PA R A G U AY POVERTY ASSESSMENT Determinants and Challenges for Poverty Reduction December 2010 » Poverty Reduction and Economic Management Unit, Latin America and the Caribbean Region Photographs: Carlos Bittar, Sonia Delgado, and World Bank Image Bank PARAGUAY PoveRt Y As ses sment DeteRmin Ant s A n D C hA l l e nG e s foR PoveRt Y R e D U Ctio n Report no. 58638-PY December 2010 CURRENCY EQUIVALENTS US$1.0 = 4915 Guaranís (Nov 2010) FISCAL YEARL January 1 – December 31 MAIN ABBREVIATIONS AND ACRONYMS BCP Central Bank of Paraguay (Banco Central de Paraguay) CADEP Centro de Análisis y Difusión de la Economía Paraguaya CCT Conditional Cash Transfers CEDLAS Center for Distributive, Labor, and Social Studies CPI Consumer Price Index DGEEC General Directorate of Statistics, Surveys and Censuses (Dirección General de Estadística, Encuestas y Censos) DIPLANP Dirección del Plan de la Estrategia Nacional de Lucha contra la Pobreza ECLAC Economic Commision for Latin America and the Caribbean EIH Integrated Household Survey (Encuesta Integrada de Hogares) ENLP National Strategy for the Fight against Poverty (Estrategía Nacional de Lucha contra la Pobreza) EPH Permanent Household Survey (Encuesta Permanente de Hogares) FEI Food Energy Intake GDP Gross Domestic Product HOI Human Opportunity Index IADB Inter-American Development Bank ICV Life Quality Index (�ndice de Calidad de Vida) ILO International Labour Organization INAN National Institute of Food and Nutrition (Instituto Nacional de Alimentación y Nutrición) INEI National Statistical and Technological Institute (Instituto Nacional de Estadística e Informatica) IPS Social Security Institute (Instituto de Previsión Social) LAC Latin America and the Caribbean MDG Millennium Development Goals MEC Ministry of Education and Culture (Ministerio de Educación y Cultura) MFI Micro Financial Institutions MSPyBS Ministry of Public Health and Social Wellbeing (Ministerio de Salud Pública y Bienestar Social) NCP Non-Contributory Pensions PPP Purchasing Power Parity RUC Unique Registry of Contributors (Registro Único de Contribuyentes) SAS Social Action Secretariat (Secretaría de Acción Social) SEDLAC Socio-Economic Database for Latin America and the Caribbean SIPASS Paraguayan Social Security System (Sistema Paraguayo de Seguridad Social) UNDP United Nations Development Programme WB World Bank WDR World Development Report Vice President: Pamela Cox Country Director: Penelope Brook PREM Director: Marcelo Giugale Sector Leader: Jose Roberto Lopez Calix Country Manager: Rossana Polastri Sector Manager: Louise Cord Task Manager: Carolina Diaz-Bonilla Table of Contents Acknowledgements i executive summAry iii POLICY CONSIDERATIONS xi introduction 1 BACKGROUND, OBJECTIVES, AND THE POVERTY ASSESSMENT PROCESS 1 chApter 1 - economic growth, poverty, And inequAlity 3 INTRODUCTION 3 UPDATING THE METHODOLOGY FOR POVERTY MEASUREMENT IN PARAGUAY 5 MEASURING POVERTY IN PARAGUAY 7 POVERTY IN PARAGUAY 8 GROWTH INCIDENCE CURVES USING HOUSEHOLD PER CAPITA INCOME 10 INEQUALITY 12 POVERTY DECOMPOSITION 12 NON-INCOME MEASURES OF WELL-BEING 14 POVERTY AND PUBLIC SOCIAL PROGRAMS 16 chApter 2 –determinAnts of poverty And inequAlity 22 WHO ARE THE POOR? DIFFERENCES BETWEEN THE POOR AND THE NON POOR 22 Education and the poor 25 Health and the poor 27 DEMOGRAPHIC CHANGES 29 CORRELATES OF POVERTY 29 Demographics 30 Education 31 Labor 31 Household size and structure 31 Dwelling characteristics, land and durables 31 Infrastructure and geography 33 HUMAN OPPORTUNITY INDEX - PARAGUAY 33 Human Opportunity Index - Education 33 Human Opportunity Index - Housing 35 Human Opportunity Index - Geographic Disparities 37 chApter 3 - urbAn lAbor mArket: trends And opportunities for employment 40 TRENDS IN LABOR MARKET INDICATORS 40 EMPLOYMENT 40 Employment by sector and economic growth 41 The level of qualification of the labor force 44 Informality 45 Under-employment 45 UNEMPLOYMENT 47 Unemployment duration 49 Youth employment and unemployment 49 JOB CREATION AND JOB DESTRUCTION DURING THE CRISIS 51 INCOME 52 The price of work: evolution and differentials of labor income 52 Determinants of income 52 Gender wage gaps in Paraguay 55 CONCLUSIONS AND POLICY CONSIDERATIONS 56 chApter 4 – rurAl fActor mArkets And poverty 60 INTRODUCTION 60 LAND MARKETS 61 FINANCIAL MARKETS 66 LABOR MARKETS 68 CONCLUSIONS AND POLICY RECOMMENDATIONS 69 chApter 5 – ex-Ante evAluAtion of the expAnsion of cAsh trAnsfer progrAms in pArAguAy 71 INTRODUCTION 71 SOCIAL PROTECTION IN PARAGUAY 72 Social insurance 74 Social assistance 74 INCIDENCE ANALYSIS 76 Targeting instrument 79 SIMULATION SCENARIOS 81 CONCLUSIONS 87 Annex 1 – chAnges in the life quAlity index (icv) 90 CHANGES TO ICV 2007 92 Annex 2 – pArAguAy: crisis, drought And poverty in 2009 93 PROLOGUE 93 INTRODUCTION AND SUMMARY 93 RECESSION WITH LESS POVERTY: A PUZZLE IN PARAGUAY 93 Poverty 94 Inequality 96 WHAT IS THE EXPLANATION FOR THE INCREASE OF RURAL POVERTY? 96 Average Incomes – Growth Incidence Curves 96 PUBLIC TRANSFERS 97 LABOR MARKET 98 Employment 98 Unemployment 99 WHY DOES URBAN POVERTY DECLINE? 100 Urban Employment 100 Urban Salaries 101 Remittances 101 CONCLUSIONS 102 references 104 list of figures FIGURE 1: PER CAPITA GDP EVOLUTION IN LAC, 1990-2008 III FIGURE 2: HUMAN OPPORTUNITY INDEX - LAC RANKING, PROJECTED FOR 2010 VII FIGURE 3: DISTRIBUTION OF WORKERS BY EDUCATIONAL LEVEL (CIRCA 2008) VIII FIGURE 4: SHARE OF WORKERS IN INFORMAL SECTOR (25-65 YEARS OLD; CIRCA 2008) VIII FIGURE 1: GDP PER CAPITA AND GDP ANNUAL GROWTH, LATIN AMERICA, CIRCA 2008. 1 FIGURE 1.1: POVERTY EVOLUTION IN PARAGUAY 1997/98-2008 4 FIGURE 1.2: REAL PER CAPITA GDP EVOLUTION IN PARAGUAY 1995-2008 4 FIGURE 1.3: PER CAPITA GDP EVOLUTION IN LAC, 1990-2008 4 FIGURE 1.4: IMPACT OF THE NEW METHODOLOGY ON POVERTY ESTIMATES 1997/98-2008 7 FIGURE 1.5: OVERALL POVERTY BY REGIONS, PARAGUAY 2003-2008 9 FIGURE 1.6: GROWTH INCIDENCE CURVE, PARAGUAY (2003-2008) IN GUARAN�ES OF 2008 11 FIGURE 1.7: GROWTH INCIDENCE CURVE BY AREA, PARAGUAY (2003-2008) IN GUARANIES OF 2008 12 FIGURE 1.8: GINI COEFFICIENT 13 FIGURE 1.9: GINI COEFFICIENT OF PER CAPITA HOUSEHOLD INCOME BY AREA 14 FIGURE 1.10: HOUSE CONSTRUCTION MATERIALS, PARAGUAY 2003 -2008 15 FIGURE 1.11: HOUSEHOLD ACCESS TO PUBLIC SERVICES, PARAGUAY 2003-2008 16 FIGURE 1.12: PROJECTION OF POVERTY AND EXTREME POVERTY IN 2009. 17 FIGURE 2.1: YEARS OF EDUCATION BY AGE. NATIONAL AND BY POVERTY, PARAGUAY 2008 26 FIGURE 2.2: SCHOOL ATTENDANCE BY AGE IN PARAGUAY, 2008 27 FIGURE 2.3: ACCESS TO HEALTH INSURANCE BY TYPE OF PROVIDER 28 FIGURE 2.4: ACCESS TO HEALTH INSURANCE BY INCOME QUINTILES 29 FIGURE 2.5: HEALTH INSURANCE COVERAGE BY AGE AND POVERTY STATUS, 2008 29 FIGURE 2.6 HEALTH CENTER VISITED WHEN SICK (IN THE LAST 90 DAYS) 30 FIGURE 2.7: OVERALL URBAN AND RURAL HOUSEHOLD SIZE, PARAGUAY 2003-2008 31 FIGURE 2.8: HUMAN OPPORTUNITY INDEX - LAC RANKING, PROJECTED FOR 2010 36 FIGURE 2.9: HUMAN OPPORTUNITY INDEX IN EDUCATION 37 FIGURE 2.10: HOI IN SCHOOL ATTENDANCE AND COMPLETING SIXTH GRADE ON TIME 37 FIGURE 2.11: OPPORTUNITY INDEX IN HOUSING 38 FIGURE 2.12: HOI IN HOUSING INDICATORS 39 FIGURE 2.13: HOI IN SCHOOL ATTENDANCE AND COMPLETING SIXTH GRADE ON TIME 40 FIGURE 2.14: HOI IN HOUSING INDICATORS (2008) 41 FIGURE 2.15: HUMAN OPPORTUNITY INDEX BY REGION 42 FIGURE 3.1: LABOR FORCE PARTICIPATION – WORKING AGE POPULATION 43 FIGURE 3.2: EMPLOYMENT RATES BY GENDER 44 FIGURE 3.3: DISTRIBUTION OF WORKERS BY REGION 2003-2008 44 FIGURE 3.4: ORIGIN OF URBAN MIGRANTS 46 FIGURE 3.5: DISTRIBUTION OF WORKERS BY ECONOMIC SECTOR 47 FIGURE 3.6: REAL GDP GROWTH 2003-2008 (CONSTANT PRICES OF 1994) 47 FIGURE 3.7: YEARS OF EDUCATION – POPULATION 21-30 YEARS OLD 48 FIGURE 3.8: DISTRIBUTION OF WORKERS BY EDUCATIONAL LEVEL (CIRCA 2008) 48 FIGURE 3.9: SHARE OF WORKERS IN EACH SKILL GROUP (BY GENDER) 49 FIGURE 3.10: SHARE OF WORKERS IN INFORMAL SECTOR (25-65 YEARS OLD; CIRCA 2008) 49 FIGURE 3.11: INFORMALITY BY EDUCATION LEVEL 50 FIGURE 3.12: UNDER-EMPLOYMENT 51 FIGURE 3.13: UNDER-EMPLOYMENT IN PARAGUAY BY GENDER 51 FIGURE 3.14: UNEMPLOYMENT RATES BY AREA AND GENDER 53 FIGURE 3.15: UNEMPLOYMENT RATES BY PER CAPITA HOUSEHOLD INCOME, 2008 54 FIGURE 3.16: UNEMPLOYMENT DURATION BY AREA AND GENDER 54 FIGURE 3.17: UNEMPLOYMENT DURATION BY EDUCATIONAL LEVEL 55 FIGURE 3.18: YOUTH UNEMPLOYMENT 56 FIGURE 3.19: PREVIOUS SECTOR OF EMPLOYMENT FOR YOUTH AGED 15 TO 25 YEARS OLD 56 FIGURE 3.20: HOURLY WAGES AND HOURS OF WORK 57 FIGURE 3.21: HOURLY WAGE AND WEEKLY HOURS BY EDUCATIONAL LEVEL 58 FIGURE 3.22: HOURLY WAGE BY INFORMALITY IN GUARANIES OF 2008 58 FIGURE 3.23: MINCER EQUATION. ESTIMATED COEFFICIENTS OF EDUCATIONAL DUMMIES (ALL WORKERS 25-55 YEARS OLD) 59 FIGURE 3.24: GENDER WAGE GAP – URBAN SALARIED WORKERS 60 FIGURE 3.25: UNEXPLAINED GENDER WAGE GAP BY PERCENTILES OF THE WAGE DISTRIBUTION OF MALES AND FEMALES 63 FIGURE 4.1: CONTRIBUTION OF THE RURAL POOR TO TOTAL POVERTY IN 1997, 2003 AND 2008 65 FIGURE 4.2: SOURCES OF INCOME BY QUINTILE 65 FIGURE 4.3: REMITTANCES AS A SHARE OF TOTAL HOUSEHOLD INCOME BY DENSITY AREA 66 FIGURE 4.4: LAND OWNERSHIP AND THE ALLOCATION OF LABOR IN RURAL PARAGUAY 66 FIGURE 4.5: LAND OWNED AND USED GINI ACROSS TIME IN PARAGUAY 67 FIGURE 4.6: LAND OWNERSHIP AND POVERTY 67 FIGURE 4.7: LAND OWNERSHIP OVER TIME 68 FIGURE 4.8: THEORETICAL OUTCOMES OF LAND RENTAL MARKETS 69 FIGURE 4.9: EMPIRICAL RELATIONSHIP BETWEEN OWNED AND USED LAND IN PARAGUAY 69 FIGURE 4.10: THE INVERSE RELATIONSHIP BETWEEN FARM SIZE AND PRODUCTIVITY IN PARAGUAY 70 FIGURE 4.11: COST OF TRANSFERRING AND REGISTERING PROPERTY AS A PERCENTAGE OF ITS VALUE 71 FIGURE 4.12: ACCESS TO CREDIT BY TITLED AND UNTITLED FARMS IN PARAGUAY 73 FIGURE 4.13: LAND OWNERSHIP AND THE USE OF MODERN INPUTS 74 FIGURE 4.14: SHARES OF HARVEST VALUES ON SELECTED CROPS (2008) 74 FIGURE 5.1. INCIDENCE CURVES. PENSIONS, SURVIVAL PENSIONS AND CONDITIONAL CASH TRANSFERS. 2008 86 FIGURE 5.2. TARGETING ACCURACY WITH RESPECT TO POVERTY MEASURES FOR ALTERNATIVE CUT-Off VALUES OF ICV. (2008) 88 FIGURE 5.3. EXCLUSION AND INCLUSION ERRORS OF CCTS WITH RESPECT TO POVERTY AND EXTREME POVERTY MEASURES. ALTERNATIVE CUT-Off VALUES OF ICV (2008) 89 FIGURE 5.4. EXCLUSION AND INCLUSION ERRORS OF CCTS WITH RESPECT TO POVERTY AND EXTREME POVERTY MEASURES. ALTERNATIVE CUT-Off VALUES OF ICV. ASUNCION AND RURAL REST, 2008 89 FIGURE 5.5.: POTENTIAL WELFARE AND COST IMPLICATIONS OF EXPANDING CONDITIONAL CASH TRANSFERS PROGRAMS IN PARAGUAY, 2008. GEOGRAPHICAL EXPANSION AND BENE�T INCREASES. 92 FIGURE 5.6. POTENTIAL WELFARE AND COST IMPLICATIONS OF EXPANDING CCT IN PARAGUAY. RURAL, 2008. 93 FIGURE 5.7. THE ATKINSON INDEX FOR SIMULATED SCENARIOS, DIFFERENT EPSILON VALUES. PARAGUAY 2008 94 FIGURE 5.8.: POTENTIAL WELFARE AND COST IMPLICATIONS OF EXPANDING THE CCT AND NON-CONTRIBUTORY PENSIONS IN PARAGUAY, 2008 94 FIGURE 5.9.: POTENTIAL WELFARE AND COST IMPLICATIONS OF EXPANDING THE NCP IN PARAGUAY, 2008. 95 FIGURE 5.10. INCIDENCE CURVES. CONDITIONAL CASH TRANSFERS (BEFORE AND AFTER EXPANSION), NON-CONTRIBUTORY PENSIONS (POTENTIAL EXP. USING ICV). PARAGUAY, 2008 95 FIGURE 5.11. INCIDENCE CURVES. NON-CONTRIBUTORY PENSIONS (POTENTIAL EXP. USING ICV AND POVERTY MEASURES). PARAGUAY, 2008 96 FIGURE 5.12. ICV 2008 BY REGION 100 FIGURE 5.13. COMPARISON ICV 2008, 2005 WITH AND WITHOUT CHILD VACCINATION VARIABLE. 101 FIGURE A2.1: STRONG IMPACT OF THE 2009 GLOBAL CRISIS ON LATIN AMERICA AND THE CARIBBEAN 103 FIGURE A2.2: GROSS DOMESTIC PRODUCT PER ECONOMIC SECTOR AND GROWTH BETWEEN 2008-2009 104 FIGURE A2.3: INTERNATIONAL COMMODITY PRICES DECLINE IN 2009 104 FIGURE A2.4: EVOLUTION OF POVERTY IN PARAGUAY (2003-2009) 105 FIGURE A2.5: EVOLUTION OF POVERTY IN PARAGUAY (1997/98-2009) 105 FIGURE A2.6: GAP AND SEVERITY OF EXTREME POVERTY, PARAGUAY 2003-2009 106 FIGURE A2.7: GROWTH INCIDENCE CURVE OF THE NATIONAL AVERAGE INCOME (2006-08 AND 2008-09) 107 FIGURE A2.8: GIC OF URBAN AND RURAL AVERAGE INCOME (2008-09) 107 FIGURE A2.9: GIC OF URBAN AND RURAL AVERAGE INCOME (2006-08) 107 FIGURE A2.10: GROWTH INCIDENCE CURVES: RURAL SECTOR 2008-09 108 FIGURE A2.11: PUBLIC TRANSFERS (INCLUDING TEKOPORÃ) IN RURAL AREAS 108 FIGURE A2.12: DISTRIBUTION OF WORKERS PER ECONOMIC SECTORS (PERCENTAGE POINTS) 109 FIGURE A2.13: RATES OF UNEMPLOYMENT PER AREA AND SEX 110 FIGURE A2.14: URBAN EMPLOYMENT, CHANGE IN URBAN EMPLOYMENT AND CHANGE IN INCOME PER HOUR 111 FIGURE A2.15: CHANGE IN SALARY BETWEEN 2008 AND 2009 PER SECTOR AND QUALIFICATIONS (%) 112 FIGURE A2.16: REMITTANCES AS PERCENTAGE OF URBAN AND RURAL HOUSEHOLD INCOME, 2009 (%) 112 FIGURE A2.17: CHANGE IN REMITTANCES AS PERCENTAGE OF FAMILY INCOME, 2008-2009 (%). URBAN AND RURAL 112 list of tAbles TABLE 1.1: KEY SOCIO-ECONOMIC INDICATORS IN PARAGUAY 2000 – 2008 3 TABLE 1.2: SOCIO ECONOMIC INDICATORS: PARAGUAY COMPARED TO LATIN AMERICAN AND THE CARIBBEAN AND TO SELECTED COUNTRIES 5 TABLE 1.3: OLD AND NEW HEADCOUNT RATES BY AREA, 2008 8 TABLE 1.4: POVERTY HEADCOUNT AND CONTRIBUTION TO POVERTY, PARAGUAY 2008 9 TABLE 1.5: POVERTY INDICATORS BY REGION AND AREA, PARAGUAY 2008 10 TABLE 1.6: INCOME SHARES BY DECILES AND RATIOS, PARAGUAY 13 TABLE 1.7: INEQUALITY MEASURES 13 TABLE 1.8: GROWTH AND REDISTRIBUTION DECOMPOSITION OF POVERTY CHANGES 14 TABLE 1.9: PUBLIC SOCIAL SPENDING – AS A PERCENTAGE OF GDP, PARAGUAY 16 TABLE 1.10: POVERTY ELASTICITY (COEFFICIENT ß OF THE EQUATION 1) 17 TABLE 1.11: PARAGUAY’S HOUSEHOLD SURVEYS 18 TABLE 2.1: HOUSEHOLD HEAD AND COMPOSITION BY POVERTY IN 2008 23 TABLE 2.2: HOUSE MATERIALS, INFRASTRUCTURE AND LAND BY POVERTY IN 2008 25 TABLE 2.3:PRIMARY AND SECONDARY ENROLLMENT RATES, PARAGUAY 1997/98-2008 27 TABLE 2.4: URBAN-RURAL AND FEMALE HEADED HOUSEHOLD SHARES 31 TABLE 2.5: CORRELATES TO INCOME BY URBAN AND RURAL HOUSEHOLDS IN PARAGUAY, 2008 34 TABLE 3.1: LABOR MARKET INDICATORS IN PARAGUAY 2003-2008 43 TABLE 3. 2: DISTRIBUTION OF WORKERS (BY GENDER, EDUCATION AND AREA OF RESIDENCE) 45 TABLE 3.3: EMPLOYMENT DISTRIBUTION BY LABOR RELATIONSHIP AND TYPE OF FIRMS 45 TABLE 3.4: CHARACTERISTICS OF HEADS OF HOUSEHOLDS IN URBAN AREAS 46 TABLE 3.5: INFORMALITY, NEW DEFINITIONS 50 TABLE 3.6: UNDER- EMPLOYMENT BY AREA 2003-2008 51 TABLE 3.7: CHARACTERISTICS OF THE UNDER-EMPLOYED (2008) 52 TABLE 3.8: UNEMPLOYMENT AND CONTRIBUTION TO UNEMPLOYMENT, PARAGUAY 2008 53 TABLE 3.9: DISTRIBUTION OF NET JOB CREATION BY LABOR RELATIONSHIP: 2007-2008 57 TABLE 3.10: MINCER EQUATION. ESTIMATED COEFFICIENTS OF EDUCATIONAL DUMMIES BY GENDER 59 TABLE 3.11: MINCER EQUATION. ESTIMATED COEFFICIENTS OF EDUCATIONAL DUMMIES 59 TABLE 3.12: RELATIVE WAGES BY DEMOGRAPHIC AND LABOR CHARACTERISTICS 61 TABLE 3.13: GENDER WAGE GAP DECOMPOSITION 62 TABLE 3.14: GENDER WAGE GAP DECOMPOSITION- JOB RELATED VARIABLES 62 TABLE 4.1: INCOME REGRESSION 76 TABLE 5.1: SOCIAL PROTECTION PROGRAMS IN PARAGUAY 80 TABLE 5.2: PUBLIC SPENDING IN PARAGUAY 1980-2009. % OF GDP 82 TABLE 5.3: EVOLUTION OF CCT PROGRAMS IN PARAGUAY. NUMBER OF BENE�CIARIES, POPULATION AND DISTRICTS COVERED 83 TABLE 5.4: GEOGRAPHICAL DISTRIBUTION OF BENE�CIARIES CCT PROGRAMS IN PARAGUAY. 84 TABLE 5.5: TARGETING ACCURACY OF ICV WITH RESPECT TO POVERTY AND EXTREME POVERTY MEASURES (2008) 88 TABLE 5.6: SIMULATED SCENARIOS OF PROGRAM EXPANSIONS. 91 TABLE 5.7: SIMULATED SCENARIOS OF PROGRAM EXPANSIONS. WELFARE INDICATORS AND ASSOCIATED COSTS. PARAGUAY, RURAL REST, 2008. 93 TABLE 5.8: WEIGHTING FACTORS USED IN THE ICV 2005 99 TABLE A2.1: HEADCOUNT POVERTY AND CONTRIBUTION TO POVERTY, PARAGUAY 2009 105 TABLE A2.2: LABOR MARKET INDICATORS IN PARAGUAY 2008-2009 109 TABLE A2.3: DISTRIBUTION OF EMPLOYMENT, 2008 AND 2009 (%) 110 list of boxes BOX 1.1: PROJECTIONS OF MODERATE AND EXTREME POVERTY FOR PARAGUAY IN 2009 17 BOX 1.2: NATIONAL HOUSEHOLD SURVEYS FOR PARAGUAY 1997-2008 18 BOX 1.3: INTER-INSTITUTIONAL COMMITTEE ON METHODOLOGIES FOR POVERTY MEASUREMENT 19 BOX 1.4: SPECIFIC ISSUES IN THE METHODOLOGICAL REVIEW OF POVERTY MEASUREMENT IN PARAGUAY 20 BOX 1.5: MAIN RESULTS OF REVIEWING, UPDATING AND IMPROVING PARAGUAY’S POVERTY MEASUREMENT METHODOLOGY. 1997-2008 PERIOD 21 Acknowledgements This study was prepared by a team led by Carolina Diaz-Bonilla (Economist, LCSPP) under the overall supervision of Louise Cord (Sector Manager, LCSPP) and Jose Roberto Lopez Calix (Sector Leader, LCSPR). We also appreciate the supervision and support of Jaime Saavedra (former Sector Manager) during the early stages. The team included Georgina Pizzolitto (LCSPP), Carlos Sobrado (LCSPP), Pedro Olinto (LCSPP), Barbara Coello (LCSPP), Juan Martin Moreno (LCSHS), and Graciela Sanchez Martinez (LCSSO), with support from Anne Pillay (LCSPP) and Lucy Bravo (LCSPP). The study at all stages benefited greatly from the comments, inputs and other support from Jose Molinas Vega (LCSPP), Amparo Ballivian (LCSPP), Cecilia Valdivieso (LCSPP), Ruth Gonzalez (LCREA), Rossana Polastri (Country Manager Paraguay), Pedro Luis Rodriguez (former Country Manager), Pedro Alba (former Country Director) and Stefan Koeberle (Acting Country Director), as well as the excellent support from the team of the World Bank Office in Paraguay. Additionally, we are grateful to the Peer Reviewers Malcolm Childress (LCSAR), Marcos Robles (Inter-American Development Bank), Elena Glinskaya (ECSH3), and Ruslan Yemtsov (HDNSP) for insightful and useful comments. We thank the Government of Paraguay for their assistance in this study throughout two years. We benefitted in several meetings from the comments and knowledge of representatives of the Ministry of Finance, in particular from our counterpart Veronica Serafini from the Social Economics Unit of the MOF, the Social Cabinet, the Social Action Secretariat (SAS), the Planning Directorate for the National Strategy to Combat Poverty (DIPLANP), the Ministry of Women, and the General Directorate of Statistics, Surveys and Censuses (DGEEC), who facilitated access to data and information. For the revision of the poverty measurement methodology we recognize the excellent work and professionalism of the DGEEC’s technical team specialized in poverty measurement and household surveys under the supervision of Zulma Sosa (Director General) and Norma Medina (Director of Household Surveys). We appreciate the invaluable support from the international consultants Javier Herrera (IRD-DIAL, France) and Nancy Hidalgo (INEI, Peru) in this process of revision. Finally, we thank all members of the Interinstitutional Committee on Poverty in Paraguay (beyond the DGEEC team and the World Bank) who supported the process of revision of the poverty measurement methodology with the goal of ensuring the maximum transparency possible in the methodologies and processes used to estimate the official poverty rate in Paraguay: Ministry of Finance, Social Cabinet, Ministry of Education and Culture (MEC), Ministry of Public Health and Social Welfare (MSPyBS), National Institute of Food and Nutrition ( INAN), Social Action Secretariat (SAS), DIPLANP, Paraguay’s Central Bank (BCP), Catholic University, Center of Analysis and Dissemination of the Paraguayan Economy (CADEP), Instituto Desarrollo, Paraguayan Chamber of Cereals and Oilseeds Exporters (CAPECO), Rodríguez Silvero & Associates, UNICEF, UNFPA, IRD-DIAL France, and INEI Peru. executive summary Paraguay, a landlocked country with a relatively large gap compared to neighboring countries and countries rural sector, had six consecutive years of economic of similar resources and with important challenges. growth between 2003 and 2008. However, the Gross Domestic Product (GDP) per capita for all these years The country’s relative position in Latin America may except for 2008 remains below the value observed in have further deteriorated in 2009 due to experiencing 1995 (U.S. $ 4,263 in constant 2005 PPP values). Even if we take into account the recent growth of GDP (at an average of 4.6 percent), Paraguay reached a GDP per capita in 2008 of U.S. $4,345 in constant 2005 PPP values. Figure 1: GDP Per Capita Evolution in LAC The GDP per capita gap between Paraguay and the average for Latin America and the Caribbean (LAC) has GDP per cap, PPP (Const. 2005 PPP $) 14,000 Argentina Poverty Assessment increased since 1990 and has continued to increase at 12,000 Uruguay a higher rate since 2003. From 1990 to 2008, despite 10,000 LAC the sustained growth in Paraguay over the last five 8,000 Brazil Peru years, the GDP per capita gap between Paraguay and 6,000 Paraguay LAC increased by 84 percent (Figure 1). Comparisons to 4,000 Peru are striking since both countries had a very similar 2,000 GDP per capita in 1990 (around PPP $4,100) while by - 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2008 Peru’s GDP almost doubled that of Paraguay. The constant per capita GDP in the face of increased values for iii Latin America leaves Paraguay with a significant income Source: World Bank, World Development Indicators 2009. one of the worst contractions in output in the region. 18 countries in the 2010 projections of the Human Preliminary 2009 growth estimates by Paraguay’s Central Opportunity Index (HOI). Moreover, despite the Bank suggest that GDP contracted by 3.8 percent, a recent economic growth, extreme poverty remains result of the severe drought early in the year and the high relative to the rest of Latin America, and may global financial crisis. However, due to data limitations have deteriorated further in the wake of the global (the last available household survey is for 2008), it is financial crisis. not yet possible to estimate the poverty impacts for 2009. Therefore, this report will focus primarily on data However, from 2000 to 2008 Paraguay shows progress available from 1997 to 2008.1 in almost all socioeconomic indicators. As mentioned, during this period Paraguay experienced an increase in Poverty rates significantly declined during the period income and a reduction in poverty. In addition, other 2003-2008, but these improvements have less impact indicators improved as well. For example, the country when compared with the mid-1990s. While overall obtained significant reductions in under-five and infant poverty for the period 2003-2008 decreased by 6.1 mortality rates (13-14 percent), as well as a significant percentage points, between 1998 and 2008 the overall 33 percent reduction in maternal mortality during birth. and extreme headcount poverty rates increased from There were also reported gains in access to improved 36.1 to 37.9 percent, and 18.8 to 19 percent, respectively. water sources (12 percent) and gross enrollment in Although the recent improvements are welcome, secondary school (8 percent). the challenge to reduce poverty and inequality in a sustained manner remains. Paraguay achieved real improvements in the welfare of its population between 2000 and 2008, despite Economic growth is more important than a better the long-term consequences and the unfavorable distribution of income in explaining the reduction comparison to LAC and countries in the region. Being in the national poverty rate between 2003 and 2008. able to understand the source of the improvements in The reduction of 6.1 percentage points in moderate its socio-economic indicators is important to build and poverty that occurred during that period, from 44 to expand the appropriate programs and policies that have 37.9, resulted primarily from growth in incomes (-4.1) made possible such advancements. The present study but also from the improvement in income distribution provides possible explanations and policy implications (-1.5). On the other hand, while rural poverty also associated with the changes observed. declined during this period, the results from the decomposition of poverty show that in this case the This study is based on the extraordinary work carried effects of growth and distribution pull in opposite out jointly with the General Directorate of Statistics, directions, although growth continues to dominate. Surveys and Censuses (DGEEC) to review and update Therefore, policies to maintain economic growth are the methodology for measuring poverty in the important for Paraguay. country. Supported by the World Bank and government officials, the DGEEC undertook the most comprehensive Paraguay appears below the average for Latin and participatory work thus far to review and update America and the Caribbean in most socioeconomic the country’s poverty measurement methodology to indicators. Compared with the average for LAC, reflect current international best practices. The previous Executive Summary Paraguay shows greater income inequality (as methodology was in line with the then-existing measured by the Gini), higher maternal and under- literature in 1997, while the revised methodology is five mortality rates, lower immunization rates, in line with the current literature. The process was access to sanitation and piped water, secondary undertaken in a manner that was transparent, credible, school enrollment, and a lower ranking in the and that helped to build a national consensus around Human Development Index. Paraguay ranks 12th of the new methodology and poverty rates. The process iv 1 Editor’s Note: The present document uses the data available at the time: from 1997 to 2008. However, the Annex (“Crisis, Poverty and Drought in 2009�) includes an update using 2009 data. included the creation of an Interinstitutional Committee in the country with poverty rates above the national comprised of representatives from the Government average. Rural households have the highest overall of Paraguay, academic institutions, private sector and poverty incidence (48.8 percent), gap (20.4 percent), and civil society, as well as international organizations. severity index (11.2 percent), and similarly for extreme The process carried out between February 2008 and poverty (30.9, 11, and 5.6 percent, respectively). They November 2009 has placed Paraguay as a pioneer at the also have the largest number of poor (1.2 million) and Mercosur level in terms of the use of new tools for the extreme poor (787 thousand), and the lowest average measurement of poverty and offers a clearer picture of income among the poor (almost 170 thousand Guarani the dimensions of poverty in the country. per capita per month). Who are the poor? Income inequality in Paraguay decreased significantly in the 2003-2008 period. Average income grew across Poverty and inequality the whole income distribution during the five years, but more so for the poorer population. All inequality The new poverty estimates for the period 1997-2008 indicators, including the Gini, show important show a slightly higher national extreme poverty rate reductions during this period. Nonetheless, the and a much higher rural poverty rate, for all years, as distribution of household per capita income is highly compared to previous calculations. The net impact skewed towards the top 10 percent of households, who of all changes made to improve the measurement of have 40.4 percent of total income. poverty is an increase in the poverty rate that varies between 0.4 and 5.6 percentage points nationally, The household income required to eliminate extreme depending on the year. Furthermore, while poverty poverty was relatively modest, estimated as an shows a general downward trend over the past five additional annual income in 2008 of around Gs. 971 years, overall poverty and extreme poverty levels for billion (1.4 percent of GDP). However, assigning such the rural areas are almost 50 to 70 percent higher, an increase to the precise households and by the respectively, than in the previous calculations. precise amount per household are difficult tasks. In order to achieve such an increase in household income, Despite the recent growth and the improvements a government program would require many more regarding poverty, by 2008 almost two out of five resources (social programs have administrative costs) Paraguayans are poor and one out of five lives in and would achieve only partial success in eliminating extreme poverty. Within the country, the incidence extreme poverty due to the lack of perfect knowledge of poverty shows important variations across regions; to identify all the extreme poor and the lack of perfect the lowest incidence is reported in Asunción (21 targeting (as in all countries), that may result in resources percent), followed by urban regions outside Central being received by the non-poor. Nevertheless, and (27 percent), Urban Central (35.3 percent), and finally only to compare orders of magnitude, the Conditional rural households (at a high 48.8 percent). Although Cash Transfer programs Tekoporã and Pro País together the report makes clear the importance of focusing on spent Gs. 120 billion in 2009, or 0.172 percent of GDP – the rural poor and extreme poor, the concentration of just over one tenth of the amount needed to eliminate Poverty Assessment poor households within main urban areas of the Central poverty if there were perfect targeting. Department is common and should also be given special consideration. Access to key goods and services Over half of the poor and more than two thirds of the Better housing quality and especially improved extreme poor are located in rural areas of Paraguay. access to services since 2003 are consistent with the With only 41.4 percent of the population, rural poverty decreases reported through 2008. However, households have a disproportional amount of the poor important improvements are still needed in the water (53.5 percent) and the extreme poor (67.5 percent). In distribution systems in order to reach the almost third v 2008, the rural region is the only one of the four regions of the country without access as of 2008. An analysis of the characteristics of poor households 2003-2008, however it is important to highlight that it shows substantial differences that can be used to increased to 3 percent in 2009, closing the gap with improve the targeting of social programs. Overall, the region’s average of 3.8 percent. Low investments in poor household heads have similar age and gender health and limited health insurance coverage (public characteristics as non poor heads, but much higher and private) are two contributing factors for the low rates of incomplete primary education, few with health indicators found in Paraguay. In the last five years, tertiary education, and much higher informality rates, public health insurance has absorbed 100 percent of while a third work in agriculture. Poor households have the growing demand for insurance due to population a higher dependency ratio, an average of 5.3 members, growth as well as coverage increase, while private and speak Guarani at home. Poor households also live health insurance has remained steady. Very few poor in low quality dwellings of inferior floors and ceilings, persons have access to health insurance, they show a with an average three persons per bedroom, but with lower number of sick visits even though they have a good access to electricity and telephone services. higher rate of health problems, and they are more likely to visit health centers than hospitals. In general, and as expected, non poor households have better quality housing and more access to public An analysis of the correlates of poverty was carried services. However, there are important exceptions, out to help deepen the understanding of how poverty mainly: (i) there is no difference in house ownership; (ii) and household characteristics are associated. The low quality wall materials are almost absent in either results of the econometric estimates show that income group; (iii) the poor have only slightly lower access to is higher, and the probability of being poor lower, if piped water and electricity in the house; (iv) the poor the individual has more education, is younger, male, is live in neighborhoods with similar water, electricity and biliangual Spanish-Guarani, does not work in agriculture telephone infrastructure as the non poor; and (v) more or the informal sector, and lives in a household with less poor households own land than non poor households, members. For urban individuals, the probability of a especially lots between 2 and 15 hectares. higher income increases if the person does not live in the Central Urban region. Land ownership is associated In the last ten years, poor households have made with higher income for rural households. significant gains in school enrollment rates, but important challenges remain. Improvements in primary Human Opportunity Index and secondary enrollment rates over this period have been pro-poor. But the improved school enrollment Another important analysis of determinants relates rates for poor children still have not reached the last year to the circumstances that explain the inequality of of secondary school. School attendance for 6-12 year old opportunities for children to access basic goods children is almost universal (98 percent) in Paraguay, yet and services. Although income inequality has been after age 12 it drops substantially, for both the poor and decreasing, it remains high for Paraguay, and may non poor, reaching less than 50 percent by age 18. Poor originate at least partially in the existing inequality thirteen year old children have an almost 10 percentage of access to basic opportunities among children. The point lower attendance rate than non poor children, Human Opportunity Index (HOI) is an operational widening to a gap of 30 percentage points by age 18. measure of opportunities that takes into account Executive Summary In addition, with less than a 30 percent attendance rate both coverage and the distribution of access to for 18 year-old poor children, a lot of work still remains basic goods and services by children. The principle to be done to help the adolescent poor continue in their of equality of opportunities stipulates that children education, and in that way increase their possibilities of should have the opportunity to access key goods leaving poverty behind. and services needed to have the opportunity to be successful in life and that such access should not Paraguay’s health indicators are below the LAC depend on circumstances over which there is no vi region’s average. Government spending in health only control (such as race, gender, family income, parents’ increased from 1.35 to 1.88 percent of GDP in the period education level, or place of residence). Even though Paraguay ranks 12th out of 18 LAC Women and ethnic groups experience the greatest countries in the 2010 HOI projections, its growth challenges in the labor market. Female workers rates are above the LAC average, reflecting a experience a longer unemployment duration (8.3 positive catch-up effect underway (Figure 2). The months), higher levels of informality (72 percent), country is doing well in school attendance and access and a strong wage gap (5.5 percent) that cannot be to electricity, however it needs to improve in the explained by observable characteristics. Ethnicity plays opportunities for completion of sixth grade on time, a more important role in wage differences, however, as access to drinking water, and sanitation. The unequal the earnings differences between minorities and non- distribution of opportunities is particularly apparent in minorities are higher than those found between males the case of access to sanitation, and as is the case for and females (for example, for non-domestic employees, the three infrastructure dimensions, the circumstances women from ethnic minorities earn 63.5 per cent of the that most matter in the inequality of access are the area average wage for women, while women who do not where the household is located and somewhat less belong to an ethnic minority earn 112 percent of the the household’s per capita income. In the case of the average wage). inequality of access to education, the circumstance with the relatively most importance is parents’ education. Compared to other Latin American countries with Finally, at the subnational level, the departments that similar levels of per capita income, Paraguay ranks are in most need of attention due to a low HOI are above the median in terms of the levels of education Itapúa, Guairá, Caaguazú, and San Pedro. of its population. Even though the country has a low level of GDP per capita, the percentage of young adults Labor Market Access and Participation with complete primary education is relatively high. Given the level of GDP per capita of the country, the net Paraguay also shows improvements in almost all secondary school enrollment rate is also high. labor market indicators during the growth of 2003- 2008. Labor force participation and employment However when analyzing the level of education of the have increased, while unemployment and informality labor force, Paraguay has a higher share of workers levels have decreased. However, both the level of with low levels of education and a high level of underemployment (28.3 percent of the working informality. Among countries in the region, Bolivia is population) and unemployment duration (7.2 months), the only country that has a higher share of its labor force particularly for urban workers, increased, suggesting with low levels of education (55.3 percent of workers in structural problems in the labor market. Bolivia have less than complete primary education), and Figure 2: Human Opportunity Index - LAC ranking, projected for 2010 HOI Level HOI growth rates Chile Mexico Uruguay Nicaragua Mexico Ecuador Poverty Assessment Costa Rica Brazil Venezuela, R.B Peru Argentina Guatemala Jamaica Paraguay 1.14 Ecuador Rep. Dominicana Colombia Colombia Brazil El Salvador Rep. Dominicana Chile Paraguay 73 Honduras Panama Costa Rica LAC Average Peru LAC Average Uruguay Guatemala Panama (0.99) El Salvador (76.5) Venezuela Nicaragua Jamaica Honduras Argentina 40 50 60 70 80 90 100 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 vii Source: World Bank sta calculations based on EPH, Paraguay. the only country to have higher levels of informality in rural areas, however, and land ownership determines (Figures 3 and 4). Although the share of informal workers how rural households allocate their labor. For example, has decreased over time, Paraguay maintains sizeable land rich families tend to dedicate more time to levels of informality (67 percent in 2008). independent agriculture activities, while the land poor dedicate more time to wage and independent work in Key challenges for the rural poor non-farm activities. This seems to be due to the high fixed cost of entrance into self-employed commercial While agriculture is still the major source of livelihoods farming, which would require access to more land and for poor people in rural Paraguay, the importance of credit. rural labor markets, and the rural nonfarm economy, has increased. Access to land, financial, and labor Because land ownership is highly unequal in markets affect the probability of a household being Paraguay, land is a likely determinant of equality of poor. Land continues to be the most important asset opportunity in rural areas via the link to investment in human capital. Some studies have shown a close link between land ownership and investments in nutrition and education (Galor, Moav, Vollrath, 2006). Land Figure 3: Distribution of workers by educational level (circa 2008) ownership inequality has remained high in Paraguay, despites several efforts to improve its distribution via High Medium Low land reform (Carter and Galeano, 1995). Distribution of workers by educational level 100% 22.3 18.9 17.0 20.7 12.9 16.3 15.5 15.6 80% 30.2 Land ownership is strongly associated with the 35.7 29.7 30.3 29.1 60% 38.5 39.5 31.7 probability of not being poor in rural Paraguay. Land 51.6 41.7 rental markets could potentially reduce the effects of 40% 54.0 54.2 55.3 land ownership inequality on poverty, but they do not 20% 42.7 43.9 47.7 51.4 26.1 28.1 seem to be distributing land to the land poor. This may 0% represent a loss of productivity, as there seems to be a Chile Argentina Peru Uruguay Costa Rica Brazil Ecuador Paraguay Bolivia strong association between farm size and productivity – small farms seem to be considerably more productive than larger farms. Source: SEDLAC database (CEDLAS and World Bank). Labor is still the largest endowment of the rural poor, and in an environment in which land and credit Figure 4: Share of workers in informal sector markets are not effective in allocating production (25-65 years old; circa 2008) factors, the poor sometimes have no choice other 80 than selling their labor to the market, even when their 70 67 69 most productive employment would be in agriculture. 64 65 66 60 58 59 60 61 48 49 49 53 Since the gap in Paraguay between the number of new 50 Percentage 41 rural workers and the number of new jobs in agriculture 40 35 38 40 30 is likely to grow, rural labor markets will become a key Executive Summary 20 alternative for exiting poverty. A dynamic rural economy 10 0 will be needed to ensure growth in the demand for labor in Paraguay. Chile Costa Rica Argentina Uruguay Panama Mexico Brazil Rep. Dominicana Honduras El Salvador Ecuador Colombia Peru Nicaragua Guatemala Paraguay Bolivia Expansion of two public cash transfer programs Note: Productive de nition of informality: A worker is considered informal if (s)he is a salaried workers in a small rm, a non-professional self-employed, or a zero-income worker. Based on each country Household Survey, for the years 2007 and 2008 except for: Nicaragua (2005), The extreme poor, and in particular the rural extreme viii Chile (2006), El Salvador (2006), Guatemala (2006), Colombia (2006). Source: SEDLAC database (CEDLAS and World Bank). poor, may require extra assistance to try to break the inter-generational transmission of poverty by important for Paraguay to focus (as in recent years) improving nutrition, health, and education. With this on policies that promote broader economic growth. goal, and in the context of the global financial crisis Some of these policies to promote growth are linked that began in 2008, the Government of Paraguay to political considerations mentioned below, such included the expansion of the conditional cash transfer as an increase in labor market productivity and the (CCT) program (Tekoporã) from 18,000 to 100,000 promotion of a dynamic rural economy with higher beneficiaries as a priority policy of its administration. rural productivity. However, one can find a more A set of ex-ante simulations suggest that the impact of detailed analysis on growth in the “Country Partnership the CCT program expansion would have had a small Strategy for Paraguay 2009-2013�. In short, it stresses but positive effect in reducing poverty and income that maintaining the macroeconomic stability inequality. The impact is relatively stronger if one that has been achieved, and reducing overall analyzes only the rural sector and only the extreme macroeconomic volatility, will remain an important poor (a decrease of 1.7 percentage points), to whom component of the growth agenda of Paraguay, as well the program was primarily targeted. The impact on as the continuation of structural reform policies such the Poverty Gap (the distance to the poverty line) would as increasing trade liberalization, a better education, be even stronger (at 12 percent for the rural extreme and financial deepening. poor) than simply the headcount rate. In addition, the results of different types of expansions show that Take into account the new poverty estimates an increase in the coverage rate is more important to in policy making and in the design of targeting decrease extreme poverty than increasing the amount instruments of the benefit. The change in ranking between urban and rural A second set of ex-ante simulations were run to analyze households due to methodological changes has the possible impact of the new law mandating the important implications for the policies and programs expansion of non-contributory pensions (NCP) to designed to reach the poor. The new poverty the elderly poor (aged 65 or more). The results show estimates are an indication that the share of resources that targeting beneficiaries of the NCP program by allocated to fight poverty and for social programs need free-riding the efforts made by the CCT programs in to increase in favor of rural areas. The government enrolling beneficiaries would yield poor targeting of Paraguay could reformulate some components results, low welfare improvements, and small cost/ of the agricultural agenda and general rural policies benefit relationships. The Life Quality Index (ICV in in order to address the much higher overall poverty Spanish) was not designed to target the elderly poor and especially extreme poverty levels revealed by the but structurally poor rural households. There is an improved methodology. overlap between the CCT programs and an eventual NCP program, but combining both programs will only Special attention should be given to targeting raise expenditure on some households, crowding out mechanisms to reach the poor. Social programs that use the opportunities of other households to gain access to geographic characteristics to identify the poor should social protection programs. take into consideration the new poverty estimates. Poverty Assessment Since the new methodology has a very different impact Policy Considerations in households according to their place of residence, any type of geographic ranking used to target social Focus on policies promoting growth programs can significantly change. Programs like Tekoporã (the Conditional Cash Transfer program) use In the long run, Paraguay shows a relatively constant a first-stage targeting based on geographic location per capita GDP even while the LAC region average (combined with a second-stage proxy targeting for continues to increase, pointing to a Paraguayan the selection of specific households) and hence, the economy that has not grown sufficiently. Given the selection of the areas to be included might change once ix importance of growth for poverty reduction, it is the new estimates are taken into consideration. The distribution of resources to fight poverty between transportation spending and time, lower treatment costs urban and rural regions should be conditional to the of small medical problems compared with hospitals, and segment of the poor being targeted. The findings because they promote the use of preventive medicine, detailed above suggest that poverty would be reduced by far the cheapest way to improve health conditions. more quickly if programs aimed at the extreme poor have a heavier rural component, programs aimed at all Increase the productivity of the labor force the poor have both urban and rural components, and through better education and less informality those focused on the non-extreme poor have a heavier Central Urban component. Other structural problems that affect poverty in the urban sector are low productivity and the low Invest in secondary education, include education level of workers. A focus on formalizing strategies to reflect changes in demand, and the labor force, by decreasing entrance costs into the invest more in small health centers formal sector and increasing the benefits to small firms of formalizing, would increase the impact on poverty Paraguay should invest in secondary education for reduction. According to the assessment by the ILO both the poor and the non poor. In addition, one should (2003), informality is the outcome of several factors such take a deeper look into the reasons for secondary school as inadequate norms or rigid laws for the development dropout and implement corrective actions targeting of firms and an inefficient system of incentives. both the poor and non poor. Improved retention at higher school years will require a multidimensional Given that the analysis shows a strong negative approach that combines a more attractive school correlation between poverty and educational system by highlighting the advantages of a secondary attainment, it is important to improve the quality degree, such as the highest quality of the labor force. of education and provide incentives for youth to complete their formal education. This would increase The analysis suggests the need to take into productivity, and also have a positive effect on Paraguay’s consideration changes in the demand for education ranking in the Human Opportunity Index, in particular to be able to adequate the supply of public schools. in light of the problems found in secondary education. The reduction in the size of Paraguayan families will For workers that are already in the labor force, one could lead to a stable and eventually decreasing school age consider training programs within their industries. population, that will affect the demand for primary education. The same population changes will impact Working women face higher rates of secondary age students (several years after the effect underemployment, higher rates and duration of on primary is felt). At the same time, if the efforts of unemployment, higher levels of informality, and increased enrollment are successful, they will increase lower wages. Women are working fewer hours than demand and the expectations for better school access. they would like, pointing at an under-used work force Finally, these changes have to be differentiated among in Paraguay. In addition, the gender gap in income the various geographical areas of the country, mainly cannot be explained by observable characteristic such between urban and rural households, but also between as ethnicity, age, education, demographics or other the Central region and the rest of the country. labor variables, suggesting that further study is needed Executive Summary to understand and correct the gap. The health data suggests that the health situation needs to be substantially improved, and in order to In rural areas: improve land rental and sales reach the poor, the number of health centers should markets, increase access to financial markets, increase. Experiences from other countries suggest that generate a good investment climate and investment in public health centers or similar health improve the human capital of the poor establishments is pro-poor and an efficient way to x improve basic health conditions in the country due to More secure and unambiguous property rights over its proximity to users and the consequent reduction in land ownership could increase the productivity of the poor. This would increase the dynamism of land homogenization of targeting instruments. In order to rental markets, allowing those markets to transfer lands give a rapid response to this crisis, and to fulfill campaign to more productive users and uses, therefore probably promises, the administration engaged in an expansion increasing the country’s agricultural productivity and of the assistance component of social protection. especially helping the poor. There is ample evidence that The need for further improvements in the targeting weak property rights or restrictions on leasing constrain instruments to be used, its update and customization land rental transactions that improve productivity and is more urgent given the expansion of social assistance the rural poor may be excluded. Some of the programs programs. If programs are to attend different groups that could increase the security of land property rights of the population, it would be advisable to design include land titling, land registration and management, specific tools for this purpose. The recent initiatives to and dispute resolution when there are overlapping reduce the overlap of programs (especially the ones claims. involving cash transfers) seem to be moving forward in the right direction to reach a unified social protection To be effective, any approach to land reform must system. The expansion of the CCT programs has already be integrated into a broader rural development provoked the uni�cation of the operations manuals of strategy - using transparent rules, offering clear and the programs. Nevertheless, the public credibility of unconditional property rights, and improving incentives such improvements also need transparency, and for this to maximize productivity gains. A reform can enhance the regular monitoring and evaluation of the programs access to land for the rural poor, but to reduce poverty is essential. and increase efficiency requires a commitment by government to go beyond providing access to ensuring Necessary efforts to undertake a new income the competitiveness and sustainability of beneficiaries and expenditure survey as market-oriented smallholders. It is important that the DGEEC conduct a new income For the rural poor, access to financial markets is as and expenditure survey to update both the food and critical as access to land. Innovative policies to enhance non-food baskets - from which the baseline poverty access to formal credit in rural areas are needed. One and indigence lines are derived. The revision of the such policy would be to establish a credit bureau methodology used for measuring poverty was only the able to collect the credit history of rural borrowers first step of a two-step strategy. The baseline poverty from Monetary and Financial Institutions (MFI’s) and and indigence lines are still based on the 1997/98 commercial banks. Integrated Household Survey, and may no longer reflect the true household consumption bundles, A dynamic rural economy will be needed to ensure particularly for the poor. The World Bank is prepared to growth in labor demand and a reduction in rural provide non-lending technical assistance and capacity poverty in Paraguay. The most basic policy element for building during the 2011-12 fiscal year to support the the government of Paraguay to ensure a dynamic rural second stage of this process, but it is important that economy is perhaps to promote a good investment the government of Paraguay provide all the necessary climate in rural areas. Enhancing the human capital of resources for the DGEEC to carry out this new survey. Poverty Assessment the rural poor would also contribute to that objective, increasing the productivity of those who opt for generating income via labor markets. Unify the fragmented set of social protection programs, improving their targeting, and investing in their monitoring and evaluation Social Protection in Paraguay should gradually unify xi its varied set of programs, but that does not mean the introduction Background, Objectives, and the Poverty Assessment Process Figure 1: GDP per capita and GDP annual growth, Latin America, circa 2008. Paraguay experienced six consecutive years of positive Mexico output growth between 2003 and 2008. The Gross Chile 1.77 3.16 Argentina 6.75 Domestic Product (GDP) per capita averaged an annual Venezuela, RB 4.82 GDP annual growth (%) Uruguay 8.89 growth of 2.7 percent for that period (with an average Panama Costa Rica 9.18 2.60 annual GDP growth of 4.6 percent), in contrast to the Brazil 5.07 Colombia 2.53 Peru 9.76 negative growth of the previous five years. However, Rep. Dominicana 5.25 Ecuador 6.51 GDP per capita for all these years except 2008 continues El Salvador 2.54 4.02 Guatemala to be below the observed value in 1995. In addition, Paraguay Bolivia USD 4345 5.77 6.14 Honduras 3.95 even taking into account the recent growth, Paraguay, Nicaragua 3.50 a landlocked country with a relatively large rural sector 0 2000 4000 6000 8000 10000 12000 14000 (45 percent of employed workers2), ranked near the GDP per capita, PPP (constant 2005 international $) bottom in terms of GDP per capita in Latin America Source: World Bank, World Development Indicators 2009. in 2008 (US$4345 in constant 2005 Purchasing Power Parity [PPP]; see Figure 1). global financial crisis. However, due to data limitations (the last available household survey is for 2008), it is not Poverty Assessment Paraguay’s relative position in Latin America may have yet possible to estimate the poverty impacts for 2009. deteriorated even more in 2009 due to experiencing Therefore, this report will focus primarily on the data one of the worst contractions in output in the region. available between 1997 and 2008.3 Preliminary 2009 growth estimates by Paraguay’s Central Bank suggest that GDP shrank 3.8 percent, a The macroeconomic stability of the last few years result of the severe drought early in the year and the and the election of a new Government have provided 2 The average for Latin America is 22.3 percent (World Development Indicators, 2007). 3 Editor’s Note: The present work comprises data available from 1997 to 2008. However, an update to 2009 is included in the Annex: 1 “Crisis, Poverty, and Drought in 2009.� Paraguay the opportunity to strengthen its poverty economic indicators. (A brief update to 2009 is reduction and human development strategies. These included as an annex.) strategies are an important aspect within the broad b. Analyze the determinants of poverty and inequality. pillars the Government has set out. The Poverty c. Include an analysis of Inequality of Opportunities Assessment can serve as a tool for the Administration through the Human Opportunity Index. through its analysis on issues of poverty, inequality, d. Explore and analyze Paraguay’s factor markets employment, land, and the recently expanded cash (land and labor), with special attention to issues of transfer programs. under-employment, informality, women’s economic participation, and access to land. The report has been developed by the World Bank and e. Simulate the ex-ante impact on poverty of an is based on the extraordinary work carried out jointly expansion of the conditional cash transfer program with the General Directorate of Statistics, Surveys (Tekoporã) as well as of the creation of a new law and Censuses (DGEEC) to review and update the mandating non-contributory pensions to the elderly methodology for measuring poverty in the country. poor. The report is part of a programmatic poverty work plan f. Assess the main implications of the analysis of (a) – that includes not only the analytical work contained (e) for public policies to accelerate poverty reduction here, but also a strong component of capacity and promote more equity. building and technical assistance that was requested in December 2007. Supported by the World Bank and This Poverty Assessment is part of a continuing government officials, the DGEEC undertook the most engagement between the World Bank and the comprehensive and participatory work thus far to Government of Paraguay. These objectives reflect review and update the country’s poverty measurement knowledge gaps and interest identified by the World methodology in line with current international best Bank during continuous consultations with key practices. The previous methodology was in line with government counterparts in the previous and the the then existing literature (1997-1998). The process current Administration. The counterparts included the was carried out in a manner that was transparent Ministry of Finance, the Social Cabinet, the Secretariat and credible, helping to build a national consensus of Social Action, and DGEEC, as well as local research around the new methodology and poverty rates. The institutes in Paraguay. Many of the objectives emerged process included the creation of an Interinstitutional as key themes in the early policy dialogue with the Committee comprised of representatives from the new administration, and in broad consultation carried Government of Paraguay, Academia, the private sector out throughout 2008. The report has benefitted from and civil society, as well as international organizations. inputs and comments from Paraguay’s government This Committee has spanned both the previous and representatives. the current Administration. The process carried out between February 2008 and November 2009 has placed The report is organized into 5 chapters. Chapter 1 Paraguay as a pioneer at the Mercosur level in terms of presents the evolution of growth, poverty and inequality the use of new tools for the measurement of poverty in Paraguay, comparisons with other countries in the and offers a clearer vision of the dimensions of poverty region, and a section on the revision of the poverty in the country. measurement methodology. Chapter 2 analyzes the determinants of poverty and inequality, characterizes With this background, the specific objectives of the who are the poor, and presents Paraguay’s results for Introduction Poverty Assessment (beyond the technical assistance the Human Opportunity Index. Chapter 3 focuses on and capacity building on the poverty measurement aspects and tendencies of the urban labor market. methodology) will be to: Chapter 4 provides an analysis of the rural dimension of poverty with a focus on rural factor markets. Chapter a. Update the recent trends in poverty, inequality, 5 simulates the ex-ante poverty impacts of extending 2 income, and employment through 2008, including two cash transfer programs in Paraguay. a gender dimension and progress in other socio- economic Growth, Poverty, and inequality Introduction overall headcount poverty increases from 36.1 percent to 37.9 percent and extreme headcount poverty From 2000 to 2008, Paraguay shows improvements in increases from 18.8 to 19 percent (Figure 1.1). almost all socio-economic indicators. During the 2000s, Paraguay shows increased income, poverty reduction Similar to the poverty headcount rates, the per and improvements in several social indicators (Table 1.1). capita GDP improvements observed from 2000 to For example, Gross Domestic Product (in PPP dollars) 2008 almost disappear if a longer period of time is increased by 15 percent and overall poverty decreased considered. A similar trend to poverty can be observed by 14 percent (6.1 percentage points). According to the for real GDP per capita, where comparison values from revised official estimates, overall poverty decreased 1995 are very similar to the 2008 estimate (Figure 1.2).4 from 44 percent of the population in 2003 to 37.9 In addition, preliminary estimates by the Central Bank percent in 2008, while extreme poverty decreased by 2.2 of Paraguay suggest a 3.8 percent decrease in GDP percentage points during the five-year period. Important for 2009. Box 1.1 presents a projection of the possible reductions in mortality rates were achieved for children impact on poverty in 2009.5 under five and infants at birth (13-14 percent) and there was a notable 33 percent reduction in maternal What might look like important improvements could mortality at birth. Improvements were also reported be more a recovery than real advancement. Given the for access to improved water sources (12 percent) apparent worsening of per capita GDP and the poverty Estudio de Pobreza and secondary school gross enrollment (8 percent). headcount from 1997 to 2002, and their subsequent improvement from 2002 to 2008, special care should be But some of the improvements disappear if a longer placed in any type of time comparisons to make sure an period of time is used. If one is to analyze the evolution appropriate period of time is included. In addition, only of poverty over a ten year period the reduction reported a subset of the national household surveys are strictly before does disappear. Indeed, from 1998 to 2008, comparable (see Box 1.2). 4 All values from 1995 to 1997 are around $ 4,200 PPP, higher than any other year with the exception of 2008. 5 Editor’s Note: The present work comprises data available from 1997 to 2008. However, an update to 2009 is included in the Annex: 3 “Crisis, Poverty, and Drought in 2009.� Table 1.1: Key Socio-Economic Indicators in Paraguay 2000 – 2008 Index 2000 a 2008 a Change b Urban population (% of total) (2000-2008) 55% 60% 9% GDP per capita (constant 2005 PPP $) (2000-2008) c $ 3,789 $ 4,345 15% GINI index (1999-2007) 0.569 0.532 -7% Extreme poverty rate (Headcount rate) (2003-2008) d 21.2% 19.0% -10% Overall poverty rate (Headcount rate) (2003-2008) d 44.0% 37.9% -14% Maternal mortality ratio (per 100,000 live births) (2002-2006) 180.0 121.4 -33% Mortality rate, infant (per 1,000 live births) (2000-2007) 27.8 24.3 -13% Mortality rate, under-5 (per 1,000) (2000-2007) 33.4 28.8 -14% Measles Immunization rates (12-23 montholds) (2000-2007) 92% 80% -13% DPT Immunization rates (12-23 montholds) (2000-2007) 68% 66% -3% Pop. with access to improved sanitation facilities (2000-2006) 67% 70% 4% Population with access to improved water source (2000-2006) 69% 77% 12% Net primary school enrollment rate (1999-2005) 96% 94% -2% School enrollment, secondary (% gross) (2000-2005) 61% 66% 8% a Actual years in parenthesis next to each indicator; b Changes in percentage (not in percentage points); c 2002 value was almost the same as 2000: PPP $ 3,715; d From 2003 the questionnaire and survey design provide highly comparable poverty estimates. Source: Dirección General de Estadística, Encuestas y Censos (DGEEC) for poverty estimates; World Bank, World Development Indicators 2009 for all others. Figure 1.1: Poverty Evolution Figure 1.2: Real Per Capita GDP in Paraguay 1997/98-2008 Evolution in Paraguay 1995-2008 60 4.600 Poverty Rate GDP per capita 2005 Real PPP 50 4.400 36.1 37.9 4.200 40 Percentage 4.000 30 3.800 20 3.600 10 18.8 19.0 Extreme Poverty Rate 3.400 0 3.200 1997-98 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: World Bank estimates based on DGEEC’s updated methodology, 2009. Source: World Bank, World Development Indicators 2009. The GDP per capita gap between Paraguay and per capita in 1990 (around PPP $4,100), while by 2008 Chapter 1 the LAC average has increased since 1990 and has Peru’s GDP almost doubled Paraguay’s. The constant continued to increase at a higher rate since 2003. per capita GDP in the face of increased values for From 1990 to 2008, the GDP per capita gap between Latin America leaves Paraguay with an important Paraguay and LAC increased by 84 percent (Figure income gap relative to neighboring countries and 4 1.3). Comparisons to a country like Peru are the most those of similar resources, and facing important striking since both countries had a very similar GDP challenges. With the exception of primary net school enrollment, Updating the Methodology for Paraguay ranks below the Latin America and Poverty Measurement in Paraguay Caribbean (LAC) average in all other selected indicators. Compared to the LAC average, Paraguay Before turning to the poverty and inequality results, shows a higher income inequality (as measured through this chapter includes a brief explanation of the new the Gini), higher maternal and under-five mortality methodology for measuring poverty in Paraguay. New rates, lower immunization rates, access to sanitation and conceptual and methodological developments in the piped water, secondary school enrollment, and a lower literature on poverty measurement, comparability ranking in the Human Development Index (Table 1.2). Comparisons to other countries in the region also rank Paraguay as having some of the worst socio-economic Figure 1.3: GDP Per Capita indicators in the region.6 Evolution in LAC, 1990-2008 GDP per cap, PPP (Const. 2005 PPP $) 14.000 Argentina Despite the long term implications and unfavorable 12.000 Uruguay comparisons to LAC and countries in the region, LAC 10.000 Paraguay has experienced real improvements in 8.000 Brazil Peru wellbeing during the 2000s. Being able to understand 6.000 Paraguay the source of the improvements in its socio-economic 4.000 indicators is important to build and expand the 2.000 appropriate programs and policies that have made - 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 possible such advancements. The present study provides possible explanations and policy implications associated with the changes observed. Source: World Bank, World Development Indicators 2009. Table 1.2: Socio Economic Indicators: Paraguay Compared to Latin American and the Caribbean and to Selected Countries Indicator Paraguay Argentina Brazil Costa Rica Peru LAC GINI index a 53 50 b 55 47 b 50 c Immunization, measles (% 12-23 month-olds) (2007) 80 99 99 90 99 93 Immunization, DPT (% children 12-23 month-olds) 66 96 98 89 80 92 (2007) Access to improved sanitation facilities (2006) 70 91 77 96 72 78 Access to improved water source (2006) 77 96 91 98 84 91 Maternal mortality ratio (/100,000) (2005) 150 77 110 30 240 130 Mortality rate, under-5 (per 1,000) (2007) 29 16 22 11 20 26 Poverty Assessment School enrollment, primary (% net) a 94 b 98 c 93 98 96 93 School enrollment, secondary (% gross) a 66 b 84 c 100 87 98 88 Human Development Index: position (2007/2008) 101 49 75 54 50 d a 2007, otherwise indicated ; b 2005; c 2006 ; d Paraguay 101st in full 2007 ranking; 28th in LAC, with only seven other countries ranking below (El Salvador, Honduras, Bolivia, Guyana, Guatemala, Nicaragua and Haiti). Source: UNDP Human Development Reports (2002 and 2009) for the Human Development Index; Ministry of Education for Costa Rica School net primary enrollment rate; WDI for all other indicators. 5 6 Only some countries in Central America and the Caribbean present an overall condition worse than Paraguay. problems with the 2006 household survey, as well as methodological proposals, and the results of the analysis a series of unexpected results in the 2005 and 2007 presented by the DGEEC, and (ii) coming to an agreement surveys, led to the request in December 2007 for World on the methodological changes to be implemented by Bank assistance to improve and update the poverty the DGEEC, finding consensus on the most appropriate measurement methodology in Paraguay. DGEEC, as methodology, its updating, and future measurement, with other Statistical Institutes in the region, estimated taking into consideration that the full methodology is poverty based on poverty lines using a methodology public and replicable. This committee can serve as an available in 1997, when the country’s first Encuesta example to achieve consensus on the methodology for Integrada de Hogares (EIH, spanish for Integrated the measurement of monetary poverty in the country Household Survey) was realized. This survey included and assure the mechanism for the regular update of a household expenditure module and constitutes still this methodology and its consistency with the best today the base line for poverty measurement. However, international practices (Box 1.3). ten years after the establishment of the 1997/98 base poverty line, new conceptual and methodological Therefore, the new poverty measurement has benefited developments in the literature (for example, on the from an ample and transparent national consensus, inconsistency of poverty lines) made a retrospective and reflects the best current international experiences. revision of the historical series necessary. In addition, The Advisory and Interinstitutional Committees have unexpected results, such as lower levels of rural than accompanied the process through both the previous urban poverty, reinforced the need to update Paraguay’s and the current Administrations. The rigorous and methodology7. detailed work of 20 months, that began in February 2008, included capacity building and ownership of the With the objective of assuring the quality of the new methodology by DGEEC’s technical team and by processes that are realized at each stage, two working members of the Interinstitutional Committee, who can groups were created: an Advisory Committee and an themselves now become instructors. For the adoption Interinstitutional Committee. The Advisory Committee, of the new methodology, DGEEC counted on the a small team led by the DGEEC and including technical and financial support of the World Bank, and international technical experts, is in charge of evaluating the support of the Interinstitutional Committee. The the quality of the data and the methodologies on which methodological adjustment that was carried out (see the poverty rates are based, such that the methodology Box 1.4 for the complete list) has placed Paraguay as is transparent, public, and replicable, adopts the best a pioneer at the Mercosur level in terms of the use of international practices and counts with the necessary new tools for the measurement of poverty and offers interinstitutional consensus. The Advisory Committee, a clearer vision of the dimensions of poverty in the therefore, evaluates the existing methodologies, makes country (Box 1.3). proposals to the Interinstitutional Committee, and implements its recommendations. The Interinstitutional A Press Conference was held on November 23, 2009 to Committee (see Box 1.3), which includes the Advisory present the summary report“Los resultados de la revisión, Committee and representatives from the government, actualización y mejora de la metodología de medición academic institutions, research institutions, business de la pobreza en el Paraguay. Período 1997-2008�. The guilds, civil society, as well as international organizations, purpose of the document is to offer information on will be responsible for: (i) knowing the diagnostics, the the main indicators of poverty in Paraguay, pertaining 7 The welfare measure used in Paraguay and throughout this study is household per capita income. Poverty is defined as having per Chapter 1 capita income below the poverty line, while extreme poverty is defined as having per capita income below the level of the extreme poverty line. The extreme poverty line is set at the cost of obtaining the minimum requirement of calories intake per person per day. The estimation of the poverty lines in based on the expenditure module included in the 1997-98 Household Survey. To calculate the extreme poverty line, it was necessary to determine the food consumption patterns of the reference population (see Box 1.5). This ‘food basket’ was then analyzed for caloric content and adjusted to ensure that the minimum daily requirements of calories are obtained. Finally, the resulting basket is valued using price data from the household survey. The general poverty line is simply the extreme line plus an allowance for non-food consumption. This allowance is estimated using the Engel coefficient by, first, determining the share of 6 total consumption devoted to non-food consumption among those whose total consumption is at or near the extreme poverty line. This percentage is added to the value of the food poverty line. The poverty lines are updated using CPI for Metropolitan Asuncion. to the period mentioned, with the objective that they However, important changes can be observed in the serve for the design, implementation and evaluation Urban/Rural headcount rates with the new estimates. of public politics tending toward improving the living The new methodology introduces important changes conditions of the Paraguayan population (Box 1.5). in the poverty estimates, especially in rural areas whose overall poverty estimates increase by 50 percent and The World Bank will provide the second stage of the extreme poverty estimates by almost 70 percent. New non-lending technical assistance and capacity building urban estimates report smaller changes: a reduction of in FY11-12. This second stage will focus on updating 8 percent for extreme poverty and 11 percent for overall Paraguay’s base poverty and extreme poverty lines, poverty (Table 1.3). The new estimates also show a much which are based on a basic basket of goods from higher poverty rate in rural than in urban households, the 1997/98 Encuesta Integrada de Hogares, and while the previous estimates were statistically the same therefore may have ceased to reflect true household between the two. consumption bundles, in particular for the poor. In order to move forward with this second stage, it is The change in ranking between urban and rural necessary that DGEEC carries out a new household households has important implications for the policies survey of family incomes and expenditures to update and programs designed to reach the poor. The new both the food and non-food consumption baskets poverty estimates are an indication for the government from which one can derive new baseline extreme and to shift some of the resources allocated to fight poverty moderate poverty lines. and for social programs from urban to rural areas. Also, the government of Paraguay should reformulate some Measuring Poverty in Paraguay components of the agricultural agenda and general rural policies in order to address the much higher overall The new methodology results in a small and poverty and especially extreme poverty levels revealed consistent increase of the national poverty by the improved methodology. headcount. The net impact of all the changes introduced to improve poverty measurement is an Special attention should be given to targeting increase in the percentage of the poor between 0.4 mechanisms to reach the poor. Social programs that and 5.6 percentage points. On the whole, the changes use geographic characteristics as the tool or as one of the imply an upward shift of both overall and extreme poverty estimates over time (Figure 1.4). In 2008, the impact in the number of poor is an increase of 285 thousand or an extra 4.6 percent of Paraguayans in Figure 1.4: Impact of the poverty and extreme poverty.8 New Methodology on Poverty Estimates 1997/98-2008 The moderate increase in the poverty estimate does not imply a change in government policy. 60 Compared to the yearly changes in poverty observed 50 Moderate Poverty New since 1997, the new estimates are equivalent to the 40 Percentage Poverty Assessment one or two year average variation already observed. 30 Old Since the vast majority of public social spending in 20 New Paraguay is a product of long term strategies (health, 10 Old Extreme Poverty education, sanitation) and social protection uses only 0 2.19 percent of government social spending9, very 1997-98 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 few adjustments are justified by the reported poverty increase estimates. Source: World Bank estimates based on DGEEC updated methodology review, 2009. 8 With the new 2008 estimates, the number of poor and extreme poor are both 285 thousand people higher. Estimates of the extreme poor increase from 884,000 to 1,169,000 people and of the poor from 2,053,000 to 2,338,000. 7 9 See Table 1.9 for social spending as percentage of GDP. Table 1.3: Old and New Headcount Rates by Area, 2008 2008 Change Old New % points % Urban Extreme Poor 11.5 10.6 -0.9 -8% All Poor 33.7 30.2 -3.5 -11% Rural Extreme Poor 18.3 30.9 12.5 +68% All Poor 32.7 48.8 16.1 +49% Source: World Bank estimates based on EPH, DGEEC Paraguay tools to identify the poor should take into consideration followed by “Resto Urbano�, urban regions outside the new poverty estimates. Since the new methodology the central region of the country (27 percent), Urban has a very different impact in households according to Central (35.3 percent), and finally, rural households (at their place of residence, any type of geographic ranking a high 48.8 percent; see Table 1.4). used to target social program can significantly change. Programs like Tekoporã (Conditional Cash Transfers) Over half of the poor and more than two thirds of use a first stage targeting based on geographic location extreme poor persons are located in rural areas of (combined with a second stage proxy targeting for Paraguay. With only 41.4 percent of the population, specific household selection) and hence, the selection rural households have a disproportionally high amount of the areas to be included might change once the new of the poor, especially the extreme poor. By 2008, the estimates are taken into consideration. Rural is the only region in the country with poverty rates above the national average. Moreover, because Finally, it is important to understand that household more people live in urban areas, the total poor are more conditions did not change because of the improvement equally divided between urban and rural households in the way poverty is measured in Paraguay: there are than the headcount rate shows. no more or less poor households due to methodological changes10. The only change is in the understanding of The distribution of resources between urban and poverty and the perception of who is poor and who is rural regions to fight poverty should be conditional not. In the past, we were underestimating rural poverty on the segment of the poor being targeted. Paraguay and overestimating urban poverty; the new statistics has an unusual distribution of the poor population: a are a better approximation to reality than the previous heavy concentration of the extreme poor among rural estimates. households (similar to many other countries in the region), but with a heavier concentration of urban households Poverty in Paraguay among the non extreme poor. Rural households make up 67.5 percent of the extreme poor, 53.5 percent of all By 2008 almost two out of five Paraguayans are poor the poor, and only 32.4 percent of the “non-extreme� and one out of five lives in extreme poverty. Moreover, poor.11 A program aimed at the extreme poor should within a longer 10 year span, poverty has not changed have a heavier rural component; while one aimed at all as the 2008 and 1997/98 estimates show similar values the poor could have similar urban and rural components. Chapter 1 (Figure 1.1). Within the country, the incidence of poverty shows important variations across regions; the The poverty ranking has not changed over the last six lowest incidence is reported in Asunción (21.5 percent), years between the different regions of the country. 10 There is an important difference between the number of poor households (the reality) and the estimated number of poor households; 8 and it is in the latter where changes took place. 11 “Non-extreme� poor are households with per capita income above the extreme poverty line but below the overall poverty line. Table 1.4: Poverty Headcount and Contribution to Poverty, Paraguay 2008 Headcount rate (%) Contribution to Poverty (%) % of Population All Poor Extreme All Poor Extreme Poor Poor National 100 37.9 19.0 100 100 Region Asuncion 8.4 21.5 6.7 4.7 2.9 Urban Central 27.1 35.4 10.6 25.3 15.1 Other Urban 23.1 27.1 11.9 16.5 14.5 Rural 41.4 48.8 30.9 53.5 67.5 Area Urban 58.6 30.2 10.6 46.5 32.5 Rural 41.4 48.8 30.9 53.5 67.5 Source: World Bank staff calculations based on EPH, DGEEC Paraguay From 2003 to 2008, Asunción has maintained the lowest poverty rate and the rural region the highest Figure 1.5: Overall Poverty by Regions, rate, with the other two urban areas ranking in the Paraguay 2003-2008 middle. Contrary to perceptions in the country, urban households outside the central region of the country are better off than urban households in the Central 60 region. That is true despite the very uncommon Rural 50 definition of urban areas used in Paraguay, allowing 40 Urban Central for households without basic services or paved roads Percentage 30 Urban Rest to be classified as urban as long as they fall within a 20 block layout. Asuncion 10 One possible explanation for the higher poverty rates in 0 2003 2004 2005 2006 2007 2008 the central urban region can be the high concentration of vulnerable households in specific urban areas. A Source: World Bank sta calculations based on EPH, DGEEC Paraguay. concentration of poor households within the main urban areas is common and should be given special 2008 was Gs.130,238, or a total national yearly value of consideration. Gs.3,632,944 million (USD 727 million) representing 5.2 percent of Paraguay’s 2008 GDP. Despite the high levels of poverty, the household income required to eliminate extreme poverty was An increase of income equivalent to 1.4 percent of GDP Poverty Assessment not very high. The cost of closing the gap between is not an unreasonable amount of money. However, household income and the poverty line can be assigning such an increase for the precise households estimated. In 2008, an additional average Gs. 69,459 and by the precise amount per household is difficult. per month (USD 13.9) per extreme poor person was For a government program to achieve such an income necessary to eliminate extreme poverty; in other words, increase many more resources would be necessary in 2008 a total of Gs. 971,357 million or 1.4 percent of (social programs have administrative costs) and only GDP in additional yearly income was necessary to a partial success would be achieved given the lack of eradicate extreme poverty. For overall poverty the perfect knowledge to identify all the extreme poor and necessary income is much higher: the necessary average the lack of perfect targeting (as in all countries), that 9 additional monthly income for each poor person in could result in resources being received by non poor Table 1.5: Poverty Indicators by Region and Area, Paraguay 2008 Poverty line Headcount # of poor Poor avg. Poverty Severity (Gs.) a rate (,000) Income Gs. a Gap Index Index OVERALL POVERTY (all poor) National 37.9% 2,324.6 224,261 14.3 7.4 Region Asuncion 474,703 21.5% 109.1 325,074 6.7 3.0 Urban Central 474,703 35.4% 588.0 323,210 11.4 5.0 Other Urban 338,902 27.1% 383.7 220,187 9.5 4.7 Rural 291,948 48.8% 1,243.7 169,898 20.4 11.2 Area Urban 30.2% 1,080.9 286,820 9.9 4.6 Rural 291,948 48.8% 1,243.7 169,898 20.4 11.2 EXTREME POVERTY National 19.0% 1,165.4 144,575 6.3 3.0 Region Asuncion 277,766 6.7% 33.7 206,140 1.7 0.6 Urban Central 277,766 10.6% 175.9 207,855 2.7 1.0 Other Urban 213,162 11.9% 169.0 147,428 3.7 1.6 Rural 197,247 30.9% 786.8 127,180 11.0 5.6 Area Urban 10.6% 378.6 180,730 2.9 1.2 Rural 197,247 30.9% 786.8 127,180 11.0 5.6 a Values in monthly per capita Guaranies Source: World Bank staff calculations based on the 2008 household survey, EPH, DGEEC Paraguay households. Nevertheless, and only to compare orders Poverty Gap Index; and (iv) the more severely of magnitude, the Conditional Cash Transfer programs extreme poor people as reported by the Severity Tekoporã and Pro País together disbursed Gs. 120,000 Index. Indeed, rural households have the worst FGT12 million in 2009, or 0.172 percent of GDP – barely above poverty and extreme poverty indicators compared to one tenth of the sum necessary to eradicate extreme the other regions reported in Table 1.5. For example, poverty if there were perfect targeting. rural households have an overall poverty Gap Index of 20.4 or around double the value in other regions, and Paraguay has more than two million poor and over for extreme poverty the Gap Index of 11.0 is around one million extreme poor. Even with substantially three times the values reported in the urban regions. lower poverty and extreme poverty lines (68 and 71 This implies that not only more poor and extreme poor percent of the poverty and extreme poverty line values, exist in this area, but they are also further from the respectively, of the Central Urban region), the number poverty and extreme poverty lines and with the worst of poor in rural areas is more than 1.2 million and the distribution among the poorest of the poor. number of rural extreme poor is more than double the Growth Incidence Curves Using Chapter 1 urban extreme poor. Household Per Capita Income Rural households have not only the highest headcount rate but also: (i) the highest number of poor; (ii) the It is possible to develop a richer characterization of lowest average income for the poor, (iii) the highest the patterns of income growth across the income 10 12 Foster, J., J. Greer, and E. Thorbecke, 1984, A Class of Decomposable Poverty Measures, Econometrica 52, 761-765. distribution by computing growth incidence curves.13 The growth incidence curves show the proportional income change for each percentile of income in a given Figure 1.6: Growth Incidence Curve, Paraguay (2003-2008) in Guaraníes of 2008 period. They have been used frequently in recent years to study the extent to which different sub-groups of Total (years 2003 and 2008) the population participate in the growth process. GICs Growth-incidence 95% con dence bounds Growth in mean Mean growth rate have two particularly interesting characteristics: (i) they 17 Annual growth rate (%) can be used to compute the average per capita income 15 growth rate experienced by different segments of the 13 population; and (ii) they are able to capture much richer 11 patterns of income inequality than those captured by 9 Gini coefficients or by the analysis of quintile income 7 shares (presented in the next subsection). 1 10 20 30 40 50 60 70 80 90 100 Per capita household income percentiles Average income grew across the whole income Source: World Bank sta calculations based on EPH, Paraguay. distribution for the years 2003-2008, but more so for the poorer population. Figure 1.6 presents the GIC at the national level for Paraguay for this period. The curve the poorest 5 percent of the urban population. In urban is well above the horizontal axis, implying income areas, the income growth has considerably benefitted growth for all the population, with an average annual the bottom percentiles of the income distribution, with rate of growth of around 12 percent. The downward a rate of pro-poor growth for this sub-period of around slope of the curve indicates that income grew faster 17 percent per year. In rural areas the growth rate of the among those in the lower percentiles than among lower percentiles are closer to the mean growth rate those in the higher percentiles. of 11.2 percent (Figure 1.7). This difference in income growth rates can explain to some extent the faster Although the pattern is similar and still pro-poor at decline in poverty rates in urban areas (-7.2 percentage a more disaggregated level, the GIC shows that the points from 2003 to 2008) with respect to the decline in poorest 5 percent of the rural population benefitted the rural poverty rate (-3.7 percentage points) observed less than those in the 10-20 percentile, and less than during the same period. Figure 1.7: Growth Incidence Curve by area, Paraguay (2003-2008) in Guaranies of 2008 Urban Rural Growth-incidence 95% con dence bounds Growth-incidence 95% con dence bounds Growth in mean Mean growth rate Growth in mean Mean growth rate 17 Annual growth rate (%) Annual growth rate (%) 17 15 15 Poverty Assessment 13 13 11 11 9 9 7 7 1 10 20 30 40 50 60 70 80 90 100 1 10 20 30 40 50 60 70 80 90 100 Per capita household income percentiles Per capita household income percentiles Source: World Bank sta calculations based on EPH, Paraguay. 13 These curves, introduced by Ravallion and Chen (2003), are simple and illustrative ways to analyze the changes in household per 11 capita income across the income distribution. Inequality the well being of all citizens, especially the poor, when there is also an overall reduction of inequality. Beyond The distribution of household per capita income these benefits, however, another issue arises. Since is highly skewed towards the richest 10 percent of the existing income inequality may originate at least households, who have more than 40 percent of total partially in the existing inequality of access to basic income. However, income inequality measured by opportunities among children, it is also important to decile income shares and ratios has improved since analyze the latter. This is done in the subsection on the 2003. Income shares per deciles increased for the Human Opportunity Index in chapter 2. lowest nine deciles, improving net distribution for the first seven deciles and for the 10th decile14 (Table 1.6). Poverty Decomposition The ratio of income shares between the tenth decile (highest income) to the first decile (lowest income) A Poverty Decomposition is useful to understand to has substantially improved from 41.7 in 2003 to 29.9 what extent a given change in poverty is due to a rise in 2008. In other words, for each dollar the average in mean income or due to a change in the distribution person in the poorest decile earned in 2003 the the of income. The growth-inequality decomposition average person in the wealthiest decile earned almost introduced by Datt and Ravallion (1992) quantifies 42 dollars. By 2008 the disparity was reduced to 30 the relative contributions of economic growth and dollars – a 28 percent reduction in inequality between redistribution (e.g., changes in inequality) to changes the two extreme deciles. in poverty. The importance of the distributional effect in poverty reduction has been highlighted by Ravallion The GINI, a more comprehensive measure of (2007). He finds that when countries are more unequal, inequality, also shows important reductions in the last overall growth translates less successfully into higher five years. With the exception of the 2005-2006 period, incomes for the poor, and he suggests that more the GINI index has been decreasing in the last five years unequal countries may often grow less rapidly in the (Figure 1.8). Taking into consideration how difficult it first place. is to decrease the GINI coefficient by 0.043 points or 8 percent in five years, a very important reduction of The growth effect in Paraguay was more important inequality is apparent (Table 1.7). All other inequality than the redistribution effect in explaining the indicators estimated in this report also show important decrease in the national poverty rate between reductions during the 2003-2008 period: for example, 2003 and 2008. Table 1.8 illustrates the growth and the Theil index decreased 20 percent,15 the Atkins redistribution components for the change in poverty coefficients decreased between 8 and 16 percent, and and extreme poverty at the national, urban, and rural the Generalized Entropy Measure reported reductions levels between 2003 and 2008. Moderate poverty in as much as 50 percent.16 Paraguay fell 6.1 percentage points in this period, from 44 to 37.9 percent, a result of both growing incomes The overall reduction of inequality in Paraguay implies (-4.1) and an improvement in income distribution an improvement of conditions for the lower income (-1.5). On the other hand, although rural poverty groups since average incomes did not decrease. also decreases during this period, the results of the Another possible positive externality of this reduced poverty decomposition highlight that in this case inequality is that, studies show, in some cases economic the growth and distribution effects tug in opposite development can have a stronger impact on improving directions, and in the case of extreme rural poverty, Chapter 1 14 Distribution improvements are achieved by increasing the share of income for deciles with less than 10 percent of total national income and by decreasing the share for deciles with more than 10 percent of total national income. 15 Different Atkinson estimates are more sensitive to different sections of the population. 16 The generalized entropy measure is defined as: were yi is the income for household i, µ(y) is the average income and n is the sample size. When α = 0, the generalized entropy measure is called the Logarithm of the mean 12 deviation, for α = 1 the Theil Index, and for α = 2 half of the variance ratio square. Table 1.6: Income Shares by Deciles and Ratios, Paraguay Decile 2003 2004 2005 2006 2007 2008 1 1.1 1.3 1.2 1.1 1.1 1.3 2 2.1 2.4 2.4 2.2 2.3 2.4 3 3.0 3.3 3.4 3.2 3.3 3.4 4 3.9 4.2 4.4 4.1 4.3 4.4 5 4.9 5.2 5.5 5.3 5.4 5.5 6 6.2 6.6 6.6 6.7 6.7 6.8 7 8.0 8.2 8.4 8.4 8.3 8.7 8 10.7 10.8 11.2 10.9 10.8 11.2 9 15.5 15.5 15.9 15.3 15.4 15.9 10 44.7 42.4 41.0 42.8 42.5 40.4 Ratios Decile: 10/1 41.7 31.8 34.2 37.8 38.3 29.9 Percentile: 90/10 12.1 10.0 10.6 11.6 11.4 10.1 Percentile: 95/80 2.3 2.2 2.1 2.2 2.2 2.2 Source: World Bank staff calculations based on EPH, DGEEC Paraguay Table 1.7: Inequality Measures Coefficients 2003 2008 Change Points % Gini 0.558 0.515 -0.043 -8% Theil Index 0.703 0.562 -0.141 -20% Coeff. of Variation 2.639 1.861 -0.778 -29% Atkinson Coefficient e=0.5 0.266 0.225 -0.042 -16% e=1.0 0.436 0.380 -0.056 -13% e=2.0 0.656 0.604 -0.052 -8% Generalized Entropy Measure c=0.5 0.573 0.478 -0.095 -17% c=1.0 0.703 0.562 -0.141 -20% Poverty Assessment c=2.0 3.482 1.732 -1.750 -50% Source: World Bank staff calculations based on EPH, Paraguay the redistribution effect dominated the growth effect. Non-Income Measures of Well-Being These results seem to correspond with the Gini results that show that although income inequality at the It is always a good practice to review other measures national level has been decreasing over time (as seen of wellbeing besides household income. Household earlier), the Gini in the rural sector is practically the infrastructure characteristics are a good indicator 13 same in 2003 and 2008 (Figure 1.9). of people’s socioeconomic status and tend to be Figure 1.8: Gini Coefficient Figure 1.9: Gini Coefficient of per capita household income by area 0.570 0.70 Rural 0.58 0.560 0.558 0.60 0.56 0.54 0.56 0.50 0.52 0.550 0.538 0.50 Gini Coe cient 0.540 0.534 0.52 0.49 0.50 0.51 0.40 0.48 0.46 0.530 Urban 0.30 0.520 0.531 0.510 0.522 0.20 0.515 0.10 0.500 0.490 0.00 2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008 Source: World Bank sta calculations based on EPH, Paraguay. Source: World Bank sta calculations based on EPH, Paraguay. more stable and reflect a more long term condition observed. The percentage of people living in poor than income. In this section, two sets of household dwellings decreased from 15.6 percent in 2003 to less characteristics are analyzed: house construction than 12 percent in 2008, a significant reduction of 4.1 materials and access to public services. The former percentage points in five years (Figure 1.10). During reflects a condition controlled almost entirely by the the same period of time the share of households with members of the household, their potential, resources, dirt floor also decreased from 17.6 to 14.1 percent, priorities and finally their decisions. The latter reflects a 3.5 percentage point reduction. Both indicators government provided services (supply side) and the show modest improvements in the last five years, but desire or ability of the household to access such services especially since 2004. (demand side). Access to several housing services, including From 2003 to 2008 a significant reduction of the electricity, water, telephones and sanitation have population living in poor quality housing was been improving since 2003. Access to electricity Table 1.8: Growth and redistribution decomposition of poverty changes Change in incidence of poverty 2003 2008 Actual change Growth Redistribution Interaction Moderate Poverty Total 44.0 37.9 -6.1 -4.1 -1.5 -0.5 Urban 37.4 30.2 -7.2 -2.6 -4.5 -0.1 Rural 52.5 48.8 -3.7 -6.2 4.1 -1.6 Chapter 1 Extreme Poverty Total 21.2 19.0 -2.2 -2.9 1.6 -0.9 Urban 13.4 10.6 -2.9 -1.7 -0.6 -0.6 Rural 31.2 30.9 -0.4 -5.2 6.1 -1.2 14 Source: World Bank staff calculations based on EPH, Paraguay improved from an already high 92 percent of the to provide electricity to all its population and start population to an almost universal access (97 percent) devoting efforts to improving the quality of this in 2008. Households also experienced an important service. Despite increased access to quality water, improvement in access to telephone services from 39 to important improvements can be achieved in the 88 percent, mostly reflecting the introduction of cellular water distribution systems in Paraguay in order to telephone services and their expansion of coverage reach almost all of the third of the country without and ease in getting a line. Regardless, better telephone access as of 2008. Having clean water within the access is a real improvement in the household’s house is not only a convenience but also improves wellbeing and should not be discounted. Paraguayan the quality of life by saving time spent in bringing households also showed important progress over the water into the house, improving sanitation (better last five years with respect to the access to quality water bathrooms), promoting basic health practices like in the house, increasing from 58 to 68 percent, and food and hand washing (and hence reducing health bathrooms inside the house, increasing from 60 to 71 problems) and improving nutrition. percent (Figure 1.11). Improvements in housing quality and especially Poverty and Public Social Programs access to services from 2003 to 2008 are consistent with the poverty decreases reported in the last Public social spending in Paraguay as a share of GDP five years by the national statistics office. Indeed, is below the Latin American average, but shows a the moderate improvements in housing conditions stronger increase in 2009. At 12.6 percent of GDP, and services provide an independent verification Paraguay’s public social spending is below the Latin to the 10 percent reduction in extreme poverty and American average of 15 percent (2009). However, after 14 percent reduction in overall poverty, making growing less than 1 percent between 2003 and 2008, both tendencies more robust and reliable. The expenditures as a share of GDP increased 2.4 percent improvement of housing conditions also shows how in 2009. Among its components, education receives individual households chose to invest part of their almost 39 percent of the resources, social security and improved wealth. social protection each use close to 17.5 percent, and the health sector receives 24 percent. The next chapter Independently of the poverty situation, the explores in more depth the link between education and government of Paraguay should keep up the efforts poverty, as well as health and poverty. Figure 1.10: House Construction Materials, Paraguay 2003 -2008 Poor Dwelling Dirt Floor 18 20 Poverty Assessment 16 18 Percentage of households Percentage of households 14 16 12 14 10 12 10 8 15.6 16.3 13.8 13.1 12.4 11.5 8 17.6 18.3 15.0 14.2 14.6 14.1 6 6 4 4 2 2 0 0 2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008 Note: Poor dwelling de ned as a shanty town house, improvised dwelling or if the households lived in a rented room. Source: World Bank sta calculations based on EPH, DGEEC Paraguay. 15 Since Paraguay’s GDP is less than half the average GDP for Latin America and the Caribbean, and Figure 1.11: Household Access to given the low public social spending levels as a Public Services, Paraguay 2003-2008 percentage of GDP, public social spending in levels in Paraguay is also among the lowest in the region. Electricity Percentage of households with access 100 However, between 2003 and 2009 social expenditure as a percent of GDP increased 3.4 percent. Chapter 80 Telephone 5 analyzes the poverty impacts of expanding two 60 Water Bathroom of Paraguay’s social protection programs: the 40 conditional cash transfer program Tekoporã, and a 20 Sewage new law mandating non-contributive pensions to 0 the elderly poor. 2003 2004 2005 2006 2007 2008 Note: “Bathroom� access de ned as the percentage of households that have a bathroom inside the dwelling. “Water� includes all households with water pipes inside their dwelling. “Telephone� includes all types of services available (land lines and cell phones). Sewage includes septic tank and connection to the sewage system. De nitions and questions used to create the variable of water quality in the house (piped water) changed in 2008 and direct comparisons are di cult to make. Source: World Bank sta calculations based on EPH. Table 1.9: Public Social Spending – as a Percentage of GDP, Paraguay Public Social Spending Year Health Social Protection Social Security (*) Education Others Total 2003 1.35 0.08 2.97 4.17 0.63 9.21 2004 1.21 0.26 2.36 3.93 0.57 8.32 2005 1.28 0.14 2.62 4.21 0.62 8.86 2006 2.06 0.48 2.43 4.36 0.36 9.69 2007 2.17 0.63 2.18 4.18 0.76 9.92 2008 1.88 1.24 2.11 4.16 0.80 10.19 2009 3.00 2.19 2.20 4.85 0.36 12.60 Note: (*) Includes non Contributive Pensions Source: World Bank staff calculations based on data from the Ministerio de Hacienda de Paraguay Chapter 1 16 Box 1.1: Projections of Moderate and Extreme Poverty for Paraguay in 2009 This box presents projections of moderate and extreme poverty using a simple poverty-growth elasticity in order to estimate the possible impact of changes in the economic growth rate on poverty. The elasticities were estimated using a stylized model with fixed effects using GDP per capita for Latin American countries and several years (see Equation 1). The model includes the temporal trend as an additional control. Standard errors were aggregate at the country level and estimates were weighted using the average population of the countries. Table 1.10 provides the coefficients of GDP per capita for the entire Latin American region. Equation 1: Data: The data on moderate and extreme poverty for countries in Latin America, used to estimate the poverty-growth elasticity, were extracted from the Socio-Economic Database for Latin America and the Caribbean (SEDLAC-World Bank) on June 1, 2010.1 GDP per capita data in constant 2005 PPP values were extracted from the World Development Indica- tors (WDI, World Bank, 2009) on the same date. Table 1.10: Poverty Elasticity (coefficient ß of Equation 1) Extreme Poverty Moderate Poverty -2.880216 -1.746187 Projection: GDP growth estimated by the Central Bank of Paraguay (-3.8 per cent) for 2009 was used as a basic input for the projection. However, the variable used in the model is the GDP per capita at constant 2005 PPP prices. Therefore, the growth rate used in the projection was adjusted using two correlations: (1) the correlation between the GDP growth rate at constant 2005 PPP prices and the GDP growth rate at constant 1994 local prices, and (2) afterward the correlation with the GDP per capita growth rate at constant 2005 PPP prices. Figure 1.12: Projection of poverty and extreme poverty in 2009 60 49.7 50 44.0 43.7 41.3 38.6 41.2 37.9 40.5 40 30 24.4 23.7 23.2 21.2 19.0 21.2 18.3 16.5 20 Poverty Projection of poverty 10 Extreme poverty 0 Projection of extreme poverty 2002 2003 2004 2005 2006 2007 2008 2009 Notes: (1) SEDLAC (Socio-Economic Database for Latin America and the Caribbean) is a joint project between the Center for Distributional, Labor and Social Studies of the Universidad Nacional de La Plata and the World Bank. Poverty Assessment 17 Box 1 .2: National Household Surveys for Paraguay 1997-2008 The first household survey in Paraguay was performed in 1983 for the Metropolitan Area of Asunción and gathered in- formation on the labor market. In 1995, the General Directorate of Statistics, Surveys and Censuses of Paraguay (DGEEC) extended the geographical reach to rural areas and added more information on standards of living conditions (educa- tion, housing, migration, demography). Since then, except for 1996, the surveys are divided into 15 aggregate depart- ments in four domains: Asunción, Urban Area of the Central Department (Urban Central), Rest of the Urban Area (Urban Rest), and Rural. The surveys are also divided into two main types: the Integrated Household Survey (EIH) in 1997-98 and 2000-2001, and the Permanent Household Survey (EPH) in 1999 and from 2002 to 2008 (see Table 1.11). As of 2002, all the Household Surveys in Paraguay are based on a new modified sampling framework that stems from the new National Population and Housing Census (CNPV, 2002). The change in the sampling procedure has introduced distortions with respect to other surveys. Therefore, results must be analyzed very carefully, in particular in comparison with previous years. The EIHs of 1997-1998 and 2000-2001, and the EPHs of 1999 and 2002 are not strictly comparable due to the variability in methodologies, reference periods, sampling and thematic scope. However, they are the most adequate source of data to monitor distributional, labor and social conditions in Paraguay on an annual and national basis. The EPHs from 2003 to date are the most comparable surveys because of their identical sampling framework, scope and design of the ques- tionnaire. Some discrepancies in the reference period persist in recent surveys mostly due to lags in the interviews and data collection. One notable case is the lag observed in 2006, when the DGEEC faced an important budget constraint that forced it to postpone data collection until March 2007. Table 1.11: Paraguay’s Household Surveys Name of Survey Year Collection Scope Areas Sampling Size Period (households) Integrated Household EIH 1997-1998 Aug 97-Jul 98 National Areas, Domains 5,000 Survey and Depts (5) Permanent Household EPH 1999 Sept 99-Dec 99 National Areas, Domains 5,000 Survey and Depts (5) Integrated Household EIH 2000-2001 Sept 00-Aug 01 National Areas, Domains 8,960 Survey and Depts (6) Permanent Household EPH 2002 Oct 02-Dec 02 National Areas, Domains 5,000 Survey Permanent Household EPH 2003 Aug 03-Dec 03 National Areas, Domains 9,520 Survey and Depts (6) Permanent Household EPH 2004 Sept 04-Dec 04 National Areas, Domains 9,520 Survey and Depts (6) Permanent Household EPH 2005 Oct 05-Dec 05 National Areas, Domains 5,000 Survey and Depts (6) Permanent Household EPH 2006 Nov 06-Mar 07 National Areas, Domains 6,210 Survey and Depts (6) Permanent Household EPH 2007 Oct 07-Dec 07 National Areas, Domains 6,612 Survey and Depts (6) Permanent Household EPH 2008 Oct 08-Dec 08 National Areas, Domains 6,000 Survey and Depts (6) Chapter 1 Areas: Urban-Rural (5) Domains: Asunción, Urban Central, Urban Rest, and Rural. Depts: San Pedro, Cuaguazú, Itapúa, Alto Paraná and Central. (6) Domains: Asunción, Urban Central, Urban Rest, and Rural. Depts: All except Boquerón and Alto Paraguay. 18 Box 1. 3: Inter-institutional Committee on Methodologies for Poverty Measurement The General Direction of Statistics, Surveys and Censuses of Paraguay (DGEEC) convened, with the support of the World Bank (WB), an Inter-institutional Committee, consisting of a group of professional users of the poverty statistics that the DGEEC produces and disseminates. The objective is to create consensus and to assure the greatest possible transpar- ency in the methodologies and procedures utilized for the official estimation of the poverty rates in Paraguay. The Inter-institutional Committee has two central objectives: a) to discuss the diagnosis, methodological proposals, and results of the analyses presented by the DGEEC on the measurement of the poverty; and b) to agree upon the meth- odological changes in the measurement of poverty to be implemented by the DGEEC, creating consensus around the most appropriate methodology, its updating and future measurement, making sure that the full methodology is public, transparent, and replicable. In the course of the five meetings of the Inter-institutional Committee: i. Paraguay’s existing methodology for poverty measurement in the last 10 years was described; ii. This methodology was compared with international experience, explaining the recent methodological advances in the academic literature; iii. New sources of information were presented, for example the new caloric norms of the WHO/FAO; and iv. The methodological changes were recommended, agreed upon, and applied to the full series of household surveys from 1997/98 to 2008. The complete process has as an objective the establishment of a new series of poverty indicators backed by the best international practices that have the acceptance of civil society and other Paraguayan public institutions, and at the same time provide capacity building to DGEEC’s technical team so that they may take “ownership� of the methodology for calculating poverty lines. In turn, the DGEEC begins to offer training to members of the Inter-institutional Committee interested in deepening their knowledge of the details of the methodology. A final report documenting the methodol- ogy of this first phase will be presented to the members of the Inter-institutional Committee and will be available in the DGEEC’s web page. Note: The Interinstitutional Committee consists at present of representatives of the Ministry of Finance, Social Cabinet, Department of Education and Culture (MEC), Department of Public Health and Social Welfare (MSPyBS), National Institute of Diet and Nutrition (INAN), Social Action Secretariat (SAS), Dirección del Plan de la Estrategia Nacional de Lucha contra la Pobreza (DIPLANP), Central Bank Paraguay (BCP), Centro de Análisis y Difusión de Economía Paraguaya (CADEP), Instituto Desarrollo, Catholic University, Cámara Paraguaya de Exportadores de Cereales y Oleaginosas (CAPECO), Rodríguez Silvero & Asociados, UNFPA, UNICEF, World Bank, IRD-DIAL France, INEI Peru, and by DGEEC’s technical team specializing in poverty measurement and household surveys. The committee has met on five occasions (May 29, 2008; October 9, 2008; March 27, 2009; June 22, 2009; and October 27, 2009). Poverty Assessment 19 Box 1.4: Specific Issues in the Methodological Review of Poverty Measurement in Paraguay I. Harmonized treatment for all household surveys 1997-2008 a. Adjustment and correction of expansion factors due to rejections/absenteeism. b. Socioeconomic stratification of sample framework c. Estimate of harmonized income series 1997-2008 d. Estimate of base year expenditures 1997/98 II. Estimate for new poverty baseline a. Review/estimate for spatial price deflator b. Determine new basic food basket i. Review caloric norm ii. Chemical composition of foodstuffs iii. Food conversion table in calories c. Definition of single reference population d. Estimate unit value of food basket items for reference population e. Review/harmonization of syntaxis for income variable f. Estimate Engel coefficients for reference population in base year. III. Update poverty lines 1997-2008 a. Build non-food CPI for 1998-2008 period using sub-group weighing for base year reference population. b. Convert food and non-food components of basket into constant values using food and non-food CPIs built accord- ing to last point. IV. Sensitivity and robustness analysis for poverty results a. Sensitivity of poverty results to OMS/FAO (1985/2001) caloric norms and to activity level adjustments of caloric requirements. b. Statistical tests to verify differences between the distinct domains considered in our current line. c. Result sensitivity to interval range defining reference population d. Elasticity of poverty incidence to changes in the value of poverty lines e. Analysis of stochastic dominance. V. Systematic elaboration of databases and documents a. Database clean-up (for example, aberrant values); b. Base documentation (labels in variables and modes). c. Technical documents for sample. Chapter 1 20 Box 1.5: Main results of reviewing, updating and improving Paraguay’s poverty measurement methodology. 1997-2008 Period The General Office for Statistics, Surveys and Censuses (DGEEC, in Spanish), the Technical Secretariat for Planning at the President’s Office, within the framework of its commitment to transparency and the creation of trustworthy statistics, presents the main results of reviewing, updating and improving Paraguay’s poverty measurement methodology. The DGEEC, as with other Statistical Institutes in the region, estimated poverty (based on poverty lines) using a method- ology available in 1997/98, when the country’s first Encuesta Integrada de Hogares (EIH, Spanish for Integrated House- hold Survey) was realized. That survey, which included a household expenditure module, constitutes the poverty mea- surement baseline in Paraguay. Ten years after the establishment of the 1997/98 poverty baseline, new conceptual and methodological developments are available that make a revision of the historical series necessary. The discussion over the inconsistency of poverty lines is recent (Lokshin and Ravallion, 2006; Simler and Arndt, 2007), and was started by Ravallion’s (1998) and Pradhan et al’s work (2001). New methodological developments, added to a series of unexpected results from the most recent surveys, such as lower levels of rural than urban poverty, reinforced the need to improve and update Paraguay’s Poverty measurement methodology. This study is the result of a rigorous and meticulous 20-month long project, started in February 2008, which included training and readying DGEEC’s technicians to improve Paraguay’s poverty measurement methodology. For the adoption of the new methodology, DGEEC counted on the technical and financial support of the World Bank, as well as the support of the Interinstitutional Committee that accompanied the entire process, and includes representa- tives from the government, academic institutions, business guilds, civil society, as well as international organizations such as ECLAC, UNDP, UNICEF, the World Bank, and others. Within the framework of the poverty measurement methodology review (by poverty line), a series of activities were un- dertaken to obtain the elements that will form the basis for estimating the new poverty lines. Among the main activities mentioned were: adjusting survey results due to the non-response slant (particularly in the income variable), reviewing the main inputs used to determine the new reference population and thus the poverty line, as well as treating income data to build income aggregates. The biggest impact of this new methodology on poverty estimates corresponds to the redefinition of the so-called “reference population�, which corresponds to the population stratum used to define the value of the Basic Basket of food and non-food items. The methodology used in 1998, called Food Energy Intake (FEI), estimated the reference popula- tion after calculating the total spending (or incomes) of those households that on average acquire the necessary caloric intake. At that time (1997/98) three reference populations were autonomously established for the following geographic domains: Asunción, Urban Other and Rural Area, located in different spending percentiles, it thus failed to measure households with the same level of wellbeing. The new proposal (Ravallion, 1998; Lokshin & Ravallion, 2006) adopts one reference population for the entire country, in this way guaranteeing consistency between population wellbeing measurements. We should point out that, within the framework of this study, the poverty rates derived from the 2006 Household Survey (EPH 2006) are being shown for the first time, as they had not been published by the DGEEC until now, due to the fact that this survey presented major problems with regards to previous ones, related to the field data collection period, reference periods for the main variables used in measuring poverty, a high rate of non-response, particularly in income- related questions and other significant errors unconnected to the sample. In no small part, these inconveniences were Poverty Assessment corrected through post-stratification adjustments and the review of income estimates. The study reveals that poverty in general has been decreasing in the last three years, although after analyzing the series between 1997 and 2008, actual poverty levels are higher than estimated with the previous methodology, particularly in rural areas. The new measurement confirms that recent poverty figures have practically not changed with regards to those of ten years ago. The new poverty measurement benefits from a wide, transparent consensus and follows the most recent international improvement and practices. It allows the DGEEC to create more precise data and has placed Paraguay as a pioneer at the Mercosur level in terms of the use of new tools for the measurement of poverty. 21 Determinants of Poverty and inequality Who Are the Poor? Differences There is very little difference in gender and age Between the Poor and the Non Poor between poor and non poor household heads. Indeed the difference in age between poor and non It is common to have preconceived ideas about who poor household heads is only 2.3 years and there is and where the poor are as well as how a typical poor no statistically significant difference when referring household looks like. Some of these ideas might be to gender, with both poor and non poor heads having true but some have proven to be false when confronted values around 71.5 percent males. Poor households with hard data. As was mentioned earlier, the poor have use Guarani as the most common language spoken a much higher probability to be living in rural areas in at home twice as much as non poor households. It is Paraguay (two to one compared to urban areas), but important to remember that most Paraguayans speak what other characteristics are more common for poor Spanish and Guarani and the variable used here refers persons or households? to a preference of spoken language. In order to have a better idea of the differences between The household head’s education levels are very poor and non poor households, the average values different between poor and non poor. Almost 45 for several characteristics of the household head and percent of household heads of poor households demographics were computed for Paraguay in 2008. started but never finished primary education The results, including the average for the country, are compared to only one quarter of the non poor presented in Table 2.1. It is important to differentiate households. Curiously, the percentage of household between the conditions of poor households (presented heads with completed primary and incomplete Chapter 2 here), the correlates of poverty, and the determinants of secondary is very similar between poor and non poverty. The numbers presented here are a description; poor households (around 39 percent). With tertiary they do not take into consideration the relationship education the differences appear again with poor between variables and neither do they establish any households having almost no participation and non causality between the variable values and the income poor households having almost 20 percent of the 22 level of the household. heads with some tertiary education. Table 2.1: Household head and Composition by Poverty in 2008 Variable Poor Non Poor National Characteristics of the household head Age 45.8 48.1 47.4 Gender (=1 if male) 72.5% 70.5% 71.1% Guarani is most spoken language at home 61.4% 31.0% 42.8% Education of the household head No education 6.5% 3.3% 4.2% Primary incomplete 44.7% 24.6% 30.6% Primary complete 25.6% 20.5% 22.0% Secondary incomplete 14.6% 18.2% 17.2% Secondary complete 7.4% 14.7% 12.5% Tertiary incomplete 0.9% 8.9% 6.6% Tertiary complete 0.4% 9.7% 7.0% Job characteristics of the household head Works in agriculture 34.9% 16.5% 21.9% Informal worker 60.0% 41.5% 47.0% Dependency rate 2.7 1.5 2.0 Household composition Spouse present in the household 76.0% 62.4% 66.5% Children 0 to 5 years old 0.8 0.4 0.5 Children 6 to 13 years old 1.4 0.5 0.8 Youth 14 to 24 years old 1.1 0.9 0.9 Adults 25 to 65 years old 1.8 1.7 1.7 Adults 66 years old and more 0.2 0.3 0.2 Average number of members 5.3 3.7 4.2 Note: Dependency rate estimated as ratio of number of occupied members of the household and the total number of members. Informal workers are defined as salaried workers in small firms, non-professional self-employed and zero-income workers. Source: World Bank staff calculations based on the 2008 EPH. Work characteristic are also different for household household heads is less than 42. Also, while each non Poverty Assessment heads of the poor and the non poor, with non poor poor salary has to support 1.5 household members, for households having more stable formal types of jobs the poor it has to support 2.7 persons. Thus the poor and having to support less household members per have lower salaries in a less secure job and to be used worker. Since poverty is more common in rural areas, by more people. it does not come as a surprise that poor households have a much higher participation in rural activities than The poor live in bigger households with more children. non poor households (almost 20 percentage points). Indeed, poor households have an average of 1.6 more Also, poor household heads have a higher propensity members than the non poor and most of the difference to work in an informal type of job (60 percent are (1.3 members) comes from children between 0 and 23 informal), while the percentage of informal non poor 13 years old. Older children start leaving the house to Table 2.2: House Materials, Infrastructure and Land by Poverty in 2008 Variable Poor Non Poor National Dwelling characteristics Owned the dwelling 76.9% 76.7% 76.7% Poor dwelling 22.5% 6.9% 11.5% Poor materials on walls 3.0% 1.4% 1.9% Poor materials on the ceiling 49.1% 27.3% 33.8% Poor floor qualities 68.7% 35.4% 45.3% Number of persons per bedroom 2.97 1.77 2.13 Access to Services Water in the house a 53.3% 62.7% 59.9% Electricity in the dwelling 93.9% 97.9% 96.8% Telephone 79.2% 91.3% 87.7% Household Assets Refrigerator 59.4% 85.1% 77.5% Air conditioning 2.4% 23.9% 17.6% Washing machine 36.6% 64.2% 56.0% Computer 2.4% 20.2% 14.9% Car 5.6% 30.6% 23.2% Infrastructure (Average access in that neighborhood/department) Water infrastructure a 56.7% 61.4% 60.0% Electricity infrastructure 94.8% 97.6% 96.8% Sewage infrastructure 3.7% 11.2% 9.0% Telephone infrastructure 83.2% 89.7% 87.8% Access to land Owned land 48.8% 34.6% 38.8% Land owned: 0.1 to less than 0.5 hectares 23.6% 17.3% 19.2% Land owned: 0.5 to less than 2 hectares 4.3% 3.4% 3.7% Land owned: 2 to less than 15 hectares 18.4% 9.2% 11.9% Land owned: more than 15 hectares 2.5% 4.7% 4.0% Notes: Poor dwelling defined as a shanty town house, and improvised dwelling or if the households lived in a rented room. Poor materials on the walls include walls made of clay, palm tree leaves, cardboard and plastics. Poor materials on the ceiling defined as households with ceilings made of wood, cardboard, palm tree leaves and zinc. Poor floor quality is defined as floors made of dirt or concrete. Infrastructure variables reflect the average access of the population in each department and neighborhood. a The question asked to create the variable “water in the house� was changed in 2008 and there is no direct comparison to other years. The results presented here are only to compare poor and non poor households. Source: World Bank staff calculations based on the 2008 household survey, EPH 2008, DGEEC. Chapter 2 establish their own household and are not included as higher rates of incomplete primary education, few household members. with tertiary education, and much higher informality rates, with a third working in agriculture. Also, poor 24 Overall, poor household heads have similar age and households have a higher dependency ratio with an gender characteristics as non poor heads, but much average of 5.3 members speaking Guarani at home. The second set of household characteristics employed to compare poor and non poor households are material Figure 2.1: Years of Education by Age. assets, services, equipment, infrastructure, and land National and by Poverty, Paraguay 2008 ownership. The comparisons and the national average are presented in Table 2.2. Since these characteristics have a high relationship with income levels, only the National Poor Non Poor 12 less expected results will be discussed. For all other 10 variables not mentioned, the non poor have better Years of education 8 quality or more access than the poor. 6 4 In general and as expected, non poor households 2 have better quality housing and more access to public 0 services, however, there are important exceptions, 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 mainly: (i) there is no difference in house ownership; Age (ii) low quality wall materials are almost absent in either Note: the values are ve year moving averages. group; (iii) the poor have only slightly lower access to Source: World Bank sta calculations based on EPH 2008, DGEEC Paraguay. piped water and electricity in the house; (iv) the poor live in neighborhoods with similar water, electricity and years of education was reached in the last five years telephone infrastructure as the non poor; and (v) more (for people 19 to 24 years old.) poor households own land than non poor households, especially lots between 2 and 15 hectares. School attendance for 6-12 year old children is almost universal in Paraguay with more than 98 percent of Education and the poor them in school. After age 12, however, attendance drops substantially for both poor and non poor children, Over time, the average number of years of education reaching less than 50 percent by age 18. Attendance completed by Paraguayans has increased substantially. continues dropping for the next four years to only 20 Around 1948, the average education for a 19 year-old percent of 22 year-old youth (Figure 2.2.a). Paraguayan was less than four years.17 The average education more than doubled in forty years reaching There is almost no difference in school attendance over eight years of completed education by 1988.18 between poor and non poor six to twelve year-olds. Since then, however, the average number of completed More than 97.5 percent of poor children ages 6 to 12 years of education has not increased and a constant are attending school, compared to 98.1 percent of non value of around 8.5 years of education can be observed poor children, representing a non-significant difference. for the “last 20 years� in Figure 2.1. However important differences can be identified after age twelve. Regardless of the almost universal school Years of education for the non poor have a very attendance up to age twelve, poor thirteen year-olds similar tendency to the national average with two have close to 10 percentage points lower attendance important differences: first, their average education rates than non poor kids, which widens to a gap of 30 percentage points by age 18. For ages 19 to 25, Poverty Assessment is around 1.5 years higher, and second, the non poor keep adding years of education until later in life normally associated with tertiary education, the gap (around 27 years of age). For the poor there are also between poor and non poor children narrows to an two important distinctions from the national average: average of 15 percentage points. first, their average values are much lower (around 2 years lower), and second, there is no plateau during Clearly the government of Paraguay has made the last twenty years but a lower limit of around 6.5 important efforts to make the first six years of education 17 That is the expected education for 19 year-old Paraguayans 60 years ago, corresponding to the 79 year-old people in 2008. 25 18 Average for a 19 year-old person in Paraguay. Figure 2.2: School Attendance by Age in Paraguay, 2008 a. National Average School Attendance b. School Attendance by Poverty Group National Poor Non-Poor 100.00 100.00 School attendance (Percentage) School attendance (Percentage) 80.0 80.0 60.0 60.0 40.0 40.0 20.0 20.0 0.0 0.0 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Age Age Note: Primary, secondary and tertiary divisions are only indicative of expected levels and not real attendance levels by the students. Source: World Bank sta calculations based on 2008 EPH, Paraguay. available to everyone. Government efforts to continue more than a two percentage points per year increase with today’s coverage rate should be kept in place. For in net enrollment rates and three out of four poor kids older kids, the dropout rate is very high and the issues of secondary age attending their appropriate school behind this are not only economic.19 Clearly, financial year. Gross secondary enrollment rates increased more reasons play a part in the dropout rates for secondary than 1.5 percentage point per year for a total gain of education20 and by age eighteen the cumulative effect 15.4 percentage points. can be up to 30 percentage points (between poor and non poor children); but with a dropout rate of over 40 Improvements in primary and secondary enrollment percentage points for the non poor at age eighteen, rates in the last ten years have been pro-poor. no monetary reason can be totally responsible for School enrollment gains in percentage points by poor the greatest share of total school abandonment.21 students have outpaced those of non poor students The government of Paraguay should take a deeper by close to three times to one in gross primary and net look into secondary school dropout reasons and secondary enrollment and by two to one in secondary implement corrective actions targeting not only the gross enrollment. The rates are even higher if relative poor but the non poor as well. improvements are compared22 (Table 2.3). Poor households have made considerable school But the improved school enrollment rates for poor enrollment improvements in the last ten years. In children have not reached the last year of secondary primary education, the gross enrollment rate improved school. With less than a 30 percent attendance rate for almost one percentage point per year and, more 18 year-old poor children, there is a lot of work to be notable, at 98.5 percent the rate is almost the same as done to take full advantage of better enrollment rates that of non poor children (Table 2.3). Improvements at for the poor. Improved retention at higher school years secondary levels are even more impressive, showing will require a multidimensional approach that combines Chapter 2 19 By 2008, school level education remains mostly public in Paraguay with 83.6 percent of primary students in public schools and 78.2 percent of secondary students in public establishments. Tertiary education on the other hand is mostly private with only 35.4 percent of students in a public institution. 20 To use a more international definition and facilitate comparisons, the term primary education is used for the first six years and secondary education is used for the next six years of school. 21 It is assumed that non poor kids have the necessary means to attend secondary school and their desire to get a job is not an economic need but a monetary incentive. 26 22 For example, the 20.9 percentage points improvement in net secondary education represents a 63 percent improvement for the poor, compared to the 7.4 percentage points or 11.2 percent improvement for the non poor or a ratio of 5.6 to one. making school more attractive by highlighting the to the region’s average of 3.8 percent.23 Basic health advantages of a secondary degree, providing a more indicators like maternal mortality have improved diverse curriculum including cultural and sports some, but the ability to deal with more complex public activities, and incorporating new and different health issues is limited as revealed by recent increases teaching techniques to better fit the various needs and in yellow and dengue fever cases. characteristics of the students. In the last five years the role of public health The government of Paraguay should take into insurance has increased in Paraguay. Total private consideration changes in the demand for education health insurance coverage has remained at around to be able to adequate the supply of schools. Several 440 thousand persons while public health insurance changes are on the way and careful planning is required coverage has increased from 643 thousand persons in to meet new demands. Changes for primary education 2003 to 1.1 million in 2008, representing 71 percent of include the reduction of family size in Paraguay that total insured persons in the country (from a 60 percent would produce a stable and eventually decreasing share in 2003) (Figure 2.3). In the last five years public school age population. The same population changes health insurance has absorbed 100% of the growing will impact secondary age students (several years demand due to population growth as well as coverage after the effect on primary is felt) but at the same time increase. increased enrollment will increase demand and put pressure to provide better school access. Finally, these Access to health insurance for the lowest three quintiles changes have to be differentiated among the various has increased by at least half in the last five years, but geographical areas of the country – mainly between remains extremely low. For example, even after the urban and rural households but also between the second quintile’s access to health insurance doubled Central region and the rest of the country. between 2003 and 2008, it remained less than 11 percent. Coverage clearly improves with income and Health and the poor has improved for people in all income levels (Figure 2.4); but even for the fifth quintile it only reaches 57.5 As mentioned in Chapter 1, Paraguay’s health percent of the people. indicators are always below the LAC average and in a few cases, among the lowest in the region. Government’s low investment in health and limited Government spending in health has increased from health insurance coverage (public and private) are two 1.35 to 1.88 percent of GDP in the last five years, but contributing factors for the low health indicators found increased to around 3 percent in 2009, closing the gap in Paraguay. Since the role of government is growing Table 2.3:Primary and Secondary Enrollment Rates, Paraguay 1997/98-2008 1997-98 2008 Change (% points) Poverty Assessment School enrollment rates Poor Non -Poor Poor Non Poor Poor Non Poor Primary (% net) 89.0 96.0 90.1 99.0 1.1 3.0 Primary (% gross) 90.1 97.0 98.5 99.4 8.4 2.4 Secondary (% net) 33.2 66.3 54.1 73.7 20.9 7.4 Secondary (% gross) 55.9 75.4 71.3 83.1 15.4 7.7 Source: World Bank staff calculations based on EPH for corresponding years. 23 Reporte del Banco Mundial n.º 35910-CR. Evaluación de la pobreza en Costa Rica, Recapturar el impulso para reducir la pobreza, 12 27 de febrero de 2007. over time and budgets are limited, the appropriate use of resources is even more crucial than ever. Understanding Figure 2.3: Access to health insurance by type of provider who uses what type of services is very important to have an efficient government health policy. 2,000,000 Health insurance coverage has an upward tendency Total over time and varies at different ages, with the highest Population with coverage 1,500,000 rates for small children, people in their 30s and again in 1,000,000 Public their 60s (Figure 2.5). This pattern leaves important gaps in health coverage: as kids become older their coverage 500,000 Private (as dependents of an insured person) is reduced by 40 0 percent (from 25 to 15 percent) and does not increase 2003 2004 2005 2006 2007 2008 until they start joining the labor force and get their own insurance. Later in life, as people become older and Source: World Bank sta calculations based on EPH, Paraguay. labor force participation in the formal sector is reduced, health insurance also decreases by 30 percent (from 33 to 23 percent) until retirement benefits including non Figure 2.4: Access to Health Insurance contributory pensions increase coverage once again. by Income Quintiles This life cycle pattern leaves two important gaps in coverage: from around 10 to 27 years of age (with the lowest values at 19 years) and then again from 40 to 60 70 year-olds (with the lowest values at 50 years). 60 57.5 Population with access (%) 2008 50 48.2 2003 The non poor health insurance coverage over time is 40 33.5 30 27.2 very similar to the national average, but with higher 21.1 20 13.3 levels and more pronounced changes (higher and 10.9 10 5.2 lower values). The poor coverage also has a similar 1.7 2.7 0 pattern, but, with values between four and fourteen, 1 2 3 4 5 Quintile of per capita household income the variations are small in absolute value. Regardless of age, very few poor persons have access to health Source: World Bank sta calculations based on EPH, Paraguay. insurance and variations over time become of almost no consequence: there is very little difference between having 95 or 90 percent of people without health Figure 2.5: Health insurance Coverage insurance. by Age and Poverty Status, 2008 The number of visits when sick decreases with a lower income level. Indeed, the number of visits in the last 60 Non Poor ninety days decreased from over 286 thousand for 50 the highest quintile, to less than 205 thousand for the Percentage with access 40 lowest quintile, a reduction of almost 30 percent. Since 30 National poor households have a higher rate of health problems 20 Poor than non poor households, the 30 percent reduction is Chapter 2 10 an underestimation of the unfulfilled need for medical 0 attention. 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 Age (years) The type of health provider used changes with income Note: Numbers include public and private health insurance. Private insurance represents 28 around 10 percentage points for the non-poor and closer to zero for the poor. Source: World level. Households in the lowest quintile use health Bank sta calculations based on EPH 2008, Paraguay. centers more than half of the time when sick. The use of health center decreases with higher income and is only 12 percent for the highest quintile (Figure 2.6). Figure 2.6: Health Center Visited As income increases, households increase the use of When Sick (in the Last 90 Days) hospitals, with public hospital used more frequently by the first four quintiles and sharply increasing as a source of medical attention for the fifth quintile. Health center visited (% of persons) Number of visits (last 90 days) miles 70 287,6 300,0 60 251,8 254,0 The more frequent use of health centers by people with 237,2 Private hospital, 50 204,5 Clínic 200,0 lower income is probably a product of the access and 40 costs associated with the visit. Regardless of the reason, 30 IPS, Hospitals smaller health centers closer to the poor have proven 20 100,0 Health center to be one of the best investments in public health a 10 Other government can make because: (i) proximity to the 0 0 I II III IV V user increases use and substantially reduces out-of- Quintile of per capita houselhold income pocket transport expenses and the time spent by the Note: “Other� includes pharmacies and non professional consultations. patients; (ii) smaller centers are more efficient treating Source: World Bank sta calculations based on EPH, Paraguay. small medical problems (compared to hospitals); and (iii) they promote the use of preventive medicine, by Overall government policies and plans should take far the cheapest way to improve the health condition into consideration this characteristic and adapt to of the people. Investment in public health centers or the changes in demands one can expect in the near similar health establishments is pro-poor and the future. Real changes should first become apparent most efficient way to improve basic health conditions in government services and programs dealing with in the country. the very young, from pre-natal care to child delivery, nutritional programs and eventually primary and Demographic Changes secondary education. Education demand for primary education can increase due to improved attendance, In five years the average size of the average household but soon the demographic changes will reduce the in Paraguay has decreased by 0.2 persons. From total demand for schools. 2003 to 2008 the average number of members per households in Paraguay decreased from 4.4 to 4.2, an On the other hand, two important demographic average reduction of 0.4 (Figure 2.7.a). The reduction characteristics, the share of urban and rural households is even greater for rural households: from 4.8 to 4.4, a and the share of female headed households, have had reduction of 0.4 members per household. The higher very little or no change over the last five years. The reduction rate observed in rural areas has decreased rural population share is decreasing, but at a low 0.4 the gap between the urban and rural households’ size to percentage points per year during the last five years. only 0.4 person per household or less than 10 percent. More important, this reduction seems to be a product of reductions in the household size and not of internal Almost all the reduction in household size is migration. The share of female headed households has Poverty Assessment concentrated in the lowest age group of the remained the same since 2003, with values ranging households. Indeed, the zero to eighteen year old between 26 and 28 percent (Table 2.4). cohort experienced an average reduction of over 0.2 persons (from 2.0 to 1.8) per household between 2003 Correlates of Poverty and 2008, 0.4 for rural households (from 2.4 to 2.0) and 0.2 for urban households (Figure 2.7.b). An analysis of the correlates of poverty helps deepen the understanding of how poverty and household Average household size in the urban areas indicates characteristics are associated. By analyzing several a stable population with nearly zero growth and rural variables at the same time in an econometric framework, 29 households rapidly approaching a similar situation. the estimated effect of each variable can be isolated. Figure 2.7: Overall Urban and Rural Household size, Paraguay 2003-2008 a. All members a. Members 0 to 18 years old 5.0 3.0 4.8 4.8 2.6 2.4 Number of members Number of members 4.6 Rural 4.4 Rural 4.4 2.2 2.0 2.0 4.2 National 4.4 National 4.2 4.2 1.8 1.8 4.0 4.0 1.8 Urban 1.4 Urban 1.6 3.8 3.6 1.0 2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008 Source: World Bank sta calculations based on EPH, DGEEC Paraguay. Table 2.4: Urban-Rural and Female Headed Household Shares Population Share Female Headed Households Year Urban Rural 2003 56.4% 43.6% 26% 2004 56.9% 43.1% 28% 2005 58.0% 42.0% 26% 2006 58.1% 41.9% 27% 2007 58.3% 41.7% 27% 2008 58.6% 41.4% 27% Source: World Bank staff calculations based on EPH, DGEEC Paraguay Estimates obtained with this technique get closer to access to basic services; composition by age and the true effect of the individual variables. The estimated number of income perceivers. Individual regressions parameters help one understand whether the variable were estimated for 2008 Urban and Rural households is positively or negatively associated with income and using the natural logarithm of per capita household assess the relative strength of association of the various income as the dependent variable. The results are factors to poverty outcomes. The analysis is limited by presented in Table 2.5. the variables used, and no direct causality effect should be assumed in the face of the statistical relationships Demographics uncovered. Other variables such as violence, access to justice, vulnerability and climate conditions were not Two results from the demographic household Chapter 2 available and are not included but are expected to have characteristics were not expected and are difficult to an impact in the households’ per capita income. interpret. For urban households, the older the household head the lower the per capita income (decreases at a The household variables used in this analysis include: decreasing rate); and even though age discrimination 30 size; selected characteristics of the head including in the labor force is a common occurrence, normally education, gender and labor; geographic location; experience, seniority and capital accumulation have a higher impact on income generated. Also for both join the labor force. But this should not be an argument urban and rural households, the lack of a spouse in the to start working earlier in life. The ideal would be to household is associated with better income. have more prepared household members joining the labor force, as opposed to having several household Other results are easier to interpret like lower income members unemployed. levels for monolingual Guarani speaker household heads with more limited access to some economic Household size and structure sectors of the country, having lower income levels. Higher income associated with male household heads Larger households tend to have lower income and can be a sign of other factors not included in the model, increase their chances of being poor. This is true for like gender discrimination. It has been found in many urban and rural households, and there is surprisingly labor studies that a significant part of income disparities little difference among different ages of the members, between male and female workers cannot be explained with children five years old and younger having by observable characteristics other than gender itself the smallest impact over income. It is important to (see wage gap analysis in chapter 3). remember that the estimates are made taking into consideration other factors included in the model, Education and the impact of having a job, normally associated with 25-65 year olds has been already “discounted� or Education can be used as a way of escaping poverty. considered by another variable. Since primary education is very common in Paraguay24 no relationship was found between primary education Dwelling characteristics, land and durables and income, but income does increase with the number of years of completed education. Also, the The relationship between dwelling characteristics possible impact of education on income is higher than and household income are very different between any other variable used in the model.25 For example, urban and rural households. For urban households, five years of secondary education imply an income26 ownership is associated with higher income, while increase of 0.49527 for urban households (0.585 for poor floor materials and crowding with lower income. rural households). Also, six years of education in urban For poor households, the poor dwelling type and dirt households is associated with a total income increase floor are associated with lower income and access to of 0.252, also the highest value in the model. telephone with higher income. Only dirt floor is present in urban and rural regions. Labor Contrary to dwelling characteristics, household Employment in the informal sector and in agriculture durables are related to higher income levels in both is correlated with lower income. Household heads urban and rural households at different levels of working in agriculture (rural) and informal jobs importance. The variables included in this group are regardless of the place of residence, have lower return typical examples were causality can be argued in for their labor (lower productivity) and increase the both directions: household with higher income are Poverty Assessment probability of being poor. in the position to buy a car, but also a car can be an asset that improves household income. Finally, land On the other hand, households can substantially (also an asset) is associated with higher income for improve their income by having more of its members rural households. 24 The average age of a poor household head is 46 years, and this age group has on average 5.5 years of education. 25 This is true for the impact of any one household member. As a total, number of members in the household working can have a bigger impact than household head’s education. 26 In this section the term income refers to the dependent variable used in the model: natural logarithm of monthly per capita household income in Guaranies. 27 The total impact over the log of per capita income is estimated by multiplying five by the Beta parameter estimated in the equation 31 and reported in Table 2.5. Table 2.5: Correlates to income by urban and rural households in Paraguay, 2008 Independent variables Urban Rural Demographics Age -0.014 *** - - Age squared 0.00015 *** - - Gender (=1 if male) 0.107 *** 0.191 *** Spouse present in the household -0.182 *** -0.329 *** Speaks only Guarani (=1 if speaks) -0.104 *** -0.196 *** Years of education Primary 0.012 ns 0.003 ns by level Secondary 0.042 *** 0.025 * Tertiary 0.099 *** 0.117 *** Labor Works in agriculture -0.095 ns -0.216 *** Informal worker -0.182 *** -0.151 *** Number of persons working in the household 0.228 *** 0.161 *** Household Children 0 to 5 years old -0.133 *** -0.180 *** structure Children 6 to 13 years old -0.193 *** -0.224 *** Youth 14 to 24 years old -0.169 *** -0.159 *** Adults 25 to 65 years old -0.156 *** -0.196 *** Adults 66 years old and more -0.172 *** -0.217 *** Dwelling Household Owned the dwelling (owner=1) 0.080 *** -0.099 ns characteristics Poor dwelling - - -0.112 ** Poor materials on the walls - - 0.171 ns Poor materials on the floor -0.113 *** -0.121 *** Crowding (number of persons per bedroom) -0.108 *** - - Water in the dwelling - - 0.063 ns Telephone - - 0.147 *** Assets Refrigerator 0.112 *** 0.191 *** Air conditioner 0.262 *** 0.505 *** Washing machine 0.086 *** 0.177 *** Computer 0.156 *** - - Car 0.229 *** 0.402 *** Infrastructure Average access to sewage in the area 0.243 *** - - Average access to telephone in the area 0.393 *** - - Land 0.5 to less than 2 hectares - - 0.095 * owned 2 to less than 15 hectares - - 0.167 *** More than 15 hectares - - 0.578 *** Geography Rest of the country (Central region omitted) -0.078 *** - Constant 13.15 *** 13.19 *** Chapter 2 Observations (n) 2,614 1,986 R-squared 0.604 0.530 Notes: (1) Dependent variable: log of per capita household income. Only head of households. (2) “ns� not significant; * significant at 10%; ** significant at 5%; *** significant at 1%. (3) Several variables were excluded due to lack of significance: Dependency rate, Poor materials on the roof, Electricity in the dwelling, Average access to water and electricity in the area. 32 Source: World Bank staff calculations based on EPH 2008, DGEEC Paraguay Infrastructure and geography show that the HOI for Paraguay falls below the LAC average, at similar levels with the Dominican Republic, Infrastructure variables are represented by average Panama, and Peru, indicating that the country has a access values for specific household services. The idea is pending agenda to improve the opportunities faced to capture the impact of regional, state or neighborhood by its children (Figure 2.8a). On the other hand, characteristics on the household’s ability to generate analyzing growth rates in the HOI (between 1999 income. Households in better developed, more and 2008), Paraguay has made higher than average connected urban areas have a higher probability improvements over time. to fully use their abilities and potential. For urban households, living in areas with more developed Human Opportunity Index - Education infrastructure (sewage and telephone coverage) as well as residing in the central area (including Asuncion), is Vast arrays of basic opportunities are relevant to associated with higher income levels. policy and critical for children’s future development. For education, the completion of sixth grade on time Human Opportunity Index – Paraguay is used as a proxy for a child’s opportunity for basic education. Primary schools must be of sufficient quality The measurement of children’s “opportunities� to provide the differentiated instruction required to is based on access to basic goods and services get all children promoted through the first six years of considered critical for individual development schooling on time avoiding grade repetition or very low and for which universal access—by public or private marks. In a world of equality of opportunity, all children, provision—is a socially valid and feasible objective. This regardless of their circumstances, should have access to section applies an operational measurement of equality basic quality education. of opportunities called the Human Opportunity Index that focuses on access to basic goods and services by Paraguay ranks above the regional average in the Paraguayan children aged 0-16 (Barros et al., 2009). overall opportunity index in education. In 2008, the HOI This measurement takes into account both average in education in Paraguay was 80.4 percent (an increase coverage and distribution of basic opportunities of 12.2 points from 1999). among circumstance groups. These groups are defined according to pre-determined circumstances at birth With regard to school attendance at ages 10 to 14, (such as race, gender, family income, parents’ education Paraguay (like most of the countries in the region) had level, and place of residence) for which children cannot reached high levels of school attendance by the mid- be considered responsible and that therefore, from an 1990s. With opportunity index in school attendance equality of opportunity standpoint, should not affect almost universal (92%) and very low dissimilarity index their access to basic goods and services. Box 2.1 details (1.9 percent), the real challenge for the country is related the construction of the Human Opportunity Index to children’s performance at schools. (HOI)28 used in this section. The HOI for completing sixth grade on time reveals Paraguay ranks 13th out of 19 LAC countries in the an important challenge for the country. The average Poverty Assessment HOI using the 2005 estimates, and 12th out of 18 probability of finishing sixth grade on time recorded in the 2010 HOI projections, however its growth impressive advances in Paraguay, as well as in other rates are above the LAC average. The ranking is countries in the region like Brazil, Colombia, El Salvador based on five basic opportunities available across and Peru. The HOI of completing sixth grade on time all LAC countries: completion of sixth grade on time, increased 10 points over the last decade, from 45.3 school attendance, and access to electricity, drinking percent, however it continues to be very low in 2008 water, and sanitation services. The 2010 projections (56.3%). The HOI of completing sixth grade on time 28 The measurement of equality of opportunities used here follows the principals described in the 2006 World Development Report by 33 the World Bank and the methodology laid out in World Bank (2009). Box 2.1: Human Opportunity Index (HOI) The HOI is a synthetic measure of equality of opportunity in basic services for children 0-16 years. Its measurement employs the following formula: Where represents average coverage or access to a service or group of services or opportunities. D measures the inequality in the distribution of opportunities or unequal coverage among children from distinct popu- lation groups of pre-determined circumstances (parental education, urban or rural area of residence, gender, race, pa- rental income (see Table B.1). is an expansion factor to weight the observations included in the samples. is the estimated probability of having access to a service or opportunity for each of the observations in the sample or group of interest. Thus, the HOI increases when average coverage increases, or when that coverage is more equally distributed. The HOI ranges from 0 (zero coverage or complete inequality) to 1 (universal coverage). The HOI penalizes inequality through an increase in the value of D. For countries with identical coverage levels, the HOI of a country will be lower with greater coverage inequality. The HOI is estimated for each of the sub-groups of basic goods or services such as access to education or drinking water, which are fundamental for future economic opportunities and on which a social consensus exists on the goal of universal coverage. In a report for Latin America and the Caribbean, the World Bank (2008) includes goods and services that form part of the Millennium Development Goals to facilitate comparisons between countries. This group of goods and services can be broadened to adapt the index to the standards of more advanced middle-income countries such as Chile, Brazil or Uruguay with objectives beyond the MDGs. The HOI can be aggregated as a simple average to include a range of goods and services. It can also be estimated by geographic areas and population groups, to compare access to opportunities and to improve the targeting of public programs and social spending. Table B.1: Pre-Determined circumstances to define circumstance groups - Paraguay Reference unit Pre-Determined Circumstances Child Gender (male/female) Area of Residence (urban/rural) Household Head (Parent) Number of years of schooling Household Characteristics Household Income Per Capita (log Guaranies) Mono-parental household (yes/no) Number of children younger than 16 Source: World Bank (2009). reflects both a much lower coverage rate (62.8%) and a and number of siblings. Gender, gender of the household Chapter 2 higher dissimilarity index (10.4%) (Figure 2.10). head and area (urban or rural) had the lowest impact of all circumstances. There are some disparities in the Table 2.6 summarizes the relative importance of each relative importance of these circumstances compare to circumstance considered in the report. For school other counties. On Average for 19 Latin American and 34 attendance, the most important circumstance variables the Caribbean countries, the urban-rural location of are parent’s education, followed by per capita income the household and the presence of both parents in the Figure 2.8: Human Opportunity Index - LAC ranking, projected for 2010 a. HOI Level a. HOI growth rates Chile Mexico Uruguay Nicaragua Mexico Ecuador Costa Rica Brazil Venezuela, R.B Peru Argentina Guatemala Jamaica Paraguay 1.14 Ecuador Rep. Dominicana Colombia Colombia Brazil El Salvador Rep. Dominicana Chile Paraguay 73 Honduras Panama Costa Rica LAC Average Peru LAC Average Uruguay Guatemala Panama (0.99) El Salvador (76.5) Venezuela Nicaragua Jamaica Honduras Argentina 40 50 60 70 80 90 100 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 Source: World Bank sta calculations based on EPH, Paraguay. household have a higher importance when explaining the inequality in school attendance. For completion Figure 2.9: Human Opportunity of sixth grade on time, parent’s education is most Index in Education important circumstance, followed by gender of the Jamaica child. This indicates the gender differences in school Chile Argentina performance between females and males. Mexico Uruguay Peru Venezuela, R.B de Ecuador Human Opportunity Index- Housing Colombia Costa Rica Panama Paraguay 80.4 Rep. Dominicana A child’s access to adequate housing conditions is a Brazil El Salvador critical element of the opportunity for a healthy life. Honduras Nicaragua Three conditions have been selected as essential: Guatemala 0 10 20 30 40 50 60 70 80 90 100 access to water, to sanitation and to electricity. The Opportunity index in Education existing literature shows a strong and negative relationship between children’s mortality rates and Source: World Bank sta calculations based on EPH, Paraguay. improved water sources and sanitation facilities. Water and sanitation are primary drivers of public health, and should be considered basic opportunities life and environmental impacts (access to the public for all children. Access to electricity is also a basic network). However, important improvements had opportunity for children. Electricity improves quality been made in these three areas over the last decade. of life with respect to alternative sources of energy for For instance, in Paraguay, 64.2 percent of children Poverty Assessment lighting, cooking, and heating, such as kerosene and lived in dwellings with access to clean water, whereas wood fuel. Studies have documented that children 64 percent dwell similarly in Latin American and the spend more time studying after electricity is provided Caribbean countries. (Gustavsson 2007); electricity also allows access to modern educational techniques that includes the use In the area of infrastructure and housing, the biggest of computers. challenges remain in the area of water sanitation service provision, an indicator focused on both the quality of In the area of infrastructure and housing, the biggest life and environmental impacts (access to the public challenges remain in the area of water sanitation service network). However, important improvements had 35 provision, an indicator focused on both the quality of been made in these three areas over the last decade. Figure 2.10: HOI in school attendance and completing sixth grade on time HOI in education (1999-2008) Coverage and D-index in education (2008) 1999 2008 Coverage 100.00 100,00 93.8 91.0 92.0 80.00 80,00 Opportunity Index Coverage, D-Index 62.8 60.00 56.3 60,00 45.3 40.00 40,00 20.00 20,00 D-Index 10.4 1.9 0.00 0,00 School Attendance Sixth grade on Time School Attendance Sixth grade on Time (Primary, 6-12 years old) (Primary, 6-12 years old) Source: World Bank sta calculations based on EPH, Paraguay. Table 2.6: Relative importance of the seven circumstance variables in the in- equality of access to education School Attendance 10-14 year olds Complete sixth grade on time Circumstances Paraguay LAC Average Paraguay LAC Average Parent's education 1.10 1.22 5.42 5.24 Gender 0.17 0.34 3.41 2.42 Gender of Household Head 0.24 0.32 0.25 1.01 Per Capita Income 0.53 0.36 2.62 2.16 Urban or Rural 0.31 0.43 0.74 1.98 Presence of Parents 0.31 0.44 1.50 1.00 Number of Siblings 0.41 0.19 3.28 2.48 Source: World Bank staff calculations based on EPH, Paraguay Table 2.7: Relative importance of the seven circumstance variables in the in- equality of access to basic infrastructure Electricity Water Sanitation Circumstances Paraguay LAC Average Paraguay LAC Average Paraguay LAC Average Parent's education 0.43 1.63 3.66 4.33 8.63 7.52 Gender 0.05 0.08 0.00 0.43 0.17 0.21 Gender of Household Head 0.08 0.49 0.06 1.22 0.80 1.53 Chapter 2 Per Capita Income 0.63 2.11 4.71 5.79 13.33 8.83 Urban or Rural 0.79 4.23 5.56 10.80 16.51 13.57 Presence of Parents 0.08 0.23 0.87 1.54 1.50 1.66 Number of Siblings 0.04 0.40 0.83 1.22 1.98 1.58 36 Source: World Bank staff calculations based on EPH, Paraguay For instance, in Paraguay, 64.2 percent of children found in other countries (see columns with the lived in dwellings with access to clean water, whereas LAC averages). Other important circumstances are 64 percent dwell similarly in Latin American and the per capita household income, followed by parent’s Caribbean countries. education. In the case of electricity, Paraguay has achieved almost Human Opportunity Index - Geographic universal access (with a coverage rate of 95.9 of the Disparities children with access and very low inequality (D-Index of 2.1 percent)). Access to electricity is the most uniform Reducing geographic disparities in the structure of across the region, with several countries reaching opportunities in education and infrastructure is another universal access (Chile) or nearly universal (Argentina, important challenge for Paraguay. Costa Rica, Mexico, and República Bolivariana de Venezuela). In education, substantial differences are apparent among departments, with the worst opportunity Only 45.7 percent of children ages 0 to 16 in Paraguay lived in dwellings with sanitation in 2008, compared with 35 percent in 1999. Sanitation coverage rate is low (only 60 percent of the children lived in dwellings with Figure 2.11: Opportunity Index sanitation), but this is not the only impediment, since in housing the dissimilarity index is also high, indicating a bigger Costa Rica inequality among children. Uruguay Chile Venezuela Argentina Brazil Table 2.7 summarizes the relative importance of each Mexico Colombia circumstance in explaining inequality in the housing Dominican. Rep. Ecuador indicators. For the first time, “area� (urban versus rural) Jamaica Paraguay 67 Panama matters as a circumstance to explain the inequality Peru Guatemana of access. It is the most important circumstance for El Salvador Nicaragua the access to sanitation, as well as for electricity and Honduras 0 20 40 60 80 100 water. The importance of the urban-rural location of Opportunity index in Housing the household in Paraguay as a determinant of the access to basic services is consistent with the results Source: World Bank sta calculations based on EPH, Paraguay. Figure 2.12: HOI in housing indicators HOI in housing indicators Coverage and D-index in housing indicators (2008) 1999 2008 Coverage Poverty Assessment 100 94 100 96 83 80 80 Opportunity Index Coverage, D-Index 72 64 60 60 60 46 42 36 40 40 24 20 20 D-Index 10 02 0 0 Electricity Water Sanitation Electricity Water Sanitation 37 Source: World Bank sta calculations based on EPH, Paraguay. (Guaira) and 100 (Cordillera) across 9 departments of the country (Figure 2.13). Figure 2.13: HOI in school attendance and completing sixth grade on time With few exceptions, access to electricity is virtually universal. Access to electricity is also very equitably School attendance (Primary 6-12 years old) distributed across the country (very low D-Indexes). Cordillera 100 Itapúa, Cordillera and San Pedro are the three Central 96 Departments that lag behind all other departments Asunción 94 (Figure 2.14.a), with HOI in access to electricity lower Paraguarí 93 Caaguazú 91 than 90 percent. Alto Parana 91 San Pedro 90 Unlike electricity, there is more variability in Itapúa 82 Guairá 78 equality of access to water and sanitation across 0 20 40 60 80 100 departments. The departments that have high HOI levels of the opportunity index in access to water are Asuncion, Cordillera and Central (Figure 2.14.b). There still exist departments where more than fifth percent Completing sixth grade on time of children are not provided access to water with an Central 65 equality of opportunity principle in mind. Among these Paraguarí 64 Asunción departments are Itapúa, Guairá and Caaguazú. 63 Alto parana 49 Caaguazú 48 Of all the housing and infrastructure variables Guairá 47 considered as basic opportunity, Paraguay performs Cordillera 45 San Pedro 43 worst on access to sanitation (Figure 2.14c). Disparities Itapúa 39 exist across all departments and are large, with 0 20 40 60 80 100 San Pedro, Itapúa, Caaguazú and Guaira with HOI in HOI sanitation lower than 30 percent and departments like Central and Asuncion with opportunity indexes higher HOI in education than 75 percent. Central 80 Paraguarí 79 When considering the overall HOI in housing, Asunción 78 Asuncion, Central and Alto Parana are the leading Alto parana 73 Caaguazú 70 Departments, while Itapúa, Caaguazú and San Pedro Guairá 69 have the lowest HOI in housing. Cordillera 67 San Pedro 62 Itapúa 60 This section explores the evolution of equality of 0 20 40 60 80 100 opportunity by departments over time. The discussion HOI that follows compares HOI levels from 1999 and 2008. As shown in Figure 2.15 the departments that had Source: World Bank sta calculations based on EPH, Paraguay. experience the biggest increased (above the national average) in HOI in the period 1999-2008 are Paraguarí, structure found for finishing sixth grade on time in Alto Parana and Cordillera. As indicated by the y-axis, Chapter 2 Itapúa, San Pedro, Cordillera and Guairá (Figure 2.13). Human opportunity index in Paraguarí increased by Children across the country have almost the same high 28.6 percentage points in nine years, followed by Alto level of access to school attendance. Children between Parana (23.1) and Cordillera (16.7). 10 and 14 years of age in almost all parts of the country 38 are provided the opportunity to attend school equally. Itapúa was not only the department with the lowest The level of the opportunity index is between 78 HOI, but also the department that experience the Figure 2.14: HOI in housing indicators (2008) a. Electricity b. Water Guairá 100 Asunción 88 Central 98 Cordillera 84 Asunción 97 Central 77 Paraguarí 97 San Pedro 74 Alto Parana 96 Alto Paraná 65 Caaguazú 92 Paraguarí 65 San Pedro 89 Caaguazú 53 Cordillera 86 Guairá 53 Itapuá 82 Itapúa 37 0 20 40 60 80 100 0 20 40 60 80 100 HOI HOI c. Sanitation (sewage network, septic tanks) d. Overall HOI in Housing Central 85 Asunción 87 Asunción 77 Central 87 Alto Parana 58 Alto Parana 73 Paraguarí 49 Cordillera 72 Cordillera 47 Paraguarí 70 Guairá 30 Guairá 61 Caaguazú 27 San Pedro 59 Itapúa 26 Caaguazú 57 San Pedro 14 Itapúa 48 0 20 40 60 80 100 0 20 40 60 80 100 HOI HOI Source: World Bank sta calculations based on EPH, Paraguay. lowest increased in the index. Both the level and the change of the HOI are below the national averages Figure 2.15: Human Opportunity Index for this department. A similar pattern is observed for by region Caaguazú. Asuncion and Central are the departments with 100.00 Asunción the highest HOI in 1999. However in term of their 90.00 performance both departments experience HOI 80.00 Central HOI 1999 changes below the national average. In the specific 70.00 case of Asuncion the change rate is negative which 60.00 Itapúa Caaguazú Cordillera implies a decreased in the Human opportunity index Poverty Assessment 50.00 Guairá Alto Parana San Pedro Paraguarí in this department. This negative change has been 40.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 mainly driven by an increase in the dissimilarity index, HOI 2008 indicating a more unequal access to basic education and services for children. Source: World Bank sta calculations based on EPH, Paraguay. 39 Urban labor market: trends and opportunities for employment Trends in Labor Market Indicators income. Female labor force participation increased by 6.2 percent at the national level over the period 2003- Paraguay shows improvements in almost all labor 2008 (Table 3.1). Even so, women show lower labor market indicators during the growth between 2003 force participation than men in all age groups (Figure and 2008. Labor force participation increased from 61.5 3.1). However, the difference is relatively small for young percent of the working age population (aged 10 to 64 people between 10 and 29. years old) in 2003 to 63.6 percent (or around 2.8 million Paraguayans) in 2008 (Table 3.1). Most of the increase was Employment registered in urban areas, especially for workers aged 20- 49 and 50-64 (Figure 3.1). In rural areas an increase in the Paraguay’s employment structure between 2003- labor force participation is observed for those aged 65 2008 has slightly increased favoring females, higher years old or more. The employment rate also increases skilled workers and urban areas. Although there are between 2003 and 2008, while the unemployment rate significantly more male than female workers employed, and informality decreased. However, underemployment the gender gap in shares narrowed between 2003 and and unemployment duration increased.29 2008 (Figure 3.2). In the year 2003, 37.8 percent of the working population were women, while in 2008 that An important change in the composition of share grew to 39 percent (Table 3.2). labor supply is the increase in female labor force participation. This increase seems to be due to the The educational structure of the working population impact of secular trends such as increased access to over the period shows more important changes in Chapter 3 education by girls, urbanization, and fertility reduction. favor of the more skilled: employment for the less Expanding female labor force participation has also educated (persons with no education or primary been a household mechanism to increase household incomplete) has decreased by 7.4 points (Table 3.2). 29 The majority of these tendencies also hold in the longer run. For example, the labor force participation and employment rates increased 40 steadily between 1997 and 2008. However, for other indicators in this chapter the comparisons across time need special attention to assure that an appropriate period of time is used. In the following sections a more detailed analysis is 2008, there has been an increase in the percentage of provided on the level of education of the labor force in workers working as entrepreneurs and wage earners Paraguay and its evolution over time. and a reduction in self-employed and workers without income (workers working in the family businesses, etc.). The share of rural areas in total employment There is also an increase in the percentage of workers decreased by 3.4 percentage points. Figure 3.3 shows working in large firms, from 15.6 to 19 percent in 2008. that Asuncion and Rural Rest have lost participation in Most of these results continue to hold when analyzing employment, while Urban Central has consolidated its the data for the last decade. position as one of the regions with the largest share in total employment. Employment by sector and economic growth The structure of employment shows a move to During the 2003-2008 period, the sectoral structure work more formal working relationship and larger of the economy also changed: a significant decrease companies or public (Table 3.4). In the period 2003- in the share of workers primary activities followed Table 3.1: Labor market indicators in Paraguay 2003-2008 Indicator 2003 2008 Change Labor force participation (population aged 10-64 years old) 61.5% 63.6% 3.46% Female labor force participation (%) 47.5% 50.4% 6.2% Employment rate 56.5% 59.9% 3.37 a Under-employment (% of working population) 26.9% 28.3% 5.1% Unemployment rate 8.1% 5.9% -2.22 a Unemployment duration (months) 5.2 7.2 2.00 b Informal workers (percentage of total employed) 71.8 65.4 -8.84% Note: (a) Indicators refer to the population between 10 and 64 years old; (b) change expressed in percentage points; (c) months. Informal workers defined as: salaried workers in small firms, non-professional self-employed and zero-income workers. Under- employment: percentage of persons with a job that work less than 30 hours per week and want to work more hours, or persons that work 30 or more hours per week but receive a wage below the current legal minimum. Source: World Bank staff calculations based on EPH, Paraguay Figure 3.1: Labor force participation – Working age population a. By area b. By gender 100% Poverty Assessment 90% 90% 80% 80% Male Number of Members Number of Members 70% 70% (2008) 60% 60% 50% Rural 2008 50% Rural 2003 Female 40% Urban 2003 40% (2008) 30% Urban 2008 30% 20% 20% 10-19 20-29 30-39 40-49 50-64 65 and more 10-19 20-29 30-39 49-49 50-64 65 and more Age Age 41 Source: World Bank sta calculations based on EPH, Paraguay. Figure 3.2: Employment rates by gender Figure 3.3: Distribution of workers by region 2003-2008 80 Male 2003 2008 Total 45 60 41 Distribution of workers Percentage (by Dominio) 40 Female 26 24 23 21 20 11 9 0 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 Asunción Urban Central Urban Rest Rural Source: World Bank sta calculations based on EPH, Paraguay. Source: World Bank sta calculations based on EPH, Paraguay. Table 3.2: Distribution of workers (by gender, education and area of residence) Employment distribution by 2003 2008 Change (% points) Gender Female 37.8 39.0 1.2 Male 62.2 61.0 -1.2 Education Low 62.7 55.3 -7.4 Medium 26.4 30.3 3.9 High 10.9 14.3 3.4 Area Urban 55.3 58.7 3.4 Rural 44.7 41.3 -3.4 Source: World Bank staff calculations based on EPH, Paraguay. Population aged 10 or more years Table 3.3: Employment distribution by labor relationship and type of firms Labor relationship Type of firm Entrepreneurs Wage earners Self-employed Zero income Large Small Public 2003 4.3 44.1 39.2 12.4 15.6 75.8 8.5 2004 4.1 43.1 39.7 13.1 16.1 76.8 7.1 Chapter 3 2005 4.6 46.5 37.2 11.7 16.9 73.9 9.1 2006 4.6 46.9 36.2 12.4 17.2 74.0 8.8 2007 5.3 48.5 36.3 10.0 17.7 73.7 8.6 2008 5.2 50.9 33.5 10.5 19.0 71.5 9.5 42 Source: World Bank staff calculations based on EPH, Paraguay Box 3.1: Migration Migration is a visible phenomenon in large cities such as Asunción, or in urban areas of the Central Department. Mi- grants are indeed the majority in urban areas, representing this past decade 45 percent of the population on average (and up to 56 percent in the Central Department), compared with the 36 percent in the rural population. Analyzing the place of origin of migrants in urban areas (Figure 3.4), most are from other urban areas (27 percent in 2008) and only secondly from rural areas (17 percent). Surprisingly, the 7 percent of recent migrants (i.e. the migration of the last 5 years) represents only one fifth of those who migrated before the last 5 years. The results are similar at the level of Asunción or the Central Department. For example, in 2008 more than 44 per cent of the population of the Central Department migrated before 2003. Figure 3.4: Origin of Urban Migrants 50 40 Foreigner 30 Urban (<5 years) 20 Rural (<5 years) Urban (>5 years) 10 Rural (>5 years) 0 2004 2005 2006 2007 2008 Source : World Bank sta estimates based on EPH, Paraguay. For the 2004-2008 period, migrants in urban areas were 10 years younger on average than the rest of the urban popu- lation (with an average age of 37 years), they have 3.3 household members in comparison with 4.3 for non-migrants, and both have over 7 years schooling (Table 3.3 for 2008). The percentage of migrants in poverty and extreme poverty declined significantly between 2004 and 2008. Poverty reduction has been lower for non-migrants, dropping from 11.9 to 10.5 per cent of extreme poverty in the period. 62 per cent of migrants work in the informal sector, a higher percentage than non-migrants. Table 3.4: Characteristics of heads of households in urban areas Non migrant Migrant Age 47.4 33.6 Years of schooling 8.9 7.7 Gender (% men) 69.2 68.2 Informal workers 57.7 62.5 Duration of unemployment 8.2 3.8 Real income pc monthly (‘000 PYG, price 2008) 112684 79377 Moderate poverty 30.2 28.7 Extreme poverty 10.5 16.3 Poverty Assessment Source: WB estimates based on the 2008 EPH, Paraguay The duration of unemployment is much shorter for migrants, with 2.4 months on average for the period versus 5.8 for non-migrants. The distribution of migrants and non-migrants in the labor sectors is almost similar, except for com- merce where non-migrants are more numerous and domestic service in which most migrants work. Lastly, recent migrants from rural areas tend to belong to the highest income quintiles. This is surprising taking into account that Otter & Villalobos (2008) show that migrants have a significant premium in the incomes only in the first two quintiles, and that the decision to migrate explains the 50 and 32 per cent of incomes for the first and second quintile.1 The purpose of this analysis is solely to start showing some of the characteristics of migration. The issue is complex and merits its own analysis, whereby it is beyond the scope of this report. 43 Nota: (1) Se trata de un análisis ex-post para la zona Metropolitana en el 2005. by a similar decrease in domestic servants were adults with complete primary education is relatively registered (Figure 3.5). On the contrary, commerce, high. Given the level of the GDP per capita of the country, construction, skilled services and industry have gained the net secondary school enrollment rate is also high. participation in total employment. Figure 3.6 shows the GDP growth for the same period by economic However when analyzing the level of education of the sector. Even with a lower distribution of workers, the labor force, Paraguay has a higher share of workers agricultural sector showed the highest growth (35 with low levels of education (primary incomplete and percent) in terms of real GDP between 2003 and 2008 complete, Figure 3.8). Among all countries in the region, (at constant prices of 1994). Bolivia is the only country that has a higher share of its labor force with low levels of education (55.3 percent The level of qualification of the labor force of workers in Bolivia has less that primary education complete). In Paraguay, 54% of the workers 25-65 years Compared to other Latin American countries with old have lower education, while only 15.5% has tertiary similar levels of per capita income, Paraguay ranks complete or incomplete. above the median in terms of the levels of education of its population (Figure 3.7). Even though the country has Educational level of the labor force has been a low level of GDP per capita, the percentage of young increasing over the years. Between 1999 and 2008 the Figure 3.5: Distribution of workers by economic sector Employment distribution by sector (% workers) 35 2003 2008 30 25 20 15 10 5 0 Agriculture Fishing Mining Manufacturing Utilities Construction Commerce Rest. & Hotels Transp. & communications Finance Business services Public Admin. Teaching Health & Social Services Other services servants Foreign Organizations Domestic Source: World Bank sta calculations based on EPH, Paraguay. Figure 3.6: Real GDP growth 2003-2008 Figure 3.7: Years of education (constant prices of 1994) Population 21-30 years old Years of education - Population aged 21-30 years old 40 Percentage change (2003-2008) 35 35 60 32 30 Chile 27 28 50 Argentina 25 23 Bolivia Ecuador Peru Panama 40 Uruguay 20 Paraguay Venezuela Mexico Rep. Dominicana Chapter 3 15 30 ColombiaBrazil Costa Rica 10 10 20 El Salvador 5 Nicaragua 10 Honduras 0 Guatemala Agriculture Fishing, Minery Electricity Construction Services 0 Forestry and and Industry Water 0 2.000 4.000 6.000 8.000 10.000 12.000 14.000 16.000 GDP per capita (constant 2005 International $) 44 Source: World Bank sta calculations based on EPH, Paraguay. Source: World Bank sta calculations based on EPH, Paraguay. proportion of workers with primary education or less percent, one of the highest levels of informality in has steadily decreased, while increasing the percentage comparison with other LAC countries. Once again, of workers in the medium level (secondary complete only Bolivia seems to have higher levels of informality. and incomplete ) and tertiary education (high skilled Although the share of informal workers has decreased workers) Figure 3.8. The increase in the percentage over time, as shown earlier, Paraguay continues to of workers with medium levels of education is higher maintain sizeable levels of informality (Figure 3.10). for males than for females. Among female workers, it is important to notice the important increase in high According to the assessment by the ILO (2003), the educational women participating in the labor market. labor market is characterized by a low compliance In 2008, while 22.1 percent of the female workers have with laws and regulations. According to the study, tertiary education, only 13.3% of the males workers informality is mainly the outcome of several factors such have a similar level of education. (Figure 3.9) as inadequate norms or rigid laws for the development of firms and an inefficient system of incentives. Informality With the objective of having a better measurement The participation rate of Paraguayan workers of 25 of the levels of informality in Paraguay’s labor market, to 65 years old in the informal sector is around 67 the National Institute of Statistics (DGEEC) with the technical assistant of the ILO introduced new questions in the 2007 household survey. These questions Figure 3.8: Distribution of workers gather information about the registration of workers by educational level (circa 2008) in a contributors’ registry (“RUC -Registro Unico de Contribuyentes�). High Medium Low Distribution of workers by educational level 100% According to all definitions, informality in the labor 22.3 18.9 17.0 20.7 12.9 16.3 15.5 15.6 80% 30.2 market in Paraguay is high. And has remained roughly 35.7 29.7 30.3 29.1 60% 38.5 39.5 31.7 unchanged and at very high levels. 51.6 41.7 40% 47.7 51.4 54.0 54.2 55.3 Under-employment 20% 42.7 43.9 26.1 28.1 0% Under-employment has been generally increasing Chile Argentina Peru Uruguay Costa Rica Brazil Ecuador Paraguay Bolivia over the last decade reaching its peak of 29 percent in 2007. The reduction in unemployment is not only related with an increase in employment but also with Source: SEDLAC database, CEDLAS (UNLP) and World Bank. Figure 3.9: Share of workers in each skill group (by gender) Male Female Poverty Assessment 80 80 70 70 Low Percentage of workers Percentage of workers Low 60 56,0 60 56,0 50 50 40 Medium 40 30 30,8 30 Medium High 21,9 20 13,3 20 High 10 10 22,1 0 0 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 45 Source: SEDLAC database, CEDLAS (UNLP) and World Bank. Table 3.5: Informality, new definitions Year Change Informality definition 2007 2008 (percentage points) Salaried workers workers not contributing to the pensions system 65.6 65.4 -0.2 Entrepreneurs and Self -employed without RUC 79.2 76.8 -2.4 Source: Dirección General de Estadistica, Encuestas y Censos (DGEEC), Paraguay under-employment. While invisible under-employment decreased in 2008, visible under-employment increased Figure 3.10: Share of workers in informal sector by almost 2 percentage points (Figure 3.12). It is (25-65 years old; circa 2008) important to notice that invisible under-employment affects 21 percent of the working population in 80 70 67 69 Paraguay. These workers are working 30 or more hours 64 65 66 60 58 59 60 61 per week, but earn less than the legal minimum wage. 53 50 48 49 49 Percentage 41 40 35 38 40 30 The increase in under-employment between 20 2003 and 2008 was driven largely by increases in 10 invisible under-employment in rural areas. Under- 0 employment in urban areas has slightly decreased. Chile Costa Rica Argentina Uruguay Panama Mexico Brazil Rep. Dominicana Honduras El Salvador Ecuador Colombia Peru Nicaragua Guatemala Paraguay Bolivia Under-employment has been decreasing in urban areas from 31.3 to 30 percent in the period, however invisible underemployment is much larger and went Note: Productive de nition of informality: A worker is considered informal if (s)he is a salaried workers in a small rm, a non-professional self-employed, or a zero-income worker. Based on up by 2 percentage points reflecting an increase in each country Household Survey, for the years 2007 and 2008 except for: Nicaragua (2005), the number of urban workers in precarious situations Chile (2006), El Salvador (2006), Guatemala (2006), Colombia (2006). Source: SEDLAC database, CEDLAS (UNLP) and World Bank. (Table 3.6). In rural areas, both visible and invisible underemployment increase during this period, leading to an increase in total underemployment of 3 percentage points reaching a level of 24.5 percent in Figure 3.11: Informality 2008. These workers or want to work more hours or by education level earn less than the minimum wage. Low Medium High Compared to non-under-employed workers, the 100% under-employed are in general younger, more likely 90% 80% female and less skilled (see last two columns of Table 70% 3.7). Historically women have shown much higher rates 60% Percentage 50% of underemployment; however underemployment Chapter 3 40% 30% among men has surprisingly been increasing in recent 20% years. Thus, the gender gap in underemployment has 10% 0% been narrowing since 2005 (Figure 3.13). The distribution 1997-98 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 of the underemployed and the non-underemployed also shows some differences by economic sector. 46 Source: World Bank sta calculations based on EPH, Paraguay. Underemployed workers tend to be more focused on work as domestic servants or in trade, agriculture and manufacturing Figure 3.12: Under-employment Invisible under-employed is more frequent for younger workers, males, working as domestic servants, or Invisible Visible Under-employment total in commerce and manufacturing. Invisible under- 35 27.3 29.0 28.3 employed have also the lowest proportion of workers 30 25.7 26.9 26.8 25.9 24.5 5.2 Under-employment with tertiary (complete or incomplete) education. 25 7.1 7.1 19.3 8.4 8.8 5.4 20 8.3 8.8 15 6.6 Relative to invisible workers, visible under-employed 23.9 10 18.6 18.0 20.2 20.5 21.2 are characterized by a higher proportion of informal 16.2 16.9 5 12.7 workers (79.6 percent), mainly working in agriculture 0 (36.5 percent) and services (20 percent). Even so, 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 these workers have higher levels of education since 19.7 percent of the workers have tertiary education Note: Under-employment includes two types of workers: (i) persons employed that work less (incomplete or complete). than 30 hours per week and would like to work more hours (visible under-employment) and (ii) persons employed that work less than 30 hours per week, would like to work more hours and their wages are lower than the minimum legal wage (invisible under-employment). Unemployment Source: World Bank sta calculations based on EPH, Paraguay The level and changes in unemployment can be explained by cyclical variations of the economy as a whole and structural factors related to secular trends in Figure 3.13: Under-employment labor supply and the performance of labor markets. The in Paraguay by gender unemployment rate rose with the economic recession of the early 2000s, fell to 5.9 percent in 2005 with the Percentage of workers under-employed recovery, and remains relatively stable since then, even 35 Female with the stronger economic growth of recent years 30 31.6 Total 28.3 (Figure 3.14). 25.8 25 20 Male Female and urban unemployment are higher than male and rural unemployment, respectively, and 15 adjust more to the economy’s performance. Figure 10 3.14 shows that the urban unemployment rate across 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 all years is always higher than the rural rate, and similarly for the female unemployment rate relative to Source: World Bank sta calculations based on EPH, Paraguay. Table 3.6: Under- employment by area 2003-2008 Poverty Assessment 2003 2008 Under- employment National Rural Urban National Rural Urban Visible 8.4 7.8 8.8 7.1 8.2 6.5 Invisible 18.6 13.7 22.5 21.2 16.3 24.5 Total 26.9 21.5 31.3 28.3 24.5 30.0 Source: World Bank staff calculations based on EPH, Paraguay 47 Table 3.7: Characteristics of the under-employed (2008) Under -employed Occupied Under- (Non Under-em- Visible Invisible employed ployed) Characteristics of the worker Age (years) 36.8 29.2 31.2 38.3 Gender (% of males) 42.1 61.9 56.7 62.6 Education Primary incomplete 29.4 24.4 25.7 26.5 Primary complete 19.5 20.5 20.2 20.2 Secondary incomplete 21.5 31.6 28.9 19.5 Secondary complete 10.0 15.0 13.7 14.1 Tertiary incomplete 12.7 7.1 8.6 10.0 Tertiary complete 7.0 1.4 2.9 9.6 Total 100.0 100.0 100.0 100.0 Job characteristics Informal worker (% of workers) 79.6 66.0 69.7 65.3 Economic sector (% of workers) Agro 36.5 9.5 16.6 29.8 Fishing 0.0 0.1 0.0 0.2 Mining 0.0 0.5 0.4 0.2 Manufacturing 5.3 16.6 13.6 11.4 Utilities 0.1 0.0 0.1 0.5 Construction 1.5 12.9 9.9 4.5 Commerce 12.7 20.1 18.1 23.3 Restaurants & hotels 1.8 1.5 1.6 2.0 Transportation & communications 1.8 4.1 3.5 4.6 Finance 4.0 0.3 0.2 1.8 Business services 3.1 2.3 2.8 3.4 Public administration 9.0 3.3 3.2 4.3 Teaching 2.0 2.3 4.1 4.5 Health & social services 12.1 0.9 1.2 2.3 Other services 10.0 2.9 5.3 4.3 Domestic servants 0.3 22.8 19.5 2.8 Foreign organizations 0.0 0.0 0.1 0.1 Total 100.0 100.0 100.0 100.0 Chapter 3 Note: Informal workers are defined as salaried workers in small firms, non-professional self-employed and zero-income workers. Under-employed includes two types of workers: (i) persons employed that work less than 30 hours per week and would like to work more hours (visible under-employment) and (ii) persons employed that work less than 30 hours per week, would like to work more hours and their wages are lower than the minimum legal wage (invisible under-employment). Source: World Bank staff calculations based on the 2008 EPH 48 male unemployment. In 2008, 7.6 percent of women in duration of unemployment has increased in the past the workforce are unemployed, a rate that is nearly 3 two years. Average unemployment duration was 7.2 percentage points higher than that of men. months in 2008, reaching its maximum in 2007 when on average a person was unemployed for eight months Unemployment is mainly concentrated in urban before finding a job (Figure 3.16). In addition, female areas: the urban unemployment rate was 7.5 percent, unemployment duration which was similar to men until corresponding to 78 percent of all the unemployed 2006, strongly overstates men in 2007 and 2008. The rise (Table 3.8). Asuncion is the region with the highest in unemployment duration since 2007 seems to be due unemployment rate (8.5 percent of the working to the increased in female unemployment duration. population), but the lowest contribution in total unemployment (13.4 percent). By contrast, almost 34 Unemployment duration is higher among the higher percent of the unemployed live in the Central Region educated population, but the differences between Urban. Rural unemployment was 3.3 percent in 2008, levels of education are lower than in 2003 (Figure 22.5 percent of the total unemployed. At the national 3.17). However, the duration is generally higher in 2008 level the unemployment rate was 5.9 percent in 2008 for a person with tertiary incomplete was on average corresponding to 167,622 unemployed Paraguayans. unemployed for 10 months. Unemployment has negative consequences for Youth employment and unemployment Paraguay’s poverty reduction and social welfare. Figure 3.15 illustrates how the poorest in urban Youth unemployment is a concern for many policymakers areas face higher unemployment rates than middle in Latin America. The successful incorporation of income and high income households. In rural areas, youth into the labor market is fundamental for the unemployment prevails among middle class. The development of their career path, and the reduction of horizontal line shows the national unemployment rate risks to which youth become vulnerable. Inactive youths (5.9 percent). In general unemployment affects even (those not in schools and not working or looking for a more disproportionately the urban poor. job) are generally at risk to become involved in crime and violence. Unemployment duration Labor force participation rates for youths in Although the national unemployment rate has Paraguay decreased by almost three percentage remained stable between 2005 and 2008, the points between 2004 and 2008 (from 63.4percent to Figure 3.14: Unemployment rates by area and gender Urban Rural 16 16 Poverty Assessment 14 14 12 12 11.0 Unemployment rate Unemployment rate 10 10 Male 7.9 8.1 8 8 7.1 7.6 6.7 6 6 5.9 5.7 5.9 Total 4 4 2 2 Female 0 0 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 49 Source: World Bank sta calculations based on EPH, Paraguay. Table 3.8: Unemployment and contribution to unemployment, Paraguay 2008 Percentage of Unemployment Total number of Contribution total Population rate unemployed unemployment National 100% 5.9 167,622 100% Region Asuncion 8.4% 8.5 22,429 13.4 Central Urban 27.1% 7.2 56,870 33.5 Other Urban 23.1% 7.6 51,412 30.7 Other Rural 41.4% 3.3 37,680 22.5 Area Urban 58.6% 7.5 130,711 77.5 Rural 41.4% 3.3 37,680 22.5 Source: World Bank staff calculations based on EPH, DGEEC Paraguay Figure 3.15: Unemployment Rates Figure 3.17: Unemployment duration by per capita household income, 2008 by educational level 12 Unemployment duration (months) 12 10 10 8 Unemployment rate 8 2008 6 2003 6 Urban 4 4 Rural 2 2 0 0 No Primary Primary Secondary Secondary Tertiary Tertiary 1 2 3 4 5 6 7 8 9 10 education incomplete complete incomplete complete incomplete complete Income decile Source: World Bank sta calculations based on EPH, Paraguay. Source: World Bank sta calculations based on EPH, Paraguay. Figure 3.16: Unemployment duration by area and gender Urban Rural National Female 10 12 9 8.5 8 10 7 7.2 Male 8 6 5.3 5.2 5.4 Months Months 5 4.8 5.6 6 4 5.0 Chapter 3 3 3.1 4 2 2 1 0 0 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 50 Source: World Bank sta calculations based on EPH, Paraguay. 60.4 percent). This is mainly the result of an increasing More than half of the youth female and male number of young people attending school and staying unemployed in 2008 were not entering the job longer in the educational system, thus delaying their market for the first time. Around 66 percent of young entrance to the labor force. However, a rising number unemployed females and 64 percent of unemployed of young people in Paraguay work in the informal males had worked before (Figure 3.19a). In 2003, these economy (three out of four working youth in 2008 percentages were lower, only 58 percent of young were informal workers), where they earn low wages unemployed females had had a previous job. Figure (almost half the amount of prime adult wages)30 and 3.18.b show the economic sector in which unemployed are often subjected to poor working conditions (only youth worked in their previous jobs. Youth were largely 22 percent of the youth have a formal contract, while employed in three sectors: services (domestic servants only 10% were contributing to the pension system). and commerce), manufacturing and construction. These results are consistent with the findings of a These employment patterns for the youth suggest that regional ILO study: “Many young workers in Latin downward fluctuations can strongly affect the youth, America are in the informal economy, engaged in poor- and women in particular. When youth enter the labor quality, unproductive and non-remunerative jobs that force to contribute to household income or because are not recognized or protected by law, and lack rights they have dependants to support, this can turn into an at work, representation and adequate social protection� important constraint to the household’s welfare. (ILO, 2006). Job Creation and Job Destruction Youths remained unemployed for longer periods During the Crisis and have major difficulties entering the labor market. Unemployment rate for youth was 7.2 percent This section presents a quick look at the distributive in 2008. Among the youth, unemployment is higher impact of the recent labor market performance. Sorting for females (7.8 percent) than for males (6.5 percent). employment changes by position of employment While unemployment rate decreased over the period, and type of economic activity reveals if the patterns the duration of unemployment increased from 5 in net job creation/destruction are the same along months in 2003 to almost 8 months in 2008 (Figure different industries. Utilities, Transportation and 3.18.a and 3.18.b). Communications, Public Administration and Figure 3.18: Youth Unemployment a. Youth unemployment rate b. Youth unemployment duration 10 8 7.5 7.6 9 7 6.1 6 Unemployment rate 8 5.3 5.1 Female 7.8 5 4.9 7 Total 7.2 4 Poverty Assessment Months 6.5 3 6 Male 2 5 1 4 0 2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008 Note: Youth includes population 15 to 25 years old. Source: World Bank sta calculations based on EPH, Paraguay. 30 Average wage in 2008 for young workers (aged 15 to 25 years old) was Gs 4,587, while the average monthly wages of workers 26-65 51 years old was Gs 9,498. Figure 3.19: Previous sector of employment for youth aged 15 to 25 years old a. Worked before and is looking for a job b. Previous economic sector of work of unemployed Public Administration Commerce Other services Construction Industry Utilities Mining Hotels & Restaurants Education Agriculture Transp & Commun Finance Domestic servants Services to rms Health & Soc Serv 68 100 66 64 80 62 Percentage Percentage 60 60 58 66 64 (2008) 40 56 60 (2008) 54 58 (2003) 20 42 (2003) 50 0 Female Male Female Male Source: World Bank sta calculations based on EPH, Paraguay. Domestic Servants are the only economic activities between the semi-skilled (complete secondary) and that have recorded a net job creation. The highest the unskilled (incomplete secondary or less) remained job destruction has occurred for the self–employed in without change (around 1.3). In terms of hours worked almost all sectors (Table 3.9). per week, the semi-skilled work the most, and show an increase throughout the period. Large differences Income can also be seen in the average hourly wages between formal and informal workers (Figure 3.22). The price of work: evolution and differentials of labor income Determinants of income Real hourly wages (deflated by the CPI) fluctuated This section presents the econometric results to little over the period 2003-2008, with the exception better explore the relationship between education of the temporary drop in 2006, but always remained and earnings. The equations estimate the logarithm higher for men than women (Figure 3.20). In 2003, of the hourly wage on educational dummies and men earned 4.5 percent more per hour than women, other control variables like age, age squared, regional and worked 19 percent more hours per week. The wage dummies, and an urban/rural dummy, for men and gap remained constant until 2003, but widened in 2006 women separately. Estimating the return to education reaching a record level of 1.21. In 2008, the wage gap has been an important econometric exercise since the narrowed to 1.05 but the hour gap grew to 20 percent. seminal work of Mincer (1974). The wage gap between the most and least skilled Returns to college education fall between 2003 has been shrinking between 2003 and 2008, but the and 2008, but have remained high, while returns to most qualified still earn 2.4 times more than unskilled secondary or primary education have on average workers in 2008 (Figure 3.21). Education continues remained constant during this period. In 2008, a Chapter 3 to be an important determinant of income. Figure worker aged between 25 and 55 with primary education 3.21 shows labor variables by educational groups, in earned on average 20 percent more than a similar which workers were classified into low, middle and worker with primary incomplete or no education, a high education categories, according to their years of worker with completed secondary education earned 52 formal education. The hourly wage for the most skilled 56 percent more, and a worker with complete college decreased during the period, while the hourly wage gap earned 65 percent more (Figure 3.23). Table 3.10 shows Table 3.9: Distribution of Net job creation by labor relationship: 2007-2008 Labor relationship Economic Sector Entrepreneurs Wage earners Self-employed Zero income Total Agriculture (4,019) 6,867 (36,804) (10,268) (44,224) Fishing (713) (420) (7,677) 118 (8,692) Mining 0 (1,261) (1,068) 0 (2,329) Manufacturing 2,952 14,250 (7,024) 10,956 21,134 Utilities 0 2,242 231 0 2,473 Construction 4,077 11,999 (2,906) 507 13,677 Commerce (2,355) 23,955 (5,081) 22,857 39,376 Restaurants & hotels (4,658) 636 (3,392) (3,459) (10,873) Transportation & Comm. 752 3,158 12,586 1,597 18,093 Finance (188) 12,490 (535) 0 11,767 Business services 4,485 4,180 3,169 (52) 11,782 Public administration 0 22,113 0 0 22,113 Teaching 416 15,668 (222) 0 15,862 Health & social services (591) (2,788) (543) 1,337 (2,585) Other services 2,691 (415) 7,466 1,788 11,530 Domestic servants 0 6,242 0 0 6,242 Foreign organizations 0 (998) 0 0 (998) Total 2,849 117,918 (41,800) 25,381 104,348 Source: World Bank staff calculations based on EPH, Paraguay Figure 3.20: Hourly wages and hours of work a. Hourly wages (Guaranies 2008) b. Weekly hours of work Poverty Assessment 10,000 Male 55 Male 8,000 50 Female Guaranies (2008) Hours of work 6,000 45 Female 4,000 40 Total 2,000 35 Total 0 30 2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008 53 Source: World Bank sta calculations based on EPH, Paraguay. the results of the Mincer equations for male and female unobservable factors have fluctuated over the period. workers in Paraguay. The results are different if the analysis is restricted to urban salaried workers. The Mincer equation is also informative with respect to two interesting factors –the role of unobservable The coefficients in the Mincer regressions are different variables and the gender wage gap. The error term for men and women, indicating that they are paid in the Mincer regression is usually interpreted as differently even when having the same observable capturing the effect of factors that are unobservable in characteristics (education, age, location). To further household surveys, like natural ability and contacts, on investigate this point we simulate the counterfactual hourly wages. An increase in the dispersion of this error wage that men would earn if they were paid like term may reflect an increase in the returns to these women. In all cases the ratio between the average unobservable factors in terms of hourly wages (Juhn et of this simulated wage and the average of the real al. [1993]). Table 3.11 shows the standard deviation of male wage is less than one, reflecting the fact that the error term in each Mincer equation. The returns to women earn less than men even when controlling for Figure 3.21: Hourly wage and weekly hours by educational level a. Hourly wages (Guaranies 2008) b. Weekly hours of work 25,000 52 Mid 20,000 High 50 (Gs 16,198) Low Skilled Guaranies (2008) 48 Hours of work 15,000 Mid 46 10,000 (Gs 8,982) 44 5,000 High Skilled Low 42 (Gs 6,632) 0 40 2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008 Source: World Bank sta calculations based on EPH, Paraguay. Figure 3.22: Hourly wage by informality Figure 3.23: Mincer equation. Estimated in Guaranies of 2008 coefficients of educational dummies (All workers 25-55 years old) 16,000 1.0 Hourly wages (in $Gs of 2008) 0.8 College 12,000 Formal 0.6 8,000 Secondary Chapter 3 0.4 4,000 0.2 Primary Informal 0 0.0 2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008 54 Source: World Bank sta calculations based on EPH, Paraguay. Source: World Bank sta calculations based on EPH, Paraguay. observable characteristics. This result has two alternative Gender wage gaps in Paraguay interpretations: it can be either the consequence of gender discrimination against women, or the result of This section explores possible explanations for men having more valuable unobservable factors than the observed levels of gender wage gaps and the women (e.g. be more attached to work). extent to which the wage gaps do not correspond to differences in observable individual and job-related The gender wage gap seems to have slightly increased in characteristics31. Table 3.9 shows relative wages by the early 2000s, decreased in 2004-2005 and increased each of the characteristics considered in the analysis. again in the last two years (Figure 3.24). In 2008, the Although these relative wages have not been wage gap decreased again reaching a similar level as controlled by differences in individuals’ characteristics in 2005. The next section presents a broader analysis of they are indicative of the heterogeneity in wages, some gender wage gaps in Paraguay. interesting patterns arise. Table 3.10: Mincer equation. Estimated coefficients of educational dummies by gender31 All workers (25-55 years old) Men Women Year Primary Secondary College Primary Secondary College 2003 0.209 0.555 0.929 0.228 0.474 0.994 2004 0.110 0.505 0.800 0.130 0.596 0.769 2005 0.177 0.634 0.661 0.101 0.508 0.890 2006 0.246 0.514 0.815 0.143 0.458 1.037 2007 0.107 0.438 0.798 0.023 0.624 0.800 2008 0.193 0.561 0.644 0.211 0.420 0.834 Source: World Bank staff calculations based on EPH, Paraguay Table 3.11: Mincer equation. Estimated coefficients of educational dummies Dispersion in unobservables All workers Urban salaried Year All Men Women Men Women 2003 0.06 -0.54 0.05 0.65 0.59 Poverty Assessment 2004 0.28 -0.26 0.19 0.64 0.57 2005 0.02 0.19 -0.11 0.60 0.57 2006 -0.45 0.35 0.00 0.60 0.56 2007 0.02 0.13 0.06 0.56 0.57 2008 -0.02 -0.07 -0.05 0.62 0.55 Source: World Bank staff calculations based on EPH, Paraguay 31 The analysis of the gender wage gap in Paraguay presented in this section used the methodology of matching comparisons proposed by Ñopo (2008). The analysis is limited to individuals between 18 and 65 years old with positive labor earnings in their primary occupation 55 and no-missing information in their individuals and job-related characteristics. Ethnic differences in wages are noteworthy in ethnicity that occurs in Paraguayan workers. Table Paraguay. Earnings differences between minorities 3.13 shows the decomposition of the gender wage gap and non-minorities are higher than those found using the matching decomposition technique proposed between males and females. The age profiles of by Ñopo (2008) that controls for socio-demographic earnings differ by gender: prime age for males (45 to characteristics32. 54 years old) is higher than for females (35 to 44 years old). Regarding education, as expected, those with An important change in the unexplained component tertiary education complete have higher earnings. of the wage gap also occurs when education is added as a control variable. It induces the unexplained The higher gender wage gap is found in urban areas. component of the wage gap to jump from 11% to 21% Among married people males earn more than females, of average females’ wages. This means that for the same but for those never married and those widowed, level of education between men and women, non- divorced or separated females earn, on average, more observables (ie, culture, preferences, discrimination, than males. etc.) increase as an explanation of the wage gap. Among employers, females also earn more than males; The inclusion of time worked substantially moves up among private employees the earnings differences the unexplained component of the wage gaps (from are almost non-existent and for self-employed, public 21.3% when controlling by race, age and education employees and, obviously, domestic servants, males to 33.6%). Table 3.14 presents some additional earn more than females. The gender wage gaps are also wage gap decompositions considering job related more pronounced for those who work either part-time characteristics. or over-time. Figure 3.25 shows the magnitude of unexplained wage The total gender wage gap (∆) for Paraguay in 2008 gaps along percentiles of the earnings distribution. It is 5.5%. The sole inclusion of ethnicity as a matching shows a higher unexplained wage gap at the low tail variable implies an unexplained component of the of the earnings distribution with a decreasing pattern wage gap that surpasses the total gap. This fact is before the 80th percentile which is followed by a slightly an indicator of the interplay between gender and increasing slope at the higher tail of the distribution. The patterns observed after controlling for different sets of matching variables are qualitatively similar. Figure 3.24: Gender Wage gap Conclusions and Policy Urban Salaried workers Considerations Paraguay shows improvements in almost all labor 1.00 0.97 0.95 market indicators during the economic growth of 0.82 0.86 0.80 0.71 0.73 2003-2008. Labor force participation and employment Gender wage gap have increased, while unemployment and levels of 0.60 informality have decreased. However, the level of 0.40 under-employment and the duration of unemployment 0.20 (especially for urban workers) have both increased, 0.00 implying an underlying weakness in the labor market. Chapter 3 2003 2004 2005 2006 2007 2008 Compared to other Latin American countries with Source: World Bank sta calculations based on EPH, Paraguay. similar levels of per capita income, Paraguay ranks 32 Each column in the Table 3.10 corresponds to a different decomposition (that adds one by one control variables to the matching set). 56 In that sense, the first column presents the decomposition controlling only for ethnicity; the second for ethnicity and age; the third for ethnicity, age and education, and so on. Table 3.12: Relative wages by demographic and labor characteristics Female Male Wage ($) 7853.9 8286.7 Ethnicity - Non Minority (%) 74.7 65.9 Age (%) 18 to 24 15.8 20.0 25 to 34 29.9 29.0 35 to 44 24.9 23.9 45 to 54 19.0 16.7 55 to 65 10.3 10.4 Education (%) None 1.8 2.2 Primary Incomplete 14.6 16.9 Primary Complete 18.2 20.0 Secondary Incomplete 14.7 24.6 Secondary Complete 19.2 21.0 Tertiary Incomplete 21.5 9.5 Tertiary Complete 10.0 5.9 Urban (%) 76.5 71.0 Marital Status (%) Married (Formal or Informal) 62.4 67.1 Widowed, Divorced or Separated 7.9 3.2 Never Married 29.7 29.7 Type of Employment (%) Employer 3.8 9.0 Self – Employed 44.8 24.1 Private Employee 31.8 57.1 Public Employee 19.6 9.8 Domestic Servants 0.0 0.0 Time worked (%) Part time 33.2 13.2 Full time 29.8 32.6 Over time 37.0 54.3 Small firm (%) 63.0 57.2 Formality (%) 25.8 21.0 Economic Sector (%) Poverty Assessment Agriculture, Hunting, Forestry and Fishing 2.3 10.7 Mining and Quarrying 0.0 0.5 Manufacturing 14.2 18.4 Electricity, Gas and Water supply 0.4 0.5 Construction 0.1 14.1 Wholesale and Retail Trade and Hotels and Restaurants 41.5 25.6 Transport, Storage 2.3 8.1 Financing Insurance, Real Estate and Business Services 6.3 6.1 Community, Social and Personal Services 32.9 16.0 57 Occupation - White Collar (%) 69.0 35.1 above the median in terms of the levels of education of its population. However when analyzing the level Figure 3.25: Unexplained Gender Wage Gap of education of the labor force, Paraguay has a higher by Percentiles of the Wage Distribution share of workers with low levels of education and a of Males and Females high level of informality. Among all countries in the Ethnicity & Age Demographic Set + Education 200 region, Bolivia is the only country that has a higher % of female wage share of its labor force with low levels of education 150 (55.3 percent of workers in Bolivia has less that 100 primary education complete), and the only country 50 to have higher levels of informality. Although the 0 share of informal workers has decreased over time, 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Paraguay maintains sizeable levels of informality. Wage percentile Source: Based on Nopo H. (2009). Table 3.13: Gender Wage Gap Decomposition + Presence + Presence of other wage + Marital Ethnicity + Age + Education + Urban of children earner Status in the HH member in the HH ∆ 5.5% 5.5% 5.5% 5.5% 5.5% 5.5% 5.5% ∆0 10.4% 11.0% 21.3% 19.8% 20.2% 21.0% 20.7% ∆M 0.0% 0.0% 0.3% -1.3% -4.6% -7.3% -10.6% ∆F 0.0% 0.0% -1.3% -0.9% 0.6% -3.0% -2.1% ∆X -4.9% -5.5% -14.8% -12.1% -10.6% -5.1% -2.5% % CS Males 100.0% 100.0% 88.8% 79.6% 66.7% 53.9% 38.0% % CS Females 100.0% 100.0% 97.1% 92.7% 85.8% 67.6% 56.3% Source: Based on Ñopo H. (2009) Table 3.14: Gender Wage Gap Decomposition- Job Related Variables Race, & Type & Time & Expe- & For- & Occu- & Small & Sec- age & & Tenure of Worked rience mality pation Firm tor education Empl. ∆ 5.5% 5.5% 5.5% 5.5% 5.5% 5.5% 5.5% 5.5% 5.5% ∆0 21.3% 33.6% 18.4% 21.4% 25.8% 29.0% 17.1% 23.7% 17.2% ∆M 0.3% 1.9% -3.1% -2.3% 1.7% -2.2% 10.1% 2.2% -5.1% Chapter 3 ∆F -1.3% -8.3% 0.8% -0.7% -4.1% 0.9% -5.6% -3.5% -2.2% ∆X -14.8% -21.6% -10.7% -13.0% -17.8% -22.2% -16.0% -16.9% -4.4% % CS Males 88.8% 67.5% 62.6% 61.4% 80.7% 75.5% 61.5% 73.7% 42.3% % CS Females 97.1% 81.4% 80.0% 80.8% 93.1% 87.7% 82.3% 91.9% 73.2% 58 Source: Based on Ñopo H. (2009) Box 3.2: Explaining the gender Wage gap using the Matching decomposition Technique The technique applied for the wage gap decompositions follows the one developed in Ñopo (2004). According to that technique, males and females are matched on the basis of their observable human capital characteristics. The resulting matched females and males make up a set that reflects a synthetic situation in which there is a labor market where both genders have exactly the same observable characteristics. Thus, the gender differences in pay that prevail in such a set of matched individuals can be regarded as unexplained by observable characteristics. On the basis of that set of matched individuals, and comparing it to the set of unmatched females and males, the gender wage gap is decomposed into four additive terms. Delta-M: reflects the fact that some males have combinations of observable characteristics that no female has achieved. Delta-F: captures the role on the gender wage gap of the fact that some females have combinations of observable characteristics that no male has. Delta-X: accounts for the differences in the distributions of observable characteristics among those females and males with the same observable characteristics. Delta-0: is the component of the gender wage gap that cannot be explained by differences in observable characteris- tics between genders. Although some authors have traditionally referred to this component as a measure of gender- based discrimination in pay in labor markets, we prefer to refer to it as a measure of unexplained differences (either because of the existence of In addition, women face the worst employment To increase productivity, and also have positive effects outcomes. Female workers experience longer on Paraguay’s rankings in the Human Opportunity unemployment duration, higher levels of informality, Index, it is important to increase the quality of education and a strong wage gap that cannot be explained by and give incentives for the youth to finish their formal observable characteristics. Ethnicity plays a more education, in particular in light of the problems found important role in wages, however, as the earnings in secondary education. For workers that are already in differences between minorities and non-minorities the labor force, one could consider training programs are higher than those found between males and within their industries. females. Parts of the structural problems affecting poverty in the urban sector are low productivity and the low education level of workers. To increase the poverty Poverty Assessment impact, attention should focus on formalizing the labor force, by decreasing entrance costs into the formal sector and increasing the benefits to small firms of formalizing. According to the assessment by ILO (2003), the labor market is characterized by a low compliance with laws and regulations. According to the study, informality is mainly the outcome of several factors such as inadequate norms or rigid laws for the development of firms and an inefficient system of 59 incentives. Rural factor markets and Poverty Introduction While agriculture is still the major source of livelihoods for poor people in rural Paraguay, the importance of Rural areas continue to be the main contributors to rural labor markets, and the rural nonfarm economy poverty and extreme poverty in Paraguay. As seen in has increased. As seen in Figure 4.2, in 2003 wages Figure 4.1, while the share of the poor and the extreme represented approximately 6 percent of the incomes of poor living in rural areas has decreased in the last the rural poor. By 2008, they represented approximately decade, the rural poor are still the majority of the poor 10 percent. A smaller increase has occurred with the and especially the extreme poor (54 and 68 percent, share of income from non-agricultural self-employment respectively). activities. It went from approximately 12 percent to 15 percent. In addition, it is well known that migration and Figure 4.1: Contribution of the rural poor to total poverty in 1997, 2003 and 2008 1997 2003 2008 Urban Rural 33% 48% 46% Chapter 4 67% 52% 54% 60 Source: World Bank sta calculations based on EPH, Paraguay. remittances have also increased as a source of income for rural households in the country (see Figure 4.3). These Figure 4.2: Sources of Income three pathways are complementary: nonfarm incomes by Quintile can enhance the potential of farming as a pathway out of poverty, and agriculture can facilitate the labor and Agricultural wage earner Agricultural self-employed No agricultural wage earner No agricultural self-employed migration pathways. 100% 90% Asset endowments and the constraints imposed by 80% 70% markets are key determinants of how rural households 60% 50% design their livelihood strategies. This chapter examines 40% how access to land, financial and labor markets affects 30% 20% the probability of a household being poor. The first sub- 10% 0% section looks into access to land, the most important 2003 2008 2003 2008 asset in rural areas. The next sub-sections discuss the Quintile 1 Quintile 5 role of credit and labor markets in determining the Source: World Bank sta calculations based on EPH, Paraguay. likelihood of poverty. Land Markets Figure 4.3: Remittances as a share of total household income Land is the most important asset in rural areas, and by density area land ownership determines how rural households allocate their labor. As seen in Figure 4.4 below, land rich families tend to dedicate more time to independent 10% Rural agriculture activities, while the land poor dedicate 8% Urban more time to wage and independent work in non-farm 6% activities. It is interesting to note that the land poor are 4% also less likely to engage in farm wage work. This is likely 2% due to the fact that agriculture in Paraguay has become less labor intensive over time. The growth of soybean 0% 2003 2004 2005 2006 2007 2008 cultivation made it more capital and land intensive. Also, since capital and land rental markets are imperfect, and land ownership may facilitate access to credit, the Source: World Bank sta calculations based on EPH, Paraguay. landless and the land-poor may not be able to engage in commercial farming which would require access to more land and credit. Figure 4.4: Land Ownership and the Allocation of Labor In Rural Paraguay Because land ownership is highly unequal in Paraguay, land is a likely determinant of equality of opportunity Agricultural self-employed Agricultural wage earner No agricultural self-employed No agricultural wage earner Poverty Assessment in rural areas via the investment in human capital .4 channel. Some studies have shown a close link .3 between land ownership and investments in nutrition Share of hours and education (Galor, Moav, Vollrath, 2006). As seen .2 in Figure 4.5, land ownership inequality has remained .1 high in Paraguay, despites several efforts to improve its distribution via land reform (Carter and Galeano, 1995). 0 Not surprisingly, as a country, Paraguay is one of the 0 10 20 30 Land own pc worst performers in terms of equality of opportunity in 61 Latin America. As seen in Chapter 2, only six countries in Source: World Bank sta calculations based on EPH, Paraguay. the region (out of 19 for which the index was calculated) perform worse than Paraguay. Figure 4.5: Land Owned and Used Gini across Time In Paraguay Land ownership is strongly associated with the probability of not being poor in rural Paraguay. As Hectars indicated in Figure 4.6 below, there is a 20 to 25 percent 0.93 owned probability of being poor if a household owns less 0.91 than 30 hectares. This probability is drastically reduced Hectars Gini Coe cient used once households reach the 30 hectare threshold. This 0.89 is likely due to the fact that at areas below 30 hectares, 0.87 households have difficulty in engaging into high value commercial farming. There is probably a minimum area 0.85 2003 2004 2005 2006 2007 2008 in which land ownership allows a household to gain access to credit, and therefore purchase modern inputs and mechanize its crop. This is consistent with the credit Source: World Bank sta calculations based on EPH, Paraguay. access story discussed in the next section. Figure 4.6: Land Ownership Within those owning land, moderate and extreme and Poverty poor have seen a significant decrease in the average land owned compared to the non-poor. Figure 4.7 shows that moderate poverty has experienced a huge .2 decrease going from an average size land of 15 hectares Probability of being poor .15 to almost 4.5 hectares in 2008. The average size of extreme poor has been divided by 2 in the last decade. .1 This has conducted to a widening of the gap between .05 poor and non poor in the recent years. 0 0 20 40 60 80 Land rental markets could potentially reduce the Land owned effects of land ownership inequality on poverty. For instance, in developed countries, there is very little link Note: The probability is only computed on households that owned some land. Source: World Bank sta calculations based on EPH, Paraguay. between land owned and land used, and more than 50 percent of land used is rented by users (Wolrd Bank, 2008). But as seen in Figure 4.5 above, the land use Gini Figure 4.7: Land Ownership generally goes hand in hand with the land owned Gini, over time indicating that land rental markets do not seem to be distributing land to the land poor. 25 Non 20 Extreme poor Another way of looking at how effective land rental poor Moderate markets are at reallocating land from large owners to 15 Percentage poor the land poor is to look at the relationship between 10 land used and land owned per unit of labor in the 5 household. In a world of perfect rental markets and Chapter 4 0 decreasing returns to scale, this relationship should be 2003 2005 2008 one in which there is perfect separability between land used and owned. That is, there would be an optimal Note: Figure based on household survey data and not an agricultural census, thus may not be constant amount of land cultivated per unit of labor in fully representative for land issues. For this gure, extreme and moderate poor outliers the household, and land owned would not determine 62 (>1000 ha) have been dropped. Source: World Bank sta calculations based on EPH, Paraguay the amount of land used per capita. Landless and land Figure 4.8: Theoretical outcomes Figure 4.9: Empirical Relationship between of land rental markets owned and used land in Paraguay 2003 2004 2005 2006 2007 2008 45º line 30 Autarchic Land Land used per unit of labor 20 Land used pc 10 Perfect Rental 45º 0 0 10 20 30 Land Owned per unit of labor Land owned pc Source: World Bank sta calculations based on EPH, Paraguay. poor households would rely on rental markets to gain labor for cultivation, and the land poor must rely on access to additional land so that they could cultivate the labor markets or non-farm independent activities. optimum amount per labor unit. Land rich households However, this allocation of the factors of production in would rent out the land in excess of the optimal amount agriculture may not be the most efficient if there are, as to be cultivated. The relationship between land owned discussed below, significant labor supervision costs in and land used per labor unit would be a horizontal large farms. line. On the other hand, in a perfectly autarchic world where there is no possibility of renting land in or out, The lack of well functioning land rental markets may this relationship would be a 45 degree line. That is, each represent a loss of productivity if there is an inverse land owning household would cultivate the area in their relationship between farm size and productivity. This possession, and the land poor would have to either relationship is well studied in the literature35. While there sell excess labor to large owners and/or to non-farm is no final consensus on the matter, the conventional business, or engage in non-farm business activities as view is that, for most crops, especially labor intensive self-employed. Figure 4.8 shows these two theoretical crops, farms that employ mostly family labor are more possibilities.34 efficient. This efficiency advantage of smaller farms is often attributed to the existence of diseconomies of How effective are land rental markets in redistributing scale (Binswanger and Deininger, 1997) involved in land from the land rich to the land poor in Paraguay? hiring labor to cultivate larger areas. Figure 4.9 below Not very effective it seems. Figure 4.9 shows the shows this empirical relationship for Paraguay. empirical relationship between land used and owned per unit of labor in Paraguay for several years between While it is almost impossible to establish that there is 2003 and 2008. As can be seen, land rental markets in a causal inverse relationship between farm size and Poverty Assessment Paraguay have remained largely autarchic. Very little productivity in the country, there seems to be a strong land is transferred from the land rich to the land poor association between the two. Small farms seem to be and the landless. Owners seem to cultivate most of considerably more productive than larger farms. Thus, their land. Therefore, larger owners must rely on hired if this relationship was indeed causal in Paraguay, land 34 A constant optimal land cultivated per unit of labor would exist also under constant returns to scale technologies, but when there are significant increasing labor supervision costs in agriculture. That is, productivity would drop for every additional hired worker in the farm, leading to an optimal amount of hired labor per unit of family labor in the household. This optimal amount could be zero in which case the most efficient farm would be the one that employs only family labor. 35 Literature that expands recently with greater survey data availability, see among others Sen (1962) Barret (1996) Berry and Cline(1979) 63 and Benjamin and Brandt (2002). Figure 4.10: The Inverse Relationship between Farm Size and Productivity in Paraguay Traditional Crops Cash Crops 2003 2004 2005 2006 2007 2008 All years 2003 2004 2005 2006 2007 2008 All years 30 30 20 20 Productivity Productivity 10 10 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Land percentiles Land percentiles Source: World Bank sta calculations based on EPH, Paraguay. reallocation from the land rich to the land poor via well dispute resolution when there are overlapping claims. functioning rental markets would likely increase the While traditionally expensive to implement, there productivity of agriculture in the country, and would are new mechanisms that can increase the security of especially help the poor. property rights over land (WDR, 2008). In addition to invigorating land rental markets, cost-effective systems More secure and unambiguous property rights over of land administration can also facilitate agricultural land ownership could increase the activity of land investment and lower the cost of credit by increasing the rental markets in Paraguay, thereby allowing these use of land as collateral, thus reducing risk for financial markets to transfer land to more productive uses and institutions. Thus, as numerous studies show, providing users. If tenure is insecure or restrictions constrain land land owners or users with security against eviction leasing, productivity-enhancing rental transactions enhances their competitiveness by encouraging land- will not fully materialize or the poor may be excluded. related investment. There is ample evidence that weak property rights and restriction on leasing weaken rental markets. In the While better establishment and enforcement of Dominican Republic, Nicaragua, and Vietnam, insecure property rights are necessary conditions for well land ownership have been shown to reduce the functioning property rights, they might not be propensity to rent. In Ethiopia, fear of losing the land, sufficient for the rural poor to gain more access to together with explicit rental restrictions, was the main land. The lack of rental transactions may also be a reason for suboptimal performance of rental markets. function of constraints in the credit market for small In India, tenancy restrictions reduce productivity and farmers. Lack of technical assistance and limited output equity. However, land rentals are increasing where they marketing opportunities may also hinder rental market had not been practiced extensively earlier—as in Eastern participation by the land poor. Large landholders may Europe; Vietnam, where rental participation quadrupled also have few incentives to rent out land because the to 16 percent in five years; and in China, where rentals land tax valuations are low (World Bank, 2007). Finally, allow rural communities to respond to large-scale out- socio-cultural factors which make rental transactions Chapter 4 migration (WDR 2008). hard to accomplish because of a lack of trust and information between operators from different classes Programs to increase the security of land property may also need to be addressed. rights have the potential to stimulate pro-poor rental 64 markets in Paraguay. Such programs would include Another way to increase land access by the rural land titling, land registration and management, and poor is through the land sales market or land reform. However, land sales are even less common in Paraguay, land productivity and hold back development. To and without the support of public policy, the rural poor overcome such inequalities, ways of redistributing are unlikely to gain more access to land via purchases assets, such as land reform, may be needed. Postwar (Masterson, 2007). First, large farmers derive more than Japan, the Republic of Korea, and Taiwan (China) show the value of production from land ownership. Access that land reform can improve equity and economic to credit, political influence and cultural traditions performance. But there are many cases where land are known to increase the reservation price of land of reform could not be fully implemented or even had potential sellers (Binswanger and Deininger, 1997). negative consequences. Evictions of tenants or changes Second, despite their productivity advantage, small of land use ahead of legislation that would have given farmers have very little access to long term credit to be greater security to tenants or allowed expropriation of able to purchase land via mortgage financing. Third, underused land often made prospective beneficiaries transaction costs for property transfers are relatively worse off or prompted land owners to resort to even high in Paraguay. As shown in Figure 4.11 below, it costs less-efficient techniques. If land is transferred through on average 3.5% of the value of the property to transfer redistributive land reform, improvements in access to its ownership. Since the averages are estimated from managerial skills, technology, credit, and markets are actual transfers between individuals who are unlikely to essential for the new owners to become competitive. be poor, and hence are unlikely to transfer small plots Some tenancy reforms have proved highly effective, of land, this figure is expected to be much higher for but measures to clarify ownership rights are needed the typical land transaction in which the poor would to avoid disincentives for investments. Land reform participate. through market exchange assisted by grants and technical assistance to selected beneficiaries shows Thus, unless land reform policies are implemented, promise, with Brazil the leading innovator, but this it is unlikely that the rural poor will gain access to approach deserves further analysis of costs and agricultural land via the sales markets. Land reform impacts. programs that subsidize land purchases by associations of poor rural households have shown to be effective To be effective, any approach to land reform must and politically feasible in different countries. One of the be integrated into a broader rural development advantages of increasing access to land via the sales strategy— using transparent rules, offering clear market is that land ownership may help the poor gain and unconditional property rights, and improving access to credit. As we discuss below, access to credit is incentives to maximize productivity gains. Yes, it highly correlated to land ownership. can enhance access to land for the rural poor. But to Imperfections in other markets, and expectations of future land price increases, affect markets for land sales more than those for rentals, implying that sales would Figure 4.11: Cost of transferring not necessarily transfer land to the most productive and registering property producers. Historically, most land sales happened under as a percentage of its value distress, requiring defaulting landowners to cede their Guatemala Chile Poverty Assessment land to moneylenders, who could amass huge amounts Colombia Ecuador of it. Sales markets are also thinner, more affected by Venezuela, R.B Panama life-cycle events, and less redistributive than those for Brazil Perú rentals. Land taxes can curb speculative demand and Costa Rica Paraguay encourage better land use, while providing revenue for El Salvador Nicaragua local governments to fulfill their functions. Mexico Honduras Argentina Uruguay Given the highly unequal land ownership of 0 1 2 3 4 5 6 7 8 Paraguay, land markets may not be a panacea 65 for addressing structural inequalities that reduce Source: Doing Business 2009, World Bank. reduce poverty and increase efficiency, reform requires explicit collateral from borrowers. They work with small a commitment by government to go beyond providing local clienteles and apply intense direct screening access to ensuring the competitiveness and sustainability and monitoring mechanisms, so that the asymmetric of beneficiaries as market-oriented smallholders. information problems which have been theorized to explain quantity rationing in formal credit markets may Financial Markets not lead to rationing in the informal segment. The provision of financial services to the poor may Formal lenders work with a wider pool of borrowers be a powerful means of providing low income across the country through the sorts of arm’s length households with the chance to escape from poverty relationships that are subject to asymmetric information and to transform their lives. It is also evident that problems. Formal market interest rates do not appear to there is a strong demand for small-scale commercial be used to ration credit locally or among individuals as financial services – both credit and savings – from low they are set ex-ante, and cannot be differentiated across income households in Paraguay (Robinson, 2001). Thus, farmers, or even across regions. Quantity rationing of improving access to financial services, but especially formal credit thus seems probable. Explicit collateral credit in rural areas of Paraguay may be one of the is required by formal lending institutions. Titled land is most tangible ways of assisting low-income households by far the preferred form of collateral, and some banks escape poverty. do not consider loan applications without it, and when they do, low borrowing ceilings are imposed for loans The majority of the rural poor in Paraguay, especially not backed by land titles. the landless and the land-poor, remain without access to financial services that are needed if they are to Land titling may not be enough to help the land-poor compete and improve their livelihoods. Broader access gain access to cheaper formal credit in Paraguay. High to financial services, especially credit, would expand transaction costs may still prevent small holders from their opportunities for more efficient technology borrowing from commercial banks. Financial contracts adoption and resource allocation. in rural areas involve higher transaction costs and risks than those in urban settings because of the greater Farmers in rural Paraguay can either borrow from spatial dispersion of production, lower population formal (mostly state banks) or from informal lenders densities, the generally lower quality of infrastructure, (usually merchants and trader-lenders). Informal and the seasonality and often high covariance of rural lenders charge high interest rates and do not demand production activities. So banks and other traditional for-profit financial intermediaries tend to limit their activities to urban areas and to more densely populated, more affluent, more commercial areas of Figure 4.12: Access to Credit by Titled the rural economy, including larger farmers. That is, for and Untitled Farms in Paraguay those for whom loan sizes are large enough to cover fixed transaction costs, and legal contracts more easily Titled Farms Untitled Farms enforced. One study shows that even though land titling 2000.0 increased the supply of credit to farmers in Paraguay, 150.0 this positive effect only happened for relatively larger 100.0 50.0 farmers (see Figure 4.12). The estimated probabilities $100 0.0 of being credit rationed in the formal market were Chapter 4 -50.0 indeed negatively impacted by the ownership of land -100.0 -150.0 titles. Nevertheless, it seems that the magnitude of -200.0 such impact is eminently dependent on farm size. 1.0 1.6 2.7 4.5 7.4 12.2 20.1 33.1 54.6 90.0 148.4 While land titles have practically no impact on credit Farm size (hectares) 66 rationing for small farmers, large farmers seem capable Source: Carter and Olinto, 2003. of neutralizing the negative effect of tenure insecurity on access to formal credit. Households owning less than fourteen hectares were still likely to be rationed Figure 4.13: Land ownership out of formal credit markets even when they acquired and the use of modern inputs land title, but land titles seem indeed to reduce the estimated minimum farm size required for participation in the formal credit market from 50 to 13.5 hectares. Despite this considerable difference, this result indicates .25 Share of modern inputs that most small farmers in Paraguay may still not gain .2 access to formal credit after obtaining a land title, since .15 more than 50% of the farm households in the country own less than 11 hectares (Carter and Olinto, 2003). .1 Therefore, few households and small farmers can meet .05 their need for credit in rural Paraguay. 0 5 10 15 20 25 Land owned pc There is a clear link between farm size and the use of Source: World Bank sta calculations based on EPH, Paraguay. modern inputs. As indicated in Figure 4.13, small farmers employ relatively little modern inputs compared to larger holders. This is likely due to the fact that large owners Figure 4.14: Shares of harvest values have access to more and cheaper credit. Therefore, on selected crops (2008) such imperfections in financial markets that give a competitive advantage to larger owners will continue Non-poor Moderate poor Extreme poor 100% to help perpetuate the patterns of land inequality in 90% Paraguay unless policies are enacted to improve access 80% 70% to financial markets for the rural poor. 60% 50% 40% Lack of access to credit is also likely to constrain the 30% 20% ability of the rural poor to engage in the cultivation 10% 0% of commercial export crops that require the purchase Cotton Soybean Wheat Mandioca Sesame Sunflower Sugar... Maize... Maize... of modern inputs. As shown on Figure 4.14 the poor are unlikely to cultivate soybean, wheat, sugar cane and sunflower. Therefore, the rural poor have not been able Source: World Bank sta calculations based on EPH, Paraguay. to profit from the dramatic increase in export crop prices in the last decade.36 agricultural lending institutions have been successful in many now-developed economies such as the Innovation in the provision of rural financial Republic of Korea and Taiwan (China). But in Paraguay, services in Paraguay is direly needed. Micro Financial government efforts to improve rural financial markets Institutions (MFIs) are sometimes offered as a solution. have a record of doing more harm than good, heavily However, MFIs cannot provide the mainstay of rural distorting market prices; repressing and crowding out Poverty Assessment finance. Promoting, improving, or even creating private financial activities; and creating centralized, rural institutions to support a wide range of rural inefficient, and frequently overstaffed bureaucracies financial transactions remains one of the fundamental captured by politics. Therefore it is not surprising that challenges facing the Paraguayan government. The public agricultural and development banks came under range of alternatives is broad. Government-sponsored heavy criticism in the 1980s (WDR, 2008).37 36 In addition to the lack of access to credit, small holders may not engage in commercial farming because of the lack of downstream market linkages, small-scale processing, and distribution channels tailored for small farmers. Models like the “productive alliances� projects which have been successful in Bolivia and Colombia may be relevant. 37 The existing debt overhang in Crédito Agrícola Habitacional, which has tens of thousands of unpaid loans, may also prevent entry of 67 credit suppliers in rural Paraguay. A well documented credit history may be more This represents a huge challenge for participation of effective in increasing access to credit than collateral the rural population in a modern economy, and may land. Credit reporting bureaus that establish hinder efforts to integrate rural households into the individual reputations can help small farmers use labor markets. If rural labor markets are to become an their past credit histories as an asset. A smallholder effective exit out of poverty, policies to boost human may begin by establishing a credit history in the capital accumulation in rural Paraguay will have to MFI sector, often using credit for nonagricultural become key priorities. purposes. The credit bureau establishes a reliable, portable signal of the borrower’s reputation. Armed Returns to education in rural labor markets are as with this signal, a borrower should then be able high as in urban markets. As seen in regression results to climb a lending ladder, moving from the more presented in Table 4.1, the coefficients for years of restricted purposes and term structures of MFI credit education and primary and secondary completion are to standard loan contracts from institutions able to nearly identical. This suggests that there is no reason bear the portfolio risk and term structures required not to focus efforts for increasing human capital for agricultural loans. For a lending ladder to work, accumulation in rural areas. two things must happen. First, a credit report must help lenders select clients and induce clients to Thus, there is no reason for rural development policy repay loans. This becomes all the more essential in Paraguay to be concentrated on support for as competition among lenders rises. Second, agriculture, rather than on improving the quality of the information on a borrower’s credit worthiness and rural labor force or on facilitating the development of reputation must flow up the rungs from MFI to rural labor markets. Note that between 1985 and 2000, commercial lenders. A study of a credit bureau that about 86% of public rural expenditures in Paraguay includes MFIs in Guatemala shows that both can went towards agricultural subsidies – the highest happen (de Janvry, McIntosh, and Sadoulet 2006). share among the nine countries surveyed in a recent World Bank publication – while investments in roads, Labor Markets communications infrastructure, and human capital have lagged (Ferranti, et al., 2005). With labor as the main asset of the rural poor in Paraguay, landless and near-landless households have The gap between the number of new rural workers to sell their labor in farm and nonfarm activities or and the number of new jobs in agriculture is leave rural areas. Making the rural labor market a more likely to grow in Paraguay. As the rural population effective pathway out of poverty is thus a major policy continues to grow rapidly, and access to land and challenge. As we discuss in this section, there are high other agricultural assets continue to be concentrated returns to education for rural workers and policies aimed in the hand of a few, rural labor markets will become at increasing the education level of the rural poor should a key alternative for exiting poverty. Even with be prioritized. Also, a better rural investment climate for migration to cities, rural populations will continue to agriculture and the rural nonfarm economy could help grow fast. Each year’s addition to the rural labor force generate higher quality jobs that are effective in lifting needs to find work in agriculture or the rural nonfarm the rural population out of poverty. economy, or to migrate to the urban economy. While agriculture employs many wage workers, labor The Paraguayan rural poor face significant challenges conditions in agriculture are not always conducive to in the labor market that are not common to other large welfare improvements, in part because of the Chapter 4 Latin American countries. Guarani, rather than Spanish, nature of the production process and in part because is the first language of 73% of the rural population. of a lack of appropriate regulation. But a greater Moreover, rural households have vastly inferior access potential for significant welfare improvement exists to education services than their urban counterparts. in the dynamic high-value crop and livestock sector 68 The average number of years of schooling in rural areas which are labor intensive and exhibit good potential is just 4.8 years compared to 8.4 years for urban areas. for employment growth. Tabla 4.1: Regresión de ingreso VARIABLES (1) rural (2) urban (3) rural (4) urban Incomplete Primary 0.201*** -0.139** (0.050) (0.061) Completed Primary 0.444*** 0.097 (0.054) (0.062) Incomplete Secondary 0.620*** 0.316*** (0.058) (0.061) Completed Secondary 0.846*** 0.595*** (0.072) (0.062) Incomplete Superior 1.075*** 0.863*** (0.069) (0.067) Completed Superior 1.467*** 1.235*** (0.085) (0.067) Years of education 0.081*** 0.094*** (0.003) (0.002) Constant 12.365*** 12.188*** 12.453*** 12.687*** (0.157) (0.129) (0.162) (0.139) Observations 8500 10627 8501 10632 R-squared 0.250 0.415 0.244 0.409 Otras variables incluidas: total de miembros en la vivienda, proporción de miembros sin ingresos, propiedad, acceso al agua, edad, edad^2, casado, hombre, migración inferior a 5 años y tener un miembro al extranjero. Fuente: Calculos del personal del Banco Mundial en base a la EPH, Paraguay. A dynamic rural economy, in both agriculture and Conclusions and Policy the nonfarm sectors, will be key to ensure growth in Recommendations the demand for labor in Paraguay. Perhaps the most basic policy element for a dynamic rural economy is a The rural poor are still by far the larger contributors good investment climate. To improve the investment to extreme and moderate poverty in Paraguay. To be climate, the government can secure property rights, effective, any poverty reduction policies will have to invest in roads, electricity, and other infrastructure, target the rural poor. In this chapter we have investigated develop innovative approaches to credit and the three main markets in which the rural poor engage financial services, as discussed above, and aid in the Poverty Assessment in exchange to better their livelihoods: land markets, coordination of private and public actors to encourage financial markets and labor markets. The choice of agro-based industry clusters. With more investment activities defining their livelihood strategies will mostly and the expansion of rural economic activities comes depend on how constrained there are in participating in the potential for higher-paying jobs, particularly these three markets. off the farm. On the farm, productivity enhancing technologies can boost incomes. With the poorest Access to land continues to be crucial for the rural most likely to remain in agriculture, increasing wages poor, and land policies will continue to have great for agricultural workers offers the greatest potential to potential to reduce poverty in Paraguay. Because land 69 lift many out of poverty. is highly concentrated in the hands of few, and land sales markets are very inactive and costly, the government the human capital of the rural poor. This will be key to should focus on policies aimed at invigorating land increase the productivity of those who opt for generating rental markets. These include enhancing security of income via labor markets. property rights via land titling, land registration and securing the right of private property. Threat to tenure security will continue to hinder the functioning of land rental markets. More active land rental markets are likely to increase the use of land by smaller and more efficient farmers. Therefore, it not only improves equity but also efficiency of resource use. Increasing the access to land ownership via market assisted land reform programs may also improve productivity and reduce poverty in rural Paraguay. However, it is key to do so in a way that does not risk tenure security and property rights. Thus, an approach based in the willing buyer acquiring land from a willing seller, in which the willing buyer receives assistance from the government via subsidies, is likely to be the most recommended. Access to financial markets is as critical as access to land for the rural poor. Although many believe that access to owned titled land will facilitate access to cheaper and more reliable formal credit, the evidence shows that this is true only for larger farmers. Small farmers, even in possession of secured titles, are still rationed out credit markets because of transaction and monitoring costs. Innovative policies to enhance access to formal credit in rural areas are needed. One such policy would be to establish a credit bureau able to collect the credit history of rural borrowers from MFI’s and commercial banks. Finally, labor is still the largest endowment of the rural poor, and in an environment in which land and credit markets are not effective in allocating production factors, the poor sometimes have no choice other than selling labor to the market, even when their most productive employment would be in agriculture. Since the gap between the number of new rural workers and the number of new jobs in agriculture is likely to grow in Paraguay, rural labor markets will become a key Chapter 4 alternative for exiting poverty. A dynamic rural economy will be needed to ensure growth in the demand for labor in Paraguay. Perhaps the most basic policy element for the government of Paraguay to ensure a dynamic rural 70 economy is to promote a good investment climate in rural areas. It should also promote policies to enhance ex-Ante evaluation of the expansion of Cash transfer Programs in Paraguay Introduction The current administration made a policy priority the expansion of the conditional cash transfer programs. Although the number of social protection programs Since the Tekoporã program was �rst launched in in Paraguay is relatively large, their coverage is September 2005, the expected coverage of the program limited.38 In terms of social insurance, pension has always been more optimistic than that shown in coverage is relatively low compared to the rest of the reality. According to the limited �gures available, the region and its administration is segmented with little Paraguay’s set of CCTs reached 18,000 households at or no portability of contributions between Pension the beginning of 2009, reaching nearly 100,000 one Funds. The other social insurance programs, like family year later. A parametric change to raise the amount of allowances or unemployment insurance, are not part the bene�t (by paying additional variable components of the set of social protection policies in Paraguay. In to certain population groups) was jointly implemented terms of social assistance, it is comprised of an atomized with the geographic expansion. set of programs with limited reach, which reflect the absence of a coordinating institution that ensures In the context of the global economic crisis of 2009, the effectiveness and efficiency in the provision of a the expansion of CCT programs was regarded as a consistent social protection policy. For example, various tool to reach, quickly, the poorest population. The similar conditional cash transfer (CCT) programs exist scaling-up of the CCT programs can be considered as Poverty Assessment (Tekoporã, PRO-PAIS II, Ñopotyvô) that are administered a coherent strategy in the context of the recessionary by the same institution (the Social Action Secretariat, world economy, which affected Paraguay’s growth SAS). Nevertheless, other public sector areas also offer more than that of other countries in the region.39 As in limited CCT programs. most CCT programs implmented during the last decade 38 In this chapter Social Protection is considered as the set of programs that provide an income transfer to beneficiaries. Two broad sets of programs comprise social protection: social insurance, normally of contributory nature; and social assistance, traditionally associated with non-contributory programs funded through general revenues. 39 The expansion of social assistance components can be interpreted as the achievement of campaign promises. This expansion was 71 transmitted through an already-existing program, Tekoporã, afterwards generically referred to as CCT, along with the rest of the similar programs. in Latin America, benefit transfers are conditioned on raise the cost of the program and worsen its targeting. compliance with a series of co-responsibilities related Finally, the comparison of alternative scenarios that to health, education, and child nutrition.40 CCT-style simulate the implementation of the non-contributory social assistance policies aim at breaking poverty cycles pension (NCP) program suggests that using the same in the medium and long term, by helping to improve targeting instrument for the CCT and the NCP programs outcomes in health, education, and child nutrition. In would not yield the best results in terms of targeting the short term, these programs help to mitigate poverty and would yield a high cost/bene�t relationship, while and improve income distribution by raising poor excluding other households from access to the coverage household’s spending capacity. Neighboring countries of the social protection programs. such as Uruguay and Argentina also introduced large expansions of social assistance programs in response to The main messages of the analytic work are: firstly, the the latest economic crisis. The strategy followed by the use of targeting programs may allow a more efficient Southern Cone countries was necessary to complement allocation of the additional social expenditure. Second, the social insurance components that proved to be if the parameters of the targeting instrument of the limited in their capacity to give a rapid response to the CCT programs are not updated, it is advisable not to population groups not covered by existing programs. increase the cutoff point of the ICV. Third, a targeted program should base the selection of beneficiaries on This chapter contributes to the discussion around a specific instrument, designed to identify the sector the implementation of social protection programs in of the population that these programs intend to reach. two ways. First, it presents a sensitivity analysis of the Fourth, the parameters of these instruments should targeting instrument (ICV, Life Quality Index; acronym be regularly updated and be adjusted to the specific in Spanish) that is used to select the majority of the CCT geographical area of application. program beneficiaries, in search of an optimal cut-off value for the ICV that minimizes the targeting errors Before continuing the chapter, it is necessary for the (inclusion and exclusion).41 Second, seven alternative reader to be aware of some aspects relating to the data simulation scenarios are presented as an ex-ante and methodology used. The data of CCT beneficiaries evaluation of the expansion of the largest CCT program are not observable by the researcher given that theyare (Tekoporã) as well as a non-contributory pensions (NCP) not captured by the household survey used here. By program that became law. replicating the methodology of the targeting instrument with household survey data (in the rural districts where The results of the simulations suggest that the impact the CCT programs were launched), it is possible to of the CCT program expansion could reduce extreme approximate the identification of eligible households poverty more than overall poverty, and the poverty and simulate their participation in the CCTs. In this way, gaps more than the poverty headcount. These impacts the results presented in this chapter may be considered would be stronger in the rural areas where the program as upper limits for the potential effects of these social was originally launched. The increase in the number of assistance programs on the indicators of wellbeing.42 beneficiaries has stronger effects than the increase in the amount of the benefit. The CCT programs (that use the This chapter is organized in �ve sections. The second ICV) seem to be well-targeted with respect to the poor section provides a brief description of Paraguay’s population, and at the current cut-off values used to social protection policies, including a sub-section on select beneficiaries, the exclusion errors would be larger CCT programs. The third section presents an incidence than the inclusion ones. The results from this chapter analysis of the social protection programs, together Chapter 5 suggest that an increase in the cut-off value would with an assessment of the targeting instruments of 40 The participation in the program is conditional on the compliance of two co-responsibilities: school attendance for children aged 6 to 18, and the presentation of the vaccination record of children aged 0 to 5 years old. 41 Inclusion errors refer to the beneficiaries that are not poor (or extreme poor) but participate in the program. On the other hand, 72 exclusion errors refer to the poor (or extreme poor) people that do not participate in the program. 42 Behavioral responses have not been considered in these micro-simulations as the beneficiaries are not observed directly. the CCT programs. The fourth section presents micro- are responsible for the administration of a wider set of simulations of the effect on wellfare indicators of seven small overlapping programs (Bertranou et al., 2003; alternative scenarios of expansion of the CCT and NCP World Bank, 2004). There have been several attempts to programs. The �fth section concludes. establish an institution that plays the coordination role, without much success as of yet. Social Protection in Paraguay The income transfer programs that comprise social Social protection programs43 in Paraguay are security in Paraguay are Retirements and Pensions, relatively new, as compared to neighboring countries, while social assistance is made up of five programs and have historically shown little integration and (Table 5.1).The size of the social security programs is capacity to provide coverage for social risks. Social larger than that of social assistance due to the amount insurance components (health and old age) have never of benefits payed by the former. After the expansion of been able to expand their coverage beyond a limited the CCT programs in 2009, the number of beneficiaries sector of the population. Social assistance went from is almost the same as that for Retirement and Pensions, being charity to having an atomized structure of little but the size of the benefit is between 5 and 8 times programs, with limited success. While social insurance smaller. The average retirement was US$ 437 per month is mostly concentrated in the population living in urban in 2009 and covered nearly 83,000 beneficiaries from areas, social assistance initiatives are concentrated in the 8 coexisting pension funds in Paraguay.44 Pensions the rural sector (i.e., conditional cash transfer programs (including non-contributory) paid out transfers of Tekoporã, PROPAIS II, Ñopotyvô). 60 percent of the average pension to almost 31,000 beneficiaries. The four CCT programs paid an average of The lack of social protection integration in Paraguay about US$ 50. Unlike the social security programs, the is reflected by the coexistence of a large number of majority of the social assistance programs use a targeting institutions that are in charge of delivering similar instrument to select the beneficiares and also require policies. Eight pension funds administer social security the fulfillment of co-responsibilities. Access to social that provides coverage to less than 15 percent of the security transfers is conditional on the beneficiary’s elderly population. A much larger number of institutions contribution history. The notable characteristics of these Table 5.1. Social Protection Programs in Paraguay Amount of monthly Program Type Criterion of eligibility Beneficiaries benefit (in US$) Retirements Social Security Contributive Record 83,000 average 437 /1 Pensions Social Security Beneficiary 31.000 average 246 /1 Noncontributive pens. Social Assistance War Veterans Tekoporã Social Assistance QLI<40 in GPI districts 80.000 range [17; 76] /2 Poverty Assessment Pro-País II Social Assistance QLI<40 in GPI districts 20.000 range [17; 76] /2 Abrazo Social Assistance QLI<40 for candidates 900 range [17; 76] /2 Ñopytyvô Social Assistance Discretional 681 average 21 /1 Data of the 2008 household survey /2 Administrative records, 2009 43 Whenever it is not explicitly mentioned, this chapter considers social protection as the programs that transfer income to participants. 44 The Social Security data in Table 5.1 was taken from the last available household survey, taking into account the difficulties in access to summaries of administrative records. It is possible to identify Pensions and Retirement separately, but a further disaggregation is impossible considering the available microdata. It is also not possible to report a number of retirees by fund, or differentiate between 73 Pensions and Non-Contributive Pensions. requirements are the heterogeneity of the number of There is strong agreement in the literature in the years required and the minimum retirement age as well diagnosis of Paraguay’s social security system as a as the lack of transferability from one pension fund to fragmented and inefficient structure. The system does another of previous contributions. not have an institution to take the coordination and planning role, while large inefficiencies derive from the The recently announced initiatives can be considered lack of portability of contributions between pension as a step forward toward a more integrated social funds (Sanchez, 2003; Betranou et al., 2003; World Bank, protection policy. Scaling up the social assistance 2004). Saldain (2003) highlights that part of the low components seems to be the chosen instrument, as coverage of the pension program is due to the labor observed in the rest of Latin America.45 CCT programs market structure. The pension system is, in practice, (Tekoporã, PRO-PAIS II, Ñopotyvõ) will be uni�ed and designed to provide coverage to salaried workers only, extended to other districts and demographic groups which represent only 40 percent of all workers. (i.e. elderly, indigenous and disabled people). The recent expansion of the non-contributory pensions Large replacement rates represent one of the main program (NCP) has been approved by Congress, despite threats to the financial sustainability of the different the President’s veto. The implementation, targeting pay as you go schemes. Saldain (2003) and Bertranou and budget to �nance this initiative are still pending. et al. (2003) highlight that these rates can reach 208 The following sub-sections describe the components percent, helping to explain why households without of social protection policy in Paraguay with more that income would be poor or extremely poor. The detail, and include descriptions of the projects under World Bank (2004) describes the pension system in evaluation. Paraguay as an expensive one in relation to population capabilities, resulting in an important source of Social Security inequality.49 Two additional factors help to explain the �scal sustainability of some pension funds (like the The development of Social Security46 in Paraguay IPS): a young age structure (Grushka and Altieri, 2003) occurred relatively late as compared with the rest of and a large amount of lost contributions due to lack of the region. The Instituto de Prevision Social (IPS, Social portability, informality, etc (Saldain, 2003). Security institute) dates from 1943. Currently, eight pay- as-you-go pension funds administer the contributory Several projects to reform the pension system have system that provides insurance against old age and been considered recently. Saldain (2003) examines health risks. Two of them (IPS and Caja Fiscal) account two projects in particular: SIPASS (Paraguayan social for nearly 95 percent of all pension bene�ts. Although security system), a proposal presented by members IPS runs surpluses, Caja Fiscal runs large de�cits.47 of the Congress, and a second proposal presented by Part of the �nancial differences are accounted for by the Ministry of Finance. Both initiatives were aimed parametric differences (retirement age, replacement at unifying the great diversity of parameters that rate48, etc), but an important part of Caja Fiscal’s de�cit regulate the difference pension funds, and with the is due to the administration of a non-contributory introduction of de�ned contribution schemes. A recent program for war veterans. proposal to extend “old age� NCP to the elderly poor 45 Some experiences: Progresa-Oportunidades (Mexico), Bolsa Escola- Familia (Brazil), Jefes de Hogar (Argentina), PANES (Uruguay), etc. 46 Social Security represents the main insurance component of the set of Social Protection programs. Most Social Security programs are contributory and thus part of social insurance. Some components of Social Security are non-contributory and thus classified as social Chapter 5 assistance (i.e. old age non-contributory pensions). In the case of Paraguay, Social Security refers both to Pensions and Health programs to which workers make contributions and get insured against such risks. Contributions to both programs are bundled and the overall coverage is relatively low. 47 According to Saldain (2003) this de�cit accounted for 1.4 percent of GDP in 1998. 48 Replacement rate can be considered a relative measure of the generosity of the pensions program and refers to the percentage of the contributed past wages “replaced� by the pension once the individual retires from the labor market 49 Pensions systems running deficits normally use general revenues to cover the financial gap, which transfers the resources of the whole 74 society to a selected group of them. If these revenues are collected through consumption taxes and the perception of the pension benefit places the pensioner in the right side of the income distribution, this type of redistribution becomes regressive. has been approved while a second proposal to extend vulnerable groups (World Bank, 2004). In 1996, the NCP to “housewives� is still to be considered by the Secretary of Social Affairs (SAS, acronym in Spanish) was Congress. Although, the approved NCP expansion created, another initiative to channel the State’s effort. states that it would focus on the poor population, little Until 2002, the SAS was responsible for administering implementation details are known. the PROPAIS project, �nanced by the IADB.50 For six years, PROPAIS �nanced a yearly average of 66 projects, Social assistance spending US$3.8 million (US$57,000 per project, on average). The projects administered by the SAS were Social assistance (SA) in Paraguay has been externally managed with limited or no audit control, progressively increasing its fiscal space. Negligible and discretionary selection of bene�ciaries (World in the 1980s, the latest available data shows that SA Bank, 2004). represented 1.7 percent of GDP, equivalent to 15.9 percent of all social spending. Table 5.2 shows that while In the year 2000, the SAS formalized an initiative called social spending increased in the 1990s and stabilized ENREPD (National Strategy to Reduce Poverty and later on, SA grew at a steady pace over the last three Inequality) and turned more into a vehicle for the debate decades, accelerating in 2009. around social protection rather than a normative project (proposals derived from this initiative progressed slowly The development of SA went through two stages: or never became effective). This initiative called for the first, moving from charity to State actions; and second, creation of a social protection network (RPS, acronym collapsing the small and overlapping programs as in Spanish) to play the coordination role among the well as the large number of unconnected institutions different institutions participating in the delivery of delivering social policy. Organizing the charity efforts social policy. One of the items outlined in the proposal within the State umbrella proved to be a difficult task that became reality was the introduction of a CCT since the late 1980s. In 1989, the creation of program. The role of the Social Cabinet (GS, acronym DIBEN (Directorate for Charity and Social Aid) was the in Spanish) was essential for the implementation of the �rst attempt to provide an institutional framework to program in 2005. The following year, another initiative strengthen Non-Governmental Organizations (NGOs), was outlined: ENLP (National Strategy to Combat provide a response to emergencies and natural poverty). As of 2008, SAS, GS and DIPLANP (Directorate disasters, and grant subsidies to the sick, indigenous or for the Plan to Combat Poverty) were in charge of the Table 5.2. Public spending in Paraguay 1980-2009. % of GDP Year (1) Social Assistance (2) Social Spending (3) Public Spending (1)/(2) (1)/(3) (2)/(3) 1980-1989 0.01 3.38 8.95 0.3% 0.1% 37.8% 1990-1994 0.07 8.55 14.31 0.8% 0.5% 59.7% Poverty Assessment 1995-2000 0.22 8.55 19.95 2.6% 1.1% 42.8% 2003-2007 0.36 8.25 19.74 4.4% 1.8% 41.8% 2007 0.55 8.72 19.55 6.0% 3.0% 47.0% 2008 0.83 8.4 16.28 9.9% 5.1% 51.6% 2009 1.67 10.47 20.57 15.9% 8.1% 50.9% Note: Columns (1) to (3) are expressed in % points of GDP. The others express quotients between (1), (2) and (3).The information in columns (1), (2) and (3) comes from the National Budget. Source: Flood, 2002 and Blanco, 2008, Central Bank, 2010. 75 50 PROPAIS stands for Paraguayan Social Investment Project. IADB is the Inter-American Development Bank. elaboration, implementation, coordination, supervision, A repeated aspect found in the literature on Paraguay and targeting of the multi-sectoral policies �nanced by is the difficulties for the beneficiaries to access services the FES (Social Equity Fund) and other components of (and thus the corresponding certi�cation) required by the ENLP (Blanco, 2008). the conditionality of the program (Soares et al., 2008; GTZ, 2008). Since participation in the program can be Like in the Social Security case, the absence of an considered as a subsidy to demand education and institution that coordinates the social assistance health services, the supply side needs to be available efforts became evident in the co-existence of several or improve throughout the expansion of the program. institutions with overlapping demographic groups. Otherwise, bene�ciaries would be penalized for the poor San Martino and Capellari (2003) identi�ed a different infrastructure of the area they live in (surely correlated set of institutions responsible for assisting the poor to the indicator used for the geographical targeting). If (15), women (4), children and youth (6), indigenous not, the program avoids demanding the compliance of groups (15), the disabled (3), emergency and natural the conditionality, turning the program into a targeted disaster (3), etc. The efforts to gather information has cash transfer with a limited intergenerational impact on led several authors to criticize the lack of transparency poverty. Once again, the lack of an institution able to in the selection of bene�ciaries, the absence of coordinate social protection policy becomes evident. monitoring and evaluation actions to help guide the decision-making process and the limited role Considered the flagship of the current administration, of auditing and control activities (World Bank, 2004; the set of CCT programs experienced one of the largest Bertranou et al., 2003). expansions both in terms of budget and number of beneficiaries, according to the limited information Cash transfers available. As mentioned above, four programs compose the set of CCTs available in Paraguay: Tekopora, PROPAIS In September 2005, the first conditional cash II, Nopotyvo and Abrazo. The �rst three share several transfer program (Tekoporã) was launched in a pilot characteristics, such as being administered by the stage (in �ve rural districts), later on complemented same institution: the SAS. The latter is administered by by the IADB-financed PROPAIS II (Table 5.3). A third, the Secretary of Youth (Secretaria Nacional de la Niñez and much smaller program, Ñopytyvô, differs from y la Adolescencia, SNNA) and is aimed at reducing the others by targeting the indigenous.51 By early child labor. The expansion experienced in 2009 was 2009, the Tekoporã program had been geographically accompanied by the uni�cation of the operations expanded to reach 19 districts with approximately manual of the three CCTs administered by SAS, as well 18,000 participants. The expansion of PROPAIS as the increase in the amount paid to bene�ciaries. The II was much slower given, among other reasons, �xed component of the bene�t jumped from G$60,000 the requirement for effective compliance with the to G$80,000 and the variable component from G$30,000 program’s participation conditionalities. In fact, co- to G$35,000. Originally destined to take into account responsabilities were not monitored for the �rst year the size of the family, the variable component was paid and half of Tekoporã’s implementation due to lack of per child below 15 years old up to four children living institutional coordination (Soares and Britto, 2007). in the household. The age limit was raised to 18 years old. Finally, an extra variable component was paid per Soares et al. (2008) performed an impact evaluation of elderly or disabled person in the household, up to two. the Tekoporã program, finding that the program had a positive effect on school attendance, consumption, Tekoporã is the oldest CCT program in Paraguay Chapter 5 and investment (in production for self-consumption). as well as the largest. Approximately 80 percent Among the negative impacts they report a reduction in of the 100,000 bene�ciary households are enrolled the male labor supply. in Tekoporã. Table 5.4 shows that the rest of the 76 51 The program Abrazo can also be accounted in the set of CCT programs. This program targets street children in urban areas, and is administered by the Secretary of Infancy. bene�ciaries belong to the IADB �nanced PRO-PAIS II the per capita income distribution. Similarly, the lower program. A negligible proportion of all bene�ciaries half of this distribution receives only 20 percent of total are being paid by the Ñopotyvô, exclusively oriented spending of the survival pensions program. The receipt to indigenous populations. The set of CCTs is paid in 67 of a survival pension53 allows a bene�ciary to be placed districts of Paraguay and distributed in 12 departments, beyond the poorest 20 percent of the distribution. as reported in Tables 5.3 and 5.4. Half of the bene�ciaries are concentrated in the rural districts of Concepción In contrast, 80 percent of the beneficiaries of CCT and San Pedro, in which the CCT programs are present programs are concentrated within the poorest 15 in 10 and 20 districts, respectively. percent of the population. No bene�ciary is to be found in the upper half of the income distribution. Incidence Analysis Evidently, the amount of the bene�t does not place bene�ciaries out of the poorest half of the per capita Paraguay’s CCT programs seem to be much better income distribution. Several characteristics of these cash targeted at the poor and extreme poor than the transfer programs need to be highlighted to explain other cash transfer programs (pensions and survival the contrast. First, given that Figure 5.1 presents after- pensions).52 According to the incidence curves reported transfer distributions and that the amount of the bene�t in Figure 5.1 more than 80 percent of spending on is much higher for pensions and survival pensions than pensions (after transfers) would be focused on the top for CCTs, the former shows incidence curves to the right quartile of the per capita income distribution. The lowest of the distribution, while the opposite occurs with the half of the distribution receives almost no transfer from latter. Pensions, and up to a point survival pensions, are the pensions program. Another interpretation of the a reflection of the income distribution position of these same graph is that the receipt of a pension allows a bene�ciaries in the past, when they participated in the household to place themselves in the upper half of labor market. The fact that these people got access to a Table 5.3. Evolution of CCT programs in Paraguay. Number of beneficiaries, population and districts covered. Date Districts Beneficiaries Population Source 2005 5 3,452 n.a UNFPA/GTZ (2008) Aug. 2006 5 4,324 n.a Soares et al (2008) 2006 12 5,386 n.a UNFPA/GTZ (2008) 5 8,162 41,338 Franco and Medina (2008) Dec. 2006 13 8,792 53,542 GTZ (2008) 17 14,014 70,000 GTZ (2008) Dec. 2007 19 17,000 103,000 Blanco (2008) Poverty Assessment Aug. 2009 n.a 45,017 n.a UNDP (2009) Nov. 2009 n.a 64,000 320,000 UNDP/GTZ/UNFPA (2009) Feb.2010 67 100,339 570,169 UES (2010) 52 Survival pensions are the (contributory) pension to which the surviving member of the household is entitled after the original pensioner dies. In the original design of the contributory schemes the survival pensions was thought to smooth the consumption of the widows who present larger life expectancy than men. The progressive inclusion of women to the labor market has modified this traditional structure. Nevertheless, the majority of survival pensioners are women, reflecting that labor and social structure of the past. 53 Unfortunately, using the household level micro-data it is not possible to distinguish the survival pensions derived from a contributory 77 pension from the non-contributory pensions. Table 5.4. Geographical distribution of beneficiaries CCT programs in Paraguay. Number of beneficiaries per districts and program as of February 2010. Tekoporã Ñopytyvô PROPAIS Total Concepcion 11,304 5,805 17,109 San Pedro 25,387 9,064 34,451 Ñeembucu 884 517 1,401 Canindeyu 11,138 495 11,633 Caaguazú 8,647 2,427 11,074 Caazapa 9,532 9,532 Amambay 1,046 1,046 Cordillera 804 804 Guairá 3,374 3,374 Capital 4,135 4,135 Alto Paraguay 690 694 1,384 Central 4,396 4,396 Total 80,291 694 19,354 100,339 Source: Unidad de Economia Social. Ministry of Hacienda (2001) Box 5.1 Targeting methods of CCT programs in Paraguay The design of the Tekoporã program uses two targeting instruments: a geographical priorization index (IPGEX, accord- ing to the Spanish abbreviation); and the life quality index (ICV, according to the Spanish abbreviation). The former targets districts and the latter identifies and filters applicant households within the pre- selected municipalities. The IPG ranked districts according to a combination of poverty measures (income and structural measures), while the ICV used a broader set of welfare measures at the household level. The IPGEX equally weights four quotients: • I is the population on extreme poverty • P is the population • S is the population on structural poverty (a household with 2 or more unsatis�ed basic needs) • sub-index d identi�es each one of the 223 districts The ICV is scoring index that can take values along the range [0, 100], which combines 18 variables grouped in 7 categories: 1. Children: • # of children (0-5) in the household 2. Health: (a) % of household members ill over the last 3 months with access to doctor; (b) % of household members insured by any health service; (c) # of children (0-5) with vaccination certi�cate 3. Education: (a) # of years of formal education received by the head and the spouse, (b) % of human capital lost by the Chapter 5 children (6-24); (c) main language spoken in the household, especially by the head and the spouse 4. Income: (a) occupational category of the head and the spouse 5. Housing: (a) overcrowding, de�ned as # of members over # of rooms; (b) predominant material on floor, exterior walls and roof of the house; (c) existence of bathroom and kitchen; (d) sanitation 6. Basic Services: (a) main water source and provision; (b) access to electricity; (c) access to phone services (fix line or mobile phone); (d) garbage treatment; (e) type of fuel used for cooking 78 7. Durable goods: (a) available fridge, AC, car, lorry, truck, motorcycle, laundry machine, or boiler contributory bene�t may be reflecting that they were households with evident signs of need. According to placed to the right of the income distribution in the Perez Ribas, Hirata and Soares (2008), in an attempt past, allowing them to smooth consumption (or income to introduce a different targeting mechanism for the at least) and retain their position in the ranking. program PRO-PAIS II, the IADB prepared an alternative proxy-means test. The �ndings of this paper was that Alternatively, if the methodology applied to identify not only the use of both multidimensional and proxy CCT beneficiaries resembles its true distribution, the means test indexes provided similar results in terms implementation of both a geographical index and of leakages and coverage, but also discovered that a multi-dimensional index like the ICV would allow the ICV performed more efficiently at lower cut-off these programs to transfer income to the most needy, values. Finally, the scaling up of the program during probably victims of the inter-generational vicious cycle 2009 implied the expansion of the CCT program (and of poverty. A third factor is that the CCT programs are its targeting mechanism) into urban areas (especially targeted at rural areas (with relative lower incomes the “Bañados� area in the outskirts of Asunción). and earnings) and that pensions are mostly found in Originally designed for rural areas, the evidence urban areas, particularly in Asunción, setting the upper showed (see Table 11 in page 29) that the density of and lower limits of the leveling out of the income potential bene�ciaries in Asunción is much lower, and distribution of the bene�ciaries of each program. may become negligible if 40 is used as the cut-off value. Unfortunately, a detailed methodology of the The use of targeting instruments is key for the adaptation of the targeting instrument was not made distributional contrast between these cash transfer available yet, although it was clear that raising the cut- programs. Nevertheless, the use of these instruments off value was the most convenient solution. The latter (geographical and multi-dimensional indexes) presents reflects the need to design idiosyncratic instruments to advantages but faces some disadvantages as well. select the target population of each program. In particular the disadvantages appear when its parameters are not updated or the instrument is over- Targeting instrument used for purposes it was not designed for. Originally set to target only household scoring less than 25 in the ICV Beneficiaries of the CCT programs are targeted, index, the people in charge of the program decided to first, geographically through the IPG, and second, raise the cut-off point to 40 in light of several complaints using the multidimensional instrument ICV. The presented to the local council given the exclusion of IPG (a weighted average of both poverty measures; Figure 5.1: Incidence curves. Pensions, survival pensions and conditional cash transfers. 2008 Pensions Survival Pensions Tekopora (CCT, 2008) Poverty Assessment 1.00 0.80 Cumulative F(x) 0.60 0.40 0.20 0.0 15 50 75 Centiles of per capita income 79 Source: World Bank sta calculations based on EPH, Paraguay. structural and conjunctural) ranked geographical units proxying poverty but much lower errors when aiming and placed rural districts at the top of the priorities. at extreme poverty. The inclusion errors get reduced The ICV provides a proxy for poverty without asking to 2 percent both using the extreme and total poverty for the person’s income. The household characteristics line. Exclusion errors, on the contrary, increase up to 27 plus a series of goods-deprivation measures gives ICV percent and 12 percent for poverty and extreme poverty, the flavour of a structural poverty indicator. Combining respectively. Finally, it is important to clarify that the the ICV dichotomy (eligible, non-eligible) with poverty alternative limit values are not neutral to budget, and line measures provides a proxy for the inclusion and that increasing it implies also increasing coverage with exclusion errors implicit in the use of ICV to target poor the corresponding budget expansion.54 bene�ciaries. Trading off inclusion and exclusion errors is not a The ICV, at a cut-off value of 40, seems to be a better linear function, suggesting the existence of a cut-off predictor of extreme poverty than total poverty. value that minimizes both errors. Figure 5.2 provides a The exclusion error is larger than the inclusion error graphical representation of both the trade-off and the using the poverty line, but the opposite occurs when existence of such minimum in the overall error function. considering the extreme poverty line. At a cut-off value At ICV=10, 3 out of 10 bene�ciary households would of ICV=40, such as the ones currently used by the CCT be wrongly targeted and there would be no inclusion programs, the inclusion error (non-poor and eligible) errors. As the ICV moves the cut-off value, the coverage would be 12 percent, and the exclusion error (poor and increases as well as the inclusion errors together with non-eligible) 16 percent. The inclusion and exclusion a decrease in the exclusion errors. At a cut-off value errors would be 17 percent and 5 percent, respectively, of 30, the inclusion errors are still negligible and the for the extreme poor, as reported in Table 5.5. exclusion errors explain most of the overall targeting errors. If an ICV value of 50 is chosen, most of the overall Increasing the cut-off value of the ICV from 25 to 40 errors would be explained by inclusion errors. At the increased the program’s coverage holding constant extreme right side, using an ICV cut-off value equal to the overall targeting errors for poverty. Considering 90, 70 percent of the bene�ciary households would be the targeting errors with respect to extreme poverty, wrongly targeted and the other 30 percent would be increasing the ICV cut-off value raised the overall errors, represented by the qualifying poor. A cut-off value of 70 in particular the inclusion ones. A lower cut-off value would provide the same percentage of qualifying poor of the ICV would provide the same overall errors in than the ICV=90 �gure. Table 5.5. Targeting accuracy of ICV with respect to poverty and extreme poverty measures (2008) Poverty Extreme poverty ICV Poor Non-Poor Total Ext. Poor Non-EP Total Qualifying 0.15 0.12 0.27 0.09 0.17 0.27 <40 Non- Qualif. 0.16 0.57 0.73 0.05 0.68 0.73 Total 0.31 0.69 1.00 0.15 0.85 1.00 Qualifying 0.04 0.02 0.05 0.03 0.02 0.05 Chapter 5 <25 Non- Qualif. 0.27 0.67 0.95 0.12 0.83 0.95 Total 0.31 0.69 1.00 0.15 0.85 1.00 Source: World Bank staff calculations based on EPH, Paraguay 80 54 In order to have an orderly presentation, coverage data and budget implications are not presented here. However, increasing the ICV cut-off value increases coverage and cost in a monotonous but not linear way, depending on the distribution of eligible households (see Figure 5.11 in the Annex). Figure 5.2: Targeting accuracy with respect to poverty measures for alternative cut-off values of ICV. (2008) Non-qualifying poor Qualifying poor Non-qualifying non-poor Qualifying non-poor 1.00 0.80 % of households 0.60 0.40 0.20 0.0 10 20 30 40 50 60 70 80 90 ICV- Targeting Instrument (Life Quality Index) Source: World Bank sta calculations based on EPH, Paraguay. To find the ICV cut-off value that minimizes the low density at smaller values of the ICV. targeting errors it is necessary to take into account the geographical area where the program could If the same ICV is going to be used in Asunción and be extended. Although the composition of the four the rural areas, a much higher cut-off value would be quadrants proposed in Table 5.5 varies in a similar way needed to minimize targeting errors in urban areas. for extreme and for total poverty, the cut-off value of The advantageous aspect is that targeting errors the ICV that minimizes overall errors is different for each would be much lower in rural areas. The fact that the poverty measure. Figure 5.3 shows that setting the ICV mode of the distribution in rural areas is shifted to the equal to 34 would minimize the overall targeting errors left of the ICV range is reflected in the lower cut-off of ICV with respect to poverty at an overall targeting values that are necessary to minimize the targeting error level of 27 percent. Similarly, the sum of inclusion errors. Nevertheless, these lower cut-off values imply and exclusion errors with respect to extreme poverty much higher levels of overall targeting errors. While would equal 14 percent if the cut-off value is placed at extreme poverty targeting errors in the rural rest of a cut-off value of 24. Paraguay are minimized at an ICV cut-off value of 23, total poverty does so at a level of 29. The associated To get a real dimension of the welfare gains of levels of exclusion and inclusion errors are 22 percent the CCT expansion, the analysis needs to restrict and 31 percent, respectively. These figures are 34 and itself to the rural areas where it was launched. The 52, respectively for Asunción. The overall targeting heterogeneous levels of poverty experienced in the errors would be 5 percent for extreme poverty and 15 different geographical areas of Paraguay and the rural- percent for total poverty. biased design of the ICV provide contrasting levels of the cut-off value that minimizes the overall targeting Simulation Scenarios Poverty Assessment errors for Asunción and for the rest of rural Paraguay. Figure 5.4 shows a much lower ICV value (lower Seven scenarios have been estimated to assess coverage and cost) that would improve the programs the impact of the CCT expansion and the potential targeting capacity in rural areas such as the ones where implementation of an old age non-contributory the program has been launched. On the contrary, as pensions program. These simulation exercises have shown in Figure 5.11 and experienced during the 2009 been made using the 2008 household survey, which scale-up of the program, the required value to improve unfortunately does not allow identification of current targeting in Asuncion needs to be much higher, not bene�ciaries of the CCT programs. Nevertheless, at 81 necessarily increasing the cost of the program, given its the time the survey collected the data in the �eld, Figure 5.3: Exclusion and inclusion errors of CCTs with respect to poverty and extreme poverty measures. Alternative cut-off values of ICV (2008) Exclusion and inclusion errors (wrt extreme poverty) Exclusion and inclusion errors (wrt poverty) 1.00 % of households 0.27 0.14 0.0 10 20 24 30 34 40 50 60 70 80 ICV- Targeting Instrument (Life Quality Index) Figure 5.4: Exclusion and inclusion errors of CCTs with respect to poverty and extreme poverty measures. Alternative cut-off values of ICV. Asuncion and rural rest, 2008 Asuncion, poverty Asuncion, extreme poverty Rural rest, poverty Rural rest, extreme poverty 1.00 % of households 0.31 0.22 0.15 0.05 0.00 10 23 29 34 52 80 ICV- Targeting Instrument (Life Quality Index) Source: World Bank sta calculations based on EPH, Paraguay. the program was still small, covering 18 thousand program’s income to the identi�ed bene�ciaries and households.55 In both cases, CCTs and NCPs, most of the allocates a similar bene�t to qualifying bene�ciaries bene�ciaries are yet to be identi�ed, which provides an living in the districts selected by the program for the additional rationale for the simulation of the potential geographical expansion. The third scenario introduces, impact of such expansions. on top of the previous one, three parametric changes to the variable component of the CCT program The set of seven scenarios are complemented by a bene�t. First, the age of the children is increased from baseline scenario which allows for the quanti�cation 15 to 18 years old (for up to four children, as before). of the potential impact. The �rst simulated scenario Second, an extra G$35.000 is paid if the household identi�es the existing bene�ciaries of the CCT programs has an elderly person (65 years old and above). Third, Chapter 5 and takes away the amount of the bene�t available from another G$35.000 is paid per disabled person in the the household income. The second scenario assigns the household.56 55 Even if the question were to be included in the 2008 round of the survey, the frequency of CCT bene�ciaries would be too low, or its standard errors too wide, especially taking into account the sample survey design. 82 56 Each parametric modi�cations has been estimated separately in a cumulative manner, but the results are presented jointly as the incremental effect is marginal or negligible. The fourth scenario simulated the provision of a non- The only indicator that evidenced a slight change when contributory bene�t (G$ 350,000) to elderly members in the bene�t of (qualifying) bene�ciaries was removed the household eligible to receive a conditional transfer from the household income is the extreme poverty gap. using the ICV as the targeting instrument, independently Figure 5.5 presents, in a radar format, the (normalized of the bene�t provided by the CCT program. In other to 1) welfare differences between scenarios 1 (without words, households in the districts covered by the CCT CCTs), 2 (Geographical expansion) and 3 (Full CCT expansion are allocated an NCP but no payment of CCT expansion). Scenario 0 is implicitly represented by the bene�t is being simulated under this scenario. line connecting the level 1 (nummeraire) of each axis. No difference is observed between scenarios 0 and 1. The �fth scenario combines the implementation of both CCT and NCP programs, but provides a pension instead The limited impact of the CCT programs until 2008 of a variable component of the CCT to elderly members was basically explained by the size of the resources in the qualifying households of the expected-to-expand involved. Once the pilot experience was scaled up, the areas. The sixth and seventh scenarios simulate an NCP impact (potential) became noticeable. The extreme for extreme poverty and total poverty, respectively, but poverty gap is the indicator that experiences the largest do not restrict its payment to the areas where the CCT variation, in relative terms. The difference between the programs have expanded. average income of the extreme poor and the extreme poverty line got shortened by 10 percent, from 6.35 The results of the simulations show that the immediate to 5.75. The poverty gap got reduced as well, less in effect of the income support part of the program’s bene�t relative terms but with the same absolute values (0.6 is much more sensitive to extreme poverty than to percentage points). While income inequality remained aggregate poverty (Table 5.6). It also shows the poverty relatively unchanged, the extreme poverty headcount and extreme poverty gaps are much more sensitive ratio decreased, in relative terms, more than the poverty than the headcount ratios in capturing the effect of headcount ratio. The aggregate cost of the geographical the expansion of the programs. The effect of the cash expansion increased by less than 10 percent, in both transfer programs on income inequality is consistent scenarios 1 and 2. This fact reflects that the parametric with previous expectations but it is only marginal on changes introduced jointly with the scaling up of the the �nal effect. The effciency of the CCT programs to program did not represent a noticeable variation influence the Gini coeffcient is explained by the use of a neither in cost nor in the effectiveness to improve targeting mechanism (IPG and ICV) but its effectiveness welfare indicators. Most of the impact is explained is bounded by the size of the resources involved in the by the increase in the number of bene�ciaries due to program. As expected, the geographical expansion, the geographical expansion. increase in the bene�t amount and the introduction of new programs imply larger expenditures. Although The importance of the impacts on the welfare indicators the �gures presented in Table 5.6 may not be reflecting is stronger when only the results in rural areas are the aggregate effective cost of the social protection examined, areas where the program was launched and programs that involves cash transfers in Paraguay, the extended. Figure 5.6 and Table 5.7 show that a slightly �gures provided may be useful to compare the relative larger effect on all welfare indicators is perceived when Poverty Assessment costs and the cost/bene�t relationships implicit in each restricting the analysis to the domain “rural rest�.57 Given scenario (alternative strategy). that restricting the analysis to this domain only ampli�es the visualization of the effects, the qualitative effect is The role of the CCT programs in the reduction of the same as for the whole country. The extreme poverty poverty as of the beginning of 2009 was negligible. gap gets shortened by 11.5 percent, the poverty gap The overall effect of providing a bene�t to 18 thousand by 6 percent; extreme poverty (headcount) decreases households was marginal in terms of the welfare by 5 percent and poverty (headcount) by 2.7 percent. indicators measured here, as well as in terms of its cost. The absolute values of these variations are very similar, 57 The sampling design of the household survey allows disaggregating into �ve domains, with reliable con�dence levels. These domains 83 are Asunción, Urban Central, Rural Central, Urban Rest, Rural Rest. Table 5.6. Simulated scenarios of program expansions. Welfare indicators and associated costs (2008) Moderate Extreme Mod. Poverty Ext. Poverty Income Annual Cost Scenario Poverty poverty Gap Gap Inequality (US$ m.)/1 0. Baseline 37.04 18.97 14.31 6.28 0.5269 588 1. Without CCT 37.94 19.05 14.36 6.35 0.5271 586 2. CCT (Geog.Exp.) 37.38 18.37 13.82 5.75 0.5249 620 3.Full CCT exp. 37.35 18.28 13.76 5.7 0.5246 625 4. NCP using ICV 37.74 18.58 14.09 6.11 0.5251 615 5. CCT+NCP (ICV) 37.21 18.15 13.65 5.64 0.5239 638 6. NCP (ext. poor) 37.57 17.31 13.60 5.69 0.5229 628 7. NCP (poor) 35.77 17.31 13.20 5.69 0.5193 675 Source: World Bank staff calculations based on EPH, Paraguay. Note 1/ The data in this column correspond purely to the benefits, without accounting for administrative costs. The information presented corresponds to the resulting expanded estimations of the household survey data base. It can differ from administrative records. the country’s average. In other words, the percentage of expenditure of the cash transfer programs in the Figure 5.5: Potential welfare and cost implications of expanding Conditional “Rural Rest� region of Paraguay is much lower than for Cash Transfers programs in Paraguay, 2008. the rest of the country. Thus, a targeted increase in the Geographical expansion CCT coverage implies a signi�cant change given the low and bene�t increases. levels of pension coverage. Without CCTs Geog. Expansion of CCTs Full CCT The impact of the simulated scenarios on income inequality is minor, regardless of the indicator used. Poverty The tables and figures presented in this paper show the impact of the alternative scenarios on income Annual cost in Extreme inequality represented by the most used indicator, the m. of US$ Poverty Gini coefficient. Figure 5.7 presents alternative values for improvements in income inequality at the most important levels of the aversion to inequality, using the Atkinson inequality index. Although the level and the Income Poverty effect are greater at higher levels of epsilon (aversion Inequality Gap to inequality), the relative ordering of these effects is the same as that observed in the Gini coefficient. Evidence also suggests that increasing the number of Extreme Poverty Gap beneficiaries has had a greater effect on the reduction of income inequality than expanding the benefits. Source: World Bank sta calculations based on EPH, Paraguay. Chapter 5 Providing an NCP benefit to the elderly living in the in a range of [-1.3,-1.7] percentage points. Income households that are already participating in the inequality improves by 1 percent, which is twice the CCT programs will result in greater costs without effect observed for all of Paraguay. Finally, the annual improving the welfare impact of the programs. Figure 84 cost observes a relatively large increase due to the fact 5.8 shows that the combined result of paying bene�ts that the original baseline is much lower compared with of the CCT program and the NCP does not produce Table 5.7. Simulated scenarios of program expansions. Welfare indicators and associated costs. Paraguay, Rural rest, 2008. Moderate Extreme Mod. Poverty Ext. Poverty Income Annual Cost Scenario Poverty poverty Gap Gap Inequality (US$ m.)/1 0. Baseline 50.6 32.93 21.63 11.82 0.5992 87 1. Without CCT 50.62 33.13 21.76 11.99 0.5997 85 2. CCT (Geog.Exp.) 49.32 31.41 20.45 10.57 0.5942 114 3.Full CCT exp. 49.22 31.25 20.31 10.43 0.5934 118 4. NCP using ICV 50.17 32.15 21.15 11.44 0.5961 110 5. CCT+NCP (ICV) 48.87 30.97 20.05 10.29 0.592 130 6. NCP (ext. poor) 50.03 30.36 20.39 10.77 0.5928 109 7. NCP (poor) 48.11 30.36 19.97 10.77 0.589 126 Source: World Bank staff calculations based on EPH, Paraguay. Note 1/ The data in this column correspond purely to the benefits, without accounting for administrative costs. The information presented corresponds to the resulting expanded estimations of the household survey data base. It can differ from administrative records. different results than the ones achieved only with the CCT program. Under scenario 5, participant households Figure 5.6: Potential welfare receive the CCT bene�t plus the difference between and cost implications of the NCP bene�t and the variable component for the expanding CCT in Paraguay. Rural, 2008. elderly, previously perceived under the scenario 3. However, the cost of the joint program increases. While the CCT expansion represents an increase of 6 percent Without CCTs Geog. Expansion of CCTs Full CCT with respect to the current expenditure on social protection, the combined effect is an increase of 8.5 Cost Poverty percetn, an extra US$ 13m per year. Finally, replacing Scenario 2: 1.31 1.1 the CCT programs with an NCP one, would represent Scenario 3: 1.36 1.0 an intermediate option in terms of cost, with almost no Annual cost in Extreme m. of US$ 0.9 Poverty impact in terms of welfare indicators. 0.8 Considering alternative targeting strategies for 0.7 the implementation of the NCP program may yield much better welfare results mobilizing a Income Poverty similar mass of resources. If, instead of targeting the Inequality Gap participants of the NCP through ICV (in the prioritized Poverty Assessment districts), an alternative mechanism is used so as to Extreme target only the extreme poor (all over the country), Poverty Gap the cost would be similar and the welfare better improved. Targeting all the elderly poor in Paraguay Source: World Bank sta calculations based on EPH, Paraguay. may result in a more expensive program that may not provide similar results. Figure 5.9 presents the results the elderly poor and extreme poor, respectively. The of using alternative strategies to implement the results of these two scenarios are the same in terms of NCP program. While the results of scenario 5 are the the headcount ratio and the gap of extreme poverty, same as the ones presented in Figure 5.8, scenarios with a slight improvement in these indicators for 85 7 and 6 show the potential impacts of targeting all poverty in general. The result of the three scenarios Figure 5.7: The Atkinson Index Figure 5.8: Potential welfare and for simulated scenarios, cost implications of expanding the CCT different epsilon values. Paraguay 2008 and Non-Contributory Pensions in Paraguay, 2008 1. Without CCTs 2. Geog. Expansion of CCTs 3. Full CCT expansion 4. NCP using ICV 5. NCP + CCT Full CCT expansion NCP using ICV CCT + NCT (ICV) 6. NCP (indigenous) 7. NCP (poor) 1.1 Poverty 0.65 1.10 1.9 0.6 1.2 1.00 0.55 Annual cost in Extreme 0.5 m. of US$ 0.90 Poverty 0.45 0.80 1.8 1.3 0.4 0.70 Income Poverty 1.7 1.4 Inequality Gap 1.6 1.5 Extreme Poverty Gap Source: World Bank sta calculations based on EPH, Paraguay. in terms of income inequality is the same. There is a when compared to the expanded CCT targeting marked difference between targeting all the elderly (scenario 3) or the combined provision of CCT and poor (scenario 7) and the other options presented in NCP (scenario 5). Under scenario 3 (expanded CCT) 60 Figure 5.9. Targeting the elderly with the ICV (only percent of the expenditure is perceived by the poorest in the prioritized districts) would increase the social 15 percent of the population, and 90 percent by the protection expenditure by 6.8 percent, while targeting poorest half of the population. Clearly the targeting all the Paraguayan elderly in extreme poverty would of the CCT program before the 2009 expansion was do so by 8.5 percent. Targeting all the elderly poor in better, but the coverage was much lower (18,000 Paraguay would imply an increase in social protection bene�ciaries). As the CCT programs are expanded expenditure of 15 percent. using the combined strategy of the IPG and ICV, the new districts that are progressively incorporated The implementation of an NCP program based only on reduce the targeting of the program. This inevitable the beneficiaries of the CCT programs may result in a trade-off between coverage and targeting is the result poorer targeting strategy than the current expansion of ranking districts according to the average needs of of the CCT programs themselves. Figure 5.10 shows its households. As the programs go down in the district that selecting beneficiaries of the NCP with the ICV ranking the targeting gets worse, but the coverage is would target the expenditure of this program to the expanded. The geographical expansion proved to be Chapter 5 right of the per capita income distribution of Paraguay. more effective in terms of welfare improvements than Under this scenario, 40 percent of the NCP expenditure parametric changes (i.e. new variable components, would be received by the poorest 15 percent of the different ICV cut-off values, etc.). More importantly population and 80 percent by the poorest half of the maybe, the geographical expansion represents an 86 population. Although the latter seems to be a good increase in equity, given that the opportunity to access targeting pattern, these results become challenged the program is expanded. Conclusions Figure 5.9: Potential welfare and cost implications of expanding the NCP Social Protection in Paraguay is characterized in Paraguay, 2008. by a series of atomized programs that have little coordination and are possibly overlapping. The repeated diagnosis in the literature for the pay-as-you- NCP using ICV NCP exp. old age NCP exp. old age go system in Paraguay is that pension funds should extreme poor poor be uni�ed to allow for portability, as well as introduce Poverty parametric changes to improve the sustainability of 1.1 the system and reduce the potential disincentives to 1.0 contribute. Social assistance observes a similar picture. Annual cost in m. of US$ Extreme Three conditional cash transfer programs with similar 0.9 Poverty characteristics are administered in parallel by the 0.8 Secretariat for Social Action (SAS). A fourth conditional 0.7 cash transfer program, run by the Secretariat for the Youth, is aimed at combating child labor. Several other Income Poverty institutions run small programs with high relative Inequality Gap administration costs and low coverage. The Tekoporã program was launched in 2005 as a pilot Extreme Poverty Gap program and was expected to expand much earlier. After struggling with budgetary limitations, the current Source: World Bank sta calculations based on EPH, Paraguay. administration increased the �scal space to scale up the series of conditional cash transfer programs (CCT) run by SAS. These programs jumped from 18,000 to Targeting NCP through the ICV (to the prioritized 100,000 beneficiaries during 2009. Unfortunately, little districts) would yield worse targeting results than information is available and the only impact evaluation using an alternative method aimed at identifying available dates back to 2006 when the program reached the elderly poor. Although it is not the purpose of this 5,000 bene�ciaries. Using a strategy of identification chapter to present such an alternative method, Figure (simulation) of bene�ciaries, this chapter simulated the 5.11 shows that the use of the ICV is far from being an expansion of the CCT programs based on information of optimal choice. Moreover, targeting NCP bene�ciaries the 2008 household survey. This ex-ante evaluation also through the ICV results in lower coverage and similar simulated the expansion of a non-contributory pension cost than targeting all the extreme elderly poor. Using program. Given that the only information available the same parameters as in Figure 5.10, under scenario with respect to the strategy to identify the beneficiares 6 (NCP to all extreme poor elderly) 95 percent of the of this program (NCP) is that it will be focused on the expenditure of this alternative implementation of elderly poor (older than 65 years old), this chapter the NCP would be focused on the poorest 15 percent offered alternative targeting strategies together with Poverty Assessment of the population. The alternative of implementing its welfare implications and associated costs. scenario 7 (NCP to all poor elderly) would allocate half of its expenditure on the poorest 15 percent of the Subject to the validity and precision of the chosen population, and the total of its expenditure on the methodology, the CCTs would be well targeted in the poorest 40 percent of the population. Targeting NCP bottom part of the per capita income distribution. The through ICV (scenario 5) would spend the full amount exclusion errors (16 percent), relative to the poverty line, of its resources over the �rst three quarters of the per would be larger than the inclusion errors (12 percent), capita income distribution. Probably not too efficient while the opposite occurs when compared with the given that it only covers the prioritized rural districts of extreme poverty line. Raising the cut-off value of the 87 Paraguay (and the �Bañados� of Asunción). ICV score from 25 to 40, increased the coverage (in the Figure 5.10: Incidence curves. Conditional cash transfers (before and after expansion), Non-contributory pensions (potential exp. using ICV). Paraguay, 2008 CCT + NCP (s.e.) Non-contributory Pensions (s.e.) Tekopora (s.e.) Tekopora (CCT, 2008) 1.00 Cumulative density F(x) 0.80 0.60 0.40 0.20 0.0 15 50 75 Centiles of per capita income Source: World Bank sta calculations based on EPH, Paraguay. Figure 5.11: Incidence curves. Non-contributory Pensions (potential exp. Using ICV and poverty measures). Paraguay, 2008 Non-contributory Pensions (s.e.) NCP (s.e. ext. poor) NCP (s.e. poor) 1.00 Cumulative density F(x) 0.80 0.60 0.40 0.20 0.0 15 50 75 Centiles of per capita income Source: World Bank sta calculations based on EPH, Paraguay. districts prioritized by the program), without modifying than what is currently used. The opposite happens the overall targeting errors (with respect to the poverty when restricting the analysis to the city of Asunción. line). The analysis presented suggests that, for the whole If the same targeting instrument is used (taking into country, the cut-off value that minimizes the targeting account the ICV parameters that correspond to the errors would be lower than the one currently used, urban sector), the results show that this value should ICV=40.58 Nevertheless, the geographical heterogeneity be higher than the one uniformly used for Paraguay. of Paraguay needs to be taken into account given that the program uses a geographical prioritization scheme The targeting of the program will not improve by as part of the targeting strategy. In the rural areas where increasing the cut-off value of the ICV, especially in Chapter 5 the program was originally launched and where the rural areas. This increase would not only increase the cost majority of its participants live, the cut-off values that but would also magnify the targeting errors. If coverage would minimize the targeting errors are much lower is to be expanded to improve the welfare impacts, 88 58 A cut-off value of 24 would maximize the targeting efficiency with respect to extreme poverty, and a value equal to 34 would do so with respect to poverty. a geographical expansion would prove to be more designed to target the elderly poor but structural poor effective than an increase in the amount of the benefit. rural households. There is an evident overlap between In addition, it should be pointed out that the original the CCT programs (variable component of the bene�t to design of the ICV was developed before 2003 using a elderly people living in the household) and an eventual principal component analysis to create the targeting NCP program, but combining both programs will only instrument for rural areas. As should be expected, the raise expenditure on some households, crowding out expansion of the CCT programs encountered problems the opportunities of others to gain access to social when applying the same instrument in urban areas. protection programs. The solution chosen to resolve this problem during the implementation was to increase the ICV cut-off value, Social Protection in Paraguay needs to gradually exclusively for urban areas. Nevertheless, for further unify its atomized set of programs, but that does not expansions of the program it would be advisable to: a) imply the homogenization of targeting instruments. update the parameters used by the ICV (and probably In order to give a rapid response to the crisis, as well as the variables), and b) customize the instrument to the to ful�ll campaign promises, the current administration area where it would be applied, at least differentiating correctly engaged in an expansion of the assistance urban and rural areas. component of social protection. The need for further improvements in the targeting instruments to be used, The results of the ex-ante simulations suggest that their update and customization to the targeted groups the expansion of the CCT programs, ceteris paribus, become more urgent in the context of an expansion may help to reduce poverty by half a percentage of social assistance programs. If programs are going point and extreme poverty by 0.7 percentage points. to attend different population groups, it would be When measuring the potential impact in rural areas, advisable that speci�c instruments are designed for the results are somewhat larger, given that the CCT that end. The recent initiative tending to reduce the programs were basically implemented in those areas, overlap of programs (especially the ones involving Poverty in rural areas may decrease by 1.4 percentage cash transfers) seem to be moving forward in the right points and extreme poverty by about 1.7 percentage direction to reach a uni�ed social protection system. points. Even though the number of households that The expansion of the CCT programs already provoked may have crossed the poverty line and the extreme the unification of the operations manuals of those poverty line are not that signi�cant, the gap between programs. Nevertheless, the public legitimation of such the average income of both groups with respect to improvements also needs an increase in transparency, the corresponding poverty line gets shortened in a for which the public availability of the monitoring and pronounced way. The poverty and extreme poverty gap, evaluation data of the programs is essential. after the simulation of the geographical expansion and parametric reform of the CCT programs, get shortened by 4 percent and 9 percent, respectively. In rural areas these gaps would decrease by even more, by 6 and 12 percent, respectively. Although the simulated inequality of income improves, the change would be marginal (0.5 Poverty Assessment percent for the whole country and 1 percent for the “Rural Rest�). The income distribution among the poor could improve in a more signi�cant way, especially among the rural poor and extreme poor. Targeting beneficiaries of the NCP program by free- riding the efforts made by the CCT programs in enrolling beneficiaries would yield worse targeting results, low welfare improvements for those households, along 89 with a worse cost/benefit relationship. The ICV was not Chapter 5 Annex– Changes in the life Quality index (iCv) According to a multivariate analysis, each variable of the survey can be broken down. Asunción, skewed participating in the construction of the ICV receives a to the right, polarly contrasts with the rest of the different weighting value. The latter varies de- pending departments in rural areas (Resto rural). The former whether the household is located in a urban or a rural shows a modal value at 80, reflecting low presence of area. Both the weighting factor and the codi�cation households below the threshold value of 40. The Resto of the variable values are de�ned in the methodology rural peaks at 25, the �rst threshold value used to target detailed in the 2005 program’s operational manual. potential bene�ciaries. Table 7 reports the weighting factors used in the ICV. Evidently, the multivariate analysis was conducted on The density distributions for the rest of the areas reflect the whole survey sample, controlling for the urban/ that the ICV methodology assigns a lower �nal value to rural nature of the household. No controls regarding the geographical regions predominately rural. other geographical segmentation was included when producing the weighting structure with which the ICV In fact the rural area of the Central region peaks at 50 was estimated. The latter resulted in a heterogeneous while both urban areas (Central and Resto) are skewed geographical distribution of densities of the ICV across to the right. The “rest� of the urban areas present a less Paraguay. Figure 11 in page 29 presents the kernel concentrated modal value, peaking at around a ICV density estimation of the ICV scores for the �ve main score value of 60. areas or geographical regions into which the results Table 5.8. Weighting factors used in the ICV 2005 Variable Value Urban Rural Variable Value Urban Rural 0 1 2.38 1.89 4.a 1 3.64 2.98 0 2 2.10 2.16 4.a 3 1.35 1.26 0 3 1.37 1.16 4.b.1 3 4.28 3.24 1.a 2 1.57 1.50 4.b.1 4 1.85 2.04 1.a 3 0.52 0.77 4.b.2 3 2.32 0.61 1.a 9 1.61 1.37 4.b.2 4 5.05 2.03 1.b 2 1.79 1.65 4.b.2 6 2.72 3.99 1.b 3 2.74 2.59 4.b.3 2 2.67 3.06 1.c 6 1.11 1.64 4.b.3 3 2.50 2.04 1.c 7 1.84 2.27 4.b.3 4 2.54 2.74 2.a 1 2.89 3.00 4.b.3 5 5.06 5.12 2.a 3 1.32 2.67 4.b.3 6 5.43 4.62 Annex 2.a 5 0.10 0.05 4.b.3 7 6.62 5.12 2.a 6 2.41 1.94 4.b.3 8 6.16 0.67 2.b.1 2 1.05 0.53 4.c.1 1 4.09 2.94 90 2.b.1 3 1.32 0.99 4.c.2 1 2.72 1.28 Table 5.8. Weighting factors used in the ICV 2005 (cont.) Variable Value Urban Rural Variable Value Urban Rural 2.b.1 4 1.48 1.23 4.d 1 5.24 5.74 2.b.1 5 1.90 1.92 4.d 2 3.67 4.76 2.b.1 6 2.69 3.03 4.d 3 1.18 2.33 2.b.1 9 4.01 4.90 4.d 6 1.18 0.90 2.b.1 1 5.15 5.42 5.a.1 1 3.60 2.77 2.b.2 2 0.47 0.77 5.a.1 2 3.57 3.01 2.b.2 3 0.74 0.82 5.a.2 1 4.81 4.49 2.b.2 4 0.87 1.79 5.a.2 2 1.51 0.82 2.b.2 5 1.80 2.52 5.b 1 4.79 2.80 2.b.2 6 2.60 3.04 5.c 2 3.14 5.04 2.b.2 9 4.03 5.67 5.c 3 1.98 2.82 2.b.2 1 5.05 5.67 5.c 4 4.29 5.92 2.b.2 1 1.92 1.42 5.d 1 0.56 1.26 2.c 1 2.51 2.11 5.d 4 2.57 0.96 2.c 2 1.55 0.52 5.d 6 0.16 4.59 2.c 6 1.44 0.99 5.e 6 3.18 3.44 3.a 1 3.27 4.17 5.e 8 0.83 0.60 3.a 3 0.52 1.05 6.a.1 1 3.30 2.79 3.a 5 2.65 2.30 6.a.2 1 3.41 5.46 3.a 6 0.94 1.03 6.a.3 1 2.71 3.28 3.a 9 0.94 0.49 6.a.4 1 3.25 3.02 6.a.5 1 2.70 3.36 Figure 5.12: ICV 2008 by region Rural Rest Urban Central Urban Rest Rural Central Asuncion .04 Poverty Assessment .03 Density .02 .01 0 0 20 40 60 80 100 ICV 91 Changes to ICV 2007 modi�cation of the coding of the occupational category variable. While in 2005, the respondent was allowed to The simulations and results in this document used declare himself either employee or laborer, in the public Paraguay’s household survey (EPH) to estimate ICV or the private sector; in 2007, these alternatives were scores. Soares et al. (2007) estimated ICV scores using reduced from four to two options. The ICV methodology the 2005 edition of EPH and compared it with an assigned a different weight to employees and to alternative score value using a proxy means indicator. laborers, regardless of the sector (public or private). The estimation of the ICV score for 2007 suffered The 2007 EPH survey grouped the four categories some alterations regarding the changes introduced according to the sector, making it impossible to recover to the questionnaire of the survey. When possible, the classi�cation present in the ICV methodology. variables have been recodi�ed to recover comparable To solve this problem, all four categories of the 2005 information. In the cases in which the variable was not survey have been classi�ed into a single one: wage present in the questionnaire, alternative versions of earners (including employees and laborers from both the ICV were estimated to be able to understand the private and public sector). Although the new coding relative importance of the absent variable. of the variable does not allow a differential weight to be assigned to employees and to labourers, it is still The EPH, in 2005, included a special module devoted to possible to provide a different weighting factor to wage gathering information on child health. Unfortunately in earners in the urban and rural areas. The weighting 2007, this module was not included in the questionnaire, value assigned to both groups was re-calculated as the losing the variable that captures the existence of a average between the 2005 ICV weighting values for vaccination certi�cate. Figure 12 compares the ICV employees and laborers. score distribution for the 2005 survey with and without the vaccination variable. Evidently, the absence of this Two more variables showed an important change variable in the algorithm reduces the total value of the between the frequency observed in 2005 and 2008, distribution, shifting the curve to the left. Although without having experienced a modi�cation in the the shift is not exactly homogeneous, the bimodal coding. First, the incidence of mobile phones was distribution is still observed with a larger gap around much larger than in 2005, reducing the percentage an ICV score value of 20. Finally, Fgure 12 shows that of households without phone service. This change in 2007 (without considering the vaccination variable), seems to be pronounced but coherent. Second, the the bimodal distribution is also present, with a much number of people that became ill in 2007 but receive larger peak at around an ICV score value of 60. no medical attention increased signi�cantly (10 percentage points). The main suspect for this increase To estimate the ICV score of 2007, another potentially is thedengue and yellow fever epidemic that occurred important change was introduced, regarding the at the end of 2007. Figure 5.13: Comparison ICV 2008, 2005 with and without child vaccination variable 2005 2007 2008 .04 Household Density Paraguay .03 Annex .02 .01 0 92 0 10 20 30 40 50 60 70 80 90 100 ICV Annex – Crisis, Drought and Poverty in 2009 Prologue by increasing the labor force participation of women and younger workers. Women entered mostly into the The Paraguay Poverty Assessment entitled “Determinants agricultural sector, in particular as autonomous workers and Challenges of Poverty Reduction� provides an or entrepreneurs. Agricultural employment acts as a analysis of the situation of poverty, inequality, income, buffer when incomes decline, but it was not enough and employment in the country between 1997 and to avoid the drop in household incomes, and may have 2008 based on the data available at the time of its put even more pressure on sectoral wages. Incomes writing. However, given the recent completion of the from remittances and from State transfers, even if a 2009 Permanent Household Survey (EPH) – the latest small percentage of family income, are more important household survey available covering the period from for the poorer households. However, they were also October to December 2009 – this brief paper presents not enough to buffer the drop in family income and an update of some sections of said study. With the the increase in rural poverty. Even so, without these availability of the 2009 EPH, now it is possible to incomes it is possible that rural poverty would have study the impact of the 2009 economic downturn on been even higher. poverty. In the urban sector, the negative impact of the crisis on Introduction And Summary the GDP of the largest urban sectors did not translate into reductions in urban employment nor, in some The severe drought at the beginning of 2009, as well as cases, reductions in the hourly wage. However, the the global financial crisis, resulted in a sharp economic quality of urban employment did deteriorate given contraction for Paraguay in 2009. The Gross Domestic the increase in unemployment, informality, and visible Product (GDP) of Paraguay dropped 3.8 per cent in underemployment, which explains the puzzle. Salaries 2009, according to the Central Bank of Paraguay. dropped in the following sectors: transportation & However, poverty and inequality declined between communication, services to enterprises, agriculture, 2008 and 2009 in the country, contrary to what would and electricity & water, but increased for less skilled have been expected. Even so, the effects were very workers in the sectors where the poor concentrate the different between the urban sector and the rural most, such as commerce and construction – a positive sector, with an increase in rural poverty and no change impact for the reduction of urban poverty. In addition, in the inequality of rural incomes, and a drop in urban remittances increased as a percentage of family income poverty with an improvement in urban inequality. This between 2008 and 2009 for the urban households of brief presents these results and attempts to explain the poorer deciles. these puzzles, exploring the changes in the GDP per economic sector, in incomes – including remittances Recession with Less Poverty: a Puzzle in Paraguay Poverty Assessment and transfers –, and in labor market indicators - including employment and unemployment. The paper highlights several stylized facts and makes clear the In comparison with previous crises, the 2008-2009 need for a more detailed analysis. global financial crisis hit the Latin America and Caribbean (LAC) region hard, although the effect The sharp drop in GDP occurred mainly in the was not as important as in other regions of the world agricultural sector, given the drought and the decline or for developed countries. According to a World Bank of international prices of goods caused by the decline in report for the region, the regional GDP experienced a global demand. As a result, incomes in the rural sector 0.6 percent contraction, the highest annual decline 93 dropped and households attempted to compensate for Latin America during the last two decades (Figure A2.1).59 If one considers the countries of the region agricultural sector60 is the most important economic separately, Paraguay’s 3.8 per cent GDP contraction in sector in terms of the GDP of Paraguay, followed by the 2009 was one of the highest in the regions. commerce sector. Together these sectors accounted for over 42 percent of the country’s GDP in 2009 (Figure 2), In Paraguay the financial crisis and the drop in over 50 per cent of employment and over 70 per cent international prices, combined with the severe of the employment of the poor. Between 2008 and drought at the beginning of 2009, had a substantial 2009, agriculture and commerce experienced major negative impact on the Gross Domestic Product declines in GDP (in value added): 17.4 and 3.4 percent, (GDP) of the agricultural sector, and a slight negative respectively (Figure A2.2). In addition, Figure A2.3 shows impact on the GDP of the commerce sector. The a drop in the international prices of goods between 2008 and 2009. The prices that declined included cotton and corn, crops in which the poor and extreme poor work. Given the importance of the agricultural sector Figure A.1: Strong impact of the 2009 global crisis on Latin America in the rural economy, in 2009 the negative impact was and the Caribbean sharper in the rural area than in the urban area. 10 Poverty Growth rate in GDP per capita (Us$ PPP) 8 6 Although the financial crisis and the severe drought 4 had a negative effect on Paraguay’s economic growth, 2 the poverty rate declined between 2008 and 2009 in 0 the country (Figure A2.4). The decline in headcount 2009-2010* 1993-1994 1994-1995 1995-1996 1996-1997 1997-1998 1998-1999 1999-2000 2000-2001 2001-2002 2002-2003 2003-2004 2004-2005 2005-2006 2006-2007 2007-2008 2008-2009 -2 poverty from 37.9 to 35.1 percent accounted for a reduction of 130,000 persons living in poverty. The extreme poverty headcount rate basically remained constant, with a reduction of only 0.2 percentage points Source: Estimates of the authors based on the Socio-Economic Database for Latin America and the Caribbean (SEDLAC and the World Bank, 2008) and World Economic Outlook (IMF, 2010). (to 18.8 per cent). Considering a longer period of time, *Projections for 2010. Figure A.2: Gross Domestic Product per Economic Sector and Growth between 2008-2009 4,500 0.20 4,000 GDP, billions Gs. GDP Growth, 0.15 3,500 const of 1994 (left) % (right) 0,10 3,000 0.05 2,500 0 2,000 1,500 -0.05 1,000 -0.10 500 -0.15 0 -0.20 & Forest Commerce Industry Transp & Communicacion General Govt Other services Construction Services to Firms Finances Electricity & Water Hotel & Restaurants Mining Fishing Agric, Livest, Annex Source: Management of Economic Studies, Department of National Accounts and Domestic Market, Central Bank of Paraguay. 59 World Bank, 2010. Did Latin America Learn to Shield its Poor from Economic Shocks? LAC Poverty and Labor Brief, LCSPP. October http://siteresources.worldbank.org/INTLAC/Resources/PovertyReport.pdf 94 60 For compatibility with the sectors used in the Permanent Household Survey, the agriculture, cattle and forestry sectors of the National Accounts were joined into a single sector. extreme poverty in 2009 was at the same level as in 1997/98 (Figure A2.5). However, the moderate poverty Figure A.3: International commodity rate reached its lowest level in 2009. Both results are prices decline in 2009 (US$ nominal) surprising given the substantial economic contraction the country had just suffered. 0.3 The reduction in the national poverty rate between 0.2 2008 and 2009 was primarily the result of the important decline in urban areas outside Asunción, 0.1 while poverty increased in rural areas. The incidence 0 of poverty in the “Urban Central� region and in the -0.1 “Urban Rest� region dropped 8 and 4.3 percentage -0.2 points to 27.4 and 22.8 percent, respectively (Table 2000-2006 2006-2007 2007-2008 2008-2009 2009-2010 A2.1). In Asunción the headcount poverty rate declined only 0.4 percentage points between 2008 and 2009, Source: Global Economic Monitor (DDP World Bank) thus closing the gap in the poverty rate between the Note: Commodities include: Beverages, Fats and oils, Grains, Other Foods and other raw materials. urban regions. However, the incidence of poverty in the “Rural� region increased 1.1 percentage points, reaching almost 50 per cent. The rural sector contributed almost 59 percent of the poor, although accounting for only 41 Figure A.4: Evolution of poverty in Paraguay (2003-2009) percent of the national population. Extreme poverty became even more of a rural issue in 2009. Asunción registered a sharp increase in the 50 45 Poverty extreme poverty headcount rate, from 6.7 to 8.8 percent, 40 35 between 2008 and 2009 (Table 1). However, the Urban Percentage 30 25 Extreme Central and Urban Rest regions showed sharp drops Poverty 20 in extreme poverty, causing the headcount rate of the 15 10 urban area as a whole to drop 1.3 percentage points to 5 9.3 percent in 2009. On the other hand, extreme poverty 0 2003 2004 2005 2006 2007 2008 2009 increased from 30.9 to 32.4 percent in the rural sector, representing 71.1 percent of the extreme poor, an increase of 3.6 percentage points with respect to 2008. The combination of the severe drought and the global financial crisis seems to have had a greater negative impact in the rural sector. Figure A.5: Evolution of poverty in Paraguay (2003-2009) The poorest were even poorer in 2009. The Foster, Greer Poverty Assessment & Thorbecke indices that measure the extreme poverty gap and the severity of extreme poverty show increases 60 particularly in the rural sector in 2009 (Figure A2.6). The 50 Poverty extreme poverty gap index, that estimates the average 40 Percentage distance of the extreme poor to the extreme poverty 30 Extreme Poverty 35.1 line, increased 1 percentage point in the rural sector. 20 The severity index, that gives even more weight to the 10 18.8 poorest of the distribution, increased 0.6 percentage 0 1997-98 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 2009 points for the rural sector. In the urban sector, although 95 the two indices rise, the values are relatively small Table A1. Incidence of poverty and contribution to poverty, Paraguay 2009 Rate of incidence (%) Contribution to poverty (%) % of the popula- Moderate Change Extreme Change Moderate Extreme tion Poverty 2008-09* Poverty 2008-09* Poverty Poverty National 100.0 35.1 -2.8 18.8 -0.2 100.0 100.0 Urban 58.7 24.7 -5.5 9.3 -1.3 41.3 28.9 Asunción 8.1 21.1 -0.4 8.8 +2.1 4.9 3,8 Urban Central 27.6 27.4 -8.0 7.8 -2.8 21.5 11,4 Urban Rest 23.0 22.8 -4.3 11.3 -0.7 14.9 13,8 Rural 41.3 49.8 +1.1 32.4 +1.5 58.7 71.1 Note: (*) The change in the poverty incidence rate between 2008 and 2009 is presented in percentage points. Figure A.6: Gap and Severity of Extreme Poverty, Paraguay 2003-2009 Extreme Poverty Gap Severity of Extreme Poverty Urban Rural National Urban Rural National 16.0 9.0 14.0 8.0 12.0 7.0 10.0 6.0 5.0 8.0 4.0 6.0 3.0 4.0 2.0 2.0 1.0 0.0 0.0 2003 2004 2005 2006 2007 2008 2009 2003 2004 2005 2006 2007 2008 2009 Note: Foster, Greer & Thorbecke Indices (FGT1 & FGT2). and can be considered constant. On the other hand, As a result, in 2009 the urban Gini declined to 0.427 rural results indicate that the poorest among the poor while the rural Gini remained at about 0.554. Hence, suffered the worst effects of the combination of the the effects of the global financial crisis and the severe crisis and drought. drought were a drop in national GDP, an increase in rural poverty with no change in rural income inequality, Inequality and a drop in urban poverty with an improvement in income distribution. Income inequality in Paraguay declined between 2008 and 2009 at the national level and for the urban What is the Explanation for the sector, while inequality in the rural sector remained Increase of Rural Poverty? Annex constant. The inequality of per capita family income, measured by the Gini coefficient, declined 0.02 points, Average Incomes – Growth Incidence Curves from 0.515 in 2008 to 0.496 in 2009. However, the 96 better income distribution occurred only in the urban The average income of the poorest dropped between sector with a decline of 0.032 points in the urban Gini. 2008 and 2009, sharply contrasting with the growth in 2006-08. The incidence curves enable the development of a more detailed characterization of the growth Figure A.7: Growth Incidence Curve patterns in income by showing the proportional change of the National Average Income of each income percentile during a given period. The (2006-08 and 2008-09) incidence curve shows that between 2008 and 2009 the incomes of the poorest households (up to almost the 0.25 2006-08 2008-09 0.20 30th percentile of the distribution) declined, as well as 0.15 those of the richest households from the 85th percentile 0.10 0.05 and higher (Figure A2.7). Meanwhile, during the food 0 -0.05 crisis (2006 to 2008) all households experienced a rise -0.10 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 in their incomes, with the poorest showing a relatively -0.15 -0.20 higher increase than the rest of the distribution (a “pro- -0,25 poor� growth). Note: The incidence curves only show households with an income di erent to cero for As expected, given that the negative economic comparability between the years. Sources: World Bank sta estimates based on the EIH and EPH, Paraguay impact mostly affected the agricultural sector, the average income of the poorest declined mainly in the rural sector. According to the growth incidence curve, average rural income dropped for 95 percent of Figure A.8: GIC of urban and rural average income (2008-09) households between 2008 and 2009, and even more for the poorest (Figure A2.8). On the other hand, most urban households saw their average incomes rise over those 0.20 Urban 08-09 Rural 08-09 two years. Even so, the urban distribution shows a sharp 0.15 0.10 drop in the incomes of the poorest, which is reflected 0.05 0 in the rise of the urban extreme poverty Severity Index -0.05 presented above. The urban distribution also shows a -0.10 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 -0.15 drop in the average incomes of the richest, which can -0.20 -0.25 explain the decline in inequality measured by the urban -0.30 Gini coefficient. Figure A2.9 presents the contrast that took place in the 2006-2008 period during which the Note: The incidence curves only show households with an income di erent to cero for incomes of all households, both urban and rural, rose. comparability between the years. Sources: World Bank sta estimates based on the EIH and EPH, Paraguay Within the rural sector, agricultural incomes decline for the entire distribution while the incomes of the non-agricultural rural sectors show mixed results Figure A.9: GIC of urban and rural (Figure A2.10). These more disaggregated results for average income (2006-08) the rural sector underscore that the negative impact on the country’s economic growth and rural poverty is 0.45 0.40 Rural 06-08 Urban 06-08 Poverty Assessment primarily a result of the negative impact of the financial 0.35 crisis and the drought on the agricultural sector. 0.30 0.25 0.20 Public Transfers 0.15 0.10 0.05 Publice transfers to rural households were mainly to 0 households from the poorest deciles. In the context 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 of the global financial crisis that began in 2008, the Government of Paraguay included an expansion of the Note: The incidence curves only show households with an income di erent to cero for comparability between the years. 97 conditional cash transfers program (CCTs) as one of the Sources: World Bank sta estimates based on the EIH and EPH, Paraguay Labor Market Figure A.10: Growth Incidence Curves Rural Sector 2008-09 Labor market indicators show mixed results for Paraguay between 2008-2009, with increases in the employment rate and in labor participation, Non-agriculture 08-09 Agriculture 08-09 but also in the unemployment rate and informality. 0.6 Although the negative consequences of the financial 0.4 crisis are still visible in most of the world, the labor 0.2 force participation rate increased 2.1 percent and the 0 employment rate increased 0.8 percentage points -0.2 in Paraguay between 2008 and 2009 (Table A2.2). Female labor force participation also increased 3.1 -0.4 percent while underemployment declined 0.7 percent. 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 However, the unemployment rate was negatively affected, rising 0.7 percentage points, as well as Note: The incidence curves only show households with an income di erent to cero for comparability between the years. informality, which rose 2.4 percent over the two years, Sources: World Bank sta estimates based on the EIH and EPH, Paraguay results that are more expected in the context of the global crisis. Figure A.11: State transfers (including Tekoporã) in rural areas Employment (% of the total family income) The positive results in terms of labor participation and the employment rate during the period of the crisis 3.0 seem to have been particularly influenced by the 2.5 considerable absorption capacity of the agricultural 2.0 sector, even given the drop in GDP in this sector. 2.5 Agriculture continued to be the sector with the highest 1.0 participation in employment (28.2 percent of workers 0.5 in 2009), and it was the one with the greatest increase 0.0 with a rise to 3.1 percentage points between 2008 and 1 2 3 4 5 6 7 8 9 10 2009 (Figure A2.12). Commerce and the restaurants and hotels also showed improvement in the distribution of Source: World Bank sta calculations based on EPH, Paraguay. workers per economic sector. priority policies of its administration. Likewise, for the Most new jobs were in the agricultural sector, even first time the Permanent Household Survey included given the sharp drop of agricultural GDP. If job a variable to capture the incomes of the CCT program, creation and destruction are classified per economic in particular the expansion of the Tekoporã program, sector and labor relations, the result shows that almost among others. Figure A2.11 shows that public transfers 110,000 workers more entered the agricultural sector in were between 1.5 and 2.5 percent of total family income comparison with 2008, followed by the commerce sector for the first three deciles of rural households. For these (almost 37,000 net workers more), and restaurants and households, income from remittances was greater as hotels. In percentage terms, hotels were the most the a percentage of family income, but the rise between most important of the three. In terms of labor relations, Annex 2008 and 2009 was almost nil for the two poorest rural the number of salaried workers declined between 2008 deciles (see section on remittances below), whereby and 2009 while the number of less formal jobs, such as public transfers were the only increase. Without these autonomous work (with a net increase of almost 66,000 98 incomes, it is possible that rural poverty would have persons) and non-remunerated workers (with a net been even higher. increase of almost 35,000 persons), rose. Table A2. Labor Market Indicators in Paraguay 2008-2009 Indicator a 2008 2009 Change Labor force participation (%) 63.7 65.0 2.1% Women’s participation in the labor force (%) 50.4 52.0 3.1% Employment rate (%) 59.9 60.7 0.8 b Visible underemployment (% of the total employed) 7.1 8.7 -0.69% Unemployment rate (% of the economically active population) 5.9 6.6 0.7 b Duration of unemployment (months) 7.2 7.5 0.3 c Informal workers (% of the total employed) 5.5 7.0 2.4% Note: (a) Indicators refer to the population aged 10-64 years; (b) the change is expressed in percentage points; (c) months. Informal workers: salaried workers in small enterprises, autonomous non-professionals and workers without income. Visible underemployment: percentage of persons with employment who work less than 30 hours per week and wish to work longer hours. Source: World Bank staff estimates based on the EPH, Paraguay Figure A.12: Distribution of workers per economic sectors (percentage points) 2008 2009 30.0 25.0 20.0 15.0 10.0 5.0 0.0 Electricity & Water & Forest Fishing Mining Industry Construction Commerce Hotel & Restaurants Transp & Communicacion Finances Services to Firms General Govt Other services Agric, Livest, Source: World Bank sta calculations based on EPH 2008 and 2009, Paraguay. The employment rate rose for women, for younger Unemployment workers, and in the rural sector, but remained relatively constant for men. Table A2.3 presents the The unemployment rate rose from 5.9 to 6.6 per cent Poverty Assessment employment rate for different groups between 2008 in the 2008-2009 period, more in line with predictions and 2009. Results show that the employment rate of given the unfavorable international context of the the rural sector rose 1.7 percentage points between financial crisis. The rise in unemployment occurred 2008 and 2009, while the urban employment rate in both urban and rural areas, but proved to be only rose 0.25 points. Women’s employment rate rose more severe for rural workers, unlike the situation 1.2 percentage points and that of youth between 10 between 2007 and 2008 when rural unemployment and 24 years of age rose almost 2.3 percentage points. rural declined while urban unemployment was rising Meanwhile, the men’s employment rate remained (Figure A2.13). The rise in 2009 was also higher for men, relatively constant (a decline of 0.09 percentage meaning that the gender gap had been closing over 99 points). the last two years. Table A 3. Distribution of Employment, 2008 and 2009 (%) Adults (10-64) Age Gender Area Total (10-24) (25-64) (65 +) Women Men Rural Urban 2008 59.9 39.0 76.3 37.9 46.6 73.4 60.8 59.3 2009 60.7 41.2 76.1 37.6 47.8 73.3 62.5 59.5 2009-2008 0.81 2.26 -0.20 -0.23 1.19 -0.09 1.65 0.25 Note: Change between 2008 and 2009 is expressed in percentage points A possible hypothesis summarizing the results of drop in family income in the rural area due to the crisis the rural area would be that the combination of the and the drought. In any case, it would be important to financial crisis and the drought had a negative effect make an in-depth analysis of this hypothesis. on men’s jobs in the agricultural sector (the most important rural employer), and on the income of rural Why does Urban Poverty Decline? families, driving women and the young to enter the labor market to contribute to household finances. The Urban Employment above results show that the men’s employment rate remained relatively constant while their unemployment The negative impact of the financial crisis on the GDP rate rose, as well as the unemployment rate for the rural of the most important urban sectors does not seem to sector. Instead, the employment rate rose for women, have translated into declines in urban employment, the young, and in the rural sectors rose. An analysis of nor, in some cases, declines in the income per hour. the job creation for women shows that between 2008 The sectors of commerce, industry and transport & and 2009 they entered primarily into the agricultural communications are the most important in the urban sector, and in particular importantly as autonomous area in GDP terms. As expressed above, although not workers and entrepreneurs. Hence, a hypothesis would as much as the agricultural one, these sectors suffered be that women and the young attempted to buffer the a decline of the GDP between 2008 and 2009 (Figure Figure A.13: Rates of unemployment per area and sex Urban Rural 80 16.0 70 14.0 60 12.0 Unemployment rate Unemployment rate 50 10.0 Female 40 8.0 30 6.0 Total 20 4.0 Annex 10 2.0 Male 0 0.0 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 1999 2000-01 2002 2003 2004 2005 2006 2007 2008 100 Source: World Bank sta estimates based on the EIH and EPH, Paraguay. 2). However, in the same period urban employment in water. However, analyzing the changes in the four these sectors rose, as well as salaries per hour in the sectors where the urban poor work most (commerce, commerce and industry sectors (Figure 14). industry, domestic service and construction), in general the salaries of less skilled workers rose more than for On the contrary, the financial crisis seems to have workers with secondary education completed (Figure resulted in a deterioration of the quality of urban 15). This growth in sectors requiring labor force with low employment due to the rise in unemployment, education helped increase the incomes of the poorer informality and visible underemployment in the and hence to reduce the urban poverty rate. All these urban sector. The rise in labor participation and the rise dynamics of the urban sector require more research from 7.5 to 8.3 per cent of urban unemployment are than is possible in this brief paper. consistent with the onset of the financial crisis if more people start looking for work to maintain the level of Remittances household consumption. The rise in the rate of urban informality (from 52.5 to 55.0 per cent) as well as the Poor and non-poor, urban and rural households rate of visible underemployment (workers employed received remittances in 2009, but the remittances were for less than 30 hours per week who wish to work longer higher (as percentage of the family income) for the hours) from 6.4 to 8.1 per cent in the urban sector can poorest deciles. In absolute values, remittances were help explain the puzzle of a decline of the GDP with a received primarily by the richer in 2009, and originating rise of employment in the urban sector. An additional mostly in Spain and the United States. However, as phenomenon was the return of migrants to Paraguay percentage of the family income, remittances impacted due to the much greater impact of the financial crisis on the entire distribution of households, but in particular in OECD countries, whereby these returning workers on the poorest deciles (Figure A2.16). Remittances to may have added to labor participation, as well as to the the urban sector were between 3 and 10 percent of employment and unemployment rates. the family income of the poorest households, with the greatest impact on the first decile of the distribution. Urban Salaries Between 2008 and 2009, remittances rose more Salaries rose for less skilled workers in urban sectors substantially as percentage of the family income for such as commerce and construction. Salaries declined the poorest urban deciles, with almost no rise for the for the following sectors: transport & communication, two poorest rural deciles (Figure A2.17). At the other services to enterprises, agriculture, and electricity & end, remittances increased for the richer urban and rural Figure A.14: Urban Employment, Change in Urban Employment and Change in Income Per Hour Employed, 2009 (thousands) Change in Employment, 2008-09 (%) Change in hourly wage, 2008-09 (%) 600 80 500 60 Poverty Assessment 40 400 20 300 0 200 -20 -40 100 -60 0 -80 Servants Transp & Communicacion Services to Firms & Forest Hotel & Restaurants Soc Serv Commerce Industry Construction Public Admin. Other services Education Finances Electricity & Water Agric, Livest, Domestic Health and 101 Source: World Bank sta estimates based on the EIH and EPH, Paraguay. households. Poor rural households, and in particular the extreme poor, were among those who received Figure A.15: Change in Salary between 2008 and 2009 per sector less help in the form of remittances. On the contrary, and quali�cations (%) the rise in remittances between 2008 and 2009 for the poorer urban households many have contributed to Incomplete Primary Complete Secondary Complete Primary the decline in urban poverty in this period. 140 120 Beyond the increase in employment, hourly income, 100 80 and remittances, there are other buffers in the urban Percentage 60 sector that could also help explain the decline in 40 20 urban poverty. Social spending increased significantly, 0 not only in terms of the budgets approved, but in -20 Commerce Industry Dom. Serv. Construction -40 dramatic increases of current execution in areas such as health, poverty reduction programs, social security and public works. For example, the construction Source: World Bank sta estimates based on the EIH and EPH, Paraguay. sector, which employs an important proportion of less qualified workers, also grew given the higher execution of the Ministry of Public Works, which executed 70 Figure A.16: Remittances as percentage percent more than the previous year. Furthermore, of urban and rural the increase of cash transfers to the regions resulted in household income, 2009 (%) an increase in works in the interior of the country. All these factors, in addition to those mentioned above, Urban Rural 12 may have influenced to improve the urban poverty 10 rate. However, as mentioned above, a more exhaustive 8 analysis is required to really understand the dynamics 6 taking place in the country, both in the urban and rural 4 areas. 2 0 Conclusions 1 2 3 4 5 6 7 8 9 10 The succinct analysis of some of the data on poverty Source: World Bank sta estimates based on the EIH and EPH, Paraguay. and the labor market presented in this brief Annex suggests that there are no sharp changes in the policy considerations presented in the Paraguay’s Poverty Assessment “Determinants and Challenges of Figure A.17: Change in remittances Poverty Reduction�. Countercyclical policies during the as percentage of family income, 2009 recession, combined with the maintenance of 2008-2009 (%). Urban and Rural macroeconomic stability, helped contain the decline of real GDP at 3.8 percent and to reduce poverty between Urban Rural 8 2008 and 2009. At the microeconomic level, the results 6 show that poverty and extreme poverty continue to be 4 primarily a rural issue, and more so in 2009. The increase in rural poverty, despite the increase in labor participation 2 Annex and despite public transfers, suggests that the capacity 0 of absorption and the flexibility of the agricultural 1 2 3 4 5 6 7 8 9 10 -2 sector, combined with the social protection network, 102 were not enough. The external shock, that affected the Source: World Bank sta estimates based on the EIH and EPH, Paraguay. agricultural sector so severely, underscored the need to enhance this sector’s productivity, generate a sound the labor force through better education (in particular investment climate, and strengthen the human capital secondary education) and less informality. Although of the rural poor so that they may diversify their sources the 2009 results show an increase in the salaries of of income and expand their employment options to be less qualified workers, given the rise in the demand for in better conditions to face these crises. The hypothesis this sort of worker, even so workers with completed that the labor participation of the young rose as a way secondary education in the commerce sector earn 40 of buffering the drop in household incomes entails the per cent more than workers who only completed their need to ensure that the educational indicators do not primary education and those of the construction sector decline, and consequently of the future productivity of earn 25 per cent more than the less skilled workers in the labor force. The Government could consider rural the same sector. work or temporary employment programs that would be ready for their implementation in rural areas in times As a last consideration, improvements in the monitoring of crisis. As example we observe that, in urban areas, and evaluation of public sector programs could improve poverty benefited from the construction boom in part the focalization and overall impact of said programs, related to the policy of investing in public works (and and help create broad support for the Government’s the greater execution of these resources in 2009), which poverty reduction strategy. increased the demand for less skilled workers (with highest probability of being poor) in the urban sector. An impact assessment would be necessary in order to properly analyze the effect of the conditional cash transfers in the rural sector, but a first reading of the data may suggest that, even if the number of beneficiaries was high, the size of the benefit was not enough to buffer the negative effect on the agricultural sector. This would suggest that, without these transfers, poverty could have increased even more in the rural sector. However, this brief analysis cannot answer the question on how well-focalized are the CCTs. Policy considerations on the social protection programs of the Paper on Poverty continue to be important: the need to unify the atomized set of social protection programs, strengthening their focalization and investing in their monitoring and evaluation. In the urban sector, the negative impact of the crisis on the GDP of the most important urban sectors did not translate into declines in urban employment Poverty Assessment in those sectors, nor, in some cases, to declines in the salary per hour. However, there was indeed a deterioration of the quality of urban employment due to the increase of unemployment, informality and visible underemployment, which explains the puzzle, although in general the effects were small. The increase in informality exacerbates the already existing structural issues that affect poverty in urban sectors. As expressed in the Paper on Poverty, it is continues 103 to be important to seek to increase the productivity of REFERENCES De Janvry, Alain, Craig McIntosh, and Elisabeth Sadoulet. 2006. “From Private to Public Reputation in Microfinance Lending: An Experiment in Borrower Response.� University of California at Berkeley. Berkeley, Amartya, S. An Aspect of Indian Agriculture, Economic CA. Processed. Weekly, Vol. 14, 1962. De Ferranti, David, Guillermo E. Perry, William Foster, Banco Central. 2009. Estadísticas Económicas. Anexo Daniel Lederman, and Alberto Valdés. 2005. Beyond Estadístico. Informe Económico. www.bcp.gov.py the City: The Rural Contribution to Development. Washington, DC: World Bank. Barret, C. B. 1996. On price risk and the inverse farm size-productivity relationship. Journal of Development Dirección General de Estadística, Encuestas y Censos de Economics, 51, 193-215. Paraguay--DGEEC. Paraguay. Benjamin, D., & Brandt, L. 2002. Property rights, labour Flood, Cristina. 2002. El gasto Publico Social en Paraguay. markets, and efficiency in a transition economy: the Mimeo. Asunción. case of rural China. Canadian Journal of Economics, 35(4), 689-716. Foster, J., J. Greer, and E. Thorbecke. 1984. A Class of Decomposable Poverty Measures, Econometrica 52, Berry, A., & Cline, W. R. 1979. Agrarian Structure and 761-765. Productivity in Developing Countries. London: John Hopkins University Press. Franco, Nelson y Norma Medina Roa. 2008. Programa Tekoporã: Características sociodemográficas y Bertranou, Fabio, Andres Mariniakis, y Gerhard Reinecke. económicas de la población objetivo. Publicado por 2003. Mercado de Trabajo, instituciones y pobreza: GTZ, SAS y UNFPA. November. Interrelaciones y desafíos de Paraguay, en “Paraguay: Empleo y Protección Social. Desafíos institucionales para Galor, Oded, Omer Moav and Dietrich Vollrath. 2009. reducir la pobreza�, editado por la Oficina Internacional “Inequality in Landownership, the Emergence of del Trabajo. Human-Capital Promoting Institutions, and the Great Divergence,� Review of Economic Studies, Blackwell Binswanger, Hans P. and Klaus Deininger. 1997. Publishing, vol. 76(1), pages 143-179, 01. “Explaining Agricultural and Agrarian Policies in Developing Countries,� Journal of Economic Literature, Grushka, Carlos y Deborah Altieri. 2003. Evaluación American Economic Association, vol. 35(4), pages 1958- actuarial y proyecciones 2000-2050 del régimen de 2005, December. jubilaciones y pensiones administrado por el Instituto de Previsión Social, en “Paraguay: Empleo y Protección Blanco, Gastón Mariano. 2008. Red de Protección y Social. Desafíos institucionales para reducir la pobreza�, Seguridad Social. Nota de progreso del Banco Mundial editado por la Oficina Internacional del Trabajo. Carter, Michael R. and Galeano, Luis, 1995. Campesinos, GTZ. 2008: Tekpora in Paraguay, Basic Social Protection Tierra y Mercado. , Centro Paraguayo de Estudios in German Development Cooperation. Germany, 2008. Sociológicos, Asunción. Hirata, Guilherme, Fabio Soares, y Rafael Perez Ribas. Carter, Michael R. and Olinto, Pedro, Getting Institutions 2008. Proyección de impactos de la expansión del ‘Right’ for Whom? Credit Constraints and the Impact of Programa Tekoporã. Informe Final. Property Rights on the Quantity and Composition of Investment. American Journal of Agricultural Economics, ILO. 2006. “Youth: Pathways to decent work� in Vol. 85, pp. 173-186, 2003 “Promoting youth employment – Tackling the References challenge.� Datt, Gaurav, and Martin Ravallion. 1991. “Growth and Redistribution Components of Changes in Poverty Lokshin, Michael M., and Martin Ravallion. 2006. Measures: A Decomposition with Applications to Brazil “Testing Poverty Lines.� Review of Income and Wealth and India in the 1980s.� Living Standards Measurement 52(3): 399–421. 104 Study Working Paper No. 83, World Bank, Washington, DC. Ñopo, Hugo. 2008. “Matching as a Tool to Decompose Blackwell Publishing, vol. 53(2), pages 275-294, Wage Gaps� The Review of Economics and Statistics, 06.Soares and Britto, 2007. 90(2): 290–299, May. Soares, Fabio V. y Tatiana Britto. 2007. Confronting Pradhan, Menno, Asep Suryahadi, Sudarno Sumarto, Capacity Constraints on Conditional Cash Transfers in and Lant Pritchett. 2001. “Eating like which ‘Joneses?’ Latin America: The cases of El Salvador and Paraguay. an Iterative Solution to the Choice of a Poverty Line International Poverty Centre, working paper #38, ‘Reference Group’.� Review of Income and Wealth 47(4): August. 473–487, December. Soares, Fabio V., Rafael Perez Ribas, y Guilherme Hirata. Perez Ribas, Rafael, Guilherme Hirata y Fabio Soares. 2008. Los logros y las carencias de las transferencias 2008. Debatiendo los métodos de focalización para de efectivo condicionadas: Evaluación del Impacto del las transferencias monetarias: �ndice Multidimensional Programa Tekoporã del Paraguay. Nota de evaluación vs. Proxy de Ingresos para el Programa Tekoporã de publicada por el Centro Internacional de la Pobreza. Paraguay. International Poverty Centre, working paper #2, March. Soares, Fabio V., Rafael Perez Ribas, y Guilherme Hirata. 2008. Informe Final de la Evaluacion de Impacto del Ravallion and Chen. 2003. “Measuring pro-poor growth� Programa Piloto Tekoporã. Secretaría de Acción Social, World Bank, MSN MC 3-306 Development Research UNFPA, GTZ, International Poverty Centre. Group. Masterson, Thomas. 2007. “Land Rental and Sales Ravallion, Martin. 1998. Poverty Lines in Theory and Markets in Paraguay,� Economics Working Paper Archive Practice, LSMS Paper 133, Washington DC, World Bank. wp_491, Levy Economics Institute, The Unidad de Economía Social, Ministerio de Hacienda, 2010. _______________. 2005. “Poverty Lines,� in The New Palgrave Dictionary of Economics, 2nd edition, Larry UNDP Human Development Reports. 2002 and 2009. For Blume and Steven Durlauf (eds.) London: Palgrave the Human Development Index; Ministry of Education Macmillan. for Costa Rica School net primary enrollment rate; WDI for all other indicators. Ravallion, Martin. 2007. “Inequality Is Bad for the Poor.� In Inequality and Poverty Re-Examined, ed. Stephen World Bank. 2004. Paraguay: Definiendo una Estrategia Jenkins and John Micklewright. Oxford, UK: Oxford para la Política de Protección Social. Reporte Nro. 27352- University Press. PA. Documento del Banco Mundial. Saldaín, Rodolfo. 2003. Análisis de las reformas World Bank. 2007. Report No. 35910-CR. Costa Rica propuestas al sistema de jubilaciones y pensiones, Poverty Assessment, Recapturing Momentum for en “Paraguay: Empleo y Protección Social. Desafíos Poverty Reduction, February 12. institucionales para reducir la pobreza�, editado por la Oficina Internacional del Trabajo. World Bank. 2008. World Development Report. Agriculture for Development. San Martino, Jorge y N. Capellari. 2003. Gestión del riesgo social en Paraguay. Informe de Consultoría, World Bank. 2009. Measuring Inequality of Opportunities Proyecto “Paraguay: Defining a Strategic Approach to in Latin America. Regional Study – Chief Economist the Social Protection Policy�, Washington, DC. World Office Latin America and the Caribbean Region. World Poverty Assessment Bank. Bank. Washington, DC. Sanchez, Andrea. 2003. Cobertura de la Protección Social en Salud, Jubilaciones y Pensiones, en “Paraguay: Empleo y Protección Social. Desafíos institucionales para reducir la pobreza�, editado por la Oficina Internacional del Trabajo. Simler, Kenneth R. and Channing Arndt. 2007. “Poverty Comparisons With Absolute Poverty Lines Estimated 105 From Survey Data,� Review of Income and Wealth, WORLD BANK Paraguay Country Office Av. España 2028 c/ Brasilia Urano Building - 5th floor Asunción Tel: 595 21 23 11 55 Fax: 595 21 23 11 96 www.worldbank.org/py