Report No. 24524-CO Colombia Poverty Report (In Two Volumes) Volume 1: Main Report November 1, 2002 Colombia Country Management Unit PREM Sector Management Unit Latin America and the Caribbean Region Document of the World Bank CURRENCY EQUHVALENTS (as of October 211, 2002) Currency Unit = Peso (Coil$) Col$] = US$ 0.000351 US$1 = Col$ 2876.400 WEIGHTS AND MEASURES Metric System FISCAL YEAR January 1 to December 31 ABBREVIATIONS AND ACRONYMS ARS Administradoras de Régimen Subsidiado CAIP Centros de Atención Integral al Preescolar CASEN Caracterización Socioeconómica Nacional CDF Comprehensive Development Framework CODHES Consultoría para los Derechos Humanos y el Desplazamiento DALYS Disability-Adjusted Life Years DHS Demographic and Health Survey DNP Departamento Nacional de Planeación ECV Encuesta de Calidad de Vida ENH Encuesta Nacional de Hogares FGT Foster-Greer-Thorbecke index GDP Gross Domestic Product GNP Gross National Product HCB Hogares Comunitarios de Bienestar IC1BF Instituto Colombiano de Bienestar Familiar ICETEX Instituto Colombiano de Crédito Educativo y Estudios Técnicos en el Exterior ICFES Instituto Colombiano para el Fomento de la Educación Superior IDP Internally Displaced Populations IDB Inter-American Development Bank INURBE Instituto Nacional de Vivienda de Interés Social y Reforma Urbana ISS Instituto de Seguros Sociales LSMS Living Standards Measurements Study NHS National Household Surveys PACES Programa de Ampliación de la Cobertura de Educación Secundaria POS Plan Obligatorio de Salud POSS Plan Obligatorio de Salud Subsidiado PPP Purchasing Power Parity PSE Public Social Expenditure SCD SISBEN classification document SENA Servicio Nacional de Aprendizaje SISBEN Sistema de Selección de Beneficiarios UNDCP United Nations Development Program for Drug Control and Crime Prevention UNDP United Nations Development Program UPAC Unidad de Poder Adquisitivo Constante VAT Value Added Tax UVR Unidad de Valor Real Vice President: David de Ferranti Director: Olivier Lafourcade Lead Economist: Marcelo Giugale Sector Manager: Norman Hicks Task Manager: Carlos Eduardo Vélez TABLE OF CONTENTS K ey F inding s ................................................................................................................................... . Executive Summary....................................................................................................................iii INTRODUCTION.........................................................................................................*********************************1 Background: Changes in the economic environment during the last two decades..............................2 The macroeconomic and sectoral environment in the 1980s and early 1990s....... ...... .... ...............3 Adverse macroenvironment in the late 1990s: Recession and risk deterioration.. .... ..................... 4 Supply shocks, structural reform, and employment generaton................. .......................................5 Major changes in the socio-demographic structure of the working population........................ ... ....6 Concentration of land and rural credit...... ..... ......... ...... ...................................................... .........7 Displaced population and armed conflict in the rural areas................. ..... .. .. . .............. .. ..........7 Illicit crops and rural development. .. .............. .. .......... ..................... ........... . ............................ 7 Chapter I. How Have the Poor Fared During the Last Two Decades and the Recent Recession?...................................................................................................9 1. POVERTY, ECONOMIC WELFARE, AND INCOME DISTRIBUTION ...................................10 1.1 Poverty trends showed substantial long-term progress with a recent setback . ................10 Urban poverty- In line with the national trend.......................................... ..................................... II Rural Poverty: Clear progress up to 1995, but much higher rates in urban areas................ ......... 11 1.2 Income per capita nearly doubles in two decades but falls in the late 1990s....................12 1.3 Rising inequality eroded welfare gains: The Sen welfare index.......................................13 Urban welfare was hurt by inequality deterioration, mostly in the 1990s..... ... ...........................15 Rural Sen Welfare Index: Rising inequality eroded welfare gains even farther than in the urban case......... . ..... ......................... ... ..... ........................... ................... .. .. ......... ..... ................... 17 1.4 Beyond average income: Assessing welfare gains from 1978 to 1999............................18 Urban welfare- Unambiguous improvement up to 1995 and unambiguous deterioration during the recession ................................ ..... .. ........... ... ... . ................ .......... ............................. ........... . 18 Rural welfare. less severe recession losses.............. ....... .... .................. .. ... .............................20 1.5 Colombia is more capable of tackling poverty today than it was two decades ago .........22 2. SOCIAL SERVICES AND SOCIAL OUTCOMES: COVERAGE AND EQUITY.......................23 2.1 Education: Increasing trend with pro-cyclical fluctuation in urban areas.........................23 Urban Schooling: Increasing trend with pro-cyclical fluctuations. ....... .. ... . .. ....... .............. ... 23 R ural Schoolm g .................. .............................. ....... ............ ...... ........................................... .... 24 Increasing frustration and inequity at the entrance ofpost-secondary education. ......................... 26 2.2 Child labor: pro-cyclical with a decreasing trend.............................................................29 2.3 Access to basic infrastructure: progressive gains .............................................................30 2.4 Child malnutrition and infant mortality: continuous improvement ..................................31 3. THE BURDEN OF VIOLENCE AND CRIME.........................................................................32 3.1 Violence: its continuous aggravation since the 1970s is mostly associated with the illegal drug trade ..........................................................................................................................32 3.2 Life expectancy: persistent improvements with gender bias associated with violence ....34 3.3 Inequality and violence: Who bears the burden of crime?................................................34 3.4 Armed conflict in rural areas, displaced population, and vulnerable groups....................36 Chapter II. Vulnerability: Are the Poor Changing Faces?.......................................3 4. CHARACTERISTICS OF VULNERABLE GROUPS...............................................................39 4.1 Urban Colombia: The faces of the urban poor have not changed, but they are becoming more apparent and polarized.............................................................................................39 Urban poverty profile and the marginal effect of key household characteristics.... ........................42 Skill endow m ent.................... ................................................................... ......................................42 D em ograp hics..... ...... .......................................................... .................. ............. ........ ...... .......43 Labor M arket.......................... .... ....................... . ................................. .....................................43 Homeownership.................................................. ...... .... ..... .... . ........ ....................... ... ............. 45 Idiosyncratic shocks ................. ................... . .......................................... ......... ............ .. ........... 45 C onclusion....................... ........................................................................ ......... ............................45 4.2 Rural Colombia: Except for the displaced population, vulnerability is similar to that of urban areas........................................................................................................................4 6 Chapter III. Driving Forces Behind Poverty Changes ................................................49 5. LABOR MARKETS AND POVERTY DYNAMICS IN BooMs AND RECESSION....................50 5.1 The decomposition of aggregate poverty changes in terms of growth and inequality......50 U rban ........................................................................ ..................................................................50 R ural............................................................................................................................................5 1 5.2 The evolution of income per capita in terms of skill endowments, wages, dependency ratios, and employment (1978-99)....................................................................................52 The dynamics of poverty in terms of basic income-generating factors... ....... ................................52 5.3 The dynamics of urban poverty in the late 1990s recession: lower wages or jobs lost?...53 Asymmetric dynamics of wages and employment by occupational choice and gender....................54 6. EXPLAINING INCREASING INEQUALITY DURING THE 1990s..........................................57 6.1 Sources of income inequality increase..............................................................................57 6.2 The reversal of inequality trends in urban Colombia: a combination of persistent and fluctuating forces..............................................................................................................60 Labor markets and the determinants of household income... ............. .. ...... ..................................60 Understanding income distribution trends using factor decompositions.........................................63 Chapter IV. Public Social Policy and Sector Priorities for the Poor.........................67 7. PUBLIC SOCIAL EXPENDITURE: INCIDENCE AND SECTOR PRIORITIES IN THE 1990s...68 7.1 The 1990s were a period of substantial PSE growth ........................................................68 7.2 Distributional impact: the poor benefit substantially of overall PSE................................70 7.3 Coverage dynamics of social programs by income groups...............................................71 E duca on..................... ........ ....... ............................................................ .....................................72 H ealthca re .................................................................. ...... ..... ......... ........... ..... ........................... 72 P ublic U tl ities........................ ........... .... .... ....................... ................. ....................................... 75 8. SUBSIDIZED HEALTH INSURANCE AND PROXY MEANS TARGETING .............................77 8.1 Main characteristics of contnbutory and subsidized regimes..........................................77 8.2 Higher and more progressive coverage in the 1990s........................................................79 8.3 Benefits and protection for the poor.................................................................................80 8.4 Targeting and Efficiency problems...................................................................................81 Chapter V. Conclusion and Policy Lessons................................................................83 9. POLICY LESSONS AND RECOMMENDATIONS FOR REDUCING POVERTY .......................83 Policy priorities: Security, growth, education, health, and infrastructure...........................................84 More security for all Colombans... ..... ........... .. . .......... .. ..... .. 84 Recover positive growth of output per capita and protect the unemployed.................... . .. . .........85 Three demands for Education: preschool education, childcare, and technical training and expansion of credit for tertiary education.... . ................................ ............. ............. .. ...... ........ 86 Infrastructure. Sewerage and Housng.......... .................... ............... ...... .. .. .. .. .......... .............. 88 Health: Transform higher insurance coverage into more treatment.. . ......... .... ...... . . .... . . .... 89 Under increasing economic insecurity, safety-net programs become a more valuable policy instrum ents ......................................... ..... .... ........... . .... ..... .............. .. .. ....... ..... .............. . 89 REFERENCES ............................................................................................................................................ 91 APPENDIX A.1: INCOME AND POVERTY LINE METHODOLOGY...........................................................101 A.1.1 Data sources and income adjustments.......................... ......................................... .......101 A.1.2 Poverty lines and price indices by region.......................................................................101 A.1.3 Intertemporal Urban Welfare Comparisons............................................................... ....102 APPENDIX A.2 CROSS REGIONAL WELFARE COMPARISONS IN RURAL AND URBAN COLOMMIA.........104 A.2.1 Cross-urban comparisons: partial convergence and improvement in left behind cities.. 104 Strong convergence in poverty counts despite persistent differences icome.. .. .... ... ................ 105 Incom e p er capita ................ .... ..... ........................ ...................... ......................... ......... .... ....106 Inequal ty.... ... .......................... ... ... ........... .. .... ...................... . .. .. ... ..............................107 Extreme poverty dynamics, growth, and income inequality . ............ ................... .. ...... . .. . .. 109 A.2.2 Rural Regional comparisons: Central and Atlantic regions win, Pacific and C entral lose ................................ ................................................................. 10 Social indicators reveal deviating paths in regional development favors the Atlantic Region......110 Strong heterogeneity in rural poverty- Oriental and Atlantic regions win, Central and Pacific regions lose.................. ... ... ...... .. .... . .. ............. ... .. ... ..................... .. . ... . .. ... .... .. Income per capita: Divergent and crossing paths in rural regional development.. .... . ........... ...112 Inequal ty...... ....... ............. ........... ..... .. .................. ..... ... .... ............ ..... .. ................ ..... 115 Extreme poverty dynamics, growth, and income inequaty.. .... ... ... .. ....... ..... .... ... ........... ....115 List of Tables Table lA Colombia at a glance 1980-1999 .......................................................... .......................................2 T able IB : C olom bia's volatility..................................................................................................................... 5 Table 2: Poverty indicators, Colombia 1978-1999 ...................................................................................12 Table 3: Income inequality indicators, Urban Colombia 1978-1999 ............................. .... ................. ..14 Table 4: Social indicators. Urban Colombia' 1978-1999 ............ ............................................................24 Table 5: Social indicators, Rural Colombia 1978-1999 ...........................................................................25 T able 6: N ational infant m ortality ..................................................................................................... ...... 31 Table 7: Poverty count for different subgroups of the population, Urban Colombia 1978-1999 .............40 Table 8: Income sources and needs: the poor versus the non-poor. Urban Colombia, 1978-1999 ...........41 Table 9: Marginal effects of selected variables on the probability of being poor: Urban Colombia ........44 Table 10: Decomposition of poverty changes, Urban Colombia ...................................... .........................51 Table 11: Decomposition of income dynamics in terms of skill endowments, adult to family size ratio, employment rate, and wages. Urban Colombia 1978-1999........................................................52 Table 12: Decomposition of income dynamics during the recession. Urban Colombia 1995-1999...........54 Table 13: Changes in wages and employment by skill level. Urban Colombia, 1995-1999 .......................55 Table 14: Inequality decomposition by household characteristics, Urban Colombia 1978-1999 ...............58 Table 15: Shorrocks' inequality decomposition by factor components of household income, Urban C olom bia 1978-1999 .................................................................................................................59 Table 16: Decomposition Income Distribution Changes for Households and Individual Workers. Changes in the Gini coefficient. Urban Colombia (1978-1988, 1988-1995) ..............................64 Table 17: Change in coverage rates and growth in coverage rates from 1992 to 1997 ..............................73 Table 18: Coverage short falls and targeting in selected social programs ..................................................75 Table 19: Relative access gap for alternative coverage targets and priorities for social expenditure .........76 List of Figures Figure 1: Colombia's inequality in the international context, 1999 .............................................................14 Figure 2: Average income per capita and the Sen welfare index, Urban Colombia, 1978,1988,1995,1999 17 Figure 3: Welfare, rural Colombia, 1978,1988,1995,1999 ...........................................................................17 Figure 4: Cumulative income distribution, Urban Colombia 1978, 1988, 1995 ..........................................18 Figure 5: Cumulative income distribution, Urban Colombia 1988, 1995, 1999 ..........................................19 Figure 6: Income changes by percentile. Urban Colombia, 1978-88, 1988-95, 1995-99 .............................19 Figure 7: Cumulative income functions, rural Colombia 1978, 1988, 1995, and 1999 ................................21 Figure 8: Income changes by percentile. rural Colombia 1988-95, 1995-99 ...............................................21 Figure 9: Share of total household income required to close poverty and extreme poverty gaps ................22 Figure 10: Average education and inequality by cohort. Colombia, 1925-1975 .........................................27 Figure 11: Efficiency and equity of educational attainment: the post-secondary bottleneck. Colombia, 1988 and 1999, 20-25 age cohort ...............................................................................................27 Figure 12: H om icide rate: 1962-1998 ......................................................................................................... 33 Figure 13: Life expectancy by gender, Colombia 1950-2000 .....................................................................34 Figure 14: Homicide rates by age and gerider, Colombia 1999 ..................................................................34 Figure 15: Socioeconomic status and propensity to hire private guards .....................................................35 Figure 16: Change in female labor force share by occupational choice and skill urban Colombia, 1995-1999 .................................................................................................................................55 Figure 17: The distribution of pension and non-pension earnings by household income percentile: Concentration curves, urban Colombia, 1999 ............................................................................60 Figure 18: Labor earnings by skill level. Urban Colombia, 1999 ...............................................................61 Figure 19: Probability of being employed or a wage-earner according to individual level of education, number of children, and human capital of other members; household heads, spouses, and other. Urban Colom bia, 1978, 1988 and 1995 ......................................................................................62 Figure 20: Long-term trends in the aggregate level of public expenditure .................................................69 Figure 21: Coverage rates for health insurance and treatment by decile, 1997............................................73 Figure 22: SCD, affiliation, and healthcare co-payments in the subsidized regime ....................................79 Figure 23: The poverty rate under high- and low-growth scenarios, 2001-201, urban Colombia ...............86 List of Boxes Box 1: Overestimated poverty by using income measures...........................................................................16 Box 2: Poverty comparisons across Colombian cities, 1978-1995..............................................................28 Box 3: Development divergence and poverty across rural regions in Colombia, 1978-1999......................29 Box 4: The reversal of two decades of progressive homeownership ..........................................................47 Box 5: Targeting mechanisms for the main social programs ......................................................................71 Box 6: Political feasibility: Colombian laymen's policy priorities are consistent with the findings of this report...........................................................................................................................................87 Box 7: Remaining puzzles and topics for further analysis...........................................................................90 Acknowledgments This report team comprised Carlos Eduardo V61ez (Team Leader), Vivien Foster, Mauricio Santamarfa, Natalia Milldn, and B6n6dicte de la Bri6re from the World Bank; Giota Panopoulou, from Sussex University; and Alejandro Gaviria from Fedesarrollo, Bogotd. Research assistance was provided by Juanita Riatio and Taizo Takeno, and document preparation assistance was provided by Tania Gomez and Anne Pillay. This report benefited from comments by Ana Maria Arriagada, Adolfo Brizi, Alberto Chueca, Maria Correia, Maria Luisa Escobar, Vicente Fretes, Jorge Garcia-Garcia, Marcelo Giugale, Elsie Garfield, Olivier Lafourcade, Norman Hicks, Vicente Paqueo, Guillermo Perry, Laura Rawlings and Eduardo V61ez-Bustillo. The teams would like to thank peer reviewers Emmanuel Jimenez and Peter Fallon for their very valuable suggestions that helped to improve this report. Preliminary versions of some papers were discussed in seminars at Banco de la Repfiblica, Departamento Nacional de Planeaci6n, Fedesarrollo, and the Ministerio de Salud in Colombia. Special acknowledgements are due to a selected group of Colombian academics and policy makers that provided useful comments to earlier drafts of this report: Juan Carlos Echeverry, Juan Luis Londofio, Hugo Lopez, Mauricio Perfetti, Mauricio Santamarfa, Alfredo Sarmiento and Fernando Urrea. Finally, we want to acknowledge the comments, questions and suggestions to earlier versions of this report given by Gary Fields and the panellists and participants in the dissemination workshops of Bogotd, Cali and Medellfn in May 2002. Our gratitude to the three leading public policy think tanks that hosted those workshops: Fedesarrollo, Planeta-Valle and Pro-Antioquia.. Country Director was Olivier Lafourcade, Lead Economist was Marcelo Giugale, Sector Manager was Norman Hicks, and Task Manager was Carlos Eduardo V61ez. Special acknowledgement regarding background paper 3 in volume 2 "The Reversal of Inequality Trends in Colombia, 1978-1995: A Combination of Persistent and Flictuating Forces" by Carlos Eduardo V61ez (World Bank), Cesar Bouillon (IDB), Jose Leibovich (Federaci6n Nacional de Cafeteros, Bogota), Jairo Ndfiez (Universidad de los Andes, Bogotd) and Adriana Kugler (U. Pompeu Fabra) - Chapter 3, Volume II-: This paper is part of the cross-country study, The Microeconomics of Income Distribution Dynamics in East Asia and Latin America, a joint IDB/World Bank research project directed by Frangois Bourguignon, Francisco Ferreira and Nora Lustig. The IDB started this initiative and provided research funding for the paper on urban Colombia. Concurrently, the World Bank (DEC) provided the financial support for the complementary rural background paper. The World Bank-LAC's Colombia Poverty Report Task supported the final stages of the urban income inequality dynamics paper and the merging of the urban and rural papers into a final product. Key Findings Colombia Poverty Report, 2002 How HAVE THE POOR FARED IN THE LAST TWO DECADES AND RECENT RECESSION? I. Poverty trends show substantial long-term progress but a recent severe setback. Poverty declined during the 1980s and during the first half of the 1990s. However, after a reduction of 20 percentage points on national poverty rates from 1978 to 1995, the trend reversed itself, and, in 1999, poverty returned to the level registered in 1988. 2. Social indicators show persistent long-term improvements. A variety of measures indicate progress, including completion rates for primary and secondary schooling, coverage rates for health insurance and basic infrastructure -water, sewerage, electricity, and telephones- as well as child labor, infant mortality and life expectancy. 3. Crime is a burden for all Colombians: homicide and domestic violence are a greater problem for the poor, property crime centers more on the non-poor. In particular, uneducated women and spouses of uneducated men bear a disproportionate share of domestic violence. By contrast, the non-poor bear a disproportionate share of the burden of property crime, extortion, and kidnapping. 4. Economic and social development gains have been resilient to growing violence and insecurity. Despite two decades of persistent security deterioration, Colombia experienced unambiguous improvement in social and economic indicators since the late 1970s. 5. Vulnerability: characteristics of the poor are persistent but the recession increased their numbers The faces of the poor are typically children of all ages, households with unemployed heads or young low-to-middle-skilled heads, non-homeowners and recent migrants -presumably displaced populations. These groups are clearly worse off than pensioners, the college educated, the elderly, and non-recent urban migrants. 6. Double benefit of economic growth: is instrumental for poverty reduction and has increased the capacity to finance pro-poor policies. During the 1980s and early 1990s, economic growth explained, almost totally, the 22-percentage point reduction in urban poverty. Moreover, thanks to the economic growth experienced up to 1995, Colombia today has a great deal more resources -a fiscal base- with which to tackle poverty. The reallocation of moderate amounts of public expenditure would have major impact on the welfare of the extremely poor. 7. Rising inequality partly eroded potential welfare gains. The expected difference in income between any two Colombians chosen at random has been increasing over time. The latter is responsible for a welfare loss - for the worse-off individual in that pair - of 18 percent between 1978 and 1995, and an additional 5 percent loss in the late 1990s. Unless special attention is paid to post-secondary education expansion, skill-wage differentials will remain too high and will keep inequality rising. GOVERNMENT ACTIONS AND PRIORITIES FOR THE POOR 8. Social services in the 1990s: huge public expenditure effort benefits the poor, but results across sectors are mixed. During the last decade, Colombia has made a sustained effort to raise public social expenditure (PSE) to 15 percent of GDP. With the exception of excessive, inequitable and fiscally unsustainable pension subsidies, public social expenditure in basic education, basic infrastructure, as well as in health and childcare, is found to be pro-poor. 9. High positive economic growth during the next decade is imperative to reverse the poverty increase brought about by the recent recession. Colombia needs to recover sustained economic growth in the magnitude of 4 percent per year until at least 2010, in order to reduce poverty back to its 1995 level. i 10. Social program priorities for the poor: childcare, health, and sewerage. Despite an increase in social program coverage during the 1990s, coverage in childcare, sewerage and health treatment are too low for the poor, vis-A-vis populations with higher incomes. 11. One social policy instrument, education, has the single most important role on social welfare through multiple poverty determinants: human capital endowments, employment rates, dependency ratios and even relative wages by skill level. Not only is the direct impact of education and technical training on income as they raise both wage earnings and the chances of being employed -particularly for women-. Education---especially of women----helps reduce fertility, thus reducing dependency rations and increasing per capita income. Education also protects against domestic violence. Moreover, having a more skilled labor force has helped to raise the relative wages of the less skilled, by reducing their availability in the labor market. 12. Safety-net programs become more valuable policy instruments in an environment of increasing economic insecurity. Contrary to its traditional stability, the volatility of the Colombian economy nearly doubled during the 1990s. Under those circumstances the "archipelago" of traditional safety net programs was both insufficient and fragile. A more integrated social protection policy to enhance risk management options should be significantly beneficial. ii Colombia Poverty Report Executive Summary Up to 1996, Colombia enjoyed high, sustained, and stable growth. Since then, Colombia has undergone its most severe economic crisis since the 1930s. After the relative slowdown of the early 1980s, adjustment policies and the coffee bonanza of 1986 helped restore internal and external balances. In the late 1980s, exports diversified away from coffee, unemployment fell, and growth was robust up until 1996. However, during the early 1990s, total public spending started to grow at an unsustainable pace. By the second half of the 1990s, the macroeconomic performance of Colombia had deteriorated significantly, with growth rates plummeting-the economy officially entered into recession-and unemployment escalating. This situation emerged during a decade in which public policy reforms were abundant and macroshocks were significant. In addition to the constitutional reform of 1991, an activist public policy enforced structural reforms in key areas such as sub-national public finance, health, pensions, labor markets, central bank framework, trade, and capital flows. Simultaneously, risk indicators worsened, mostly because of increasing economic volatility, financial sector fragility, and persistently high levels of crime and violence that stretched the judiciary system beyond its limits. Moreover, de-facto authoritarian regimes enforced by paramilitaries and local guerrilla warlords gained control in some isolated rural areas. The detrimental effects of the recent economic slowdown on the welfare of the average Colombian can easily be anticipated. However, it is not clear how large the welfare losses are, and, in particular, how much have they affected the poor and other vulnerable groups? What types of vulnerabilities have emerged in the new economic environment? How appropriate was the government's policy in terms of social investment and social protection? Which are the sectors that need urgent improvement? This report appraises the impact of economic development, or lack thereof, on the welfare of the Colombian population - the poor in particular- during the last two decades. At the same time, the report identifies priority areas for public policy action vis-A-vis the most vulnerable groups. Welfare assessment covers three key areas: income, access to social services, and personal security. Moreover, the study compares welfare indicators between urban and rural areas and across other regional partitions. In summary, this report addresses two main questions: How are the poor doing and why? and What is the government doing for the poor? iii Following are the main findings and policy recommendations of this report: HOW ARE THE POOR DOING AND WHY? Poverty trends: substantial long-term progress with a recent setback First, poverty trends in Colombia show substantial long-term progress, albeit with a recent setback. As measured by the national poverty line, 64 percent of the population lived in poverty in 1999, and 23 percent were in extreme poverty. Poverty declined during the 1980s and during the first half of the 1990s. However, after a total reduction of 20 percentage points from 1978 to 1995, the downward trend reversed itself, and, in 1999, poverty returned to the level registered in 1988. Meanwhile, extreme poverty declined much faster than the poverty rate between 1978 and 1995, failing by more than 50 percent (from 45 to 21 percent). Similarly, the recession did aggravate the extreme poverty rate, but the level it reached was still well below that of 1988. The rural population still faces more adverse economic conditions, particularly in the extreme poverty count, which is nearly three times larger than that of the urban dwellers. However, unlike in urban areas, extreme rural poverty declined much faster up until 1995, and, moreover, the economic recession hit the urban extreme poor more severely. It is important to note that consumption- based measures suggest that income-based poverty indices overestimate urban and rural poverty by large margins - between 15 and 20 percentage points. Social indicators: persistent long-term improvements Second, during the last two decades, several indicators showed a clear and positive social development trend. Although school enrollment rates progressed somewhat slowly and suffered a reversal during the recent economic downturn, completion rates for primary and secondary schooling improved substantially, reaching 90 and 59 percent, respectively, for teenagers and young adults. In urban areas, the average educational attainment increased by 2.7 years and illiteracy rates were nearly halved. Compared to improvements in urban areas, improvements in rural education are relatively larger when judged by the dynamics of enrollment rates for all school ages -especially those for 18- to 22-year-olds. Moreover, rural enrollment rates seem more resistant to the recession period than do urban rates. At the national level, child labor followed a decreasing trend; its decline was particularly strong during the recession, , thus following a pro- cyclical trend, as it has in other Latin American countries. Growth in basic infrastructure coverage, i.e., water, sewerage, electricity, and telephone, results from the catch-up by cities least covered in 1978 as well as from progressive extension to the poorer population. However, sewerage and telephone connections in urban areas still lag behind electricity and water, and the missing percentages mask thousands of individuals, concentrated in regional pockets of poverty, where these basic needs are not yet met. Finally, during the last four decades, life expectancy improved by approximately one year every two calendar years. Despite this, regional differences are considerable, and most of the gains benefited females; rising homicide rates for young males partly offset the impact of health gains on male life expectancy. Burden of Crime: homicide and domestic violence for the poor, property crime for the non-poor Third, violence and the continual increase in crime since the 1970s have eroded the welfare of all Colombians. Poor households bear the brunt of homicide and domestic violence, and young Colombian males face an extremely high risk - even by Latin American standards - of being murdered. In particular, uneducated women and spouses of uneducated men bear a iv disproportionate share of domestic violence. In contrast, the better off bear a disproportionate share of the burden of property crime, extortion, and kidnapping. They are more likely to be victimized, modify their behavior because of fear of crime, feel unsafe, and invest in crime avoidance. After tripling from 1970 to 1991, homicide rates decreased moderately in the 1990s, while extortion, kidnapping, car theft and armed robbery rates kept growing. The violence trend is mostly associated with the illegal drug trade and the presence of illegal armed groups, and Colombia is now the world's largest producer of cocaine and has recently become a significant supplier of opium poppy and heroin. Moreover, the social costs of violence are amplified by the demand of public resources to attend to the victims and prevent further security deterioration. The armed conflict in rural areas has also created an increasing trend in the numbers of internally displaced population -at least 1 million- further increasing the pressure on public resources. Public expenditure on justice and security has more than doubled its share of gross domestic product (GDP) during the last decade, and private expenditure on security seems to be increasing even more rapidly. In addition to its direct burden, violence disrupts the market economy and imposes a considerable psychological cost on those populations who are not directly victimized. The disproportionate concentration of property crime victimization among the better-off members of the population has serious, undesirable economic consequences such as lower levels of investments and growth, and higher migration rates among the most educated. The concentration of crime victimization among small business owners has a perverse effect on economic efficiency, reducing investment and employment in poor urban communities. For its part, the concentration of homicide and domestic violence among the worse-off, besides its obvious effects on the well-being of the victims, imposes a psychological and economic burden on their relatives and increases their chances of enduring pathological and dysfunctional behaviors, especially among children. This ultimately erodes socioeconomic mobility and contributes to the perpetuation of poverty. Thus, as long as poor, middle-class, and rich Colombians continue feeling defenseless outside their homes, and poor Colombian women also experience. similar feelings inside theirs, the possibility of recovering economic prosperity and equity seems more elusive. Economic and social development gains have been resilient to growing insecurity Fourth, three main welfare fundamentals: 1. household income, 2. social development, and 3. personal security, do not evolve in complete harmony. Sporadically, they diverge and turn out to be in partial contradiction. Despite two decades of persistent security deterioration, Colombia experienced unambiguous improvement in social and economic indicators. Moreover, social investment in human resources showed some healthy robustness relative to recent market instability. Only recently, probably as a result of the economic recession, school enrollment has been declining. Vulnerability: with the exception of IDP, characteristics of the poor are persistent. Nevertheless the unemployment of household heads is becoming increasingly important Fifth, during the last two decades, the populations most vulnerable to poverty have maintained a relatively constant set of characteristics; albeit, those characteristics have become more polarized over time. Poverty, here defined in the monetary sense, is generally associated with at least one of the following states: low skills, lower employment rates, higher dependency ratios (number of children per adult), and lower wages. The first three predominate among poor households in Colombia. During the 1990s, the likelihood of escaping poverty was increasingly dependent upon: having less children, having more working-age members with postsecondary education, and having access to employment for members other than the household head. Contrary to what could be expected, the presence of the elder reduces the risk of poverty. Homeowners and middle- to high-age-headed households are doing better. Poverty may also arrive in less typical v circumstances, that is from low-frequency events that bring devastating welfare effects: for example, when the head of the household loses employment, or a disabled member is present in the household, or the household has recently migrated, the latter presumably because of displaced populations in the late 1990s-.Population displaced by the armed conflict constitutes one of the most vulnerable groups. In addition to the burden of displacement, their demographic characteristics exacerbate their vulnerability and deteriorate their living conditions In contrast to the urban situation, rural poverty remains much more severe and represents a higher share of total poverty. Moreover, the fact that household characteristics alone appear insufficient to explain the higher level of rural poverty relative to urban, underscores the importance of other non- demographic determinants of rural productivity, namely, infrastructure, natural resources, and technology, in order to explain the rural-urban development gap. In summary, the faces of the poor are typically children of all ages, young low-to-middle-skilled household heads, recent migrants, and non-homeowners. These groups are clearly worse off than pensioners, the college educated, the elderly, and non-recent urban migrants. Economic growth was instrumental for poverty reduction Sixth, income per capita growth is the dominant factor behind the gains and losses in urban poverty from 1978 to 1995. In this period, income per capita for the average Colombian household became 92 percent higher. During the 1980s and early 1990s, economic growth explained almost completely the 22-percentage-point reduction in urban poverty. In the same way, during the recent recession, poverty increased because of the combined effect of negative growth and increasing inequality. In particular, the extreme poverty rate appears more sensitive to growth than the "normal" poverty rate. Nevertheless, the unfortunate declining trend in the elasticity of labor demand to growth is creating more obstacles to transforming future economy- wide growth into income per capita growth through higher household employment ratios. This problematic trend justifies a critical assessment of all the policy instruments -both macro and micro - that might be distorting the labor skill intensity of private investment. Finally, through its detrimental effects on growth, the armed conflict in Colombia is jeopardizing any attempt to reduce poverty. More education and lower fertility were crucial for income per capita growth The household characteristics that explain most of income per capita growth during the last two decades have been the rise in education levels of the labor force and the reduction in the dependency ratios - through smaller family size- but not many changes in employment ratios or wages. However, among households with low-skilled heads, income per capita gains are also explained by a third factor: higher real wages. Not only is the direct impact of education on income is important -in terms of having better skills to sell in the labor market-, indirect effects are crucial as well. Education beyond high school buys substantial improvements in wages. Better education -especially of spouses- improves their chances of a high household employment rate -a key factor in escaping poverty. Education -especially women's- helps to reduce fertility, thus increasing per capita income. Finally, education also protects against domestic violence. In summary, the evidence shows that, in Colombia, education seems to play the single most important role among all social policy instruments. Job losses generate most of the poverty rise during the economic recession During the recessive period, most of the observed urban poverty increase of 7.5 percentage points was explained by job losses, and the remainder by lower wages. The asymmetric dynamics of job creation and wage adjustment between wage earners and self-employed sectors explain the outcome. Without any wage-downward flexibility, the market adjusted through job losses for wage earners during the recession - 6 percent of urban households were affected! Some of those vi households managed to become self-employed, and took a severe drop in labor earnings, and the rest were left without any labor income. Rising inequality partly eroded potential welfare gains Seventh, the rising trend of inequality reduced the potential welfare gains of Colombians up to 1995 and aggravated the welfare losses during the economic recession in the late 1990s. In Latin America, a region of relatively high income inequality, Colombia ranks near the top; in 1999, the Gini coefficient had reached a value 0.57. That is, the expected difference in income between any two Colombians chosen at random had been increasing over time, and it was responsible for a welfare loss - for the worst-off individual - of 18 percent between 1978 and 1995, and an additional 5 percent loss in the late 1990s. The dynamics of urban and rural inequality tend to diverge: while urban inequality rose continually during the whole period - deteriorating substantially during the 1990s -, rural inequality fluctuated. Despite the importance of between- rural-urban inequality, within-group inequality has an increasingly, dominant role in explaining the national trend. Education is the most important variable in understanding income inequality. By income sources, labor income is the driving force behind the observed trend, but pensions is the most unequalizing income component among non-labor income sources. Greater skill-wage differentials raised income inequality in'the 1990s To some extent, the inequality reductions of the 1960s and 1970s have been lost during the last two decades. This result is a combination of counterbalancing effects of equalizing and unequalizing forces that include wage differential by skills, non-labor income - pensions mainly - family size, supply of more educated workers, changes in labor participation, and occupational choice between being self-employed a wage-earner. Decomposition analysis shows that, paradoxically, the persistent equalization of education level of the labor force brought about expected lower income inequality for rural areas. However, the opposite happened in the urban centers, since skill-wage differentials were much higher. Moreover, increasing participation of less skilled women generated asymmetric effects between household income and individual wage distributions: regressive for the latter yet progressive for the former. Furthermore, while households appear to exacerbate static inequality among workers, they also attenuate the changes in individual income inequality produced by each income determinant factor. Sub period analysis reveals that the dynamics of income inequality might be decomposed as a combination of persistent and fluctuating forces. The former closely associated with demographic and the labor supply side, namely, the increase in education endowments, the reduction in family size, and the increase in labor force participation, and the latter with short-run fluctuations of labor markets. The aggregate effect of persistent inequality determinants leads us to expect an increasing deterioration of equality in the long run, unless the trend of increasing skill wage differentials is reversed. For these reason, special attention should be paid to postsecondary education, since the increasing earning differential of urban workers with postsecondary education relative to high school graduates is partly attributed to the sluggish supply of these high-skilled workers. GOVERNMENT ACTIONS AND PRIORITIES FOR THE POOR? Public policy instruments may tackle poverty by attacking its source or by attacking its most detrimental consequences, or both: that is, attacking the lack of income-earning opportunities and/or attacking the lack of access to basic social services. The profile of vulnerability to poverty in Colombia provides a guide for evaluation and choice of policy instruments. Lack of access to basic social services usually brings irreversible losses of human capital to vulnerable cohorts, especially children. Consequently, since those groups have also been found to be the most vii vulnerable to poverty in Colombia, subsidizing human capital accumulation helps to break the vicious cycle of poverty by putting assets in the hands of the children of the poor, as improving social mobility reduces inequality and poverty in the long run. At the same time, human capital investment constitutes a strategic component of long-term growth, reinforcing long-term poverty reduction. On the other hand, income opportunities for the poor in the short run are strongly correlated with the business cycle and the unemployment level. Therefore, the welfare of the poor is quite sensitive to macroeconomic management --fiscal balance, interest rates, exchange rates, and inflation-- through its impact on both the level and the variability of aggregate economic activity. Since volatility has shown an increasing trend during the last decade, safety net instruments, such as direct transfers in cash, kind, or physical assets, are becoming more relevant remedies for transitory insufficient income. Social services in the 1990s: huge public expenditure effort with mixed results across sectors Eight, during the last decade, Colombia has made a sustained effort to raise public social expenditure (PSE) to 15 percent of GDP, with considerable benefits for the poor. The 1990s were a period in which total public expenditure and social public expenditure grew rapidly and in similar proportions and, at the same time, major public sector reforms were introduced in subnational public finance, mainly decentralization, pensions, and health insurance. In spite of its growth, Colombian PSE is still 10 percent below the Latin American average. Ninety percent of PSE is devoted to pensions, education, and health. Despite a very limited coverage of the retired population, pensions is the largest category (40 percent), with minimal benefits for the poor, and with a fiscally unsustainable, explosive growth that will eventually shift the composition of PSE away from human capital investment. Comparisons across social services (education, childcare, health, housing subsidies, and public utilities) reveal substantial heterogeneity in the degree of administrative centralization, as well as targeting financial mechanisms. Between 1992 and 1997, all social programs, with the exception of childcare and health treatment, increased their coverage rates and became more pro-poor. In most sectors, however, public expenditure growth outpaced that of service delivery. Although overall incidence of PSE is broadly neutral, the poor benefit substantially, and its estimated impact on poverty reduction seems substantial. As a result of the universal health insurance reform of the early 1990s, insurance coverage almost doubled and became much more. Moreover, the simultaneous introduction of means-targeted subsidized health insurance for the poor reduced the poor's out-of- pocket expenses in the event of illness. Simultaneously, satisfaction with services improved across all income groups. Nevertheless, health immunization coverage suffered a severe drop during the last decade, and health treatment rates remained stagnant across income groups, despite the considerable addition of resources infused by the reform. High positive economic growth during the next decade is imperative to reverse the poverty increase brought about by the recent recession Ninth, recovering high economic growth is a necessary condition to return to the poverty reduction path that the Colombian economy enjoyed up until the recent recession. Average annual GDP growth would have to be at least 4 percent during this whole decade in order to reduce poverty to the pre-recession level. Poverty-reducing growth will depend crucially upon the recovery of the private sector, the job provider for Colombia's poor households. Such outcome will require several changes. In order to recover a favorable investment climate, the public sector is to be reformed so as to ensure fiscal sustainability. For this purpose, pension reform would generate substantial public resources in the medium term without hurting the poor. Also, real rates of interest must be set at efficient levels. The elasticity of employment to economic growth must also be recovered, especially for low-skilled workers, via recovery of the construction sector. This recovery will depend upon not only policies to re-establish a vigorous flow of viii mortgage credit, but also upon labor market policies that avoid inefficient payroll taxation and an excessive minimum wage. Social program priorities for the poor: childcare, health, and basic infrastructure Tenth, despite an increase in individuals covered, there are still substantial shortfalls in the coverage rate of some social programs, concentrated primarily among the lower-income groups. At the end of the 1990s, the highest priorities for social service expansion among the poor, when characterized as those which register the widest coverage gaps vis-A-vis middle- and high-income groups, appeared to be childcare, sewerage, health treatment, and health insurance. For the upper-low and middle-income groups, priorities are similar, with the important addition of tertiary education. However, at the current marginal cost, a substantial increase in the coverage of the latter is prohibitively expensive, and rather the aggressive expansion of college credit should prove to be an efficient solution, since its private returns are very attractive and the private sector has shown a very dynamic response market demand. Safety-net programs become more valuable policy instruments in an environment of increasing economic insecurity Eleventh, in Colombia the poor are disproportionately exposed to economic insecurity, or to the risk of facing an unexpected temporary loss of income. After being considered one of the most stable economies in the region, the volatility of the Colombian economy has nearly doubled during the 1990s. Under those circumstances the "archipelago" of traditional safety net programs was both fragile and insufficient. Moving towards a more integrated social protection policy would be significantly beneficial for the poor, who are exposed to increasing economic risk during the current economic slowdown. This policy should rest on three main pillars: a set of social assistance programs -for the chronically poor-, a social insurance system -for those exposed to idiosyncratic risk- and, a counter-cyclical public expenditure component -for covariate risk-. While raising family health insurance coverage among the poor would improve social insurance, recovering homeownership among the poor could enhance self-protection, and improving street safety in poor neighborhoods would raise protection of life and property. At the same time, Colombia should expand its menu of instruments of social assistance in order to face the challenges of some vulnerable groups that are receiving too little or no help. In order to help the most severely affected -in particular households with unemployed heads-, safety net mechanisms should address the undesirable temporary effects of the recession on current welfare and help prevent irreversible or permanent losses of human capital accumulation. Other key social programs for specific vulnerable groups. The extension of individual health insurance to family coverage-on both the subsidized and contributory regimes- after Ley 100 (1993) represents a major improvement in protection from economic insecurity. However, health treatment rates for the poor is still much lower relative to those of the non-poor. Homeownership, especially for the poor,- has plummeted during the recent recessive period. Thus, a self-protection device through asset ownership is being eroded, increasing the vulnerability of the poor. Housing policy should focus on: allocating more resources to affordable housing - Vivienda de Interis Social- redesigning targeting methods to control leakage and modifying mortgage credit markets regulation to avoid the excessive and inefficient real interest rates in effect since the mid-1990s. Unemployment of low-skilled household heads has increased considerably during the recession, with devastating impact on poor households. Therefore, low-skilled workers should be given priority in current emergency employment and subsidized health insurance programs. Poor households, and uneducated women in particular, are disproportionately exposed to domestic violence; hence, both the judiciary system and the public family welfare organization (ICBF, ix Instituto Colombiano de Bienestar Familiar) should coordinate their efforts to improve prevention of these types of episodes and protect the individuals exposed to evident risk. Children and youngsters of all ages are the demographic groups most exposed to poverty. Public programs should attend to their demand for social services to prevent irreversible losses of human capital investment with perverse effects on social mobility. Increasing the coverage of nutrition, childcare, and preschool education will benefit not only the children at risk but will give many poor families the opportunity to increase the labor force participation of non-head members and improve the odds of escaping poverty. Finally, young high school graduates face a disproportionate share of unemployment. Labor skill development through technical training and special exemptions of payroll taxation could help to improve the employment opportunities and improve their odds of escaping poverty. Moreover, young males' exposure to enormous homicide risk, with documented post-traumatic effects on survivors and witnesses, justifies specific preventive policies from both the judiciary and the public family welfare institute. On the other hand, decreasing the risk of unwanted pregnancy among young females in poor neighborhoods, with could have negative effects on social mobility, is a good reason to expand reproductive health programs among this demographic group. CONCLUSIONS The robust poverty-reducing trend that Colombia enjoyed up to the mid 1990s, has been interrupted by a severe recession that raised poverty rates back to the levels observed in 1988. Even if the positive growth path is promptly reestablished it will take more than a decade to return to the poverty levels of 1995. Moreover, increasing violence and crime has eroded the welfare of all Colombians, in particular the welfare of poor households who bear most of the burden of homicide and domestic violence, and of young males who face a very high risk of homicide. This represents a serious blow to welfare improvement and a major challenge in terms of policy choices to help the poor recover their levels of welfare and, more importantly, prevent irreversible losses in their human assets acquisition that could increase structural poverty in the long run. The most typical vulnerable groups are children of all ages, households with younger and less educated heads, displaced population, and non-homeowners. In order to help the poor, government must combine short and long term actions to attack both the source of poverty -lack of income- and its most detrimental consequences. Accordingly, in order to increase household income per capita primary objective in the short run is to attain higher economic growth and lower unemployment. In the long term, priority should be given to efficient acquisition of human capital assets that ensure growth and induce pro-poor behavior -namely, lower fertility and higher female labor participation-. In terms of fighting the undesirable long term consequences of poverty, the government should maintain subsidized provision of essential social services to the poor to promote efficient human and nonhuman asset acquisition, and, in the short run, it must consolidate and expand social safety net protection mechanisms. All this requires a coordinated government action in five policy areas: security, economic growth and employment, childcare and education, urban infrastructure, and healthcare. More importantly, those sectoral priorities should be politically feasible, given the fact that, according to Latinobarometro, nearly four out of every five Colombians believe that public policy should focus primarily on violence, unemployment or education. x In summary, political institutions in Colombia face top policy challenges in six main areas: * Security: Address governance issues to provide security of life and property to the common citizen. * Growth and employment: achieve macroeconomic and fiscal stability to recover sustainable growth rates in order to increase job creation and return to the poverty reduction path. * Social Protection: match the increasing economic volatility by developing an integrated social protection policy to enhance risk management instruments. * Childcare and education: improve efficiency and equity in the provision of child care and education -in preschool and technical training- in addition to expanding educational credit to tertiary education. * Infrastructure: expand coverage of sewerage and recover the trend of increasing homeownership. * Health: Improve efficiency in health treatment and reduce targeting leakages in the subsidized insurance regime. xi пх INTRODUCTION Until 1996, Colombia enjoyed high, sustained and stable growth. After 1996, however, Colombia is suffering its most severe economic crisis since the 1930s. Its macroeconomic performance has deteriorated significantly. Growth rates have declined sharply, the public sector has increasingly gone into deficit, and the unemployment rate has jumped to about 20 percent. These adverse economic situations emerged in a decade in which public policy reforms were abundant and macroshocks were significant. In addition to the constitutional reform of 1991, active public policy enforced structural reforms in key areas: subnational public finance, health care and insurance, pensions, labor markets, central bank framework, trade, and capital flows. Simultaneously, risk indicators worsened in key areas. After being considered as one of the most stable economies in the region, recent performance has shown greater volatility. International risk ratings of the Colombian e onorny were recently downgraded, increasing the cost of external borrowing. Financial sector performance has deteriorated and become more fragile, with significant bailout costs. The judiciary system has been stretched to its limits by increasing levels of crime and violence. Moreover, informal authoritarian regimes enforced by paramilitaries and local guerrilla warlords are in effect in some rural areas. The effects of those events on the welfare of the average Colombian can easily be anticipated. However, it is not clear how large the welfare losses are. How much have these losses affected the poor and other vulnerable groups? What types of vulnerabilities have emerged in the new economic environment? How appropriate was the government's response in terms of social investment and social protection? What are the main public policy challenges and opportunities to attack poverty in the next decade? This report appraises the impact of economic development on the welfare of the Colombian population -the poor in particular- during the last two decades. The report also identifies priority areas for public policy action to help the most vulnerable groups. Welfare assessment covers three key areas: income, access to social service, and personal security. Welfare assessment also compares those indicators between urban and rural areas and across other regional partitions.' In summary, this report addresses two main questions How are the poor doing and why? and What is the Government doing for the poor!' This volume is divided into five chapters. Chapter I shows bow t he poor have been are faring during the last two decades, including the recent recession. Sections evaluate welfare outcomes in Issues related to economic insecurity and social protection are presented in another simultaneous Bank report. See Rawlings, L. (2002) 1 three dimensions: economic welfare, access to social services, and exposure to crime and violence. Chapter 2 identifies the "faces" of the poor, i.e., the demographic and social characteristics of those groups most vulnerable to poverty. The Colombian poverty profile provides a guide for both the diagnostics and the evaluation and choice of policy instruments. Chapter 3 analyzes the effects of labor market forces upon the poor. In particular, the chapter investigates what is driving earning opportunities and how these opportunities modify the dynamics of poverty directly, and indirectly, through income inequality. Special attention is placed on (a) differentiating the impact of increasing unemployment and wage reductions on poverty during the recent recessive period and (b) understanding to what extent income per capita growth is explained by more schooling, lower fertility, higher participation in the labor markets, and higher real wages. Next the sources of increasing trends of income inequality are examined, or the impact of changes in returns to skills relative to changes in the distribution of skills endowments and changes in labor market participation. Chapter 4 examines to what extent public social expenditure benefits the poor, and identifies priority sectors for coverage expansion. One section is devoted to studying the performance of the subsidized health insurance system, an innovative social policy instrument introduced in the 1990s. Special attention is given to its interaction with the new proxy means targeting mechanism, SISBEN. Chapter 5 presents the conclusions and the main policy lessons and recommendations. Volume II contains background papers. Background: Changes in the economic environment during the last two decades On the basis of GNP per capita-$2,200 in 1997 (Table 1A)-Colombia is considered a lower- middle income country by international standards. Colombia, with a population of nearly 42 million as of 1999, is the third most populous in Latin America, with an expanding urban sector that accounts for nearly 62 percent of the total population. Similar to other Latin American countries, Colombian exhibits a high level of inequality, and, consequently, poverty incidence is relatively high vis-A-vis its level of development. Table IA. Colombia at a glance, 1980 to 1999 1980 1988 1995 1999 Population (millions) 28.4 33.6 38.6 41.5 Population growth gate 2.2% 2.0% 2.0% 1.8% GDP growth rate 4.1% 4.1% 5.2% -4.3% Unemployment Urban 7.71%* 10.3% 8.7% 19.7% Rural 1.46%* 4.6% 5.0% 10.9% Inflation Rate 26.4% 28.1% 19.5% 9.2% Government debt (% of GDP) -0.7% -1.4% -2.3% -5.8% Real effective exchange rate 146.68 87.34 100.00 102.68 Goods & Services (% of GDP) Exports 11.0% 12.0% 14.8% 18.2% Imports 12.3% 10.2% 21.3% 19.2% * 1978 2 The macroecoomk ana rectoralen qronmenlt n the 1980s ana early 199s Even though Colombia successfully avoided the large macroeconomic imbalances that plagued Latin America at the end of the 1970s, it suffered from a shortage of capital in the early 1980s. The previous prosperity resulted from careful management of the coffee revenues from the earlier decade. Thereafter, the 1980s started with a strong economic slowdown; that is, the growth rate fell from an average of 5.4 percent between 1975 and 1980 to 1 percent in 1982. This was accompanied by a rising deficit in the balance of payments as well as in the public sector, and a financial crisis. Some banks collapsed, and various financial intermediaries were nationalized. After the slowdown of the early 1980s, adjustment policies and the coffee bonanza of 1986 helped restore internal and external balances. To confront the crisis, the Colombian government reversed some of the liberal policies adopted previously, accelerated the devaluation, increased tariffs by 8 percent (average tariffs reached 34 percent in 1985), restricted public spending, instated a VAT tax, purged the financial sector, and provided support for social housing to stimulate growth. In 1986, the growth rate returned to more than 5 percent, the current account balance was positive, and the public spending deficit remained under control. In the latter part of the 1980s, the economy diversified away from coffee, unemployment fell, and growth was robust. Exports rose from 15.6 to a high of 22.7 percent of GDP between 1986 and 1991. Nontraditional exports increased from 32.1 to 49.1 percent of total exports during the same period, while oil, gas, and other minerals also rose from 18.6 to 32.6 percent of total exports. Unemployment had peaked at 14.7 percent in 1986, but subsequently oscillated between 9 and 12 percent in the late 1980s (Nifiez and Bernal, 1998). The growth rate also remained relatively high in the latter half of the 1980s, averaging at an annual rate of 4.5 percent between 1986 and 1990. The early 1990s saw an array of liberalization policies in trade and capital flows. Average tariffs fell from 44 percent in early 1990 to only 11.8 percent in March 1992 (Ocampo, et al., 1998). The exchange rate quickly became overvalued, partly because of large inflows of capital-a mix of fresh investment in new oil fields, increased borrowing by firms, and some repatriation of drug money-encouraged by the country's favorable investment grading. The strong peso and trade liberalization hurt farmers hard and pushed up prices of non-tradables, especially real estate assets. Moreover, the International Coffee Pact collapsed in 1989, leading to a substantial fall in coffee prices. At the same time, total public spending rose at a pace incompatible with sustainable fiscal policy. Until the early 1990s, prudent management of the Colombian economy allowed for low government debt levels that, together with low inflation rates by Latin American standards, led to steady-although moderate-growth rates. However, public spending in the 1990s pushed its share from 24 to 36 percent of GDP between 1990 and 1998. This corresponds partly to constitutional reform calling for the decentralization of public finances, which led to increasing reassigning of social spending to local governments; this reassigning was not followed by corresponding cuts in central government spending. In addition, public expenses in crime fighting and security increased owing to the escalation of crime and violence, while public sector pensions demanded increasing outlays (these public sector pensions were excluded from the social security reform in 1993). The government financed its budget deficit mainly through a combination of tax reforms-including greater reliance on VAT and increased measures to reduce tax evasion-and financing through commercial bank credit. In addition, after 1995, Colombia increased its reliance on external financing through bond issues. Total gross external central government debt increased by 125 percent between 1994 and 1999, while internal debt increased by 260 percent. The fiscal position continued to deteriorate as the economic slowdown adversely affected tax revenues. In 1998, the current account and fiscal deficits both ran at 5 3 percent of GDP. The country lost its investment-grade rating, which raised its cost of credit even more.2 Moreover, when the credit bubble burst, property prices plummeted, several banks filed for bankruptcy, and the Government was burdened with even more liabilities as it restructured the financial sector and provided mortgage debt relief. In addition, the January 1999 earthquake that affected Colombia's main coffee-growing region led to unforeseen government spending. At this rate, the public deficit was heading toward unsustainability. Adverse macroen Ponm en mi /e late /990s.- Recessian and rid deterioration The lack of fiscal discipline pushed Colombia toward a slowdown after 1996, and into recession in 1998-99. The GDP in 1999 contracted by more than 4 percent, and private investment fell to 4 percent of GDP, from 13 percent in 1995. Fiscal imbalances in an environment of passive monetary authority3 -Banco de la Repdblica- created persistent pressure to raise the rate of interest, inducing stock imbalances for households, firms and the Government itself. For households, Echeverry (2002) shows that between 1991 and 1997, mortgage credit exploded from 7 percent of GDP to 13 percent, which in turn raised debt service and interest payments.4 Once the real state bubble exploded, housing prices fell significantly inducing significant losses in net household wealth. Moreover, between 1991 and 1997, formal resources available for business investment shrank because of the increase -11 percentage points-5 in the level of mortgage debt service relative to total private savings. In addition, Colombia had to cope with the hurtful effects of increasing violence and internal armed conflict,6 and 1999's devastating earthquake in the Central coffee-growing region. After being considered one of the most stable economies in the region, recent performance of the Colombian economy has shown greater volatility.7 The Colombian business cycle has suffered dramatic change; cycle duration is now much shorter, with higher peaks and deeper troughs. Before 1996 the Colombian economy experienced relatively long business cycles of six to eight years in duration. However, the last cycle's duration was less than five years; not only the duration changed, but the amplitude of the last cycle was 50 percent% larger than the amplitude of cycles experienced between the late 1970s and early 1990s.8 Moreover, according to Partow (2002), the volatility of the Colombian economy during the 1990s increased relative to the two previous decades. Contrary to the Latin American trend, volatility nearly doubled for a number of macroeconomic and fiscal variables, with the exception of the terms of trade (see Table IB). The increasing volatility shown by the Colombian economic factors led to the recent downgrading of the risk-based ratings given to the Colombian economy, increasing the cost of external borrowing. Financial sector performance has deteriorated and has become more fragile, with significant bailout costs (Partow, 2002; Acosta, 2000). 2 Government debt servicing costs rose from 2.2 in 1994 to about 4.4 percent of GDP in 2000. 3 The 1991 Constitution determined that the Central Bank of Colombia -Banco de la Repiblica- be set as an independent body devoted exclusively to controlling and reducing inflation. 4Echeverry (2002) estimates that between 1991 and 1997, interest payments relative to private savings rose by 3.3 percentage points. Echeverry (2002) 6 The cost of the internal armed conflict and violence suffered by Colombia in terms of GDP growth has been estimated by Echeverry (2002) at 1.3 percentage points and by Rubio (1996) at 2 percentage points. 7 Rodrik (1999) estimates the probabilities of entering episodes of high volatility for various countries, and shows that Colombia faced a probability very close to zero for the last 30 years. 8 See Echeverry, Santamaria, and Escobar (2002). According to this article the main cause behind the change suffered by the Colombian business cycle is the Central Bank reform of the early 1990s. Countries with monetary policy set by an independent body are should experience greater volatility but lower resulting inflation, which is expected to bring reater economic growth. A number of authors-among them Gavin and others (1996), Herrera, Perry, and Quintero (1999), and IDB (1997))have pointed out that domestic volatility in many countries of the region declined during the 1990s. 4 Table 1B. Colombia's volatility 1970s 1980s 1990s 1990s: 1980s* GDP volatility 1.8 1.5 2.7 81% Fiscal volatility 1.3 1.4 2.7 91% Exchange rate volatility 1.1 1.5 2.4 57% Monetary volatility 6.8 3.5 3.7 56% Terms of trade volatility 13.0 10.4 5.8 -44% *1990s. 1980s indicates the change in volatility in the 1990s relative to the 1980s Note: Data represent the standard deviation, in percentage terms, of changes in GDP, in the fiscal balance/GDP, in the real exchange rate index, in MI growth, and in the terms of trade index Source: Partow (2002) In remote areas, the rule of law is under severe challenge and, overall, the judiciary system is stretched to its limits. Crime and violence-political, drug related, and normal street crime- remain high.'o The widespread practice of extortion and kidnapping-which pervades all strata of society-is increasingly weakening property rights over physical assets and thereby undermining the market economy. Moreover, de facto authoritarian regimes enforced by local warlords-paramilitaries or guerrillas-effectively rule some rural areas. Judicial uncertainty is leading to unpredictability of public mandate in key policy areas, thus reinforcing market uncertainty. A number of recent judicial rulings created increasing uncertainty about the "rules of the game" in several key economic markets, such as labor, mortgage credit, and the private provision of education. In addition, constitutional rulings on pensions rights have cast doubts about the contingent liabilities of the public sector." As a result, political mechanisms appear to have lost effectiveness in solving the fiscal crisis. For example, it is relatively uncertain that the reforms of the pension system eventually passed by the elected legislative branch-which are urgently needed to alleviate the unsustainable fiscal deficit-will become a public mandate, since the reforms may be overturned by the Constitutional Court. Sapply xhock, strucaral /reform, and employmentgeneraion During the last decade, the Colombian economy experienced supply shocks and several structural reforms that significantly modified the labor market environment. Reforms include the Labor Reform of 1990 and the Pensions and Health Insurance Reform of 1993, as well as a string of trade liberalization measures that dismantled import quotas and reduced average import tariffs from 40 percent in 1990 to 11 percent in 1993. Furthermore, free trade agreements were signed with Mexico and Venezuela, and the capital account was opened in 1993. The economy also enjoyed supply shocks caused by major discoveries of oil reserves, which led to expectations of exchange rate appreciation and a jump in export revenues in the second half of the 1990s.12 I its Poverty in Colombia report (1994), the Bank warned about the danger of missing this opportunity to attain a higher growth path if prudent fiscal policy was not followed and public resources were not invested in high return projects that create human and physical capital.13 Finally, decentralization transferred resources to the municipalities to pay for public education and health, under the presumption that service delivery should improve. The Bank also warned that institutional weakness of the social sector at the local level could further increase community disparities.14 io Homicide rates are roughly 59 per 100,000, among the highest in the world. " Below, see Box I about mortgage credit. 12 Ex-post oil revenues were below expected levels owing to lower oil prices in this period. "3 World Bank (1995), Chapter 2, p. 28 14 World Bank (1995), Chapter 3, p. 52. 5 There was satisfactory growth performance up to 1995, but declining elasticity of employment to GDP. Although the growth performance of the Colombian economy was satisfactory between 1978 and 1995, the elasticity of urban employment to GDP declined as the growth rate picked up in the early 1990s. The country's GDP per capita grew at an average annual rate of 1.8 percent, with very low instability.'5 Economic growth was higher during the second sub-period: On average, the GDP growth rate went from 3.3 percent in 1978-88 to 4.3 percent in 1988-95.16 However, labor demand response was less intense, so that employment growth fell from 5.2 to 3.2 percent for the corresponding sub-periods. Increasing difficulties for low skill job creation in the 1990s. Lower elasticity of job creation relative to growth should be associated with weaker labor demand, although not exclusively. Several macroevents and structural reforms brought (a) an exchange rate appreciation and labor legislation reforms in the early 1990s that increased the relative cost of labor relative to capital and made job creation for less skilled workers more difficult; (b) a tendency of domestic industry to invest in more capital-intensive technology, as exposure to international competition rose owing to tariff reductions and regional trade integration; (c) a gradual recomposition of productive activities toward more capital-intensive activities, as production shifted from agriculture and industry to mining and services; and (d) an increasing corporate tax rate during the early 1990s. Only one factor helped to reinforce the demand for low-skilled labor: the five- fold increase in construction activity in the early 1990s, closely related to exchange rate appreciation, which derived from unprecedented capital inflows that increased the relative price of non-tradables to tradables.17 However, the construction bubble collapsed during the late 1990s and consequently reduced the dynamism of labor demand for less skilled labor. Rising costs of wage-earning job creation relative to self-employment. The substantial rise in payroll taxation also made the generation of wage-earning jobs increasingly difficult vis-h-vis self-employment. In fact, during the 1988-95 sub-period the loss of job creation dynamism was mostly concentrated in wage earners and less in self-employment. Between 1978 and 1988, the rate of growth of self-employment was 1.5 times the rate for wage earners; however, from 1988 to 1995, the former rate becomes twice (!) the latter. Although the labor reform of 1990 (Ley 50) somewhat reduced labor costs by diminishing the expected value and the risk of cost of dismissal-"cesantias"-subsequent labor legislation reforms produced the opposite effect. The reform of social security regimes, including health insurance and pensions (Ley 100 de 1993), resulted in an almost doubling of payroll contributions (!) and produced a substantial increase in the labor cost gap between wage earner and self-employed workers.'8 Major changes hn the soci'-dem ographk structure of/he wor,kigpopulaion. During the last two decades, the socio-demographic characteristics of the working population registered positive changes. There were changes in fertility, educational attainment, and female labor force participation. Fertility rates decreased throughout the entire period. School attainment IS A standard deviation of 0.02 percent. Moreover, a cross-country study by Rodrik (1999) shows that the probability of the Colombian economy entering episodes of "high volatility" in the last 30 years was close to zero. 16 Nevertheless, cycles were not completely absent and the economy went through a moderate recessive period during the first half of the 1980s. In the second half of the 1980s, macroeconomic policy kept a competitive exchange rate and a moderate public deficit. Interest rates fell and some trade restrictions were lifted. Nontraditional exports grew at a high pace. After a low level of activity in 1991, the economy recovered once again in 1994 and 1995. 17 In addition to the liberation of the capital account in 1993, the economy suffered supply shocks derived from major discoveries of oil reserves, which resulted in a jump in oil export revenues and expectations of exchange rate predation in the second half of the 1990s. See Cardenas and V61ez (1997). Payroll contributions increased 13 percentage points (!), up to 13.5 percent for pensions and 12 percent for health insurance; this was on top of preexisting payroll taxes of 9 percent, earmarked for labor training, and social welfare programs. In summary, including remunerated annual leave and prima semestral, the reforms of 1993 lifted total payroll contributions to 59.4 percent for regular workers. 6 rose and became more egalitarian. The working population gained labor market experience and became more educated. Declining gender earnings differentials, higher educational attainment, and lower fertility rates helped to increase female labor force participation throughout the entire period. Male heads and other household members began to move away from wage-employment toward self-employment, especially during the 1990s. Concentraiol oflandandrura/credi? Land and credit: deconcentration started in the middle 1980s. Ownership of land went through a period of concentration from 1974 to 1984. Since then land concentration decreased as the land-value Gini coefficient decreased from 0.61 in 1984 to 0.59 in 1999. International comparisons show that in 1996 concentration of land in Colombia was less than that in Czechoslovakia, Paraguay, Brazil, Argentina, and Panama.19 In a period of subsidized interest rates and rationing of credit, credit was particularly concentrated among large-scale producers between 1974 and 1984, but since then competitive interest rates have tended to deconcentrate it (Leibovich and Ndfiez, 2000). Dirplaced,popula.o7 and arm ed cofict in te rural areas From 1985 to 1999, between 1.2 and 1.5 million Colombians are estimated to have been forced to leave their homes, owing to the escalating military conflict.20 Most of those displaced persons remain internally displaced people (hereinafter IDP) and reside in medium and large cities. Forced displacement appears to be concentrated in a few departments. Even though studies of IDP in Colombia are few and their results are not comparable, the main expulsion areas appear to be Antioquia (Central region); Santander and Meta (Oriental region); C6rdoba, Magdalena, and Sucre (Atlantic region), and more recently Choc6 and Narifio (Pacific region) and Caqueti (Central region) (Erazo, et al., 2000). According to The Economist (2001a and 2001b), these departments are also the ones that receive the greatest amount of petroleum royalties and the ones in which drug traffickers acquired most land during the 1980s.22 1/kt crops and rural developmeni Colombia is now the world's largest producer of cocaine and has recently become a significant supplier of opium poppy and heroin. Production of cocaine grew from 158 metric tons in 1995 to 326 metric tons in 1999, with the area under cultivation increasing from 50,900 to 160,119 hectares in the same period (UNDCP 2000).23 In addition, Colombia hosts most of the clandestine manufacturing laboratories, which reportedly processes 80 percent of the world's demand for cocaine.24 In 1994, coca became Colombia's fifth most valuable agricultural activity, behind cattle, poultry, coffee, and sugar (Jaramillo 1998). Production of opium reached 102 metric tons in 1998, placing Colombia as the world's fourth largest producer (3 percent of the world's production). The area of production increased from 5,200 hectares in 1995 to 7,400 in 19 Deininger and Squire (1998) quoted by Leibovich and Ndiiez (2001). 20 From CODHES Boletin no. 28, February 22, 2000, Bogot6; quoted in Partridge and Arboleda (2001) The Population Displaced by Armed Conflict in Colombia, processed, Colombia Country Unit, Washington, D.C, World Bank. 21 On January 25, 1999, the Armenia earthquake in Quindfo affected this department, as well as Risaralda, Caldas, and Tolima in the Central region, and Valle del Cauca in the Pacific region, leaving 150,000-200,000 homeless and also causing some displacement. 22 "A Survey of Colombia. Drugs, War and Democracy." The Economist. April 21, 2001. It is estimated that drug barons may have acquired more than I million hectares in cattle farms and other estates at that time. 2 The figures were 34,000 hectares and 51 tons in 1988. 24 The UND.C.P reports seizure of 470 labs in 1996 and 14 tons, 208 in 1997 for 41 tons of coca paste. For heroin, 81 labs were seized in 1997. 7 1998 (UNDCP 2000). The value of the poppy harvest was similar to that of beans, cassava, and sorghum in 1994. Colombia is also a minor producer of cannabis and is the second largest Latin American producer after Mexico. Coca production tends to concentrate in (or near) the Oriental and the extreme south of the Central region. Typically, the coca leaf production areas are located far away from urban centers and in nontraditional agricultural areas. Most of the cultivated areas are in the departments of Putumayo, Caquetd, Guaviare, and Norte de Santander, and Bolivar, where production is rising. While drug eradication efforts have been successful in departments such as Caquetd and Guaviare, on a national level coca plantation has not diminished. The coca-growing areas overlap with the territories of the guerillas and paramilitaries. In summary, during the last two decades Colombia has faced increasing difficulties in several key areas, resulting in a more challenging environment for economic and social development. First, worsened risk indicators, mostly owing to increasing economic volatility, financial sector fragility, and persistently high levels of crime and violence have stretched the judiciary and public security systems beyond their limits. Moreover, increasing concentration of illegal drug production--cocaine-in rural areas provided the economic base for de facto authoritarian regimes enforced by paramilitary and local guerrilla warlords. In turn the escalation of the armed conflict is pushing an increasing number of displaced population into urban areas. Second, this situation has emerged during a decade in which public policy reforms were abundant and supply shocks significant. Major discoveries of oil reserves produced a jump in export revenues in the second half of the 1990s and led to expectations of exchange rate appreciation. In addition to the constitutional reform of 1991, an activist public policy enforced structural reforms in key areas, such as sub-national public finance, health, pensions, labor markets, central bank framework, trade, and capital flows. A lower elasticity of job creation relative to growth reveals the increasing difficulties for low skill job creation emerged in the 1990s, especially for wage-earning jobs. Finally, on the bright side, the socio-demographic characteristics of the working population registered positive changes in declining fertility, higher educational attainment, and an increase in female labor force participation. 8 Chapter I How Have the Poor Fared Drng the Last Fwo Decades and the Recent Recessinrn? Poverty trends in Colombia show substantial long-term progress with a recent setback. Poverty trends declined during the 1980s and during the first half of the 1990s. However, after a reduction of 20 percentage points from 1978 to 1995, poverty reversed itself and, in 1999, it returned to the level registered in 1988. Economic and social development gains have been resilient to growing insecurity. In other words, the three welfare trends given by household income, social development, and personal security do not match but, rather, diverge and are in partial contradiction. Despite two decades of persistent security deterioration, Colombia showed unambiguous improvements of social and economic indicators. Moreover, social investment in human resources showed some healthy robustness relative to recent market instability. Social development trends were positive in most sectors. In urban education, completion rates for primary and secondary schooling improved substantially. In rural education enrollment rates improved, especially for the 18-22 year olds. Child labor maintained a decreasing trend, despite being pro-cyclical, as in other Latin American countries. Similarly, child malnutrition and infant mortality have had persistent declining trends. Life expectancy improved approximately one year every two calendar years. Access to basic infrastructure had clear progressive gains. Urban coverage of most public utilities improved, but access to sewerage still has room for improvement. Violence and the continuous worsening of crime since the 1970s has eroded the welfare of all Colombians. However, poor households bear most of the burden of homicide and domestic violence, while young Colombian males face an extremely high risk of being murdered. In particular, uneducated women and/or spouses of uneducated men bear a disproportionate share of domestic violence. In contrast, the better off bear a disproportionate share of the burden of property crime, extortion, and kidnapping. The violence trend is mostly associated with the illegal drug trade and the presence of illegal armed groups. The armed conflict in rural areas has also created an increasing trend in the numbers of internally displaced population-at least 1 million-further increasing the vulnerability of these groups. Moreover, the concentration of property crime among business owners-big and small-has a perverse effect on investment, growth, and employment for poor households. Thus, as long as poor, middle-class, and rich Colombians continue feeling defenseless outside their homes and poor Colombian women experience similar feelings inside theirs, the possibility of recovering economic prosperity and equity becomes more distant. 9 This chapter explains how are the poor have fared during the last two decades by examining the evolution of three core dimensions of welfare: economic welfare, access to social services, and security. The first section deals with economic welfare and poverty. Special attention is devoted to identifying the specific effects of different periods of the business cycle on the welfare of the poor. The main questions to be address are the following: Are the poor better off now than two decades ago? How much did the poor lose during the recent recession? Did all income groups gain equally during growth periods? Who lost more during the recent recession? How do those conclusions vary between urban and rural areas, and across urban centers and rural regions? The second section examines the evolution of access to basic social services and the level of social living conditions, i.e., access to education and public utilities, educational attainment, illiteracy, child labor, and malnutrition and infant mortality. This represents an essential non-monetary angle of welfare, and it shows not only how successful Colombia has been in raising access to basic social services but also how far key final social outcomes have improvecL. Equity aspects of access to basic social services will be examined in Chapter IV. The third section studies the burden of violence and crime. Little is known about the distribution of crime and violence in Colombia and about the various ways in which individuals and households have reacted to the escalation of violence. Hence, this chapter explains how violence and crime have differentially affected the various sectors of Colombian society (by income groups, age groups, gender, regions). The distribution of crime and violence across the spectrum has not only fairness implications, but important policy implications as well. In particular, the distribution of crime can affect individual coping mechanisms and the level of public social spending on security. It might also have implications for life expectancy across gender and age groups. In addition, crime can create new vulnerable demographic groups, as is the case for the population displaced from rural areas where illegal armed forces are currently operating. 1. POVERTY, ECONOMIC WELFARE, AND INCOME DISTRIBUTION 1.1 Poverty trends showed substantial long-term progress with a recent setback Table 2 presents the evolution of the basic poverty indicators --the poverty headcount, the poverty gap, and the P2 index -from 1978 to 1999. The poverty gap describes the average distance of the poor from the poverty line; by expressing it as a percentage of the poverty line, it is in fact approximating the value of the resources necessary to bring the poor to the poverty line. The P2 index (FGT) captures the degree of income inequality among the poor. Table 2 also includes measures of average income per capita inequality and rural-urban decomposition of the Theil entropy measure. Poverty trends in Colombia show substantial long-term progress with a recent setback. According to the estimates presented in this report, about 64 percent of the population is estimated to have been living in poverty in 1999 and 23 percent in extreme poverty. Poverty declined during the 1980s and during the first half of the 1990s. However, it reversed itself in 1999, returning to 1988 levels. Something similar happened to the poverty gap and FGT. Extreme poverty, on the other hand, shows small but important differences. It declined much faster than the poverty rate between 1978 and 1995, falling by more than 50 percent (from 45 to 21 percent). Again, the recession did aggravate the extreme poverty rate, but the level it reached was still well below that of 1988. Average income per capita almost doubled-rising by 92 percent from 1978 to 1995--only to subsequently fall by nearly 3 percent in the recessive period. 10 tlrhan,poverly. -In hhe with the naiotnal tr-end After a continuous and significant reduction in urban poverty from 1978 to 1995, the economic recession pushed poverty indicators back to 1988 levels. Urban poverty decreased at rates close to 1.5 percentage points per year between 1978 and 1995. Unfortunately, the 1999 rates are again close to those of 1988, even slightly higher for the poverty gap. The recession not only increased the percentage of poor people in urban areas, but the poor also became poorer, an observation confirmed by the increase of extreme poor people, reported in Table2. Economic recession hits the extreme poor more badly and erases the substantial gains of the 1978-95 period. Extreme poverty followed a similar U-turn pattern, although extreme poverty had sharply decreased with income growth (the 1978-95 period showed a 70 percent decrease). Half of those gains were lost in the last four years. Colombia's economic recession seems to have hurt more severely the very poorest segments of the population. The behavior of the P2- intensity of poverty-measure provides one more confirmation: It rose by 22 percent between 1995 and 1999, indicating both deeper poverty and a higher level of income inequality among the poor. All these results are very robust to the poverty line chosen.2 In number terms, the evolution of poverty rates implies that in 1978 almost 5.4 million persons lived in poverty and 2.3 million in extreme poverty in the largest seven cities. In 1995, the poor numbered 5.7 million and the extreme poor 1.15 million. The figures for 1999 are 8.0 million and 2.1 million, respectively. RaralPoverty.- Clearprogress up to 199I but much higher rates in arban areas After a substantial reduction in the 1980s, the rural poverty headcount ratio has remained relatively stable since 1988 (see Table 2). Similar to urban areas, rural poverty fell sharply (14 percentage points) between 1978 and 1988.26 On the other hand, in contrast with urban areas, where poverty dropped by almost 4 percentage points between 1988 and 1995, rural rates only saw a decline of I percentage point. While urban incomes grew at an average annual rate of 2.9 percent in the 1988-95 period, this figure was only 0.7 percent in rural areas. Nonetheless, during the economic downturn, rural areas were not as negatively affected, as mean income per capita continued to grow and poverty increased by only 0.2 percentage points versus a 3 percentage point increase in urban areas. Extreme rural poverty declined much faster than in urban areas up until 1995. Both the extreme poverty rate and the US$2/day poverty rate declined much faster than the poverty rate during the first 2 sub-periods, a phenomenon not unlike that found in urban areas. Thereafter, these rates remained relatively stable, just like the poverty rate did. While the US$2/day poverty rate fell by a 23 percentage points between 1978 and 1988, extreme poverty rates fell by a similarly substantial 20 percentage points.27 Between 1988 and 1995, when poverty was reduced by only 1 percentage point, extreme poverty and the US$2/day poverty rates declined by 11 and 7 percentage points, respectively. Both of these rates have recently stagnated, with the extreme poverty rate remaining the same and the US $2/day poverty rate slightly decreasing between 1995 and 1999. 25 Similar trends were found for computations of poverty and extreme poverty rates for the US$1 per day line as well as a wide range of hypothetical poverty lines. 26 The urban poverty rate fell by 15 percentage points during this same period. 27 The extreme poverty rate in urban areas fell 10 percentage points during the same penod. 11 Table 2. Poverty indicators: National, Urban, and Rural Colombia 1978-99 1978 1988 1995 1999 National Poverty rate 80% 65% 60% 64% Poverty Gap 46% 32% 29% 34% FGT P(2)2 32% 20% 17% 22% Extreme poverty rate 45% 29% 21% 23% US$ 2 per day poverty 44% 28% 22% 24% Mean income per capita 112 183 216 210 Urban Poverty rate 70% 55% 48% 55% Poverty Gap 35% 23% 19% 26% FGT P(2)2 21% 13% 10% 15% Extreme poverty rate 27% 17% 10% 14% US$ 2 per day poverty 23% 13% 8% 12% Mean income per capital 157 235 295 277 Rural Poverty rate 94% 80% 79% 79% Poverty Gap 61% 43% 40% 44% FGT P(2)2 44% 29% 25% 29% Extreme poverty rate 68% 48% 37% 37% US$ 2 per day poverty2 73% 50% 43% 42% Income per capital Median 37 63 68 65 Mean 52 90 95 102 1. Thoussand 1999 pesos, based on monthly household income. 2. Foster-Greer-Thorbecke index. 3. Based on Purchasing Power Parity Convertors from WDI database. Relative stability of rural poverty counts hides recent deterioration of the poverty gap and poverty intensity (P2). The recent relative stability of the poverty count as measured by the three different poverty lines hides a widening of both the poverty gap and the P2 index in the last 4 years. These two indicators had progressed faster than the poverty count until 1988, but since then the poverty gap and P2 index have risen by 3.2 and 3.5 percentage points respectively, as shown in Table 2. Given that most of the population lies below the poverty line, it follows that much of rural Colombia has felt the negative impact of the recession. Most worrisome is the larger rise in the P2 index, implying that the brunt of the recession is being carried by the very poor. This is also indicated by the fact that the US$2/day poverty rate has increased slightly, although the extreme poverty rate has declined. 1.2 Income per capita nearly doubles in two decades but falls in the late 1990s Urban average household income per capita doubled between 1978 and 1995, but the recession erased at least six years of economic progress. A simple way to assess welfare changes is to consider the evolution of average household per capita income. Average household income per capita almost doubled: It went from C$157,08028 (US$250) in 1978 to C$277,469 28 All monetary figures are in 1999 C$. The PPP adjusted exchange rate is US$1 = C$629. 12 (US$441) in 1999. This overall increase hides very different patterns among the sub-periods. Average income per capita grew at an average yearly rate of 4.1 percent during 1978-88, 2.9 percent between 1988 and 1995 but fell by 1.2 percent between 1995 and 1999, reflecting the severity of the crisis. The economic downturn has erased at least 6 or 7 years of economic progress for the average urban household. Rural average household income per capita: strong improvement in the 1980s, slowdown in the early 1990s and some recovery after 1995. Table 2 shows that the rise in mean household income per capita between 1978 and 1988 was 72 percent, rising at an average annual growth rate of 5.6 percent per year, faster than the growth rate in urban areas (4.1 percent); 1988 was an exceptional year across all rural regions, with agriculture playing a major role. The devaluation helped agricultural exports, which was further supported by high tariffs for cereals and higher coffee production and prices. Later, mean income rose at a much slower rate of only 0.7 percent a year between 1988 and 1995 (versus 2.9 percent in urban areas), but continued rising between 1995 and 1999, this time at a faster average annual rate of 1.4 percent (versus minus 1.2 percent in urban areas). This overall evolution hides striking differences between regions and income groups, as will be discussed below. Mean household income in rural Colombia does not suffer during the recessive period. As opposed to the fall experienced in urban areas, mean (monthly) household income per capita rose during the economic downturn; however, median household income per capita declined. After behaving (rising) much like average mean income, the median income fell at an average annual rate of 0.8 percent during the recessive period. The fall in the median income and the rise in inequality in 1999 indicate that more than half of the population was indeed negatively affected by the recession. Over the years, the rural-urban gap fluctuates and falls in the late 1990s. It is also notable that throughout the 21-year period studied, rural average income per capita remained at roughly one third of that in urban areas. However, urban-rural disparities in economic growth produced fluctuations in the relative income ratio (Table2). Despite impressive rural growth between 1978 and 1988 that brought relative income to a qualified maximum of 38 percent, the better performance of the urban sector up to 1995 reduced rural Colombia's relative income to a minimum of 32 percent. This was then partially reversed by negative urban growth in the late 1990s, bringing rural relative income to 37 percent in 1999. 1.3 Rising inequality eroded welfare gains: The Sen welfare index Income inequality trends deteriorated mostly in the 1990s. Several authors have identified the mid 1960s as the breaking point in the regressive trend of income distribution during the first half of the twentieth century. By the late 1970s, the Colombian economy had completed two decades of consistent reduction in income inequality and had improved its standing with respect to other Latin American countries. After the persistently raising inequality of the first half of the twentieth century, substantial inequality reductions were observed during 1960s and 1970s as the economy grew. Colombia appeared as exemplary of Kuznets' inverted U-shaped curve. However, during the late 1970s and the 1980s, inequality levels plateaued, and during the last decade took a clear U-turn erasing the equity gains of the two previous decades. 29 Urrutia (1984), Reyes (1987), Ocampo (1992), and Londoflo (1995). Declining wage differentials of the 1970s fostered significant improvements in inequality (Misi6n de Empleo 1986). According to Ocampo, et al. (1998) the determinants behind these developments were: (a) the reduction of the rural labor force surplus, owing to fast migration in the 1950s; (b) the robust pace of capital accumulation and modernization in the rural sector; and (c) the larger and well-targeted investment in education and healthcare delivered through the "Frente Nacional." On the progressivity of public social expenditure, see Selowsky (1976) and V6lez (1996). 13 Figure 1. Colombia's income inequality in the international context, 1999 Brazil (1999) 1009.Mexic d Uruguay oombia 5 8China Co 70%~ 7w_ -7 60% 50% > 30% - ~ E 20% - ~ 10% - 0%. 20 26 32 33 36 37 40 41 46 51 57 Gini coefficient % (income inequality) Source World Bank Indicators, 2002 Table 3. Income inequality indicators: National, Urban, and Rural Colombia, 1978-1999 1978 1988 1995 1999 National Gini 0.53 0.54 0.56 0.57 Entropy Measure EO 0.51 0.54 0.57 0.61 Entropy Measure El 0.54 0.60 0.70 0.68 Entropy Measure E2 1.05 1.52 2.60 1.80 P90/P1O 12.14 11.50 11.08 14.38 P75/P25 3.63 3.31 3.43 3.72 Share q5/Share q1 17.17 17.58 17.16 20.17 Inequality Urban-Rural Decomposition (Theil) Within 0.42 0.50 0.59 0.58 Between 0.12 0.10 0.11 0.09 Urban Gini 0.47 0.49 0.52 0.54 Theil index (Eo) 0.34 0.41 0.48 0.54 P90/P1O 8.00 8.50 8.44 12.00 Rural Gini 0.45 0.47 0.45 0.50 Theil index (Eo) 0.36 0.40 0.36 0.46 P90/P1O 7.36 8.85 7.22 9.67 Colombia's income inequality is extreme in the international context but relatively moderate in Latin America, a region of relatively high income inequality where it ranks just above the median. Figure 1 shows that the Colombian population faces more income inequality than 99 14 countries containing 94 percent of the world's population. Nevertheless, in the Latin American context, Colombia's inequality appears relatively moderate. Out of 17 countries listed in the World Bank data for 1999, Colombia ranks seventh after Nicaragua, Brazil, Honduras, Bolivia, Paraguay, and Chile. Under almost any possible measure, inequality has been deteriorating during the last two decades. Inequality changes did not follow the same patterns as average income per capita. Table 3 shows seven possible measures of income inequality, and all seven are worse in 1999 when compared to 1978. While the Gini coefficient steadily increased over the entire period-by 4 percentage points-the share of the top quintile of the distribution relative to the poorest 20 percent-first quintile-increased from 17 to 20 times.30 Five out of seven inequality indicators show acceleration in the latest 1995-99 sub-period. This pattern of inequality dynamics is also shown by the stability of the P90/Pl0 and P75/P25 ratios until 1995-around 8.4 and 2.9, respectively-before they shot up to 12.0 and 3.5 in 1999. All three entropy measures (Eo, El, E2) also increased continuously until 1995. Thereafter, Eo continued to increase (from 0.48 to 0.54), confirming once more the increasing inequality among the poor. Despite the importance of rural-urban differentials, within-group inequality has an increasingly dominant role in explaining the national trend. Rural-urban decomposition of the Theil index shows that most of the rise in national inequality during the last two decades is associated to greater within inequality. In fact, between inequality explains only one-sixth of total inequality and has been relatively stable with a very weak decreasing trend. For this reason and since the dynamics of urban and rural inequality tend to diverge during the period of analysis, we will present below a separate discussion of the dynamics of urban and rural poverty and inequality. 1rban we/fare was hur- by inequaliy deleroration, mostly in the 1990s The rising trend of inequality reduced the potential welfare gains up to 1995 and aggravated the welfare losses during the economic recession. Income inequality is of concern since high levels erode potential welfare gains. The welfare of the average Colombian is represented by average income per capita. At the same time, the expected difference in income between any Colombians chosen at random is proportional to inequality and measured by the Gini coefficient. The welfare of the worst-off individual in any pair of Colombians picked at random is equal to the so-called Sen welfare index. 3' Figure 2 depicts the evolution of the mean household per capita income along with the Sen welfare index S. Average income increased by 88 percent between 1978 and 1995, and has leveled off if not decreased during the second half of the 1990s. But the Gini coefficient jumped 5 percentage points between 1978 and 1995 and 2 more points in the late 1990s. Had inequality remained constant, the Sen welfare index would have increased proportionally with income. The increase in inequality was responsible for an 18 percent loss in welfare gains between 1978 and 1995, and an additional 5 percent loss in the late 1990s. 30 This increase in the Gim coefficient means that the expected difference in income between two Colombians picked at random has increased by 8 percent. 3 The Sen welfare index is defined as S = u(1 - G), where u is the sample mean income per capita and G is the Gini coefficient. Therefore, S is a fraction (1 - G) of mean income u-the welfare of the average individual-and has a simple intuitive and relevant interpretation. S is equal to the expected welfare of the worst-off individual taken from any couple of individuals chosen at random, (Yitzhaki, 1984). 15 Box 1. Overestimated poverty by using income measures. Welfare measures based on consumption suggests that current income measures of urban and rural poverty are upwardly biased and urban mequality is underestimated Colombia's LSMS type surveys allows us to check any bias in welfare measures based on quarterly household surveys. While the figures calculated using Encuesta Nacional de Hogares -ENH- provide us with a clear picture of the trends of poverty and income inequality during the last two decades, it is worth noting that such figures, when computed using 1997 Encuesta de Calidad de Vida (ECV97), show a significant upward bias in poverty estimates. The ECV97 registers much lower poverty rates than those found in ENH for both 1995 and 1999. The urban and rural rates are 31 and 63 percent, respectively, while the ENH data in 1995 and 1999 register urban poverty rates of 48 and 55 percent, respectively, and a rural rate of 79 percent in both years. The poverty gaps are also significantly different-the ECV97 urban poverty gap was only 7 percent, while the other surveys register 19 and 26 percent. Urban extreme poverty rates and the P(2) measure are much closer in value, however, only differing by about 3 or 4 percentage points, whereas rural rates still differ by ten or more percentage points. In terms of mean income per capita and income inequality, differences are staggering only for urban data. Mean income per capita for ECV97 is over C$150,000 higher than that of ENH 1995 or 1999, while the Gini reaches 0.60. Only the P75/P25 measure is similar in the ECV97 and ENH surveys; all other inequality measures depict the ECV97 sample as much more unequal. These vast differences could be partly attributed to top coding in 1995 urban ENH data. Table BI. Poverty and inequality based on ECV and NHS surveys, Colombia. 1997 Urbano2 Rural 2 Urbano Rural 1995 1999 1995 1999 Mean income per capita 442,502 102,004 294,522 277,469 94,851 101,630 Income inequality Gini 0.598 0.451 0.522 0.545 0.446 0.502 Entropy Measure EO 0.68 0.37 0.48 0.54 0.36 0.46 Entropy Measure El 0.81 0.40 0.62 0.60 0.42 0.52 Entropy Measure E2 2.91 0.77 2.15 1.50 1.01 1.22 P90/Pl0 13.0 7.7 8.4 12.0 7.2 9.7 P75/P25 3.6 2.8 2.9 3.5 2.7 3.0 Share q5/Share ql 23.0 10.7 13.3 17.2 10.2 14.4 Poverty Poverty rate 31% 63% 48% 55% 79% 79% Poverty Gap 7% 29% 19% 26% 40% 44% FGT P(2) 12% 17% 10% 15% 25% 29% Extreme poverty rate 10% 22% 10% 14% 37% 37% 1. Based on Encuesta de Calidad de Vida, 1997. 2.Based on National Household Surveys. 3. Foster-Greer-Thorbecke index. 16 Figure 2. Average income per capita and the Sen welfare index, Urban Colombia, 1978, 1988, 1995, and 1999 280,000 230,000 --Mean household income per capita 180,000 - 0 -Sen welfare index 130,000- -e 1975 1988 1995 1999 Year Figure 3. Welfare, Rural Colombia, 1978, 1988, 1995, and 1999 105,000 95,000 85,000 lcr g 75,000 Mean income 65,000 -*-Sen Index 55,000 45,000 35,000 25,000 1978 1988 1995 1999 Rural Sew We//are ndex.- Rtsg eVua/ly eroded we/fare gazns even/ar/her //an in he urban case Although income has risen substantially since 1978, a concurrent rise in income inequality in the late 1990s has reduced potential Sen welfare gains by nearly 20 percent. After an improvement in the 1988-1995 period, inequality deteriorated in 1999 to reach levels much higher than those registered in 1988 (Table 3). In contrast with urban income inequality, which rose throughout all periods, rural inequality actually increased more from 1978 to 1988, but then decreased between 1988 and 1995, reaching levels close to those of 1978. However, since then, inequality has reached unprecedented levels for any inequality measure: The Gini coefficient jumped almost 6 percentage points, well above the previous highest level, reached in 1988. Figure 3 depicts the evolution of rural welfare measured by mean household income per capita and the Sen welfare index. The increase in mean income per capita between 1978 and 1999 was 95 percent, but the 32 Section 7 below and Chapter III, Volume II, presents a more thorough explanation of inequality changes via microsimulation of structural parameters and endowment changes in a model of household income generation, labor force participation, and occupational choice. 17 simultaneous increase in inequality resulted in a 19 percent loss in Sen welfare. On a positive note, a more favorable distribution of income in 1995 led to welfare gains 4 percent greater than if inequality had remained the same. However, in the recessive period of the late 1990s, the Sen welfare index fell despite simultaneous real average income gains. 1.4 Beyond average income: Assessing welfare gains from 1978 to 1999 While the previous section revealed the substantial changes---overall improvements and recent reductions---of average income, those patterns do not make clear whether all income groups experienced changes in income in the same direction and in similar proportions. That is, the income dynamics for the entire distribution of household income was not established. This section provides an answer to that question in both urban and rural areas. It reveals that urban Colombians are better off now than in 1978, but not better than in 1988. Nonetheless, it shows that in rural areas recession losses were less severe, despite a comparable welfare trend during the 1980's. Urban we/fare: Unambigwous impropement up to 1995 a7d unambiguous deterzorain duri;g the recession Until 1995, urban welfare levels improved substantially and unambiguously. As shown in figure 4, welfare levels in urban Colombia improved continuously and unambiguously for every percentile of the population between 1978 and 1995. This figure depicts the Cumulative Density Function of income -CDF-, the fraction of the population with income per capita below different thresholds -50, 150 or 250 thousand Colombian pesos. The proportion of the population below the 1999 poverty line (at C$175 thousand) decreased from more than 70 percent to approximately 50 percent during that interval (figure 6).3 During the 1978-88 period, improvements in earnings reached 47 percent for nearly all percentiles, except at both extremes of the distribution, where the improvements were slightly higher. In the next period, 1988-95, earnings kept increasing by 17 percent for nearly the entire population. As a result, between 1978 and 1995, the 5'h and 10th lowest percentiles increased their income increase by 75 percent, while the 90t and 95t0 experienced 75 and 91 percent increases, respectively. Figure 4. Cumulative income distribution, Urban Colombia 1978, 1988, and 1995 100% 2 1978 7 90% .980%- 70% 1995 o6 60%- *t50% 40% ' 30% 20% o 10% 1995 Poverty line $166,204 0% 1 I I ; I 0 50 100 150 200 250 300 350 Per capita monthly household Income, 1999 thousands pesos 33 The 1999 PPP adjusted exchange rate is US$1 = C$629 18 Figure 5. Cumulative income distribution, Urban Colombia, 1988, 1995, and 1999 90% o 80% 0 i 70% 5 60%- 50%- C t 40%I 0 5 30%1 18-5 20%a E 40% O 1999 Poverty line $175,831 0% 0% - . . .. . . . 0 50 100 150 200 250 300 350 Per capita monthly household income, 1999 thousands pesos Figure 6. Income changes by percentile: Urban Colombia, 19788, 1988-95, and 1995-99 L-1978-1988 i1 -19 8-1995 --1995-1 999 40%- 20% __ 0%- Vo 10%/ 20%/ 3016%. 0 0 -20%- -40%- Income percentile In 1999 urban welfare deteriorated with respect to 1995, but the comparison is not straight forward. Figure 5 shows the CDF for 1988, 1995, and 1999. The cumulative distribution of income for 1999 lies to the left of the curve for 1995 for all monthly per capita household incomes below C$350 thousand. That is, income fell for all percentiles, except those between the 80h and the 95th (see figure 6). The poor lost more than the rich: the 5h and 1Ot percentile both lost 25 percent of income, while the median income decreased by 11 percent. By contrast, the 90h income percentile rose by 9 percent, and the 95t gained 1 percent. Since income did not fell for all, immediate conclusions about social welfare deterioration are precluded.4 However, introducing moderate value judgments to assign relative social weights across income groups 3 That is, "first-order dominance" does not hold 19 allows to draw clear conclusions about the evolution of social welfare between 1995 and 1999. In fact, if the poor are consider to be equally or more deserving than the rich, the losses of the former become dominant and social welfare unambiguously deteriorates between 1995 and 1999.11 Only pro-poor welfare measures show a deterioration from 1988 to 1999. Comparisons between 1988 and 1999 are relatively more complicated because income fell for all Colombians below the 45h percentile, but increased for the rest (see figure 5). This implies that social welfare measures with substantial weights on the poor -such as the poverty measures- would lead to the conclusion that 1988 is better than 1999. Although the poverty count were similar in both periods (55.0 percent in 1988 and 55.4 percent in 1999), both the poverty gap (at 26 percent) and the poverty intensity index P2 (at 15.4 percent) worsen in 1999. On the other hand, social welfare measures with less pro-poor distributional weights tend to favor the 1999 distribution when compared to 1988. For instance, average income per capita remains higher in 1999 at C$277 thousand (versus C$235 thousand in 1988), and even when the Sen welfare index's correction for inequality is undertaken.36 In summary, urban Colombians are better off now than in 1978, but not better than in 1988. After continuous and significant reduction of poverty from 1978 to 1995, the economic recession pushed poverty indicators back to 1988 levels. The extreme poor suffered the brunt, having had a good share of the substantial gains of the 1978-95 period erased and having underwent a deepening in the intensity of poverty in 1999.37 Raral wefare. less severe recesshn losses Rural welfare behaved similarly to that in urban areas, yet recession losses were less severe. Rural welfare is unambiguously better in 1988, 1995, and 1999 than in 1978. As shown by the cumulative distribution functions in figure 7, relative to 1978, the other three years studied, 1988, 1995, and 1999, reveal unambiguously large gains in welfare with first-order stochastic dominance. The percentage of people under the 1999 poverty line (C$125 thousand per month) decreased from 95 percent to 79 percent in the 21-year period studied. Figure 8 shows even larger for those well below the 1999 poverty line: the proportion of people below the 1999 extreme poverty line (C$49 thousand per month) fell from 65 to 37 percent during the last two decades. Welfare improved for great majority of the rural population-except for the top 7 percent-from 1988 to 1995. The 1995 cumulative distribution of income is always to right of 1988, except for after the 97h percentile, where the two lines cross (figure 8). Interestingly, as it was suspected above, those at the bottom of the distribution benefited the most during this period. Hence, if the poor are at least as deserving as the non-poor, 1995's welfare is above 1988's.38 Most of the rural population is worse off in 1999 than in 1995. The cumulative distribution function for 1995 and 1999 (figure 7) reveals that welfare turned for the worse for the bottom 70 percent of the population during the recessive period, while for the top 30 percent, it made a slight gain. The proportion whose household income per capita put them below the 1999 extreme poverty line increased from 34 to 37 percent in 1995 and 1999, respectively. The fact that most of the population experienced a decrease in welfare in the 1995-99 period despite an increase in mean household income is also revealed by a drop in the Sen welfare index, as shown in Figure 3. 3 That is, if the marginal utility of income is decreasing "second-order stochastic dominance" does hold. This is shown companng the generalized Lorenz curves of those two years (see chapter 1, volume II). For a full intertemporal comparison considering alternative welfare weights see Table A. .& in the Append 3 For a full intertemporal companson considering alternative welfare measures see Table A. 1.2. 38 Generalized Lorenz curves for 1995 and 1988 do show 1995 as having generalized Lorenz domunance over 1988. See Volume 11, Chapter 2. 20 Figure 7. Cumulative income functions, Rural Colombia, 1978, 1988, 1995, and 1999 100%i o 90% 7 80% 0 /1988,_________0'__ C. 70%- O 50% 0% > 30%- 9 a 20% 1999 Extreme E 10% povertyline 1999Poverty M 0 $49,901 Ine. $125497 0 Per c4ita mdA1y holg9hold me, 49 thSOPands350 pesos Figure 8. Income changes by percentile: Rural Colombia, 1988-95, 1995-99. 35% 25% -1988-1995 -1995-1999 15% Co 5%- -5%. 1M% 20r'A 3M. 4 60% 701% 80 90% a. -15% -25% Income percentile Nevertheless, in 1999 rural welfare was above the 1988 level for 95 percent of the population. Relative to 1988, however, welfare did increase for Colombians in rural areas in 1999, with the exception of those at the bottom 5 percent of the income distribution. Therefore, relative to urban Colombia, rural areas sustained a less severe deterioration of welfare during the 1990s.39 In summary, it is unambiguously true, by any welfare measure, that, relative to 1978, rural Colombians were better off in the other three years studied. But whether welfare underwent a relative improvement within the 1988-99 period is dependent upon the social welfare indicator chosen. That is, 1995 is clearly better than 1988 in terms of all poverty measures, income per capita, and the Sen welfare index. However, compared to 1988, 1995 is not better for every 3 We saw above that nearly the 50 percent urban poorest were worse off in 1999 than in 1988. 21 income group. Nevertheless, 1995 is better if one considers the poor more deserving than the non- poor. In 1999, welfare in terms of income per capita expanded, relative to 1988 and 1995. However, 1988 is favored relative to 1999 in terms of both the poverty gap and intensity measure. Furthermore, 1995 is favored relative to 1999 in terms of all poverty measures and the Sen welfare index. Thus, although mean income per capita may have increased from 1995 to 1999, by most welfare measures, rural Colombians were not better off in 1999 relative to 1995. 1.5 Colombia is more capable of tackling poverty today than it was two decades ago a Does Colombia have the resources necessary to tackle poverty and extreme poverty? How large is the income gap of the poor and extreme poor relative to the size of the economy, and to the resources in the hands of the public sector? The evidence demonstrates that relative to the size of the economy, closing these gaps is not beyond the means of the Colombian economy. Figure 9 shows that in 1999 eliminating extreme poverty would require less than 31 percent of total household income. Moreover, closing the poverty gap-that is, getting 55 percent of Colombians out of poverty-would require 25 percent of total household income. Unlike two decades ago, today Colombia is a great deal more capable of tackling poverty. The growth of the economy up until 1995 helped to reduce the magnitude of both the poverty and extreme poverty gaps. From 1978 and up to 1995, both the economic growth and the reduction in the poverty gap from 46 percent to 29 percent reduced the share of total household income necessary to close the poverty gap from 58 percent to 20 percent, just a third of its original value. However, the increase in the poverty gap in 1999 to 34 percent increased the required share to 25 percent. The share of household income necessary to close the extreme poverty gap followed a similar trend. From 1978 to 1995 it was reduced from 11 to 2 percent. The recent setback slightly increased the required resources to 2.5 percent. Undoubtedly, Colombia has enough resources to tackle poverty. Figure 9. Share of total household income required to close poverty and extreme poverty gaps 60% 12% | F--Poverty =-O= Extreme Povery 50% 10% 10% 2% 0 40% -8-/. ( ) 0 T 30o % E T 20o e 4% 25 0 8 10% rl 20 1978 1988 1995 1999 Share 1978 1988 1995 1999 To close the Extreme Poverty Gap 10 .7 % 4.3% 2.1% 2.5% To close the Povert Gap 58% 25% 20% 25% 4 This same issue was raised by Selowski (1979) in his seminal study about poverty and public expenditure in Colombia. 22 Today the Colombian public sector has more than enough resources to tackle the poverty problem without abandoning other core mandates. In 1999, public social expenditure was 15 percent of. GDP, while total public expenditure was approximately three times larger. In summary, this evidence shows that the reallocation of 10 percent of public expenditure-a moderate amount-would have a major impact on the welfare of the poor. 2. SOCIAL SERVICES AND SOCIAL OUTCOMES: COVERAGE AND EQUITY 2.1 Education: Increasing trend with pro-cyclical fluctuation in urban areas &/-an Schooling: Acreas.ng trend wfhpro-cyclialfluctuatins In the last two decades, average educational attainment increased by 2.7 years and the illiteracy rate was nearly halved. Illiteracy rates were halved between 1978 and 1999 in urban Colombia, from 11 to 6 percent. For individuals over 12 years of age, illiteracy rates fell from 5 to 2 percent during the same period. The average educational attainment of the adult population increased from 6.2 to 8.9 years over the entire period. School enrollment: Slow progress partly reversed during economic recession. School enrollment is an educational indicator closely related to the economic cycle. Table 4 reports enrollments for three different age groups of boys and girls (7-11 years, 12-17, and 18-22) and captures primary, secondary, and higher education enrollments. For all three age categories these rates grew very slowly between 1978 and 1995 but decreased in the latest four years. The recent drop in the university-age population's enrollment is especially worrisome: In 1978, 31 percent of 18- to22-year-olds were enrolled in school; this rate peaked at 41 percent in 1995 but and then declined to 36 percent in 1999. The 64 percent of young adults not enrolled in school belong mostly to the economically active population and face rates of unemployment twice as high as the rest of the adult population, a fact that reveals their diminishing chances of avoiding poverty in the long run. Completion rates improved more rapidly for primary and secondary schooling. The share of individuals within the intermediate age group-12 to 17-who have completed primary school rose from 67 to 90 percent between 1978 and 1999, while among the older group-18 to 22-the proportion of high school graduates increased from 21 to 59 percent. During the 1990s, coverage became more progressive for secondary and tertiary education. Coverage rates in primar , secondary, and tertiary education improved for all income quintiles between 1992 and 1997. However, the poor benefited much more from additional coverage in the case of secondary education and were only benefited moderately in tertiary. The concentration coefficient was minus 0.563 (!) for secondary education and 0.403 for tertiary (still slightly better than the previous 0.445). 41 Increased by 8, 12, and 6 percent respectively. See Chapter VI, tables 1 8 and 1.9. 23 Table 4. Social indicators: Urban Colombia 1978-99 1978 1988 1995 1999 Average education > 18 years 6.2 7.7 8.4 8.9 Illiteracy rate2 5.3% 3.3% 2.8% 2.6% School enrollment Ages 7 to 11 91.8% 94.8% 96.5% 95.3% Ages 12 to 17 76.9% 80.5% 84.4% 82.2% Ages 18 to 22 31.2% 35.8% 41.0% 36.3% Complete primary school (ages 12 to 17) 67.0% 78.7% 77.7% 89.8% Complete high school (ages 18 to 22) 21.6% 35.3% 48.7% 59.2% Child labor Ages 12 to 16 12.0% 11.5% 9.9% 9.5% Ages 12 to 14 5.8% 5.0% 5.2% 3.7% Child Malnutrition3 Stunting, low height for age 16.8% 10.1% 8.4% 6.7% Wasting, low weight for height 22.4% 16.6% 15.0% 13.5% Low weight for age 4.9% 2.9% 1.4% 0.8% Crime4 Homicides (Per 100,000 pop.) 26 62 65 59* Access to public utilities Electricity 63%** 99% 100% 99% Water 63% 97% 98% 99% Telephone NA 62% 71% 84% Sewerage 51% 95% 96% 97% 1. Urban Colombia represents 67% of Colombia urban area: Barranquilla, Bucaramanga, Bogota, Manizales, Medellin, Cali and Pasto 2 For population 12 years old & older. 3. For population under 5 years old; represents national data for 1977, 1986, 1995, and 2000 4. Levitt and Rubio, 2000. * 1998 figure **1974 figure. Source: Selowsky (1979) Sources: Authors' calculations based on Encuesta Nacional de Hogares; Profamilia, and Encuesta Nacional de Demograffa y Salud Rura/Schoo/ig Average school attainment for adults more than doubled, from 2.2 years in 1978 to 4.6 in 1999, but the urban-rural gap remains unchanged (Table 5). Tantamount to improvements in illiteracy rates, most progress in adult average education levels was made between 1978 and 1988, when adults gained an average of 0.16 years of schooling per year. Progress since 1988 has remained stable, with an average increase of 0.08 years of schooling per year. Throughout the period studied, there has been a constant gap of a little more than four years between rural and urban education. 24 Table 5. Social indicators, Rural Colombia, 1978-99 1978 1988 1995 1999 Illiteracy rate* 29.4% 18.5% 14.5% 14.8% Average education > 18 years 2.18 3.74 4.31 4.63 School enrollment Ages 7 to 11 66.2% 85.4% 90.1% 90.5% Ages 12 to 17 43.5% 57.2% 63.7% 66.0% Ages 18 to 22 9.0% 14.6% 19.2% 20.6% Complete primary school (ages 12 to 17) 20.7% 49.3% 62.7% 66.8% Complete high school (ages 18 to 22) 1.3% 9.1% 14.2% 21.1% Child labor Ages 10 to 16 25.5% 27.9% 22.3% 19.3% Ages 10 to 14 19.1% 23.1% 15.6% 13.4% Access to public utilities Electricity NA NA 87% 84% Aqueduct NA NA 64% 62% Telephone NA NA NA 15% Sewerage NA NA 32% 32% * For population 12 years old & older. In contrast with urban areas, during the recessive period, rural school enrollment rates remained constant, if not rose, for all age groups. Despite the general slowdown in improvement rates, considering that urban areas suffered drops in enrollment rates for all age levels from 1995 to 1999, the economic recession does not appear to have affected rural schooling as severely. Nevertheless, all schooling rates continue to lag behind those in urban areas. Enrollment rates for children of primary and secondary school age registered their,most significant gains between 1978 and 1988, followed by moderate gains between 1988 and 1995 and only a slight improvement in the 1995-99 period. The 1999 primary school enrollment rate-90 percent-is now only 5 percentage points lower than its urban counterpart, an impressive feat, especially when compared to the 1978 urban-rural gap of 26 percentage points. Children of secondary school age are also increasingly attending school: 66 percent in 1999 versus 44 percent in 1978, and the urban-rural gap was halved from 32 to 16 percentage points in 1978 and 1999, respectively. The overall improvement in enrollment rates for the 18- to--22-year-old age group is notable, as it has doubled in the past 21 years, up from 9 percent in 1978 (versus 31 urban) to 21 percent in 1999 (versus 36.3 urban). This enrollment rate progressed at similar rates between 1978 and 1988 and 1995, but then stagnated during the recessive period. In contrast, the urban rate suffered a decline of almost 5 percentage points between 1995 and 1999. 25 Increarthy frustrathn and lneqAuly at the entrance offpost-secondary educatnow In the last 25 years, Colombian school-age cohorts have persistently reached higher educational attainment with less inequality. Figure 10 shows that, on average, the number of years of education that the 1975 cohort attained by 1995 was approximately 10 years, 4 years greater than the one achieved by the cohort born 40 years earlier. Furthermore, the inequality of education within each cohort-measured by the coefficient of variation-fell by more than half during the same period.42 Therefore, the inequality index of the stock of education among income groups tends to fall over the years. According to our calculations, the inequality of the stock of education was 18.1 in 1978, 18.6 in 1988, and 16.4 in 1995.43 The substantial improvements in equity and efficiency in basic and secondary education in the 1990s are creating an inequitable bottleneck at the gate of post-secondary education. Nevertheless, progress has not been homogeneous in all stages of the educational system. As Figure 11 shows, despite of the significant improvements in equity and efficiency in completion of primary and secondary school, there is an increasingly binding bottleneck at the entrance to post-secondary education, with mounting frustration for the lower-middle class. Figure 11 shows that in 1988, for the 20 to 25 age cohort, 99 percent entered primary school, and 92 percent finished it. Meanwhile, 81 percent entered secondary school but only 44 percent finished it. Barely 19 out of 100 of those in the 20 to 25 age cohort in 1988 entered post-secondary education. Since then, the educational system as a whole has become more efficient in every stage of the educational ladder (the points for the 1999 the curve represent a shift up). Promotional and completion rates have increased for both primary and secondary school between 1988 and 1999: from 92 to 95 percent in completion of primary and from 44 to 63 percent in completion of high school. The transition rate from primary to secondary schooling slightly improved from 8 to 9 percent in 1988 to from 9 to 10 percent in 1999. On another note, the slope of the "blue" curve shows that access to higher levels of the educational ladder is more unequal; that is, in each for higher levels, the proportion of poor students entering the higher grades is lower than in the previous stage. Nevertheless, the shift to the left from 1988 to 1999-from the blue to the red curve-reveals substantial improvements in equity for all grades up to high school; the only exception is the transition between high school and post-secondary education, college or technological. Figure 11 shows that from 1988 to 1999, both the promotion and equity gaps have widened: the promotion gap from 24 to 35 percent and the equity gap from 16 to 25 percent. In other words, the gap between secondary school and university is becoming increasingly regressive. In summary, in the last decade, Colombia enjoyed clear improvements in educational attainment, both in level and equity. However, there is an increasingly frustrating and inequitable queue at the entrance of post-secondary education. 42 From 0.75 to 0.33 43 As measured by the concentration coefficient, similar to the Gini coefficient. 26 Figure 10. Average education and inequality by cohort: Colombia, 1925-75 Average education and Inequality by cohort 11 07 10 -- 3P . 0 80 8 50% from24%%o 3 ~. ~. . ~b ,nishnisecie o198 ~M ~~ 40 20%3 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 Inequality of cohort attainment (Concentration Coefficient of the promoted subset) 27 Box 2. Poverty comparisons across Colombian Cities, 1978-95: Partial convergence and improvement in left behind cities, especially Barranquilla- M There are two policy angles that justify a close examination of cross regional differences in Colombia. First, after local democracy was established in the 1980s and increasing decentralization advanced during the 1990s, more social public policy decisions are being taken and executed at the local level. Second, regional sources of growth have been identified as key determinants of overall economic dynamics, inducing differential poverty dynamics across regions (Sanchez and Nifiez, 2000a). Over the years, urban areas showed strong convergence in extreme poverty counts. The maximum to minimum difference in poverty rates dropped from almost 30 percentage points in 1978 to only 8 points in 1999. From 1978 to 1995, significant poverty reduction was visible across all cities; however, economic recession in the late 1990s reversed that tendency, except in Barranquilla. Somewhat similar is the evolution of income per capita. Across cities, there is a pattern of convergence in average income per capita, with the exception of Bogoti. Once income is adjusted by differentials in local cost of living, convergence is even stronger and the relative advantage of BogotA drops from 40 percent to only 14 percent. Throughout the entire period, income inequality rose sharply in most cities. The worst rise occurred in Barranquilla and BogotA -- 14 and 10 percentage Gini points, respectively. However, during the 1990s, the dynamics of inequality across urban centers were more diverse. While inequality mostly increased in the four largest cities-particularly in BogotA, by 8 Gini points-it decreased in the middle-sized cities of Bucaramanga, Pasto, and Manizales. Figure B2.1 Poverty and growth across urban In the last decade, the strong heterogeneity of centers in Colombia, 1988-99 extreme poverty dynamics is explained by the strong variability in growth and inequality across -14% the main urban centers. While Barranquilla, -12% - Manizales, and, Pasto saw reductions of 12, 13, and t5 - _8 percent, respectively, BogotA saw a moderate increase and the rest only saw reductions between 2 -D8% oand 5 percent. Figure B2.1 shows that BogotA and 6% Cali-the cities where poverty reduction was 0C disappointingly low relative to growth-are those a 0~ Eo Redam 0 Cal that had a rise in inequality strong enough to e a2% counterbalance the effect of growth on poverty 0% .... . 0 dynamics. The contrast of Barranquilla and BogotA 2% is especially interesting. Despite increasing -10% 00/0 10% 20% 30% inequality, income growth brought significant Growth (income per capita) benefits for the extreme poor in Barranquilla --a 13 percent reduction in their headcount. But the opposite happened in BogotA, where extreme poverty remained almost unchanged despite having the second highest income growth--23 percent-- clear evidence that economic growth in Barranquilla was more pro-poor than it was in BogotA. Cross-city comparisons of social indicators also reveal a pattern of convergence, with strong catch up by the laggard cities since 1978, particularly Barranquilla, on the Caribbean coast. During the last two decades social indicators converged in basic education, electricity, and water but, heterogeneity persisted in post-secondary education, child labor, telephone connections, and sewerage. In spite of the economic prosperity enjoyed during the past decade, Barranquilla is still behind in some basic infrastructure services: 17 percent of the population does not have sewerage and 51 percent does not have a telephone connection. As to which Colombian city enjoys the highest welfare as of 1999, the answer depends on the choice of social weights: by average income it is BogotA, by poverty count it is Pasto, and by extreme poverty it is Bucaramanga. In dynamic terms, Barranquilla showed the highest progress-in all development indicators- while Cali faced the most significant losses. 28 Box 3. Divergence and poverty across rural regions in Colombia, 1978-99: Oriental and Atlantic regions win, Pacific and Central regions lose Extreme poverty remains quite heterogeneous across rural areas, with the most severe problems concentrated in the Central and Pacific regions. The Oriental and Atlantic regions show substantial gains during the last two decades. Something similar might be said about poverty dynamics in the 1990s. Contrary to in the 1980s when extreme poverty fell across all rural regions, during the 1990s extreme poverty reduction came to a halt in the Central region and it was reversed in the Pacific region -minus 3 percent-. Nevertheless the Atlantic and Central regions enjoyed sharp reductions in extreme poverty --24 and 23 percent, respectively- . Figure B3.1 Extreme poverty reduction and growth Growth differentials seem to be the dominant across rural regions in Colombia. 1988-1999 explanatory factor in all four regions (Figure B3.1) --positive and significant for the Atlantic and -30% Oriental regions, where vigorous growth continued SAotan d during the late 1990s and negative in the other two oQnel * regions, specially for the Pacific. Nevertheless, 0% -: increasing income inequality in the Central o_ region-10 percentage points-also a contributed negatively to extreme poverty dynamics. In the 0% 0/ Atlantic region the deterioration of income inequality was clearly dominated by growth. E. 0% Similarly to the main urban center in the same .a .kregion-Barranquilla-the more pro-poor quality L ' of growth in the Atlantic region dominated the 10% adverse effect of increasing inequality. Social -40% -20% 0% 20% 40% 60% indicators also reveal deviating paths of Growth (income per capita) development, and the Atlantic region advances the most. Primary school enrollments converged during the last two decades, with the largest increase in the Pacific region. During the recent economic downturn, enrollment dropped in both the Pacific and Oriental region but remained steady in the Atlantic and rose 4 percentage points in the Central region. Secondary school enrollment rates also converged, with the exception of the Atlantic region, which always enjoyed considerably higher rates, around 12 percentage points higher than the rural average. The education system in the Atlantic region thus seems somehow better able to retain students after they reach secondary school age. Something similar is happening with child labor, which is lowest in that area: only half the rate found in the other three regions. Nevertheless, during the late 1990s, all regions, except for the Pacific, experienced large drops in child labor, of about 6 percentage points. * see detailed information in Appendix A.2. 2.2 Child labor: pro-cyclical with a decreasing trend Urban child labor tends to fall slowly and seems pro-cyclical. Concurrent with improvements in school enrollment for adolescents, urban child labor shows a decreasing trend during the last two decades. The proportion of children aged 12-16 years who work decreased from 12 percent in 1978 to 10 percent in 1999, a low 2 percentage points over 21 years. Child labor is sensitive to variations in the level of economic activity. As in other Latin American countries, the proportion of working children in urban Colombia seems to fall during recessive periods of the economic cycle."4 The rationale is that the opportunity cost of sending children to school is higher during economic booms. Although income shortage becomes higher during economic recession, 4 Cunningham and Maloney (2000) 29 incentives for sending children to school may increase as opportunity costs decrease. For example, children in school receive a guaranteed meal per day along with daycare, allowing parents to work or take care of home activities. In addition, the benefit of the alternative choice falls with the declining probability of children finding jobs during recessions, a fact confirmed below when unemployment rates by age groups are presented. Defined in stricter terms, as a proportion of children from 12 to 14 years of age participating in the labor market, child labor remained practically unchanged from 1978 to 1995 at around 5 to 6 percent, and fell abruptly with the 1999 recession to 4 percent. Just as in our urban findings, rural child labor rates peaked during the 1988 boom and have been falling since. However, in contrast with urban areas, the decrease was most substantial between 1988 and 1995, and then slightly less so between 1995 and 1999. The 1988 boom was largely driven by the exceptionally high coffee prices and the good weather, which led to an increase in the demand for agricultural labor. Thereafter, 1995 saw the beginning of the construction crisis. Considering the large rise in adult unemployment that occurred during the economic downturn, it is surprising that child labor participation rates have not declined to a larger extent. Perhaps the fact, as will be seen later, that rural regions have not experienced a decline in mean household income per capita-except for in the Pacific region-may explain why the recent fall has only been in the magnitude of around 3 percentage points. Despite their decline, rural child labor levels remain very high when compared to levels in urban areas. Regardless of the 30 percent drop in child labor between 1988 and 1999, child labor rates remain well above those in urban areas, at 19 percent for those 10-16 years of age, versus 10 percent in urban areas, and 13 percent of the 10-14 year-old group employed versus 4 percent in urban areas.45 About 31 percent of children 15-16 participated in the labor force in 1999, a rate just under half that of adults 17 to 65 years old (65 percent). 2.3 Access to basic infrastructure: progressive gains Urban coverage of public utilities continues to improve. 4 Table 4 reports access of the population to different public utilities (water, electricity, telephone, and sewerage), as an alternative measure of poverty, focusing on indicators of basic needs. Since coverage for electricity and water reaches 99 percent, only small pockets of poverty-approximately 150,000 people- have no access to these basic services. Nevertheless, some alternative evidence suggests that such indicators for water coverage may be highly biased upward, in the sense that water services are actually highly heterogeneous in both continuity and quality.47 Sewerage was not available to 5.2 percent of the population in 1988, a proportion nearly halved in 1999 to 3 percent. While almost 38 percent of the urban population did not have access to a telephone in their homes in 1988, this figure was reduced to 16 percent in 1999. The growth in coverage for public utilities was progressive, although income disparities in telephone access are still a cause for concern. Progressive marginal connections to public utilities, between 1988 and 1999, helped the lower-income quintiles to catch up with the middle- and high-income groups. For example, sewerage coverage increased by 7.6 and 3.4 percentage points for the two poorest quintiles of the household income distribution, compared to just 0.3 percentage points for the two richest ones. The improvement in access to telephone was 45 Child labor participation rates in urban areas are for children 12-14 and 12-16; thus, the urban-rural difference in child labor participation rates may be even larger. 4 A more detailed account of the evolution and incidence of access to public utilities is given in Chapter 5, Volume II, The Distnbutional Impact of Public Expenditure. 47 In fact, evidence from the DNP suggests that aqueduct service in Colombian urban areas (defined as cabeceras municipales) is highly inadequate: Over 60 percent of the urban population is believed to be exposed to health risks due to substandard drinking water, while the piped water supply is also often highly discontinuous (World Bank, 2000). 30 particularly spectacular for poorer households, with increases of more than 30 percent for the two lowest quintiles. Access to safe water grew by 5 percentage points for the first quintile, which is reaching full coverage. The ratio of coverage of the richest quintile to the poorest thus fell from 1.07 to 1.01 for safe water, 1.15 to 1.06 for sewerage, and 2.55 to 1.40 for telephone. Public utility coverage in rural towns is comparable to that in large urban centers; however, notable disparities remain in sparsely populated areas. Economies of scale in the provision of basic public utilities in urban concentrations facilitate greater coverage with comparable unit costs (compare Tables 4 and 5). In contrast with urban areas, service coverage decreased slightly from 1995 to 1999. Coverage rates for electricity, aqueduct, and sewerage decreased between the two years for which data are available. However, in cabeceras municipales rurales-rural towns--electricity and aqueduct coverage rates-both at 97 percent-are comparable to those of urban areas, whereas the corresponding figures in sparsely populated areas are only a small fraction, 19 and 49 percent, respectively. 2.4 Child malnutrition and infant mortality: continuous improvement Malnutrition was more than halved between 1977 and 2000. Data used come from 1977 and 1986 surveys by the National Institutes of Health as well as the 1995 and 2000 Demography and Health Surveys (DHS) for Colombia. The proportion of children under five years of age who are underweight fell from 17 to 7 percent. This still represents around 120,000 malnourished children, a high number considering that the effects of malnutrition will cause irreversible welfare losses to linger throughout their lives.48 The proportion of children displaying stunting (low height for age), an indicator of chronic malnutrition, remained at 14 percent in 2000, three-fifths of the 1977 proportion, a more moderate improvement. The proportion of children facing acute malnutrition (low weight for height) dropped from 5 to I percent between 1977 and 2000. These rates remain below the Latin America and Caribbean region average, conditional on income. Table 6. National infant mortality (under 1 year of age, rates per 1000 born) Source 1981-82 1983-84 1986-87 1989-90 1990-95 1995-2000 1993 census 44 41 37 37 34 NA DHS survey NA NA NA NA 28 21 Source: Profamilia and Encuesta Nacional de Demograffa y Salud Infant mortality has been falling steadily over the last 20 years. As shown in Table 6, rates decreased from 44 per 1,000 in 1981 to 34 in 1990-9549 and fell afterward. Data from the 1995 and 2000 DHS surveys confirm the continued decline, with another decrease of 7 per 1,000. In summary, during the last two decades, Colombia achieved clear social development gains and showed some resilience during the recent recession. In education, completion rates for primary and secondary schooling showed substantial improvement, reaching 90 and 59 percent for teenagers and young adults, respectively. However, school enrollment rates progressed somewhat slowly and suffered a small reversal during the recent economic downturn. Illiteracy rates had a similar evolution. In health, life expectancy increased 20 years during the last four decades, with more than proportional gains for females. Equally important, child malnutrition and infant mortality were significantly reduced. Despite being pro-cyclical-as in other Latin 4 Malnourished children are less likely to perform well in school and more likely to become disabled and/or economically dependent and consequently become an economic liability for their families and society as a whole. 49 Estimation from 1993 census (Departamento*Nacional de Planeaci6n). 31 American countries-child labor exhibited a decreasing trend during times of fervent economic growth; nevertheless, it did decline even faster during the recent recession. Finally, basic infrastructure showed progressive gains. Most of the growth in coverage of water, sewerage, electricity, and residential telephone connections results from catch up by cities least covered in 1978 and the progressive extension of coverage to the poorer deciles. Sewerage and telephone connections still lag behind. The missing percentages mask thousands of individuals, concentrated in regional pockets of poverty, where these basic needs are not met yet. 3. THE BURDEN OF VIOLENCE AND CRIME 3.1 Violence: its continuous aggravation since the 1970s is mostly associated with the illegal drug trade After almost tripling from 1970 to 1991, homicide rates decreased in the 1990s, while extortion, kidnapping, car theft, and armed robbery rates kept growing.50 Homicide rates almost tripled during the 1970s and 1980s; After peaking in 1991, when almost one in a thousand Colombians was murdered, homicide rates fell by more than 20 percent between 1992 and 2000 (Figure 12). Notwithstanding the recent decline, rates remain three times as high as those in Brazil or Mexico, and, worldwide, are surpassed only by homicide rates in El Salvador. While almost all of the decline can be attributed to a decrease in homicide rates in Bogota, Medellin, and Cali, homicide rates in Medellin are still more than double the national average. Despite a decline in the overall homicide rate, violence spread over the country between 1990 and 1997, in the sense that homicide rates decreased in the most violent areas, yet increased in areas considered to be less dangerous. In contrast with homicide rates, mitigation, extortion, car theft, and armed robbery rates kept growing during the 1990s, leading Colombia to display a remarkable rate of victimization, at more than 35 percent of households in 1997. Nevertheless, Colombia still has one of highest homicide rates in the world, but ranks twelfth in victimization out of 18 Latin American countries for which the information is available. According to the best empirical evidence available, drug trade and the simultaneous presence of illegal armed groups seems to be the main explanation behind the staggering homicide rates prevalent in Colombia today. Multiple empirical studies have tried to explain the determinants of homicide and crime in Colombia.5' Five alternative hypotheses have been tested: (a) the illegal drug trade, (b) impunity, (c) the presence of extra-governmental groups (guerrillas and paramilitaries) that have taken over traditional governmental roles in parts of the country, (d) poverty/income inequality, and (e) the possibility that Colombia's decades of internal strife have created a population that is innately more "violence-prone."52 Sanchez and Nifiez (2000) performed a comprehensive test of the alternative hypothesis, with the most complete data set available for crime across different municipalities in Colombia.3 Their main findings indicate that socioeconomic variables such as inequality, poverty, political exclusion, and lack of education have an adverse effect on the crime rate. However, jointly they explain only 6-12 percent of the total variability in the homicide rate. The rest-nearly 90 percent-is mainly explained by the intensity of illegal drug trade activities and its interaction with the presence of illegal armed groups; that is, guerrillas and paramilitaries. According to Levitt and Rubio (2000), 50 See Levitt. and Rubio (2000). 51 SAnchez and Nifiez (2000b) mention at least 10 studies: Comsi6n de Estudio de la violencia (1987, Montenegro and Posada (1995), Gavina (1998), Echeverri and Partow (1998), Echandia (1999), Sarmiento (1999), Moser (1999), L6pez and Garcia (1999), and Rubio (1997, 1999). 52 Levitt and Rubio (2000). 53 As mentioned above, vanability of homicide rates across municipalities is very high. 32 these findings are consistent with international evidence that indicates a positive covariance between homicide rates and periods of intense illegal drug trade. Figure 12. Colombian homicide rate, 1962-98 80- 60 40 20 - 0_ 1960 1965 1970 1975 1980 1985 1990 1995 2000 Source: Colombian National Police. Violence generates increasing social costs and demands substantial public resources. Apart from the 30,000 lives lost per year in the war against illicit drugs, Colombia has sacrificed not only police and military officers but also several presidential candidates, politicians, intellectuals, and journalists.54 Estimates of the human capital loss due to homicide are at least 1 percent of GDP.55 Moreover, investment and educational and labor market opportunities become severely limited in an atmosphere of violence and insecurity. At the same time, the demand for public resources to fight crime has escalated. Public expenditure in justice and security has more than doubled its share of GDP during the last decade. While current public expenditure in security and justice is around 5 percent, in 1990 it was close to only 2 percent of GDP. Simultaneously, private expenditure in security seems to be increasing even more rapidly: The ratio of policemen to private guards decreased from 2.5 in 1980 to only 1.0 in 1995.57 Additionally, the World Health Organization estimates that violence's effects on health care costs in Colombia are in the magnitude of 5 percent of its GDP, an undoubtedly alarming proportion. s4 Approximately 600 lives per year are lost in massacres, half of them in the Department of Antioquia. DNP-UNDP (1999). 5 Capital-cost estimates are based on decreases in disability-adjusted life years (DALYs), Trujillo and Badel (1998) and Londoflo (1996) quoted by Levitt and Rubio (2000) Additional costs in terms of medical assistance are provided byBonadilla etal. (1995). 5,See section II in Levitt and Rubio (2000) for detailed estimates. 5 Comisi6n de Racionalizaci6n del Gasto y de las Finanzas Pablicas (1996) quoted by Levitt and Rubio (2000). 5 World Report on Violence and Health, World Health Organization, 2002. 33 3.2 Life expectancy: persistent improvements with gender bias associated with violence Life expectancy has increased by nearly 21 years during the last four decades, with a strong advantage for women and considerable regional differences. As shown in Figure 13, women have benefited most from these gains: The gender gap was 3.4 years in the 1950s, started widening to 6.6 in the late 1970s, and rose to 8.3 in 1995 before returning to 6.3 at present. During the last decade, the gender gap varied widely between departamentos, with the maximum in Antioquia (14.2 years, 162 percent the national average) and the minimum in Nariflo (4.4 years). Regional comparisons across departamentos show that life expectancy is much higher in the Atlantic region.59 Apparently, longer life expectancy seems more prevalent in the least prosperous areas. Finally, the departments with higher life expectancy also reveal lower gender gaps. Most of the gender difference is associated with the extremely high risk of being murdered among the young Colombian males. Colombian men between 15 and 35 years of age are 15 times more likely to be homicide victims than women of the same cohort and twice as likely to be victims than men over 45 years of age (see Figure 14). Figure 13. Life expectancy by gender, Figure 14. Homicide rates by age and gender, Colombia, 1950-2000 Colombia, 1999 9D .!20- U.22M - 50 Im z 2 15 0~ D 15 - F44 .75- 3.3 Inequality and violence: Who bears the burden of crime? The poor are more likely to be victims of homicide, and the non-poor are more likely to be victims of property crime and kidnapping.a~ Our main results concerning the effect of relative household socioeconomic status on the probability of victimization are as follows.. The probability of victimization is very similar for the first three quintiles, slightly higher, though not significantly, for the fourth quintile, and substantially higher for the top quintile. On average, the richest 20 percent of the population has a probability of victimization of at least 6 percentage points higher than the poorest 60 percent. This result holds up after controlling for household attributes and city fixed effects, though the significance drops somewhat in the latter case.6' However, similarities in the probability of being the victim of a violent crime hide important so With the exception of males in the department of la Guajira. a This subsection is based on Gaviria and Vdlez (2001), Chapter 4, Volume II of the report. 6! V61ez, L.F., et al. (2000) reach a similar conclusion in their in-depth analysis of victimization in the city of Cali. Here, the typical victims of crime are young individuals from the middle and upper social strata. 34 differences in specific kinds of cnmes: the higher incidence of homicides in the lowest quintiles and the higher incidence of kidnappings in the highest. The probabilities of having a household member murdered are 2.4 percent for households in the bottom quintile, 1.2 percent for households in the third quintile, and 0.6 percent for households in the top quintile. For having a household member kidnapped, the corresponding probabilities are 0.0 percent, 0.1 percent, and 2.8 percent. Female-headed households are at least 3 percentage points more likely to have been victims of a violent crime (a difference of more than 100 percent with respect to mean values), which can be explained in part by the fact that one of the main reasons women head these households is that the former (male) head was murdered (i.e., there is reverse causality).62 The intensity of crime concentration among business owners reveals a perverse effect on economic efficiency, i.e., reducing investment and employment in crime-infested areas. Households that conduct a business out of their homes are 10 percentage points more likely to be victimized than households that do not. This finding indicates that property crime is especially taxing on small entrepreneurs, implying that the economic effects of property crime, from slower economic growth to stifled job creation, could be substantial. Surprisingly, households in which the head is unemployed are much more likely to be victimized, which may reflect greater exposure to risk caused by job searching or loitering, the main activities of the unemployed. Regardless of the reason, the fact is that a heightened victimization risk appears to be a hitherto unknown cost of being unemployed, at least during the current escalation of joblessness in Colombia. Figure 15. Socioeconomic status and propensity to hire private guards 60% Propensity to hire_ 40%- pnivate guards 40% Probability of Victimization 20% 0% 1 2 3 4 5 Quintiles Source: Gaviria and V6lez (2001, Volume II). The non-poor feel unsafe and are more likely to modify their behavior and invest in crime prevention accordingly. Households from the top quintile are much more likely-15 to 18 percent-to report feeling unsafe because of fear of crime than are households in extreme 62 Whereas more than 3 percent of all female-headed households had a family member murdered in the six months prior to the survey, less than 1 percent of all male-headed households experienced the same ordeal. Kidnapping rates do not differ much between these two types of households and assault rates are slightly higher in households headed by women. 35 poverty, a result consistent with the higher probability of victimization among the richest households mentioned above (Figure 15). Households from the bottom quintile are at least 10 percentage points less likely to report feeling unsafe, and households from the intermediate quintiles have similar propensities to report the same feeling (with the exception of the second poorest quintile). Migrant households are more likely to feel unsafe in their new city of residence (perhaps because they have to face an unfamiliar and often hostile environment, especially if they are IDP (Internally Displaced People). Female-headed households are also slightly more likely to feel unsafe, though the difference is not significant and is relatively small when compared to the higher propensity of those households to have a member victimized. Households from the top quintile are much more likely to have antitheft devices in their homes, to participate in neighborhood watch programs, and to hire private guards. For this last strategy, the difference between the top and bottom quintiles is almost 50 percentage points, and the difference between the fifth and fourth quintile is more than 20 percentage points. Households from the top quintile are also more likely to avoid road trips than households from any of the other quintiles. The same is true for not going out at night, though the difference in this case is not statistically significant. Domestic violence mostly affects uneducated women. Poor women are much more likely to be victims of domestic violence than women from more advantageous backgrounds. The probability of being a victim of domestic violence increases as much as 16 percentage points as income level moves from the top to the bottom quintile (see Vol. 2, Chapter 4). But the effect of relative socioeconomic status shrinks substantially and loses its significance completely after controlling for both the years of education of the potential victim as well as those of the household head, suggesting that poor education, rather than lower socioeconomic status, appears to be the main risk factor of being a victim of domestic violence. Crime and feeling unsafe have perverse effects on economic efficiency. The disproportionate concentration of property crime among the better off could have huge economic consequences, from lower levels of investments and growth to higher migration rates among the educated. The intensity of crime concentration among business owners reveals a perverse effect on economic efficiency; that is, reducing investment and employment in poor urban communities. Exposures to homicide and domestic violence have perverse psychological and economic effects, yet quality family life can play a protective role. For its part, the concentration of homicide and domestic violence among the worse off, besides its obvious effects on the well- being of the victims, imposes a psychological and economic burden on their relatives and increases their chances of enduring pathological and dysfunctional behaviors, especially among children and youngsters, and ultimately reduces socioeconomic mobility and contributes to the perpetuation of poverty. Brook, et al. (1999, 1998) show that, in the Colombian case, vis-A-vis other developed nations, quality of family interaction plays a particularly efficient role in the reduction of the probability of drug use, crime involvement, and post-traumatic syndrome among Colombian youth living in high-risk neighborhoods.63 3.4 Armed conflict in rural areas, displaced population, and vulnerable groups The displaced include a majority of women and children. As was shown in Section 1, the estimate of Internally Displaced People (IDP) is at least 1 million during the last 15 years. The displaced include a majority of women and children, who fled as the threats to their safety became closer and closer, as well as an overwhelming majority of rural households, whose main income came from agricultural activities. 63 They conclude that "consideration should be given, therefore, to developing prevention programs aimed at family bonding, not only for its intrinsic value, but also for its long-range implications for decreased (... drug addiction ...), delinquency, and ultimately violence" Brook, et aL. (1999). 36 IDP either try to stay close to their land in the departmental cabeceras or they migrate to the safer anonymity of larger urban centers, such as Bogotd-which may receive up to 25 percent of the IDP-Medellfn, Barranquilla, and Cartagena. These forced migrants, who are trying to escape the various armed groups, fuel urban growth. A survey of 200 IDP in Bogota, Medellin, and Cartagena suggests that households that opt to reach departmental urban or peri- urban centers may be undertaking "preventive displacement" as their environment becomes generally unsafe and they predict an escalation of conflict at the local level. Those who seek larger and more distant cities may be trying to escape direct threats to their lives when fleeing is the sole option to remain alive. The latter are typically leaders in their community and/or depend on agriculture as their main source of income. They must often leave quickly, and the majority is forced to abandon their assets, i.e., land (Erazo, et al., 2000). The increasing trend in the numbers of IDP and the limited absorption capacity of the urban sector, especially in the context of the recession, are putting an increasing number of vulnerable groups at risk. IDP face a high risk of unemployment; agricultural skills have little value in urban labor markets, thus limiting their income-generation opportunities. In addition, inadequate housing and poor access to public services characterize the urban slums or spontaneous settlements where they arrive. Major economic losses, tremendous psychological hardship, and the added strain on household structure- atomization of families, single headship-put them at risk of resorting to begging, prostitution, and delinquency and falling into a vicious circle of extreme poverty. The well being of children of IDP households, present and future, is of special concern as households' food security is endangered and the opportunities for school enrollment and attendance shrink. The full economic and social effects of considerable forced migration-on approximately 10 percent of the rural population-are uncertain. One effect may be the concentration of landholdings in the areas of "expulsion" and the increasing weakness of property rights on any kind of real assets. In turn, the latter give rise to perverse efficiency effects on the remaining rural population through its obvious discouragement of private investment. Another may be the creation of a new highly vulnerable group, composed of returnees or re-settled IDP who may not be able to return to their original income-earning strategies owing to the loss of assets or household members. On the other hand, population loss could possibly lead to local labor shortages in rural areas and, thus, rising wages. In summary, violence and the continual worsening of crime since the 1970s have eroded the welfare of all Colombian households and imposed a considerable burden on Colombian society as a whole. The trend toward persistent deterioration of violence is mostly associated with illegal drug trade and the presence of illegal armed groups. Poor households bear most of the burden of homicide and domestic violence; at the same time young Colombian males face an extremely high risk of being murdered. The latter fact helped to duplicate the gender gap of life expectancy from the 1970s to the 1990s. Uneducated women and spouses of uneducated men bear a disproportionate share of domestic violence. On the other hand, the better off bear a disproportionate share of the burden of property crime, extortion, and kidnapping and are more likely to be victimized, to feel unsafe, and to modify their behavior accordingly, including increasing investment in crime avoidance. Moreover, this situation induces serious indirect efficiency costs on the economy through migration of educated Colombians to other countries and reductions in domestic private investment, detrimental to growth and the generation of employment for the poor. For its part, the concentration of homicide and domestic violence among the worse off, besides its obvious effects on the well-being of the victims, imposes a psychological and economic burden on victims' relatives and increases their chances of enduring pathological and dysfunctional behaviors, especially among children and youngsters, ultimately 37 eroding socioeconomic mobility and contributing to the perpetuation of poverty. Quality family life has been found to be an efficient protective mechanism; thus, preventive programs aimed at reinforcing family bonding should be developed. Equally important is the regulatory environment to promote stability and the peaceful resolution of conflict within the family. Another important component of the burden of violence is the increasing number of internally displaced people-at least 1 million-associated with the armed conflict in the rural areas. Finally, the social costs of violence are amplified by the demand for public resources---current and future--to attend the victims and restore security. Although public expenditure in justice and security more than doubled during the last decade, the improvements in security and law enforcement have been limited. Finally, progress in economic and social development was mostly resilient to growing insecurity. To assess how poor Colombians had been doing during the last two decades, this chapter examined three axes of welfare development: poverty, social development, and personal security. These three main welfare fundamentals have not evolved in complete harmony. Sporadically they diverge and turn out to be in partial contradiction. Despite two decades of persistent security deterioration, Colombia experienced clear, unambiguous improvement in social and economic indicators. Although most of the economic welfare and social indicators in education, health, and infrastructure show quite positive developments during the last two decades, the simultaneous escalation of violence-with its perverse effects on life expectancy among males and increasin5 overall social cost-reveals a considerable deterioration of living conditions in urban Colombia. Moreover, social investment in human resources showed some healthy robustness relative to the market instability of the late 1990s. Only recently-probably owing to the economic recession- has school enrollment been declining. 6 See DNP-UNDP (1999), Chapter 1, for estimates of Human Development Index incorporating life expectancy estimates. 38 Chapter /Z V/nerability. A re the Poor Chanpg Faces? During the last two decades the populations most vulnerable to poverty have maintained a relatively constant set of characteristics. However, over time, such characteristics have become more polarized. Insufficient household income per capita-hence poverty-is tautologically associated with at least one of the following conditions: low skills, lower employment rates, higher dependency ratios-children per adult per household--and lower wages. All four are predominant among poor households in Colombia. During the 1990s, the likelihood of a household escaping poverty became increasingly dependent on having fewer children, having more working-age members with post-secondary education, and having access to employment for members other than the head. Contrary to what could be expected, the presence of the elderly within a household reduces the risk of poverty. Homeowners and middle- to high- age-headed households face a lower risk as well. Poverty may also arrive in less typical circumstances; that is, from low- frequency events that bring devastating welfare effects such as the loss of employment for the head, the presence of a disabled member in the household, or being a recent migrant or an IDP -in the late 1990s-. In contrast to the urban situation, rural poverty remains much more severe and represents a higher share of total poverty. In summary, the faces of the poor are typically children of all ages, young, low to medium- skilled household heads, recent migrants, and non-homeowners. These groups are clearly worse off than pensioners, the college educated, the elderly, and non-recent migrants. 4. CHARACTERISTICS OF VULNERABLE GROUPS Chapter I already established the evolution of poverty, social indicators, and security during the last two decades. Logically, the next step is to identify the poor, or characterize their "faces"; that is, establish the demographic and social characteristics of the groups most vulnerable to poverty in Colombia. The main questions to be addressed include: What is the set of characteristics that identify the poor today? Have those characteristics remained stable over time? Or have new clusters of poverty emerged? Which characteristics have been the most detrimental? Which detrimental characteristics are more common or rare? Which characteristics protect against poverty? How does vulnerability to poverty change over the lifecycle? The identification of the poverty profile of different demographic groups has clear policy consequences. In particular, it provides a guide for the selection of appropriate public policy instruments and their target groups. 4.1 Urban Colombia: The faces of the urban poor have not changed, but they are becoming more apparent and polarized Facts versus common beliefs about vulnerability: The typical faces of the poor are children of all ages, younger household heads, recent migrants, and non-homeowners. They are clearly worse off than pensioners, the elderly, and non-recent migrants. Who are the most vulnerable among various urban population groups? Table 7 reports the incidence of poverty in some population groups traditionally considered more vulnerable to poverty: migrants, women, children, non- homeowners, pensioners, the elderly, and the disabled, along with the overall urban poverty rate for comparison purposes. Popular beliefs are confirmed in relation to children of all ages and 39 non-homeowners. Children under 18 consistently present higher poverty rates than the entire population, with an increasing proportion of infants and preschoolers in poverty. Homeownership clearly provides protection against poverty. Migrants used to fare better than or similarly to the overall population until 1995, but recent migrants are more likely to fall into poverty in recession years. This must be partly associated with a shift from voluntary to forced migration--displaced peoples-owing to the armed conflict in rural areas. On the other hand, incidence rates are nearly identical among men and women over the entire period. The disabled have remained consistently poorer than the rest of the urban population since 1988, although incidence rates are decreasing for this group. Pensioners and the elderly do far better than the rest of the population, and their relative standing seems to be improving over time. Table 7. Poverty count for different population subgroups, Urban Colombia, 1978-99 Population 1978 1988 1995 1999 share in 1999 Urban Colombia 70% 55% 48% 55% 100% Households with unemployed heads 91% 80% 72% 78% 8% Children under 2 yrs. 80% 71% 63% 72% 4% From 2 to 6 yrs 81% 70% 63% 69% 10% From 7 to 13 yrs 80% 70% 62% 69% 13% From 14 to 17 yrs. 73% 61% 55% 64% 8% Migrants/just moved' NA 50% 50% 64% 1% Non-homeowners 77% 64% 57% 63% 51% Disabled 69% 68% 60% 60% 1% Migrants<5%2 NA 51% 46% 60% 1% Women. 69% 55% 48% 55% 53% Migrants