Report No. 29594-SV El Salvador Poverty Assessment Strengthening Social Policy December 29, 2005 Poverty Reduction and Economic Management and Human Development Sector Management Units Latin America and the Caribbean Region Document of the World Bank CURRENCY EQUIVALENTS (Exchange RateEffective April 10,2001) Currency Unit = Colon US$1 = 8.75 Colones FISCAL YEAR January 1-December31 Vice President: David de Ferranti Country Director: Jane Armitage Director PREM: Ernest0May Director HD: Ana-Maria Arriagada Lead Economist: C. Felipe Jaramillo Sector Manager PREM: Jaime Saavedra Sector Manager HD: Helena Ribe Task Manager: Andrew D.Mason ... 111 EL SALVADOR Poverty Assessment Toward a National SocialPolicy TABLE OF CONTENTS PREFACE ................................................................................................................................................ viii ... EXECUTIVE SUMMARY...................................................................................................................... x PART 1:POVERTY AND VULNERABILITY INEL SALVADOR. 1991-2002 1. INTRODUCTION ................................................................................................................... 1 2. RECENT ACHIEVEMENTS. CURRENT CHALLENGES .................................................. 5 3. EXPLAININGRECENT POVERTY REDUCTION ............................................................. 18 PART 2: CRAFTING A NATIONAL SOCIAL POLICY 4. ELEMENTS OF A POVERTY REDUCTION STRATEGY.................................................. 33 5. EDUCATION.......................................................................................................................... 34 6. HEALTH.................................................................................................................................. 48 7. SOCIO-ECONOMICINFRASTRUCTURE........................................................................... 64 8. SOCIAL PROTECTION ......................................................................................................... 82 PART3: TOWARD A NATIONAL SOCIAL POLICY FOR EL SALVADOR 9. THE WAY FORWARD .......................................................................................................... 99 REFERENCES......................................................................................................................................... 109 TABLES 1.1 Key Socio-Economic Indicators inEl Salvador. Early 1190sto Early 2000s................................... 2 1.2 Socio-economicindicators -How ElSalvador Compareswith LatinAmerica and Lower- MiddleIncome Countries. 2000-2002............................................................................................... 3 2.1 The Evolution of Poverty, National. Urban, and Rural. 1991-2002.................................................. 8 2.2 The Depthand Severity of Poverty. National. Urban. andRural. 1991-2002................................... 8 2.3 Income Inequality inEl Salvador: Gini coefficients of Householdper Capita Income, 1991-2002........................................................................................................................................... 15 2.4 Chronic andTransient Poverty inRuralEl Salvador, 1995-2001..................................................... 16 iv 3.1 Elasticities of Poverty Reductionwith Respect to per Capita Income Growth and to Changes inIncome Inequality. 1991-2002 ........................................................................................ 18 3.2 Patterns of per Capita HouseholdIncome Growth. 1991-2002 ......................................................... 19 3.3 Changes inthe Sectoral Structure o f Poverty Between. 1991192-2000............................................. 20 3.4 Marginal Contribution o f Education to Householdper Capita Income. 1991-2002.......................... 23 3.5 Municipalities and Populationmost affected by the 2001 Earthquakes inEl Salvador. 1999-2002........................................................................................................................................... 28 5.1 Expenditureper student inUS$ PPP and as % o f GNP per capita. 2000 .......................................... 38 5.2 Reasons for not Attending School. by Area. 2002............................................................................. 44 6.1 Key Health, Key Indicators, early 1990s- early 2000s..................................................................... 49 6.2 Health Service Access Indicators, early 1990s-early 2000s............................................................ 49 6.3 Public and Private Health Spendingand Sources o f Financing, 1996-2000...................................... 55 6.4 Distributiono fPersons that Visit MSPAS Facilities by Income Group, 2002.................................. 57 6.5 Population Coveredby MedicalInsurance, 2002 (percent o f total population) ................................ 59 6.6 Establishment that Visitedwhen Ill,2002 (percent o f Ill) ................................................................. 60 6.7 Motives for Not Visiting MSPAS Establishments, 2002 (percent o f Ill).......................................... 60 7.1 Infant and Child Mortality, by Access to Water and Toilets (March 1993 to February 1998) ................................................................................................................................................. 64 7.2 Sources o f Water inRural Areas, 1992 -2002.................................................................................. 66 7.3 Institutions inthe Rural Water and Sanitation Sector, by Role ......................................................... 69 7.4 Public Spending and Investment inWater Supply and Sanitation, 1999 - 2003 ............................... 72 7.5 Public Expenditure on RoadInfrastructure, 1997 - 2003 .................................................................. 80 8.1 Spending on El Salvador's Social Safety, 2000................................................................................. 87 A.l RisksRiskGroups, Programs, Coverage and Costs, 2000................................................................ 95 9.1 Estimated Costs of ImplementingaNational Social Policy for Poverty Reduction ......................... 107 FIGURES 2.1 The Evolution o f Total and Extreme Poverty. National Level. 1991-2002.................................... 7 2.2 Sensitivity o f the 1991-2002 Poverty Decline to Different Measurement Approaches .................... 9 2.3 RuralPoverty Trends. 1995 -2001................................................................................................... 10 2.4 Net Enrollment Rates inBasic Education in El Salvador, by Area and Income Quintiles ...............11 2.5 Infant Mortality per Socio-Economic Group .................................................................................... 12 2.6 Access to Basic Services inEl Salvador, 1991-2002 ....................................................................... 13 2.7 Gini Coefficients inL A C (Distribution o f equivalized household income circa 2000) .................... 14 V 2.8 Distribution o f Householdper Capita Income inEl Salvador. 1991-2002..................................... 15 3.1 Trends inPoverty With and Without Remittances. usingEHPM and Basis Data............................. 27 3.2 Quality o f Housing inEl Salvador. 1999-2002 ............................................................................... 29 3.3 Coffee Prices inUS$ per Pound (January 1980 -August 2003 ........................................................ 30 3.4 Change inPercentage o f Underweight Children. 1998.2002. by Department and Intensityof coffee Cultivation........................................................................................................... 31 5.1 Net Basic and Secondary Enrollment Rates, by Poverty Status, 1991 - 2002 .................................. 35 5.2 Basic Completion Rates (lst & 2"d cycles) by Quintile, 1995-2002............................................... 35 5.3 Average Years o f Schooling, by Quintile, 1995 -2002 .................................................................... 35 5.4 Education Sector Expenditures inEl Salvador .................................................................................. 37 5.5 Education Spending as a Proportiono f GNP, 1999/2000.................................................................. 38 5.6 Allocation o f Recurrent Spending across Educational Levels........................................................... 40 5.7 Distribution o f Public Subsidy to Basic Education, by Quintile, 1995 and 2002 .............................. 41 5.8 Distribution o f Public Subsidy to Secondary Education, by Quintile, 1995 and 2002 ...................... 42 5.9 Basic and Secondary Completion Rates, by Quintile, 2002 .............................................................. 43 5.10 PAES Test Scores and per Capita Income, by Department, 2000.................................................... 46 6.1 Total Health Spending (Public and Private) as a Proportion o f GDP, 2000 ...................................... 53 6.2 Public Health Spending as a Proportion o f GNP 2000 ...................................................................... 53 6.3 Health Spending per Capita, 1997- 2000 ......................................................................................... 54 6.4 MSPAS Expenditures, 1996-2002, as a Percent o f GDP and o f Total Government Spending ............................................................................................................................................ 55 6.5 MSPAS Expenditures by Level o f Attention and Administration, 1996 - 2003................................ 56 6.6 Distribution o f Public Subsidy to Primary HealthCare, by Quintile, 2002....................................... 58 6.7 Distribution o f Public Subsidy to HospitalCare, by Quintile, 2002.................................................. 58 7.1 Annual Increase inAccess to ImprovedWater ................................................................................. 66 7.2 Access to ImprovedWater Sources, 2000 ......................................................................................... 67 7.3 Access to Improved SanitationFacilities, 2000................................................................................. 68 7.4 Central Government Spending on Water Supply and Sanitation....................................................... 72 7.5 Distance to Paved Roads inRuralAreas (in Kms.) ........................................................................... 77 7.6 Distance to the Nearest Market (inKms.) .......................................................................................... 78 7.7 Average Annual Increase inKilometers o f PavedRoads inthe 1990s inLatin America .................79 vi 2.1 Challenges of Poverty Measurement inEl Salvador......................................................................... 5 4.1 Promoting Economic Growth for Poverty Reduction inEl Salvador................................................ 33 8.1 Social Protection -Managing Social Risk, Promoting Long-term Growth and Development ..................................................................................................................................... 82 8.2 Strengthening Poor Children's Education, Health, and Nutrition through Conditional Cash Transfers ................................................................................................................................... 92 8.3 ProgramaUnidos por l a Solidaridad - DevelopingCommunities inPartnership with Salvadorans inthe Diaspora ............................................................................................................... 94 7.1 Rural ImprovedWater Providers, by Municipality .......................................................................... 71 Acknowledgements This report was prepared by a team led by Andrew D. Mason (LCSHS) under the general supervision o f Helena Ribe (LCSHD). The team comprised Omar Arias, Carlos Sobrado (LCSPP); Emanuela di Gropello, Jesus Maria Fernandez (LCSHD); Jose Marques (Synthesis Consultores Intemacionales); Anabella Larde de Palomo, Margarita Beneke de Sanfelid, Mauricio Shi, Alvaro Trigueros (FUSADES); Claudio Gonzalez-Vega, Jorge Rodriguez-Meza (Ohio State University), Tito Yepes, Jose Montes, Ana Cristina Torres, Reyes Aterido, David Glejberman, Nancy Gillespie, and Mark Feige (Consultants). Logistical support, as well as assistance inprocessing the document, was provided by Christina Alquinta and Claudia Isem (LCSHD). We would like to thank Juan Jose Daboub, Evelyn Jacir de Lovo, Carmen Regina de Arevalo, Eduardo Zablah-Touche, Yolanda de Gavidia and Miguel Angel Siman for their ongoing interest, contributions, and support for this work. We are also grateful to Evelyn Jacir de Lovo and to the participants o f the Government o f El Salvador's MDG initiative, including Consultative Committee members from the UNDP, IDB, FUSADES, FLACSO, FUNDE, and FUNDAUNGO, and Technical Committee members, Orlando Martinez Orellana, Aida Argiiello de Morera, and Juan Albert0 Castellanos Veciana. We would also like to thank Tom Hawk, Emesto Nosthas, Guillermo Garcia, and Carlos Alfred0 Molina (FISDL), for their valuable feedback and inputs at various stages. Carlos Salvador Melgar and MiguelAngel Corleto (DIGESTYC) also provided useful and timely support with El Salvador's national household surveys. The report has benefited enormously from comments and other inputsprovided by our governmental counterparts and non-governmental partners during a series o f meetings and consultations. The team would also like to thank the organizers and participants o f the April consultative workshop, "Los retos de crecimiento y la reducci6n de la probreza en el nuevo milenio", including Antonio Cabrales and his team at FUSADES. The workshop provided important feedback to the report team at a critical stage o f drafting the study. Finally, we are grateful to our peer reviewers Margaret Grosh (HDNSP) and Jesko Hentschel (LCSHD) for their timely and insightful feedback at various stages, to Guillermo Perry and William Maloney (LCSCE) for their constructive feedback at the concept and decision stages, and to Jaime Saavedra (LCSPP) for his feedback and support throughout the course o f the study. vii PREFACE This Poverty Assessment i s part o f an ongoing analytical and advisory program developed by the World Bank in responseto a request from the Government o f El Salvador (GOES) for advice and assistance on poverty reduction, social sector policy and poverty monitoring. The request grew out o f the Government's interest in examining the poverty and social situation in the aftermath of the 2001 earthquakes. The Poverty Assessment fits into a broader program of work that has included both analytical and technical support. Together, these activities have helped set the stage to support the Government's flagship poverty reduction program, the Red Solidaria, a multi-sectoral program launched by President Saca earlier this year to strengthen the human capital and productive capacity o f the poorest Salvadorans. The first phase of the Bank's programmatic support on poverty and social policy began in 2001 and included three main elements: e Analysis o f the poverty situation inEl Salvador and policy advice on how to improve the poverty impact o f social policy and social sector spending; e Analysis of vulnerable groups in El Salvador and o f existing programs to confront vulnerability, along with recommendations on how to strengthen the country's social safety net; and e Technical assistance on data collection to enable analysis o f the poverty impacts o f the earthquakes. Phase Iculminated in June 2002 with the delivery to the GOES o f two documents -Policy Note on Poverty and Social Expenditures and Policy Note on Elements to Strengthen the Social Safety Net and follow-up discussions with members of El Salvador's Economic and Social Cabinets. The analytical work was accompanied by technical assistance to the Salvadoran statistical agency, DIGESTYC, designed to enable analysis o f the poverty impacts o f the 2001 earthquakes and other poverty-related issues at the municipal level - a more geographically disaggregated level than had been available in the past. This first phase provided an important analytical foundation, and generated further demand from the Government for poverty analysis and dialogue focused on strengthening the impact o f El Salvador's policies and programs addressing poverty and social sector outcomes. The second phase of the program was carried out in response to a request from the GOES inlate 2002 for more in-depthanalysis on selected themes and led to the preparation o f this Poverty Assessment, Strengthening Social Policy. Specifically, the Government asked that the Bank: (i) update the poverty profile with special emphasis on the poverty impact o f the 2001 earthquakes; (ii)undertake a detailed and up-to-date analysis o f social sector spending and its impact on the poor; and (iii) examine the likely fiscal costs o f achieving further poverty reduction and socio- economic progress. For this phase o f the work, the GOES also requested that the Bank broaden its definition o f "social sector" spending to include public spending on rural water and rural roads, to make the work consistent with the Government o f El Salvador's own definition o f the social sectors. I n line with the availability o f houscliold survey data during thc preparation phasc of this report, most of the quantitative analysis focuses on tlie 1901-2002 period; analysis of policies generally cxtcnds until mid-2004, thc end of the sitting govcrnmcnt's term of`office. Alongside the analytical efforts, the Bank's Poverty Assessment team provided technical assistanceand support inother areas, including: ... Vlll 0 Technical assistance to DIGESTYC, as part o f broader GOES efforts to construct a municipal-levelPoverty Map; 0 Technical support in the production o f a 2004 study o f the Millennium Development Goals (MDGs) in El Salvador. This report was initiated by the country's Commissioner for the Coordination o f the Social Areas, and included a wide range o f national and international stakeholders, including the World Bank, UNDP, IDB, and several think tanks and NGOs in El Salvador. The Bank provided a range o f technical inputs and guidance to the analytical exercise which culminated in the launching o f a report in May 2004 in San Salvador; and 0 At the request of counterparts in the Secretariat o f the Presidency and in the Social Investment Fund for Local Development (FISDL), the Bank team initiated efforts to support the strengthening of El Salvador's social safety net. Phase I1o f the Bank's programmatic support also included consultations on the key findings o f the Poverty Assessment with a range o f governmental and non-government stakeholders in May, 2004, including as part o f a national workshop on growth and poverty reduction co-sponsored by the Bank and FUSADES, a well-known Salvadoran think tank. This workshop provided an important opportunity to discuss and receive feedback on the main findings and messages o f the report which, inturn, contributed to the study's further development and refinement. A third phase of the Bank's programmatic support on poverty and social policy is currently underway. The current Government of El Salvador came to office in June 2004 with a strong commitment to further poverty reduction and social progress. As part o f this commitment, the Government has launched Plan Oportunidades, a new multi-faceted social strategy with five components: (i) Fosalud, focused on extendingbasic medical services, (ii) Programa JovenES, focused on youth development; (iii)Red Solidaria, focused on increasing access to basic education, health, and nutrition services to the poorest Salvadorans; (iv) Conectate, focused on improving education quality through the harnessing o f information and communication technologies; and (v) Tu CrBdito, focused on the strengthening micro-credit as a tool for poverty reduction. As part o fthese current efforts, the GOES has sought the collaboration of the WorldBank andthe Inter-American Development Bank on the development o f the Red Solidaria and related education initiatives, among other things, to assist the Government in applying the lessons o f recent analytical work and international operational experience. Inmany ways, this current work represents a culmination o f these recent years o f collaboration and partnership with the GOES on poverty and social policy issues. The World Bank looks forward to continued collaboration on the country's efforts to reduce poverty and to improve the well-being o f the Salvadoran people more generally. ix EL SALVADOR Poverty Assessment Toward a National Social Policy Executive Summary I. Introduction More than a decade ofprogress- El Salvador experienced a major political, economic, and social transition during the 1990s. The civil war o f the 1980s came to end in 1991, and since that time regular, broad-based elections - most recently in March, 2004 - have helped lay the foundations for a stable democracy. This political transformation was accompanied inthe 1990s by structural economic reforms and stable macro policies, which in turn paved the way for the dollarization o f the economy, and for El Salvador's participation in the Central America Free Trade Agreement (CAFTA), signed in late 2003. El Salvador has registered a number of important socio-economic gains in the decade-plus since the returnto peace. The economy grew by an impressive 6 percent per year from 1990-1995 and, then, by an average rate o f 2.8 percent from 1996-2002. Moreover, economic growth, coupled with reforms in some key sectors and increases in social sector spending inthe second half o f the decade, has contributed to substantial reductions in poverty and improvements in basic socio- economic indicators over the period. Indeed, the poverty headcount has fallen by more than one- third; education levels have risen; access to basic services, such as safe water, has increased; and basic health outcomes, such as life expectancy, infant and child mortality, and child nutrition have all improved. In several areas, including inbasic education, infant mortality, and access to potable water, the gaps betweenthe poor and the non-poor have also declined over the period. -but challengesremain- Notwithstanding its achievements, El Salvador faces important challenges as it looks toward consolidating and extending these gains. For example, despite important gains, a number o f key social indicators in El Salvador - especially in education and in access to safe water - are still poorer than the average for the Latin America region and for Lower Middle Income Countries. Moreover, growth will continue to be a cornerstone o f poverty reduction in the country. Yet, slower growth in the second half o f the 1990s and the early 2000s has meant that average per capita incomes have barely grown over the last several years, observably slowing poverty reduction and socio-economic progress since 2000. While social sector spending can help to catalyze progress, low overall spending levels and tight a fiscal situation limit the scope for action in the near term. Moreover, efforts to reinvigorate growth, poverty reduction, and socio- economic progress have been complicated by several external factors -two powerful earthquakes in2001, the region's "coffee crisis,'' andthe slowdown inthe U.S.and global economies. -highlighting the need to strengthensocialpolicies- The country's recent achievements in reducing poverty and raising social indicators have also spawned a new set o f challenges for the future. For example, while growth in the 1990s was relatively broad-based - Le., incomes o f the moderate poor grew roughly at the same rate o f that X o f average incomes - the incomes o f the poorest 20 percent o f the population grew relatively little, leading to a moderate increase in income inequality in El Salvador. The fact that the poorest Salvadorans have not beenable to take advantage o f broad socio-economic progress over the last decade highlights the need for active measures oriented toward reaching the country's most vulnerable groups to support them in escaping from poverty. Because poverty is less widespread, continuing to make social gains will require more refinedand targeted measuresthan those that have increased education, improved health, and raised access to basic services in the past. This study examines the recent evolution o f poverty and social indicators inthe country, the key factors explaining recent progress, including domestic factors and the effects o f exogenous shocks, such as the 2001 earthquakes and the coffee crisis. Special emphasis i s placed on the role o f the social sectors - which in El Salvador is taken to include rural water and roads, as well as the traditional human development sectors (education, health, and social protection). Based on the diagnosis, the study outlines an approach for increasing the effectiveness of El Salvador's poverty reduction efforts - through the development o f a coherent national social policy. Indeed, the collection o fthe evidence suggests that to be most effective inreducingpoverty inthe coming years, it will be important for El Salvador to craft a social policy - based on a broad national consensus-that 0 Develops the human capital o f all Salvadorans 0 Strengthens people's access to markets and basic services, and 0 Provides support and protection to the poorest, most vulnerable Salvadoran families. -and increase investment in the social sectors The evidence also shows that in spite of recent increases, social sector spending in El Salvador remains low. By enabling the poor to participate more fully in the opportunities created by development, social sector investments are critical to poverty reduction (as well as to the growth agenda). Thus, the country needs to continue strengthening the social sectors and increasing the levels and effectiveness o f social sector spending if it is to build successfully on recent accomplishments. This, itself, raises a challenge for which developing a national consensus will be critical: how to finance the country's commitment to poverty reduction and broader improvement insocial indicators. 11. RecentAchievements, CurrentChallenges Substantialpoverty reduction, but a slowing trend Income poverty declined significantly in El Salvador during the 1990s -although progress has slowed considerably inthe 2000s in the face o f the coffee crisis, the earthquakes of 2001, and the slowdowns in the global and domestic economies. The share o f the population whose income falls below the poverty line declined by 27 percentage points from 64.4 percent in 1991 to 37.2 percent in 2002, while the fraction living in extreme poverty declined by almost 16 percentage points from 31.2 percent to 15.4 percent during the period-although poverty fell fastest in the first half o f the decade. Since 2000 total poverty has declined more slowly, while extreme poverty has barely changed. Moreover, income inequality increased slightly since the mid-1990s. Poverty continues to be disproportionately rural. About half o f Salvadorians living in rural areas are poor, a quarter o f which live in mere subsistence, while 28.5 percent o f the urban population i s poor and only 9 percent extremely poor. In terms o f numbers o f poor people, the rural poor made up about 55 percent o f all El Salvador's poor in 2002, while urban poor made up45 percent xi of the poor. Extreme poverty is particularly concentrated in rural areas. In 2002 there were nearly two extreme poor living in rural areas (137) for every one extreme poor living in urban areas. Progress in education and health has narrowedgaps- Substantial progress was also made in human development over the period. In education, not only have enrollment rates for the population as a whole increased, but the gaps betweenthe poor and non-poor have narrowed in net enrollment rates, in primary graduation rates and in average years o f schooling. Moreover, gender gaps in schooling have essentially been eliminated at the primary and secondary levels. There has been considerable progress in health as well - in life expectancy, infant and child mortality, child nutrition, and health service access. While more limiteddata exist on how health outcomes changed among the poor relative to the non-poor, there is some evidence that the gap may have closed, at least for some indicators. Data from the National Survey o f Family Health (FESAL) indicate, for instance, that the poor have made relative gains with respect to infant mortality. - a n d highlighted new challenges Despite recent achievements, twenty-two percent of 7-15 year-olds from the poorest families still did not attend school in2002. Access to secondary schooling remains low and the gap between poor and non-poor remains large; secondary enrollment rates are only about 20 percent among the poor (compared with 36 percent among the non-poor). In addition, when the poor do enroll, in school, they are less likely to continue to completion. And while the quality o f education is a system-side challenge inEl Salvador, education quality appears to be particularly low among the poor. Important challenges remain with respect to improving the health and nutritional status o f the poor as well. Basic indicators are still significantly lower among the poor than the non-poor, in spite o f advances. The poor also continue to have less access to quality health and reproductive healthcare. For example, while 80 percent o f the wealthy mothers deliver in a hospital, only 4 of 10 low-income mothers have institutional deliveries. Moreover, problems o f access have been exacerbated by un-repairedearthquake damage to hospitals and other health facilities. Despiteprogress,gaps remain in access to basic services and to markets Access to basic services also improved over the period, with increased access to safe water, sanitary installations, and electricity. Important gains were also made in reducing the isolation o f poor, rural households. Indeed, data collected as part o f a unique panel o f rural households (called BASIS) indicates that average distances and travel times to paved roads and to markets declined substantially for all rural households, and most dramatically for the poorest families. Still, further progress is needed. About halfthe poorest households still lack access to safe water, particularly in rural areas. In addition, over 20 percent o f the poorest households lack adequate sanitation facilities - with potentially important impacts on family and environmental health - and about 4 out o f 10lack access to electricity. Inspite o f significant progress indeveloping road and transport infrastructure, the poor still face greater distances and costs associated with accessing services or markets. Average distances to paved roads are still 80-90 percent higher among the extreme poor than the non-poor inrural areas, whereas average distances to the nearest market are roughly one-third higher among the extreme poor than the non-poor. xii 111. ExplainingRecentProgress What have been the key factors responsible for the broad improvements in poverty and social indicators since the early 199Os? A few factors stand out: Thepoor have benejitedfrom growth- There is now a global empirical literature showing the importance of economic growth for poverty reduction, and the same can be seen in El Salvador. The elasticity o f total poverty with respect to growth was -1.3 over the 1991-2002 period, indicating that a one percent increase in per capita income was associated with decline in the poverty headcount of 1.3 percent, holding the distribution of income constant. For the 1991-2002 period, average incomes of the poor grew by 3.1percent a year, just below the nationwide average o f 3.7 percent. But there were important differences in income growth among the poor during the period. Per capita incomes o f the poorest Salvadorans lagged during much o f the period, averaging only 1.3 percent per annum. These differences in income growth betweenthe moderate poor and the extreme poor contributed to increases inaggregate income inequality, particularly since 2000, when the poorest Salvadoran experienced real declines intheir per capita incomes. -a reduced reliance on agricultural incomes- Structural changes in employment and household earnings also affected poverty reduction over the period. Most households have multiple sources of income, but the share o f the people whose households reliedon agriculture as their main source o f income dropped dramatically - from 38.9 percent to 18.1 percent - over the 1991-2002 period. This structural shift in households' sources o f income had an important effect on poverty reduction. The incidence o f poverty among households whose main income source was agriculture declined very little between 1991 and 2002 -just over 1percentage point, from 75.3 to 74.1. In contrast, households who found other main income sources increased their well-being significantly. In fact, shifts away from agricultural earnings contributed to over 12 percent o f national poverty reduction over the period. In spite of the large structural changes, agricultural households still make up the largest single poverty group. More than one-third o f El Salvador's poor still count agriculture as their main income source, although less than one-fifth o f all households now depend on agriculture as their primary source o f earnings. In addition to households' earned incomes, remittances also played an important role inpoverty reduction over the period. Receipt o f remittances grew during the period, and the EHPM data indicate that households that receive remittances have per capita incomes that are 50 percent higher than otherwise identical households. Yet, only a minority of poor households receive remittances - e.g., only 11 percent o f households in the poorest quintile in 2002. In fact, while remittances have been vital to those who have received them, simulation analysis suggests that their aggregate impact on poverty reduction appears to have beenrelatively modest compared to growth inhouseholds' earned incomes. -and increasedattention to the social sectors- Recent improvements inpoverty and social indicators can also be attributed, inpart, to increased attention to the social sectors in El Salvador. Recent progress in education, for example, i s related to a series of education reforms designed to increase access to schooling, particularly among the poor in rural areas. In addition, there have been important increases in social sector spending. Historically, public spending on the social sectors has been very low by regional standards and inrelation to El Salvador's income level. But recent efforts have brought spending levels - and sector outcomes - closer to international benchmarks. Moreover, patterns o f public spending have tended to be pro-poor, contributing to closing of gaps in basic education, health and access to safe water. While there still i s clear scope for strengtheningthe social sectors as ... Xlll part o f a national social policy - including through further increases in spending - progress in education, health, water and rural road access has beennoteworthy. -but lost ground, at least temporarily, due to recent shocks InJanuary andFebruary, 2001, ElSalvador was hit by two powerful earthquakes that resultedin over 1,000 casualties and left more than 300,000 families homeless. The EHPM data indicate that the earthquakes did not result in an increase in income poverty at the national level - although they didhave important short-term impacts inmost-affected localities. Affected households' incomes rebounded quickly, however - thanks in part to relief, rehabilitation, and reconstruction activities - and by 2002 poverty in the affected areas had generally declined to their pre-earthquake levels. The earthquakes appear to have adversely affected basic education outcomes as well in affected areas (e.g., Usulutan, Cuscatlan). Nearly 4 percent o f children left school between 2000 and 2001 - at least temporarily. While enrollment levels have recovered somewhat, they have not rebounded as quickly as household incomes. In 2002 only about half the affected areas had attained enrollment rates at or above their pre-quake levels. Likewise, households involved in the coffee sector have been negatively affected by the coflee crisis. Although coffee households have not experienced absolute declines in their incomes (many o f them have been able to reduce their reliance on agriculture) they have experienced slower real income growth than non-coffee households. Moreover, there is some evidence to suggest that the crisis has adversely affected child nutrition in coffee growing departments (e.g., Ahuachapan). In contrast to the earthquakes, the coffee crisis does not appear to have had a systematic effect on children's schooling - although it may have negatively affected schooling in specific localities. Many poor are still unable to take advantage of economicopportunity Why were some households better able than others to take advantage o f growth and emerging economic opportunities over the past decade? Analysis o f the EHPM and BASIS surveys indicate that education, ownership o f land or other physical assets, and access to markets and basic services (via proximity to all-weather roads and transportation) were all key factors in households' ability (or not) to participate in recent economic progress. Households' involvement informal-sector salaried employment, innon-agricultural enterprises, andtheir ability to diversify their income portfolio - related, in part, to education, assets, and access to markets - were also important. Indeed, the evidence suggests that many o f the poorest, most vulnerable Salvadorans still lack sufficient human capital, as well as access to markets and basic services, to be able to advantage o f new and emerging opportunities. Ensuring that the poorest, most vulnerable citizens are equippedto participate in future economic progress will, thus, need to be at the core o f a forward-looking strategy for reducing poverty inEl Salvador. IV. Elementsof a NationalSocialPolicy Growth will remain an important engine of poverty reduction- Economic growth will continue to be a key feature o f El Salvador's accomplishments inreducing poverty. There will thus be high returns in terms o f poverty reduction to Government o f El Salvador taking the necessary measure to reinvigorate growth in the 2000s. In this context, the recent World Bank Country Economic Memorandum for El Salvador (2003) identifies four priority areas that will, if addressed, help to promote faster growth inthe country. These include: (i)increasingthe educationlevelsofthe Salvadoranpopulation; (ii) developing the country's economic infrastructure; (iii) fostering greater technology adoption and local innovation; and (iv) improving the investment climate, including efforts to reduce violence and increase the rule o f xiv law. Through their impact on growth, these measureswill be critical to further poverty reduction as well. -but enabling all Salvadoransto benefitfrom growth callsfor a national socialpolicy that- Nonetheless, the fact that the many o f the poorest Salvadorans have not participated fully in recent economic progress highlights the importance o f implementing a coherent and strategic social policy as part o f a poverty reduction strategy. Building on El Salvador's impressive achievements since the early 1990s, an effective social policy would focus on 0 strengthening the human capital o f all Salvadorans, and 0 improving people's access to markets and basic services. Because there are some among the very poor who still lack the capacity to participate in and benefitfrom traditional sector programs, an effective social policy would also 0 put inplace programsto support andprotect the poorest, most vulnerable families. -strengthens the human capital of all Salvadorans- Education. A key element o f buildingthe human capital o f all Salvadorans involves ensuring that all people have access to education with quality. Strengthening the human capital and, thus, the capabilities o f the poor will be critical to ensuring that they are prepared to take advantage o f emerging economic opportunities in El Salvador. Increasing the education levels o f all Salvadorans, including the poor, will also be important to hlfilling the growth agenda elaborated inthe Country Economic Memorandum. Indeed, there are significant complementarities between the growth and poverty reduction agendas with respect to education. It will be important for the government to focus on measures to: 0 continue to improve access to basic education among the poor 0 reduce grade repetition and drop-outs and increase completion rates at the basic level 0 introduce efforts to induce more poor grade 9 graduates into secondary schooling (as part o f a broader effort to increase secondary enrollments and achievement for all Salvadorans), and 0 improve education quality at all levels o f education. In light of the broad national consensus on education and the ongoing reform process in the sector, efforts should focus on buildingon recent progress. Special emphasis should be placed on addressing the constraints and challenges faced by the poor - among the most important o f which continue to be the direct and indirect costs o f schooling. Health. Good health is a fundamental part of people's well-being and is valued for its own sake. Sound health is also essential for people to participate fully and productively in society and inthe economy. A second key element o f building the human capital o f all Salvadorans thus involves ensuring that allpeople have access to quality healthcare. Fewer than 20 percent o f all Salvadorans have access to medical insurance either through the Salvadoran Social Security Institute (ISSS) or through private sources. Less than 10 percent and 3 percent o f the moderate and extreme poor, respectively, are insured. A major responsibility o f El Salvador's Ministry o f Public Health (MSPAS) is therefore to cover the population without xv insurance. And while there have been important advances in recent years, it i s estimated that nearly one-quarter o f the population, mostly the poor, still has limited or no access to public health services. Continuing to improve basic health outcomes among the poor will thus require enhanced efforts to ensure access to affordable healthcare among those who lack it, as well as building greater trust inthe public healthcare systemthrough enhancing the quality ofservices. Thiswill involve: 0 ensuringthat health fee waivers are made available to those who cannot afford to pay 0 assessing ongoing institutional reform and decentralization efforts to identify ways o f strengthening the outreach and quality o f the public health system - particularly at the primary level 0 reforming the health sector's human resource policy to ensure a sufficient supply o f well- trained healthcare professionals andquality care, particularly inpoor, rural areas, and 0 completing rehabilitation of the hospitals damaged by the 2001 earthquakes to ensure adequate supply o f secondary and tertiary health services, and Continuing to make progress in improving health outcomes, particularly among the poor, will require efforts on several fronts. In contrast to education, there is still no broad consensus on the direction o f health sector reforms. As such, a critical element in making sustainable progress in health will be buildinga national consensus on the way forward. -improves access to markets and services- Basic infrastructure, such as rural water and rural roads, provide both direct and indirect benefits in the fight against poverty - benefits that complement and reinforce investments in human capital development. Access to potable water and suitable sanitation facilities in rural areas i s increasingly recognized as an important health input, one that critically affects the incidence o f disease and infant and child mortality. Improved access to all weather roads translates into reduces isolation, lowers the costs for goods and services (including accessing education and health facilities), and increases access to markets, enabling people to better take advantage o f emerging economic opportunities. And, as with education, road investments make potentially important contributions to both the poverty reduction and the growth agendas. Investments to increase both access to safe water and to rural roads, will thus continue to be a key element in El Salvador's fight against poverty. Rural water. The challenge o f ensuring universal access to safe water and sanitation is made more complex by a complex institutional and regulatory environment in the sector. While the decentralized nature o f rural water provision has potential benefits, the current regulatory environment provides uneven "rules o f the game" across water sector institutions, leading to inefficiencies inthe system, considerably variability inthe quality o f water systems, and at times, exorbitant water costs to the poor who access water through gray market purchases. As such, a key policy challenge inthe water sector will be to reinitiate and complete water sector reforms. Reform efforts were initiated inthe late 1990s and then set aside, inpart, becausethe earthquakes temporarily shifted the sectors focus to system reconstruction. Many technical inputs to a reform are already reasonably developed. The incoming administration thus has the opportunity to build on these inputs to complete the reform and put the water sector on a sound footing to improve access to potable water among the poor. xvi Rural roads. Recent progress has been impressive and provides useful lessons for the way forward, particularly with respect to the importance o f road maintenance. Indeed, by ensuring longevity o f recent rural road investments, returns to rural roads maintenance are extremely high and should be considered a priority along with continued investments in road expansion. Inthis context, efforts to ensure that the poor benefitfrom rural roadinvestments should focus on: 0 continuing expansion and improvement o f the rural road network 0 developing a program for the maintenance o f secondary and tertiary road networks to ensure the sustainability o f recent investments inruralroads, and 0 continuing development o f local and municipal capacity to build and maintain rural road infrastructure. -and supports andprotects society'spoorest, most vulnerable households Building a Social Safety Net. The idea o f developing a coherent social safety net is a relatively new one in El Salvador -yet there is growing awareness o f the issue and it is now a core element inthe platform of the newly elected government. Social safety programs have an important role to play inEl Salvador's poverty reduction efforts for several reasons. 0 Inspite ofbroad socio-economic advance, the poorest, most vulnerable Salvadorans have largely not participated in or been able to take advantage o f recent progress. Continuing lack o f access to education, health, and other basic services raises the risk that these poor will be further left behind and that their weak capacity to take advantage of emerging economic opportunities will be passedon to their children via weak human capital. 0 ElSalvador is a shock-prone country. While all Salvadorans are potentially vulnerable to shocks, as the recent earthquakes showed, the extreme poor tend to be most vulnerable, both regarding income and basic education, health, and nutritionaloutcomes. 0 While most households have their own "private" safety nets - most notably migration and remittances - they tend not be sufficient, particular among the poor. Private risk management strategies onlypartially insure families against risWshocks and, in the case o f remittances, only a fraction o f those that need support most receive them. Givenits combination o f chronic extreme poverty and vulnerability to shocks El Salvador would benefit from a safety net that focuses on: (i)improving service access to the poorest, most vulnerable groups who have been outside the reach o f traditional line ministry interventions; and (ii) protecting the poorest from the most harmful impacts o f natural and economic shocks. While existing safety net-type programs are generally pertinent, focusing on key risks faced by the poorest, most vulnerable Salvadorans, their impact is modest. Coverage is low, due to low levels o f public spending. Moreover, within the existing budget envelop, resources tend to be under-allocated to priority risk groups. Insufficientinformation on which to target programs and on program impact, coupled with lack o f a strategic vision for the sector and coordination across numerous atomized programs, further constrains the impact o f existing programs, reducing the cost effectiveness and impact o f the El Salvador's safety net. To ensure that El Salvador's safety net contributes appropriately to sustainable poverty reduction, it will be important to 0 Develop a coherent, strategic vision for the sector xvii 0 Establish an institutional mechanism for coordinating programs, that would support o greater emphasis on high priority, high return areas (e.g., early childhood interventions), and o identification and scaling up of cost-effective models, and 0 Continue recent efforts to strengthen the information base for program targeting and programs' abilities to monitor and evaluate their impact. To be effective, it will be important that any institutional mechanismhave a strong mandate, the authority to coordinate across traditional ministerial boundaries (since the safety net would need to address a number o f cross-cutting issues), and command sufficient resources and technical capacity to effectively play a role. V. Strengthening Social Policy -The Way Forward A needfor increasedresources- The evidence suggests there are potentially important gains - in terms o f poverty reduction and increased economic participation o f the poor - available through strengthening national social policy. Greater productivity among the poor would also contribute to enhanced economic growth and development. To have an impact on remaining poverty will require additional resources, however - significantly so in some cases. Given that the fiscal space for increased spending is limitedin El Salvador, it will be important to make strategic about next steps and priority areas for action. -most notablyfor education and a social safety net The largest commitments of additional resources are associated with improvements in education and in the development o f a social safety net. It i s estimated that achieving universal basic education and 70 percent net enrollments in secondary schooling, while ensuring quality improvements, would require an additional 1.8 percent o f GDP in education sector spending; developing an adequate safety net would require spending o f an additional 1 percentage point o f GDP over time. Only spending on road investments, which essentially doubled between 1999 and 2003, would not require noticeable increases. While commitment o f additional resources in any single sector may appear within reach, the joint costs o f implementingthe proposed elements o f a national social policy pose a greater challenge -animpliedspendingincreaseofbetween3.2 and3.6 percent ofGDP. Inthiscontext, serious considerationshould be given to measures that would increase government revenues. But with or without additional revenue measures, some effort to consider relative priorities and phasing of policies and investments are warranted. Education and rural roads sectors arepoisedfor high-return investments- Given that additional investmentsin education and roads will likely contribute both the poverty reduction and economic growth agendas, these are likely to be particularly high return investments. Inaddition, recent developments inboth sectors - progress on the reform agenda in education, the success o f Caminos Rurales Sostenibles - suggest the sectors are poised to use additional resourcesproductively. -while other sectors would bene@tfrom stronger strategic direction With respect to health and water and sanitation, there i s less consensus about direction o f sectors or about sectoral reforms. In this sense, it will be important to build a national consensus on the ways forward, before committing substantial additional resources. Indeed, reforms that define a xviii strategic direction, including the appropriate roles for the private as well as the public sector, could pave the way to significant efficiency gains and, thus, a greater leveraging o f available resources. Inthe case of development of a social safety net, since this is a relatively new policy area inEl Salvador, it will be important to resource programs sufficiently to ensure an impact. At the same time, it may make sense to increase public resources incrementally, to ensure that an appropriate institutional framework i s developed and cost-effective service models and delivery mechanisms are identified. Indeed, phasing-in o f spending increases - to provide the Government o f El Salvador a clear and predictable time horizon for increasing its revenue base - may be a pragmatic approach to implementing multiple elements o f the country's social policy. Leveraging resources Independent o f spending levels adopted, ongoing efforts should also be made to leverage resources through creative partnerships -between public-private entities, between the Central and Local Governments, and between Salvadoran emigrants and their communities, friends and families who remain in the country. While El Salvador has made important advances inthis area recently, there remains considerable potential through such partnerships to enhance poverty reduction efforts and national development as a whole. Improving datafor enhancedpoverty monitoring andprogram targeting Finally, reaching the poorest, most vulnerable Salvadorans will require improved data and information with which to monitor poverty and target programs to at-risk groups. Several priorities stand out, including: 0 collection o f a new population census to ensure that El Salvador's poverty profile fully reflects recent demographic and geographic changes due to migration as well as other factors 0 further strengthening of the national household survey (EHPM), which together with a new census will enable more accurate identification o f the poor and better targeting o f poverty-related interventions, and 0 improving program monitoring and evaluation to identify effective service delivery models and enable adjustment o f program design (when necessary) to enhance program effectiveness. Indeed, enhancing the Government's ability to target poverty reduction programs will be central to El Salvador meeting its emerging and future poverty challenging, and will help to ensure that the country is making the most effective use o f its scarce fiscal resources in the fight against poverty. xix ELSALVADOR Poverty Assessment Towarda National SocialPolicy PART 1: POVERTYAND VULNERABILITYINEL SALVADOR,1991-2002 I. INTRODUCTION A return topeace, coupled with economic reforms- 1.1 El Salvador experienced a major political, economic and social transition during the 1990s. The civil war o f the 1980s came to end in 1991, and since that time, regular, broad-based elections - most recently in March, 2004 - have helped lay the foundations for a stable democracy. A process o f structural reforms implementedduring the 1990s also transformed the Salvadoran economy by liberalizing trade, strengthening the private sector, privatizing several state enterprises, reforming the pension system and introducing improvements in the competitiveness environment for private investment. Stable macro policies, coupled with these structural reforms, paved the way for the Monetary Integration Law approved in 2000 that established the base for the dollarization o f the economy, and for El Salvador's participation in the Central America Free Trade Agreement (CAFTA), signed inlate 2003. -has helped spur growth and reducepoverty since the early 1990s- 1.2 El Salvador has registereda number of important socio-economic gains inthe decade-plus since the return to peace. The economy grew by an impressive 6 percent per year from 1990- 1995 and, then, by an average rate of 2.8 percent from 1996-2002. Moreover, economic growth, coupled with reforms in some key social sectors and increases in social spending in the second halfo f the decade, has contributed to substantial reductions inpoverty and improvements inbasic socio-economic indicators over the period (Table 1.1). Indeed, the poverty headcount has fallen by more than one-third; education levels have risen; access to basic services, such as safe water, has increased; and basic health outcomes, such as life expectancy, infant and child mortality, and child nutrition have all improved. Inseveral areas, including inbasic education, infant mortality, and access to potable water, the gaps between the poor and the non-poor have also declined over the period. 1 I c Indicator 1990-93 2000-02 Change GDPper capita, 1990-2002(1995 28% increase Constant US $) PovertyHeadcount, 1990/91-2002 66% (DIGESTYC) 43% (DIGESTYC) 35% decline (Total, %) 42% decline Poverty(Extreme,Headcount, 1990/91-2002 %) 33% (DIGESTYC) 19%(DIGESTYC) 42% decline 3 1%(WB) 15%(WB) 5 1% decline , Net Basic EnrollmentRate, 1990/91- 74% 85% 15%increase ii2002, EHPM (%) Net SecondaryEnrollmentRate, 61% increase 1990/91-2002, EHPM (%) I I ' Life Expectancy, 1990-2002(years at 7% increase birth) 70 Infant Mortality Rate(deathsper 41 (FESAL 1993) 25 (FESAL2002) 39% decline 1,000 live births) 46 (WDI 1990) 33 (WDI2002) 32% decline Under-5 Mortality Rate(deaths per 52 (FESAL 1993) 31(FESAL2002) 40% decline 1,000 live births) 60 (WDI 1990) 39 (WDI2002) 35% decline Chronic Malnutrition (height for age, 16%decline 1 childrenunder 5), 1993-2002(%) I 23.3 19.6 I Access to Safe Water, 1990-2002(%) 55% 76% 38% increase 1.3 Notwithstanding its achievements, El Salvador faces important challenges as it looks toward consolidating and extending these gains. As can be seen in Table 1.2, despite important gains, a number o f key social indicators in El Salvador - especially in education and in access to safe water - are still poorer than the average for the Latin America region and for Lower Middle Income Countries. -but progress has slowed in recentyears, highlighting key challenges. 1.4 Moreover, growth will continue to be a cornerstone o f poverty reduction in the country. Yet, slower growth inthe second halfo fthe 1990s and the early 2000s has meant that average per capita incomes have barely grown over the last several years, observably slowing poverty reduction and socio-economic progress since 2000. While social sector spending can help to catalyze progress, low overall spendinglevels and tight a fiscal situation limit the scope for action in the near term. In addition, efforts to reinvigorate growth, poverty reduction, and socio- economic progress have been complicated by several external factors - two powerful earthquakes in 2001, the region's "coffee crisis," and the slowdown in the U.S. and global economies. The earthquakes alone caused more than 1,000 casualties, left more than 300,000 families homeless, 2 destroyed infrastructure and had considerable short-term impacts on family incomes and access to basic services in affected areas. Indicator ElSalvador LatinAmerica and Lower Middle the Caribbean Income Countries Net Primary EnrollmentRate, 2001 89 94 91 i(%) Net Secondary EnrollmentRate, 46 65 nla 2001 (??) Life Expectancy, 2002 (years at birth) 70 71 69 Infant MortalityRate, 2002 (deaths 33 28 32 per 1,000 live births) Under-5Mortality Rate, 2002 (deaths 39 34 40 per 1,000 live births) Access to SafeWater, 2000 (%) 76 86 81 ~ Thepoorest Salvadoranshave not benejZtedproportionatelyfrom recentgains- 1.5 The country's recent achievements in poverty reduction have also spawned a new set of challenges for the future. For example, while growth in the 1990s was relatively broad-based - i.e., incomes o f the moderate poor grew roughly at the same rate o f that o f average incomes - the incomes o f the poorest 20 percent of the population grew relatively little. These differences in income growth among the 1990's poor have fueled a moderate increase inincome inequality inEl Salvador. Perhaps more importantly, from a policy perspective, the fact that the poorest Salvadorans have not beenable to take advantage o f broad socio-economic progress over the last decade highlights the need for proactive measures oriented toward reaching the country's most vulnerable groups to support them inescaping from poverty. -and future success will requirefocused efforts. 1.6 Inaddition, recent social gains havebrought to the forefront more persistent, more difficult- to-address challenges. For example, the considerable declines in child mortality over the last decade have come largely through declines in so-called "soft" mortality, that is, through reductions inpost-neonatal and 1-4year old mortality. This has resulted in an increasing share o f the persistent "hard" - or neonatal - mortality, a more difficult health sector challenge.' Similarly, bringing and keeping children from the poorest, most vulnerable households in the education system will likely require more refined and targeted measures than those that have succeeded inincreasing enrollments in the past. In sum, recent progress in reducing poverty and improving social indicators has created new challenges that will require increased and deliberate efforts -and creativity -to ensure ongoing success. This report uses a multi-dimensional approach to analyzingpover@- 1.7 This Poverty Assessment focuses on poverty, in its multi-dimensions, and on how social policy along with increased investments inthe social sectors can be usedby the Government o f El Salvador (GOES) to accelerate poverty reduction inthe coming years. To do so, this study brings 'Thisi s true not only inEl Salvador, but throughout CentralAmerica. 3 together several newly commissioned background studies along with a rich body o f recent empirical analysis on poverty and social policy in El Salvador. This work has also benefited immensely from ongoing discussion and consultations with government counterparts, members o f non-governmental think tanks, and civil society groups. -highlighting the roles of socialpolicy- 1.8 This study examines the recent evolution ofpoverty and social indicators inthe country, the key factors explaining recent progress, including domestic factors and the effects of exogenous shocks, such as the 2001 earthquakes and the coffee crisis. Special emphasis i s placed on the role of the social sectors - which in El Salvador i s taken to include rural water and roads, as well as the traditional human development sectors (education, health, and social protection). Based on the diagnosis, the study outlines an approach for increasing the effectiveness o f El Salvador's poverty reduction efforts -through the development o f a coherent national social policy. Indeed, the collection o f the evidence suggests that to be most effective inreducing poverty inthe coming years, it will be important for El Salvador to craft a social policy - based on a broad national consensus-that: 0 Develops the human capital of all Salvadorans 0 Strengthens people's access to markets and basic services, and 0 Provides support andprotectionto the poorest, most vulnerable Salvadoran families. -and public spending- 1.9 The evidence also shows that in spite of recent increases, social sector spending in El Salvador remains low. By enabling the poor to participate more fully in the opportunities created by development, social sector investments are critical to poverty reduction (as well as to the growth agenda). Thus, the country needs to continue strengthening the social sectors and increasing the levels and effectiveness o f social sector spending if it is to build successfully on recent accomplishments. This, itself, raises a challenge for the country about which developing a national consensus will be important: that is, how to finance the country's commitment to poverty reduction and broader improvement in social indicators as part o f its development strategy. -as well as the needfor better data. 1.10 Finally, meeting the challenges associated with further reducing poverty and improving social indicators will require an improved database upon which to base policy decision. This includes collection o f a new population census and continued improvements in the national household survey instrument, steps that will be vital to identifying the poor, targeting poverty reduction policies and investments, and ensuring that they are effective. Sharpening the Government's ability to target anti-poverty programs is particularly important to ensuring that El Salvador is able to make the most o f its scarce fiscal resources. 1.11 This study is organized as follows. Chapter 2 analyzes the evolution o f poverty and vulnerability inEl Salvador from 1991-2002. Chapter 3 examines the factors that help to explain the evolution of poverty and other socio-economic indicators, including economic growth, recent structural changes in the economy, shocks such as the 2001 earthquakes and the coffee crisis, as well as other factors, such as recent increases inremittances. Chapters 4-8 examine, inturn, key social investments- in education, health, rural infrastructure, such as water and roads, and social protection - their prospective roles in attacking poverty in the coming years, and the likely fiscal costs of the requisite investments. Chapter 9 summarizes the lessons o f the analysis for crafting a national social policy to accelerate poverty reduction and social progress inEl Salvador. 4 11. RECENTACHIEVEMENTS, CURRENT CHALLENGES 2.1 This chapter focuses on the evolution o f poverty, in its multi-dimensions, from 1991-2002, o n vulnerability, and on the key remaining challenges facing El Salvador in its poverty reduction agenda. Several key messages emerge from the chapter. First, El Salvador experienced significant declines inpoverty during the 1990s- althoughprogress has stalled inthe 2000s inthe face of the coffee crisis, the earthquakes of 2001, and the slowdowns in global and domestic economies. The finding o f significant poverty reduction is robust to different the use of different data sets and different methodological approaches. Second, despite broad socio-economic gains for the "moderate" poor, the poorest Salvadorans tended not to share fully in the income and socio-economic gains of other groups. Enabling this group to escape poverty thus remains a critical challenge for poverty reduction in El Salvador. Third, in spite of broad socio-economic gains, many Salvadoran households remain vulnerable to at least temporary spells of poverty, due to a number of factors, including the prevalence of economic and natural shocks. This vulnerability, again, frames an important policy challenge for El Salvador moving forward. Theevolution of uoverQ A significant reduction inpoverty in the 1990s- 2.2 An analysis of El Salvador's national household survey in 1991, 1995, 2000, 2001 and 2002 indicate that poverty declined significantly in El Salvador during the 1990s - although progress has slowed considerably in the 2000s in the face of the coffee crisis, the earthquakes of 2001 andthe slowdowns inglobal and domestic economies (Box 2.1). Box 2.1: Challenges of Poverty Measurement inEl Salvador A key objective o fthis Poverty Assessment is to generate a consistent poverty profile for El Salvador over the 1991-2002 period, usingthe country's national household survey (EHPM), and income and expenditure survey. The EHPM surveys have national coverage and are statistically representative for urban and rural areas as well as at the Department level. The 2001 survey included an over-sampling o f households in the municipalities most affected by the earthquakes, and over-sampled departments are statistically representative. The EHPM sample sizes are generally around 13,000 households (except for the 2001 over- sample, which is almost twice that). The EHPM collects detailed data on household incomes but not on household consumption (now considered international best practice for poverty measurement). In the absence o f detailed consumption data with which to build a welfare aggregate, the analysis here relies on household income estimates generated from the EHPM data. The analysis also uses the Government o f El Salvador's official poverty lines, based on the standard methodology o f minimumcaloric intake. The extreme poverty lines correspond to the estimated monthly cost o f basic food baskets that provide a minimum caloric requirement (about 2,200 KcaVday) for a family o f four. The total poverty lines are obtained by multiplying the extreme lines by a factor o f two (the inverse o f a long-standing estimate o f the share o fnon-food consumption) to allow for non-food expenditures. The official poverty lines use different consumption baskets for urban and rural areas to achieve the same minimumcaloric requirement. The cost o f these food baskets is updated yearly. In2002, the value o f the per capita extreme poverty line was estimated at $277.9 and $173.6 colones per month for urban and rural areas, respectively, up from an estimated $202.2 and $119.1 colones per month in 1991. Official poverty numbers published by DIGESTYC, the Salvadoran statistical agency, are household level figures. A household is considered poor if its total per capita monthly income is below the total poverty line, and extremely poor if its per capita income is below the extreme poverty line. The analysis presented here departs from the official approach in two ways. First, the poverty measures reported here are for individuals rather than households, consistent with international good practice. Second, the income aggregates used here are more comprehensive than the ones used by DIGESYTC. This is because they 5 include in-kind income and other components o f household non-monetary income that are particularly important for rural agricultural households; when necessary, standard imputations are also performed to enhance the comparability of the poverty estimates over time. The income aggregates used inthis Poverty Assessment bring the income measures closer to per capita consumption, approximating to the extent possible the preferredmeasure o f well being. While methodological differences yield poverty estimates that differ slightly from official figures, observed poverty trends are similar. Nonetheless, some caution i s warranted in interpreting the poverty numbers, as consistent poverty measurement in El Salvador faces significant data challenges. First, since the most recent census dates from 1992, recent household surveys may not reflect accurately the changing demographic and geographic composition o f the population, especially given the substantial internal and international migration over the last decade. This problem is hard to address ex-post and could potentially bias the estimates o fpoverty and social indicators. Second, there are important differences in the survey instruments from the beginning to the end o f the 1990s that make decade comparisons o f poverty and inequality figures difficult. Specifically, non- monetary income is not captured in the early 1990s surveys, so survey income measures tend to overstate the true level ofpoverty at the beginning of the period as well as overstate actual declines over the 1990s. Systematic income data adjustments are thus performed on the earlier surveys to enhance comparability.' Finally, the official poverty lines used in El Salvador are not based on a detailed consumption aggregate and keep the value o f the non-food portion o f the general poverty line fixed relative to the food basket. At any point intime these may not reflect accurately the consumption basket (food and non-food expenditure) o f a meaningful minimum subsistence level. Indeed, the real value o f the official poverty lines have declined in recent years, although there i s not obvious upward tendency in households purchasing power. While the extreme urban and rural poverty lines increased by 36.3 and 44.7 percent during 1992-2002, the food and non-food components o f the consumer price index (CPI) rose by 72 and 62 percent, respectively. Moreover, while it is common practice to have poverty lines cross-regional price variation, especially in large countries, the use o f different food baskets for urban and rural areas is harder to justify on welfare grounds. The value o f the urban basket i s about 60 percent higher than the value o f the rural basket (down from 70 percent in 1991), essentially reflecting consumption o f higher quality foods in urban areas. This may lead to inaccurate estimates of national poverty and overstate the gap between poverty in urban and rural areas. A variety o f adjustments and simulations have been carried out to ensure the robustness o f the analysis to these problems. In this context, it i s important to note that while specific point estimates o f poverty are sensitive to the measurement issues described above, the finding o f a significant poverty reduction in El Salvador between 1991-2002 remains robust. 2.3 The share of the population whose income falls below the poverty line declined by 27 percentage points from 64.4 percent in 1991 to 37.2 percent in 2002 (Figure 2.1). The share of people living in extreme poverty declined by almost 16 percentage points, from 31.2 percent to 15.4 percent, during this period. The incidence of poverty and extreme poverty declined by 13 and 5 percentage points, respectively, between 1995 and 2002 due to the slowdown in economic growth andthe adverse shocks that hitthe country. Specifically, non-monetary incomes are imputed based on hedonic regression models estimated for 2002 and applied to household characteristics inthe 1991 survey. 6 Figure 2.1: The Evolution of Total and Extreme Poverty, National Level, 1991-2002 70 60 50 E 40 Q, p" 30 2 20 10 0 I 1991 1995 2000 2001 2002 -but the rate ofprogress has slowed- 2.4 Poverty continued to decline during the early 2000s - although by only two percentage points, while extreme poverty remained unchanged. This i s remarkable in light o f the slowdown in domestic and world economic growth, the 2001 earthquakes and the steep decline in world coffee prices. These results are in line with other studies with poverty estimates for El Salvador covering the 1990s. 2.5 Overall, poverty reduction has been larger in urban than in rural areas (Table 2.1). The incidence of total urban poverty fell by 30 percentage points (16 for extreme urban poverty) during 1991-2002, while total ruralpoverty declinedby 20 points (12 for extreme ruralpoverty). About two thirds o f the decline in urban poverty occurred in the first half o f the 1990s. Extreme urban poverty has changed little since the mid-1990s. Incontrast, after the mid-1990s poverty has fallen more rapidly in rural than in urban areas. Total rural poverty fell by 15 percentage points between 1995 and 2002. - a n d , despiteprogress,poverty remains morepronounced in rural areas. 2.6 By 2002, about 37 percent o f Salvadorans were poor and 15 percent were extremely poor. Poverty continues to be disproportionately rural. About half o f Salvadorans living in rural areas are poor, a quarter o f whom live in mere subsistence, while 28.5 percent o f the urban population is poor and only 9 percent extremely poor. In terms o f numbers o f poor people, the rural poor made up about 55 percent o f all El Salvador's poor in 2002, while urban poor made up 45 percent o f the poor. Moreover, extreme poverty remains concentrated in rural areas. Indeed, in 2002 there were nearly two extreme poor living in rural areas (1.87) for every one extreme poor living inurban areas. For example, a recent Government of El Salvador initiated review of progress toward the Millennium DevelopmentGoals (MDGs) finds similar patterns of poverty reduction, using both the official, national poverty line and US $l/day poverty measures (Gobierno de El Salvador 2004). See also ECLAC (2003), Szekely et a1(2002), Wodon et a1(2002), SegoviaandLarde(2000). 7 ' I 1991 1995 2000 2001 2002 Total Poverty National 64.4 II 50.6 Ii 39.6 38.9 37.2 ~ Urban 58.7 38.9 28.5 I 29.3 1 I 28.5 1 ~ ~ I I Rural 69.5 64.8 55.2 52.4 49.8 ~ Extreme Poverty National 31.2 20.5 15.9 15.7 1 15.4 Urban 25.1 11.6 8.1 9.1 9.1 I Rural 36.6 31.2 26.8 25.1 24.5 2.7 Overall, other income poverty measures show similar trends (Table 2.2). Both the poverty gap (PI), which measures the average distance o fthe poor from the poverty line, and the severity ofpoverty measure (P2), which puts higher weights on the extremely poor, declined by almost half during 1991-2002. The depth and severity o f rural poverty has improved much less than that o f urban poverty over the same period. Moreover, progress in reducing the depth and severity o f poverty has slowed considerably since 2000; whereas the national poverty gap declined from 17.0 to 16.5 between 2000 and 2002, the severity o fpoverty measure remainedunchanged. I 1991 1995 2000 2001 2002 ~ ~ 'Poverty Gap (PI) National 29.8 22.5 I 17.0 I 17.0 I 16.5 I ~ ~ 1 Urban Rural 33.7 32.0 26.6 25.3 24.2 Severity of Poverty (P2) I 1 National I 17.7 1 13.6 1 10.0 1 10.2 I 10.0 1 1 Urban 14.7 7.7 5.3 1 6.1 6.2 Rural 20.5 20.6 16.6 16.0 15.5 8 Measured trends inpoverty reduction appear robust 2.8 There has been some discussion recently within El Salvador about whether observed declines inpoverty are robust to changes inthe underlying assumptions o f poverty rneas~rement.~ O f particular concern is the fact that the real poverty line has grown more slowly than the CPI, apparently reflecting differences in price changes in the basic consumption basket o f the poor relative to the average Salvadoran's consumption basket. Yet, comparison o f poverty trends using different methodological approaches, as well as different data sources, suggest that the observed declines inpoverty are indeed real. 2.9 Sensitivity analysis o f the poverty estimates were undertakenusing EHPMdata from 1991, 1995, 2000, 2001 and 2002, using alternative measures o f household income and different methods o f inflatingthe poverty line over time. As a way o f summarizing this analysis Figure2.2 shows alternative poverty trends for El Salvador using: (1) DIGESTYC's poverty estimates, using their traditional income aggregate and the official poverty line; (2) the poverty estimates presented above, using an enhanced income aggregate and the official poverty line (see Box 2.1); and (3) poverty estimates using the enhanced income aggregate and the 1991 official poverty line inflated using the general consumer price index (CPI). As can be seen from the figure, while there are differences in the point estimates o f poverty, the declining trends in poverty are generally similar across measures. Figure 2.2: Sensitivity o f the 1991-2002 Poverty Decline to Different Measurement Approaches 70 60 0 b n 1 c) 50 40 30 20 1991 1995 2000 2001 2002 --tDlGESlYC +World Bank, Official Line World Bank, CPI Adjusted Line 2.10 Indeed, analysis of rural poverty using an independent data set suggests that declines in rural poverty may have been even greater than those measured by the EHPM. Researchers at FUSADES have analyzed a unique panel data set called BASIS that tracks incomes o f the same rural households in 1995, 1997, 1999 and 2001.5 The BASIS data collect information on the incomes o f rural households in greater detail, and thus likely with greater precision, than the EWM in the sense that it better captures in-kind and other forms self-employed andor See, for example, UNDP 2003. See Beneke de Sanfeli6 and Shi (2004). 9 agricultural income. The FUSADES researchers found that between 1995 and 2001 rural poverty declined by between 23.1 and 28.6 percentage points depending on how the poverty line was deflated (Figure 2.3); extremepoverty was found to decline by similar magnitudes. Figure 2.3: RuralPoverty Trends, 1995-2001, Using BASIS Panel Data and Alternative Poverty Lines 80 70 60 50 C 40 0) 30 20 10 0 1995 1997 1999 2001 +Total Poverty, Official Line -a- TotalPoverty, CPIAdjustedLine Extreme Poverty, Official Line Extreme Poverty, CM Adjusted Line 1I Source: Beneke de Sanfelid and Shi 2004 Progress in human development Improving education outcomes- 2.11 Substantial progress was also made in human development over the period. For example, net enrollment rates in basic education of 7-15 year old children rose from about 74.3 percent in the 1991 to 85.1 percent by 2002 (Figure 2.4). The improvement inbasic net enrollment rates has beenfaster inthe worst-off regions and income groups, demonstrating some convergence inbasic education enrollments across localities and across poor and non-poor households. Indeed, the gap inenrollments between children inthe poorest families and those inthe top income quintilehas narrowed from 22 percentage points in 1991 to 14 percentage points in 2002. Moreover, El Salvador has essentially succeeded in attaining gender equality in education at the basic and secondary levels.6 2.12 Nonetheless, several important challenges exist with respect to education. For example, 22 percent o f 7-15 year-olds from the poorest families still did not attend school in 2002. Access to secondary schooling remains low and the gap between the poor and non-poor remains large. At all levels of education, when the poor do enroll in school, they are less likely to continue until completion. Moreover, the quality of education in El Salvador remains a challenge, particularly with respect to education among the poor. EHPMdata indicate that though 22 percent ofwomen hadno schooling compared to 20 percent ofmenin 2001, for girls age 6-14 years, 31 percent hadno schooling compared to 32 percent o f boys o fthe same age. Also, the same proportion of men and women, 21.4 percent, had completed the first 6 grades o f schooling in2001. 10 Figure2.4: Net EnrollmentRates in BasicEducationin ElSalvador, by Area and Income Quintiles (% children,cohort 7-15 years) c 100 80 81991/92 60 .I995 40 02000 20 2002 c 0 Urban Rural National 81991192 H1995 02000 2002 1st. Q. 2nd. Q. 5th. Q. Source: World Bank staff based on EHPMsurvey data. 2.13 As in education, there was considerable progress in health since the early 1990s. Life expectancy increased from 65.6 years in 1990 to 70 years in 20017 And as shown in Table 1.1, infant mortality declined and under-5 mortality declined substantially during the period. Chronic malnutrition ("stunting") also declined, albeit more slowly, during the period; and marginal improvements were seen inthe incidence o f underweight children under 5. 2.14 Although data suggest that access to reproductive health care, as measured by the proportion o f births attended by trained healthcare personnel grew from 51to 69 ercent between the 1988-1993 to 1997-2002 period, trends in maternal mortality are not clear! According to data from the National Survey o f Family Health (FESAL), the maternal mortality rate declined from 158 to 120 (deaths per 100,000 live births) between the 1983-1993 and 1988-1998 periods, but then rose again to 172 during the 1993-2002 period. But special care needs to be taken in interpretingthis data, as the maternal deaths are relatively rare events and the confidence intervals for these statistics makes it unclear whether real trends can be observed from the data. Data suggestthat in2003 HIV/AIDS affected about 0.7 percent o fthe pop~lation.~ -and gains in health- 2.15 While detailed data on health outcomes are not available from the EHPM surveys, the FESAL surveys suggest that the poor made relative, as well as absolute, gains in at least some health indicators. Inthe case o f infant mortality, for example, the gap between the poor and non- WorldDevelopmentIndicators(WDI) database. National Survey of Family Health, or La Encuesta Nacional de SaludFamiliar (FESAL), various years. WorldDevelopmentIndicators(WDI) database. 11 poor has declined somewhat (Figure 2.5). , lo Figure 2.5 Infant Mortality per Socio-Economic Group I Low Income Middle Income High Income I Source: Marques (2004). -but challengesremain, both in terms of service access and quality. 2.16 Nonetheless, as in education, important challenges remain with respect to improving the health and nutritional status o f the poor. Important gaps still remain in health status betweenthe poor and the non-poor. The reported incidence o f illness is higher among the poor than the non- poor. And the poor continue to have inferior access to quality health care than the non-poor. Indeed, addressing the challenges o f human development - both education and health - for the poor will be a critical part o f a national strategy to reduce poverty. These challenges are addressedinmore detail inPart 2 o f this report. Imurovements in access to basic services Thepoor have benejitedfrom increasedaccess to basic services- 2.17 The EHPM data suggest notable improvements in access to basic services (Figure 2.6). Water access reached 76 percent o f households in 2002 up from 55 percent in 1991, including an increase o f 3 percentage points inthe last 3 years. Water access has almost doubled inrural areas (almost half o f the population now has access), and again has been relatively larger for the worst- off localities and the poorest families. Access to sanitary installations has also expanded consistently to reach 93 percent o f households in 2002 up from 78.1 percent in 1991, particularly in the rural areas (from 59.2 to 84.3 percent). Almost all urban households have access to some form o f sanitary installation. Progress is also quite visible in the access to electricity that now reaches 87.5 percent o f households (18 percentage points gain over the period). While access i s almost universal inurban areas, three out o f ten rural households still lack electricity. -but significant numbersstill lack water, electricityand adequate sanitation. 2.18 At the same time, there is still much room for improving access o f the worst-off population loIt i s important to note that the FESAL publications (and website) use a different methodology for analyzinghealthoutcomes across different socio-economicgroups than i s usedinthe analysis ofthe EHPM data. Analysis of the EHPMdata generally focuses either on comparingoutcomes across the poor andnon- poor, measured against the official Salvadoran poverty line, or on examining outcomes across the five income quintiles of the population. In the case of the FESAL, outcomes are differentiated across three different socio-economic levels: "low", "medium", and "high". 12 to these basic services, particularly with regards to quality and reliability. By 2002 about half o f the households in the poorest quintile lackedaccess to water compared to access rates of over 90 percent among those inthe richest quintile(Figure 2.6). Over 20 percent o f families inthe poorest quintile lack adequate sanitation facilities, while about 4 out of 10 inthe poorest quintile still lack access to electricity. It should also be kept inmindthat these figures mask significant differences inthe quality of access. Figure2.6: Access to(% households) in El Salvador, 1991- Basic Services Sanitation (Oh) Electriclty 100 80 100 60 80 60 40 40 20 20 0 0 Urban Rural National Urban Rural National Urban Rural National ( 0 4 Water by Quintiies Sanitatlon by Qulntiles 1%1 Electricity by Qulntiles 100 120 80 100 80 80 60 40 40 20 20 0 0 1st Q 2nd (1 5th Q lrst Q 2nd Q 3rd Q Irst Q 2nd Q 3rd Q bl1991/92 HI995 02000 HZ002 Source World Bank staff based on EHPMsurvey data 2.19 In 2002 about 15 percent o f households got their water from a non-private source (e.g., a public tap/well or neighbors) and 45 percent (78 percent in rural areas) used latrines, some o f which are not hygienically adequate. While quality of housing has improved, the challengesin rural areas remain significant. 2.20 Finally, with economic progress there has been an upgrade inthe quality o f housing. About one-fifth of homes have seen an improvement inthe quality of walls or floors inthe last ten years. The fraction o f dirt floor homes, for example, has declined to 23 percent. Roughly 68 percent of homes are built with pure or mixed concrete (no adobe or bahareque). The biggest needs remain in rural areas where still between 40 and 50 percent of homes has dirt floors or are built with unreliable materials. 2.21 As in the case o f human development, access to basic services, like water and sanitation, as well as the presence o f critical socio-economic infrastructure, like roads (that connect the poor to markets, services and economic opportunity) will also play an important part o f a strategy to reduce poverty in El Salvador. Like the challenges o f human development, issues o f basic socio- economic infrastructure, specifically related to rural water and roads, are discussed in greater detail inPart2 of the report. 13 Income inecrualitv in El Salvador Income inequality appearsto have increased between 1991 and 2002- 2.22 The data presented above on human capital and on access to services indicate that there has been some convergence between the poor and non-poor in a number of non-monetary indicators of well-being between 1991 and 2002. The same does not appear to be true, however, in the case o f income. Indeed, El Salvador is among several LAC countries where income inequality appears to have increased during the last decade (Gasparini 2003). On average, income inequality is higher in Latin America than in other regions o f the world. And income inequality in El Salvador is now close to the regional average (Figure 2.7). Figure2.7: GiniCoefficientsinLAC (Distribution of equivalizedhouseholdincomecirca 2000) t I n 50 n Source: Gasparini (2003). 2.23 In 2002 the poorest 20 percent o f households received 3 percent o f income down from 3.7 percent in 1991, while the middle (third) quintile increased its participation slightly (from 12.4 to 12.8 percent) and the other quintiles saw little change in their income shares (Figure 2.8)" "GovernmentofElSalvadorestimatesofthequintilesharesofincomesuggestthat,ingeneral,theshare o f income going to the poorest 20 percent ofthe population has tendedto be relatively stable, rangingfrom 2.68 to 3.69, and averagingroughly 3.0 over the 1991-2004 period. These estimatessuggest that the income share going to the poorest 20 percenthas rebounded slightly since 2002, reaching 3.4 in2004 (data fromSecretaria Te'cnicade la Presidencia,Government of El Salvador). 14 Figure2.8: Distributionof HouseholdPer CapitaIncome inElSalvador, 1991-2002 (% Shares ofIncomeQuintiles) 7 3.7 7 0 1991192 _ _ _ _ I 3.0 _I I .u 2002 ~~ .4 2.8 Ellrst. Q. m2nd. Q. 03rd. Q. 55.4 84th. Q. 55.7 m5th. Q. 7 Note: 1991figures inclu non-monetary income estimations. Source: World Bank staff basedon EHPMsurvey data. 2.24 Inequality, measured by the gini coefficient, also increased from 1991 to 2002 (Table 2.3). The gini coefficient of the distribution o f household per capita income went up from about 0.51 to 0.52 while the gini for monetary incomes (Le., excluding self-consumption and in-kindincome) increased from 0.503 to 0.52 over this period.'* Table 2.3: Incomeinequalityin ElSalvador Gini coefficients of HouseholdPer CapitaIncome, 1991-2002 1 1991/92 I 1995 I 1999 I 2000 I 2001 I 2002 National Urban 0.509 0.518 0.507 0.523 0.519 0.519 Rural 0.474 0.463 0.454 0.473 0.471 0.473 0.476 0.444 0.441 0.449 0.466 0.464 Monetary income WB PA LAC Flagship inequalityreport ] I 0.503 I 0.493 I 0.499 I 0.510 I 0.505 1 0.520 0.537 0.512 0.524 0.532 0.529 0.525 Labor income I 11 II 11 II I1 *1991/92figures 0.536 0.509 0.517 0.524 0.524 0.530 include non monetary income estimations Note: SeeAnnex 1 for the differentcomponents included in monetary income. The Flagship definition o f monetary income includes: labor income (salary and extra hours from primary and secondaryjobs), income from independentjob (excluding agricultural profits, national remittances, rents and pensions. Labor income: includes salary and extra-hours from primary and secondaryjobs, and income from independentjobs. Source: World Bank staff basedon EHPMsurvey data. -driven by changes since 1995. 2.25 This rise in income inequality was not uniform throughout the decade. Economic growth in the first half o f the 1990s was accompanied by lower income inequality, while growth during the second half o f the decade was inequality-increasing. The gini o f per capita monetary income ''The distribution of per capita household income deteriorated in the early 1990s and remained stable thereafter. However, changes in the inequality o f household per capita income should be viewed with caution due to the imputationo fnon-monetaryincomes inthe early 1990s surveys. 15 declined in 1995 but then increased in 2002. Labor income inequality shows a similar trend. The rise in monetary income inequality during the 1990s mainly results from greater urban-rural income disparities over time, since overall the gini coefficients in urban and rural areas were generally the same at the beginning and end o f the period. As with inequality at the national level, however, both urban and rural income inequality (for all income) first declined and then increased over the course o f 1991-2002.l3 Vulnerability to povertv Most rural Salvadoransface considerable income volatility and at least temporarypover- 2.26 The existence o f the BASIS data enables analysis o f the volatility o f household income, which is not possible using repeated cross-sections from the EHMP. In their analysis o f the BASIS data, Beneke de SanfeliG and Shi (2004) find considerable income volatility in rural El Salvador. For example, over the 1995-1997-1999-2001 period, about 80-85 percent o f households in the BASIS survey moved at least one decile upward or downward from period to period. And between 30 and 45 percent o f households moved at least more than 2 deciles from one periodto another. In terms o f magnitudes o f income changes, roughly two-thirds o f the rural households inthe sample and nearly three-quarters o f the poor experienced a fall o f more than 30 percent of their real income from one period to the next over the entire period (Beneke de Sanfeli6 and Shi 2004). -and one infour was chronicallypoor between 1995 and 2001. 2.27 This income volatility has led to considerable movement in and out o f poverty over time, although there is a significant core o f rural households - about 25 percent - that are chronically poor. At the same time, roughly 62 percent of households in the BASIS survey experience temporary bouts o f poverty (Table 2.4). Only 13 percent o f the sample didnot have incomes that fell below the poverty line during the entire 1995-2001 period. In fact, the analysis suggests that during the 1999-2001 nearly 89 percent o f the rural sample had a greater than 50 percent chance of being poor for at least one period over the subsequent five years - although the analysis also shows that the percentage o f households that are vulnerable to poverty has declined since 1995 (Beneke de Sanfeli6 and Shi 2004). Table 2.4: Chronic and Transient Poverty in Rural El Salvador, 1995-2001 PercentwithAverage PercentwithAverage Percentof Per CapitaIncomes Per CapitaIncomes Classification Households Belowthe PovertyLine Belowthe PovertyLine ChronicPoor 25.1 25.1 - TransientPoor 61.9 27.9 33.9 Poor 3 o f 4 Periods 24.8 21.1 3.8 Poor 2 o f 4 Periods 19.7 6.0 13.8 Poor 1o f 4 Periods 17.2 0.9 16.4 Non-Poor 13.1 13.1 Percent o fHouseholds 100 53.0 I 47.0 l3 Both Beneke de Sanfelid and Shi (2004) and Rodriguez-Meza and Gonzalez-Vega (2004) estimate gini coefficients for rural El Salvador, using the BASIS data and "full" income definitions that are very close to those estimated using the EHPM. For example, for 2001 Beneke de Sanfeliu and Shi find a rural gini o f 0.467, while Rodriguez-Meza and Gonzalez-Vega find a rural gini o f 0.478. Inboth studies, inequality is found to rise slightly between 1995 and 1997, declining slightly thereafter. 16 Moreover, thepoor are most vulnerableto income shocks. 2.28 Moreover, Rodriguez-Meza and Gonzalez-Vega (2004) find, analyzing the BASIS data, that although the rich in rural areas recover very quickly from an income shock, it takes the poor a much longer time to recover. This is consistent with a broad empirical literature on household consumption and income smoothing, including evidence from other Central American countries, that shows that the poor have a more difficult time protecting (or "self-insuring") themselves from income shocks.14 This i s not completely surprising. On one hand, the rural poor do not have extensive access to formal insurance mechanisms inEl Salvador. And while remittances are important, the high volatility o f income and the presence o f considerable transient poverty suggest that informal insurance arrangements are insufficient to protect the rural poor from shocks. This suggests the need to strengthen the country's social safety net, both to protect the poor from shocks and to enable them to pull themselves out o f poverty. Conclusion 2.29 This chapter has examined the evolution o f poverty, in its multi-dimensions, from 1991- 2002. Review of the recent trends inpoverty and social indicators highlightseveral things. e El Salvador experienced significant declines in poverty during the 1990s - although progress has stalled inthe 2000s inthe face o f the coffee crisis, the earthquakes o f 2001, and the slowdowns in global and domestic economies. e Despite broad socio-economic gains for the "moderate" poor, the poorest Salvadorans have not sharedfully inthe income and socio-economic gains o f other groups. e Moreover, even if they are not chronically poor, a significant proportion o f Salvadoran households remain vulnerable to at least temporary spells o f poverty. 2.30 Enabling the poorest and most vulnerable Salvadorans to participate in economic progress, and through this, to escape poverty thus remains a critical challenge in El Salvador in spite o f recent progress.l5 To effectively address this challenge, it is important to understand the main factors that affect poverty reduction - factors such as economic growth, structural change in the economy, investments in human capital, and measures that increase people's access to markets and basic services. These issues are explored indetail inthe chapter that follows. l4For evidence on household "self-insurance" (consumption smoothing) in Latin America, see Skoufias (2002) on Mexico; Klugman, Kruger, and Wither (2003) and Kruger, Mason, and Vakis (2003) on Nicaragua; and Glewwe and Hall (1998) on Peru. For global evidence, see Skoufias and Quisumbing (2002, various), Morduch (1999, various), Jalan and Ravallion (1999, China), Ravallion and Chaudhuri (1997, India). l5The recent Government of El Salvador sponsoredreview of the MillenniumDevelopmentGoals (MDGs) similarlyhighlights that while there hasbeensignificant povertyreductionandsocio-economic progressin El Salvador since the early 1990s,the countrycontinuesto face importantchallengesinraisingthe welfare of all Salvadorans(Gobiemo de El Salvador 2004). 17 111. EXPLAINING RECENTPOVERTY REDUCTION 3.1 What has been responsible for the broad improvements in poverty and social indicators since the early 1990s? This chapter uses new empirical analysis from the EHPMand BASIS data sets to try to explain the key factors that influenced poverty reduction since 1991. The chapter investigates several main areas, including: (1) the effects o f economic growth; (2) the effects o f structural changes in the economy over the period; (3) the factors that affect household incomes and income growth; and (4) the role of growth o f households' own-earned income vs. receipt o f remittances. This chapter also examines the impact o f recent shocks, such as the earthquakes o f 2001 and the coffee crisis, on poverty and key social indicators ineducation and health. 3.2 Several key messages emerge from this chapter. First, growth was an important force for poverty reduction in the 1990s, as were structural changes in the economy. Indeed, growth was reasonably broad-based and effective in reducing poverty. Nonetheless, not all households benefitedequally from growth duringthe period. While the incomes of the moderately poor grew roughly at the rate o f average per capita income growth, the poorest Salvadorans experienced relatively little growth. Why were some households better able than others to take advantage o f growth and emerging economic opportunities over the past decade? The analysis suggests that education, ownership o f land or other physical assets, and access to markets and basic services (via proximity to all-weather roads and transportation) were all key factors inhouseholds' ability (or not) to participate in recent economic progress. Yet, many o f the poorest, most vulnerable Salvadorans still lack sufficient human capital, as well as access to markets and basic services, to be able to advantage of new and emerging opportunities. The collection o f evidence thus suggests that ensuring that the poorest citizens are equipped to participate in future economic progress will be important to a forward-looking strategy for reducing poverty inEl Salvador. Growth and uovertv reduction Economicgrowth has been an important driver inpoverty reduction- 3.3 There is now a large global empirical literature showing the importance of economic growth for sustained poverty reduction (e.g., Dollar and Kraay 2002; Shorrocks and van der Hoeven 2004). The same i s demonstrably true in El Salvador. Indeed, new analysis o f the EHPMdata indicates that bothper capita income growth, and its distribution, has been important to progress in the country between 1991 and 2002. Table 3.1 reports the empirical estimates o f (1) the elasticities o f extreme and total poverty with respect to income and (2) the elasticities of extreme and total poverty with respect to inequality as measuredby the Gini coefficient. Table 3.1: Elasticities of Poverty Reduction with Respectto Per Capita Income Growth and to Changes in Income Inequality, 1991-2002 Extreme Moderate Elasticity with respect to: Poverty Poverty Total Poverty Per Capita Income -1.5 -1.1 -1.3 Inequality (ie., gini coefficient) 2.8 1.1 1.9 Source: World Bank staff calculations, usingEHPMdata, 1991-2002 3.4 As can be seen from the table, the elasticity o f extreme poverty with respect to growth was -1.5 over the period, indicating that a one percent increase in per capita income was associated with a decline inthe extreme poverty headcount of 1.5 percent, holding the distribution o f income 18 constant. Inthe case o f total poverty the elasticity was -1.3, meaning that a one percent increase in per capita income was associated with a 1.3 percent decline in the total poverty headcount (again holding the distribution o f income constant). -but the distribution of the benefits of growth has also mattered- 3.5 The pattern of growth clearly also matters. This can be seen by the fact that the elasticity o f extreme poverty with respect to inequality was 2.8, indicating that a one percent increase inthe gini coefficient (Le., an increase in income inequality), holding income levels constant, was associated with an increase in the extreme poverty headcount o f 2.8 percent. At the same time, the elasticity o f total poverty with respect to inequality was 1.9. In other words, a one percent increase in the gini coefficient was associated with a 1.9 percent increase in the total poverty headcount. 3.6 More detailed analysis o f household income growth over the period, usingthe EHPM data, indicates that growth was reasonably broad-based during the period, contributing to El Salvador's decline in poverty. As can be seen in Table 3.2, the per capita incomes o f the poor grew on average by 4.6 percent from 1991-1995 and by 3.8 percent between 1995 and 2000. For the entire 1991-2002 period, average incomes of the poor grew by 3.1 percent a year, just below the nationwide average o f 3.7 percent. 3.7 Highper capita income growth among the poor during the 1991-1995 period contributed not only to poverty reduction in the first half o f the 1990s, but to poverty reduction in the second half o f the decade as well. This is because while income growth among some o f the moderate poor was insufficient to push them over the poverty line in the first part o f the 1990s, it moved them close enough to the poverty line that slower growth post-1995 was sufficient to pushthem over the poverty line in the latter half o f the decade. This fact points to the importance o f sustained growth in achieving significant reductions inpoverty over time. I I1 1991-1995 1995-2000 2000-2002 1991-2002 Growth inMeanp/c Income I I I I ~ National Average 5.8 3.9 -1.0 3.7 The Poor (total) 4.6 1 3.8 -2.0 3.1 ~ 1 ~ Lowest Quintile 0.2 4.3 -3.6 1.3 I I I 1 Growth inMedianp/c/ Incomes 1 7.0 3.8 1 -0.2 4.2 i National Urban 8.1 1 2.5 -0.2 4.0 1 Rural 1.5 2.8 1.6 2.1 ~ -and income growth of thepoorest Salvadoranshas lagged- 3.8 There were important differences in income growth among the poor, however, during the period. While per capita incomes o f the moderate poor grew at roughly the same rates as average 19 incomes duringthe 1990s, average incomes o f the poorest Salvadorans lagged duringmuch o f the period. These differences in income growth between the moderate poor and the extreme poor account inpart for the increase in aggregate income inequality, particularly since 2000, when the poorest Salvadoran experienced significant declines inreal per capita incomes. 3.9 Patterns o f household income growth, across rural and urban areas, also conform largely to the patterns o f poverty reduction seen during, before and after 1995. As can be seen from Table 3.2, per capita incomes o f urban households grew considerably faster than those o f rural households during the 1991-1995 period, whereas income growth was much more balanced during the 1995-2000 period. Moreover, although urban per capita income growth has essentially stalled since 2000, rural incomes have continued to grow, leading to continued - albeit much slower -poverty declines. Structural shifts in the economv Contributionto Percent Population Average Per Capita Headcount National Reductionin Shares Income ' Index Povertv Povertv due to Intra- Pop- 1991 2002 1991 2002 1991 2002 1991 2002 'lation Agricultureand Gains Shifts 38.9 18.1 248,709 242,743 75.3 74.1 45.5 36.2 1.7 57.3 Mining I I I Industry, 15.6 15.6 415,209 541,340 61.1 35.2 14.8 14.7 14.8 0.2 ~ Manufacturing Services 23.3 33.2 518,061 719,685 53.6 25.3 19.4 22.7 24.2 -19.3 IConstruction 5.8 6.9 333,128 477,727 68.9 36.7 6.2 6.8 6.9 -2.7 Public 5.8 5.2 479,938 834,551 52.1 12.5 4.7 1.7 8.4 1.2 Administration Social, Community 7.0 9.8 555,673 899,834 , 47.3 22.9 5.2 j 6.1 6.3 , -4.7 1 Services I Other 2.3 1 4.2 '262,010 402,461 79.5 32.1 2.8 3.6 4.0 -5.5 Undesignated 1.2 7.2 333,439 510,693 68.0 41.8 1.3 8.1 1.2 -14.9 I Total 100.0 100.0 378,811 584,291 64.4 37.2 100.0 100.0 67.4 11.6 l6 Because o f the lack of recent census data with which to re-weight the survey samples over time, care should be taken in interpreting the sectoral labor shares and their changes over time. At best, they should be taken as indicative. 20 Source: World Bank staff estimates, usingEHPM data (1991, 2002) -with benefitsfor those whofound employment in other sectors. 3.11 As can also be seen from the table, households whose main income source came from outside o f agriculture experienced significant per capita income growth over the period -between 30 and 74 percent, depending on the sector. In contrast, households whose main source o f income came from agriculture had marginally lower per capita incomes on average in 2002 than in 1991. This suggests that households whose main source of income remained agriculture over the periodlost considerable ground over the period." Indeed, whilepoverty declined among householdsin every sector- 3.12 The incidence of poverty declined in every sector over the period - although the pattern across sectors is striking (and consistent with income changes over the period). Nearly all sectors registered significant declines in poverty rates between 1991 and 2001. The only exception is among those who continued to earn the majority o f their incomes from agriculture. Among those households, the incidence o f poverty only declined marginally, from 75.3 to 74.1 percent. All other sectors experienced double-digit declines intheir poverty rates. 3.13 Decomposition analysis can provide a clearer sense o f how important within-sector increases in incomes and population shifts across sectors have been to poverty reduction in El Salvador between 1991 and 2002. This can be done using a simple formula that enables one to decompose reductions in aggregate poverty into: (i)declines in poverty due to within-sector improvements in incomes over the period; (ii) shifts in the main source o f household earnings from one sector to another over the period; and (iii)interactions between within-sector improvements and inter-sectoral shifts.l8 3.14 Decomposition indicates that roughly two-thirds o f the decline in poverty over the decade came as a result o f within-sector increases in income, that is, among households whose main income source remained the same in 1991/92 and 2000 (Table 3.3). Indeed, the largest declines in national poverty came from improvements in living standards among households who continuedto earn inthe services sector (24.2% o f all poverty reduction) and from improvement in living standards of those whose main income source came from industry and manufacturing (14.8%). In contrast, only 1.7 percent of the decline in poverty came from within-sector improvements among households that continued to rely primarily on agriculture. -there were only limited gains to those whose main incomes remained in agriculture. 3.15 Nearly 12 percent o f the reduction in national poverty came from shifts in household earnings from one sector to another. l9 By far, the most important example of this was the large shift away from agriculture as the main earning source ofhouseholds. This substantial change in the structure of household incomes contributed to over halfo f the "gross" decline inpoverty over the 1990s-althoughthese gains were offset somewhat by the fact that some poor households that shifted their main source o f income away from agriculture to other sectors remained poor. Nonetheless, this latter phenomenonwas much smaller than the poverty reduction associated with within-sector improvements in incomes or with shifts away from agricultural earnings over the period. ~ l7 Some caution i s neededininterpretinghow this relates to individual households' incomes. The EHPM data are repeatedcross sections and, so, changesinspecific households' incomes cannotbe traced. For details on the methodology see Ravallion andHuppi (1991). l 9The remaining21% ofnationalpoverty reductionbetween 1991and2002 came frominteractions betweenwithin-sectorgains andshifts across sectors. 21 3.16 It i s important to note that even with the large shifts away from agricultural earnings from 1991-2002, agricultural households are still disproportionately represented among the poor. More than one-third of El Salvador's poor still count on agriculture as their main income source, although only about 18 percent o f all households depend primarily on income or earnings from agriculture. Household determinants of income 3.17 What are the key factors driving household income growth and, thus, poverty reduction during the period? To get at this question, several statistical analyses were undertaken usingthe EHPM and BASIS data sets to better understand how different factors affect households per capita income and per capita income growth. Themainfactors associated with households beingpoor are: 3.18 Inthe case of the EHPMdata, reduced form equations where (the log of) householdincome per capita is estimated as a function of household and regional characteristics, including family demographics, human capital, employment status, type o f job, sector o f employment, gender, regiono f residence, and migration(as proxied by the receipt o f remittances). Quantile regressions were used to assess the extent to which different factors might affect households differently across the conditional income distribution. Estimations were performed both for 1991 and 2002 to identify possible changes over time in the returns to specific factors, such as education or sector o f employment. The analysis can thus provide some insights into which groups o f households benefited more or less from the recent growth and economic change that has taken place inEl Salvador. 3.19 Overall, income per capita is found to be higher among households with a well educated head, who are employed in high productivity services (e.g., financial, real estate), live in Metropolitan San Salvador, and who receive remittances, holding constant all other factors. In contrast, per capita income is lower - and thus the probability o f being poor higher - among families where the household head is employed in agriculture, works in the informal sector or i s unemployed, has high dependency ratios, large numbers o f old adults and children, or lives in relatively poor administrative departments. Inmany cases, the returns to particular factors have changed over time and differ according to a household's place in the conditional income distribution. For example: a low level of educational achievement; 3.20 Education. In 2002, the per capita income o f a household with a primary-educated head was 13 percent higher, on average, than for a household with an uneducated head, up slightly from 1991 (Table 3.4). The gain from having a high-school-educated head relative to having a primary educated head was about 26 percent in2002, up from about 20 percent in 1991, while the average gain from a household head having a university education increased from 26 to 38 percent over the period. 22 Table 3.4: Marginal Contribution of Educationto HouseholdPer Capita Income, 1991-2002 Percentage increasein householdper capitaincomeassociatedwith the householdhead (or spouse) havingattainedthe nexthigher education level Quantile 0.5 Year 0.1 0.25 (median) 0.75 0.9 Headof Household I Primary Schooling 1991 12.9 11.8 9.8 12.9 16.4 2002 21.3 19.1 13.4 12.3 11.8 Secondary Schooling 1991 15.7 1 20.0 1 19.5 21.9 32.0 I 2002 16.5 1 26.5 j 25.8 24.7 20.3 1 University 1991 1 20.0 22.5 26.5 1 36.3 I 47.9 1 ~ ~ 2002 29.6 I 31.6 37.9 44.8 61.4 Spouse 1 i I 1 1 ~ I Primary Schooling 1991 5.1 1 5.7 5.6 6.4 8.7 ~ ~ I 2002 6.1* 8.8 7.6 11.0 7.0 Secondary Schooling 1991 13.3 17.4 21.1 27.4 23.7 2002 9.1 10.6 13.9 18.1 25.8 University 1999 11.5 14.8 17.3 21.7 26.3 2002 35.6 31.3 30.2 36.0 21.3 3.21 Having a well-educated spouse brings additional income gains, controlling for other factors. While the average income gains from a spouse's primary education remained below 10 percent over the period, having a high school-educated spouse contributed an additional 14 percent to per capita income in 2002 (down from 21 percent in 1991). The marginal contribution of a spouse's college education to household per capita income, relative to secondary education, reached 30 percent in2002, up from 17percent earlier inthe decade. 3.22 These averages mask substantial differences in the premia from education across the income distribution, however. Among households in the lowest quantiles, the per capita income gain associated with a primary educated head grew during the period, and by 2002 exceeded the gains to primary educated heads among households in the upper quantiles. However, for higher levels o f education, the income premia from the household head's education was generally lower and grew less quickly at the bottom o f the conditional income distribution. This i s most dramatic 23 at the university levels where returns to university education grew from roughly 20 to 30 percent from 1991-2002 among those in the lowest quantiles, but from 36-48 percent to 45-61 percent among those inthe highest quintilesduringthe period. working in agriculture- 3.23 Economic activity. Working in industry or services rather than in the agricultural sector is associated with higher per capita income - and increasingly so. The income penalty associated with a household head working in agriculture or mining (compared to white collar services) was roughly 40 percent lower in 2002, up from 33 percent in 1991. Moreover, the affect is strongest among the lowest quantiles, where the penalty is as high was 45-46 percent in 2002. The econometric analysis confirms that relative incomes grew comparatively faster outside o f agriculture. -the informal sector- 3.24 Informal employment. Informal sector employment o f a household head yields lower per capita income than salaried employment. In 2002, households with self-employed heads, for example, had per capita incomes 43 percent lower, on average, than those with salaried jobs, a gap that had grown from roughly 28 percent in 1991. Self-employment appears to carry a slightly lower income penalty for households in the bottom than in the top quantiles, however. Social security affiliation, a proxy for formal sector salary employment, i s associated with an approximately 24 percent income premium, on average. This premiumhas been roughly constant over the period, although the premiumhas increased somewhat among the upper quantiles. - o r not at all; 3.25 Unemployment. Having a head who i s unemployedreduced expected per capita income by 19 to 23 percent, on average, in both 1991 and 2002. The relative impact o f unemployment on households at the bottom and top o f the conditional income distribution changed considerably over the period, however. In 1991, households at the top o f the distribution experienced a greater than average income penalty from unemployment o f a household head, but those at the bottomo f the distribution bore the greatestpenalty in2002. and living outside of San Salvador; 3.26 Region of residence. There has been some convergence in income levels across El Salvador's administrative regions over time. That is, the income penalties associated with living in regions outside of Metropolitan San Salvador have declined since 1991. While the income disadvantage associated with living outside o f Metropolitan San Salvador ranged from 14 to 21 percent in 1991, depending on the region, it had fallen to about 7 percent by 2002. 3.27 Gender of the household head. In general, the income penalty associated with living in a female-headed householdin 1991had disappeared by 2002. An exception was among households in the upper quantiles, who experienced a smaller, but still significant income penalty (5-9 percent) in2002, compared to identical male-headed households. Receiving remittancesfrom relatives abroad reduces the risk of poverq- 3.28 Remittances. Receipt o f remittances, proxied by migration o f a household member, gave a solid, positive boost to per capita household income in both periods. Indeed, households that receive remittances have per capita incomes 50 percent higher than otherwise identical households. This effect is strikingly similar across quantiles o f the distribution and did not vary significantly over time. This 50 percent boost to income, other factors heldconstant, clearly has a critical impact for those who receive remittances. As is discussed in greater detail below, remittances have played an important - albeit not the dominant role - in El Salvador's decline in poverty between 1991and 2002. 24 -as does greater control of assets, access to road and transport infrastructure- 3.29 Analysis of the correlates of per capita income was also done using the BASIS rural panel data (Beneke de Sanfelih and Shi 2004). Analysis o f the BASIS data was done using a slightly different set o f explanatory variables, as the BASIS dataset contains more detailed information in some areas, such as on household assets and on households proximity to all-weather roads and markets. Consistent with the EHPM findings, the BASIS analysis indicates that between 1995 and 2001 rural per capita household incomes were consistently higher in households with higher education levels and who received remittances, while they were lower among those households with a greater number o f dependents. The analysis also finds that households tend to have higher incomes when they control for more assets (e.g., land) and when they have greater access to road and transport infrastructure (e.g., buses). Consistent with the findings above, rural households engaged in the production o f basic grains tend to have lower per capita incomes, while those engaged in a greater number o f non-agricultural enterprises and (in the latter part o f the period) non-traditional agricultural products, tend to have higher per capita incomes (controlling for other factors). 3.30 The BASIS data was also used to analyze the determinants o f household income growth, taking advantage o f the fact that the BASIS has data on the same households at 4 different periods o f time. The analysis o f household income growth reinforces the cross-sectional findings. Specifically, average levels o f education inthe householdin 1995 had a strong positive impact on household income growth over the period, as didfamilies' control o f assets.20Consistent with the cross-sectional findings, households who lived greater distances from paved roads or from transportation infrastructure (e.g., bus stops) in 1995 experienced slower income growth during the 1995-97 period, perhaps due to higher transaction costs (in terms o f time and money) associated with engaging in the market economy. Moreover, reductions in the distances o f households to paved roads and bus stops - i.e., improvements in access to market services - was associated with faster income growth, particularly from 1995-1999. -and having more diversified sources of income. 3.31 Households producing basic grains in 1995 tended to increase their income more slowly over the period, while those with micro-enterprises tended to rise more quickly. Indeed, diversification o f income sources - for example, via an increase in the number o f a household's non-agricultural enterprises or in the cultivation o f non-traditional agricultural products - was strongly related to income growth between 1995 and 2001. Increases in cultivation o f non- traditional agricultural products was also associated with positive income growth from 1995 to 2001, albeit less strongly than in the case o f undertaking new non-agricultural enterprises. Finally, households that received remittances at the beginningo f the period- as well as those that receivedincreases inremittances over time -also experienced faster growth. 3.32 It i s worth noting that according to the analysis o f the BASIS data, access to credit, either formal or informal, did not show up as a major correlate or determinant o f household per capita income over the 1995-2001 period. It is not completely clear why this is the case. But one possibility might be that income from remittances provided a measure o f liquidityandor working capital to support households' efforts to develop micro-enterprises and diversify into non- agricultural enterprises. This issue deserves further investigation. Analysis of the BASIS data also shows that higher education reduces households' vulnerability to poverty (Beneke de Sanfeli6 and Shi 2004). Households with low average levels of education are more likely to be chronically poor, controlling for other factors. In contrast, households with higher levels of education are less likely ever to be poor. Education also helps families to manage economic risk and shocks more effectively; for example, households with more educated members are likely to be poor only temporarily, ifthey get hitby a negative income shock due to family, environmentalor economic reasons, 25 The role of remittances in reducina uovertv, 1991-2002 Remittancescan give a significant boost to household income- 3.33 Analysis o f both the EHPM and BASIS data indicate that remittances have made an important contribution to householdincome andrecent household income growth. As analysis of the EHPM indicated, households that received remittances in 2002 had per capita incomes 50 percent higher than other households, controlling for other factors. Butjust how important have remittances beento the overall poverty reduction story inEl Salvador? -and their importance has increased in recent years 3.34 The EHPM data indicates that the contribution o f remittances has indeed increased in recent years. For example, they comprised 5 percent o f per capita income in 1999 and 7 percent in2002, on average, and about 10 percent ofper capita income among households inthe second and third quintiles o f the income distribution. The BASIS data show slightly higher figures - although inthe same order o f magnitude. Remittances and "other family assistance" grew from 8 percent o f average family income to about 16 percent between 1995 and 2001, while remittances from outside El Salvador comprised 6 and 13 percent o f family income in 1995 and 2001, respectively.21Thus, labor earnings and self-employed incomes still make up the vast majority o f Salvadoran's incomes, on average. Nonetheless,relativelyfew of thepoorest Salvadorans receive remittances- 3.35 Moreover, the data suggest that the proportion o f households that receive remittances has increased since the 1990s as well. The EHPM indicate, for example, that 22 percent o f the population receives as least some remittances (although only 11percent o f the poorest Salvadoran households received remittances in 2002). Similarly, the BASIS data indicate the proportion o f rural panel households receiving remittances increased from 13 to 30 percent between 1995 and 2001. As in the case o f the national survey, the BASIS data show that a relatively small percentage o f very poor households (19 percent) receive remittances. -and their role in aggregatepoverty reduction, while important, does not appear dominant. 3.36 The dynamics o f income and poverty reduction are difficult to capture with precision. Nonetheless, to assess the order o f magnitude o f the impact o f remittances on poverty reduction, poverty rates were recalculated (using both the EHPM and BASIS surveys) using an income aggregate calculated with and without remittances. As can be seen from Figure 3.1, in the absence o f remittances, the decline in total poverty between 1991 and 2002, estimated using the EHPM data and the World Bank income aggregate, would have been 23.9 percentage points (from 66.8 to 42.9 percent) rather than 27.2 percentage points (from 64.4 to 37.2 percent), a difference o f 3.4 percentage points. The decline in extreme poverty from 1991-2002 would have been 13.2 percentage points (33.7 to 20.5 percent) as opposed to 15.8 percentage points (31.2 to 15.4 percent), a difference o f 2.6 percentage points. 3.37 Estimates o f rural poverty, using the BASIS data indicate a similar order o f magnitude changes. Inthe absence o f remittances, the decline o f rural poverty from 1995-2001 would have been 24.5 percentage points (77.8 to 53.3 percent) rather than 28 percentage points (74.6 to 46.6 percent), a difference o f about 3.5 percentage points. 3.38 Interpreting these numbers should be done with some caution. On one hand, inthe absence o f out-migration and remittances, families probably would have increased their labor supply and, thus, their earnings (perhaps through the entrance of the out-migrant into the domestic labor *'Salvadorwhichsuggest that These householdsurvey-basedfigures are roughly inline with the macro-economic figures for El worker remittances from abroadgrew from an average of 2.6 percent of GDP during the 1980sto over 11percent during the 1990s(World Bank 2003). 26 market). This would mean that the incomes o f those receiving remittances would be higher than suggested by a calculation without remittances. This would mean correspondingly lower poverty rates. On the other hand, the existence o f additional income via remittances has likely led to domestic demand for goods and services that would not have existed in the absence o f those remittances. This increased demand for goods and services resulted in higher incomes (and perhaps less poverty) for some number o f families that did not receive remittances. Just how those two (opposing) impacts net out is difficult to assess with certainty. 3.39 Nonetheless, these simulations suggest that while remittances have made an important contribution to poverty reduction inrecent years, they have not played the dominant role inrecent achievements. Figure 3.1: Trends in Poverty With and Without Remittances, using EHPMand BASIS Data" a. NationalPoverty Trends, 1991-2002, b. Rural Poverty Trends, 1995-2001 Total and Extreme Poverty (EHPM) Total Poverty (BASIS) I 80 90 70 80 70 -e60 50 60 c 40 c 50 30 8 f 40 20 30 20 10 10 C 1991 2002 0 I905 2001 4 T o l a l Poverty (mlh) +Total Poverty (wiltwut) Extrem Pov~rty(mlh) -+--Extrem Poverty (wiwout) +Total Poverty (with) *Total Poverty (wllhout) Source: World Bank staff calculations Source: Benekede Sanfeliu and Shi2004 Natural and economic shocks 3.40 The 2001 earthquakes. InJanuary, 2001, El Salvador was hit by two powerful earthquakes that resulted in over 1,000 casualties and left more than 300,000 families homeless. As many as 85 percent o f rural households in the main earthquake-affected areas experienced some damage to their homes and as much as 40 percent o f households in affected areas experienced damage severe enough to make their homes uninhabitable (Beneke de Sanfeli6 and Shi 2004). Data limitations make accurate assessment of theeffects of the 2001 earthquakesdiyJcult- 3.41 Making an accurate estimate o f the poverty impacts o f the earthquakes i s complicated by the lack of sufficient data to perform careful before-and-after analysis. Nonetheless, a reasonable assessment can be based on a comparison o f poverty trends during 1999-2002 between localities hardest hit by the quakes and those that were less affected, using the 2000 and 2001 EHPM surveys, the latter o f which includes an over-sampling in the most affected municipalities. Most affected municipalities are defined as those with at least 30 percent o f homes damaged (the cutoff 22Estimatesusingofficial DIGESTYC poverty line. 27 usedto identify the municipalities ofthe 2001 EHPMover-sample. 3.42 Table 3.5 presents data on the most affected departments and municipalities, along with the poverty rates, by department, over the 1999-2002 period. The most affected municipalities in these departments house almost a million people (16 percent o f the country's population extrapolated from the 1992 census) and comprise a majority o f the departments' populations (except inL a Libertad). Total The worst affected PovertyRates Departments Municipal. Pop. Municipal. % Pop. % 1999 2000 2001* 20002 La Libertad 22 707,406 10 45.5 153,595 21.7 32.7 28.6 31.5 26.2 Cuscatlan 16 205,372 10 62.5 159,032 77.4 38.6 32.9 44.3 35.9 L a Paz 22 298,977 18 81.8 233,484 78.1 49.7 43.1 42.9 42.7 San Vicente 13 163,462 10 76.9 140,902 86.2 59.2 54.3 52.6 52.7 Usulutan 23 340,576 16 69.6 206,866 60.7 55.2 49.2 54.1 46.1 Total 96 1,715,793 64 66.7 893,879 52.1 43.6 38.3 41.5 36.7 -nonetheless, their impacts onpoverty appear to have been short-term and localized- 3.43 Examination o f national poverty data makes clear that the earthquakes did not reverse progress in reducing poverty at the national level (Figure 2.1, Table 2.1). Nonetheless, they did have important local impacts in the short-term. Significant poverty impacts can be observed onZy in those departments comprised of most localities hardest hit by the quakes. Indeed, while the incidence o f poverty continued to decline in most departments in 2001, about 3 percent of the population inthe most affected departments fell into poverty following the earthquakes. Poverty rates increased the most in Cuscatlan followed by Usulatan and L a Libertad, while L a Paz and San Vicente experienced little change up or down. (Similar patterns are observed, albeit in smaller magnitudes, with respect to extreme poverty.) 3.44 Taking the less affected localities as a control group to estimate the counterfactual poverty decline that would have been observed in the absence o f the quakes, it is possible to obtain a basic difference-in-difference estimate o f a 5 percentage point poverty increase in the most affected departments possibly due to the earthquakes. Inreality, this should be taken as a lower bound estimate o f the local impacts o f the quakes on income poverty for at least two reasons. First, the definition o f most affected localities used for the over sample survey (30 percent o f damaged homes) i s ultimately arbitrary and does not capture the plausible continuum of impacts. Second, negative spillover effects to neighboring municipalities or departments could have affected poverty reduction in the control Both these effects are likely to lead to an upwardbias inthe counterfactualestimate. -as reconstruction efforts provided a boost to affected local economies 3.45 Nonetheless, household incomes rebounded quickly inthese localities, and poverty inthese locations generally declined to pre-earthquake levels again in 2002. In fact, the declines in 23Only 2 departments that had relatively few affected municipalities registeredsmall increases inpoverty after the earthquakes: Chalatenangoand San Salvador. 28 poverty in these localities during 2002 were much higher than in the rest of the country, so that the gap in poverty rates between the two groups reverted to nearly pre-earthquake levels. The negative impacts o f the earthquakes on poverty thus appear to have been quite temporary with, among other things, rehabilitation activities providing a boost to the economies in the most affected areas.24 There may be a longer-term impactson school enrollments- 3.46 However, the earthquakes appear to be responsible, in part, for some declines in school enrollments inbasic education. Almost 4 percent o f children left school between 2000 and 2001 - at least temp~rarily.~~ While school enrollments have slowed nationwide in the 2000s, enrollments tended to decline in the departments with most-affected localities. For example, net enrollment rates in Usulutan and Cuscatlan declined by 4 and 6 percentage points, respectively, after the quakes. And while enrollment levels have since rebounded, in2002 only about half the affected localities achieved enrollment rates at or above their pre-earthquake levels. -and on housing stock in quake-affected areas. 3.47 Perhaps the most lasting impacts o f the earthquakes have been in the deterioration of housing quality inthe worst affected localities. The quality o f roofs declined inabout 9 percent o f houses nationwide (4 percent in rural areas) and the percentage o f houses built with low quality (improvised) material doubled following the quakes (Figure 3.2). The collapse o f housing infrastructure was substantially higher inthe most affected localities. As can be seen inthe figure, housing stock showed a slight improvement in2002.26 Figure3.2: Quality of HousinginElSalvador, 1999-2002 ("/) Concrete Roofs (W Walls of low quality rnaterlals 80 70 60 50 40 30 20 I O 0 Urban Rural National Urban Rural National Source: WorldBank staff basedonEHF'M survey data. A dramatic drop in world coffee prices since 1997- 3.48 The coffee crisis. Untilrecently, coffee was the main export crop o f El Salvador; it is still important in some regions o f the country, generating employment and income for several thousand families. While a declining share o f El Salvador's labor force is involved inthe nation's coffee economy, decliningworld coffee prices have dampened demand for labor on large coffee 24 Beneke de Sanfehi and Shi (2004) find that most rural households experienced only temporary dislocation from their productive activities as a result of the earthquakes. The average work interruption reportedby rural householdswas about 40 days. 25 In the rural areas most affectedby the earthquakes, about 40 percent of studentsreportedstopping school temporarily(Beneke de Sanfeliu and Shi 2004). School-leavers reportedan average of 37 days out of school as a result ofthe earthquakes. 26 The impactonbasic services infrastructurewas less significant. CuscatlanandLaPaz are the only departmentswhere water access was interruptedfor about 4 percentofhouseholds. By 2002, however, water access rates inthe most affecteddepartments were significantly abovepre-earthquake levels. There were no significantnegative impactson access rates to sanitationand only a marginaldeteriorationo f electricity access inLa Paz. 29 farms, forcing workers to seek alternative sources o f income, and have adversely affected the incomes o f smallholder farmers who cultivate coffee as part o f their families' income portfolio. 3.49 World coffee prices have a history o f volatility, but the current price drop was significantly larger than usual. Indeed, since reaching their peak in May 1997, prices for Salvadoran and Guatemalan coffee have declined to among their lowest real values in the last 50 years (Figure 3.3).*' Figure3.3: CoffeePricesinUS$Per Pound(January/1980-August/2003) 250 4 1 Average Price in Survey Years 200 -Coffee, Robusta, 150 Uganda and Cote dlvoire, ex-dock New York 100 -Coffee, Other Milds, E Salvador and Guatemala, ex-dock 50 NewYork 0 Source: IMF,http://www.imf.org/extemal/np/res/commodindex.asp. -slowed income growth, and increased income volatility, among workers in the coffee sector. 3.50 Recent analyses o f the impacts o f the coffee crisis, using BASIS data from 1995, 1997, 1999, and 2001 (Trigueros and Avalos 2003; Beneke de Sanfeli6 and Shi 2004) reveal that: Households who had at least some members working in the coffee sector have seen their incomes negatively affected by the coffee crisis. This has not manifested itself as a decline in real income since 1997 (the height o f recent coffee prices), but rather as slower real income growth than that o f otherwise equivalent rural households outside the coffee sector. Coffee sector households have also experienced considerably more volatile incomes than non-coffee sector households since the mid-l990s, but have managed to preserve and grow their incomes through shifts out o f coffee into non-agricultural enterprises. There is suggestive evidence that the coffee crisis has also adversely affected child nutrition. Specifically, between 1998 and 2002, the incidence o f underweight children declined more rapidly in non-coffee growing regions o f El Salvador than in coffee-growing regions (Figure 3.4). Inthe case o f Ahuachapan, the department with the highest land intensity of coffee production, the incidence o f underweight children actually increased over the period. *'For a detailedanalysis ofthe current situation and crisis ininternationalcoffee marketssee Lewin and Giovannucci(2002). 30 Figure3.4: ChangeinPercentageof UnderweightChildren,1998-2002, byDepartmentand Intensityof Coffee Cultivation age change in underweig mwSALVADOR amVICW MIGUEL m W T A A N A m c m AmCUSCAT-4'4 s mCHALATENANGO mL.4 PPZ . M O W fiUSULUTAN Source: Trigueros andAvalos 2003, basedon FESALandPROCAFEdata. 0 There is no systematic evidence that the coffee crisis (or participation in the coffee economy more broadly) has negatively affected children's school participation; however there may have beenadverse impacts inspecific locations. Implications for uovertv reduction Dolicv Focus on enabling the most vulnerabletoparticipate infuture economicprogress 3.51 This chapter has examined key factors that have influenced poverty reduction since the early 199Os, including economic growth, recent structural changes in the economy, and the factors affecting householdincome growth (both own-earned income and the role of remittances). The chapter has also examined the impact o f recent shocks - the earthquakes o f 2001 and the coffee crisis - on recent socio-economic progress. The collection o f evidence highlights the following: Growth was an important force for poverty reduction in the 199Os, as were structural changes in the economy. Indeed, growth was reasonably broad-based and effective in reducing poverty. e Nonetheless, not all households benefited equally from growth during the period. While the incomes o f the moderately poor grew roughly at the rate o f average per capita income growth, the poorest Salvadorans experienced relatively little growth. Analysis o f the EHPM and BASIS surveys indicate that education, ownership o f land or other physical assets, and access to markets and basic services (via proximity to all- weather roads and transportation) were all key factors in households' ability (or not) to participate in recent economic progress. Households' involvement in formal-sector salaried employment, in non-agricultural enterprises, and their ability to diversify their 31 income portfolio -related, inpart, to education, assets, and access to markets - were also important. 0 The evidence suggests that many o f the poorest, most vulnerable Salvadorans still lack sufficient human capital, as well as access to markets and basic services, to be able to advantage o f new and emerging opportunities. Ensuringthat the poorest, most vulnerable citizens are equipped to participate in future economic progress will, thus, need to be at the core of aforward-looking strategy for reducingpoverty inElSalvador. 0 Inthe chapters that follow, these possible elements of a national social policy to reduce poverty inEl Salvador are explored ingreater detail, includingapproaches to o strengtheningthe human capital of all Salvadorans o improvingpeople's access to markets andbasic services, and o supporting and protectingthe poorest, most vulnerable families. 32 PART 2: CRAFTING A NATIONAL SOCIAL POLICY IV. ELEMENTSOF A POVERTYREDUCTIONSTRATEGY Growth will remain an important engine of poverty reduction- 4.1 As noted in the previous section, during the 1990s growth was relatively effective in reducing poverty. The evidence suggests that economic growth will continue to be a critical element of El Salvador's efforts to reduce poverty in the coming years as well. Thus, there will be high returns in terms o f poverty reduction to the Government of El Salvador taking the necessary measures to reinvigorate growth in the 2000s. In this context, the recent World Bank Country Economic Memorandum (CEM) for El Salvador (2003) highlights several key areas for policy and investment to enhance growth (Box 4.1). Box 4.1: PromotingEconomicGrowth for PovertyReductionin ElSalvador The World Bank's recent Country Economic Memorandum for El Salvador (2003) examines both the macro- and micro-economic underpinnings o f El Salvador's recent growth performance to identify the key policy levers for promoting rapid and sustainable growth inEl Salvador. The report presents the results of cross-country data analysis on the macro and structural determinants o f growth, as well as analyses of microeconomic foundations o f economic development related to such factors as the quality of inputs, infrastructure, institutions, and the array o f regulatory and other policies that shape the business environment. The study identifies four priority areas that will, if addressed, will lay the foundation for increased economic growth inEl Salvador over the next decade: Increasing the education levels of the population. Education provides a critical foundation for future productivity growth in the Salvadoran economy. Indeed, macro-economic analysis indicates that concerted efforts to improve the education levels o f the Salvadoran population, along with the quality o f education, hold the highestpotential to increase economic growth rates inthe medium-to-long term? Developing the country's economic infrastructure. As with education, development o f the country's economic infrastructure, such as roads, ports, and telecommunications infrastructure will go far toward increasing its economic competitiveness and, hence, future growth. Fostering greater technology adoption and local innovation. Closely tied to the education agenda, the ability to adopt technology and to innovate locally will be critical in the coming years to El Salvador's attracting foreign investment and increasing domestic economic productivity. In addition to gains from higher levels o f better quality education, there appear to be gains associated with greater levels of domestic research and development to promote production innovations. Improving the investment climate, including efforts to reduce violence and increase the rule of law. El Salvador has one o f the highest crime rates o f the Western Hemisphere and this is perceived as a significant cost on business; there i s also a perception o f judicial insecurity for businesses which adversely affect investments, economic productivity, and growth. The diagnosis of poverty presented in chapters 2 and 3 o f this report suggest important complementarities between the growth and poverty reduction agendas in El Salvador. Specifically, the analysis suggests strong complementarities growth and poverty reduction with respect to investments in education and in economic infrastructure, both o f which demonstrate strong impacts o f incomes and the household (micro) and national (macro) levels. 28See also Loayza, Fajnzylber, and Calderbn, 2002. 33 -but enabling all Salvadoransto benefitfrom growth callsfor a national socialpolicy 4.2 Given the nature of the remaining poor, however, and the fact that many o f the poorest, most vulnerable Salvadorans have beenunable to take advantage o f recent economic progress, it will become increasingly important that El Salvador put in place a deliberate and strategically designed set o fpolicies and investments to ensure that the remaining poor are better positioned to benefit from future socio-economic progress. The analysis presented so far suggests several broad areas where increased and strengthened government attention appears warranted: instrengthening education-or more broadly stated-indeveloping the human capital of all Salvadorans; e in strengthening people's access to markets and basic services, especially among the poor; and e providing support and protection to the poorest and most vulnerable Salvadorans to ensure that they are better prepared to take advantage o f future growth and development. 4.3 In the chapters that follow, the possible strategic elements of a National Social Policy to reduce poverty are explored. Opportunities and challenges in education, health, socio-economic infrastructure, and development o f a social safety net are examined inturn. V. EDUCATION Education increases incomes and reducespeople's vulnerability topoverty 5.1 Education contributes to poverty reduction in a number o f critical ways - through its impact both on economic growth, and on household incomes and income growth, by reducing vulnerability to poverty and increasing economic mobility o f individuals and families. Building on the discussion in the previous section, this chapter focuses on: (1) recent advances in extending education to the poor; (2) the role that public actions have played in recent progress, with special emphasis on public education spending; (3) key education challenges for El Salvador, moving forward; and (4) policy directions for ensuring that education works for all Salvadorans. Recent twopress Significant progress inproviding access to education among thepoor- 5.2 El Salvador has made considerable progress in raising enrollment and completion levels since the early 199Os, especially in rural areas. These gains have translated into considerable progress in access to education among the poor. As noted in Chapter 2, the enrollment gap between the poor and non-poor for basic education has closed between 1991 and 2002 - both in absolute and relative terms (Figure 5.la). Access to secondary education by the poor has also begun to increase in recent years - although gains have been larger among the non-poor (Figure 5.lb). In addition, more children from poor households are completing the first two cycles o f basic education, and this had led to some narrowing o f the gap in completion rates (Figure 5.2). All of this has added up to an increase inthe average years o f schooling among the poor (Figure 5.3), and a narrowingo f the education gap inElSalvador. 34 Figure 5.1: Net Basic and Secondary Enrollment Rates, by Poverty Status, 1991-2002 a. Basic Net Enrollments b. Secondary Net Enrollments 100 70 90 80 60 :: 0 70 20 50 --1 2* - 40 5e 40 30 W 30 - 20 7? 20 J I O 10 0 0 1991 1995 2000 2001 2002 1991 1995 2000 2001 2002 Extreme Poor -Relative Poor I N o nPoor +Average Source: WorldBank staff calculations, usingEHPM data (variousyears). figure 5.2: Basic CompletionRates figure 5.3: Average Years of Schooling, st & 2ndcycles), byQuintile, 1995-2002 by Quintile, 1995-2002 9 8 7 1 6 a 5 4 3 I 2 I 1995 2000 2002 1995 2000 2002 Year VeSr tile W 2ndOuintileu3rd Ouintile 04th Ouintile 5th Ouintile1 Source: di Gropello (2004) Source: di Gropello (2004) -although gaps remain wide in school completion and at secondary enrollment 5.3 These accomplishments have - and will continue to be - important to raising the capacities o f the poor to increase their incomes, to better protect themselves from shocks, and to take better advantage o f emerging economic opportunities. Nonetheless, recent accomplishments frame future challenges for the sector with respect to El Salvador's poverty reduction efforts. For example, in 2002, 22 percent o f 7-15 year-olds from the poorest families still did not attend school; access to secondary schooling also remains low among the poor; large gaps incompletion rates exist between the poor and the non-poor; and, as will be discussed in greater detail below, raising education quality, particularly for the poor, remains an important challenge. 35 Recent Dublic action 5.4 The public education system in El Salvador is divided into four levels: three years o f pre- primary education usually for children aged 4-6; nine years o f basic education for children 7-15, which in turn is divided in three cycles o f three years each; two or three years o f secondary education depending o f the modality: general studies two years and technical-vocational education three years; and the higher level o f education which includes university and technological education. Most students attend public institutions up to and including the secondary level- although the importance o fprivate education increases as the level o f education increases. Specifically, in terms o f enrollments, the public sector covers 81 percent o f pre- primary, 89 percent o f basic, and 68 percent o f secondary. The private sector plays the dominant role at the university level. Education reforms in the 1990s- 5.5 Recent progress in raising education outcomes for the poor are very much connected to recent education reforms, which grew out o f a broad-based consensus inthe country. During the 1990s, the administration o f the public education systemwas substantially de~entralized.~'Inthe early 1990s, MINED established EDUCO, a community-based school program managed through community associations, ACEs, which has been the main instrument to expand pre-primary and basic education in rural areas.30The ACEs have the responsibility for managing school resources includingthe hiringand firing o f teachers. Inthe mid-1990s the authorities began transferring the positive experience o f the ACES to the traditional system where School Councils (CDE) were created with the responsibility for: i)planning and implementing school activities; ii)managing the resources transferred to schools; and iii)raising additional finds to support school activities. Other instruments were developed at the school level to empower the schools, such as the Institutional Education Project (PEI), which sets forth the school's medium term objectives and the associatedAnnual SchoolPlan (PAE). -with an emphasis on improving access and increasing q u a l i F 5.6 MINED'Sstrategic objectives and targets are set forth in the Government's Ten Year Education Plan (1995-2005), which has received broad based support from the various sectors o f Salvadoran society. The Plan calls for: i)increasing coverage at all levels, but with emphasis on the third cycle of basic education and secondary education, particularly in rural and peri-urban areas; ii)creating and strengthening the different programs that facilitate access to education and permanence in school o f the poor; iii)improving the curriculum content and teaching-learning methods; iv) establishing learning evaluation systems as an instrument to improve continuously the quality of education; v) establishing monitoring, evaluation and incentive systems for teacher performance; and vi) strengthening school autonomy over resource management. 5.7 In 2000, MINED authorities reviewed the progress made with the implementation of the Plan and prepared their strategic program -Education Challenges in the New Millennium, which details the challenges faced by education in each key area, the objectives to be accomplished and the strategic guidelines. The program's principal objectives are to increase education quality, strengthen community participation, facilitate access to education and promote values and *'For a review of the sector reform effort during the 1990s see Marques, Jose Silverio and Ian Bannon. 2003. "Central America: Education Reform in a Post-Conflict Setting, Opportunities and Challenges," World Bank, Conflict Preventionand ReconstructionUnit,Working Paper 4, April. 30 The number o f EDUCO students inpre-primary and basic education increased from 10,72 1 and 42,85 1, respectively, in 1993 to 47,551 and 202,719 in 2002. Currently, EDUCO students represent about 16 percent of pre-primary and 18 percent of basic education students in public schools nationwide, and 44 percento f all basic-levelstudents inruralpublic schools. 36 personal development. It details the programs, targets and financial requirementsto achieve these objectives. 31 -and increasedpublic spending 5.8 Along with education reforms, El Salvador has significantly increased its public spending on education in the last several years. For example, the Government increased education sector spending from 1.9 percent o f GDP in 1992 to 2.3 percent in 1996 and to 3.2 percent in 2003 (Figure5.4). Education's shareofthe overall Central Government budgetwas also raised by over 7 percentage points in the secondhalf of the 1990s- from 12.4 percent in 1996 to 19.5 percent in 2003. Consistent with its strategy o f giving autonomy to the schools, in 1996 MINEDestablished an Education Quality voucher (or bono) that transfers resources to the schools (pre-primary, basic and secondary) to buy key goods and services to improve education quality. The voucher system was accompanied by a series o f other reforms that were gradually implemented, including: i)a teacher training/development program that is administered in a decentralized manner with strong school participation and the support o f Pedagogic Advisors, who replaced the previous school supervisors; ii)a decentralized system of administrative support to the schools; iii)incentives for teachers in rural areas; and iv) a performance-based incentive system for all teachers. At the same time, MINEDstrengthenedthe capacity o f its Departmental Offices to support the schools. Figure 5.4: EducationSector ExpendituresinElSalvador 4.0% 25.0% 3.5% 20.0% 3.0% 2.5% 15.0% 2.0% 1.5% 10.0% 1.0% 5.0% 0.5% 0.0% I 0.0% 1996 1997 1998 1999 2000 2001 2002 2003E -Education ExpenditureI GDP +Education Expenditure I Government Expenditure 1 Source: di Gropello (2004). Butpublic spending remains low by international norms- 5.9 Nonetheless, international comparisons indicate that public education spending is still low by international standards. As a percentage o f GDP, El Salvador still remains near the bottom o f 31Since initial preparationof and in-country consultations on this report, the Government inEl Salvador has developed and launchedanew nationaleducationplan, Plan 2021. This planbuilds on and expands the Government's existingvision for educational development. Plan 2021 focuses onachieving universal access to basic education o f quality, beginningat the pre-schoollevel. It focuses, as well, on promoting higherlevels o feducation, oftenthroughnon-traditionaleducationand outreach modalities, and strengthening educationinscience andtechnology(Ministry of Education,Government of El Salvador). 37 a list o f Latin American and Caribbean countries and well below the regional average (Figure 5.5). Spending i s also low relative to El Salvador's per capita income level. Moreover, a recent analysis by FUSADES o f public education spending (in PPP dollars) also indicates that per student spending is low by international standards (FUSADES 2002; Table 5.1). In 2000, of those countries studied, only the Philippines, Honduras and India spent less per student on pre- primary education, while only Honduras and India spent less per student on primary education. Moreover, at the secondary level, El Salvador ranked the lowest among the countries studied for per student spending. Figure 5.5: EducationSpending as a Proportion of GNP, 1999/2000 7 0 6 0 5 0 0) 5 4 0 P c 0) 2 3 0 Q) Q 2 0 1 0 0 0 w -1 Source: di Gropello (2004). US$PPP % of GNPper capita Pre- Pre- Primary Primary Secondary Primary Primary Secondaq Argentina 1054 1307 1781 10 13 17 Brazil 820 880 1087 13 13 17 Chile 1929 2151 2337 15 17 18 ElSalvador* 174 220(**) 260 9 11 (**) 13 Philippines 74 440 570 2 13 16 Honduras* 89 124 373 India 28 182 334 2 11 20 Jordan 528 690 1176 15 20 34 Malaysia 332 820 1334 4 10 16 Mexico 979 1108 2320 13 14 30 Paraguay 482 690 12 17 Uruguay 1104 974 1536 12 11 17 Zimbabwe 353 647 15 28 Average OECD 3788 3907 5312 17 20 29 38 5.10 To put these figures inperspective, it is also useful to compare public per student spending with private - i.e., household - spending on education in El Salvador. Matching data on household education spending with data on public per student spending in 2000 suggests that at the basic level, households spendon average 3.0 times more per student than the Government of El Salvador (Mason, Torres, and Aterido 2004). The lion's share of household spending per student goes to indirect costs of schooling, such as textbooks, materials, transportation and so on. -andprivate costs are high, especiallyfor poor households. 5.11 Poor households spendbetween 1.8 and 2.7 times, on average, what the government spends per student at the basic level. While the poor tend to spend less per student inabsolute terms than the non-poor, the poor tendto spendmore as aproportion of their per capita income. 5.12 Household spending per student at the secondary level i s even larger relative to public spendingthanat the basic level, due to boththe higher indirect costs of schooling as well as to the costs o f tuition. At the secondary level, households spend about 4.3 times more per student than the government on average, while the poor spendbetween3.O-3.6 times as much. Public spending has emphasized basic education- 5.13 The largest share o f public expenditures on education has gone - and continues to go - to basic education (Figure 5.6). From 1996-2003, the share o f the total education budget going to basic education has hovered around 60-65 percent - although its share has tended to increase slightly over time. This share is high by Latin American and international standards - but is generally consistent with the Government's commitment to broaden basic education coverage, especially in rural areas and among the poor (di Gropello 2004).32 The share o f secondary education is about 7 percent of the recurrent budget, which i s low by Latin American standards, and reflects the relatively low priority El Salvador has given this level in the past. The tertiary education share is similar to that of secondary education and, again, i s low by Latin American standards.33 32 This is true even correctingfor differences in the lengtho f El Salvador's basic education cycle and the rimaryeducation cycle ofother countriesinthe region. p3 Over the 1990-96 period, selectedLatinAmerican and Caribbean countries directedmore than21 percent of their education expenditurestoward tertiary education(di Gropello 2004). 39 Figure 5.6: Allocation of Recurrent Spending across EducationalLevels34 100% .-e - 80% -5 60% w f 40% e e-g c 20% L 0% I 1996 1997 1998 1999 2000 2001 2002 2003 Source: di Gropello (2004). 5.14 Patterns o f public education spending, along with the sector reforms focused on expanding basic education coverage in rural areas, have contributed to recent education gains among the poor. In addition to the overall focus o f public spending on basic education, analysis o f the benefit incidence o f public spending on basic education indicates that spending is "pro-poor", where pro-poor is defined as the poor receiving a greater share of the benefits o f public spending than their share o f the population. This can be seen from Figure 5.7, where the benefit incidence o f public spending on basic education in 1995 and 2002 is broken down by quintile. In both years, the poorest quintiles clearly receive the largest share o f the benefits o f public spending and more than their quintile population share (of 20 percent). Moreover, comparison of the benefit incidence in 1995 and 2002 indicates that public spending on basic educationhas become slightly more pro-poor over time. 34Only recurrent costs are disaggregated across levels, so the shares shown here are calculated interms of the overall recurrent budget. This includes central administration costs and other recurrent costs that could not be attributed to any of the levels. 40 Figure5.7: DistributionofPublic Subsidyto BasicEducation, By Quintile,1995 and 2002 60 50 Share of Basic Education Subsidy, 2002 National Income Share 40 30 20 10 0 1 2 3 4 5 Quintiles Source: Adapted from di Gropello (2004). -which has become increasinglypro-poor- 5.15 These patterns of pro-poor spending at the basic level are driven in large part by the strongly pro-poor patterns o f benefit incidence in rural areas.35 In 2002, for example, rural households in the poorest quintile received over 40 percent of the benefits of public spending on basic education received by all rural households (di Gropello 2004). In contrast, the benefits of public spending on basic education in urban areas go more to the middle 3 quintiles, with the middle quintile (20 percent) receiving roughly 28 percent of the benefits. Even so, it is important to note that public spending on basic education serves, at the margin, to improve the national distribution of income: this, in the sense that the urban poor receive a share of the benefits of spending that is larger than their share o f national income (see Figure 5.7). -although spending on secondary educationstill tends to benefit the middle class. 5.16 Incontrast to basic education, public spending on secondary education tends to benefit the middle quintiles, especially those inthe 3rd and 4th quintiles (Figure 5.8). As in the case of basic education, the distribution o f benefits of public spending on secondary education has become more pro-poor between 1995 and 2002. Moreover, public spending on secondary education serves, at the margin, to reduce income disparities, as the poor (particularly the moderate poor) receive a greater share of the benefits than their share innational income (Figure 5.8). 35This i s an impact of EDUCO, which focused on the most isolated, rural communities as well as from other MINEDprograms that gave priority to the rural areas where poverty was more pervasive andwhere education enrollments were lowest. Moreover, a GOES school lunch program has focused almost exclusively on rural areas (Marques 2004). 41 Figure5.8: Distributionof PublicSubsidy to SecondaryEducation, By Quintile,1995 and 2002 60 I =Share of Secondary Subsidy, 1995 50 Share of Secondary Subsidy, 2002 -A- National Income Share 40 30 20 10 0 1 2 3 4 5 Quintiles Source: Adapted from di Gropello (2004). 5.17 As in the case o f basic education spending, looking at the distribution o f the benefits o f public spending separately for rural and urban areas is instructive. In fact, in rural areas alone, the distribution of public spending can be considered pro-poor, as those in the 1st and 2nd quintilesreceivednearly 25 and 30 percent ofthe total benefits, respectively, in2002 (di Gropello 2004). In urban areas, however, the benefits o f public secondary spending were concentrated among those inthe 3rdand 4th quintiles. Onaoinnchallenges 5.18 In spite of recent advances, important work remains to be done to ensure that education is available and "works" for all Salvadorans. Despite increases in public spending, expenditure levels remain low by regional standards. Moreover, cross-country analysis indicates that El Salvador's performance in several areas - basic literacy, secondary school enrollments and average years of schooling - remain lower than expected given the country's per capita income levels. Most dramatically, in 2000, secondary school enrollments were still only half o f what would have beenexpected, given El Salvador's income. 5.19 In terms o f an agenda for education to reduce poverty, El Salvador faces several key challenges. Specifically: 0 despite overall improvements in basic education coverage and a narrowing of the gap between the poor and non-poor, over 10 percent o f all poor children still do not attend school and another 10percent are not on track for their age;36 36In2002, the gross basic enrollmentratewas about 89 percentbetweenthebottomtwo quintiles, indicatingthat approximately 11percent o fpoor children do not attend school (di Gropello 2004). The difference between the gross enrollment rate and the net enrollment rate (Figure 5.la) indicate the proportion o f children who are not "on track", Le., not attending the correct grade for their age. 42 a only about 20 percent o f poor children continue on to secondary school, compared to 36 percent among the non-poor (Figure5.lb); less than 50 percent of the poor who enroll in school do not complete the second cycle o f basic education, while less than 25 percent complete the 31dcycle o f primary (Figure 5.9); only about 16 percent o f the poor who enroll insecondary school complete that cycle; Figure5.9: Basicand Secondary CompletionRates, by Quintile, 2002 90 80 70 60 CI 2 50 40 30 20 10 0 Basic - 1st & 2nd Basic - 3rdCycle Secondary Cycles Source: di Gropello(2004) and, while the quality o f education is a system-side challenge inEl Salvador (di Gropello 2004; Marques 2004), education quality appears to beparticularly so among the poor. The costs of schooling are still a barrier among thepoor- 5.20 Access and afordability. A large share o f Salvadoran youth does not attend school because they cannot afford it. Basic education in public schools is, in principle, free o f charge. In practice, however, students have not only indirect costs, such as time that i s not dedicated to work, but also direct costs with transportation, materials and contributions to the different school activities. Untilrecently, school directors frequently asked parents to contribute with payments to help finance school spending. In some cases, parents even paid the salary of the teachers until MINEDcould assign teachers to the school. Inother cases, parents would contribute to school maintenance, purchase o f equipment or materials, or to finance extra-curricular activities. In late 2003, however, the National Assembly prohibited schools from asking parents for tuition payments (cuotaspur matricula), and MINEDbegan transferring additional resources to schools (bono degratuidad)to compensate partly for this loss of financing. 5.21 As for public secondary schools, students are charged a monthly tuition that varies between US$ 1to US$ 15. The amount o f the tuition is determinedby the School Councils. Low-income students may be totally or partially exempt. Students in the technicalhocational education track usually pay higher fees than those inthe academic track. 5.22 When asked, as part of the EHPM 2002, why they did not attend school, about 38 percent of the youth cited either that it was too expensive, that they needed to work in the labor market or 43 at home, or they reportedfamily reasons associated directly or indirectly with affordability (Table 5.2). The percentage o f youth giving these money-related reasons for not attending school increases very rapidly with the age o f the cohorts: for those in upper basic and secondary education age group, they reach 48% and 64%, respectively. The magnitudes and patterns are similar between urban and rural areas. ' I No School Doesnot I nearbyor no School want; It is Too Needs to Domestic Family Sum available damagedby Sum Parentsdo , not ~ Years Expensive work works reasons 1-4 Place earthquakes '6-7 notwant interested Other Total 1 2 3 4 5 6 I 8 9 10 11 12 ~ Total 20.5 9.8 3.7 4.2 38.2 2.9 0.1 3.0 1 10.8 17.0 31.0 1I 100.0 -but lack of availability of schools does not appear to be a major problem. 5.23 In contrast, school availability does not appear to be a major reason for not attending school: only 3% o f the youth give as reason for not attending school that there is "no school nearby or no space available", with the percentage being somewhat higher in rural areas relative to urban areas (3.8 percent versus 1.8 percent). The percentage o f youth at 3rd cycle (of basic education) age and at secondary school age giving this latter justification i s lower than for the primary age group. This is surprising in view o f apparent classroom constraints at these levels, particularly in peri-urban areas (di Gropello 2004); it may be that other reasons, such as the costs o f schooling, overshadow the issue o f school availability in the minds o f children and their families. It should be noted that the unavailability o f schools owing to the 2001 earthquakes is no longer a significant reason for children not attending school, as over 95 percent o f schools damaged by the earthquakes have now beenreconstructed. Lack of enthusiasmfor schooling by children may reflect lowperceived return to education 5.24 Lack o f interest in schooling does appear to an important reason for not attending school, however. Over 20 percent o f children age 7 and above indicate "lack o f interest" as a factor affecting school attendance, with over one-third o f children age 13-15 years (the age for the 3rd cycle o f basic education) reporting lack o f interest as the reason for not attending school (Table 5.2). Why do so many youth have such little interest in studying? A recent study by the World Bank estimates that the private rate o f return to an additional year o f schooling differs significantly by level o f schooling, a factor that may influence perceived value and, thus, interest 44 in schooling.37 For primary schooling, the yearly return is estimated at close to 7 percent, while for the 3rd cycle o f basic it is only estimated at 4.2 percent. At the secondary level, estimated returns are 11 percent per year, while returns to higher education are even higher. The study concludes that without public intervention, it i s unlikely that increases in 3rd cycle or secondary education coverage will be quickly forthcoming. 5.25 Existing public programs to reduce the demand constraints facing the poor have had a positive, albeit limited, impact. There are three scholarships program: 1) the FANTEL scholarships for higher education studies in El Salvador or abroad, which are financed with resources from the privatization o f the telecommunication company, ANTEL; it awards about 200 scholarships a year; 2) the Rodriguez Porth scholarships for low-income/high-achieving secondary graduates that finances about 80 scholarships a year;38and 3) the secondary education scholarships financed with the support o f a World Bank loan; the latter has awarded about 4,000 scholarships over the last 4 years and about 500 scholarships in 2003. While scholarships and related measures to reduce demand-side constraints have great potential to improve schooling outcomes among the poor, their impact to date has been limited largely be a relative lack o f resources.39 Quality of education may be lowerfor poor children 5.26 Quality. An ongoing GOES priority for the education sector is the improvement o f school quality. And while improving quality i s an important challenge for the entire sector, data indicate that this i s a particular challenge with respect to the poor. Data on El Salvador's PAES test scores, by department, indicate that average educational achievement is systematically worse in lower income departments (Figure 5.10). While these patterns could reflect other factors besides education quality - for example, the household learning environment - at least some portion o f this difference is likely attributable to differences inthe quality of education across poor andnon- poor areas. ~ ~~ 37 World Bank, El Salvador Country Economic Memorandum, Report No. 26238, 2003, Chapter VI, Volume 11. 38 The Presidency o f the Republic also gives Excellence Awards (US$ 114 per student) to the 500 most outstandingbasic education students. 39Inaddition, inthe case of the FATEL scholarships for higher education, they arenot targeted toward the poor, per se. 45 Figure 5.10: PAESTest Scores and Per CapitaIncome,by Department, 2000 PAES Scores and Per Capita Income, by Department, ZOO0 5.4 - A A, 160 Per Capita Income (&$/month) Scores -Log. (Scores) Source: Marques (2004). 5.27 Analysis o f the EHPM data indicates, moreover, that returns to education in the labor market are systematically lower among the poor than among the non-poor. Specifically, the returns to each year o f basic and secondary education are about half a percentage point lower for workers from households in the poorest quintile than for those from the wealthiest households. Moreover, the gap in the marginal returns to college education between the poorest and richest households is nearly 1 percentage point (Arias 2004).40 Again, these differential returns to education among the poor and the non-poor may reflect one or more of several factors that cannot be measured easily in the available data, such as school quality, labor market connections, or the household learning environment in poor versus non-poor households. Nonetheless, the evidence suggests that at least some portion o f this difference is due to a lower quality of education receivedby the poor, on average. Directions for public policv Fostering an educationsystem that worksfor all Salvadorans,including thepoor- 5.28 The new government's plan - Safe Country, 2004-2009 - calls for the development and implementation o f a new National Education Plan 2004-2021, which can continue to command a broad-based consensus and that will serve "as a renewed vision for sector development" for strengthening education access and quality. To promote a system of education that works for all Salvadorans, includingthe poor, it will be important for the Government to take measures to: 0 Continue to improve access to basic education among the poor. 0 Increase completion and graduation rates at basic level. 40While half a percentagepoint or one percentage pointper year may not sound like a large amount, the effect is cumulative and, so, buildsup over the course o f the many years o f education from basic to universitylevel. 46 0 Undertake efforts to bring more grade 9 aged students from poor households into the secondary education system, as part o f a broader effort to increase secondary enrollments and achievement for all Salvadorans. 0 Improve education quality at all levels o f education. 5.29 These measures will be important to strengthen the capabilities o f the poor and to ensure that the poor are able to take best advantage o f emerging economic opportunities in El Salvador. Moreover, these measures coincide largely with the education agenda spelled out in the World Bank's recent Country Economic Memorandum (2003) to promote faster growth in El Salvador. So there are important synergies with respect to education-for-growth and the education-for- poverty reduction agendas. -calls for measures to improve access, affordability, and quality of education- 5.30 To ensure that the poor have increased access to education, it will be important to take measures to ensure that the most poor and vulnerable households can afford to send their children to school. Inthis context, increased investmentsintargeted scholarships and other programs to reduce the demand-side constraints facing poor children and their families is warranted. 5.31 Although households do not commonly report lack o f availability o f schools as an important constraint to school participation, evidence from a recent World Bank study on the education sector (di Gropello 2004) does suggest that this is still an issue in some circumstances. For example, although the recent expansion o f primary school infrastructure in rural areas has largely removedprimary school availability as a major barrier among the rural poor, there i s some evidence that a lack o f classrooms remains a problem in some peri-urban areas. Moreover, lack o f availability o f schools in some rural areas does remain an important issue at the secondary level. As such it will also be important to strengthen programs o f school construction and rehabilitation in areas where supply constraints appear to be binding. In this regard, there may also be scope for developing public-private partnerships to exploit excess capacity o f private schools in some areas (di Gropello 2004). -which will require additional resources- 5.32 Meeting the education goals outlined above will require additional resources. Simulation analysis was undertaken to estimate the likely fiscal costs o f achieving universal primary school completion (Le., the first two cycles o f primary education) as well as to assess the likely costs o f attaining universal basic coverage (Le.' all three cycles) pZus 70 percent coverage at the secondary school level by 2015. Cost estimates for achieving universal primary school completion (MDG2) were made assuming the introduction o f additional scholarships for the poor as well as the implementation o f quality enhancing measures. Inthe case o f achieving universal basic coverage and 70 percent enrollments at the secondary level, several scenarios were run using alternative assumptions about the roles o f public versus private provision o f education and about the use o f traditional versus non-traditional models o f the provision o f education (e.g., EDUCO). 41 5.33 The estimates suggest that achieving 100 percent completion rates at the primary level could be achieved if spending on primary education were increased by about 0.5 percent o f GDP between now and 2015. If spending increases were only focused on primary schooling, this would imply increasing education sector spending from 3.2 percent o f GDP (realized in 2003) to 3.7 percent in 2015. Financing the more ambitious goals o f universal basic coverage, plus 70 percent enrollment at the secondary level, while ensuring the necessary quality improvements, would require larger increases in education spending. Indeed, the simulations suggest that it would be necessary to increase the overall education budget to about 5 percent o f GDP by 2015. 41For further details on assumptionsand methodology,see di Gropello (2004). 47 As spending increases, it would be necessary as well to increase the share o f upper basic and secondary education spending in the total education budget (to about 17% and 15%, respectively, by 2015). -and strategicpartnerships between theprivate andpublic sectors. 5.34 While increased public spending will be important to improving education outcomes in El Salvador - both nationally and among the poor - the different scenarios estimated highlight the importance o f the private as well as the public sector in making education work for all Salvadorans. Indeed, the analysis suggests some potential for cost savings as well as scope for spending efficiency gains through strategic partnerships between the public and private sectors in the financing andprovision ofeducation services.42 VI. HEALTH Good health isfundamental topeople's well-being and a critical input intopoverty reduction 6.1 Good health is a basic part o f people's well-being and is valued for its own sake. Good basic health is also essential for people to participate fully and productively in society and inthe economy and is, thus, a critical input into poverty reduction. There i s no simple relationship between health sector policy and programs and health outcomes, however. Health outcomes are a function o f a multiplicity o f factors, including the availability and quality o f health care services, access to potable water, family health knowledge and health care practices, as well as household income (ability to afford health care, including proper medications, and so on). Health sector actions are, thus, one o f several key inputs into improving basic health outcomes among the poor. 6.2 Inchapter 7, the role of access to potable water and sanitationinhelpingto improvehealth outcomes and reduce poverty is discussed. This chapter focuses more on the health sector. Specifically, this chapter examines: (1) recent developments inhealth outcomes; (2) the role that public actions have played in recent progress, with special emphasis on public education spending; (3) key challenges for El Salvador with respect to ensuringbasic health for the poor; and (4) policy directions for meetingthose challenges. Recent developments Goodprogress in improving most health care indicators- 6.3 Duringthe 1990s, El Salvador made progress inlife expectancy, infant mortality rates and children immunization, but progress in reducing child malnutrition has lagged. Table 6.1 shows key health indicators for El Salvador, Costa Rica, and LAC. El Salvador's life expectancy and infant mortality rates improved during the 1990s and these indicators are now similar or better than LAC. Moreover, they are higher than would be expected given El Salvador's per capita income levels (Todd and Hicks 2003). The prevalence o f chronic child malnutrition (height for age) and o f underweight children under 5 (weight for age) have also declined, although more slowly. While region-wide comparisons are not possible, data from Costa Rica indicates that malnutrition remains considerably higher in El Salvador. Indeed, nearly one in five Salvadoran children still experience chronic malnutrition (or "stunting"). HIV/AIDS prevalence is estimated 42Among other things, there appearsto be scope to take advantageof existingexcess capacity (supply of classroomspace) amongprivate schools inEl Salvador to promote greater educationcoverage, including amongthe poor. For further details, see di Gropello (2004). 48 at 0.7 percent of the population, which is about average for Latin America and for Lower-Middle Income Countries more broadly.43 75.4 77.6 13.7 9 26 11 6* d a 2.8 d a d a 0.6 1 Latin America ! 67.9 70.7 39.5 27.5 84 34 19* d a 9* d a d a 0.7 1 Lower Middle Income 65 69 55 33 83 41 17* d a 10* d a d a Countries OS7 6.4 Several other indicators, which can generally be taken to reflect health service access and delivery, have also improved over the decade (Table 6.2). The proportion of births attended by trained health providers has increased, as has the prevalence of contraceptive use. DTP immunization in 2002 was 93 percent for children under 12 months, a drop from 98 percent in 1990 -although it is still above the Latin America and Caribbean average.44 Table 6.2: Health Service Access Indicators, early 1990s- early 2000s !1 Birthsattendedbv 1 i Measles 1 DTP Immunization Contracedve ~ ~ 1 skilled health staif Immunization % o f childrenunder PrevalenceL(% o f (% oftotal) % Childrenunder 11 12 months women age 15-49) ~ ,1 I 1I 12months 1993 ' 2002 1993 1 2002 1993 2002 1993 1 2002 1 ElSalvador I 1I ' I I 1 I 51 69 98 99 98 I 93 53 67 Costa Rica I 1L Lower d a 98* 90 82 95 88 d a 80 Latin America I d a d a 76.7 1 91.1 70.7 88.6 d a 40 1 Middle Income Countries , d a d a d a I 84 d a 83 d a I 61 1 43This average hides highprevalence among one particularly highrisk group, men who have sex with men. A recent PASCA Multi-site Survey found the HIV prevalence to be 17.7 percent in El Salvador, higher than in Honduras (13 percent), Panama (10.6 percent), Guatemala (11.5 percent), and Nicaragua (9.3 "ercent).N o data are available yet for Costa Rica. According to MSPAS there has been no recorded case o f measles since 1996, when the triple viral vaccination (MMlUmeasles, mumps, and rubella) was introduced; in 2002, the ministry introduced the multipurpose vaccination which protects children against five different diseases. 49 -but there remain notable differences across regions and socio-economic groups 6.5 Despite the general progress in health indicators, substantial differences remain in health outcomes across different regions and socio-economic groups. The prevalence o f child mortality is 52 percent higher in rural areas than in urban areas (Fiedler 2004). And although the gap has been closing in recent years (Figure 2.5), infant mortality is almost twice as highfor low-income families as for those with higher incomes (FESAL-2002/03). Chronic malnutrition among the poor is about 40 percent higher than average national levels, with a wide variance across regions. Over 15 percent o f children o f low-income families are underweight, while less than 4 percent o f children from wealthier households experience low weight for age.4s 6.6 Similar patterns across socio-economic groups are found with respect to health service access. Salvadorans in the highest income quintiles are one-third more likely to receive medical care when illthan those in the poorest q~intiles.~`FESAL-2002/03 indicates that about 20 percent o f expectant mothers from low-income households had n o prenatal care visits, compared with only 4 percent o f expectant mothers from better-off households. And, while 93 percent o f the better-off mothers delivered in a hospital, only 52 percent o f low-income mothers had institutional deliveries. Recent vublic action El Salvador'spublic health system is comprisedof multiple institutionsat multiple levels 6.7 The health sector in El Salvador comprises the MSPAS, the Salvadoran Social Security Institute (ISSS), the military health facilities, the private sector and over 30 NGOs that provide mostly primary healthcare services. Total health spending has been distributed as follows: 22.1 percent MSPAS; 18.4 percent ISSS; 56 percent household out o f pocket expenditures; 2.4 percent private insurance; and 0.2 percent NGOS.~'ISSS is financed by payroll taxes and caters to formal sector employees and their families covering about one million people (17 percent o f total population); MSPAS is financed from government revenue and external loans and grants. 6.8 MSPAS has 30 hospitals (27 general hospital and 3 specialized), 367 health centers, 171 health clinics (casus de salud), and 47 rural health and nutrition centers, and 2 centers for emergency attention; the ISSS has 16 hospitals, 70 health centers and 170 enterprise clinics; the military has three hospitals; and the private sector has 36 hospitals. ISSS has rented several MSPAS facilities, including most recently a full hospital (Hospital de Especialidades) in San Salvador48;ISSS rental o f MSPAS hospitals was scheduled to end at the end o f 2004, however. 6.9 The public health infrastructure was severely damaged by the 2001 earthquakes. The first earthquake on January 13 left two MSPAS hospitals severely damaged and another six hospitals partially damaged, requiring full or partial evacuation. The most badly damaged hospitals continued to operate under provisional structures, while the rest reduced their operations. The second quake on February 13 left more than 2,000 hospital beds out o f service, which compares 45 As noted in Chapter 2, the FESAL publications differentiate outcomes across three different socio- economic levels: "low", "medium", and "high". This contrasts with the standard analyses undertaken using the EHPM data. Those analyses compare outcomes across the "poor" and "non-poor", basedon the official Salvadoran poverty line or examine outcomes across income quintiles o f the population. 46 An Earthquake Emergency Reconstruction and Health Service Extension Project, Project Appraisal Document (Report NO. 22626-ES), October 31,2001. 47 InstitutionalDevelopment Strategic Plan, MSPAS, October 2001 48 The Hospital de Especialidades (or Specialty Hospital) was built after the 1986 earthquakeswith a grant from France. It was to replace the Rosales Hospital that was severely damaged by the earthquake. The authorities decided not to close the Rosalesand since MSPAS hadno resources to operate both hospitals, it decided to rentthe HospitaldeEspecialidadesto ISSS. The proceeds from hospital rental have gone to the SalvadoranTreasury. 50 to the 4,843 MSPAS hospital beds in 2000.49 Altogether, the two earthquakes damaged 24 hospitals, 98 health centers, 4 health clinics, and 5 rural health andnutrition center^.^' Achieving consensus on the direction of health sector reforms has been elusive 6.10 Incontrast to the education sector, there has been no national consensus on the direction o f reform o f the health sector. Duringthe 1990s there were several proposals to reform the sector that did not prosper. In 2000, a Presidential Commission for Health Sector Reform produced a diagnostic o f the sector and some guidelines for sector reform5' but failed to reach agreement on key issues, such as those related to the financing o f the sector. In September 2003, the President appointed another Commission to follow up on the Reform o f the Health Sector, which has organized five working groups (Comisiones Especiales) to seek a broader consensus on the direction o f the reform. These working groups are on: financing, human resources, organization, social participation, and legal issues. 6.11 Historically, hospitals in the major urban centers and particularly in the capital city o f San Salvador absorbed the bulk o f public health resources in detriment o f primary healthcare and the rural areas, where most o f the poor live. During the 1990s, MSPAS sought to improve healthcare services in rural areas by expanding infrastructure, hiring health promoters and training midwives. While progress was made, especially in terms o f immunizations and infant mortality rates, many remain without access to health services. Indeed, the presence o f MSPAS promoters in the communities has had only limited impact on access to basic health care among the poor. Inaddition, efforts to increase demand for basic health services by the poor have been hampered by perceptions o f low service cpality, in part because o f the low and irregular availability o f and long waits for medication. Efforts have been made to decentralize the system- 6.12 MSPAS' Institutional Development Strategic Plan o f October 2001 seeks to address these problems by changing the organization and delivery o f health services through: i)a community- based approach to primary care provision; ii)a functional planning approach that specifies the functions (such as the projected volume o f procedures, equipment requirements, patient flows, service definition, numbers and categories o f staff, and inter-hospital referrals) that will take place in the health facilities; and iii)closer ties between hospitals and primary care providers by decentralizing a subset o f management functions, improving the referral system and raising quality o f care.53 6.13 The cornerstone o f the Plan is the establishment o f Basic Systems o f Integral Health Attention (SIBASI) or health district units, which would replace the Departmental management units, and the introduction of a new model of health service management inwhich the financing and the provision o f services may be performed by different institutions. The SIBASI would be the basic operative structure of the reformed public health service. MSPAS will continue to have the responsibility for the financing but the provisiono f services may be done by an NGO, as well 49 For estimates of the damage, Earthquake Emergency Reconstruction and Health Services Extension, Project Appraisal Report, World Bank, page 3; for number of existing hospital beds, Institutional DevelopmentStrategic Plan, MSPAS, October2001. Dataprovidedby MSPAS, October 2004. 51The agreed guidelinesare: Consolidatea National Health System; consolidatean attentionmodel based on the promotion, prevention and primary health care; consolidate a mixed provision model (public and private); consolidate a model under the direction of the MSPAS; institutionalize social participation and decentralization as transversal systems; invest in human resources for health; strengthen intersectoral coordination; and guarantee basic services to all Salvadorans. See Earthquake Emergency Reconstruction and Health Service Extension Project, Project Appraisal Document (Report NO. 22626-ES), October 31,2001. See also Table 6.7, below. 53Ibid. 51 as by the private sector. In addition, the Strategic Plan calls for developing and implementing performance agreements and contracts and an incentive scheme to make health providers accountable to both the MSPAS and service users and introducing community participation in basic health and nutrition service delivery.54 -and to strengthen community-based approaches toprimary careprovision 6.14 MSPAS has established 28 SIBASI and has prepared the operational protocols. The purpose is to provide integral services to the population through the coordination o f primary healthcare and general hospital care in the delimited geographic areas under the responsibility of each SIBASI. Within the SIBASI area, a network o f health providers and a general hospital will work in a coordinated manner, avoiding duplication o f efforts, and referring cases to the three specialized hospitals in San Salvador, through a National Reference Center. The impact o f the health providers' work on the health status o f the population under their responsibility will be monitored. The management functions relatedto planning o f activities, provision o f services and decisions concerning human, technological, and financial matters are being transferred to the SIBASI. Each SIBASI has a Managing Committee, which is advised by a Social Consultation Committee. The Managing Committee comprises representatives o f the SIBASI's health providers and the SIBASI manager. Technical, financial and administrative teams support the manager. The SIBASI health providers include NGOs, the ISSS, the MSPAS, the private sector and others. The organizational development o f the 28 SIBASI varies. In some SIBASI there i s already a strong NGO presence and community participation; in others the organization i s still incipient. But lack of data has made it difficult to assess the impact of recent reforms 6.15 Implementation o f the Strategic Plan and, in particular, the SIBASI model has potentially important implications for the performance o f El Salvador's public health system. Yet, relatively little systematic information exists to date on the extent to which the Plan has been implemented and, more importantly, on how the reform is contributing to (i) the overall performance of health system, and (ii) improvedaccess and quality o f healthcare for the poor. Inthis context, it will be important to review the institutional reform process to assess its impact on the availability and quality o f care, the balance between primary healthcare and hospital care, and the overall management o f the public health system. Total health spending (publicplus private) is high by regional standards- 6.16 Health spending. Total Salvadoran health spending, including private and public spending i s relatively highby LatinAmerican standards. The World Development Indicators (2003) reports that El Salvador spent 8.8 percent o f GDP on health in 2002 (3.8 percent public spending and 5 percent private spending), nearly 2 percentage points higher than the average for Latin America and the Caribbean (LAC) of 7 percent (Figure 6.1). This figure represents, in part, highprivate spending on health care. When only public sector spending is taken into account, El Salvador's rank in the region falls - although public health spending is still slightly above the regional average (Figure 6.2). 54 "SIBASI: Marc0 Conceptualy Operacional", versionrevisada, MSPAS,2001. 52 Figure6.1: TotalHealthSpending (Publicand Private) as a Proportionof GDP, 2000 PO 130 t; 80 22a3 60 4 0 2 0 nn Note: Total health expenditure i s the sum o f public and private health expenditures. It covers the provision of health services (preventive and curative), family planning activities, nutrition activities, and emergency aid designated for health but does not include provision o f water and sanitation. Private expenditure includes out-of-pocket and private health insurance plans. Source: World Bank, WDI, 2003. Figure6.2: Public HealthSpendingas a Proportionof GNP 2000 Source: World Bank, WDI, 2003 53 -although public health spendingper capita is relatively low 6.17 Given El Salvador's low GDP per capita, its total health spending per capita is much lower than Costa Rica or LAC, h~wever.~'Figure 6.3 shows that in 2000, Costa Rica and L A C spent 52 and 42 percent more on a per capita basis than El Salvador. Since public spending inEl Salvador is much lower than Costa Rica, per capita public spending is just a fraction o f that country, which impact on the coverage and quality o f public health services, particularly o f the poor. Figure 6.3: Health Spending Per Capita, 1997-2000 2 600 500 400 3 E 300 5 200 0 100 Source: World Bank, WDI, 2003. 6.18 MSPAS estimates total health spending at 8.0 percent of GDP in 2001 (Table 6.3). This i s somewhat lower than the figure for 2000 reported in the World Development Indicators (2003), although the split between public and private spending is generally consistent across the two sources. O f total health spending, 46 percent was public spending - o f which 47.4 percent came from the treasury (e.g., MSPAS, military, teachers, etc.), 44.5 percent from ISSS, 2 percent from external sources, and 6.1 percent from other sources (including, for example co-payments by beneficiaries). 55 In 2000, WDI shows total population of El Salvador and Costa Rica at 6.3 million and 3.8 millions, respectively; El Salvador GDP at US$ 11billion and Costa Rica GDP at US$ 14billion (in 1995 US$); and per capita GDP inEl Salvador, Costa Rica and LAC at US$ 1,752, US$3,912 and US$3,856, respectively (all in 1995 US$). 54 Table 6.3: Public and Private Health Spending and Sources of Financing, 1996-2000 i 1996 1 1998 2001 1 Total Spending % GDP 7.6 1 8.3 1I 8.0 1 ~ ' Public Spending YOof Total I I1 41 I 42.5 46.4 Central Government Financing ', % public (44.6) (47.1) (47.4) ~ ~ spending External resources,% public spending (11.9) 1 (6.8) ! (2) 11Other, %public spending b' (0.6) (4.4) 1 (6.1) Private Spending % of Total I! 59 I1 57.5 1 53.6 I1 1 1 Private Insurance % private spending (2) I (3.3) 1 2.6 I Out-of- Pocket, %private spending 1j (97.7) (96.7) 1 97.4 I ~ a/ Central Government FiscalResources. b/ Other sources may include co-paymentsby the beneficiaries. Source: Marques (2004), citing CuentasNacionales en Salud, MSPAS, mayo 2003. 6.19 The MSPAS budget has remained a relatively constant share o f GDP in recent years, hovering around 1.6 percent. As a share of the Central Government budget, however, it has increased somewhat -from 7.8 percent in 1996 to 9.7 percent in2003 (Figure 6.4). For 2004, the proposed MSPAS budget is equivalent to 1.6 percent of GDP and to 9 percent o f the Central Government budget.56 Figure 6.4: MSPASExpenditures,1996-2004, as a Percentof GDP and of Total Government Spending E 12.0 2.0 E> ,a 10.0 n Q) n 0 .e 1.5 5 9 E 8.0 5 2 7 2 - a 8 Q) 6.0 1.0 g c 4.0 =$ X = 2.0 0.5 % 0 WX W 199619971998199920002001200220032004 1- % of GDP +% of Total Budget 1 Source: Marques (2004), based on data from Ministry of Finance (1996-2001: executedbudget; for 2002-2003, approvedbudgets; for 2004, proposedbudget). 56The World Health Report (2002) shows that health expenditures by the General Government, which includes the Ministry of Health and other autonomous public institutions such as the social security, as a percentage o f total General Government expenditures was much higher in El Salvador that other Central America Countries in 2002: El Salvador, 26.2 percent; Honduras, 18.3 percent; Costa Rica, 18.2 percent; and Guatemala, 16.4percent. 55 Investment spending has increasedfollowing the 2001 earthquakes- 6.20 Patterns of public spending by level of service. Traditionally, MSPAS has classified healthcare in three levels: Level Ior primary healthcare; Level I1 or general hospital care; and Level I11or specialized hospital care. Figure 6.5 presents the distribution o f MSPAS approved budget for these levels o f health care, for administration, and "other" expenditures that include transfers to other entities and investment expenditures. While the bulk of MSPAS resources continued to be allocated to hospital care (56 percent on average during the 1996-2004 period), the share going to hospitals has tended to decline; specialized hospitals' share o f total budget declined from about 19 percent in 1996 to 16 percent in 2004 and general hospitals' share declined from 41 percent to 34 percent during the same period. The share o f spending on primary health care, increased from 27 percent in 1996 to nearly 35 percent in 2000, but declined again thereafter, equaling 28 percent o f total spending in 2004. According to MSPAS data, the share allocated for administration has declined slightly over the period, from 6.6 percent o f total spending in 1996 to 5.3 percent in 2004. "Other" spending, which includes investment spending, among other things, nearly tripled its share in the total budget, from 6.6 percent in 1996 to 16.2 percent in 2004. This has been driven to a large degree by increases in investments, particularly on earthquake reconstruction. While greater investments have accompanied the declining share o f spending on hospitals, pre-earthquake spending trends also suggest that the share o f spending on primary care i s currently lower than it would have been inthe absence o f the quakes. Figure6.5: MSPASExpenditures by Levelof Attention and Administration,1996-2004 100% 80% 60% 40% 20% 0% 1996 1997 1998 1999 2000 2001 2002 2003 2004 Note: Refers to approved budget. Primary healthcare includes level I(public health and primary care in health units); general hospital care includes level 11; specialized hospitals include level 111. Administration spending refers to that incurred by MSPAS at the central government level. Other includes transfers to other entities and investment expenditures. Source: InstitutionalFinance Unit (Unidad Financiera Institucional), MSPAS. 56 -nonetheless, public health expenditureshavegenerally beenpro-poor- 6.21 Has public health spending been pro-poor? As will be discussed further below, only 17 percent o f the Salvadoran population is covered by the ISSS, and the poor make up only a small share o f those covered. Since public spending on the ISSS makes up nearly half o f total public health spending, it is unlikely that the poor benefit from a disproportionate share o f total government spending on health. Nonetheless, the evidence indicates that MSPAS spending has been relatively pro-poor. 6.22 This is due, inpart, to the fact that MSPAS facilities play a relatively more important role inthe health care ofthe poor. Indeed, while the poormakeupabout 37 percent ofthe population, they account for 53 percent of all visits to MSPAS facilities (Table 6.4). The poor's share of MSPAS service utilizationvaries quite considerably across levels o f service, however. As can be seen in Table 6.4, although the poor account for 73.5 percent o f all visits to Health Units, they make up less than half(47.5 percent) of visits to hospitals. Table 6.4: Distributionof Personsthat Visit MSPASFacilities by Income Group, 2002 i Hospitals 1 Health i Healthunits I Total Centers I iI National 100.0 100.0 100.0 100.0 Poor 47.5 55.6 73.5 53.3 Extreme poor 21.3 26.8 44.0 25.3 Relative poor 26.2 28.8 29.5 28.0 N o poor 52.5 44.4 26.5 46.7 -particularly at theprimary care level- 6.23 Analysis o f the distribution (or benefit incidence) o f public spending to primary health care i s strongly pro-poor, where, as in the case o f education, "pro-poor" is defined as the poor receiving a larger share o f the benefits o f public spending than their population share (Figure 6.6). As can be seen inFigure 6.6, the poorest 20 percent of the population receive roughly 29 percent o f the benefits o f primary health care spending, while the second quintile receives about 23 percent o f the benefits. 57 Figure6.6: Distributionof Public Subsidyto PrimaryHealthCare, By Quintile, 2002 60 I =Share of PrimaryHealthSubsidy I 50 I +National Income Share I s 40 I ' n 30 l g 9 I 20 ~ I 0 i 1 2 3 4 5 ~ Quintiles Source: Marques(2004); estimationsby FUSADES,usingEHPMdata. 6.24 While the distribution of the benefits of public spending to hospital care is enjoyed relatively more by those inthe middle quintiles, spending can also be considered to be reasonably pro-poor (Figure 6.7). While those inthe first quintile receivejust under their population share o f the benefits of public hospital spending (19.4 percent), the bottom 40 percent of the population receives nearly 43 percent o f all the benefits o f public spending on hospitals. Figure6.7: Distributionof Public Subsidyto HospitalCare, By Quintile, 2002 60 =Share of HospitalSubsidy 50 -A- NationalIncomeShare 40 30 20 10 0 1 2 3 4 5 Quintiles ~ Source: Marques(2004); estimationsby FUSADES,usingEHPMdata. 58 -and in rural areas. 6.25 As in the case o f education, the pro-poor nature o f public spending in health is driven largely by the benefit incidence o f public spending among rural households. In rural areas, for example, about 45 percent o f the benefits o f MSPAS spending on primary health care went to those in the poorest quintile in 2000, while nearly 42 percent o f MSPAS spending on hospitals went to the poorest quintile (Mason, Torres, and Aterido 2004). The corollary of this i s that in urban areas, MSPAS spending on primary health care benefits the middle three quintiles most, whereas spending on hospital care disproportionately benefits the top three quintiles. Ongoing challenges 6.26 Despite recent progress, El Salvador continues to face significant challenges in improving health outcomes, especially among the poor. Increased access for those who currently lack it, affordability, and service quality are all key issues. Moreover, several o f El Salvador's accomplishments have created a new set o f health care challenges. This is because with some health issues, such as malnutrition and infant and child mortality, it is often more straightforward to improve indicators when the problem is really extensive than when the problem exists at lower levels. For example, recent gains ininfantand child mortality have come as a result o f significant reductions in so-called "soft" mortality - inthe post-neonatal or 1-to-4 year age groups. Indeed, with these gains, neonatal mortality - or so-called "hard" mortality - has become an increasing proportion o f all infant and child mortality. But successfully addressing "hard" mortality is more difficult, and will require improvements not only in the outreach and quality o f El Salvador's primary healthcare system, but inits hospital care as well. 6.27 In the paragraphs that follow, the challenges o f access, affordability, and quality are addressedinturn. Nearly a quarter of theSalvadoranpopulation lacksfull access to basic health services 6.28 Access and Affordability. The coverage o f the health system is still inadequate. Table 6.5 shows that fewer than 20 percent o f Salvadorans have access to medical insurance. The ISSS covers 17 percent o f the Salvadoran population; private and institutional insurance another 1.8 percent; and the remaining 81.3 percent has no insurance. The uncovered population is even higher among the poor - 90.9 percent among the moderate poor and 97.3 percent among the extreme poor. It is the responsibility o f MSPAS to cover the population without insurance, especially the poor. Nevertheless, it has been estimated that 24 percent o f the population still do not have access to health services or only have limitedaccess." Table 6.5: Population Covered by Medical Insurance, 2002 (percent of total population) Source: Marques (2004), usingEHF'M data. 57 Human Development Report 2003 (UNDP). 59 6.29 Indeed, substantial differences remain between the access o f the poor and the better o f f to health services. Salvadorans in the highest quintiles are one-third more likely to receive medical care when illthan those in the poorest quintiles.58 And as noted above, there remain important gaps between the poor and non-poor in access to reproductive health care services, such as pre- natal visits and institutional deliveries (FESAL 1998). When ill,14 percent o f the poor go to private or NGO facilities rather than to MSPAS facilities (Table 6.6). 1 Poverty Level MSPAS Priv ISSS ate "ge;' ~ , Total(all) 67.5 10.2 17.1 5.3 Poor 81.3 4.5 9.6 4.6 Extreme Poor 86.2 1.2 7.5 ' 5.1 Relative Poor 77.4 7.1 11.2 4.3 NonPoor 56.5 14.7 23.1 5.7 Both direct and indirect costs can be barriers to health care access among thepoor 6.30 One-quarter o f the poor and nearly one-third o f the extreme poor do not seek help from MSPAS facilities because o f the financial costs (Table 6.7). In an effort to reduce financial barriers to healthcare access among the poor, payments to health centers were legally eliminated by Executive Decree inJune, 2002. Reports from the field indicate, however, that inpractice fees continue to be charged at many health centers. The actual extent o f this practice requires further investigation. In addition, the indirect costs o f using health centers, such as travel time and the costs o f transport also affect affordability o f health care services among the poor. Indeed, the poor report travel times to the nearest health facility that are 17-50 percent longer than those reported by the non-poor, indicating higher time (and probably financial costs) o f accessing basic health service^.^' Poverty Level Not necessary Lack of money Center too far; Does not trust; Center has no prefers to take I medicines: domestic lack of attention medicine; other Total 43.8 15.6 23.3 17.3 Poor (all) 42.2 25.3 19.1 13.3 Extreme Poor 36.2 30.1 18.8 14.9 Relative Poor 47.7 20.8 19.4 12.1 Non Poor 45.1 7.3 26.9 20.7 I 58An EarthquakeEmergency Reconstruction andHealth Service Extension Project, Project Appraisal Document (Report NO. 22626-ES), October 31,2001. 59World Bank staff calculations, using 2002 EHPM data. 60 Low quality of care appears to undermine trust in thepublic health system 6.31 In addition, one in every three poor person who gets ill,does not trust the public system andor prefers to seek cures in alternative ways. Indeed, several factors appear to adversely affect service quality and, thus, people's confidence in the public health system. This includes lack o f sufficient trained staff in public facilities. This problem is compounded by what appears to be a lack o f adequate incentives to induce enough medical professionals to participate in the public system. Only halfo f the medical staff inpublic institutions work full-time or receive all o f their earnings from MSPAS; moreover, key clinicalpersonnel spendonly about one-third to one- half o f their time in non-clinical activities.6o Together, these factors represent important challenges for improving outreach and access o f the public healthcare system. 6.32 Concerns about access, affordable, and quality are not limited to the provision o f health services, but to pharmaceuticals as well. A study by the Management Sciences for Health (MSH) found that for a sample o f critically neededmedicines, only 84 percent were available at MSPAS primary healthcare facilities and 70 percent at hospitals. The study also found that El Salvador routinely purchases pharmaceutical products at prices higher than the international median. Moreover, products o f substandard quality were found in "50 percent of samples collectedfiom publicfacilities ...and 28.6percentfrom private pharmacies". 61 In 2003, M S H proposed a new systemto manage pharmaceuticals, which involvedjoint procurement by the specializedhospitals and the SIBASI to obtain a better price and the involvement o f a single distributor to manage the purchase orders, storage and distribution; MSPAS would monitor the quality o f pharmaceutical products ordered and purchased by the network facilities. While the new system experienced some initial logistical problems, the MSPAS is continuing to address these important challenges. The 2001 earthquakes damagedpublic healthfacilities, affecting careprovision 6.33 Efforts to improve access and service quality have been complicated by the earthquakes of 2001, As noted above, the public health infrastructure was severely damaged by the earthquakes. Two MSPAS hospitals were severely damaged and six others were partially damaged in the first quake, with as much as one-third o f all MSPAS health facilities affected. Indeed, estimates suggestthat as many as 40 percent o fMSPAS hospital beds were out o f service inthe wake ofthe second earthquake. 62 The most badly damaged hospitals continued to operate under provisional structures, while the rest reduced their operations. Moreover, the government has engaged in ongoing earthquake reconstruction efforts, supported by the IDB, the Central American Bank for Economic Integration (CABEI), Japan and the World Bank.63Nonetheless, the effects o f the earthquake still affect El Salvador's ability to provide healthcare, including the types o f care that will be needed to address the challenges o f reducing neonatal (and thus so-called "hard" infant and child) mortality. ~~ 6o .Non-clinical activities include staff administrative duties, coordinationwith local authorities, and training. For further detail, see "Baseline and Best Practices Assessment o f seven SIBASI inEl Salvador," Eric Seiber, PHRPlus, December, 2002 61Cited inMarques (2004), based on reports posted on the web page o f Management Sciences for Health www.msh,org/searn/3.1.2b.htm). 62 Marques (2004), citing Earthquake Emergency Reconstruction and Health Services Extension, Project Appraisal Report, World Bank, and Institutional Development Strategic Plan, MSPAS, 2001, figures. 63 MSPAS has developed a program to reconstructhehabilitate its facilities, including 23 o f 30 national hospitals and 82 health centers. Existing IDB loans and CABEIfunds, are being utilized to cover the costs o f provisional structures and for the rehabilitation o f the 15 less badly damaged hospitals. For the reconstruction o f the eight hospitals with the greatest damage, the government has obtained the support o f the World Bank and Japan. The World Bank Emergency Reconstruction project will also support the extension o f basic services to remote areas and the strengthening o f primary health care services and facilities inareas affected by the 2001 earthquakes. 61 Directions for uublic uolicv I t will be important to build a consensus on how to improve access and coverage- 6.34 To continue to improve basic health outcomes for the poor, a key priority will be to builda national consensus on the strategic direction for El Salvador's national healthcare policies and programs. In this regard, the new Government's plan - Safe Country, 2004-2009 - calls for the reform o f the health sector in a "concerted manner which may lead to an efficient national health system, finctioning in a decentralized manner, with universal coverage and free o f charge to all persons that can not afford." -that recognizesboth demand and supply constraints 6.35 Consistent with these goals, a key policy challenge will be to continue to increase healthcare access and coverage among the poor who currently lack access. Accomplishing this will require addressing both demand- as well as supply-side constraints to access. 0 Addressing demand-side constraints. A key element o f reducing demand-side constraints involves increasingthe affordability o f healthcare services. 0 In the short-term, this should include increased efforts to enforce the Decree eliminating fee payments to health care centers. Consideration should also be givento the use of targeting mechanisms - either on the basis o f geographic location or household poverty status - to ensure the efficiency and sustainability o f such financial support to the poor. 0 Inthe longer-term, consideration should be givento broader approaches to providing financial protection against health shocks to the poor. This would include consideration o f approaches to improving access among the poor to the formal social security system and to private health insurance mechanisms, as well as exploration o f ways to "pool" or insure against family healthrisks at the community level. 0 Addressing supply-side constraints. Several supply-side measures will be important to strengthening El Salvador's health care system and, through it, health care access among thepoor. These include: 0 Assessment o f the ongoing institutional reform and decentralization plan to identify whether any mid-term adjustments would help to strengthen the reform process, particularly with respect to primary health care delivery;64 0 Review and reform o f the health sector's human resource policy, including incentives and rewards systems, to ensure a sufficient supply o f well-trained professional staff andquality care, particularly inpoor, ruralareas; 0 Reconstruction and rehabilitation o f the hospitals damaged by the 2001 earthquakes to ensure adequate supply o f secondary and tertiary health services; and 0 Measures to reduce the costs (and improve the quality) o fpharmaceuticals, particularly basic medicines. 64 Such an assessmentmight also be valuable as part of the process of building national consensus on reforms. 62 6.36 Since private sector provision o f healthcare services plays such an important role in El Salvador, strengthening the regulations governing private sector provision of services will also b e important to strengthening the overall health sector as well as to ensuring the availability o f quality care to all Salvadorans. 6.37 In spite of recent increases in public spending and levels of public spending that, as a percentage of GDP, compare well with other countries, public spending o n health per capita remains low by international standards. Indeed, as in the case o f education, continuing to make progress in improving health outcomes, particularly among the poor, will require additional resources. Estimating the resources needed to achieve specific health outcomes is difficult, due to the complex relationship between multiple factors (including, but not limited to health sector inputs) and specific health outcomes. Nonetheless, it is possible to provide a reasonable range of estimates for achieving universal access to basic healthcare services. Assuring basic health services to those who lack access will require additional resources 6.38 Recent World Bank estimates of the annual cost o f providing a minimum package of essential health and nutrition services in El Salvador, based on differentiated capitation, is estimated at between U S $15 and US$ 20 per person.65 At the same time, the cost of the basic health and nutrition package established by M S P A S (the SESYN, or Essential Heath and Nutrition Services) and based on universal capitation, which i s the basis for providing basic health care in the IDB's Support to the Modernization Health Program, is US$ 31 per person. 66 Taking U S $20 per person as a lower-bound estimate, andUS $31 per person as an upper bound estimate for the basic package o f essential health and nutrition services, the annual cost of covering the 1.6 million Salvadorans w h o still lack regular health services would range from between US$ 32 million and US$50 million per year, the equivalent of 0.2-0.3 percent of GDP, respective~y.~' 65The package o f essentialhealth and nutrition services referredto here is designed to ensure the provision o f basic maternal and child care services to poor Salvadorans and includes the following interventions: (i) maternal (pre and postnatal) care; (ii)sexual and reproductive health, including family planning and the control o f sexually transmitted diseases; (iii)child health and nutrition (for children under 5 years); (iv) health education, detection and treatment o f the most prevalent communicable diseases (tuberculosis, HIV/AIDS, Dengue) inthe general population; and (v) community health prevention, detection and referral o f the most prevalent and non-communicable diseases in adults. The package also includes a basic vademecum o f pharmaceuticals according to the list o f essential drugs regulatedby the MSPAS. For more information, see Earthquake Emergency Reconstruction and Health Service Extension Project, Project Appraisal Document (Report NO. 22626-ES), October 31,2001. 66 Earthquake Emergency Reconstruction and Health Service Extension Project, Project Appraisal Document (Report NO. 22626-ES), October 31, 2001, page 13, provided a preliminary cost estimate of US$ 15 per person per year based on the Guatemala and Honduran experiences. Recent studies prepared for the World Bank project point to a slightly higher cost o f approximately US$20 per person per year. These studies for the World Bank project are based on a differentiated capitation, where the cost o f the package for children, adults, women and man are different, while the cost estimates for the IDB project are basedon universal capitation, inwhich a single unit cost for all people, i s estimated. 67The Human DevelopmentReport 2003 (UNDP) estimates the cost o f US$ 31per personper year; with a total annual cost o f US$ 50 million to reach 1.6 million people, the equivalent o f 0.3 percent o f GDP. To eliminate malnutrition in children under 5 years, the UNDP estimates it would cost an additional US$ 23 million a year, equivalent to 0.15 percent o f GDP. This estimate is based on the number o f children with malnutrition in urban areas (29,655) and rural areas (49,841) and assuming an annual cost o f eliminating malnutrition equivalent to the urban minimum basket (US$ 385) and rural minimum basket (US$ 240). The implied total cost - at the margin - o f ensuring basic health and nutrition services for the 1.6 Salvadorans who still lack regular health coverage, according to the Human Development Report 2003, i s thus 0.45 percent of GDP. 63 VII. SOCIO-ECONOMIC INFRASTRUCTURE Access to clean water and sanitationfacilities important to health andpoverty reduction- 7.1 This chapter focuses on the role of socio-economic infrastructure investmentsas part of a strategy to reduce poverty. The chapter focuses specifically on rural water and sanitation and on ruralroads. Access to potable water and suitable sanitation facilities inrural areas is increasingly recognized as an important health input and, through that channel, a key component of a country's poverty reduction efforts. Water and sanitation are particularly recognized for their role and impact on infant health.68For example, the Ministry o f Health reported diarrhea, a key cause o f which is contaminated water, as one o f the most common infant diseases in2001, evidencedby the 20 percent o f medical consults for children under 1 year o f age, and 15 percent for children between ages one and four. Inaddition, the 1998 FESAL data indicate that the problems of infant and child mortality are significantly more severe in households without access to improvedwater than in those with access (Table 7.1). The same is true of households who lack sanitation facilities. While the data presented in Table 7.1 are descriptive, the same relationshi s are valid controlling for income and a range o f other household and community characteristics. 8 Table 7.1: Infant and ChildMortality, by Access to Water and Toilets (March 1993 to February 1998) Total Neonatal Post neonatal 1-4 years old 0-4 years old Yes 30 17 13 4 34 No 40 16 24 12 51 Total Neonatal Post neonatal 1-4 years old 0-4 years old Yes 30 15 14 3 32 No 37 17 20 10 47 Source: Yepes (2004). 7.2 Investments in rural roads are also a key part o f a poverty reduction strategy. Increased access to all-weather roads brings numerous benefits, such as improved access to schools and healthcare facilities and other basic services, reduced transactions costs (which enable greater access to and participation in market-based activities), as well as access to non-farm employment opportunities. -as is the accessprovided by all-weather roads 7.3 As discussed in Part Io f the report, there is a robust empirical relationship between proximity to all-weather roads and transport infrastructure and the growth o f household incomes inrural El Salvador (see also Beneke de Sanfeliu and Gonzalez-Vega 2000; Beneke de Sanfeli6 and Shi 2004). Analysis o f reductions in poverty at the canton level supports household-level Leipziger, Fay, Wodon, and Yepes, 2003. "Achieving the Millennium Development Goals: The Role o f Infrastructure". 69Ibid. 64 findings about the linkbetweenrural road investmentsand household incomes, especially among the very poor.70 A recent study used the BASIS data to match households in cantones that received road improving investments with those that had no such investments between 1999 and 2001 (Yepes 2004). Cantones with and without road improving investments were matched according to a number o f characteristics, including average distances o f households to paved roads in 1999 to ensure appropriate comparability. The study finds that although total poverty declined at similar rates in cantones with and without road improvements, extreme poverty declined significantly faster in cantones that received road investments. The impact was both statistically and economically significant. Extreme poverty declined 5.1 percentage points more in cantones with improved roads than in those without. These findings suggest that over the 1999-2001 period, at least, road investments had a particularly salutary effect on the incomes of the extreme poor. 7.4 As in earlier parts of the report, this chapter discusses: (1) recent developments in extending rural water and rural roads to the poor; (2) the role that public actions have played in recent progress; (3) ongoing challenges for improving access among the rural poor; and (4) policy directions for ensuring that investments inroads and water contribute powerfully to El Salvador's poverty reduction efforts. The chapter examines first rural water and sanitation and then turns to ruralroads. Water and Sanitation Recent developments 7.5 El Salvador has made significant progress inthe expansion ofwater services and increased water access over the past decade. Access (or "improved" access) i s defined as water supplied by public tap or pipe inside or outside the dwelling. "Reasonable access" to drinking water - defined at as the availability of at least 20 liters a person per day from a source within one kilometer o f the dwelling71- increased significantly inrural areas, contributing to a closing o f the gap with respect to the cities. While there have been significant gains in access to clean drinking water since 1991- 7.6 As shown in Figure 2.6, household access to improved drinking water increased at the national level from 54.9 percent in 1991 to 75.9 percent in 2002. In urban areas, access grew from 80.2 to 92.6 percent, while in rural areas, access increased by over 20 percentage points - from 27.5 to 47.6 percent - over the period. Although access in rural areas continues to be low relative to urban areas, the gap is closing (Figure 7.1). The average annual growth in rural coverage was 2.2 percent between 1991 and 1995, 1.3 percent between 1995 and 1999, and 2.8 percent between 1999 and 2002 - equaling or exceeding urban growth ineach sub-period. 70Cantones are administrative sub-divisionsofmunicipalities. "WHO andUNChildren'sFund, 2000. Global Water Supplyand Sanitation Assessment Report. 65 Figure 7.1: Annual Increase inAccess to ImprovedWater (YO) 3 2 5 2 c$C 1.5 n0 1 0.5 0 1991-95 1995-99 1999-2002 Years Source: Yepes (2004). 7.7 The main sources o f drinking water in rural areas also improved over the period (Table 7.2). Drinkingwater from improved sources increased by 20 percentage points (row A). Water access from wells remained largely the same during the 1990s, but then increasing slightly inthe 2000s. Correspondingly, access via unimproved water sources, such as springs, river, rainwater, or vendors, declined from 49.1 percent in 1992 to 26.3 percent in2002. Table 7.2: Sourcesof Water inRuralAreas, 1992-2002 49.1 40.3 34.8 26.3 -8.8 -5.5 -8.5 Source: Yepes (2004), usingEHPMdata (various years). 7.8 Efforts to improve rural water access between 1991 and 2002 have yielded significant benefits to the poor. For example, the proportion of poor who obtain their water from inferior sources, such as rivers and other sources, declined about 60 percent over the period. -half of rural householdsstill lack access to improved water source 7.9 Yet, important challenges remain. According to the WorldDevelopment Indicators (2004) data, El Salvador still ranks relatively low by regional standards with respect to the proportion of the population coveredby improvedwater sources (Figure 7.2). Indeed, analysis o f data from 31 Latin American and Caribbean countries found that in 2000 access to safe drinking water in El Salvador i s considerably lower than would be expected given its per capita income (Todd and 66 Hicks 2003).72 Moreover, about 19 percent o f the rural population - and over one-third o f the extreme poor in rural areas - still get their water from rivers, lakes and springs where there are often high levels o f pollution.73 These groups have a particularly elevated risk o f waterborne disease, such as diarrhea and gastroenteritis. Another nearly 9 percent o f the extreme rural poor (who lack independent water access) purchase water from water salesmen, sometimes at three times the regular piped water tariff. Household data suggest that the cost to the extreme poor of water purchases from these sources is staggering, constituting 18 percent of their monetary income, on average (Yepes 2004, based on EHPMdata). Figure 7.2: Access to ImprovedWater Sources, 2000 (percent of population) Source: WorldDevelopment Indicators(2004). Lack of sanitationfacilities is also most acute in rural areas 7.10 In addition, important challenges remain inthe provision o f sanitation services. According to the World Development Indicators (2004), while El Salvador fares relatively well in the provision of sanitation services, both nationally and in rural areas (Figure 7.3), over 20 percent o f the poorest households still lack access to adequate sanitation facilities (Figure 2.6). Moreover, there persist considerable differentials in quality o f access. The use o f latrines i s still very common throughout the country, with the availability o f sewer infrastructure low in a number of rural areas.74 At the same time, existing sewage collection systems often lack sufficient capacity to handle wastewater flows, especially in poor neighborhoods, resulting in overflow problems. These wastewater overflows contribute to the high levels o f environmental pollution (Yepes 2004). 72Using WorldDevelopment Indicator (2002) data from 2000, Todd and Hicks (2003) findthat El Salvador '' ranks fifth lowest inthe region among "under-performers'' inwater access. 73Significant water pollution is due to a weak regulatory environment governing the flow, quantity and uality o fwater along with poor (and frequently nonexistent) treatment o f wastewater (Yepes 2004). More than 70% of inhabitants lack sewer service inabout 10 municipalities, while more than 50 percent lack service inmany others (Yepes 2004). 67 Figure 7.3: Access to Improved SanitationFacilities,2000 (percentof population) I 100 90 80 70 c) 60 f8 50 e, 40 30 20 10 0 Source: WorldDevelopmentIndicators (2004). Recent public action A plethora of agencieshave responsibilities.for water and sanitation 7.11 The ruralwater and sanitation sectorin El Salvador characterized by a multiplicity of i s institutions and agencies providing services. The lead Government o f El Salvador institution in the rural water and sanitation sector is ANDA, which has formal responsibility for water and sanitation in both urban and rural areas. ANDA was founded 43 years ago to manage all water and sanitation systems in the country, and it undertakes most functions associated with the water and sanitation sector, including planning, construction, operation, maintenance and commercial processes. ANDA also plays the role o f the sector's regulatory and enforcement agency. 7.12 In its first stage o f operations, ANDA focused on urban areas, leaving provision o f water largely the responsibility o f the Ministry o f Public Health (MSPAS). In addition, the Departamento de Acueductos Rurales, later converted to Plan Nacional de Saneamiento Bcisico Rural (PLANSABAR), had responsibiliy5for rural water systems, constructing 308 aqueducts over a 15-year periodthat ended in 1996. Some rural systems were also initiated and managed by communities or local committees, for example, ANDAR (Asociacidn Nacional para la Defensa, Desarrollo y Distribucidn deAgua a Nivel Rural). 7.13 More recently, ANDA has become involved in the direct provision o f water services to rural areas, though largely through the extension o f urban networks to adjacent rural areas. Indeed, ANDA has not developed the capacity to work in rural areas that are not connected to urban systems. So, most of the connections in rural sectors served by ANDA are at the peripheries o f cities. Inthe absence o f ANDA working on purely rural water systems, a large and rather heterogeneous group o f institutions and agencies have played this role, including FISDL '*In 1996, when the programendedand administration o fthe PLANSABAR water systemspassedon to ANDA, most ofthe aqueductswere facing sustainability problems -both interms offinancial weakness and lack of maintenance. See Cuellar, N.and S. de Larios (2001), "Acceso a1aguapotable en EL Salvador: tendencias,perspectivasy desafios", Prisma No. 42. 68 (Fondo de Inversidn Social para el Desarrollo Local) 76, MSPAS, municipal governments, decentralized utilities, NGOs, the private sector andlocal water committees (Figure 7.3). ANDA FISDL MSPAS Municipalities D e ~ NGOs ~Sector~ ~ Private Committees ~ ~ d Number ofEntities i, 1I 1 1 1 1 1 256 13 1 42 I 30 I 102 GeneralPlanning X UndergroundWater Exploration X X X X Water Quality Control X X Tariffs Auuroval X X ControlandRegulation X Sector's financing X X X X X Operationand Maintenance X X X X Constructionof Water Infrastructure X X X X X Constructionof SanitationInfrastructure X X X X X -with sometimes overlappingroles and responsibilities 7.14 As can be seen from Table 7.3, different entities play different roles in the water and sanitation sector. In addition to ANDA's wide-ranging role, the Central Government i s involved inthe water and sanitation sector through FISDL, which plays roles in securing fundingfor the sector and in the construction o f water infrastructure, and through MSPAS, which i s currently responsible for assuring water quality. 7.15 Municipalities, decentralized utilities and NGOs also play an increasingly important role, often with the support o f ANDA and/or FISDL. For example, FISDL supports municipalities to promote local development through (i) raising and managing financial resources from the national budget, donors, NGOs and Salvadorian emigrants, and (ii) strengthening institutional capacity to implementdevelopment projects, includingwater projects. 7.16 Since the 2001 earthquakes, ANDA and FISDL have been working toward an assisted decentralized scheme, which includes a set o f initiatives to help municipalities improve investment in and management o f water and sanitation utilities. ANDA and FISDL provide financing and support the strengtheningo f institutional capacity. NGOs act as counterparts inthe investments. Municipal authorities and/or local committees also contribute, working to organize communities and orienting them on a range o f issues, including on rational water use with 76 FISDLwas founded in 1990 as a Social Investment Fundto compensate the poorest Salvadorans affected bythe structural adjustment process. In 1996, the Funditwas mergedwith Secretaria de Reconstruccidn Nacional (SRN) andPrograma de Municipalidades en Accidn (MEA), whose goals where to provide financing to local governments and strength their institutional capacities. In 1999, the Government o f El Salvador's development plan, La Nueva Alianza, empowered FISDL to promote and facilitate local development as part o f its goal to strengthen municipal infrastructure and access to basic services. 69 sustainable tariffs. At present, 13 decentralized water utilities with differentlegal structures have beenestablished and are operating. 7.17 While the private sector also participates in the water and sanitation sector, its role remains largely limited to construction o f water and sanitation infrastructure - although it has provided a small share o f capital insome of the decentralized utilities. 7.18 A recent study estimated the relative contributions of ANDA and the other water sector institutions in the provision o f improved water access over the 1991-2002 period (Yepes 2004). Estimations were undertaken usingdata on rural coverage (and its change over time) from ANDA and comparing it with data on rural coverage from the 1991 and 2002 EHPM surveys, Adjustments were made to correct for definitional and other differences inthe two data sources.77 Differences in rural coverage rates between ANDA's data (which captures only ANDA's investments) and the EHPM data (which captures coverage regardless of the source) were then taken as reasonable estimates o f the contribution that municipalities, decentralized utilities and local providers have made to improve water access over the period. The analysis suggests that ANDA has contributed to about three-quarters of the increase inruralcoverage between 1991and 2002, whereas other water sector institutions (e.g., local, municipal, decentralized and NGO providers) have collectively contributed the remaining 25 percent o fthe increase. And while access has improved,progress in rural areas has been uneven. 7.19 While the current institutional arrangements have obviously contributed to increased access to water and sanitation services since the early 199Os, the multiplicity (and heterogeneity) o f agencies are a function o f an outdated regulatory framework that provides for uneven "rules o f the game" for different institutions with different roles in the sector. This has resulted, among other things, inunevenspatial patterns o f water sources and access across the country. As can be seen from Map 7.1, ANDA has continued to serve primarily the areas surrounding the larger cities. Increased water access outside ANDA's primary service areas has, thus, largely beenthe result o f work by the other water entities, such as the municipalities, decentralized water providers and NGOs. As such, a significant proportion o f improved drinking water access in rural El Salvador does not rely on direct Central Government management via ANDA - although it is true that many o f the other water entities continue to rely on a variety o f types o f support from ANDA andor FISDL (e.g., direct financial support, input subsidies, support to capacity building, etc.). 77See Yepes (2004) for details. 70 Figure7.4: Central GovernmentSpendingonWater Supply and Sanitation (US$ millionand percent of GDP) 0.0 1 =CentralGovernmentSpending (US$,millions) --t 1 Percent of GDP Source: ANDA, FISDL,and MSPASdata. Sanitation 2.4 8.5 9.3 10.5 12.1 0.9 As a Percentof GDP Total Expenditure 0.4 0.6 1.o 1.o 0.9 0.6 Total Investment 0.1 0.3 0.4 0.4 0.2 0.1 7.22 Investment in the water and sanitation sector averaged 0.3 percent o f GDP over the 1990- 2003 period. Investments grew from 0.1 percent o f GDP in 1990 to about 0.4 percent o f GDP in 2000-01, but then fell again to 0.1 percent by 2003 (Table 7.4). Seventy percent o f investment was used for rehabilitating, improving and building water systems (aqueducts and public water 72 wells) and the remaining 29 percent for sanitation (sewerage, septic tanks, and latrines).78 Over the period, investments constituted the second-highest most important expenditure item, on average, after spending on operations, including physic-chemical treatment of water, and maintenance o f distribution systems. Investments in water supply and sanitation reached a maximum of 39 percent o f total spending in 2000, but declined dramatically thereafter, dropping to 5 percent o f total expenditures in2003. 7.23 To put these investment figures in perspective, during discussions for the National Sanitation Plan7' in 1993, PAHO estimated required investments o f between 0.67 and 0.83 percent of GDP in order to achieve 100 percent coverage of water and sanitation services, inboth rural and urban areas.8o This requirement was estimated based on universal access, making allowance as well for wastewater treatment. These figures were estimated under an economic growth scenario that was optimistic relative to actual growth rates inthe second half o f the 1990s. For this and other reasons, investments have been lower than anticipated, averaging less than half of the lower boundof the resource targets set to attain the established goals. -with heavy reliance on externalfinancing of sector investments 7.24 Investments in the sector have also been highly reliant on external funding sources. Between 1990 and 2002, external resources comprised about 63 percent o f total investment inthe sector (World Bank 2004). In this context, an average o f 48 percent o f the sector's investment over the periodwas financed by external loans, while another 15 percent was financed by external grants. The remaining investments (about 37 percent) were funded using domestic resources; about 21 percent, on average, were funded with ANDA's own resources, while the remainder (about 16 percent) was funded with Central Government andor FISDL resources. Substantial increases in rural investmenthas narrowed coveragegaps 7.25 In spite o f fluctuations in total spending, average spending on investments in rural ureas has increased substantially in recent years - albeit from a low base. Whereas investment inrural areas was $2.9 million between 1995 and 1998, it rose to $35.8 million between 1999 to 2002 - due to both increases in domestic funding(via ANDA and FISDL) and external support (through loans and donations from foreign governments and multilateral lending organizations). *' Although no formal analysis exists o f the distribution o f the benefits of public spending in water and sanitation in rural areas, the data on changes in water access between 1991 and 2002 indicate that coverage grew faster among the poor than the non-poor and inrural areas than in urban areas over the period (see Figure 2.6).82 As such, spending over the past decade has served to narrow the poorhon-poor andurbadrural coverage gaps. ''World l8 Bank 2004. Elaborated by National Committee o f Water and Sanitation Sector Institutions (CONIAPOS), composed b the Ministryo fHealth, Ministryo fPlanning, andANDA. "See "Analisis del Sector deAguay Saneamientoen El Salvador", Washington D.C., 1994. 81 Yepes (2004). A new World Bank Public ExpenditureReview for ElSalvador find that growththe availability o fpiped water at the national level between 1991 and 2002 has tended to benefited the first three quintiles most, withthe largest gains accruingto the middleandpoorest quintiles (World Bank 2004). Increases inthe availability o f sanitation facilities were found to benefitthe poorer quintiles more than the wealthier quintiles. The analyses does not distinguishbetweenrural andurban areas, however. 73 Ongoing challenges Increase access to water and sanitationfacilities, especially in rural areas- 7.26 Although access to water and sanitation services have increased, El Salvador faces continuing challenges in ensuringthat the poor have reasonable access to potable water, as part o f a national social policy to reduce poverty. El Salvador still ranks relatively low among Latin American countries in terms o f safe water access - indeed, lower than would be expected given its per capita income. Several remaining challenges will be important in framing a pro-poor policy agenda for rural water and sanitation: 0 Nearly half o f the poorest Salvadorans lack access to improvedwater sources; 0 In rural areas, over one-third of the extreme poor still get their water from rivers, lakes and springs where there are often high levels o f pollution, raising their exposure to waterborne diseases; 0 Over 20 percent o f the poorest households also still lack access to adequate sanitation facilities, with potentially important impacts on family and environmental health; and 0 Moreover, nearly 9 percent o f the extreme poor in rural areas purchase water from informal sources at as much as three times the cost o f the regular water tariff, with significant impacts on their abilities to purchase other basic consumption items. 7.27 While access clearly remains the central issue for water and sanitation, affordability and water quality (closely relatedto access) are criticalfor the poor as well. -and rationalize the complex institutional and regulatory environment 7.28 The challenge o f ensuring universal access to safe water and sanitation i s made more difficult by a complex institutional and regulatory environment in the sector. ANDA currently provides services in 182 o f 262 municipalities in the country, and defines the regulation in municipalities and micro-regions under its control. This has resultedin duplications o f efforts in some places, as well as uneven "rules o f the game" between institutions working under ANDA jurisdiction and those outside. While the current multi-actor approach to rural water provision has potential benefits, the current regulatory environment can lead to inconsistent incentives for ANDA as regulator vs. service provider. It also contributes to inefficiencies in the system, considerably variability in the quality o f water systems, and at times, exorbitant water costs for the poor. 7.29 Indeed, in the current environment, the expansion o f urban water networks to provide surrounding rural areas has sometimes exacerbated the difficulties facing the poor in accessing affordable water. For example, as peri-urban areas have become more populated, some families located near urban water networks have purchased up to 300 meters o f pipe from ANDA to access water from existing urban networks. These ad-hoc connections often fuel the supply side of gray water markets in which the rural poor pay considerably higher prices for water than the regular water tariff. Directions for uublic uolicv ClariJj,roles, responsibilities,and rights in the sector- 7.30 Inthis context, a key policy challenge inthe sector will be to reinitiate and complete water sector reforms to update and rationalize the institutional and regulatory environment. The reform 74 would be intended to level the playing field and to clarify the functions and responsibilities in the sector. Among the key challenges for the reformprocess include: 0 Clarification o f the roles and responsibilities o f the various institutions operating in the water and sanitation sector, includingthe roles o f centralizedversus decentralized service providers; 0 Creation o f an independentregulatory body, focused exclusively on policy and regulation and its enforcement; and 0 Clarificatioddefinition o f water use rights. 7.31 Much o f the sector reform agenda is not new. Indeed, reform efforts were initiated in the late 1990s and then set aside, in part because the earthquakes temporarily shifted the sector's focus to system reconstruction. Many technical inputs to a reform are already reasonably developed. The incoming administration thus has the opportunity to build on these inputs to complete the reform and put the water sector on a sound footing to improving access to potable water access among the poor. Yet progress on reforms i s likely to require efforts to build consensus around the need and scope for reforms, as well as the identification o f appropriate and cost-effective service delivery models for covering poor, often isolated, rural areas. -building on lessonsfrom recent experience 7.32 Some recent lessons from experience may also contribute to sector reforms that are effective in assuring access to safe water to the poor in rural areas. For example, there may be lessons for an appropriate service model, based on the experience o f the thirteen decentralized utilities that were established in partnership between ANDA, FISDL and local actors after the earthquakes. A recent evaluation o f several o f the decentralized utilities suggests that these utilities have functioned well, with water systems that are technically appropriate to the locality, good water quality and a high degree o f financial sustainability (Yepes 2004). These utilities were established under very specific circumstances in the aftermath of the earthquakes. Nonetheless, the approach appears to have promise, and could be used to inform design o f an appropriate regulatory framework for the establishment and scaling up o f small scale and local water enterprises, as part o f the sectoral reform. 7.33 The issue o f drinking water contamination is an increasingly important problem - especially to poor Salvadorans who lack access to improved water, and for whom consumption o f contaminated water has direct and adverse health consequences. In this context, it will be important that there be efforts to align policies regarding drinking water access, wastewater management and decontaminationo f water sources as part o f a sector reformprocess. 7.34 Finally, from a poverty reduction perspective, there may be added value in establishing an institutional mechanism for better identifying and targeting water and sanitation investments to those who currently lack access. Additional resources are required- 7.35 Costs. Inrecent years, the water and sanitation sector has also been under-resourced if it is to achieve universal coverage in water and sanitation within a reasonable timeframe. Total spending has averaged about 0.8 percent o f GDP, falling to 0.6 percent in2003 after hovering at about 0.9-1.0 percent between 2000 and 2002. Yearly investments have averaged about 0.3 percent o f GDP over the last decade, although they have only reached between0.1-0.2 inthe last 75 two years. Much o f this spendinghas been fundedby external resources that has varied year-to- year based on donors' project cycles.83 7.36 As noted above, in the context o f the National Plan o f Sanitation in 1993, it was estimated that to achieve universal coverage inwater and sanitation services - inboth rural and urban areas -annualinvestmentrequirementsrangedfrom0.67to0.83percentofGDP-morethantwicethe percentage actually spent over the period, on average. A decade later, this estimate should be interpreted with some care, as considerable progress in providing water and sanitation services has been made since the early 1990s. In addition, completing and implementing water and sanitation sector reforms may bring efficiency gains that lower the costs o f water and sanitation service provision. Moreover, and related to reforms, the identification o f cost-effective delivery models, such as the decentralized utility model, could result in the ability to achieve more with less. -equivalent to about 0.3% of GDP. 7.37 Given the uncertainty about the nature and timing o f water and sanitation sector reforms and about the most cost-effective and appropriate service delivery models, estimating with precision the necessary increases in investments is difficult. Nonetheless, recent estimates suggest a possible range of costs associated with attaining universal coverage in water and sanitation. Usinginternational estimates o f efficiency prices and differentiating between the costs of investments in rural and urban areas, Yepes (2004) estimates an average annual investment cost o f US$21 million between now and 2015, or about 0.14 percent o f GDP per year. Because these calculations use estimates o f efficiency costs, rather than actual costs, they may be interpreted as lower-bound estimates. Following an analysis presented in the recent Human Development Report (2004), and using unit cost estimates provided by ANDA (although not differentiating between the unit costs o f rural and urban areas), it is estimated that the additional investment costs of achieving universal water and sanitation coverage would range from US$ 45 to US$ 50 million between now and 2015, or as much as an additional about 0.3 percent o f GDP per year.84 Rural Roads Recent develouments Considerable improvementsin road access, especiallyfor the extremepoor- 7.38 As with water, there have been considerable improvements in access to roads in rural El Salvador, especially among the very poor. Improvedroad access translates into reduced isolation, lower costs for goods and services, and increased access to markets and services and participation ineconomic activities. 7.39 Several access indicators, measured in terms o f distances and travel times o f rural households to pavedroads and to the nearest markets, were collected as part o f the BASIS survey data for 1995, 1997, 1999 and 2001. Estimates o f average distances and travel times were made separately for the extreme poor, the moderate poor and the non-poor over the period (Yepes 2004). As can be seen in Figure 7.5, the average distance to a paved road is highest for the extreme poor and lowest for the non-poor. In addition, distances declined for all categories o f 83This uncertainty about size and sustainability o fresources flows has also complicatedthe development within the water and sanitation sector. 84Inthe absence ofsector reforms and the identification o fa cost-effective service delivery model for rural areas, the costs are likely to be closer to the upper- than the lower-bound estimate. 76 households over the 1995-2001 period.85Distances to paved roads among the rural poor declined slightly more than for the rural non-poor, however, meaning that the distance gap between poor and non-poor closed slightly during the period. For example, the average distance to a paved road among extremely and moderately poor households declined by roughly 1.7 kilometers between 1995-2001, whereas it declined by just over 1 kilometer for the non-poor over the period. Figure7.5: Distanceto PavedRoadsinRuralAreas (in kms.) 1995 1997 1999 2001 Source: Yepes (2004), estimated usingBASIS Data, 1995-2001. 7.40 Average distances to the nearest market also fell during the period (Figure 7.6). Average distances to the nearest market for both the very poor and the non-poor inrural areas declined by about 5 kilometers over the period, whereas the data suggest that the moderate poor may have experienced a slight increase. 85 Itshouldbe notedthat declines indistance to pavedroads do not reflect geographic movements of householdsinthe BASISpanel, but rather reflect growth o f the pavedroadnetwork. 77 Figure 7.6: Distance to the Nearest Market (in kms.) 1995 1997 1999 2001 Source: Yepes (2004), estimated usingBASIS Data, 1995-2001. -with notable effects on access and reducedtravel times-to markets - 7.41 Travel times capture not only physical distances but also the quality o f roads, the quality and quantity o f transport and other factors that affect the speed o f travel (e.g., traffic congestion). Data on changes in travel times largely mirror those on changes in physical distances. As with distances to markets, the extreme poor experienced particularly large declines in travel times to the nearest market between 1999 and 2001 (Yepes 2004). These data suggest that the more and better roads and transport were particularly effective inreducing transactions costs and increasing access to markets among the extreme poor in recent years. This is consistent as well with the findings presented earlier in the chapter regarding relatively large declines in poverty between 1999 and 2001 among the extreme poor who lived in areas that received road-improving investments. 7.42 Despite measurable improvements, important challenges remain, as many poor rural households remain isolated - in absolute, as well as inrelative terms. In2001, average distances (and travel times) to paved roads remained 80-90 percent higher among the extreme poor than the non-poor. Similarly, distances (and travel times) to markets remained 33-36 percent higher among extremely poor households than among the non-poor. Recent public action. 7.43 There are currently three main institutions responsible for rural road development and maintenance: the Ministry o f Public Works (MOP), the Fund for Road Conservation (FOVIAL) and FISDL. MOP is responsible for the rehabilitation, modernization and development o f primary, secondary, and tertiary road networkss6. FOVIAL is responsible for maintenance o f the primary road network, financed partially through a special tax on gasoline. FISDL works to strengthen municipal management capacity for maintenance o f local roads, which link 86Under Salvadoranlaw, primary, secondaryandtertiary roads are designatedas part of the "National Priority Network" (or Red Nacional Prioritaria). 78 municipalities, cantons and villages.*' In coordination with the Ministry of Agriculture, FISDL also oversees rehabilitationof rural roads through institutional agreementswith municipalities. A signi9cant expansion ofpaved roads, with emphasison rural areas- 7.44 Figure 7.7 presents a comparison of the growth of paved roads in selected Latin American countries over the 1990s. The data suggest that, from a regional perspective, expansion o f El Salvador's road network was relatively slow during much of the 1990s. The pace o f road investments increased noticeably, at least rural areas, at the endof the decade. Between 1999 and 2003, the Government embarked on a significant road investment program inwhich 40 percent o f the nation's principal road network was rehabilitated and repaired. This included paving or improving nearly 780 kilometers of roads as part o f Central Government's Caminos Rurales Sostenibles (Sustainable Rural Roads) program, rural road repair of earthquake damaged roads, and partnerships between FISDL and local governments to increase rural roadaccess.88 Figure 7.7: Average Annual Increase inKilometersof PavedRoadsin the 1990s inLatinAmerica I 6.O 5.0 -2 4.0 3.0 t 2.0 1.o nn I Source: WorldDevelopment Indicators (2004). 7.45 Between 1997 and 2003, GOES invested US $1,256 million on projects in construction, rehabilitation and maintenance of the country's road network (Table 7.5). Of this amount, 82 percent was executed through MOP, 11percent through FOVIAL, and 7 percent through FISDL. Public resources spent on roads fluctuated somewhat over the period, peaking at 1.7 percent o f GDP in2002, but declining to 1.3 percent in2003.89 7.46 The available data do not permit separate analysis of spending on rural roads prior to 1999. However, between 1999 and 2003, public spending on rural roads totaled U S $184 million. This Under Salvadoranlaw, local roads(caminos vecinales o locales) are designatedas part ofthe "Local Network"(or RedLocal). 88FISDLundertookjust over 1,900 roadimprovementprojects duringthe 1999-2003 period. Unlikeinthe cases ofeducationandhealth, cross-countrydata are not availableon spendingon road investments. As aresult, no internationalcomparisons -or benchmarking-are possible. 79 represented about 21 percent of total public spending on roads over that period. Fifty eighth percent o f public spending rural roads went to projects implemented under Caminos Rurales Sostenibles and to rural road maintenance executed by MOP; 37 percent o f spending on rural roads went for road improvement projects executed by FISDL;" and 5 percent o f public spending went for routine maintenance o f unpavedroads, carried out by FOVIAL. Spending on rural roads stayed fairly constant as a share o f national income over the period, remaining at 0.2-0.3 percent o f GDP. Table 7.5: PublicExpenditureon Road Infrastructure, 1997-2003 Notes: a/ Inter-urbanandurban b/ Periodic androutine maintenance o fpavedandunpavedroads cl Rural maintenance and Caminos Rurales Sostenibles d/ Routinemaintenanceof unpavedroads e/ Localroadimprovement Source: World Bank 2004, usingMOP, FOVIAL andFISDL data. -and an increasedfocus on road maintenance 7.47 In addition to the priority placed on developing rural roads through Camino Rurales Sostenibles, an important recent development has been the establishment o f an institution, FOVIAL, dedicated to road maintenance. Since 2002, FOVIAL has overseen routine and periodic maintenance on El Salvador's primary road network. Many countries under-invest in operations and maintenance, and indoing so raise the cost o f road infrastructure investmentsover time. In that sense, FOVIAL's maintenance activities represent good practice in governmental efforts to strengthen road sector investments - conserving resources by extending the life of recently paved or rehabilitatedroads. 90FISDL activities often mobilize counterpart funds from participating municipalities and local government, thus leveragingCentralGovernment investments. Suchcounterpart funds are not captured in the CentralGovernmentexpenditurefigures presentedhere. Inthat sense, these figures underestimate,to some extent, public spendingon rural roads over the period. 80 Still, needjor an institutionalizedprogramfar rural raads maintenance. 7.48 Although FOVIAL spent nearly 10 percent of its budget on rural roads maintenance in 2003 (Table 7.5), the Government of El Salvador does not yet have an institutionalized program for rural roads maintenance. There would be high returns, however, to developing a scheme to guarantee the longevity and sustainability of rural roads investments, particularly in secondary and tertiary networks (Le., rural maintenance efforts that would mirror FOVIAL's efforts on the primary network). Such a program could be implemented, at least in part, through partnerships between the Central Government and local governments - along the lines o f recent rural road maintenance activities supported and monitoredby FISDL and the German agency, GTZ. 7.49 Ongoing challenges and directions for policy. Despite measurable improvements, important challenges remain, as many poor rural households remain isolated- in absolute as well as in relative terms. As noted above, in 2001, average distances for members o f extremely poor households remained over 5 kilometers, about 80-90 percent further than among non-poor households. Similarly, distances to the nearest markets exceeded 45 kilometers for the extreme poor, more than one-third higher, on average, than for the non-poor. 7.50 Recent progress to rehabilitate the nation's principal road network, to repair earthquake damage roads and to promote greater rural road access through CaminosRurales Sostenibles and other programs has been impressive and provides useful lessons for the way forward, particularly regarding the importance o f road maintenance. Indeed, by ensuring longevity o f recent rural road investments, returns to rural roads maintenance are extremely high and should be considered a priority along with continued investments in road expansion. In this context, efforts to ensure that the poor benefit from ruralroad investments should focus on: 0 Continuing expansion and improvement o f the rural roadnetwork; 0 Developing a program for the maintenance o f secondary and tertiary road networks to ensure the sustainability o f recent investments inrural roads; and 0 Continuing development o f local and municipal capacity to build and maintain rural (and local) road infrastructure. Only a modest increase in spending is necessary 7.51 Costs. The costs associated with a program to continue expansion and improvement o f the rural road network, as well as to develop a maintenance capacity for rural roads depends on the specific goals and the timeframe set for achieving them. One possible set o f objectives - that would be consistent with the country's goals to strengthen poverty reduction through infrastructure development - would be: (i) to pave or rehabilitate an additional 1,000 kilometers o f rural roads; and, (ii)to ensure annual maintenance o f 3,000 kilometers o f rural (secondary, tertiary, and local) roads. Analysis o f El Salvador's road investments suggests that the unit costs o f the country's road development are in line with international norms (Yepes 2004). Using recent average unit costs o f road paving, rehabilitation, and maintenance as benchmarks, it i s estimated that a 5-year program to (i) andor rehabilitate an additional 1,000 kilometers o f pave rural roads and (ii) initiate a rural road maintenance program would cost the Government o f El Salvador roughly an additional US $25 million per year over average 2003 spending levels. This implies a spending level o f approximately 1.4-1.5 percent o f GDP per year, up from 1.3 percent o f GDP in 2003 (and on average between 1999 and 2003). 81 VIII. SOCIAL PROTECTION 8.1 Social protection is traditionally defined as a range o f measures adopted by governments to provide basic income security to their people, to protect them from unanticipated shocks, and protect and develop the human capital of society's poorest most vulnerable members o f society, those outside the reach of traditional government and private services. Social protection is typically delivered via a range o f social insurance andor social assistance programs (Box 8.1). Box 8.1: SocialProtection-ManagingSocialRisk, PromotingLong-term Growth and Development People in developing countries face a range o f risks. Some risks, such as economic recessions, harvest losses, natural disasters and wars, affect whole societies or large groups. Others, such as the illness o f family members, loss o f household breadwinners' jobs, and crime, may only affect individual households. Social protection is defined as a range o f public interventions that support society's most poor and vulnerable members and help individuals, families, and communities manage risks better. Social protection includes a range o f informal, market-based, and public mechanisms, including regulation, government financing, and the direct provision o f services. Intended to augment, not replace, family, community, and market-based risk management mechanisms, such interventions complement national economic policies and support strategies for poverty reduction and human development. Public social protection measures often are categorized into two main groups: social insurance and social assistance. Social insurance includes an array o f insurance-type mechanisms, such as pensions and unemployment and health insurance that are intended to cushion the impact o f shocks affecting income, employment, or health, and thereby prevent families from falling into poverty. Social assistance includes a variety o f safety net programs, such as workfare, assistance to the disabled and indigent, and cash transfers, all o f which help individuals and families deal with temporary or chronic poverty and achieve higher standards o f living. Labor market policies and institutions often play a central role in social protection, both by influencing the nature and extent o f the risks workers face and by providing a framework for social protection programs such as pensions, unemployment insurance andworkfare. Well-designed social-protection programs can compensate for missing insurance markets or other private risk-mitigation instruments, andindoing so, they create opportunities among the poor for more productive investments and higher incomes. In addition, certain types o f safety net tools-such as "conditional transfers," where payments are made contingent upon family investments in children's health or schooling-both provide short-term income support and strengthen longer-term investments in children's education, health, well-being andproductivity. Source: "Volatility, Risk, and Innovation: Social Protection in Latin America and the Caribbean, Spectrum, World Bank, Fall2003. Socialprotection can soften the impact of shocks among society's most vulnerable- 8.2 Social protection can - and should - play an integral role in a National Social Policy to reduce poverty in El Salvador. In particular, safety nets programs are important as part of El Salvador's poverty reductions strategy for several reasons. Specifically: Although the moderate poor have been able, to a large extent, to share in recent growth, the poorest, most vulnerable Salvadorans have been much less able to take advantage of recent progress. Between 1991 and 2002, per capita income growth among the poorest 20 percent of Salvadoran households was just over one-third of average per capita income growth for the p e r i ~ d . ~Indeed, a new analysis of the B A S I S data finds some ' evidence of "poverty traps" among the poorest households in rural El Salvador (Gonzalez-Vega and Rodriguez-Meza 2004). Persistently low levels of education and ''See table 3.2. 82 continuing poor access to basic services mean that the most poor and vulnerable households lack the capabilities to take advantage o f new and emerging economic opportunities and pullthemselves out of poverty. 0 El Salvador is prone to a variety of shocks, including earthquakes, droughts, floods, as well as external economic events. While all Salvadorans are potentially vulnerable to shocks, as events like the earthquakes o f 2001 attest, the extreme poor tend to be most vulnerable. Evidence from the BASIS panel data indicates that rural incomes remain quite volatile in El Salvador, and that poor households tend to rebound from transient income shocks muchmore slowly than non-poor households (Beneke de SanfeliG and Shi 2004; Gonzalez-Vega and Rodriguez-Meza 2004).92 In addition, evidence from the 2001 earthquake and the coffee crisis suggests that even when households' incomes rebound quickly, the effects of shocks on non-income dimensions o f well-being, such as education and nutrition, may be more persistent, taking more time to recover. While most households have their own "private" safety nets - most notably migration and remittances - they tend not be sufficient, particular among the poor. International evidence suggests that private risk management strategies only partially insure families against riskhhocks. Moreover, in the case o f remittances, only a fraction o f the population-and thus those that needsupport -receive them.93 -andprovide them with thesecurity to investproductivelyin their ownfutures. 8.3 More generally, a growing body o f global evidence i s making increasingly clear that too much vulnerability and "uninsured" risk in a society can hinder a country's ability to grow and reduce poverty, by reducing both the ability o f the poor to invest inproductive enterprises and the quality o f their investments (Ravallion 2003). Poor, uninsured, and risk-averse farmers and entrepreneurs, for example, often invest in low-risk, low-return enterprises in an attempt to prevent or mitigate the impact o f unanticipated shocks, leading to the type o f poverty traps observed in El Salvador. Moreover, too much uninsuredrisk can lead the poor to under-invest in their children's education and health - saving on schooling costs or counting on child labor to augment and diversify their earnings "portfolio". But in doing so, they sacrifice their children's future productivity and increase the changes o f an intergenerationalcycle o f poverty. In contrast, by virtue of its insurance and safety net functions, sound social protection programs can actually helpto strengthencountries' ability to reduce poverty. 8.4 This chapter summarizes the key risks and risk groups inEl Salvador, based on the findings presentedearlier inthe report as well as on the findings o f a recent World Bank Social Safety Net Assessment (World Bank 2002). It then discusses the current state o f El Salvador's social safety net, highlightskey challenges for developing a safety net as an integral part o f a National Social Policy to reduce poverty andvulnerability, and outlines policy directions for the near-te~m.~~ Withfalling poverty rates, more deliberate,targeted measures are needed. 8.5 Several key messages emerge from the chapter. As is highlighted elsewhere in the report, El Salvador's recent accomplishments have created new challenges with respect to reducing 92The findings ofhighvolatility of income inruralareas, coupledwith the fact that the incomes of the poor recover from shocks more slowly thanincomes ofthe non-poor, suggest that formal and informal insurance marketsdo not functionwell inrural areas of El Salvador, andeven less well amongthe poor. 93For amore detaileddiscussionofthe role and extent of remittances, see Chapter 3. 94Since initial preparationo f andconsultationson this report, anew Governmenthas developedand at strengtheningaccess to basic educationandhealth services - and thus the humancapital- of children launcheda multi-facetedsocial strategy that includesthe Red Solidaria,a unifiedsafetynet programaimed living inextremely poor Salvadoranhouseholds. 83 poverty. When poverty was more widespread, growth together with broad sectoral strategies were appropriate and effective in raising the living standards o f the poor. Yet, as the proportion o f the population that is poor has declined, it has become clearer that there are sub-groups of the populations who have not beenable to take advantage o f growth or to benefit from traditional line ministryprograms. For these, ElSalvador's most poor and vulnerable families, specific and well- targeted measures are needed to ensure that they are more prepared to take advantage of future progress and opportunities. Moreover, the evidence suggests that focusing strategically on priority risks and risk groups; it is possible to have a significant positive impact with a safety net at a relatively low cost -perhaps less than 1%o f the GOES budgetper year inthe short-term. Kev risks Many risks are age-specific- 8.6 As discussed in detail in previous chapters, poor Salvadorans face elevated risks - with respect to health, nutrition, mortality, low education levels, lack o f access to productive assets and earnings opportunities - affecting both their current and future well-being. These risks often manifest themselves in different - and programmatically relevant - ways over the lifecycle. Specifically: Youngchildren, age 0-6. Young children inpoor families face the risk of poor health and nutrition and lack of access to health early childhood development services, leading to atrophied physical and cognitive development. Atrophied child development, caused either by malnutrition or the lack o f early stimulation, has severe negative consequences on school performance and, eventually, on productivity and earnings in the job market, thus contributing to the intergenerational transmissionofpoverty. Children andAdolescents, age 7-15. A key risk among poor children inthis age group is the lack of basic education, due either to non-attendance, late entry, repetition, desertion, as well as child labor, again with adverse effects on children's capabilities and economic prospects. The SSNA also identifiedfamily disintegration, domestic violence, and child abuse and abandonment as an important risk among poor children inthis age group.95 Young Adults, age 16-23. Poor youth in this age group face high risk o f low accumulation o f human capital due to low levels o f schooling, with negative consequences for employment, labor productivity and earnings. Economic inactivity among this age group is often associated a higher likelihood o f violence, gang activity, and substance abuse. Teen pregnancy is also a growing risk among poor young Salvadoran women, with adverse impacts interms o f schooling, economic prospects, and family life.96 Prime working-age adults, age 24-64. The main risk faced by poor people in this age group i s low productivity and income due to low levels o f human capital and poor access to productive assets, commonly with adverse second-round effects in terms o f families' investments intheir children's human capital. 95 In 1999, there were 1,510cases ofchildabusereportedinthe country (World Bank2002). 96 Every year, between 30,000 and 40,000 adolescents become pregnant, which represents a third of all pregnancies(World Bank2002). 84 0 income insecurity due to the lack o f a pension or other forms of old-age security. 9 Older adults, age 65 and over. The main risk faced by poor people in this age roup is 8 -though a lack of access to key services affects people of all ages 8.7 There are clearly important risks facing the poor face are not linkedto a specific part of the lifecycle. As discussed earlier, the poor as a group have less access to basic services in general and report a higher incidence o f illness than the non-poor, regardless o f their age. The poor also tend to face greater personal and legal insecurity than the non-poor (World Bank 2002). And when economic or natural shocks occur, these shocks affect the well-being o f entire families, regardless o f age. Priority should be given to building human capital in theyoung, investments with high returns. 8.8 At the same time, framing risks interms of the lifecycle is helphlinhighlightingpotential criteria for prioritizing safety net related programs and investments - particularly when fiscal constraints are likely to selectivity among potentialprograms. Two attractive criteria include: (1) investments, the absence o f which will cause irreversible damage; and (2) investments which, if made, can reduce the intergenerational transmission o f poverty. Both o f these criteria point inthe direction o fprograms and investments to support interventions to buildhuman capital early inthe lifecycle -i.e., among infants, children, and adolescents inEl Salvador's poorest, most vulnerable households. This prioritization is also consistent with a large body o f global evidence showing higheconomic and social returns to investments inyoung and school-age children. El Salvador's current social safetv net Drom-ams Overjifty safety net typeprograms- 8.9 Unlikethe sectoral issues discussed inchapters 4-7 of this report, the risks outlined above are cross-cutting and do not fit neatly into one public institution or another. Indeed, most countries do not have a single institutional entity responsible for safety net type programs, and El Salvador is no exception. Understanding the scope and reach o f a country's social safety thus requires a multi-sectoral approach and considerable "leg-work." In this context, the World Bank SSNA team, in collaboration with the main public and private social service institutions, identified 52 social safety net-type programs operating in El Salvador in 2000. These are programs implemented by a range o f range o f governmental, including the Ministries o f Education, Health, and Labor, the National Secretariat for the Family, the Salvadoran Institute for the Protection o f Minors, and, FISDL, as well as a variety of non-governmental actors (Le., private sector, NGOs). -with responsibilities that are not always clearly demarcated- 8.10 It is important to note that outside o f safety net programs implemented in response to specific shocks - such as an earthquake or hurricane - there i s not always a clear line o f demarcation between the programs that make up a country's safety net and those implementedby public institutions in the course o f their normal work or line activities. One practical way o f assessing whether a program is a line program or i s part o f the social safety net is to consider which obstacles to service access it seeks to eliminate, those from the supply side (e.g., lack o f school infrastructure) or those from the demand side (e.g., cost o f services, knowledge about service benefits and availability, etc.). Lineprograms typically act on supply (e.g., increasing the number of schools or teachers), whereas safety net programs often focus on removing obstacles ''Amongtheworking poor, only 6.1% is coveredby the pension system. About 351,000 elderly persons are not covered by a formal pension program, nor do they receive any alternative benefit (World Bank 2002). 85 on the demand side (e.g., by providing support to families who do not send their children to school due to the high direct or indirect costs). Where relevant, this rule-of-thumb was used. 8.11 The SSNA found that most o f the risks outlined above are addressed by one or more programs. (A detailed list of programs is presented in Annex 1.) The greatest number o f programs address the risks faced by children under 6 and their mothers (6 programs), efforts to increase learning and retention o f children in school (7 programs), the situation o f child workers (5 programs), family disintegration (6 programs), andyoung offenders (3 programs). There i s also support for the training and development of adolescents and young adults (6 programs) and the unemployed (2 programs). Problems o f discrimination against women (3 programs) and disabled persons (2 programs) are also addressed by existing programs. For the elderly, besides social insurance, only two small support programs were identified. For the provision o f basic services there are the FISDL programs, the FONAVIPO program o f targeted subsidies for housing purchases, the Program for Attention to Earthquake Victims, and in particular, the program Unidosfor El Salvador ("United for El Salvador", the earthquake reconstruction program). -and most are small-scale an&or pilot programs. 8.12 In sum, there were found to be a considerable number o f safety net type programs being implemented in El Salvador, by a range o f institutions and organization. With a few exceptions, however, most o f the programs are small scale interventions andor pilot projects, as opposed to wide-reaching or national level programs. Moreover, the programs tend to operate in a largely uncoordinated manner, with relatively little sharing o f operational lessons across projects or agencies. 8.13 Public spending on the safety net. The Government's line ministryprograms and the social safety net programs ought to be thought o f as complements - and spending the same. Line programs aim to provide basic services with adequate quality to the population at large. The social safety net programs should target those population groups, generally the poorest, who do not otherwise have access to these programs, or who face other obstacles to access (e.g., cost, distance, lack o f know knowledge about service availability and benefits) to basic line ministry services. Inprinciple, ifthe poor no hadproblems in gaining access to line ministryprograms, or to private pr public insurance mechanisms, then safety net programs would not be needed. In practice, the very poor still tend to access to traditional sectoral programs as well as formal insurance mechanisms (whether public or private). Public spending on safety netprograms is about 0.5% of GDP- 8.14 The SSNA estimated total spending (public plus private) on El Salvador's social safety net, based on estimates o f the amounts spent on each o f the 52 programs identified (World Bank 2002). Special efforts were made to avoid double counting o f expenditures that were better defined as part o f the Government's basic line ministry programs. The estimates suggest that total spending on El Salvador's safety net from public and private sources was about U S $135.4 million in 2000 (Table 8.1). Forty-eight percent o f that, or US $65 million, came from the public NGOs, and household^.^^ sector, while the remainder came from non-governmental sources, including the private sector, Thus, in 2000 the Government of El Salvador spent about US $65 million on safety net related programs - the equivalent o f 0.5 percent o f GDP and 3 percent o f Central Government spending. ~ ~~ 98Private sources o f spending on the safety net include private sector/workers' contribution to Social Security (Le., pensions and health), contributions to INSAFORP (Znstituto SalvadoreAo de Formacidn Professional), as well as estimates o f NGO financing o f some social protection programs (e.g., implemented through FUSAL and other non-governmental agencies). Private spending, as discussed here, excludes remittances - although remittances are often considered a key element o f households' private safety net. 86 Spending YO YO Category (US$OOO) Total GDP Total Spending 135,382 100.0 1.02 Public 65,000 48.0 0.49 Privateli 70,382 52.0 0.53 By Age Group 0-6, 10,100 7.5 0.08 7-15 37,558 27.7 0.28 16-23 10,138 7.5 0.08 24-64 29,309 21.6 0.22 65+ 12,528 9.3 0.09 ' Populationat large " 35,749 26.4 0.27 -which is low by regional standards- 8.15 Comparison o f these estimates with those from other recent studies suggests that public spending on safety net programs inEl Salvador i s low by regional standards. Public spending on safety nets in Central America as a whole is about 1.5 percent o f GDP, about 1.4 percent o f GDP for the Latin America region, and about 2.9 percent o f GDP in OECD countries, on average (Marques 2003; Spectrum 2003). 8.16 Analysis o f the BASIS data also indicates that transfers make up only a very small proportion o f the incomes o f rural households - both on average and among the poor. Between 1995 and 1999, total transfers from public and private institutions averaged between 0.8 and 1.1 percent o f household incomes (Beneke de Sanfeli6 and Shi 2004). This figure doubled to about 2.2 percent in2001, due to receipt o f assistance following the earthquakes. Perhaps more striking i s the fact that between 1995 and 1999 total receipt of transfers made up only 1.3 2.2 percent o f - the incomes o f the poorest rural households. This figure also doubled - to 4.1 percent - in2001. To put this inperspective, remittances made up about 16 percent o f rural household incomes, on average, in 2001, and nearly 10 percent o f incomes among the poorest rural households (Beneke de Sanfeli6 and Shi 2004). -especially amongprioriiy groups such asyoung children. 8.17 The SSNA's expenditure analysis also highlightedpatterns o f spendingon risk groups over the lifecycle (Table 8.1). Only 7.5 percent o f total spending on safety nets (0.08 percent o f GDP was spent on programs for children age 0-6, perhaps the area with the highest economic and social returns on investments. At the same time, another priority age group, 7 to 15 year olds, received more than a quarter o f the benefits o f spending - this due mainly to the Healthy School 87 program, which is nationwide in scope." As with 0 to 6 year olds, 16 to 23 year olds received 7.5 percent o f the resources spent on safety net type programs. Twenty-four to 64 year olds received 22 percent of spending, while 9.3 percent went to people 65 and over. Finally, about 26.4% o f the total spending on safety nets resourceswent to address risks affecting the population as a whole, much o f this through infrastructure and local development activities implemented through FISDL.loo Ongoing challenges. 8.18 Assessment o f the existing programs suggests that they are generally pertinent, focusing key socio-economic risks faced by the poorest, most vulnerable Salvadorans. At the same time, the impact of the safety net remains quite modest relative to its potential for several reasons, including Low coverage associated with low levels o f public spending Under-allocation o f resources to highestpriority risk groups 0 Weak program targeting Lack o f program coherence and coordination, and Low cost-effectiveness o fprograms insome cases Coverage of the safety net remains limited- 8.19 Coverage. Although there is a significant number o f programs, coverage o f the safety net remains limited relative to the size o f the groups at risk. For instance, according to FESAL (2003) 28 percent o f children under 5 in low-income households experience chronic malnutrition; but only a fraction o f those are covered by existing programs (Annex 1). Coverage will thus need to be increased substantially if El Salvador's safety net is to have significant and lasting impact. With some exceptions (e.g., the Healthy Schools program, which is nationwide), most programs are pilots or modest interventions that - without minimizing their merits - will have difficulty making a significant and lasting impact. Low program coverage o f safety net is closely related to low levels of public spending on the safety net programs. As a percentage o f GDP, El Salvador currently spends only about one-third on its safety net than do other countries in Central America (and Latin America as a whole)."' IfElSalvador is goingto implementaneffective social safety net, it will requireincreased commitment o f public resources. -notably with respect to the developmental risksfaced by 0-6year olds 8.20 Allocation of resources to high priority risk groups. Based on the criteria o f irreversibility and reducing the inter-generational transmission o f poverty, interventions to address the 99Startedin 1997, this Healthy Schools Program, implementedby theNational Secretariat for the Family, seeks to reducestudent desertion rates, compensate for inadequate school infrastructure, improve nutrition, andreduce the highrates of illness among the children. The programoperates in 100%ofruralpublic schools, reaching650,000 childrenfrom 4 to 17years old in2000. It shouldalso benotedthat this programalso benefitedthe 25 percent o fpoor children, aged4 to 6, who attendedpreschoolin2000 (World Bank 2002). looAdditional detailon spending levelsby programand risk group canbe found inAnnex 1. lo'Public spendingon El Salvador's safety net was roughly 0.5 percent of GDP in2000, comparedto 1.5 percent of GDP, on average, for the CentralAmerican countriesand 1.4percentof GDP, on average for the LatinAmerica regionas awhole (see Table 8.1, above, for the data onElSalvador; comparativefigures for the regioncanbe found inMarques 2003 andSpectrum 2003). 88 developmental risks o f 0-6 year olds are particularly under-resourced. Although this age group makes up over 20 percent o f the poor, early child interventions only received 7.5 percent o f total spending on the safety net in 2000 (equivalent to 0.08 percent o f GDP). Investmentsin early child development (e.g., early child health, nutrition, and education), along with investments in school aged children and adolescents, require an increase in priority v i s - h i s resource allocation to reduce irreversible damage to poor children's human capital. Moreover, early childhood investmentshave relatively have higheconomic and social rates o freturn, so increasingemphasis on these groups would also help to ensure that safety nets are focused on high rate-of-return investments. Targetingmechanisms need to be strengthened- 8.21 Targeting. While most programs use some targeting mechanisms, and while they tend to be pertinent in terms of the risk they address, they also generally lack criteria or tools for effectively targeting their actions and services to the groups with the greatest need. Since the 2001 earthquakes, the Government o f El Salvador has been working, with World Bank technical assistance, to improve its poverty monitoring through the development o f a poverty map. Data from the poverty map, combined with more direct selection mechanisms, will be important to targeting more effectively of the poorest and most vulnerable Salvadorans. Effective targeting will be necessary to ensure the most effective use of resources in reaching poor, vulnerable populations. -as doesprogram coordination and coherence. 8.22 Lack of program coherence and coordination. El Salvador's safety net i s currently less impactful than it could be, in part, because its many small uncoordinated programs lack coherence or strategic vision. Indeed, at present, there i s no institutional body or mechanism that provides strategic direction or supports identification o f priorities, coordination between programs working on similar issues or in similar localities, nor promotes exchange o f information on good practice. The effectiveness o f the safety net would benefit from an institutional process that could help provide a coherent vision, strategic direction, and potentially supporting services (e.g., analysis, data on targeting, etc.). 8.23 Cost-effectiveness. Insufficientdata are available with which to do a detailed analysis o f program cost-effectiveness. Nevertheless, review o fthose programs that benefit children reveal a wide range o f unit costs, suggesting both that some programs may be more efficient than others, and that there i s not yet a proven cost-effective model for intervention. The fact that many o f the existing programs are small means that, whatever other benefitsthey bring, they are probably not able to exploit economies o f scale inprogram implementation. Itwill be increasingly important to identify both effective programs and cost-effective models o f service delivery to promote the "scaling up" o f successful models and to assure the maximum impact o f the safety net for a given budgetenvelop. Directions for public policv A two-pronged strategy- 8.24 Giventhe combination o f chronic poverty and vulnerability and shocks in El Salvador, the country would benefit from a safety net that has two strategic prongs: (1) one prong designed to improve service access to the poorest, most vulnerable groups who have been unable to benefit from El Salvador's traditional sectoral interventions; and (2) another prong that provides assistance to those, especially among the poor, who are detrimentally affected by unanticipated shocks, An important feature o f the safety net is that i s should act as a complement to, not a substitute for the regular programs o f line ministries, private insurance, and households' own risk management strategies. 89 8.25 Strengthening the social safety net will require additional effort and resources to extend program coverage and improve impact, through focusing more strongly on priority risk areas, such as on for poor children and adolescents at risk. To do so, it will also be necessary to improve data and information collection for program targeting, as well as to develop monitoring and evaluation systems to enable the identification o f effective models o f intervention and to ensure program impact. Perhaps most importantly, it will be necessary to establish an institutional framework or process to support the development o f a coherent social safety net as part o f a broader National Social Policy to reduce poverty inEl Salvador. -to improve coveragefor priority groups- 8.26 Strengthening coverage, impact on priority groups. Recent dialogue with counterparts in the Government of El Salvador produced a consensus that the criteria of "irreversibility o f damages" and reducing that reducing the probability o f the intergenerational transmission o f poverty were appropriate. I O 2These criteria also work to focus safety net related resources to areas o f potentially high economic and social returns. Inthis context, priority should be given to programs that focus on (inorder of priority): a 0 Early childhood development, including health, nutrition, early childhood education interventions for poor children, age 0-6 (as well as their mothers) 0 Basic education, raising both enrollment and graduation rates among poor children, age 7-15, and 0 Youth development, including interventions to promote constructive participation o f youth in education, productive work, sports and recreation, community service, and counseling, for poor youth inthe 16-to-23 age range. -andprotect and support those most affected by shocks. 8.27 A safety net this i s designed to protect the poor from the adverse impacts o f shocks will also need to have in place a plan, as well as the institutional capacity for rolling out and scaling up interventions quickly inthe event ofunanticipatedshocks. Experience from other countries in Latin America as well as in Asia in the aftermath o f the East Asian crisis suggests that the effectiveness o f a country's responseto a shock depends inpart on their preparedness, ex ante, to respond (de Ferranti and others 2001). Effective targeting will be important to ensurethe most effective use of resources- 8.28 Improving targeting. Following the 2001 earthquakes, the Government o f El Salvador began an effort, supported by the World Bank, to collect household income and expenditure data at the municipal level (as opposed to the department level) to enable development o f a poverty map for poverty monitoring and program targeting purposes. This effort has been undertaken in two rounds, focusing first on those municipalities most affected by the earthquakes and, subsequently on the remainder o f El Salvador's municipalities. The poverty map generated from these data is an important step toward refining poverty monitoring and targeting anti-poverty and safety net programs. 8.29 At the same time, serious constraints to effective poverty monitoring and targeting remain. Perhaps the most important is the lack o f a recent population census. The most recent census was fielded in 1992 just after the return to peace. But there has been a significant population movement since then, and the census is not currently reliable, nor useful in supporting either poverty monitoring or program targeting. In addition, there are still ways to improve El lo'The criteria andpriorities were discussedwith the Social Cabinetof the GovernmentinDecember 2001, 90 Salvador's household survey. Integrated household surveys that collect data on household consumption as well as income and expenditure - like the World Bank supported Living Standards Measurement Studies (LSMS) surveys are now considered international best practice for poverty measurement. This is in large part because household per capita consumption, rather than income, i s now considered the preferred measure of household welfare. Thus, the collection o f a new population census, together with an enhanced multi-purpose household survey, would go fare to enable effective targeting o fpoor and vulnerable households. 8.30 More direct targeting methods can also be used. In some countries in the region, two-stage targeting has been used to target conditional cash transfers, transfers to poor households that are made conditional upon family investments in their children's education and health, and nutrition. Inthe first stage of targeting, poor localities are identifies; inthe second stage, poor households within the poor localities are identified as beneficiaries of the transfers. This was the case in Mexico's PROGRESA program. Elsewhere, approaches to designing "self-targeting" mechanisms have been used. In cash-for-work (or "workfare") programs, for example, best practice dictates that the offered wage be below the prevailingmarket wage in the locality where the public works program is being undertaken. By settingthe wage below the prevailing market wage, these workfare programs attract those for whom the opportunity cost o f their time is very low - either because they are unemployedor significantly underemployed. Such approaches may also be worth exploring in El Salvador in the context o f designing better targeted safety net programs. -as will strengthened monitoring and evaluation of programs. 8.31 Identifiing cost effective program models. On the basis o f the information collected for the SSNA (2002), it became clear that most safety net programs have deficient or non-existent information and monitoring and evaluation (M&E) systems. Indeed, existing monitoring and evaluation o f programs is weak. Strengthening such systems is important to ensure that program benefits reach their intended beneficiaries and not other groups, that benefits delivered have the desired impact, that administrative costs are reasonable, and that program unit costs can be monitoredto determine the most cost-effective models. 8.32 In addition to building up a domestic information base and M&E capacity to identify cost effective approaches, there are many lessons from recent experience in the Latin America region from which El Salvador could potentially learn.lo4For example, several Latin American countries have been successful in improving the human development outcomes o f poor children through the use o f conditional cash transfer programs, financial assistance made to poor households conditional upon family investments in children's education, health, and nutrition (Box 8.2). Such a model mightbe appropriate inElSalvador ingetting children from the poorest, most vulnerable households into and through basic education and/or through secondary schooling. `03At the nationallevel, informationandM&Esystems are necessary to support strategicresource allocation decisions (see Znstitutionalfiamework, below.) For areview ofrecent experiences insocial protectioninLatinAmerica andthe Caribbean, see Spectrum, "Volatility, Risk, and Innovation: SocialProtectioninLatinAmerica and the Caribbean," World Bank, Fall2003. 91 Box 8.2: StrengtheningPoor Children's Education, Health, and Nutrition through Conditional CashTransfers Evaluation results, available for two conditional cash transfer programs in Latin America, PROGRESA in Mexico and the Red de Proteccidn Social pilot in Nicaragua, show that conditional cash transfers are an effective instrument for increasing the human capital o f the poor through investments in health and education. Conditional cash transfer programs provide cash assistance to poor households on the condition that families make specified investments inthe education, health, and nutrition o f their children. Education. Conditional cash transfer programs have raised enrollment rates for both boys and girls. In Mexico, primary school enrollment rates increased around 1 percentage point from a high pre-program level o f about 90 percent. At the secondary school level, enrollment rates rose 7.2-9.3 percentage points for girls from baseline enrollment rates o f 67 percent, and from 3.5-5.8 percentage points for boys from a baseline o f 73 percent. InNicaragua, program impacts are even more impressive. Average enrollment rates o f children ages 7-13 ingrades 1to 4 intreatment areas increasednearly 22 percentage points as a result o f the program, from a low starting point o f around 70 percent. Program impact on attendance rates are more mixed. In Nicaragua, the RPS produced an increase o f 30 percentage points in the share o f children who had fewer than six unexcused absences during a two-month period. The PROGRESA evaluation showed more pronounced effects on school enrollment than on attendance rates. Child health and nutrition. Evaluations show improvement inhealth and nutrition too. Growth-monitoring visits o f PROGRESA beneficiaries up to three-years-old have increased between 30-60 percent, and beneficiaries up to six years old have a 12 percent lower incidence o f illness compared with control group children. In Nicaragua, around 60 percent o f children under three-years-old participated in nutrition monitoring before the RPS was implemented. After a few months o f program operation, more than 90 percent o f children in RPS areas benefited from nutrition monitoring compared with 67 percent in control areas. The RPS increasedtimely immunization among children 12-23 months oldby 18 percentage points. Consumption levels also grew faster for households participating in conditional cash transfer programs. In Mexico, the average consumption level in PROGRESA households increased 14 percent. Median food expenditure after more than a year o f program operation was 11 percent higher than in control group households. In Nicaragua, control households experienced a sharp decline in consumption due in part to low coffee prices and a drought, whereas the average per capita annual household expenditures in RPS areas did not change. This suggests that conditional cash transfer programs may also help poor people shield their consumption intimes o f crisis. Source: RawlingsandRubio(2003) Establishing an appropriate institutionalframework will be important- 8.33 Institutional framework. Management o f the social safety net requires a coherent and effective institutional process that fulfills the functions o f policy leadership, program coordination, and supervision o f program actions and impact. This institutional role would need to be defined by the government. One model might be to convene a Social Cabinet with responsibility for Social Policy definition and leadership including (but not limited to) social protection. To fulfill its role, the social cabinet would want the support o f a technical team that would provide technical inputs to help establish strategic priorities, undertake analysis for targeting o f safety net interventions, support monitoring and evaluation o f program execution, and define adjustments to the strategy over time, as necessary. Such a technical team could coordinate internal discussions about the kind o f interventions that should be promoted and the appropriate trade-offs between unit costs of interventions and program coverage. The technical team could also undertake responsibility for development o f programs to deal with emergencies 92 due to natural disasters or sectoral crises, as well as explore options to expand insurance coverage for the poor (e.g., pensions, health, unemployment, agricultural, etc.). 8.34 It is important to note that the above describes but one possible approach. Indeed, setting inplace an institutional process does not imply the establishment of a new agency or a set of centralized programs. Rather, what is important i s to put in place the mechanisms to ensure coherence across programs, that safety net interventions complement, not supplant, sectoral initiatives, that programs are well aligned with priority needs, and that effective approaches are identified and scaled up. -as will the commitment of additional budgetary resources. 8.35 Costs. Strengthening El Salvador's social safety net as part o f a National Social Policy to reduce poverty will require additional resources, in light o f low current spending levels. But because the idea o f developing a coherent social safety net is relatively new in El Salvador, it i s proposed that the increase inpublic resources is relatively modest at first, focusing predominantly on expanding outreach and coverage to highest priority groups (i.e., young child in very poor households), as appropriate institutional framework is established, systems for targeting and monitoring and evaluation are established, and effective and cost-effective program models are identified, The intention would be increase public spending incrementally toward regional norms over time, alongside a process of developing increasingly cost effective service models and delivery mechanisms. There is scopefor leveragingpublic resources throughpublic-private partnerships- 8.36 Efforts should also be made to leverage public resources through the further development o f public-private partnerships aimed at assisting and protecting El Salvador's poorest, most vulnerable families. As noted above, over half o f the total resources spent on safety net type programs have come from non-governmental sources, including from the private and NGO sector. In light o f budgetary constraints, it is prudent to continue to identify areas for public- private partnership, thus leveraging public investments in the safety net. Indeed, promoting the active participation o f communities, emigrants, local governments, service organizations, and private sector offers an opportunity to expand the outreach and impact o f the safety net through creative resource mobilization efforts. Recent initiatives, like Programa Unidos Por La Solidaridad (The Unitedfor Solidarity Program), inwhich emigrant groups join with the FISDL to finance social investment efforts in local communities, could provide inspiration for such partnerships (Box 8.3). - a n d through a strategicfocus onpriority risk groups 8.37 The SSNA analysis suggests that by focusing strategically on priority risk groups - specifically children, adolescents, and youth - and buildingon existing programs, it is ossible to have a noticeable impact in the short-term for approximately US $20 milli~n/year.'~~ This is equivalent to one-third o f all public spending on safety nets in 2000, less than 1 percent o f the GOES budget in that year, and less than 0.4 percent o f GDP. At this level o f funding, no less than two-thirds of the additional expenditure should be directed towards poor, at-risk children under 7 years of age for activities to strengthen early physical and cognitive development, with The SSNA (World Bank 2002) estimatedthat by buildingon existingprograms andfinding, an additionalUS $20 million peryear couldfacilitate an increaseinthe number ofpoor children andat-risk youth coveredby early childhooddevelopment, basic education, andyouth development programsbymore than200,000 over a 5-year period. The SSNA made very specific suggestions aboutwhichprogramsought to havebeentargeted, basedonprograms that were either operatingor about to come on stream at the time. While the basic principlesoutlinedinthe SSNA still hold, analysis of the specific programs shouldbe updatedto ensure that funding increases account for the mostrecent developments. The forthcoming PHRD grant on SocialProtectionandLocalDevelopment(FISDL) canbe usedto support suchanupdate. 93 the remainder focused on programs to increase education access and participation among poor and at-risk children and youth, age, 7-15 and 16-23 (e.g., through expansion of scholarships or conditional education-oriented transfers). Box 8.3: Programa UnidosPor La Solidaridad-DevelopingCommunitiesinPartnershipwith Salvadorans inthe Diaspora More than 2.5 million Salvadorans live outside their country - about 1 percent all international migrants worldwide. Out-migrants tend to maintain close ties with their communities o f origin, sending back a steady stream o f income earned abroad. This income is significant, making up as much as 14 percent o f El Salvador's GNP. Traditionally, are generally thought to be used to purchase o f a variety o f consumption goods. But two new trends are emerging: (i)resources from abroad are positively related to the establishment o f household micro-enterprises; and (ii) insome cases, remittanceshavebeenusedto support small, but significant, components o f local social investment projects. Inthis context, the Government ofElSalvador, through the SocialInvestment Fundfor LocalDevelopment (FISDL), has begun to facilitate the active involvement o f Salvadoran communities abroad in local development processes. In the early stages, the FISDL provided information services and promoted links between out-migrants and local communities, to remind communities abroad o f the challenges and priorities facing El Salvador's municipalities. More recently, FISDL has provided concrete mechanisms at the local and governmental levels to facilitate and expand the social investments coming from Salvadoran communities abroad. T h i s process became Programa Unidos Por La Solidaridad (The United for Solidarity Program). The objective o f these efforts is to create a "triangle o f solidarity" among Salvadorans living abroad, local organizations, and the Central Government o f El Salvador, aimed at maximizing resources for local poverty reduction. At the operational level, Salvadoran associations and committees abroad publicly submitproject proposals, specifying the amount o f they are prepared to contribute to its implementation. Periodically, FISDL reviews a group o f proposals, organizes them according to the percentage o f counterpart funds, and assigns complementary resources, giving priority to those projects that offer the greatest counterpart funding inrelationto the total proposal. To date, based on 14 rounds o f proposal reviews, Programa UnidosPor La Solidaridadhas implemented 45 projects worth US$ 11.5 million. This has included US $2.13 million in support from Salvadoran organizations abroad, US $2.28 million from municipal governments, U S $6.92 million from FISDL, and U S $0.12 million other institutions o f the Central Government. These works have benefited the country's poorest communities. In addition to contributing to improve public infrastructure and services - and temporary employment associatedwith construction activities -the program has facilitated the opportunity for Salvadorans inthe Diasporato participate inthe development o f their original communities. Source: FISDL 8.38 Any additional demand for resources in the short-term would require tough strategic choices regarding the budget and m a y lack cost effectiveness in the absence of a clear strategic direction, institutional support mechanism, and "good practice" delivery models. Nonetheless, in the medium to longer-term, as the appropriate institutional processes are developed and the most effective delivery models identified, it is recommended that spending levels move incrementally toward regional norms - roughly 1.5 percent of GDP, on average. Regardless of public funding levels, public resources can and should be leveraged through the continued development of creative public-private partnerships. 94 m i m m 6 6 m m r il 8E -1 8E E z 0 9 9 Ln 0Ln 2 z C e L W PART 3: TOWARD A NATIONAL SOCIAL POLICY FOR EL SALVADOR IX. THE WAY FORWARD Economic growth will remain a key to reducingpaver- 9.1 The experience of the 1990ssuggests that economic growthhas been-andwill continue to be - a key feature o f El Salvador's accomplishments inreducing poverty. This means that there will be high returns in terms of poverty reduction to Government of El Salvador taking the necessary measure to reinvigorate economic growth in the 2000s - through policies and investments that: (i)increaseeducation levels ofthe population, (ii)develop the country's economic infrastructure, (iii) foster greater technology adoption and local innovation, and (iv) improve the investment climate (including efforts to reduce violence and increase the rule of law) (World Bank 2003). -but specific policies are needed to ensurethat all benefitfrom growth: 9.2 The fact that many of the poorest, most vulnerable Salvadorans have been unable to take advantage o f recent growth suggests, however, that it will be increasingly important for El Salvador to put in place a coherent set o f policies and investments to ensure that the poor can share in and benefit from future economic progress. To build effectively on past achievements, it will thus be important for ElSalvador to craft a national socialpolicy that: 0 Buildson recent accomplishments to strengthen thehumancapital ofall Salvadorans, and 0 Strengthens people's access to markets and basic services. 9.3 For many o f the remaining poor, better access to quality education and health care, and greater access to markets and basic services, such as safe water, will be sufficient to enable them to escape from poverty. Nonetheless, there are others who will still lack the capacity to take advantage o f new and emerging opportunities andwho will be especially vulnerable in the face o f shocks. A thirdkey element o f a national social policy will thus be to: 0 Put inplace the instruments to assist and protect the poorest, most vulnerable members o f society. 9.4 Such an approach would help to further reduce poverty by enabling the poor to participate more fully in economic growth. By strengthening the human capital o f the poor and enabling them to be more productive, such a policy would contribute to stronger growth performance o f the Salvadoran economy as well. This chapter summarizes the main policy challenges and pulls together the key elements o f a possible national social policy for poverty reduction. In this context, the chapter highlights the pressing need for improved data for poverty monitoring and targeting o f poverty reduction programs. It also outlines several important strategic choices facing El Salvador as it moves forward. Develop the human capital of all Salvadorans- 9.5 Education. A key element of building the human capital of all Salvadorans involves ensuring that all people have access to education with quality. Strengthening the human capital and, thus, the capabilities o f the poor will be critical to ensuring that they are prepared to take advantage o f emerging economic opportunities in El Salvador. In addition, increasing the education levels o f all Salvadorans, including the poor, will be important to fulfilling the education agenda elaborated in the recent Country Economic Memorandum (World Bank 2003) 99 on promoting faster growth in El Salvador. Indeed, there are significant complementarities between the growth and poverty reduction agendas ineducation. 9.6 Inspite ofrecent advances, El Salvador faces a number of ongoing challenges with respect to assuring that all Salvadorans get a quality education: 0 over 10 percent o f all poor children still do not attend school and another 10 percent are not on track for their age 0 only about 20 percent o fpoor children continue to secondary school 0 less than half o f all poor who enroll in school complete the second cycle o f basic education; less than 25 percent complete the 3rdcycle o f basic education; and only about 16percent o fthe poor who enroll in secondary school complete that cycle 0 and while the quality o f education i s a system-side challenge in El Salvador, education quality appears particularly low among the poor Increase education levels- 9.7 Inthis context, it will be important for the government to focus onmeasures to: 0 continue to improve access to basic education among the poor 0 reduce grade repetition and drop-outs and increase completion rates at basic level 0 introduce efforts to induce more poor grade 9 graduates into secondary schooling (as part o f a broader effort to increase secondary enrollments and achievement for all Salvadorans), and 0 improve education quality at all levels o f education. 9.8 Inlight ofthe broadnationalconsensus on education and the ongoing reformprocess inthe sector, efforts should focus on buildingon recent progress, with special emphasis on addressing the constraints and challenges faced by the poor. -by addressing the affordability of schooling by thepoor- 9.9 For example, a large share o f Salvadoran youth report that they do not attend school because they cannot afford it - in spite o f the fact that basic education is, in principle, free o f charge. Even when there are no tuition costs, families bear the costs o f transportation, textbooks, and other materials. As such, it will be important to take measures to ensure that very poor and vulnerable households can afford education. Increased public investment in targeted scholarships and other programs to reduce the financial constraints facing poor children and their parents, as well as measures to increase the quality o f schools inpoor areas, is thus warranted. -increasing the availability of schools, where supply constraints impede access- 9.10 Although it is less common that households report lack o f school availability as an important constraint to participation, evidence suggests this is an issue at the post-primary level in some localities (e.g., peri-urban areas). As such, it will also be important to strengthen programs o f school construction and rehabilitation inareas where supply constraints appear binding. Inthis context, there may be scope for developing public-private partnerships to take advantage of excess private school capacity in some localities. 100 -and increasing the quality of education. 9.11 While improving the quality o f education is part o f a national agenda, evidence suggests that education quality i s a particular challenge among the poor. Test scores tend to be lower in lower-than in higher-income departments. Moreover, analysis o f returns to education indicates that returns to the same level o f education are lower for poorer, more vulnerable individuals and families than for wealthier ones. In addressing education quality, special efforts should thus be made to understand and confront any special challenges that might affect poor students and/or schools inpoor areas. Ensure access to basic health services- 9.12 Health. Good health is a fundamental part o f people's well-being and i s valued for its own sake. Soundhealth is also essential for people to participate fully and productively insociety and in the economy. A second key element of building the human capital of all Salvadorans thus involves ensuring that all people have access to quality healthcare. 9.13 As with education, key challenges remain to ensure that the poor receive quality healthcare: 0 basic health indicators are still low among the poor o infant mortality is almost twice as highfor low- than for high-income families o chronic malnutritionis about 40 higher than average among the poor o over 15 percent o f low-income children under 5 are underweight, comparedwith four percent of children from wealthier households 0 the poor continue to have poorer access to quality healthcare than the non-poor, although they report a greater incidence of illness 0 the issue of access to quality care is exacerbated by un-repaired earthquake damage to hospitals and other health facilities 0 the cost of medicines remains highby international standards, while quality of drugs is often low. 9.14 Fewer than 20 percent o f all Salvadorans have access to medical insurance either through the Salvadoran Social Security Institute (ISSS) or through private sources. Less than 10 percent and 3 percent o f the moderate and extreme poor, respectively, are insured. A major responsibility o f El Salvador's Ministry o f Public Health (MSPAS) is therefore to cover the population without insurance. And while there have been important advances in recent years, it is estimated that nearly one-quarter o f the population, mostly the poor, still has limited or no access to public health services. -by addressing issues of cost, and confidence in thepublic health system-. 9.15 In this context, poor families report several reasons why they do not access public health care, related to the financial costs, distances to facilities, and to lack o f trust in the public healthcare system. For example, one-quarter o f the poor and nearly one-third o f the extreme poor indicate that they do not use MSPAS facilities due to the financial costs. This includes not only fees for health services (which inprinciple have been eliminated by Executive Decree), but costs o f travel and transport, as well as basic medicines. Indeed, the poor report travel times to the nearest health facility that are as much as 50 percent longer the those reported by the non-poor. And a recent study found that pharmaceutical prices in El Salvador are high by international 101 standards. Moreover, one in every three poor person who gets illindicates that they do not trust the public system and/or prefers to seek alternative cures. These issues needto be confronted to ensure improved outreach by and access to the public healthcare system. -focusing on both demandand supply-side issues. 9.16 Continuing to improve basic health outcomes among the poor will thus require enhanced efforts to ensure access to affordable healthcare among those who lack it, addressing both demand- and supply-side constraints, as well as building greater trust in the public healthcare systemthrough enhancing the quality of services. This will involve: ensuringthat healthfee waivers are made available to those who cannot affordto pay assessing ongoing institutional reform and decentralization efforts to identify ways o f strengthening the outreach and quality of the public health system - particularly with respect to primary health service delivery reforming the health sector's human resource policy to ensure a sufficient supply o f well- trained health care professionals and quality care, particularly inpoor, rural areas, and completing rehabilitation o f the hospitals damaged by the 2001 earthquakes to ensure adequate supply o f secondary and tertiary health services, and undertaking measures to reduce the costs (and improve the quality) o f pharmaceuticals, particularly o f basic medicines. 9.17 Continuing to make progress in improving health outcomes, particularly among the poor, will require efforts on several fronts. Yet, in contrast to education, there is still no broad consensus on the direction o f health sector reforms. Inorder to ensure consistent and sustainable progress inhealth, therefore, a critical element o f a national social policy will be the forging o f a national consensus on the way forward inthe health sector. Improve access to markets and services- 9.18 Basic infrastructure, such as rural water and rural roads, provide both direct and indirect benefits in the fight against poverty - benefits that complement and reinforce investments in human capital development. Access to potable water and suitable sanitation facilities in rural areas is increasingly recognized as an important health input, critically affecting the incidence o f disease and infant and child mortality. Improved access to all weather roads translates into reduces isolation, lowers the costs for goods and services (including accessing education and health facilities), and increases access to markets, enabling people to better take advantage o f emerging economic opportunities. And, as with education, road investments make potentially important contributions to both the poverty reduction and the growth agendas. Investments to increase both access to safe water and to rural roads, will thus continue to be a key element in El Salvador's fight against poverty. 9.19 Several important challenges remain to ensure that the poor have access to safe water and sanitation and, through improved road networks, to markets and a variety o f basic services. With respect to rural water and sanitation: e Nearly half o f those inthe poorest Salvadorans lack access to improvedwater sources. 102 0 In rural areas, over one-third of the extreme poor still get their water from rivers, lakes, and springs where there are often high levels o f pollution, raising their exposure to waterborne diseases. 0 Over 20 percent o f the poorest households also still lack access to adequate sanitation facilities, with potentially important impacts on family and environmental health. Moreover, nearly 9 percent o f the extreme poor in rural areas purchase water from informal sources at as much as three times the cost of the regular water tariff, with significant impacts on their abilities to purchase other basic consumption items. 9.20 With respect to rural roads: 0 many poor rural households remain isolated from all-weather road networks. Indeed, average distances (and travel times) to paved roads remain 80-90 percent higher among the extreme poor than the non-poor, and average distances (and travel times) to markets remain 33-36 percent higher among extremely poor households than among the non-poor. -by completingregulatory and institutional reform in the water sector- 9.21 Rural water. The challenge of ensuring universal access to safe water and sanitation i s made more complex by a complex institutional and regulatory environment in the sector. The water and sanitation sector i s characterized by a multiplicity o f institutions and agencies providing services. ANDA, the lead agency in the water sector, was established and continues to be configured primarily to serve urban areas (or rural settings adjacent to urban areas). In this context, a number o f other institutions, including FISDL (the Social Investment Fund for Social Development), MSPAS, municipal governments, decentralized utility companies, NGOs, and local committees have come to play important roles inrural water provision. 9.22 While the decentralized nature o f rural water provision has potential benefits, the current regulatory environment provides uneven "rules o f the game" across water sector institutions, leading to inefficiencies in the system, considerably variability in the quality o f water systems, and at times, exorbitant water costs to the poor who access water through gray market purchases. As such, a key policy challenge inthe water sector will be to reinitiate and complete water sector reforms. 9.23 Reform efforts were initiated in the late 1990s and then set aside, in part, because the earthquakes temporarily shifted the sectors focus to systemreconstruction. Many technical inputs to a reform are already reasonably developed. The government thus has the opportunity to build on these inputs to complete the reform and put the water sector on a sound footing to improving access to potable water access among the poor. -and continuingprogress in expanding-and maintaining the rural road network. - 9.24 Rural roads. Three institutions are primarily responsible for rural road development and maintenance: the Ministry o f Public Works (MOP), FOVIAL, and FISDL. MOP i s responsible for the rehabilitation, modernization and development o f primary and secondary networks. FOVIAL is responsible for maintenance o f the primary roadnetwork, financed partially through a tax on gasoline. And FISDL works to strengthen municipal management capacity for maintenance o f the tertiary network, which links municipalities, cantons, and villages. 9.25 Recent progress to rehabilitate nation's principal road network, to repair earthquake damage roads, and to promote greater rural road access through Caminos Rurales Sostenibles and other programs has been impressiveand provides useful lessons for the way forward, particularly 103 regarding the importance o f road maintenance. Indeed, by ensuringlongevity o f recent rural road investments, returns to rural roads maintenance are extremely high and should be considered a priority along with continued investments in road expansion. In this context, efforts to ensure that the poor benefit from rural road investments should focus on: 0 continuing expansion andimprovement o f the ruralroadnetwork, 0 developing a program for the maintenance o f secondary and tertiary road networks to ensure the sustainability o f recent investments inrural roads, and 0 continuing development o f local and municipal capacity to build and maintain rural road infrastructure. Support andprotect society'spoorest, most vulnerable households- 9.26 Building a Social Safety Net. The idea o f developing a coherent social safety net is a relatively new one in El Salvador - although it is now a core element in the platform o f the current Salvadoran Government. Social safety programs have an important role to play in El Salvador's poverty reduction efforts for several reasons. 0 Inspite ofbroadsocio-economic advance, the poorest, most vulnerable Salvadorans have largely not participated in or been able to take advantage of recent progress. Continuing lack o f access to education, health, and other basic services raises the risk that these poor will be further left behind and that their weak capacity to take advantage o f emerging economic opportunities will be passedon to their children via weak human capital. 0 ElSalvador is a shock-prone country. While all Salvadorans are potentially vulnerable to shocks, as the recent earthquakes showed, the extreme poor tend to be most vulnerable, both regarding income and basic education, health, and nutritional outcomes. 0 While most households have their own "private" safety nets - most notably migration and remittances - they tend not be sufficient, particular among the poor. Private risk management strategies only partially insure families against risWshocks and, in the case o fremittances, only a fraction of those that need support most receive them. 9.27 Given its combination o f chronic extreme poverty and vulnerability to shocks El Salvador would benefit from a safety net that focuses on: (1) improving service access to the poorest, most vulnerable groups who have been outside the reach o f traditional line ministry interventions; and (2) protecting the poorest from the most harmful impacts o f natural and economic shocks. -by developing a coherent, coordinated, and well-targeted social safety net 9.28 A recent World Bank Social Safety Net Assessment (2002) identified 52, mostly small, social safety net-typeprograms implemented by a range o f government institutions (e.g., MINED, MSPAS, the National Secretariat for the Family) and non-governmental actors (e.g., private sector organizations, NGOs). Review o f these programs suggested that they are generally pertinent, focusing key socio-economic risks faced by the poorest, most vulnerable Salvadorans. Yet, the impact o f these programs tends to be quitemodest due to: 0 Low public spending levels and low programcoverage 0 Under-allocation o f resources to highest priority risk groups 0 Weak program targeting 104 0 Lack o f program coherence and coordination, and 0 Low cost-effectiveness o f programs insome cases 9.29 To ensure that El Salvador's safety net contributes appropriately to sustainable poverty reduction, it will thus be important to: 0 Develop a coherent, strategic vision for the sector 0 Establish an institutional mechanismfor coordinating programs, that would support: o greater emphasis on high priority, high return areas (e.g., early childhood interventions), and o identification and scaling up of cost-effective models, and 0 Continue recent efforts to strengthen the information base for program targeting and programs' abilities to monitor and evaluate their impact. 9.30 To be effective, it will be important that any institutional mechanismhave a strong mandate with the government, the authority to coordinate across traditional ministerial boundaries (since the safety net would needto address a number of cross-cutting issues), and command sufficient resources and technical capacity to effectively play a role.lo6 Improve datafor poverty monitoring andprogram targeting- 9.31 Reaching the poorest, most vulnerable Salvadorans will require an improved data and information upon which to monitor the poverty situation and target programs to at-risk groups. Indeed, enhancing the Government's ability to target poverty reduction programs will be central to El Salvador meeting its future poverty challenging and, in doing so, i s able to make the most effective use o f its scarce fiscal resources. 9.32 Since the earthquakes o f 2001, the Government o f El Salvador has been working, with technical assistance from the World Bank, to collect household survey data (through the EHPM) that would enable development o f a poverty map that would facilitate targeting o f interventions at the municipal level. Prior to 2001, the EHPM data were statistically unbiased only at the department level. These efforts and the forthcoming municipal-level poverty map certainly represent a step forward in El Salvador's ability to identify and target poor and vulnerable households. Nonetheless, efforts to build a sufficiently detailed and disaggregated poverty map continue to be hinderedby lack o f data. -by collectinga newpopulation census- 9.33 Among the most important constraints is the lack o f an up-to-date population census. The most recent census dates from 1992 just after El Salvador's return to peace. This is the last comprehensive accounting o f El Salvador's population distribution and, as such, still underpins even recent estimates o f the population - including the sampling frame for the most recent household surveys. Given the substantial changes in the demographic and geographic composition of the population, especially the substantial internal and international migration, lo`As notedinChapter 8, since initialpreparationof and in-country consultations on this report, a new Government inElSalvadorhas developedandlauncheda unified social safety net program, the Red Solidaria,that aims at strengtheningaccess to basic education, health andnutritionservices and, thus, the human capitalofthe poorest Salvadorans. The programadheres to a numberofthe principles outlined in this report. 105 serious questions exist as to how well recent household surveys capture the geographic distribution o f poverty and vulnerability. lo' - a n d continuing to strengthenthe household survey andprogram monitoring data. 9.34 Other key data and information challenges exist vis-&vis poverty monitoring, and poverty reduction programdevelopment and targeting. These include: The EHPM surveys collect detailed data on household incomes but not on household consumption; yet per capita consumption is considered a more reliable measure o f household welfare and use o f consumption is now considered international best practice for poverty measurement. In several key areas, including development of a social safety net and rural water provision, further work needs to be done to identify the most effective and cost effective models for service delivery. Yes, especially inthe case o f a social safety net, insufficient data exist with which to assess and compare the efficacy o fvarious program approaches. 9.35 In order to strengthen the database for poverty monitoring, program development, and program targeting, it will thus be important to: implementa newpopulation census as soon as is feasible continue efforts to strengthen the national household survey instrument (EHPM) for poverty measurement and program targeting; priorities include consideration o f transforming the EHPM into a consumption survey to enable poverty measurement in accordance with international best practicelog use the new census and enhanced household survey instrument, together, to develop a more detailed, geographically disaggregated "map" for poverty targeting than is currently available log develop indicators and collect o f data on selected safety net (and other poverty-targeted programs) that would enable identification o f the most cost effective service delivery models, as well as ensure effective program monitoring, evaluation, and (if necessary) adjustment o fpoverty targeted and other social programs. Making strategic choices Financing a national socialpolicy will require decisions about strategicpriorities 9.36 The collection o f evidence suggests there are potentially important gains - in terms o f poverty reduction and increased economic participation o f the poor - available through the implementation o f a national social policy. Greater productivity among the poor would also contribute to enhanced economic growth and development. At the same time, given that the lo' the Since initialpreparationof and consultations on this report, the Governmentof El Salvador has scheduledthe fielding of a newpopulation census for 2007. logInthis context, abroaderreview ofElSalvador's povertymeasurementmethodologyshouldbe undertakento ensure that it adheres to internationalbest practice. This shouldinclude not only an assessmentofwhichvariables shouldbe collectedinthe householdsurvey,butwhat methods shouldbe usedfor developingand inflating the poverty line over time. logSee, for example, Poggi, Lanjouw, Hentschel, and Lanjouw (1 998) andLanjouw, Elbers, and Lanjouw (2002). 106 fiscal space for increased spending is limitedin El Salvador, it may be necessary to be selective and make strategic choices about priority areas for action. 9.37 Table 9.1 summarizes, sector-by-sector, the estimated costs o f implementing the elements o f a national social policy discussed in the earlier chapters - including investments to increase access to and the quality o f education, to ensure access to a package o f essential health and nutrition services, to ensure access to safe water and sanitation services, to extend and maintain rural road networks, and to implement a safety net to support and protect the poorest, most vulnerable Salvadorans. To have an impact on remaining poverty will require additional resources, significantly so insome cases. Current Public Target Public Proposed Spendingas a Spendingas a Increaseas a Sector Objective Percenta e of Percentageof Percentageof GDP ? GDP GDP Education cycle); 70 percent net secondary school 100 percent completion o f basic education(3rd 3.2% 5.0% 1.8%* enrollments by 2015 Health Provide access to a minimumpackage of essential health and nutrition services to those 3.8% 4.0-4.1Yo 0.2-0.3%3 who still lack regular health services Water Ensure access to safe water and sanitation among the poor who currently lack access I 0.8%4 I 0.9- 1.1% O.1-0.3% Roads Rehabilitating an additional 1,000 kilometers o f rural roads over 5 years; annual 1.3% 1.4-1.5% 0.1-0.2% maintenance of 3,000 kilometers of rural (secondary and tertiary) roads Safety Net Ensure access to basic service by the poorest, most vulnerable Salvadorans 0.5% 1.5% 1.O% Totals 9.6% 12.8-13.2% 3.2-3.6% primary education (basic c cles 1 & 2) by 2015, it is estimated that education sector spending would have to increaseby 0.5% o f GDP. The estimated health costs are focused on assuring accessto a basic package of health Y and nutrition services to those who currently lack regular health service access in the context o f a strategy to reduce poverty; financing a sector-wide strategy for strengthening the health sector more generally, would require additional financing commitments. 41990-2002average. 9.38 The largest requirements for additional resources are associated with improvements in education and in the development o f a social safety net. It is estimated that achieving universal basic education and 70 percent net enrollments in secondary schooling would require an additional 1.8 percent o f GDP in education sector spending; developing an adequate safety net would require spending o f an additional 1 percentage point o f GDP over time. Only road investments, which essentially doubled between 1999 and 2003, would not require substantial increases inspending. 9.39 While commitment o f additional resources in any specific sector may appear within reach, the joint costs o f implementing the proposed elements o f a national social policy pose a greater challenge - an impliedspending increase o f between 3.2 and 3.6 percent o f GDP (Table 9.1). In this context, serious consideration should be given to measures that would increase government revenues. But with or without additional revenue measures, some effort to consider relative priorities and phasing o fpolicies and investments are warranted. 107 Education and rural road investments arepoised toprovide high returns in the short-term 9.40 Given that additional investments in education and roads will likely contribute both the poverty reduction and economic growth agendas, these are likely to be particularly high return investments. Inaddition, recent developments inboth sectors-progress on the reform agenda in education, the success o f Caminos Rurales Sostenibles - suggest the sectors are poised to use additional resources productively. Building consensus on health and water reforms would raise returns in those sectors 9.41 With respect to health and water and sanitation, there is less consensus about direction o f sectors or about sectoral reforms. Inthis sense, it will be important to build a national consensus on the ways forward, before committing substantial additional resources. Indeed, reforms that define a strategic direction, including the appropriate roles for the private as well as the public sector, could pave the way to significant efficiency gains and, thus, a greater leveraging o f available resources. Establishing an appropriate institutionalframework will be critical to an effective safety net. 9.42 In the case o f development o f a social safety net, since this i s a relatively new area o f policy and program development in El Salvador, it will be important to resource programs sufficiently to ensure an impact. At the same time, it may make sense to increase public resources incrementally over time, to ensure that an appropriate institutional framework i s developed and cost-effective service models and delivery mechanisms are identified. Indeed, phasing in of spending increases to provide the Government o f El Salvador a clear and predictable time horizon for increasing its revenue base may be both a sensible and pragmatic approach to implementingmultipleelements o f a national social policy. There is scopefor leveragingprivate resourcesfor effective poverty reduction. 9.43 Finally, independent o f spending levels adopted, ongoing efforts should also be made to leverage resources through productive and creative partnerships - between public and private entities, between the Central and Local Governments, and between Salvadoran emigrants and their communities, friends and families who remain inthe country. While El Salvador has made important advances in this area recently, there remains considerable potential through such partnerships to enhance poverty reduction efforts and national development as a whole. 108 REFERENCES Arias, Omar, 2004, "Poverty in El Salvador During the 1990s: Evolution and Characteristics, " draft, Poverty Reduction and Economic Management Unit, Latin America and the Caribbean Region, World Bank, Washington, D.C.* Beneke de Sanfelih, Margarita y Mauricio Shi, 2004, "Dinhmica del ingreso rural en El Salvador, borradorJina1, FUSADES, San Salvador, El Salvador.* Beneke de Sanfelih, Margarita, and Claudio Gonzalez-Vega. 2000. "Dynamics o f Rural Household Incomes inElSalvador: 1995-1997 Panel Results," Paper submitted to the 2000 Annual Latin American and Caribbean Economic Association, Rio de Janeiro, October 12-14, 2000. CEPAL, 2003, "Panorama social de America Latina, 2000-2001,"Santiago de Chile. Conning, Jonathan, Pedro Olinto, andAlvaro Trigueros. 2001. "Managing Economic Insecurity inRuralElSalvador: The role of asset ownership andlabor market adjustments," reviseddraft. http://www.williams.edu/Economics/wp/conningsalvador.pdf Cuellar, N.and S. De Larios, 2001, "Acceso a1agua potable en El Salvador: tendencias, perspectivas y desafios", Prisma, No. 42. DiGropello, Emanuela, 2004. "El Salvador: Education Strategy Paper," HumanDevelopment Department, Latin America and the Caribbean Region, World Bank, Washington, D.C. Dollar, David and Aart Kraay 2002, "Growth is Good for the Poor," Journal of Economic Growth (U.S.), Vol. 7, NO.3, pp. 195-225. FESAL (EncuestaNacional de Salud Familiar), El Salvador's National Family Health Survey, various years, www.fesal.org.sv . Glewwe, Paul and Gillette Hall. 1998. "Are Some Groups More Vulnerable to Macroeconomic Shocks than Others? Hypothesis Tests Based on PanelData from Peru," Journal of Development Economics, Vol. 56, pp. 181-206. Gobierno de El Salvador, 2004, El Salvador: Primer Informe de Pais: Avance de 10s Objetivos de Desarrollo del Milenio, San Salvador, El Salvador. Jalan, Jyotsna, and Martin Ravallion. 1999. `LArePoor Less Insured? Evidence on Vulnerability to Income Risk inRural China," Journal ofDevelopment Economics, Vol. 58, pp. 61-81. Klugman, Jeni, Diana Kruger, and Kate Withers. 2002. "Consumption Risk and Smoothing DuringDisasters: The Case ofHurricane MitchinNicaragua," drafl, Poverty Reductionand Economic Management Network, The World Bank, Washington DC. L a EncuestaNacionalde Salud Familiar (FESAL), various years. Lanjouw, Jean O.", Chris Elbers, and Peter Lanjouw, 2002, "Micro-Level Estimationo f Welfare," Policy Research Working Paper 2911, World Bank, Washington, D.C. * This backgroundpaper to the El Salvador Poverty Assessment will be postedon the World Bank website. 109 Leipziger, Danny, Marianne Fay, Quentin Wodon, and Tito Yepes, 2003, "Achieving the MillenniumDevelopment Goals: The Role ofInfrastructure," draj', Financialand Private Sector and Infrastructure Department, Latin America and the Caribbean Region, World Bank, Washington, D.C. Lewin, Bryan, and Daniele Giovannuci. 2002. "Global Supply & Demand: New Paradigms in the Coffee Markets," draj', RuralDevelopmentGroup, Environmentally & Socially Sustainable Development Network, The World Bank, Washington, DC. Loayza, Norman, Pablo Fajnzylber, and Calderbn, 2002, "Economic Growth inLatinAmerica and the Caribbean: Stylized Facts, Explanations, and Forecasts", processed, Development Research Group, World Bank, Washington, D.C. Marques, Jose, 2004, "El Salvador: Evaluation o f Public Expenditure on Education and Health," Human Development Department, Latin America and the Caribbean Region, World Bank, Washington, D.C.* Marques, Jose Silverio and Ian Bannon, 2003, "Central America: Education Reform in a Post- Conflict Setting, Opportunities and Challenges", Conflict Prevention and Reconstruction Unit, Working Paper 4, World Bank, Washington, D.C., April. Management Sciencesfor Health, 2004, Findings o f a Study o f InternationalPharmaceutical Prices, www.msh.or~/sea/3.1.2b.htm. Mason, Andrew D.,Ana Cristina Torres, and Reyes Aterido, 2003, "Benefit Incidence o f Public Education and Health Spending inEl Salvador," drafl, Human Development Department, Latin America and the Caribbean Region, World Bank, Washington, D.C. Ministerio de Salud Phblica y Asistencia Social (MSPAS), 2001, "Institutional Development Strategic Plan,'' Government o f El Salvador, San Salvador, El Salvador. Ministerio de Salud P~blicay Asistencia Social (MSPAS), 2001, "SIBASI: Marco Conceptual y Operacional," versi6n revisada, Government o f El Salvador, San Salvador, El Salvador. Morduch, Jonathan. 1999. "Between the State and the Market: Can Informal Insurance Patch the Safety Net?" The WorldBank Research Observer,Vol. 14, no. 2 (August), pp. 187-207. PanAmerican Health Organization (PAHO), 1994, "Andlisis del Sector de Agua y Saneamiento en El Salvador", Washington D.C. Poggi, Javier, Jean Olson Lanjouw, Jesko Hentschel, and Peter Lanjouw, 1998, "Combining Census and Survey Datato Study Spatial Dimensions o f Poverty," Policy Research Working Paper 1928, World Bank, Washington, D.C. Ravallion, Martin and Monika Huppi, 1991, "Sectoral Structure o f Poverty Duringan Adjustment Period: Evidence for Indonesia inthe mid-l980s," WorldDevelopment, 19: 1653-78. Rawlings, Laura B. and Gloria M. Rubio, 2003, "Evaluating the Impact o f Conditional Cash Transfer Programs: Lessons from Latin America," World Bank, Washington, D.C. * This backgroundpaper to the El Salvador PovertyAssessment will be postedon the World Bank website. 110 Rodriguez-Meza, Jorge, and Claudio Gonzalez-Vega, 2004, "Household Income Dynamics and Poverty Traps inEl Salvador," draji, Ohio State University,Columbus, Ohio.* Segovia and Larde, 2000, "Hacia unperfil de lapobreza en El Salvador en la decada de 10s Noventa," MECOVI final report, reprint. Shorrocks, Anthony and Rolph van der Hoeven (eds.). 2004. Growth,Inequality, and Poverty: Prospectsfor Pro-poor Economic Development. Oxford UniversityPress, Oxford: NewYork. Skoufias, Emmanual. 2002. "Household Vulnerability to Risk: Estimates from the PROGRESA EvaluationSample inRuralMexico," draft, International FoodPolicy Research Institute, Washington, DC. Skoufias, Emmanuel, and Agnes R. Quisumbing. 2002. "Consumption Insurance and Vulnerability to Poverty: A Synthesis o f the Evidence from Bangladesh, Ethiopia, Mali, Mexico, and Russia," draji, International FoodPolicy Research Institute,Washington, DC. Spectrum, 2003, "Volatility, Risk, and Innovation: Social Protection in Latin America and the Caribbean",World Bank, Washington, D.C. Fall. Todd, Jessica, and Norman Hicks, 2003, "A Brief Examination o f the Relationship between GDP/capita and Performance on Selected Social Indicators in the Latin America and Caribbean Region," draft, Poverty Reduction and Economic Management Unit, Latin America and the Caribbean Region, World Bank, Washington, D.C. Trigueros, Alvaro and Carolina Avalos, 2004, "Analysis o f the Poverty and Social Impacts o f the Coffee Crisis (Price Shock) in El Salvador," draft, FUSADES, San Salvador, El Salvador.* UNDP, 2003, "Desafios y opciones en tiempos de globalizacibn," Informe sobre Desarrollo Humano El Salvador 2003, San Salvador, El Salvador. Vakis, Renos, Diana Kruger, and Andrew D. Mason. 2003, "Coffee and Shocks: Lessons from Nicaragua," draft, Poverty Reduction and Economic Management and Human Development Sector Management Units, Latin America and the Caribbean Region, World Bank, Washington, D.C. WHO and UNChildren's Fund, 2000, Global Water Supply and Sanitation Assessment Report. Wodon, Quentin et al, 2002, "Poverty in Latin America: Trends 1986-1998 and Determinants," processed, Poverty Reduction and Economic Management Unit, Latin America and the Caribbean Region, World Bank. Washington, D.C. World Bank, 2001, Earthquake Emergency Reconstruction and Health Service Extension Project, Project Appraisal Document, Report No. 22626-ES, Washington, D.C., October 31. World Bank, 2002, "El Salvador Social Safety Net Assessment," Report No. 2419&ES, Poverty Reduction and Economic Management Sector Unit, Latin America and the CaribbeanRegion, Washington, D.C. * This backgroundpaper to the El Salvador Poverty Assessment will bepostedon the World Bank website. 111 World Bank, 2003, "El Salvador Country Economic Memorandum," Report No. 26238, Volume I,CentralAmericaCountryManagementUnit,LatinAmericaandtheCaribbeanRegion, Washington, D.C. World Bank, 2003, "El Salvador Country Economic Memorandum,", Report No. 26238, Volume 11, CentralAmerica.Country ManagementUnit,LatinAmerica andthe CaribbeanRegion, Washington, D.C. World Bank, 2004, "Republic of El Salvador: Public ExpenditureReview", Central America Department, LatinAmerica and the CaribbeanRegion, Washington, D.C. World Bank, 2005, Social ProtectionProject, Project Appraisal Document, Report No. 32648- SV, September 26. WorldDevelopment Indicators 2002, World Bank, Washington, D.C. WorldDevelopment Indicators 2003, World Bank, Washington, D.C. WorldDevelopment Indicators 2004, World Bank, Washington, D.C. WorldDevelopment Indicators 2005, World Bank, Washington, D.C. Yepes, Tito, 2004, "Infrastructure and Poverty Reduction in Rural El Salvador," drajl, Poverty Reduction and Economic Management and Human Development Sector Management Units, Latin America and the CaribbeanRegion, World Bank, Washington, D.C. 112