of Taxes and Expenditures in Colombia.” Policy Muralidharan, Karthik, Abhijeet Singh, and Alejandro J. Ganimian. 2017. Research Working Paper 9171, World Bank, Washington, DC. “Disrupting Education? Experimental Evidence on Technology-Aided Instruction in India.” NBER Working Paper 22923, National Bureau of Núñez, Jairo, Julieth Parra, and Gabriel Piraquive. 2017. “Desigualdad Economic Research, Cambridge MA. de la Riqueza y el Ingreso en Colombia.” Departamento Nacional de Planeación, Bogotá. Nakicenovic, Nebojsa, and Rob Swart, eds. 2000. Emissions Scenarios. Intergovernmental Panel on Climate Change. Cambridge, UK: Cam- Observatorio Dinámicas del Territorio (2014). “Riesgos, Desastres y De- bridge University Press. sarrollo en Bogotá” (Rev. 19), Instituto Distrital de Gestión de Riesgos y Cambio Climático (IDIGER) 2015, Secretaría Distrital de Planeación Nantulya, Vinand, and Michael Reich. 2002. “The Neglected Epidemic: (SDP). Road Traffic Injuries in Developing Countries.” British Medical Journal 324 (7346): 1139–41. OECD (Organisation for Economic Co-operation and Development). 2014. “OECD Territorial Reviews: Colombia 2014.” Paris, OECD. Narayan, Ambar, Roy Van der Weide, Alexandru Cojocaru, Christoph Lakner, Silvia Redaelli, Daniel Gerszon Mahler, Rakesh Gupta N. Ra- ———. 2017a. The Governance of Land Use - Country Fact Sheet Germa- masubbaiah, and Stefan Thewissen. 2018. Fair Progress? Econom- ny. Paris: OECD. ic Mobility across Generations around the World. Washington, DC: World Bank. ———. 2017b. The Governance of Land Use - Country Fact Sheet Chile. Paris: OECD. Nataraj, Shanthi, Francisco Perez-Arce, Krishna Kumar, and Sinduja V. Srinivasan. 2014. “The Impact of Labor Market Regulation on Em- ———. 2018a. Divided Cities:  Understanding Intra-Urban Inequalities. ployment in Low-Income Countries: A Meta-analysis.” Journal of Eco- Paris: OECD. nomic Surveys 28 (3): 551–72. ———. 2018b. “Educación en Colombia: Aspectos Destacados.” Reviews of National Policies for Education, OECD, Paris. 98 ———. 2019. OECD Economic Surveys: Colombia 2019. Paris: OECD. Patino, J. E. (2020). Analyzing long-term availability of urban green space by socioeconomic status in Medellin, Colombia, using open BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References ———. 2020. PISA 2018 Results (Volume V): Effective Policies, Successful data and tools. 2020 IEEE Latin American GRSS & ISPRS Remote Schools. Paris: OECD. Sensing Conference (LAGIRS), 87–92. https://doi.org/10.1109/LA- GIRS48042.2020.9165672 OECD and EC (European Commission). 2020. “Cities in the World: A New Perspective on Urbanisation.” Highlights, OECD Urban Studies, OECD Pérez, Jorge Pérez. 2020. “The Minimum Wage in Formal and Informal and EC, Paris and Brussels. Sectors: Evidence from an Inflation Shock.” World Development 133. OECD and World Bank. 2012. Tertiary Education in Colombia. Reviews Peters, Jeffrey C. 2016. “The GTAP-Power Data Base: Disaggregating the for National Policies for Education. Paris and Washington, DC: OECD Electricity Sector in the GTAP Data Base.” Journal of Global Economic and World Bank. Analysis 1 (1): 209–50. Olsson, Lennart, Humberto Barbosa, Suruchi Bhadwal, Annette Cow- Phillips, David, Ross Warwick, Maya Goldman, Karolina Goraus, Gabrie- ie, Kenel Delusca, Dulce Flores-Renteria, Kathleen Hermans, Esteban la Inchauste, Tom Harris, and Jon Jellema. 2018. “Redistribution via Jobbagy, Werner Kurz, and Diqiang Li. 2019. “Land Degradation.” In VAT and Cash Transfers: An Assessment in Four Low and Middle-In- Climate Change and Land: an IPCC Special Report on Climate Change, come Countries.” CEQ Working Paper 78, Commitment to Equality Desertification, Land Degradation, Sustainable Land Management, Institute, Tulane, LA. Food Security, and Greenhouse Gas Fluxes In Terrestrial Ecosystems. Geneva: Intergovernmental Panel on Climate Change. Pigato, Miria A., ed. 2019. Fiscal Policies for Development and Climate Action. Washington, DC: World Bank. Ong, Paul, and Evelyn Blumenberg. 1998. “Job Access , Commute and Travel Burden Among Welfare Recipients.” Urban Studies 35 (1) 77–93. Piper, Benjamin, Yasmin Sitabkhan, Jessica Mejía, and Kellie Betts. 2018. “Effectiveness of Teachers’ Guides in the Global South: Script- Ortiz Royero, Juan Carlos. 2012. “Exposure of the Colombian Caribbe- ing, Learning Outcomes, and Classroom Utilization.” Occasional Pa- an Coast, including San Andrés Island, to Tropical Storms and Hurri- per, RTI International, Research Triangle Park, NC. canes, 1900–2010.” Natural Hazards 61 (2): 815–27. Popkin, Susan J., Bruce Katz, Mary K. Cunningham, Karen D. Brown, Packard, Truman, Ugo Gentilini, Margaret Grosh, Philip O’Keefe, Robert Jeremy Gustafson, and Margery A. Turner. 2004. “A Decade of HOPE Palacios, David Robalino, and Indhira Santos. 2019. “Protecting All: VI: Research Findings and Policy Challenges.” The Urban Institute, Risk Sharing for a Diverse and Diversifying World of Work.” Working Washington, DC.  Paper 141572, World Bank, Washington, DC. Quirós, Tatiana Peralta, and Shomik R. Mehndiratta. 2015. “Accessibility Paez, Antonio, Ruben G. Mercado, Steven Farber, Catherine Morency, Analysis of Growth Patterns in Buenos Aires, Argentina: Density, Em- and Matthew Roorda. 2010. “Accessibility to Health Care Facilities in ployment, and Spatial Form.” Transportation Research Record 2512 Montreal Island: An Application of Relative Accessibility Indicators (1): 101–109. from the Perspective of Senior and Non-Senior Residents.” Interna- tional Journal of Health Geographics 9: 1–15.  Ramirez, M., and K. Perez. 2019. “Case Study on the Climate Resilience of Sustainable Livestock Production using Silvopastoral Systems.” Patino, Jorge E., Juan C. Duque, Josep E. Pardo-Pascual, and Luis A. World Bank, Washington, DC. Ruiz. 2014.  “Using Remote Sensing to Assess the Relationship Be- tween Crime and the Urban Layout.” Applied Geography 55: 48–60.  Ramirez-Villegas, Julian, Mike Salazar, Andy Jarvis, and Carlos E. Navar- ro-Racines. 2012. “A Way Forward on Adaptation to Climate Change in Colombian Agriculture: Perspectives towards 2050.”  Climatic Change 115 (3-4): 611–28. 99 Reggiani, Aura, Pietro Bucci, and Giovanni Russo. 2011. “Accessibility Rozo, Sandra V., and Hernan Winkler. 2019. “Is Informality Good for Busi- and Network Structures in the German Commuting.” Networks and ness? The Impacts of IDP Inflows on Formal Firms.” Policy Research BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References Spatial Economics 11 (4): 621–41.  Working Paper 9035, World Bank, Washington, DC. Riley, Shawn J., Stephen D. DeGloria, and Robert Elliot. 1999. “A Ter- Saez, Emmanuel, Benjamin Schoefer, and David Seim. 2019. “Payroll rain Ruggedness Index That Quantifies Topographic Heterogeneity.” Taxes, Firm Behavior, and Rent Sharing: Evidence From a Young Work- Intermountain Journal of Sciences 5 (1-4): 23–27. ers’ Tax Cut in Sweden.” American Economic Review 109 (5): 1717–63. RIMISP (Centro Latinoamericano para el Desarrollo Rural), IDB (In- Sanchez, Thomas W. 2002. “The Impact of Public Transport on US Met- ter-American Development Bank), and UNDP (United Nations De- ropolitan Wage Inequality.” Urban Studies 39 (3): 423–36.  velopment Programme). 2015. “Fortalecimientos de Capacidades Institucionales de Gobiernos Intermedios para la Gestión y Gober- Sánchez-Cuervo, Ana Maria, and T. Mitchell Aide. 2013. “Consequences nanza Territorial.” RIMISP, IDP, UNDP, La Paz. of the Armed Conflict, Forced Human Displacement, and Land Aban- donment on Forest Cover Change in Colombia: a Multi-Scaled Analy- Robinson, Sherman, Andrea Cattaneo, and Moataz El Said. 2001. “Up- sis.” Ecosystems 16 (6): 1052–70. dating and Estimating a Social Accounting Matrix Using Cross Entro- py Methods.” Economic Systems Research 13 (1): 47–64. Sanchez-Serra, Daniel. 2016. “Functional Urban Areas in Colombia.” OECD Regional Development Working Paper 2016/08, OECD, Paris. Rodríguez-Pose, Andrés. 2018. “The Revenge of the Places That Don’t Matter (and What to do About it)”. LSE Research Online. Schejtman, Alexander, and Julio A. Berdegué. 2004. “Desarrollo Territo- rial Rural.” RIMISP, Santiago. Rodríguez Becerra, Manuel. 2019. Nuestro Planeta, Nuestro Futuro. Bo- gotá: Penguin Random House. Sebastian, Ashwini Rekha, Viviana Maria Eugenia Perego, and Juan Car- los Muñoz Mora. 2020.  “Integrating Venezuelan Migrants in Colom- Rojas, Mariano. 2016. “The Relevance of Happiness: Choosing Between bia’s Agri-Food Sector.” World Bank, Washington, DC. Development Paths in Latin America.” In Handbook of Happiness Re- search in Latin America, edited by Mariano Rojas, 51–62. Dordrecht, Sepúlveda Rico, Carlos E., Denis López Camacho, and Juan Miguel Gal- Netherlands: Springer. lego Acevedo. 2014. Los Límites de la Estratificación: En Busca de Al- ternativas. Bogotá: Editorial Universidad del Rosario. ———. 2019. “Relative Income and Happiness in Latin America: Implica- tions for Inequality Debates.” In The Economics of Happiness, edited Shah, Ritesh, and Liyun Choo. 2020. “Accelerated Education Evidence by Mariano Rojas, 107–26. Cham: Springer Nature Switzerland. Review. Strengthening the Evidence Base for Accelerated Education.” UNHCR, Geneva. Romero, Germán, Andreas Álvarez-Espinosa, Silvia Calderón, and Ale- jandro Ordóñez. 2018. “Redistributive Impacts of a Carbon Tax in Co- Tacoli, Cecilia. 1998. “Rural-Urban Interactions: A Guide to the Litera- lombia: the Link Between Models of Microsimulations and General ture.” Environment and Urbanization 10 (1): 147–66. Equilibrium.” Lecturas de Economía 89: 163–98. ———. 2003. “The Links between Urban and Rural Development.” Envi- Roson, Roberto, and Martina Sartori. 2016. “Estimation of Climate ronment and Urbanization 15 (1): 3–12. Change Damage Functions for 140 Regions in the GTAP 9 Database.” Journal of Global Economic Analysis 1 (2): 78–115. Tax Experts Commission. 2021. “Tax Expenditure Report.” 100 Tolbert, Charles M., and Molly Sizer Killian. 1987. “Labor Market Areas Veiga, José. 2003. Cidades Imaginárias. O Brasil é Menos Urbano do Que for the United States.” Staff Reports 277959, Economic Research Ser- se Calcula. Campinas: Autores Associados. BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References vice, U.S. Department of Agriculture, Washington, DC. Villegas Rodríguez, Ernesto. 2014. “Las Unidades de Planificación y UNDP (United Nations Development Programme). 2011. “Rural Co- Gestión Territorial Como Directriz para la Zonificación Urbana.” El lombia, Reasons for Hope.” In Human Development Report. UNDP, Ágora 14: 551–81. New York. Viollaz, Mariana, and Hernan Winkler. 2020. “Does the Internet Reduce ———. 2015a. “Mainstreaming Climate Change in Colombia: Screening Gender Gaps? The Case of Jordan.” Policy Research Working Paper for Risks and Opportunity.” UNDP, New York. 9183, World Bank, Washington, DC. ———. 2015b. “Objetivos de Desarrollo del Milenio: Informe de 2015.” World Bank. Forthcoming. “Analyzing the Impacts of a Large-Scale Pro- UNDP, New York. gram to Improve Internet Access in Colombia.” World Bank, Washing- ton, DC. World Bank. 2005. World Development Report 2006: Equity UNGRD (Unidad Nacional para la Gestión del Riesgo de Desastres). and Development. Washington, DC: World Bank. 2019. “Evidencia Empírica sobre la Relación Pobreza - Desastres en Colombia.” Bogotá: UNGRD. ———. 2009.  World Development Report 2009: Reshaping Economic Geography. Washington, DC: World Bank. UN-Habitat (United Nations Human Settlements Programme). 2013. Planning and Design for Sustainable Urban Mobility: Global Report on ———. 2012. World Development Report: Gender Equality and Develop- Human Settlements 2013. New York: Routledge. ment. Washington, DC: World Bank. ———. 2017. “Implementing the New Urban Agenda by Strengthening ———. 2013. Improving Accessibility to Transport for People with Limit- Urban-Rural Linkages.” UN-Habitat, Nairobi. ed Mobility (PLM): a Practical Guidance Note. Washington, DC: World Bank. UN-Habitat. (2016). “Evaluación de la eliminación de la estratificación socioeconómica en bogotá y las ciudades colombianas y propuesta ———. 2014. Colombia: Land Policy in Transition. Report 27942-CO. de implementación del nuevo sistema de asignación de subsidios y Washington, DC: World Bank. contribuciones en servicios públicos domiciliarios”. Informe Final. (tech. rep.). UN-Habitat Alcaldia de Bogota. ———. 2015. Indigenous Latin America in the Twenty-First Century: The First Decade. Washington, DC: World Bank. United Nations. (2018). World Urbanization Prospects: The 2018 Revi- sion [key facts] (tech. rep.). United Nations DESA/ Population Division. ———. 2016a. Taking on Inequality: Poverty and Shared Prosperity 2016. https://esa.un.org/unpd/wup/Publications/Files/WUP2018-Key- Washington, DC: World Bank. Facts.pdf ———. 2016b. World Development Report: Digital Dividends. Washing- USGS (U.S. Geological Survey). 1996. GTOPO30. Sioux Falls (SD): Unit- ton, DC: World Bank. ed States Geological Survey Center for Earth Resources Observation and Science (EROS). ———. 2017. “Unbreakable: Building the Resilience of the Poor in the Face of Natural Disaster.” World Bank, Washington, DC. Van der Mensbrugghe, Dominique. 2017. Mitigation Adaptation and Natural Resources Applied General Equilibrium (MANAGE) Model ———. 2018a. “Colombia Policy Notes.” World Bank, Washington, DC. Documentation. 101 ———. 2018b. “Diagnóstico Rápido del Rol del Sistema de Protección ———. 2020c. “Primary Health Care Vital Signs Profile Assessment for Social Colombiano en la Gestión de Riesgo de Desastres.” World Colombia.” World Bank, Washington, DC. BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References Bank, Washington, DC. ———. 2020d. “Propuesta de Reforma al Financiamiento de la Edu- ———. 2018c. “Migration from Venezuela to Colombia: Short and Medi- cación Superior.” Unpublished. um-Term Impact and Response Strategy.” World Bank, Washington, DC. ———. 2020e. World Development Report: Trading for Development in the ———. 2019a.  “Colombia Gender Assessment.” World Bank, Washing- Age of Global Value Chains. Washington, DC: World Bank. ton, DC. ———. 2020f. “Adaptive Social Protection: Building Resilience to Shocks”. ———. 2019b. “Colombia: Learning Poverty Brief.” EduAnalytics, World World Bank, Washington, DC. © World Bank. Bank, Washington, DC. ———. 2021. “Colombia COVID-19 High-Frequency Survey.” World Bank, ———. 2020. “The Human Capital Index 2020 Update : Human Capi- Washington, DC.  tal in the Time of COVID-19”. World Bank, Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/34432 ———. 2021b. “Realizing the Future of Learning: From Learning Poverty License: CC BY 3.0 IGO to Learning for Everyone, Everywhere”. Washington, DC: World Bank. ———. 2020a. “Análisis de los Impactos Ambientales, Económicos, y World Bank, CIAT (Center for Tropical Agriculture), and CATIE (Tropi- Sociales del Impuesto al Carbono y Mecanismo de No-causación.” cal Agricultural Research and Higher Education Center). 2014. “Cli- World Bank, Washington, DC. Unpublished. mate-Smart Agriculture in Colombia. Supplementary Material.” CSA Country Profiles for Latin America Series, World Bank, Washington, DC. ———. 2020b. “Los Impactos de la Crisis del COVID-1 en la Educación y Respuestas de Política en Colombia.” World Bank, Washington, DC. Figures, Tables, 102 BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References and Boxes List of Figures Figure 1.1. Gini Coefficient for Colombia (2008–2020) and Compared to OECD and LAC Countries (circa 2018) 9 Figure 1.20. Effects of the Fiscal Policy Measured by the Reduction in the Gini 15 Figure 1.2. Colombia’s Distance to the Frontier from LAC and OECD Countries across Inequality Dimensions 9 Figure 1.21. Level and Composition of Taxes, average 2014–18 (percent of GDP) 15 Figure 1.3. Differences in Poverty Headcount by Group, percentage points, 2019 9 Figure 1.22. Contribution to Total Collection of Personal Income Taxes, by decile of income (percent) 15 Figure 1.4. Percent of Children Born in Families with Parents in the First Two Quartiles of Figure 1.23. Income Level at which Individuals Pay the PIT (% of median income and % of GDP per capita) 15 Educational Attainment who Manage to Reach the Highest Quartile (relative mobility) 9 Figure 1.24. Pension versus Labor Income: Pseudo-Gini Evolution by Income Source, 2008–2019 15 Figure 1.5. Percent of Children with a Higher Education Level than their Parents (absolute mobility 9 Figure 1.25. People in Pensionable Age by Pension Status and by Per Capita Income Decile, 2019 15 Figure 1.6. Intergenerational Income Persistence 9 Figure 1.26. Aggregate Resources Destined to Pension Benefits by Income Decile, 2017 15 Figure 1.7. Gini coefficient (right) and Share of People Reporting Unfair Income Distribution 9 Figure 1.27. Household Income Changes in 2050 (percent) with Respect to the Baseline, by rural and urban Figure 1.8. Well-Being Score (0-10) and Standard Deviation of Subjective Well-Being 9 income quintiles 16 Figure 1.9. Assessing Equity in Colombia through the Assets-Based Framework 10 Figure 1.28. Wage Effects in 2050 (percent) with Respect to the Baseline, high-impact scenario 16 Figure 1.10. Learning Gaps Across Ethnic Groups 11 Figure 2.1. The Vicious Cycle of Human Capital and Poverty 22 Figure 1.11. Years of Schooling and Learning Gaps across Departments 11 Figure 2.2a. The Impact of Current Investments in Children on Future Income Productivity (HCI) 23 Figure 1.12. Use of Financial Services 12 Figure 2.2b. Human Capital Index for Colombia and Comparators 23 Figure 1.13. Savings Rate by Region 12 Figure 2.3. Human Capital Index vs Poverty Rate by Region 23 Figure 1.14. Labor Market Gaps between the Vulnerable and the Average Worker Figure 2.4. Human Capital: Women versus Men 23 (with respect to comparison group in 2019, percentage points) 13 Figure 2.5. Learning Gaps across Ethnic Groups 24 Figure 1.15. Hourly Earnings Gap between Groups, 2019 (% differences in earnings) 13 Figure 2.6. Years of Schooling and Learning Gaps across Departments 24 Figure 1.16. Share of Firms Citing Labor Regulations as a Major Constraint (%) and Mandatory Social Security Contribution Rate 13 Figure 2.7. Percentage of Children Aged 0–5 from Strata 1, 2, or 3 Enrolled in ICBF Services by Department 24 Figure 1.17. Distribution of Wage Earnings in Colombia 13 Figure 2.8. Percentage of 15-Year-Old Students in Private School By Decile 24 Figure 1.18. Gaps in Education between Municipalities and Regions: Average Test Scores by Municipality 14 Figure 2.9. The Two Main Sources of Spending on Quality 24 Figure 1.19. Unmet Basic Needs in 2005 and 2018: Indigenous and Afro-Descendent Populations 14 Figure 2.10. Health Insurance Coverage in Colombia 25 Figure 2.11. Lost Years of Potential Life by Department, Relative to Bogotá (2015–2020) 25 103 Figure 2.12. Combined Number of Physicians and Nurses per 1,000 Population by Department 25 Figure 4.5b. CIT and PIT Collection Relative to Corporate Profits and Employee Compensation (percent) 44 Figure 2.13. Inequalities in Access to Health Care 25 Figure 4.5c. Corporate Tax Rate in 2020 (percent) 44 BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References Figure 2.14. Health Facility Usage by Wealth Quintile 25 Figure 4.6. General Government Spending, 2017 (percent of GDP) 44 Figure 2.15. Distribution of Beneficiaries of Flagship Social Assistance Programs by Income Decile 26 Figure 4.7. Composition of General Government Spending, 2017 (percent) 44 Figure 2.16. Share of Students at or below the Minimum Reading Level 27 Figure 4.8. Contribution of Direct and Indirect Taxes and Transfers to Reducing Inequality in Colombia 44 Figure 3.1. Labor Market Outcomes in Colombia, 2008–19 33 Figure 4.9. Reduction in the Gini Coefficient from Direct Taxes and Transfers (percentage points) 44 Figure 3.2. Labor Market Gaps across Different Groups, 2019 33 Figure 4.10. Contribution to the Total Collection of Personal Income Taxes, by decile of income (percent) 45 Figure 3.3. Labor market Gaps across Different Groups, changes 2008–19 33 Figure 4.11. Income Level at which Individuals on Average Start Paying the PIT 45 Figure 3.4a. The Quality of Jobs in Terms of Formality and Earnings in LAC, circa 2018 33 Figure 4.12. Effective PIT Tax Rate on Annual and Life Income and Distribution of Wages (percent) 45 Figure 3.4b. Wage vs. Household Income Inequality, 2008–19 33 Figure 4.13. Marginal and Effective PIT Rates (percent) 45 Figure 3.5a. Wage Workers Pre-Pandemic and Loss of Jobs at the Onset of the Pandemic (May 2020) 33 Figure 4.14. VAT Rate across Countries in 2020 (percent) 46 Figure 3.5b. Ratio of Number of Employed in Each Quarter and Each Group, compared to first quarter Figure 4.15. Share of Total Consumption Affected by Standard and Reduced VAT Rates 46 of 2020 (Q1 2020=100) 33 Figure 4.16. Relative and Absolute Progressivity of Consumption 46 Figure 3.6. Labor Productivity (share of high-income countries’ labor productivity) 34 Figure 4.17. VAT, Relative and Absolute Progressivity (percent) 46 Figure 3.7. Labor Productivity, Wages, and Informality by Sector 34 Figure 4.18. Tax Expenditure on VAT Exemptions and Zero or Reduced Rates: Relative and Figure 3.8. Changes in Labor Shares vs. Labor Productivity Levels across Sectors, 2008–19 34 Absolute Progressivity (percent) 46 Figure 3.9. Minimum Wages in Colombia 34 Figure 4.19. Absolute Progressivity of Spending on Selected Goods and Services with Zero VAT Rate or Exempted (percent) 46 Figure 3.10. Social Security Contribution Rate and Labor Regulations as Constraints for Firms 34 Figure 4.20. Proxy C-Efficiency Ratio 46 Figure 3.11. Internet Use Gap by Income and Area vs. GDP per capita, 2019 35 Figure 4.21. Resources Available after Compensating Individuals for the Loss of VAT Exemptions Figure 3.12. The Task Content of Jobs, 2011–19 35 and Reduced Rates, up to a given decile of income 46 Figure 3.13. Gaps in the Share of Workers in Occupations Intensive in the Tasks of the Future 35 Figure 4.22. Relative and Absolute Progressivity of Familias en Acción (percent) 47 Figure 3.14. Decomposition of Hourly Earnings Inequality 35 Figure 4.23. Relative and Absolute Progressivity of Colombia Mayor (percent) 47 Figure 3.15. Percentage of Jobs that Can be Done from Home 36 Figure 4.24. Distribution of Selected Subsidies, by stratum (Percent) 47 Figure 3.16. Jobs Amenable to Working from Home, by workers’ characteristics 36 Figure 4.25: Distribution of Strata, by income decile (percent) 47 Figure 3.17. Fixed Internet Penetration among Municipalities Targeted by the Figure 4.26. Relative and Absolute Progressivity of Electricity Subsidies 47 Proyecto Nacional de Fibra Optica 38 Figure 4.27. Relative and Absolute Progressivity of Gas Subsidies 47 Figure 4.1. Redistribution by Fiscal Policy in Colombia 43 Figure 4.28. Relative and Absolute Progressivity of Water Subsidies 47 Figure 4.2. Redistributive Properties of Fiscal Policy, by country 43 Figure 4.29. Individuals Eligible for Retirement, Receiving a Pension, and Contributing to a Pension 48 Figure 4.3. General Government Tax Revenue, average 2014–18 (percent of GDP) 44 Figure 4.30. Relative and Absolute Progressivity of Pension Spending 48 Figure 4.4. Composition of General Government Tax Revenues, average 2014–18 (percent) 44 Figure 4.31. Internal Rate of Return for Different Wages 48 Figure 4.5a. Corporate Tax Rate in 2020 (percent) 44 Figure 4.32. Distribution of Transfers to Colpensiones, by decile of income 48 104 Figure 4.33. Impact of Direct and Indirect Taxes, Transfers, and Subsidies on Individual Income (percent) 49 Figure 6.6. Household Income Changes in 2050 (percent) with Respect to the Baseline, by rural and urban income quintiles 71 Figure 4.34. Distribution of Selected Taxes, Transfers, Subsidies, and Spending by Income BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References Decile (percent of GDP) 49 Figure 6.7. Wage Effects in 2050 (percent) with Respect to the Baseline, High-Impact Scenario 71 Figure 4.35. Concentration Curve 49 Figure A1.1 Subjective Well-Being by Level of Economic Development at the Country Level 79 Figure 4.36. Effective Tax Rates under Different Option of Alternative Tax Schedule, and Distribution Figure A1.2. Subjective Well-Being (Cantril Ladder, 0-10) in Colombia across Regions 79 of Income (percent) 50 Figure A4.1. The New Combination of Functional Subregions Map 85 Figure 4.37. Estimated PIT Collection under Current PIT Schedule, and Different Reform Option (percent of GDP) 50 Figure A5.1. MANAGE Production Structure 86 Figure 5.1. Inequality among Colombia’s Regions Compared to Other Countries 55 Figure 5.2. Regions in Colombia Used in this Chapter 56 Figure 5.3. Extremes in Socioeconomic Vulnerability 57 List of Tables Figure 5.4. Unmet Basic Needs of the Indigenous and Afro-Descendant Populations 57 Table ES.1. Policy Options for a More Equitable Society in Colombia 7 Figure 5.5. Vulnerability across Dimensions 57 Table 1.1. Subnational Human Capital Index and Components for Colombia 11 Figure 5.6. Functional Territories in Colombia 58 Table 2.1. Subnational Human Capital Index and Components for Colombia 23 Figure 5.7. Intra-Urban Inequality in Colombia 59 Table A3.1. Task Typology 81 Figure 5.8. Contributors to Intra-Urban Inequality 59 Table A3.2. Decomposition of the Variance of the Logarithm of Hourly Earnings 81 Figure 5.9. Intra-Urban Inequality in Larger versus Smaller Cities 59 Table A3.3. Decomposition of Changes in the Gini of Hourly Earnings 81 Figure 5.10. Intra-Urban Concentration of Socioeconomic Vulnerability in Colombian Cities 59 Table A4.1. Vulnerability across Regions 82 Figure 5.11. Access to Schools in Colombian Cities 60 Table A4.2. Amenities at the Block Level 82 Figure 5.12. Quality Gaps in Education between Municipalities and Regions in Colombia 60 Table A4.3. Descriptive Statistics of Changes in Poverty, Inequality, and Expenditures for Territories of Different Core Size 83 Figure 5.13. Access to Health and Sports Facilities 60 Table A4.4. Ordinary Least Squares Model for Monetary Poverty 83 Figure 5.14. Access to Public Services for Small Municipalities 60 Table A4.5. Ordinary Least Squares Model for GINI Change 83 Figure 5.15. Cities’ Center-Periphery Patterns in Terms of Access to Facilities 60 Table A4.6. Ordinary Least Squares Model for Average Household Expenditure Change 83 Figure 5.16. Spread of Health Facilities in Medellin and Cali 60 Table A4.7. Three-Stage Least Squares Model for Monetary Poverty and Net Effects 83 Figure 5.17. Comparison of Current Stratification Categories and the Adjusted Multidimensional Poverty Index, 2018 61 Table A4.8. Three-Stage Ordinary Least Squares Model for Multidimensional Poverty Measure and Net Effects 83 Figure 5.18. Mismatch between Current Stratification System and Socioeconomic Clusters, 2018 61 Table A4.9 Urban-Rural Channels 84 Figure 6.1. Sectoral Productivity for Each Sector by Climate Change Scenario 69 Table A4.10. Three-Stage Ordinary Least Squares Model for Monetary Poverty with Channels 84 Figure 6.2. Composite Real GDP Effects in 2050 (percent, with respect to the baseline) 70 Table A4.11. Three-Stage Ordinary Least Squares Model for Multidimensional Poverty Measure Figure 6.3. Real GDP Changes Due to Sector Productivity Changes 70 with Channels 84 Figure 6.4. Real GDP Changes Due to Labor Productivity 70 Table A4.12. Three-Stage Ordinary Least Squares Model for Monetary Poverty Change for New Figure 6.5. Real GDP Changes Due to Changes in Hydroelectric Generation 70 Combination of Functional Subregions 85 105 Table A4.13. Net Effects of Monetary Poverty Change 85 Table A4.14. Three-Stage Ordinary Least Squares Model for Multidimensional Poverty Measure BUI L DI N G AN EQ UI TABL E SO CI ETY I N CO LO MBI A | References Change for New Combination of Functional Subregions 85 Table A4.15 Net Effects of Multidimensional Poverty Measure Change 85 Table A4.16. Three-Stage Ordinary Least Squares Model for Monetary Poverty Change with Channels 85 Table A4.17 Three-Stage Ordinary Least Squares Multidimensional Poverty Measure Change with Channels 85 Table A4.18. Net Effects of Monetary Poverty Change and Multidimensional Poverty Measure Change 85 List of Boxes Box 2.1. The Colombian Health Care System 25 Box 3.1. The World Bank’s Jobs Diagnostic Report for Colombia 33 Box 4.1. The Personal Income Tax in Colombia 45 Box 4.2. Main Cash Transfer Programs and Subsidies in Colombia 47 Box 5.1. Framework for Prioritizing Policy Interventions: Institutions that Unify, Infrastructure that Connects, and Interventions that Target 55 Box 5.2. The Impact of Armed Conflict on Territorial Inequalities in Colombia 57 Box 5.3. The Exclusion of Afro-Colombians 57