Report No. 21833-VE Bolivarian Republic of Venezuela Investing in Human Capital for Growth, Prosperity and Poverty Reduction March 30, 2001 Colombia, Mexico and Venezuela Country Management Unit LCSPR Management Unit Latin America and the Caribbean Region Document of the World Bank CUIRRENCY EQUIVALENT (As of March 23. 200(1) Cuirrcnicy I nil: Itiolivar (lis) US$1: 705 Bs FISCAL YEAR June 15, 2000 ACRONYMS AND ABBREVIATIONS UBN/NBI Unsatisfied Basic Necds/Necesidade.s Btiscas Insalisfichas CEM Country Economic Memorandum ECLAC/CEPAL United Nations Economic Commission for Latin America and the CaribbeanlComisi6n Econ6niica para America Lat/ina CORDI PLAN Ministry of Planning and Development/Oficina de Est ado de Coordinacidcn y Plan~ficacicin, currently Ministeno de Planificacitn y Desarrallo CTV Venezuelan Central Workers Union/C'entral de Trabajadore.s de Venezuela FIDES Inter-Governmental Decentralization Fund/Fondo Interguhernamentalpara la De.scenfrulizacidn FONVIS Venezuelan Social Investment Fu,ld/Fondr, Venezolanso de Inversion Social FUS Unified Social Fund/Fonida Unico Social GDP Gross Domestic Product HDI Human Development Index HIV/AIDS Human Immunodeficiency Virus/Acquired Immunodeficiency Syndrome IDB/IADB Inter-American Development Bank ICRG International Country Risk Guide IESA Institute for Advanced Studies in Administration/Inst/tuto de Estudios Superiores de la Administracidn INN National Nutrition Institute/instilm/o Nacional de Nitrici/dn IVSS Venezuelan Social Security Institute/lnstiunto Venezolano de Seguridad Social LACSO Lahoraiorio de Ciencias Sociales LFS Labor Force Survey/Encuessa de Ilogares por Muestreo LOSSSI Comprehensive Social Security System Law/Ley O(Jgdnicea del 'istema de Seguridad Social Integral LSMS Living Standards Measurement Survey PL/LP Poverty Line/Linea de Pohreza Ml] Ministry of Health NGO Non-Govemmental Organization OCEI Central Statistical and Information Oflice/O/icina Central de E.sradistica e Informalica OECD Organisation for Economic Co-operation and Development OLS Ordinary Least Squares OPAM Macroeconomic Analysis and Program ming Office/kicinad c/ Prograrnacion y An6lisis Macroecondmico OPEC Organization of the Petroleunm Exporting Countries PAHO Pan-American Health Organization PAMI I'rogram for Mother and Child Care/Programa de Atencicdn Materno-hnfantil PDVSA Venezuelan Petroleum Company/Peir6leos de Venezuela SociedadAnonima PPP Purchasing Power Parity UNDP/PNUD United Nations Development Programme/Programa de las Naciones Unidaspara el Desarrollo UNESCO United Nations Educational, Scientific and Cultural Organization VAT/IVA Value Added Tax/lmpue.uvo al Valor Agregado WHO World Health Organization Vice President LCR: David de Ferranti Director LCC IC: Olivier Lafourcade Director LCSPR: Ernesto May Sector Manager: Norman L. Ilicks Sector Leader: Marcelo Giugale Team Leader: Vicente Fretes-Cibils BOLIVARIAN REPUBLIC OF VENEZUELA: INVESTING IN HUMAN CAPITAL FOR GROWTH, PROSPERITY AND POVERTY REDUCTION* Table of Contents Executive Summary ............................................................. Chapter I. Introduction ............................................................ 1 Chapter II. Income, Poverty and Inequality ...........................................................3 I. The Economic Environment . ............................................................3 II. Poverty and Inequality-Levels and Trends .........................9....................9 III. Conclusions and Recommendations ....................................................... 18 Chapter III. Social Expenditure-Shocks, Policy Response, and The Institutional Framework . ..................................................... 19 I. Trends in Public Sector Expenditures ..................................... .................. 19 II. Public Expenditure Responsiveness to Oil Price Shocks ....... . ................. 20 III. Composition of Social Sector Expenditures . ......................... .................. 23 IV. Improving the Social Safety Net System ............................... .................. 26 V. The Institutional Framework .................................................. .......... 34 VI. Conclusions and Recommendations ...................................... .................. 40 Chapter IV. How to Improve Human Capital and Prevent its Deterioration? ... 41 I. Investing in Education ............................................................ 41 II. Investing in Health and Nutrition Services ............................ .................. 55 HI. Investing to Reduce Crime and Violence .............................. .................. 69 Annex A. Summary of Economic Plan and Objectives for 2000 ........... .................. 81 Annex B. The Link Over the Oil Cycle ................................................... ......... 83 Annex C. Education-Methodology for Estimating Rates of Returns .................... 86 Annex D. Health Sector and Nutrition-Statistical Appendix .............. .................. 87 Annex E. Crimes and Violence-Statistical Appendix .......................... .................. 90 * This Report was prepared by V. Fretes-Cibils (Team Leader), M. A. Cuevas, H. A. Patrinos and K. Scott. The Report has benefited from comments provided by A. M. Arriagada, B. Carlson, M. M. Colmenares, C. Corbett, D. Cotlear, M. Giugale, N. Hicks, P. Marquez, F. Saez, A. Solimano, E. Velez, L. Wallace and D). Yuravlivker. A. Legovini and J. S. Hentschel were the peer reviewers. M. Lane coordinated the production of the Report. EXECUTIVE SUMMARY To provide the Government with more analytical information for making policy decisions to revert the deterioration of living standards and reduce poverty, this Report draws on limited available data to analyze selected economic and social issues. These issues include (i) better understanding of poverty and inequality and their relationship to real income; (ii) improving the allocation of social expenditures, and increasing the efficiency and effectiveness of social programs; and (iii) preventing the continued deterioration of social and human capital, while simultaneously promoting its accumulation. The economic performance of the Bolivarian Republic of Venezuela (Venezuela) over the last decade has been dismal, and economic policy-makers have been unable to cope with the oil cycles and the general declines in real oil prices and revenues. This has resulted in low economic growth, with bursts of inflation and recurrent recessions (Graph 1). In the 1990s, real GDP per capita growth averaged -1.2 percent per year and the inflation rate averaged about 40 percent per year (with a declining trend over the last five years). For the same period, real non-oil GDP per capita growth averaged -2 percent per year. Graph 1. Econonuc Cycles, 1979-2000 15.00% - Unemployment Rate 10000/0 - g / -5.00% - , i-trw K 10.00% /' ' I ,'~~~K ~ GDP GrovAth 05.00% oi , , X ' OIG , 8 -10.00% O \ 0 0\ 0 h o Moreover, Venezuela's dependence on the oil sector deepened, with the share of oil-GDP in total GDP increasing from about 21 percent in the early 1990s to about 26 percent in the late 1990s. At the same time, the share of agriculture in total GDP decreased (from about 6 percent in 1990 to about 5 percent in 2000) and manufacturing decreased (from about 13 percent to about 11 percent), while the service sector held at about 47 percent of real GDP.' The share of non-oil exports to total exports remained, I IThroughout the Report, the statistical data for year 2000 are preliminary. on average, below 25 percent, reflecting the lack of export diversification away from the oil sector, and the negative impact of the overvalued domestic currency on external competitiveness over time. In this environment, real wages fell significantly and unemployment rose. Between 1990 and 1999, average real wages fell by about 23 percent-despite the existence of mandatory minimum wages for most of the period. The drop affected all sectors of the labor market, including the oil sector. Even so, unemployment rose, affecting those individuals that (i) had relatively little human capital; (ii) were young and female; and (iii) were urban residents and/or recent migrants. Not surprisingly, labor Graph 2. Informal Sector productivity also decreased, falling at an Helped Buffer Unemployment average annual rate of about one 65% 17% percent. This reflects low real economic 60% 15% growth and the informal sector's role as Informality Rate a buffer. Employment in the informal 13%/o 9 50%1% sector reduced potential unemployment so% in the formal sector, and the counter- I\ 5 ~~~~~~~9% cyclical pattern of changes in the 40% informal sector's share of total 35% 7% employment softened the impact of 30% Unemployment5% recessions and economic contractions on-I 80 8 8 unemployment (Graph 2). _ _________ Poverty and Inequality As a result of negative per capita real economic growth, falling real wages, higher unemployment, and lower labor productivity, poverty and inequality increased during the 1 990s, although remaining below the average for the LAC region throughout this period (Graphs 3 and 4). Other welfare indices and social indicators also showed a significant drop. -1 i- Graph 3. Poverty is Up... but Below the LAC average 40 -- X~ 305_g 40 ~.30 30 25 4'20 o15 rv o o 0 Co o O 00o -+.-LAC REGION _ Ven: Offic: flat A Ven: PPP: Flat Ven: Offic: Y source vK Ven: PPP: Y source Graph 4. Inequality is Also Up, but Below the IAC Average 0.62 U 0.57 0.52- 0.47- 0.42 s0 ON N Cl 00 00 00 a,NO xl x a, .9 as & -_--LAC REGION -4-Ven: PC, Flat A Ven: PC, YSource Note: The LAC Region estimates are based on data representing 71 percent: of the population in the region although it includes only 12 countries. The data for Venezuela is per capita income taken from the Labor Force Survey, which is used to estimate poverty lines and inequalities. More specifically, "Offic." stands for the official poverty line of the Government, while "PPP" stands for the PPP US$2 per day poverty line. "Flat" refers to an adjustment made for underreporting of income wherein a coefficient of adjustment is applied to all incomes; the coefficient is the ratio of labor incomie in the national accounts to wage income in the household survey. "Y source" is a (ifferent adjustment for underreporting of income wherein separate coefficients are calculated for different income sources. This uses a coefficient applied to wage income-wage income in national accounts to wage income in the survey-and self-employment or employer income-labor non-wage income in national accounts and labor non-wage income in household survey. Source: Inter-American Development Bank, OCEL, and World Bank. - 111 - How strong was the impact of real income and inequality on poverty? To answer this question, we estimated changes in the poverty rate as a function of changes in real per capita income and changes in the Gini coefficients. The estimated parameters (i.e., the average elasticities) indicate that changes in real income are negatively correlated with changes in the poverty rate, while changes in inequality are positively correlated with changes in the poverty rate.2 A 10 percent growth in real per capita income, ceteris paribus, would reduce the poverty rate by about 12 percent on average. To reverse the deterioration in living standards and reduce poverty, Venezuela needs to restore and sustain real income growth. This could be achieved by boosting investment and increasing labor productivity, which in turn involves preventing any further deterioration in social capital and accelerating the accumulation of both human and social capital. However, a word of caution is necessary. Given the problems with data and methodologies on poverty and inequality, a survey or system of surveys (e.g., Living Standards Measurement Surveys) must be instituted to facilitate measurement, monitoring, and policy formulation. This should (i) facilitate the construction of a consumption-based measure of welfare, linking consumption levels to other facets of welfare-such as levels of health and education, access to social services, and the incidence of government spending; (ii) include elements of a panel design to track the same households over time; and (iii) cover both the rural and urban populations. Social Expenditures But what else can explain the deterioration in living standards and social indicators? Have changes in social expenditures contributed to the deterioration? Is the problem how Venezuela has responded to the fluctuations in oil prices? To answer these questions, we examined the relationship between government expenditures, social expenditures, and oil prices (in real terms), filtering and decomposing these relationships over both the trend and the cycle components. The analysis shows that over the long- term (1970-1998) there is a positive and statistically significant relationship between real government expenditures (both total and social) and real oil prices. However, over the period of continuous real oil price declines (1981-1998), the analysis shows that the relationship between real government expenditures (both total and social) and real oil prices is negative. Nevertheless, the estimated coefficient for social expenditures is larger in absolute terms than for total expenditures, implying that increases in the trend of social expenditures are larger than for the trend of total expenditures in response to declines in the oil price trend. These results, therefore, indicate that in the long-term real government expenditure is closely and positively associated with real oil prices, and that 2 Indeed, the impact of real income growth on poverty in Venezuela is consistent with the general findings in "Growth Is Good for the Poor" by Dollar, D. and Kraay, K. (World Bank, March 2000). In this paper, the authors find that income of the poor rises one-for-one with overall economic growth. This general relationship between income of the bottom fifth of the population and per capita GDP holds in a sample of 80 countries covering four decades. The paper also finds that policy-induced growth (including openness to foreign trade, good rule of law, fiscal discipline and low inflation) is as good for the poor as it is for the whole economy. The paper, however, does not find evidence that formal democratic institution or public spending on health and education have systematic effects on incomes of the poor. - iv - real social expenditures have been "protected" with respect to downward trends in real oil prices. For the cyclical movements, the results indicate that while there is a. statistically significant and positive relationship between the cyclical component of real government expenditures and that of real oil prices, no statistically significant relationship is found between the cyclical component of real social expenditures and that of real oil prices. While adjustments in total real expenditure are related to the trend components of real oil prices, they are also affected by the real oil price cycle, supporting the hypothesis of pro- cyclical fiscal behavior. However, real social expenditures are not significantly affected by "temporary" changes in real oil prices. Therefore, the direct impact of the oil price does not seem to be at the root of the observed deterioration in social sector output performance. Were social expenditures "protected" during fiscal contraction episodes and did some social sectors fare better than others did? To answer these questions, we estimated elasticities of social sector expenditures (i.e., changes in social sector expenditures with respect to changes in total expenditures). Over the period 1981-1998, social expenditures and education expenditures were indeed "protected", as carn be inferred from the estimated elasticities (Table 1). Social development (which includes social safety net programs) and social security expenditures have negative estimated elasticities, indicating that in years when total government expenditures fell, expenditures in these sectors actually increased. But health and housing were not "protected", as changes in total expenditures led to more than proportionate changes in expenditures in these sectors. What happened during shorter fiscal contraction episodes? For the 1980s, the elasticities show that social expenditures were not used disproportionately as adjustment variables to achieve expenditure cuts. While public expenditures on education were the least "protected", expenditures on social development increased despite the overall contraction in public sector expenditures. During the 1990s, however, education and social development were relatively more "protected," whereas health and housing made disproportionate expenditure sacrifices. Outlays on social security, by contrast, moved in the opposite direction, rising in the face of overall expenditure cutbacks. Table 1. Social Expenditures Are Protected Fiscal Contraction Episodes 1981-1998 1992-1995 1982-1989* Elasticity % change Elasticity % change Elasticity Total expenditure ---- -8.57 ----- -10.00 ---- Social expenditure 0.76 -7.58 0.88 -8.87 0.89 Education 0.76 -2.98 0.35 -9.19 0.92 Health 1.29 -11.64 1.35 -6.19 0.62 Housing 2.53 -22.37 2.63 -8.12 0.81 Social Development -0.36 -1.57 0.18 27.07 -2.71 Social Security -1.09 2.33 -0.27 -2.80 0.28 Note: The elasticities are calculated with respect to changes in total expenditures; (*) means that the years: 1984 and 1987 are excluded (expansionary years). - v Have social safety net programs cushioned the impact of the contractions? A detailed look at these programs raises questions about whether they were targeted at the poor and if the most needy received the majority of the benefits. The State level distribution of resources was reasonably good for some programs and poor for others, but we do not know whether the poorest actually received benefits.3 Moreover, as poverty indicators are weak on the depth and severity of poverty, the distribution of resources at the State level may not be optimal-it is possible that the distribution of extreme poverty is substantially different from that of overall poverty. An analysis of the institutional framework for the social sectors indicates that there are multiple institutions and officials responsible for the design, implementation, monitoring, and supervision of social sector policies. Programs have not had clear and common objectives, and have lacked appropriate incentive-performance mechanisms. Furthermore, the public has (i) had limited mechanisms to influence the design of policies; (ii) not exercised sufficient participation in the monitoring and supervision of social programs; and (iii) lacked the necessary means to regularly enforce accountability of elected officials or bureaucrats. This situation, which is not unique to Venezuela, has resulted in inefficient public resource use. Various "social investment funds"-aimed at coordinating the design, promotion, execution, and financing of specific social programs-have been created at different times to try to overcome these deficiencies. The hope was that given their structure and relative flexibility, the social funds could respond more effectively than the more traditional institutions. Indeed, the funds were initially charged with implementing emergency programs designed to alleviate the temporary distress caused by economic shocks and adjustment programs. Unfortunately, however, a multitude of funds was superimposed on the pre-existing institutional framework, making the overall institutional infrastructure even more complex. Rather than correcting the lack of participation of target populations in the design of social policy, the funds solidified the creation of parallel structures with overlapping mandates. In this complex and overwhelming setting, and with weak administrative capacity, the system gave way to "informality" and discretionality, thereby increasing the possibilities for corruption. In summary, in Venezuela the level of expenditures has not been the most pressing problem, as social expenditures have been relatively "protected" vis-a-vis oil price fluctuations and fiscal adjustments. The problem has lied primarily with the efficiency with which resources have been allocated and used. Social sector institutions have lacked the capacity to formulate, coordinate and implement policies with clear objectives and responsibilities. As a result, a large number of social funds have emerged, but have only made the overall institutional infrastructure even more complex. Furthermore, coordination and incentive monitoring and evaluation mechanisms have not existed, and the funds have been weak or simply not worked at the intra- and inter- institutional levels. This has been exacerbated by weaknesses in the broader institutional and legal framework, and public sector culture (including weak administrative capacity). 3 To know that, one would need to estimate the correlation between poverty and benefit receipt by individual or household. Required data is not available. - vi - Specific proposals to improve the efficiency in the allocation and use of existing resources, thereby enhancing social sector performance, could include (i) defining clear social policy objectives; (ii) strengthening the Social Ministries by prioritizing administrative capacity building, reducing discretionality and thereby limiting possibilities for corruption; and (iii) simplifying, streamlining, and increasing the administrative efficiency of social programs to improve their cost-effectiveness, notably by merging and simplifying social sector institutions and funds. Accelerating Human and Social Capital Development Education. Despite inefficiencies and ineffectiveness found in social programs, an analysis of the education sector suggests that the proportion of the labor force completing basic education continues to increase. The relative increase in school attendance between 1990 and 1995 was higher among those populations classified as having the lowest probability of attending (older rural boys) than among those with a higher probability of attending (younger urban girls). This suggests that, in general, some education programs have reached their target populations. Even so, the quality of basic education remains below international standards, significant inefficiencies and inequalities remain in the school system, and the incidence of public expenditure in education continues to be regressive. At grade school and high school levels, internal efficiency is low and shows no signs of improving. The repetition rates observed during the 1980s and 1990s show an average of about 10 percent of students from the first to eleventh year repeats at least once. The dropout rate in the first six grades has been rising since 1989, despite the targeting of this group by social programns. Although it mnay sound contradictory, it is possible that the dropout rate is positively correlated with increased attendance, because those who attend school exclusively to benefit from the social programs are also those with a higher propensity to drop out. During the last two decades, the highest dropout rate has taken place between the seventh and ninth grades, and this could be explained by the fact that Venezuelans, mainly the relatively poor, have not yet fully adapted to the nine-year basic education system, and the incentives to keep their children in school beyond the sixth grade may not yet be sufficiently strong. For higher education, the internal efficiency (measured as the proportion of the population that graduates in the prescribed time frame) is higher in universities than in institutes and junior colleges, and higher in private than in public institutions. One hypothesis is that the low internal efficiency observed in institutes and junior colleges is connected with the strong presence of young people with fewer resources. As for external efficiency, analysis demonstrates that the returns to schooling have been declining at all levels over the last 15 years, partly reflecting the economy's dismal performance (Table 2). Nevertheless, rates of return on education remain relatively high and investment in education is economically and socially profitable (especially for primary education). The private rate of return to primary education, on average, does not seem to differ much from the private rate of return to university - vii - education. But the social rate of return of primary education usually (although not always) outstrips the social rate of return of university education. Table 2. Education is Still a Good Investment (percent) Private Rates of Return 1980 1982 1985 1988 1990 1993 1996 Primary (vs. none) 19.1 15.5 17.3 14.8 17.7 14.7 18.9 Secondary (vs. primary) 12.6 16.5 17.5 11.0 8.4 9.7 12.2 Technical (vs. secondary) - - 17.3 19.7 14.5 19.5 25.0 University (vs. secondary) 17.0 16.4 17.2 16.4 14.0 18.0 14.8 Social Rates of Return 1980 1982 1985 1988 1990 1993 1996 Primary (vs. none) 15.3 11.9 11.9 11.7 12.8 9.0 12.2 Secondary(vs. primary) 11.0 11.8 14.1 9.9 7.7 8.2 10.1 Technical (vs. secondary) - - 13.6 16.8 11.0 11.8 9.6 University (vs. secondary) 11.0 10.7 12.2 13.2 9.8 11.3 6.9 Note: -- indicates that the data are not available. Against this background, some policy recommendations for the Government include the following: * Continue to implement reforms at the primary education level and monitor the impact on repetition and dropouts; * Accelerate development of the new curriculum for the third phase of basic education in an effort to decrease the high dropout rates; * Intensify efforts to increase investments in lower levels to improve the quality of basic education and make investment in education more progressive; * Pay more attention to children from low income families, given that social programs appear to have helped boost school attendance but have had little effect on reducing repetition and dropout rates; * Develop new financing mechanisms that generate greater internal efficiency in public university spending; and * Facilitate the expansion of private universities through scholarship programs for students from low income families, given that university education enjoys high private rates of return, and private universities are more efficient. Health. Turning to health and nutrition, the analysis suggests that Venezuela has experienced demographic and epidemiological changes, altering the profile of disease. Some health indicators have remained stable, and others have improved over time, but the causes of death have changed-cardiovascular diseases, cancer, and injuries from - viii - accidents and violence have become the leading causes. Moreover, other diseases have emerged, such as malaria and HIV/AIDS. Yet despite longer term improvements, some health indicators have worsened slightly in recent years-for example, infant and maternal mortality rates have fallen in the last few decades, but increased during the first half of the 1990s (before again declining). Although most health indicators in Venezuela compare well with those of other countries in the region, the infant and maternal mortality rates remain higher than in countries with even lower per capita income. While the worsening economic conditions over the last decades may have hurt demand for health services, the decline in government spending in the health sector (particularly during the 1990s) may have also contributed to restricting the quantity and quality of health services. In addition, there is some evidence that the system is not reaching the most vulnerable groups and is highly inefficient. Public facilities and skilled personnel are distributed unequally throughout the country, with a large proportion of the facilities and personnel concentrated in the Federal District (FD) metropolitan area. The geographic distribution of the budget also shows signs of inequity-several States with high poverty receive low levels of per capita budget, as do several States with high geographical dispersion. Personnel expenditures crowd out the purchases of other inputs, leaving little for maintenance, equipment, drugs, and other supplies: in 1997, about 80 percent of the Ministry of Health and Social Development budget was allocated to salaries, about 15 percent to materials, services, and repairs, and less than 1 percent to equipment, construction, and maintenance, and recently these percentages have not changed significantly. The current system for public budgeting and provider payment does not encourage the efficient allocation of resources or internal efficiency in service delivery. Hospital and ambulatory services receive budgets based on prior year expenditures rather than on their performance in terms of volume, quality of services, or impact on health status. While data do not permit a detailed analysis of resource allocation, it can be inferred that many cost-effective health interventions (childhood immunizations, integrated programs of prenatal and delivery care, improved domestic hygiene, and infant feeding practices) have been neglected. With respect to human resources, the health manpower pyramid is distorted. There is a high proportion of specialists relative to general physicians, a high proportion of physicians relative to nurses, and a high proportion of lower level, untrained manpower relative to nurses and other paramedical staff. The lack of sufficient middle- level personnel implies that physicians (who are more costly) perform technician or nursing level tasks. The low level of hospital autonomy is also a constraint. Hospital directors do not have sufficient budgetary and administrative decision-making authority to establish or maintain quality and efficiency. Hospitals are not given the responsibility for managing expenses related to the purchase of equipment, utilities, and maintenance contracts. With few budgetary controls, hospital personnel have no incentives to manage resources - ix - efficiently. Furthermore, hospital directors generally only work part-time because of low salaries, dividing their time between managing the hospital and privately practicing medicine. To overcome key problems in the health sector, improve the provision of health services, and facilitate human capital accumulation while increasing labor productivity, Venezuela must deepen its health and nutrition reforms, particularly regarding financing issues. To improve the allocation and efficiency of health services financing, the Government could: * Include incentives for performance, efficiency, and equity in resource allocation; * Assign priority to high-impact programs, such as those addressing maternal and childhood problems and control of endemic diseases; * Link public sector budgetary transfers to the provision of minimum health services for the poor; * Enlarge the participation of private providers in the health sector; and * Emphasize cost-effective interventions grouped under a basic health service package. To improve equity in the sector, the Government could: * Reduce differences in access to health services between income groups by improving targeting of programs to the poor. To enhance the efficiency and effectiveness of service delivery, the Government could: * Develop institutional capacity for fulfillment of new responsibilities at the central and State levels, particularly in relation to policy-making, strategic planning, resource allocation, and monitoring of both resource use and results. Crime and Violence. The rising levels of crime and violence add another dimension to Venezuela's social development agenda. Analysis indicates that for the 1982-1997 period, there was a significant increase in the level and dispersion of total crime. In 1982, only Caracas, FD and Carabobo exceeded the national crime rate, but by 1997 the national rate had increased and was surpassed by the FD and four States (Carabobo, Aragua, Nueva Esparta and Monagas). Both violent crimes against people (homicide rates) and crimes against property (robbery rates and auto theft) increased significantly. Crime is concentrated in Caracas and in the cities with the largest populations, including Valencia and Maracay. - x - Is the decline in income behind rising crime? The answer appears to be no, as there is no statistically significant relationship between crime (including violent crime), economic growth, and poverty. But crime and violence do have a marked effect on the already poor quality of life of the poor, both in terms of personal costs and on the additional costs to the sectors that are key to alleviating the impact of poverty-namely education and health. The rise in crime and violence is, however, linked to the poor administration of public security and justice, and citizen control policies. IThis "lack of justice", primarily for the poor, needs to be addressed to complement efforts to increase social and human capital. Violence (that disproportionately affects the poor) destroys the basic mechanisms for maintaining and creating social and human capital, thus becoming one of the most important factors of exclusion. There is some evidence that the rise in crime and violence has resulted from higher inequality and social exclusion, particularly for poor young people.4 Almost 60 percent of the victims and perpetrators of crime are between the ages of 15 and 29, live in poor neighborhoods, have a limited education, and have restricted access to the formal labor market (i.e., the average employed person in the formal labor market is about 37 years old). High youth unemployment (i.e., more than 20 percent for the 15-20 years old group), high dropout rates, lack of stable job opportunities, chaotic relations with police, and the weakness of the boundary between the legal and the illegal, substantially influence the decision of these young people to commit crimes. To tackle this problem, the Government could consider creating additional opportunities for this segment of society by enhancing social capital (through improving public services, particularly safety, in poor neighborhoods, and facilitating community-based actions and youth services) and human capital (through better quality and quantity of education, and programs to reach youth at risk and keep them in school). In addition, domestic violence is a serious problem and a frequent occurrence, affecting poor women's capacity to earn an income, as well as having consequences for the children exposed to such violence. To reduce this type of violence, the Government should establish judicial instruments to penalize perpetrators of domestic violence and create special institutions for treating victims of domestic violence. Finally, the lack of data and the weakness of official statistics make it difficult to more precisely establish the causes of violence and crime in Venezuela. Decision- makers urgently need better and more timely data if they are to formulate policies for effectively combating the problem. The direct and indirect costs of violence in Venezuela have been estimated at over 11 percent of GDP in a recent study commissioned by the Inter-American Development Bank. - xi - Chapter I. Introduction 1. Venezuela needs a comprehensive poverty assessment, but the current lack of adequate data would limit the depth and quality of such an assessment. Nevertheless, the Government cannot wait for such a comprehensive assessment before formulating a strategy to reverse the deterioration of living standards and reduce poverty. For this reason, this Report draws on available data to analyze selected economic and social issues that could help the Government make more analytically informed policy decisions of pressing concerns. These issues include (i) better understanding of poverty and inequality and their relationship to income; (ii) improving allocations of social expenditures, and increasing the efficiency and effectiveness of social programs; and (iii) preventing the continued deterioration of social and human capital, while simultaneously promoting its accumulation. In addition to providing the basis for our policy dialogue with the Government on these issues, the Report also identifies areas where more information is required. 2. The second Chapter discusses the performance of the Venezuelan economy over the last decade and the impact on real wages, the structure of the labor market, unemployment, and productivity. Against this background, it reviews the levels and trends of poverty and inequality, drawing on various indicators, including poverty lines, Gini coefficients, key social indicators and human development indexes. It analyzes the relationship between poverty, income growth, and inequality, and contrasts the levels and trends of poverty and inequality in Venezuela with those in other countries of the Latin American and the Caribbean region. To reverse the deterioration in living standards, the Chapter concludes that Venezuela needs to restore and sustain real income growth. This could be achieved through increasing investment, preventing a further deterioration in social capital, and accelerating the accumulation of both human and social capital. To improve poverty and inequality monitoring, the Chapter also concludes that a system of surveys must be instituted. 3. In light of increased poverty and inequality over the last decade, the third Chapter reviews the allocation of resources to the social sectors in Venezuela and the impact of oil shocks on social expenditures. It attempts to establish whether the level and variability of social expenditures have affected social sector outputs, including poverty, inequality, and human capital development. It provides an overview of the distribution of social expenditures and the responsiveness of specific categories of expenditures to fluctuations in total expenditures. It also attempts to determine the effectiveness and efficiency of selected social safety net programs in protecting the poor and preventing the deterioration of human capital, and reviews the institutional framework for formulating and implementing social policies. The analysis concludes that a key factor behind the poor performance of the social sectors is inefficient public resource use, stemming from policy and institutional constraints. 4. While social assistance programs can mitigate poverty in the short-term, and protect living standards against major real income variations, long-term poverty reduction will depend on rapid economic growth combined with substantial investments to accelerate the accumulation of human capital. The last Chapter reviews the evidence on investments and outcomes in education, and in health and nutrition-the key investments for building human capital and increasing labor productivity. Although educational attainment has improved over time, in general, the analysis reveals that investment performance is inefficient and ineffective, and that public investment in education continues to be regressive. As for the health sector, key problems are (i) a large inequity in terms of access to health services and distribution of resources; and (ii) a low level of efficiency and lack of coordination among the institutions providing health services. 5. Finally, the last Chapter explores some of the causes and factors behind the increase in violence and crime in Venezuela, which not only raises the transaction costs of "doing business" but also harms social and human capital development. The analysis indicates that the rise in crime and violence over the last decades has resulted from higher inequality and social exclusion, particularly for poor young people. Thus, there is a need to create additional opportunities for this segment of society by enhancing social capital (better and safer neighborhoods, community actions, and youth services) and enhancing human capital (programs to keep youth in school). - 2 - Chapter II. Income, Poverty and Inequality 6. This Chapter discusses the performance of the Venezuelan economy over the last decade and its impact on real wages and the structure of the labor market. Against this background, the Chapter reviews the levels and trends of poverty and inequality over the same period, drawing on different reported indicators, including poverty lines, Gini coefficients, key social indicators and human development indexes. The Chapter also explores the relationship between poverty, income growth and inequality, and contrasts the levels and trends of poverty and inequality in Venezuela to those in the Latin American and Caribbean region. The Chapter ends with key conclusions and some policy recommendations for reversing the deteriorating trend of living standards in Venezuela and for improving poverty monitoring. I. The Economic Environment 7. The performance of the Venezuelan economy over the last decade has been dismal, and economic policy makers have been unable to cope with the oil cycles and the general decline in real oil prices (and real oil revenues) over the long-terrn.' This has resulted in low economic growth with bursts of inflation and recurrent recessions. Over the last ten years, real GDP per capita growth averaged -1.2 percent per year and the inflation rate averaged about 40 percent per year (with a declining trend over the last five years). For the same period, real non-oil GDP per capita growth averaged about -2 percent per year. At the same time, as the economy failed to diversify, it became more vulnerable to the boom and bust cycles generated by the oil sector (see Graph 1). 8. Following a short period of economic growth during the early 1990s, Venezuela's economic conditions deteriorated considerably during 1993-1994 due to a number of adverse shocks (falling oil export prices, political instability, and a major banking crisis) and an inadequate policy response to these shocks (see Table 1). As a result, non-oil GDP fell, inflation accelerated, and international reserves declined. In response to these conditions, the government embarked on an economic program in earl) 1996. Key policy measures included (i) unification of the exchange rate, initially under a floating system, followed by a system of exchange rate bands; (ii) liberalization of interest rates; (iii) abolishing most price controls; and (iv) adjusting domestic fuel prices.2 By 1997, the economy recovered, but inflation and unemployment remained high. The economy, This Chapter was prepared by Vicente Fretes Cibils and Kinnon Scott. For details on Venezuela's economic performance since 1970 see "Venezuela CEM: Living with Oil" (World Bank, Report No. 12849-VE, June 1995). Also, see "Stylized Facts and the Characteristics of the Labor Supply in Venezuela" (World Bank, Report No. 17901-VE, June 1998); and "Reinventing Venezuela: Towards the Fifth Republic" (institute For European-Latin American Relations, Dossier No. 71, December 1999). Throughout the report, the statistical data for the year 2000 are preliminary. 2 For details on the economic program, see "Agenda Venezuela.: Desemperio de la Economia Venezolana en 1996" (Oficina de Programacion y Analisis Macroecon6mico-OPAM, Ministerio de Hacienda, Report No. 97/1, March 1997, Caracas). - 3 - Graph 1. Economic Cycles, 1979-2000 15.00% Unemp1oment Rate 10.00% / 5.00% - O.O00% - ors t\ '' 1 " 'v' \2 \ / -5.00% - \,' Oil-GDP (tjiwth GDP ' -10.00% O N ON oN N O O 00 00 0c 0 X 0 X 0 X 7 0o CN al 0 0 however, deteriorated in 1998 and tumed into a deep recession in 1999 (see paragraph 10). The deterioration in 1998 was again caused by a sharp decline in international oil prices (i.e., the price of Venezuelan oil declined from an average of US$16.4 per barrel in 1997 to US$10.6 per barrel in 1998). This was compounded by a contagion of external financial crises (that limited Venezuela's access to international markets) and by high domestic real interest rates. General elections in late 1998 added uncertainties that negatively affected investment and growth. Table 1. Macroeconomic Indicators, 1974-2000 Average 1987-89 1990-92 1993 1994 1995 1996 1997 1998 1999 2000 Real GDP Growth 1.0 7.6 0.3 -2.9 3.4 -0.2 6.4 0.2 -6.1 3.2 Non-Oil GDP Growth -0.3 7.4 -1.5 -5.0 2.2 -2.5 4.2 -0.9 -5.4 2.7 Unemployment Rate 8.3 8.6 6.3 8.5 10.2 11.75 11.35 11.3 14.9 14.2 (Average) Investment/GDP 17.4 15.3 18.7 14.2 15.9 16.6 21.0 21.9 18.1 17.5 PIrivate 9.4 5.5 9.4 7.5 6.0 5.5 9.5 12.0 10.6 9.4 Investment/GDP Fiscal Balance/GDP -5.4 -1.7 -3.0 -13.2 -6.9 7.2 1.9 -6.8 0.6 2.0 Non-Oil Budget -17.6 -17.2 -10.7 -23.7 -12.3 -6.7 -8.3 -7.6 -6.0 -10.7 Balance/GDP Current Account/GDP -2.5 4.7 -3.3 4.4 2.6 12.5 3.9 -3.4 3.6 10.8 RealEff.Ratelndex 121 106 115 111 139 117 153 188 211 216 (1990=100)* Lending Interest Rate 20 38 65 60 41 37 22 45 31 26 Inflation 52 33 46 71 57 103 38 30 20 13 * An increase denotes a real appreciation. - 4 - 9. Over the last ten years, the structure of the economy has changed: V'enezuela has become more dependent on the oil sector with the share of oil-GDP in total real GDP increasing from about 21 percent in the early 1990s to about 26 percent in the late 1 990s.3 At the same time, the share of agriculture in total real GDP decreased from about 6 percent in 1990 to about 5 percent in 2000 and manufacturing decreased from about 13 percent to about I I percent.4 The service sector (including government, commerce and other services) remained at about 47 percent of real GDP. Finally, reflecting the failure of exports to diversify away from the oil sector, and the ivapact of the overvaluation of the domestic currency on the external competitiveness over time, the share of non-oil exports to total exports remained, on average, below 25 percent. 10. The new Governrnent, in office since early 1999, inherited an economy heavily dependent on the oil sector with major structural imbalances, including a growing fiscal deficit, a deficit in the external current account, and high unemployment. The scale of the crisis made it necessary to act quickly, and the Government adopted a set of economic policies, including the decision to (i) reduce the rate of public expenditures (by limiting public sector wage increases to an average of 17.5 percent and by cuttinig investment outlays of the government and public enterprises, including Petr6leos de Venezuela S.A. (PDVSA); (ii) increase public sector revenues (by introducing a tax on bank debits and financial transactions for one year and by reducing the sales threshold of firms subject to VAT); (iii) cut back oil production-in agreement with other oil producers to increase the oil price in the international market; and (iv) maintain the stance of exchange rate policy in the context of the exchange rate band system.5 11. More importantly, however, the administration gave priority to political reforms, including the introduction of a new Constitution. This, and the lack of a convincing and comprehensive economic program, increased uncertainty, which led to a process of "dis- investment"-real private investment declined by 18 percent in 1999 when compared with 1998 (and total investment as a percentage of GDP was 18 percent in 1999 compared to 22 percent in 1998). This deepened the economic recession (the worst since 1989) and increased unemployment to more than 16 percent by end-1999. At the same time, however, the Government's policy (together with other OPEC and non-OPEC oil In 2000, oil revenues represented about 56 percent of public revenues and nearly 84 percent of goods exports. However, the role of oil is probably understated in these statistics, since there are indirect links through the use of oil as an input in several industries. Moreover, the macroeconomic stability of the economy is closely related to the price of oil. A one dollar reduction in the price of a barrel of Venezuelan basket oil leads to about US$0.8-1.0 billion reduction in public sector revenues (equivalent to about 0.8-1.0 percent of GDP at the current exchange rate). Since this structure may reflect Venezuela's comparative advantage, there is no a priori reason for agriculture and manufacturing to be large. However, these are more labor-intensive sectors than the oil sector, and therefore have larger labor employment generating impacts. In June 1999, the Government announced its "Economic Transition Program 1999-2000". The last Chapter outlines the main tenets of economic policy, which includes (i) economic stabilization policy; (ii) sectoral policies; and (iii) employment policy. While the program includes specific goals, it has been criticized by being "too" general on how these goals would be achieved and by not being widely distributed and communicated to the public. - 5- producing countries) has succeeded in increasing oil prices (at least temporarily), which has helped improve the short-term stance of key macroeconomic indicators, including the overall fiscal deficit and net international reserves. 12. In December 1999, following several days of heavy rains, floods and mudslides, thousands of Venezuelans were killed (estimated up to 30,000) or left homeless.6 The poor, living in precarious conditions on the sides of hills and mountains, were particularly hurt by the magnitude of this natural disaster. Infrastructure and basic services were substantially damaged in the affected States (mainly Vargas, Miranda and Falc6n). It has been estimated that the total cost of the natural disaster amounts to about US$3.2 billion, with the direct cost of the lost infrastructure amounting to about US$1.9 billion. While the reconstruction of damaged infrastructure will take a few years, there will not likely be significant negative effects on total output. 13. Against this background, during February-April 2000, the Government announced the Economic Plan for 2000 (and beyond), which included, among other variables, specific goals for economic growth, unemployment, inflation, fiscal deficit and net international reserves. To achieve these targets, the Government also announced a series of policy actions that include (i) a reduction of the VAT; (ii) elimination of the tax on financial transactions; (iii) fiscal incentives (tax credits) for activities in agriculture, tourism and manufacturing; and (iv) promotion of exports through the establishment of lines of credits, and revitalization of industrial parks.7 14. While the economic plan includes important initiatives, the main challenges for the Government remain to sustain stability and growth, without major social disruptions, to diversify the economic structure over the longer term, and to improve the living standards of the population. To restore and sustain stability over the long termn, the non- oil fiscal deficit must be reduced through the adoption of measures to enhance non-oil revenues and control expenditures. Moreover, the strong links between oil price volatility and public sector expenditure must be reduced (this could be achieved by imposing an inter-temporal budget constraint through the full implementation of the oil stabilization fund). To complement these policies, the external position should be strengthened through a combination of exchange and monetary policies. With regard to the exchange rate, to ensure external competitiveness of non-oil exports, a more flexible exchange rate regime could be adopted to adjust prices to significant and protracted changes in the terms of trade, thereby avoiding the full adjustment through real wages. In turn, a restricted monetary policy of high interest rates cannot be used as a substitute for 6 For details on the causes, impact and economic evaluation of the natural disaster, see "Assessment Report" (ECLAC, February 2000, Caracas). For more details see Annex A: Summary of Economic Plans and Objectives for 2000. Also, in early 2000, the Government announced detailed programns for the economy and different sectors, including Agriculture, Tourism, Industry and Trade, Aluminum, Petroleum, Education, Health and Mining. -6 - non-oil fiscal adjustment and a mis-aligned real exchange rate, as high real interest rates will delay economic recovery and growth.8 Real Wages and the Labor Market 15. In the environment created by the low and unstable real income growth, with high dependency on the oil sector, how did the return to labor services (a major "asset" of the poor) and the structure of employment change? Real wages in Venezuela were "responsive" during the period under analysis. In fact, between 1990 and 1999 average real wages fell by about 23 percent, discounting the hypothesis of real wage rigidity, despite the existence of mandatory minimum wages for most of the last decade (see Graph 2). The fall affected all sectors of the labor market, including the oil sector. Nevertheless, the fall in wages was insufficient to avoid increases in unemployment, Graph 2. Productivity and Real Wages, particularly affecting those 1979-1999 individuals that (i) had 110 ------------------------------------ relatively little human capital 105 ------------------------------------ (i.e., only primary education); 100 -----------verageLaborProductivity (ii) were young (and, hence, o 95 - -------- ---------- inexperienced) and female; go and (iii) were urban residents - 8 -- ------- and/or recent migrants.9 In w 80 ---------- light of the extent of the X 75 ---------- external and internal shocks, 70 ---------- the variations in 65 ----------------------_______--- unemployment were smaller 60 than otherwise could have e * 9 * _ co _ 4 m 40 been, somewhat limiting the Year negative consequences on employment of economic instability. 10 16. As a result of the fall in real wages, as expected, labor prodluctivity also decreased. Over the last ten years, labor productivity has declined at an average annual rate of about 1 percent. What factors could have contributed to the decline in productivity? In addition to low real economic growth, factors related to the role of the For details on some policy suggestions, see "Venezuela: Selected Policy Notes on Economic and Social Development" (World Bank, Report No. 19040-VE, March 1999). For example, the average unemployment rate for females with little education undes the age of 22 was in excess of 25 percent for 1997-1999 period. 10 Based on estimates of the structural cyclical relationships between wages, employment, unemployment and output, the Venezuelan labor market adjusted to (adverse) output shocks relatively more through real wages than employment/unemployment adjustments as a consequence of high instability in prices. For more details, see "Stylized Facts and the Characteristics of the Labor Supply in Venezuela" (World Bank, Report No. 17901-VE, June 1998). - 7 - informal sector as a buffer 1 offer an explanation." Hrlpe 3. InfmaloSet Employment in the informal HelpedBufferUnemployment sector has reduced potential 65% 17% unemployment in the formal 60% 15% sector, and the counter- S Informality Rate cyclical pattern of changes in ,55% 13% the informal sector's share of Y o%/ total employment has allowed I\ 11% E recessions and economic 0 51 contractions to have a lesser 2 } 9% - impact on unemployment (see 7% Graph 3) 12 UY ( 35% - nemployment 30%' 5% 17. Consistent with the c O N OD O 0) a e> O changes in the structure of the economy indicated above, the structure of employment over the last ten years has also changed. More specifically, the share of employment in the tradable sectors has declined, mainly as a consequence of the shrinking participation of agriculture and manufacturing in total employment.'3 The share of employment in non- tradable sectors has increased, resulting mainly from informal activities. 18. In this environment of low and unstable real economic growth with falling real wages, increasing unemployment, and decreasing labor factor productivity, a major In addition to economic factors, demographic factors have influenced the performance of the Venezuelan labor market. Over the last ten years, the total population grew at an average rate of 2.2 percent annually, while the working-age population grew at 2.8 percent. For the same period, the economically active population or labor force grew at a rate of about 4 percent annually and employment grew at about 3 percent. The increases in female and youth participation rates are responsible for the faster growth of the labor force. The more pronounced growth in female participation rates occurred for women between 25 and 44 years of age (from about 40 percent 1985 to about 50 percent in 1997) and for those who are better educated (with high school or college education). This indicates both a long-term trend marked by effects associated with more opportunities for higher educated women, as well as short-term effects associated with pressure on the female members of the family to supplement household income. In addition, the highest youth participation rates occur in the informal market as youths between the ages of 15 and 19 with only primary education or less schooling joined the labor force to supplement household income, but could find employment mainly in the informal sector. 12 The counter-cyclical argument of the pattern of changes in the formal sector participation versus the informal sector participation has been supported by a simple regression analysis, that finds a highly (statistically) significant negative correlation between the rate of participation in the formal and informal sectors. 13 A full demand and supply analysis is required to determine the causes of the changes in the structure and its implications for labor demand. While we can conclude, in general, that real wages are flexible in the labor market, without demand analysis we can not determine the reallocation of labor from declining to expanding sectors of the economy. This would require relative wage flexibility across sectors to produce real wage differentials needed to induce labor movement among sectors. - 8 - question emerges: how much have the living standards of the Venezuelan people deteriorated over the last ten years or so? II. Poverty and Inequality-Levels and Trends'4 19. In spite of considerable effort put into measuring poverty in the past decades, there is little consensus on the actual poverty levels in Venezuela. The problems have been (i) the data sources (see Box 1); and (ii) the variety of definitions of poverty employed (among them, a simple basic needs approach, a more refined basic needs approach, a capabilities approach and the poverty line method). Given this dwiversity, it is not surprising that figures on poverty that emerge from different studies are dissimilar. The unfortunate result of these analyses has however been a debate over data and methods that has often obscured the debate over how to alleviate poverty. Despite the limitations of national poverty statistics, which prevent a complete understanding of poverty, the cumulative effects of the economic deterioration on Venezuela's poor have been serious, as reported by official poverty statistics using two different methodologies, "Poverty Line" (LP) and Basic Unsatisfied Needs (NBI) (see Box 2 and Table 2). 20. Systematic measurements of poverty over time have been carried out using the poverty line method with data from the biennial labor force survey (Encuesta de Hogares por Muestreo, LFS). With this data source (and recognizing its limitations), the different methodologies have led to different results. However, as can be seen in Table 2 and Graph 4, there is agreement that poverty has increased over the past years, irncluding the last decade (although there is no consensus on the actual levels of poverty nor on the depth of poverty). 15 21. Even if the methods selected to estimate poverty rates in Venezuela are problematic, there are important findings for analytical and policy considerations. Changes in the poverty line not only give different levels of poverty but provide substantively different findings concerning poverty during the period. For example, poverty is unambiguously higher in the late 1990s than in the early 1980s. But, for the period 1995-1997, there is no statistically robust difference in poverty levels. 14 Given the limitations of data for poverty analysis, this Section uses different sources of data including data published by the Central Bank of Venezuela, OCEI, IDB, IMF, UNDP and The World Bank. It also uses data included in published working papers and studies of several authors. 15 The variations in results could be much more striking than shown in Table 2. Estimated poverty measures using the LFS for the period 1980 to 1999 indicate that methodological choices are responsible for the different poverty estimates available. These methodological choices may, for example, include (i) adjustment for under-reporting of household income; (ii) economies of scale for using resources within households; and (iii) alternatives income/poverty measurements. (For more details, for example, see "Falling Apart: Temporal Poverty and Household Income Inequality Patterns in Venezuela" (Ruprah, 1. and Marcano, L., draft, January 2000). 9- Box 1. Tbe Data for Poverty Analysis-Limitations and Recommendations One of the problems carrying out poverty analysis in Venezuela is the lack of appropiate data. The Encuesta de Hogares por Muestreo (LFS) is designed for the study of labor force variables and not poverty or welfare studies. It is used for poverty analysis simply because it is the only available data set that has been collected more or less consistently over time (from 1979 to the present). In addition to significant changes in the way the survey has been administered, important changes have taken place, including (i) the number of households/persons surveyed has fallen significantly (from about 130,000/730,000 households/persons in 1979-11 to about 16,000/80,000 households/persons in 1997-II); (ii) the stratification clustering prior to mid-1994 was based on the 1980 population census and, after mid-1994, on the 1990 census; and (iii) the set of questions has changed, particularly regarding income (e.g., from 1979-1993, "income" was labor income, and from 1994 and onwards, "income" included labor and non-labor income). These data sets have limitations for welfare analyses that need to be understood. Measuring Welfare-Income vs. Consumption. In theory, income and consumption would be equivalent over a person's lifetime (barring inheritances). In the short run, however, the two are not the same: income is a measure of potential well-being; consumption is actual or achieved well-being. A case can be made for using either one of these concepts for measuring welfare, but the main reason for preferring consumption over income is related to a practical problem: income information is more difficult to collect. The difficulties include: First, at the higher end of the distribution, incomes tend to be underestimated, as richer respondents are less willing to share income data. In the LFS, the households not providing income data tend to be more highly educated, urban, have higher levels of home ownership and be slightly older than the population in general. The underestimation of income is serious, and adjusting to national accounts provides a poor approximation, because it assunes that (i) national accounts are a "gold standard"; and (ii) all households underestimate by the same percentage or at least underestimate in the same manner by type of income. Furthermore, there is the added difficulty that, until 1994 the LFS only collected labor income, and thus total income was underestimated. Second, at the lower end of the distribution households often engage in self-employment activities that are closely interlinked with other household activities, and so it is difficult to determine a net income figure. In sum, there are problems at both ends of the income distribution-the underestimation by the richest groups leads to underestimation of inequality (although it should not affect poverty measurement), but the inaccuracy of the lower end leads to an overestimation of poverty and inequality. Third, if the LFS data were used for welfare analysis, the poverty line must then be determined exogenously. For a line based on food consumption, this means that the line is not necessarily based on true consumption but it is a theoretical line designed to minimize cost. This assumes that the resulting diet is palatable, that the goods are available and the all households face the same prices for all the goods (assumptions which are not necessarily supported by empirical evidence). The above argues for using consumption as a measure of welfare and not income. First, consumption would allow a food basket based on actual consumption patterns, "revealing" consumption preferences. Second it would avoid the serious underestimation problem-a comparison of the cumulative density functions of total income and total consumption in Venezuela showed that consumption dominated income, i.e., consumption is higher than income. Third, avoiding the underestimation problems removes the need to carry out potentially distorting adjustments for under-reporting. Finally, a consumption-based measure would solve the difficulties in determining welfare of primarily agricultural households and/or those involved in informal sector activities. Recommendations. To improve the data for poverty analysis, key recommendations include: First, determine how best to gather data at the household level for a consumption-based measure of welfare. This would decrease the present high levels of missing data and avoid the biases caused by richer households not being represented in the data set. Second, link consumption levels to other facets of welfare, including levels of health and education, access to social services and infrastructure, and the incidence of government spending. Third, institute a survey or system of surveys that will provide consistent data on such issues over time, including some elements of a panel design-i.e., to track the same households over time to address issues such as which households move in (or out) of poverty, coping mechanisms and the like. Fourth, include the entire rural and urban populations in the samples. (Even though the population is highly urban, there is evidence that poverty is extremely high in rural areas, and omitting these groups leads to an underestimation of both poverty and inequality). - 10- Box 2. Measurement of Poverty Official poverty statistics are reported using two different methodologies, the Linea de Pobreza ("Poverty Line"-LP) and the Necesidades Basicas Insatisfechas ("Unsatisfied Basic Needs"-NBI). The LP, calculated and published by OCEI, is based on the standard CEPAL methodology, in which the cost of a market basket (canasta normativa) of basic necessities is calculated, with adjustments made periodically for estimated changes in household income and expenses. The proportion of the population living in poverty or extreme poverty is then defined by those receiving less than the respective threshold values. The NBI method uses household data (based on the 1990 Census) on five social indicators to create a "poverty map" which defines tlhe proportion of the population at the State and national levels living in conditions of "poverty" (defined as households characterized by one of the criteria) or "extreme poverty" (households characterized by two or more criteria). The five criteria are: physical housing characteristics; housing density; level of basic services (running water and sanitation); children 1-12 years old who are (not) attending school; and the number of household members unemployed. Official statistics are included in Table 2 in this Chapter. At the State level (not shown), poverty estimates vary widely. OCEI estimates of the proportion of the population living in extreme poverty in 1997 varied from 9.8 percent in Caracas to 44.9 percent in Delta Amacuro and Apure, respectively. In addition to the general limitations regarding methods and data, the specific limitations of eztch of these methods of measuring poverty are well-known and discussed elsewhere, but are highlighted here. B3riefly, variations in the LP measure are based on variations in prices of basic goods and incomes, both of which are difficult to accurately measure, even in the best of cases, in a highly inflationary environment. However, by only infrequently incorporating changes in the composition of the canasta normativa and its underlying weights, the LP measure seriously understates the ability of consumers to substitute -both within the canasta and items outside it-cheaper sources of food and nutrition for more expensive ones, leading to an overestimation of poverty incidence. Other serious problems include the limited geographic coverage of the underlying survey used to collect primary household data, significant sampling problems that have been reported with this survey, and the omission of transfers, in-kind income, and other income sources in estimating income. The NBI method also has some serious limnitations, resting as it does on outdated social indicators that were collected a decade ago in the 1990 census. The periodic updates of poverty measures using the NBI method apply periodically revised population projections within the country to the original composition and incidence of social indicators. The resulting poverty indicators thus basically measure changes in population numbers and composition, rather than the changing incidence of the underlying social indictors which actually measure poverty. That the resulting estimates of poverty rates are so stable in the face of significant economic changes is thus not surprising. Table 2. Poverty Over Time Using Income and Poverty Line Methods Period Poverty Headcount Poverty Gap Marquez 1DB Ruprah UCAB Official Marquez IDB Ru.prah Official & & & & Alvarez Marcano Alvarez Marcano 1980-1984 27 23 9 34 31 12 14 4 11 1985-1989 39 36 18 54 47 14 20 6 19 1992-1996 45 49 20 66 53 16 25 5 22 1997-1999 -- -- 24 67 52 _- - 9 24 Usine NBI Method Period Official UCAB 1995-97 50 1998-99 60 49 Source: Marquez aid Alvarez, 1996,IDB, 1999; UCAB, 1999; Ruprah and Marcano, 2000. The last study uses the official poverty line and per capita income adjusted to national accounts by income source Note: Period indicators are eitlher averages of yearly indicators or a figure represenitative of the perod, depending on data availability - 11 - Graph 4: Powrty in Venezuela, 1990-1998 40 35 30 - o25 Et 20 10 ON _ Cs 'C N CC ON O ON ON O C ON ON O +*---Offic: flat -U--PPPP: Flat Offic: Y source X PPP: Y source Note: The data is per capita income taken from the Labor Force Survey. "Offic." stands for the official poverty line of the government, while "PPP" stands for the PPP $2 per day poverty line. "Flat" refers to an adjustment made for underreporting of income wherein a coefficient of adjustment is applied to all incomes, the coefficient is the ratio of labor income in the national accounts to wage income in the household survey. "Y source" is a different adjustnent for underreporting of income wherein separate coefficients are calculated for different income sources. This uses a coefficient applied to wage income- wage income in national accounts to wage income in the survey-and self-employment or employer income-- labor non-wage income in national accounts and labor non-wage income in household survey (see Ruprah and Marcano, 2000). Source: Inter-American Development Bank, OCEI, and World Bank. Another View of Poverty-Social Indicators and Human Development Index 22. The above discussion has mainly focussed on monetary measures of welfare, which allow us to identify the characteristics of the households (and their members) that are more represented in the poor and non-poor households (see Box 3). More importantly, however, as poverty is a multidimensional phenomenon, there are many other facets of peoples' lives that affect their welfare, including education, health and nutrition, housing, crime and violence, and absence of legal protection. One common alternate measure of poverty is, therefore, to examine social indicators. Such social indicators are often seen as measuring a person's ability to meet its basic needs, and are often combined into a composite index of basic needs. - 12 - Box 3. Some Characteristics of the Poor Poverty rates are problematic to estimate. To overcome these problems, some characteristics of the households that belong to the lowest 40 percent of the income distribution are described below. Salient characteristics of the households that are most likely to belong to this category are: headed by women, uneducated, unemployed and inactive, employed in the informal sector, and living in the rural areas and particularly in the States of Amazonas, Apure, Delta Amacuro, Guarico, Portuguesa, Sucre. Trujillo and Yaracuy. Gender. Households headed by women are over represented within this category. Age. Households headed by individuals older than 50 years are over represented in this category. In contrast, households with heads in the 30-50 year-old range are less represented-this is expected because 30-50 years of age approximates the prime income-generating years. Children. Households in the lowest income deciles have, on average, twice as many children as households in the middle income deciles. Education. Schooling is positively and significantly associated with income. The lower income groups have lower education, as measured by number of school years attended by the head of the household. Employment status and sector. Among those employed, those employed in the private informal sector are highly represented in the lowest income deciles. Households in which the head is not in the labor force or is unemployed are also highly represented in the lowest income deciles. In 1997, in middle income households, there were two working adults per child and two dependents. This compares with less than one working adult for every two children and four dependents in the lowvest income distribution. Family size. Households with numerous family members follow a U-shape distribution. Specifically, households with family members of 5 or more are highly represented in the lowest income cteciles. This reflects the fact that larger families have higher dependency ratios with more children and more elderly members per prime-age family member. This is also observed for affluent families. Hovwever, single person families are more represented in households in the lowest incomes deciles, which suggests that limited income generating possibilities may be a major reason why adults, not only live in small households, but live alone. Location. Poor households have systematically and substantially higher representation, in relative terms, in rural areas than in urban areas. Source: Encuesta de Hogares por Muestreo, OCEI. 23. Key social indicators for Venezuela are listed in Table 3, and indicate mixed progress over the period 1985-1999. On one hand, life expectancy and adult literacy have increased but, on the other, per capita protein intake has decreased. However, during the 1990s, there has been a significant deterioration in social indicators. Despite the mixed results and the deterioration during the 1990s, Venezuela's social indicators are still ranked relatively well compared to the average for all Latin American countries. The Human Development Index (HDI) estimated by OCEI and UNDP for the period 1990-1999 provides a composite index for welfare measurement in Venezuela (see Table 4).16 According to this index, living standards in Venezuela deteriorated 16 For more details on the Human Development Index in Venezuela, see "Indice y Entorno del Desarrollo Humano en Venezuela" (OCEI and UNDP, 1999). - 13 - significantly during the 1990s. This deterioration, however, resulted from the income component of the index, as the stagnant or slight improvements in life expectancy and education could not outweigh the large deterioration in real income. Table 3. Key Social Indicators Units 1985 1997 1999 LAC** Demography Population Thousands 17077 22800 23738 501700 Population Growth Annual % 2.4 2.1 2.0 1.8 Urban Population % of Total Popn. 82 86 87 74 Life Expectancy at Birth Years 69 73 73 70 Total Labor Force Thousands 6170 9111 9483 212000 Education, Health & Nutrition Adult Illiteracy Rate Ratio 13 8 8 12 Population per Hospital Bed Persons 3300 667 -- 455 Per Capita Protein Intake g/day 66 58 -- -- Basic Needs Access to Safe Water: Urban % of Urban Popn. 88 79 79 -- Access to Safe Water: Rural % of Rural Popn. 65 79 79 -- Access to Safe Water % of Population 84 79 79 72 Crime and Violence Reported Crime Incidents per 100,000 inhab. 819* 1040 -- -- Notes: *=Figure is for 1982; **=Figure is for 1998 or latest year available Table 4. Human Development Index Year Lfe lfe Expectancy literacy IEnrollent Achievement Purchasing Income 11) Expectancy Indicator |_Rate Indicator Power Indicator 1990 71.2 0.77 90.9 0.73 5.2 0.96 0.82 1993 71.2 0.77 90.9 60.9 0.81 4.9 0.91 0.83 1996 72.4 0.79 90.9 60.4 0.81 2.7 0.48 0.69 1999 72.4 0.79 90.9 61.2 0.81 2.0 0.35 0.65 Source OCELUNDP, Human Development Index for Venezuela, 1999 Income Inequality 24. Gini coefficients are used to measure inequality but the income levels are adjusted for under-reporting and for economies of scale within households. As in the case of poverty, there is a variation in inequality indicators (Graph 5). As expected, the adjustment by income source increases the level of inequality, as does taking into account household economies of scale (showing that larger households are poorer). - 14 - Graph 5. hicome Inequality 0.60 0.55- 0.50 - 00.45- 0.40 0.35 _~ o~ _~ o~ oa ol oa _, oa - PC: Flat - PC: Y source theta=0.5: Flat - theta=0.5:Y source Note: "PC" stands for per capita income, "O.5" refers to using an economies of scale indicator with a theta value of 0.5. "Flat" refers to an adjustment made for underreporting of income wherein a coefficient of adjustment is applied to all incomes, the coefficient is the ratio of labor income in the national accounts to wage income in the household survey. "Y source" is a different adjustment for underreporting of income wherein separate coefficients are calculated for different income sources. This uses a coefficient applied to wage income-wage income in national accounts to wage income in the survey-and self-employment or employer income-labor non-wage income in national accounts and labor non-wage income in household survey (see Ruprah and Marcano, 2000). Source: Inter-American Development Bank, OCEI, and World Bank. 25. The conclusions about inequality are, nevertheless, somewhat stronger than those about poverty. The basic finding is that inequality increased between the early 1980s and late 1990s, and this finding is statistically robust, despite differences in computational methods. The actual level, however, is still sensitive to assumptions. While, there are no clear patterns of changes in inequality during the 1980s and early 1990s, the increase in income inequality is statistically robust during the mid and late 1990s, regardless of indicator or methodology used. The Impact of Changes in Real Income Growth and Inequality on Poverty 26. To determine the impact of real income and inequality on poverty in Venezuela, changes in the poverty rate are estimated as a function of changes in real per capita income and changes in the Gini coefficients (for the 1979-1999 period and for the un- adjusted and adjusted set of corresponding variables).'7 The estimated parameters (i.e., the elasticities) indicate that changes in real income are negatively con-elated with changes in the rate of poverty and, by contrast, changes in inequality are positively correlated with changes in the rate of poverty. 8 The average elasticities and their 1 7 The estimated equation regresses the change in the log of the poverty rate (dlnPR) on the change in the log of real per capita income (dlnPCY) and the change in the log of the Gini coefficient (dlnGINI); or dInPR = a + b dlnPCY + c dInGINI + e. The estimated parameters for the adjusted variables are: b=-1.23 (t4.09), and c=0.67 (t=1.36). 18 Indeed, the impact of real income growth on poverty in Venezuela is consistent with the general findings in "Growth Is Good for the Poor" (Dollar, D. and Kraay, K., World Bank, March 2000). - 15 - implications for changes in poverty Table 5. Elasticities and Simulations rates are presented in Table 5. For ChangeinPovertylkademaut example, a 10 percent growth in real per capita income, ceteris i Total 2ag i tc paribus, would reduce the poverty i 10/ 20% 30% rate by about 12 percent on average. °% -12.3% -24.6% -36.90/o While the simulations are 5% - 8.90/o -21.20/o -33.5% illustrative of the importance of 10% - 5.6% -17.90/o -30.20/o both real income growth and its Ne: Elasticityofpoty wrttrealincxris-1.23(t-stat-4.088,df=18); distribution on poverty, dynamic elasticityofpovertywr.tLiniooeff isO.670(t-stat'-1.359,df=18);RL0.49 elasticities may have important asymmetric effects on poverty- Estimated Poverty Headcomt i.e., they tend to be higher in the %Change Total ChanginPerCapitahIome downturn than in the recovery of inDGbd 100/% 20% 30% the business cycle.19 This would 0% 61.4% 52.8% 44.20/o imply that poverty is likely to be 5 63.7/o 55.1% 46.5% reduced faster if, for the same 10%___ 66.1% 57.5% 48.90/o average real income growth, the Note: Baseyew 199 Off variance of real income growth is smaller. How Poverty in Venezuela Compares to the Latin-America and Caribbean Region 27. In the latter half of the 1980s, Venezuela experienced significantly lower poverty than did Latin America as a whole (see Graph 6). 0 By the second half of the 1990s, however, the levels in Venezuela moved much closer to those of the region. Income inequality in Venezuela, as measured by the Gini coefficient, does not appear to have followed the regional patterns. The two levels were similar only in the mid 1980s (see Graph 7). Since then, inequality in Venezuela has declined while the regional indicator increased; it then increased in Venezuela while the regional level stayed relatively flat. Thus, inequality in Venezuela is likely to be still below that of the region (although the fall in inequality in the late 1980s was more than compensated for the increase during the mid 1990s). In this paper, the authors find that income of the poor increases one-for-one with the overall growth of the economy, and this general relationship between income of the bottom fifth of the population and per capita GDP holds in a sample of 80 countries covering four decades. The paper also shows that (i) policy-induced growth is as good for the poor as it is for the overall economy; (ii) openness to foreign trade, good rule of law and fiscal discipline benefit the poor to the same extent that they benefit the whole economy; and (iii) in contrast, high inflation is more harmful to the income of poor than to overall GDP. In contrast, the authors do not find evidence that formal democratic institutions or public spending on health and education have systematic effects on incomes of the poor. 19 See "How Effective Has Aggregate Income Growth Been in Reducing Poverty and Inequality in Latin America" (de Janvry, A. and Sadoulet, E., November 1999). 20 See "Poverty and Policy in Latin America and the Caribbean" (Wodon, Q., World Bank, March 2000). - 16- Graph 6. Poverty in Venezuela and the IAC Region 40 c35 30- 20 E25- % 15 10 cc ON C o r N ON oo oN O O O -+-LAC REGION _Ven: Offic: flat Ven: PPP: Flat Ven: Offic: Y source Ven: PPP: Y source Note: The LAC REGION line is based on data representing 71 percent of the population in the region although it includes only 12 countries. In Graph 6, the four lower lines are data from Venezuela and are the same as in Graph 4 in this Chapter. Source: Inter-American Development Bank, OCEI, and World Bank. Graph 7. Inequality in Venezuela and the LAC Region 0.62- 0.57 0.52- 0.47 042 00 00 _O ON m O - LAC REGION - Ven: PC, Flat A Ven: PC, Y Source Note: The LAC REGION line is based on data representing 71 percent of the population in the region although it includes only 12 countries. Source: Inter- American Development Bank, OCEI, and World Bank. - 17 - III. Conclusions and Recommendations 28. In the environment provided by the low and unstable real income growth, with high dependency on the oil sector and inconsistent economic policies, real wages and employment opportunities have decreased significantly over the last decade. The deterioration in real per capita income, employment and labor productivity-despite the problems and limitations with the data and the methodologies-indicate that poverty in Venezuela has increased. Indeed, while there is no consensus on the level of poverty, poverty is unambiguously higher in the late 1990s than in the early 1980s. Moreover, inequality increased between the early 1980s and late 1990s. This finding is statistically robust, despite changes in computational methods. Poverty and inequality have, however, remained below the average for the LAC region throughout the 1990s. Other composite welfare indexes and social indicators also show that the living standards in Venezuela deteriorated significantly during the 1990s. 29. Given the relationship between poverty, real income and inequality found in the analysis, to reverse living standards deterioration, Venezuela needs to sustain real income growth and to reduce inequality through accelerating investment (particularly private sector investment) and to improve labor factor productivity. Labor factor productivity could be improved by preventing human (and social) capital deterioration and, simultaneously, accelerating its accumulation. The following chapters provide a more detailed analysis on these issues and present some suggestions on how to improve human capital development. To measure and monitor poverty and inequality, a survey (e.g. a LSMS) or system of surveys must also be instituted. In addition to providing information for policy formulation, the surveys should (i) facilitate the construction of a consumption-based measure of welfare, linking consumption levels to other facets of welfare-such as levels of health and education, access to social services and infrastructure, and the incidence of government spending; (ii) include elements of a panel design to track the same households over time; and (iii) cover the entire population (i.e., include both urban and rural areas). - 18- tChapter III. Social Expenditure-Shocks, Policy Response, and The Institutional Frameworkl 30. Against the background of increased poverty and inequality over the last decade, this Chapter analyzes the relationship between the allocation of resources to the social sectors in Venezuela and the impact of oil shocks on social expenditures. It attempts to establish whether or not the level and variability of social expenditures have affected social sector outputs, including human capital development. This Chapter provides an overview of the distribution of social expenditures and the responsiveness of specific categories of expenditures to fluctuations in total expenditures, and attempts to determine the effectiveness and efficiency of selected social safety net programs in protecting the poor. Moreover, the Chapter reviews key elements of the institutional framework for formulating and implementing social policies. The analysis concludes that the poor performance of the social sectors in Venezuela can partly be attributed to inefficient public resource use due to policy and institutional constraints. I. Trends in Public Sector Expenditures Total Public Expenditures2 31. Total real government expenditures fluctuated significantly over the last three decades (see Graph 8). 32. With the OPEC oil price hike of Graph 8. Real Total Expenditures Had 1973, total government expenditures Big Swings During (1972-2000) increased by more than 150 percent in real terms. Afterwards, total 130 expenditures experienced marginally a downward trend, fluctuating with the oil 110 - cycle. Throughout the early 1990s, total public expenditures rose, fueled by fi economic growth (average real GDP = 70- growth of 7.4 percent from 1990-92) along with high real oil revenues (resulting from the price and quantity 30 effects of the Gulf war). Since 1992, C C C 0 co 4 Venezuela has experienced a number of recessionary years (1993-A, 1996, and This Chapter was prepared by Vicente Fretes Cibils with collaboration from Mario Adolfo Cuevas. It draws on a background paper prepared by Danielle Anne Dukowicz supported by PRMPO. 2 For the analysis, total public sector expenditures are total government expenditures and exclude expenditures in the oil sector (carried out by PDVSA) and net assistance to banks. Also expenditures in this and following sections are based on real expenditures in 1984 prices, and the price of oil is taken to be the principal source of fluctuations in oil revenues. - 19 - 1998-9) resulting from internal and external shocks closely related to the oil cycle;3 total government expenditures fell in absolute terms. Public Sector Social Expenditures 33. Real social sector expenditures increased since the early 1970s (Graph 9).4 Generally, the social sectors seem to have been relatively "protected" in the face of declining real oil prices, Graph 9. Real Social Sector Expenditures and during the last two decades, Increased, but With Swings (1972-2000) social sector expenditures have generally exceeded 1970s 175 expenditure levels. Moreover, social 0 150 expenditures reached a local peak in 1992 when the Government allocated Z 125 - about 40 percent of total expenditures .3 loo / (about 9.7 percent of GDP) to social s 'a= 75- sectors. Subsequently, social sector expenditures fell in absolute real ~ 50 terms partly as a result of the 5_2s slowdown in the Venezuelan _ ,., .,,,. . economy, although expenditures 0h O ON o as O ON oNO remained at relatively high levels. In the last three years social expenditures increased considerably in real terms. In 2000, the Government allocated 11 percent of GDP to social expenditures, exceeding the 1992 allocation by over 1 percent of GDP. 34. This suggests that the level and the trend of social sector expenditures in real terns cannot, by itself, explain the deterioration in the output performance of the social sectors in Venezuela. The reasons for deteriorating performance must therefore be found elsewhere. Could the volatility of expenditures account for low output performance? The next Section explores this possibility, with reference to volatility originating in the oil sector. II. Public Expenditure Responsiveness to Oil Price Shocks 35. The expenditure (total and social sector) and oil price time series in real terms are decomposed into their trend and cyclical components using the Hodrick-Prescott filter technique.5 First, two regressions have been estimated:6 3 For more details, see Chapter II, Section I. 4 Social sectors include education, health, housing, social development, and social security. 5 See "Postwar U.S. Business Cycles: An Empirical Investigation" (Hodrick, R. and Prescott, E.; Journal of Money, Credit, and Banking; Pages 1-16; February 1997). - 20 - TTEt = aO + 3o TPricet + uo TSEt = al + 0i TPricet + ul where TTEt is the difference of the Table 6. Real Total Expenditures Vary Positively logarithm of the government with Real Oil Prices... but Not Real Social expenditures trend component, TSEt is Expenditures the difference of the logarithm of the Dependent Variables social expenditures trend component, Real Expenditures: Total Social and TPricet is the difference of the A-Trend Component logarithm of the price of the Explanatory Variable Venezuelan oil basket trend Real Price of Oil 0.57 0.49 component; uo and ul are the error t-statistic (17.1) (16.4) terms. The regressions were run using Observations 28 28 annual data covering the period from Adjusted 2 0.91 0.90 1970 to 1998, with the results B-Cyclical Component presented in Table 6.A. The estimated Explanatory Variable Ps in these specifications can be Real Price of Oil 0.50 0.21 interpreted as approximate measures of t-statistic (3.7) (1.2) the oil price-elasticity of total and Observations 28 28 social expenditures at the zero Adjusted R' 0.32 0.08 frequency-these are approximations to the long-run elasticities of total and social sector expenditures with respect to the price of oil. The results of the estimated Ps indicate a positive and statistically significant relationship between the changes in the trend components in real oil prices and in real total government and social expenditures. However, the smaller positive 3 coefficient Graph 10. Trend Components of Real associated with social sector expenditure Expenditures and Real Oil Prices suggests that as the real oil price trend changes upwards or downwards, the social spending trend does also, although at a lower 2Ao rate than total expenditure trend. 36. A similar series of regressions was 2.15 run for the period of continuous oil price 1.90 declines (1981-1998). In both cases the estimated coefficients (Ps) are negative; 1.65 implying that as the changes in the trend of oil price are negative, the changes in both the 1.40 /, total and social expenditure trends are positive. However, the estimated coefficient I-Total &p. Iend -Sodia] E&p. Trend--Oil Pri= e Trend for social expenditures is larger in absolute terms than for total expenditures, implying 6 This exercise does not attempt to provide a full model for the determination, structural or otherwise, of the trend component of expenditures, but simply to measure the responsiveness of expenditure trends to price trends. - 21 - that changes in the trend of social expenditures is larger than that of total expenditures in response to changes in oil price trend. 37. The estimated trend components of total, social expenditures and oil prices are presented in Graph 10. This shows first, from 1970 to 1981, significant increases in the trend of real total and social expenditures accompanying the increase in real oil price trend; and second, from 1982 to 1998, a marginal decline in the trend of real total expenditures and a marginal increase in the trend of real social expenditures, as the trend of real oil prices falls significantly. Thus, the econometric results and the corresponding Graph show that the Government has increased and "protected" real social expenditures over the long term in the face of falling real oil prices. 38. To analyze the cycle-relationship between total government expenditures, social expenditures and the price of oil, two additional regressions have been estimated: CTEc = a2 + 02 CPricec + u2 CSEc = a3 + 03 CPrice, + U3 where CTEC is the difference of the Graph 11. Cyclical Components of Real logarithm of the approximate cyclical Expenditures and Real Oil Prices component of total real government expenditures, CSEc is the difference of the logarithm of the approximate cyclical component of real social expenditures, and CPrice, is the difference of the logarithm of 0.13 the approximate cyclical component of the /?vA A / real price of the Venezuelan basket of oil; 0.00 and u2 and U3 are the error terns (Table 6.B). The Ps in the cycle 4.13 specifications can be interpreted as approximate measures of the oil price- . .......... elasticity of public sector expenditures at the cycle frequency. The estimated - - - - - - - - - - - - parameter for the first equation (total Total lp. Cycle-OilPriceCyde government expenditures) is significant, but the estimated parameter for second equation (social sector expenditures) is not. These results imply that while there is a positive and statistically significant relationship between the cyclical component of total real expenditures and that of real oil prices (see Graph 11), such a strong relationship is not found between the cyclical component of real social expenditures and that of real oil prices. 39. The above results suggest that while changes in total real expenditure trend are positively related to changes in the trend component of the real oil price (i.e., there is a "permanent" adjustment of total expenditures to oil prices), there is also a cycle component of the real price of oil affecting total real expenditure (i.e., there is a "temporary" adjustment of total expenditure to oil prices), supporting the hypothesis of - 22 - pro-cyclical fiscal responses. In contrast, while changes in social real expenditure trend are also positively associated with real oil prices over time, social expenditure adjustments in response to what may be perceived as being cyclical (or "temporary") changes in oil prices are not significant, supporting the hypothesis that social expenditures are not pro-cyclical (for more details, see Annex B). III. Composition of Social Sector Expenditures 40. Given that neither the level of expenditures nor volatility-inducing developments in the oil sector are obvious explanatory factors for the deterioration of the output performance of the social sectors in Venezuela, could it not be that perverse intra- sectoral dynamics are causing the deterioration? In this section, a general overview of intra-sectoral expenditure dynamics in the last decade is provided. By discussing the most salient features of resource allocation within the social sectors, this Section seeks an explanation for the faltering performance of social sector institutions in Venezuela.7 41. The share of social sector expenditures in the budget has generally increased over the last decade-from 34 percent in 1988 to about 38 percent in 2000 (see Table 7). Moreover, social sector expenditures have increased by over 3 percent of GDP since 1988. This reinforces the notion that the level of expenditures in the social sectors (in Table 7. Government Expenditure By Sector Annual Averages 1988 1989-92 1993-95 1996 1997 1998 1999 2000 Budget Shares (as % of Total Exp) Total Social Sectors 34.0% 35.1% 36.9% 32.3% 38.8% 36.3% 38.5% 37.9% -eduration 16.3% 13.3% 164% 9.7% 13.2% 15.6% 17.8% 16.8% - health 7.4% 6.6% 5.9'% 3.6% 5.9O/ 5.8% 5.9% 5.3% - social protection 0.9% 4.7% 5.2% 8.5% 5.9% 3.4% 3.6% 3.5% -housingandurbandevelopment 5.6% 6.4% 4.1% 4.8% 68% 4.1% 3.1% 5.2% -socialsecurity 2.8% 3.2% 4.0% 4.8% 5.5% 7.3% 8.1% 7.1% Other Social Sectors 1.0% 0.9% 1.4% 0.8% 1.5% 0.2% 0.0% 0.0% Total Non Social Sectors 66.0% 64.9% 63.1% 67.7% 61.2% 63.7% 61.5% 62.1% TOTAL 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% (as % of GDP) Sorial Seior Expenditures 7.3 8.9 7.9 8.0 8.5 8.2 9.0 11.0 Total Expenditure 21.5 25.4 21.4 24.8 21.9 22.6 23.3 29.2 Source: CORDIPLAN, IMF real terms) as a proportion of total expenditures and as a proportion of GDP, can not fully explain a fall in the quality or quantity of social sector outputs. However, it is worth noting that the social sector spending increase appears to be the result of the payroll component accounting for a large share of total social expenditures. Public sector unions have prevented a reduction of the labor force, so personnel expenditures have continued to rise despite an economy-wide decrease in real wages (see Chapter II). 7 Lack of data prevents a more detailed analysis of types of expenditures (e.g., current expenditure -wages and salaries-and capital expenditure) in the social sector and within each subsector. - 23 - Over the past 10 years, education has typically absorbed the largest share of resources (averaging 45 percent of total social expenditures), followed by health (17 percent). Expenditures on education and health have been fairly constant with the exception of a few years, particularly 1996 (see Graph 12). In contrast, the share of expenditures on social protection programs (social development and social security) has increased substantially, from 11 percent of total expenditures in 1988 to about 30 percent in 2000. The decentralization process (implemented since 1989) partly explains this increase, as power and funding for some social sector programs were transferred to sub- national levels of government (States and municipalities). Finally, the share of housing and urban development in government expenditure has fluctuated widely, between about 8 and 19 percent of social expenditures. To gain further insight into the intra-sectoral Graph 12. Structure of Social Expenditures I 00% / / 100% 40% ° 20% 0% 00 r* ' ) x- 00 O 00 O ~ O 00 O O Education 0 Health I Social development E Social securty 1 Housing U Other dynamics of expenditure, the responsiveness of specific categories of expenditures to fluctuations in total expenditure is measured by estimating elasticities. To establish whether or not certain expenditure categories exhibit asymmetric responsiveness, periods of expenditure contractions are isolated and analyzed separately. 42. The elasticities have been calculated such that: ei= X / X where ei is the elasticity of expenditures in the ith sector, Xi is the percentage change in expenditures in the ith sector and X equals the percentage change in total expenditures.8 8 The higher the magnitude of the elasticity, the greater the responsiveness of that category of expenditure to variations in total expenditures. Sectors that are favored in periods of real contraction of total government expenditure have elasticities of less than unity, that is, they would be expected to experience reductions of a proportionately smaller magnitude than the total expenditure cutback, or even real increases (if the elasticity is found to be negative). An elasticity - 24 - 43. Over the whole period of analysis, social expenditures and education expenditures were indeed "protected", as can be inferred from estimated elasticities of 0.76 (see Table 8). By contrast, the health and housing sectors were not "protected", as changes in total expenditures led to more than proportionate changes in expenditures (elasticities greater than 1). Social development and social security expenditures have negative estimated elasticities (-0.36 and -1.09 respectively), indicating that in years when total government expenditures fell, expenditures in these sectors actually increased. Table 8. Social Expenditures Are Protected Fiscal Contraction Episodes 1981-1998* 1992-1995 1982-1989* Elasticity % change Elasticity % change Elasticity Total expenditure ---- -8.57 ---- -10.00 ---- Social expenditure 0.76 -7.58 0.88 -8.87 0.89 Education 0.76 -2.98 0.35 -9.19 0.92 Health 1.29 -11.64 1.35 -6.19 0.62 Housing 2.53 -22.37 2.63 -8.12 0.81 Social Development -0.36 -1.57 0.18 27.07 -2.71 Social Security -1.09 2.33 -0.27 -2.80 0.28 Note: The elasticities are calculated with respect to changes in total expenditures. (*) means the years 1984 and 1987 are excluded (expansionary years). 44. How about during specific fiscal contraction episodes? For the 1980s, the elasticities show that social expenditures were not used disproportionately as adjustment variables to achieve expenditure cuts. While public expenditures on education were the "least" protected, expenditures on social development increased despite the overall contraction in public sector expenditures. During the 1990s, however, education and social development were relatively more "protected," whereas health and housing experienced disproportionate expenditure sacrifices. Outlays on social security, by contrast, moved in the opposite direction, rising in the face of overal] expenditure cutbacks. It is difficult to explain the reasons behind the dispersion of lthe estimated elasticities-in both fiscal contraction episodes, the reported elasticities ranged for a difference of about 3. They indicate, however, that total expenditure cuts are associated with a reallocation of expenditures among the social sectors, even though the relative amount of resources devoted to the social sectors, as a group, remains protected. 45. In an efficient allocation of resources, the change in the relative responsiveness of the various categories of expenditures to variations in total expenditures should reflect the evolution of strategic priorities. Coupled with correct prioritization of objectives and good strategic design, increased budgetary flexibility is highly desirable. But, in a setting of exactly zero means the sector is completely protected with respect to fluctuations of total government expenditure. Elasticities greater than unity suggest that the relevant category of expenditure adjusts more than proportionately to changes in total expenditures (e.g., a given percentage fall in total expenditures would lead to a greater percentage fall in that category of expenditure). - 25 - of ill-structured and poorly functioning institutions that lack clear strategies, these adjustments and dispersions likely do not represent optimal responses. IV. Improving the Social Safety Net System 46. One area where the shifts in resources have been especially apparent in recent years is in the programs that constitute the social safety net. These programs are aimed at mitigating the impact of economic crises by helping households maintain welfare levels and by preventing the deterioration of human capital. During the latter half of the decade, the programs received a significant share of resources. Issues related to the composition of these programs, however, such as the beneficiaries selected and the benefits provided, appear to have had limited overall effectiveness.9 The Social Safety Net in Venezuela 47. As a result of the economic conditions, Venezuelan policy makers made efforts to create a comprehensive social safety net system in the mid-1990s (see Table 9). While many of the programs included in the social safety net existed prior to this effort, the focus was to expand and revise existing programs and to generate new programs to act as an effective safety net for the poor. The different programs incorporated into this safety net ranged from day care centers to monetary transfers to the elderly, as well as service provision in the health, education, nutrition and labor sectors. 48. Significant resources were dedicated to these programs in 1996 (see Table 10). While the overall nominal level of spending increased in 2000-with much of the increase due to the expansion of benefits in the pension program-spending on several programs declined. In US$ nominal terms, excluding social security, the remaining safety net programs declined by about 35 percent between 1996 and 2000. In relative terms, spending has also declined: in the first year of the systemn, spending on the social safety net programs represented about 1.5 percent of GDP; this declined to about 0.5 percent for the last three years of the decade. But the spending decline did not correspond to an increase in economic growth, and hence a decline in need, as in the last five years only in 1997 was economic growth strongly positive (see Chapter II). The lack of external and internal sources of data make it difficult for detailed overall assessment of the effectiveness and efficiency of the social safety net in decreasing the costs of adjustment and alleviating the worst effects of the economic crises. Nonetheless, available information does provide some insights. - 26 - Table 9. Key Social Safety Net Programs Program name Description Family Subsidy Cash transfer to poor families Strategic Nutrition Subsidized prices for basic food goods School Nutrition One meal to pre-school through 6th graders in poor localities School Uniform Provide school uniform to children enrolled in pre-school through 6th grade in poor localities School Snack and School snack or meal to school children in poor localities as well as non- Cafeteria enrolled children in poor localities Medicine Subsidy Subsidized medicines to those using the public health system, with part of subsidy paid by private sector pharmacies PAMI Health and basic nutrition services to poor pregnant and lactating women and their young children Medical Supplies Provision of medical equipment to public hospitals and clinics Day Care Health, nutrition and child development for poor children ages 0-6, as well as support to mothers Transport Subsidy Indirect and direct subsidies for transportation systems or tickets for students at all levels of education Youth Training Training to poor youths ages 15-24 who have not obtained educational credentials Old Age Protection Cash transfer to those over age 65 who are not eligible to receive a pension, also some nursing homes Social Development Temporary employment in poor communities to carry out rehabilitation of local infrastructure Social Security Increase in value of basic pension under the Social Security Institute Source: Fichas Tecnicas, former Ministry of the Family, 1996. - 27 - Table 10. Social Safety Net Programs Spending, 1996-1999 (In US$ Millions) 1996 1998 1999 Beneficiaries Program name 1997 (actual) (actual) (preliminary) (million) Family Subsidy 415.5 95.7 ------- 2.9 Strategic Nutrition 28.9 26.8 28.2 4 SchoolNutrition 36.0 89.9 106.4 1.1 School Uniform 93.7 87.7 78.5 1.1 to 2.5' School Snack and 62.0 94.3 86.1 2.9 Cafeteria Medicine Subsidy 19.9 7.5 7.8 0.6 PAMI 22.6 38.5 37.1 0.3 Medical Supplies 6.6 40.0 41.4 0.01 to 8.7 Day Care 80.8 96.8 95.3 0.35 Transport Subsidy 43.2 49.3 46.4 5.6 Youth Training 21.8 21.6 13.1 0.04 Old Age Protection 26.6 32.4 32.9 0.1 Social Develop 26.0 13.4 9.5 ------ Social Security 60.2 399.8 979.6 0.2 to 0.3 Total 943.8 1093.6 1562.5 ------ Total Without Social 883.6 693.8 582.7 Security Source: Fichas Tecnicas, former Ministry of the Family, 1996. (*) means that the range in the nurnber of beneficiaries is due to differing methods of calculation. 49. Concentration of Resources. Although the social safety net is comprised of 14 programs, six of these receive the bulk of resources. Specifically, family subsidy, pension increase, school nutrition, school snack/cafeteria,10 school uniforms, and day care programs represented about 80 percent of total expenditures in 1996 and about 90 percent in 1999. In 1996, family subsidies (a cash transfer to poor families) was the most important but it was no longer funded in 1999-by which point top importance went to pensions for eligible individuals under the Venezuelan Social Security Institute ("social security"; see Graph 13). to The school snack and cafeteria programs are actually separate but are bundled in some financial reports. Thus they are treated as one here, but when important, are treated separately later on. - 28 - Graph 13. Share of Social Safety Net Spending-Key Programs (1996-1999) 70.0 - 60.0 -0_' c ;v 50.0 r :-.-. .:.I - 3 . *19 O . . - - . - . - : : . : - 1 > . *~~~~~~~ 1998 adj CL lj~~~~~~~~~~~~~~~~~~~~~~~~~~~1 40. _0999 prop z 30 ..0. 2 0.0 < l$ :g ; - :- .. . 3 0.0Xi__ Family School School School Snack Day care Soc. Secunty Subsidy Nutrition Uniform Source: IDB and World Bank 50. This concentration of resources-as well as significant and broad overlap in goals among the main programs in the social safety net (see Table II )-has resulted in a system that is focussed on a narrow segment of the population (school-age children and pensioners). The available evidence indicates that these groups do have some level of need but the evidence does not clearly support the relative importance given to them.'1 Table 11. Goals, Benefits and Beneficiaries of Main Social Safety Net Programs Program Goals Benefit type Beneficiary Family Subsidy Income support, enrollment Cash Elousehold School Nutrition Child nutrition, school attainment In-kind Individual: Child School Uniform School enrollment In-kind Individual: Child School Snack, Cafeteria Child nutrition, school attainment In-kind Individual: Child Day Care Child development, mother labor opportunities In-kind Child and mother Social Security Increase base pensions Cash Individual: Pensioner Source: Fichas Tecnicas, former Ministry of the Family; Lee et al; 1DB and World Bank. It is of course, plausible, that the observed concentration of resources is actually a reflection of a well-targeted system wherein the most vulnerable groups and their respective needs have been identified. However, determining the beneficiaries of any program is complicatedi. There are both direct beneficiaries (who the program is defined to assist) and indirect beneficiaries (who may or may not have been contemplated in the original program design). Moreover, as public spending tends to crowd out private spending, the real beneficiary and benefit may be quite different from that defined in the program. - 29- 51. More specifically, education data show that, after a rise in enrollment rates in the 1980s and early 1990s, rates fell in the mid 1990s (see Graph 14). Interventions were thus based on the need to increase school enrollment and nutrition. Two-thirds of the main programs defined their beneficiaries to be children of school age, practically all of whom were enrolled in a formal school program.12 Resources were targeted to poor children, using schools in poor areas as a proxy for targeting. Given the problems with the poverty data (as noted in Chapter II) it is not clear that such targeting has been successful. 13 52. As for nutrition, the evidence of the importance or need for such a strategy is mixed. There does appear to be evidence that malnutrition has been on the rise-or at least stagnation in gains in the reduction of malnutrition among children in the mid- 1990s. Accordingly, the decision to focus on malnutrition as a key element in the social safety net appears sound.'4 However, the data do not show that the malnutrition problem is concentrated among the school age population. Therefore, it cannot be concluded that the right group has been targeted for nutrition interventions. 12 The school nutrition program was aimed at children from pre-school through 6th grade, enrolled in school. Some element of geographical targeting was also employed to reach poorer areas. The school cafeteria program targeted Ist-6th grade students, for school snacks the target group was the preschool age group. Most of the beneficiaries were in pre-school although part of the program provided the benefit to poor children not enrolled in pre-school. School uniforms were also targeted to children from Ist-6th grade, again in poor localities. The family subsidy was targeted to children in Ist-6th grades in poor localities. The only exception to the targeting of children of school age were the day care centers which were explicitly for children not in other schooling or pre-school programs and the pension program (for retirees). Geographic targeting requires accurate data on poverty dis-aggregated to small areas. The data available is probably inadequate for this purpose and, in all likelihood, has resulted in both errors of inclusion and exclusion. The lack of counterfactual evidence, and more recent data, makes it impossible to determine the impact of the initial benefit of the program, as well as the more recent school feeding programs. 14 There are two important caveats to this: First, the manner in which malnutrition is measured in Venezuela tends to overstate the nutrition problem compared to results obtained using WHO standard definitions of malnutrition. Relying on such definitions of malnutrition has probably led to a greater concentration of resources being devoted to problems of malnutrition than might be necessary. Second, there are biases in the malnutrition data, regardless of the definition used. These biases are due to the fact that no national, statistically significant sample of children exists; instead data are collected from children who visit health clinics. This method of data collection may lead to either over- or under-estimates of malnutrition. On the one hand, malnourished children are sicklier and are more likely to be taken to a health center than healthy, well-nourished children. Hence, the sample of children is biased towards those with the problem one is trying to measure leading to overestimation of the problem. On the other hand, the poverty level of the family in which a child lives, the poverty that has led to malnutrition in the first place may prevent the child from being taken to medical care. Where distances to health clinics are great, transportation costs high and the opportunity cost of the parent high, the very children that are poorest and most malnourished may not be included in the statistics, thus leading to an underestimation of malnutrition. So, regardless of the measure used there may be serious measurement error. Unfortunately, the direction of the bias is not clear, which limits any conclusions that can be drawn about the appropriateness of the emphasis on malnutrition and/or the impact of public spending. - 30 - Graph 14. Enrollment Rates by Age Group (1984-1996) 0.0~~~~~~~~~~~~~~~~~~. 8~~~~~~~~~~~~~~~~U18 70 0- 60. 0 Source: UNESCO Data Base 53. The evidence supporting the need to concentrate resources on the elderly is weaker than that for school age children. First, while some data exist showing that households with heads over 50 years of age have a greater level of poverty,'5 it is not clear that generally targeting the elderly is an effective mechanism. In 1995, just over half of all households with heads over age 50 were not poor. Thus, additional information on older individuals is needed if age is to be a criterion for benef.ts. 54. Second, there is doubt as to whether the elderly benefiting from this program are the poorest among that group. The pension support program tops up the pens,ion amounts of those eligible to receive pensions. But pension coverage is only around 25 percent, and it is not clear that those eligible for pensions are the most needy. In the first place, those eligible for pensions tend to be those who have worked in the formal sector and have long years of services. Thus, they are individuals who have had greater income stability, if not greater incomes, over their working years, in contrast to people working in the informal sector or with long periods of unemployment or being inactive. The former group is less likely to be poor than the latter. Moreover, in a country where pension coverage is far from universal, a policy of increasing coverage and not simply concentrating more resources on those already receiving a pension, could have more of an irpact on poverty. In short, the group presently receiving the greatest share of social safety net spending is not demonstrably the group with the greatest needs. 55. Types of Benefits. A key feature of the main programs of the social safety net has been the preference for in-kind, as opposed to, cash benefits. Two thirds of the programs relied on in-kind benefits (school nutrition, school snack and cafeteria, school uniforms, daycare centers). The pension and family subsidy were the only two to rely on cash benefits, and the family subsidy has been phased out. those ebKeeping in mind the caveats about such poverty data noted in Chapter 11. - 31 - 56. In-kind benefits have both advantages and disadvantages. In terms of advantages, for at least some of the types of benefits provided, the in-kind benefits assure that the defined, primary beneficiary reaps the benefits from the program. For the school feeding programs and the daycare centers, the educational benefits directly accrue to the child in the program. The nutritional benefits, however, might or might not be fully enjoyed by the targeted child as the interaction of public and private spending affects this. To the extent that the meal provided to the child at school replaces household spending on that child, the nutritional impact is not as large as would be expected.'6 A disadvantage of in-kind benefits is very high administrative costs, which lower the overall resources spent on the in-kind benefits. For example, for some in-kind benefit programs where up to a third of the costs is administrative, compared to the family subsidy where administrative costs take up less about 10 percent of total costs. 57. The specific program costs also show that there is potential for cost savings and increased effectiveness. For similar programs, the costs are widely different. The benefit of the school nutrition program and the school cafeteria program was the same-one meal a day equivalent to 30-40 percent of needed daily calories. Yet, as can be seen in Table 12, the costs of providing this benefit were not the same, either in total (US$107 under the school cafeteria program compared to US$74 under the school nutrition program) or in terms of the cost of the provided food itself (US$41 under the school cafeteria program compared to US$62 under the school nutrition program). In other words, the same food costs one third more under the school nutrition program, but the school cafeteria program has such high administrative costs that, overall, its program is significantly more expensive. Clearly there are costs savings that could be obtained. 16 This interaction of public and private spending also limits the net effect of cash transfers. For the pensioner, the net benefit of the pension increase, depends on the extent to which the recipient controls the resources and the extent to which receiving a pension lowers other assistance from private or public sources. - 32 - Table 12. Costs of Main Programs are High (1997) Program US$ per year per Administrative Costs (% Food Procurement Costs beneficiary total costs) (% total costs) School Snack 31 32.5 78.2 School Cafeteria 107 32.5 38.1 School Nutrition 74 n.a. 84 .02 Family Subsidy 118' 10.0 n.a Pension support 1093 n.a n.a. Day Care 248 7.7 47.4 Source: Lee et al, IDB and World Bank. The value of the benefit received by the household was six US$16 dollars payments per child per year (up to three children) in 1995-97. In 1998, this was down to two such payments per child. 2Refers to only one State (Barinas), no consolidated information available. 3Refers to 1998, includes only value of benefit given out, does not include cost of providing benefit. 58. Administrative costs in the entire system are higher than needed. The duplication of effort in the provision of food also hampers the effectiveness of school feeding programs. The two school feeding programs that are providing the same benefit are run by different institutions, implying two sets of administrative staff and overhead, two different procurement mechanisms, and the like. Merging these two programs could generate economies of scale and gains in efficiency. 59. A further concern with in-kind benefits under the social safety net program is related to the emphasis on food. The large quantities of foods that are procured under the social safety net programs means that the interaction between the programs and prices is important. In 1997, for all social safety net programs, food purchases of corn flour represented about 13.5 percent of the amounts of commercial flour availablie for human consumption in the country, purchases of oils for the programs represented about 19 percent of oils available, and purchases of fluid milk under these programs was about 24.5 percent of the available total. The magnitude of the purchases under the social safety net programs of certain food goods has had both positive and negative impacts. 60. On the positive side, the programs providing food have been able to purchase in bulk with related discounts. The strategic nutrition program, school snacks and PAMI all purchase goods in a centralized manner with formal national and regional bidding processes. This has led to significant savings in food costs. For other programs, such as the school feeding and the school cafeteria programs, which have used much more decentralized procurement processes, there appears to be scope for cost reductions if more centralized processes wereX adopted. This is especially importanit given the 1 7 magnitude of these programs vis-a vis the other food programs. 61. On the negative side, seve al existing economic policies undermine the effect of the nutrition programs. Administered prices, trade protection (both domestic and through the Andean Group), market structure and prices factors (including an overvalued 1 7 There is some evidence that these programs have distorted the demand for food products and negatively affected local producers (see Lee et al, 1999). - 33 - domestic currency), have increased the price of food in the country. This has depressed consumption and has contributed to the malnutrition problem that the social safety net programs are attempting to resolve. And, at the same time, the increases in food costs reduce the number of beneficiaries that can be served by the programs. A liberalization of prices would lead to a significant increase in the number of beneficiaries in several programs (see Table 13). Table 13. Impact on Programs of Price Liberalization Programs Beneficiary Months Percent Actual 1997 Additional Simulated 1997 Increase Day Care Centers 4,209,264 53,468 1.3 PAMI 355625,740 105,866 3.0 Strategic Nutrition 48,990,588 12,852,399 26.3 School/Snack 10, 066,740 10,066,740 4.0 School Cafeterias 3,330,432 9,768 0.0 School Nutrition 13,1045,144 13,104,144 6.6 Source: Lee et al, 1999. 62. A final concern related to the many programs (not just those with in-kind benefits) is that there are wide discrepancies in both the criteria used to assign benefits and the content of programs with the same goals. Multiple providers of the same or similar services often vary on beneficiary selection criteria. Moreover, the high levels of decentralization have led to the application of differing eligibility criteria within the same program but in different areas. This indicates a potential lack of effectiveness in targeting that may damage the public's perception of the programs. This type of confusion also arises in the way the programs provide support to the elderly. Under the pension support programs, the elderly received ten times the amount as under the other cash transfer program. 63. New Plan. Plans for substantially revising the social safety net were drawn up in 1999 and a new mechanism for providing social safety net (and other assistance) programs is being designed. The new plan creates one entity, the Fondo Unico Social (FUS), that would administer six of the original 14 programs plus additional ones supporting cooperatives and an early childhood development program. The Family Subsidy program is to be phased out; the other remaining programs will be administered in the manner in which they have been in the past (e.g., school feeding programs), while the future of others is still undefined. Given the major shifts, the evaluation of the prior programs and the main findings reported here are imnportant tools for improving the effectiveness of the new social safety net. V. The Institutional Framework 64. This Section discusses the nature of the social sector institutional framework, including the impact of the decentralization effort and the role of some of the "social funds" prior to the new Constitution approved in December 1999. Given the complex - 34- structure and the lack of capacity and coordination to formulate and implement social sector policies, the institutional framework has been a leading cause of inefficient resource use, contributing to the declining quality and quantity of social sector outputs in Venezuela. 65. At the highest level, the Council of Ministers (led by the President) has been responsible for general policy-making. Several "sectoral Cabinets", composed of selected Ministers and senior officials, were created to support the operations of the Council of Ministers.'8 For many years, the Social Cabinet had responsibility for developing social sector policy. The Social Cabinet functioned as a relatively open "umbrella" organization in that a large number of officials could attend. This made meetings of the Social Cabinet informative, but it also meant that the discussion and debate were broad and that political considerations tended to overshadow the more technical aspects of policy-making in the social sectors. Moreover, the lack of clear links between the interests of the various members of the Social Cabinet and the interests of program beneficiaries suggests that "principal-agent problems" have likely arisen as a result of the structure of the policy-making process. 66. The National Monitoring and Supervisory Council for Social Programs was created in January 1991 in an attempt to enhance the Government's capacity to monitor both the performance of social sector institutions and the implementation of social sector programs. This Council consisted of four members, designated by the parties represented in Congress, under the leadership of a member designated by the President. The National Monitoring and Supervisory Council was replaced in June 1996 by the National Council for Monitoring of the Social Protection Component of the Agenda Venezuela. This body consisted of 19 members, seven of them designated by Congress, and the others by the President, the Ministry of Defense, Family Affairs, Cordiplan, Ocepre, National Food Council, Fedecamaras, the Church, Central Workers Union (CTV), the mass media, the Association of State Governors, and the Council of City Mayors. Given the complexity of its structure, the Council was not able to function coherently either to appropriately represent different interests, or to protect the poor.19 67. The new Government has recognized that the complexity of the structure and the lack of capacity to formulate, coordinate and implement social sector policies affected social sector performance in Venezuela. As part of the comprehensive Ley Organica de La Administraci6n Central and the new Constitution, both approved in 1999, the new Government has introduced important changes that will significantly affect the formulation and implementation of social policies in Venezuela (see Box 4). It is, however, too soon to make a substantive evaluation of the impact of these changes. Specific guidelines for the operation of sectoral Cabinets, including the Social Cabinet, were issued through Decree 292 of July 1994. 19 A proposal was circulated during 1998 to create a National System of Social ]ndicators, as a mechanism to improve the Government's understanding of social sector issues, to enhance policy design and to tailor implementation to the specific conditions of a locality or targel group. Such a proposal has not been implemented to any significant degree. - 35 - Box 4. The August 1999 Reform Package from the Perspective of the Social Sectors The Government of Venezuela has introduced a package of reforms aimed at improving Central Government performance. Although it is premature to establish whether or not these reforms have succeeded, this Box presents the general orientation of the reform package, with particular reference to the social sectors. The reform package (introduced through Decree 253 of August 1999) is justified by the need to (i) strengthen the links between the State and society; (ii) increase transparency and accountability; (iii) improve coordination, control and evaluation of the activities of governmental institutions; (iv) change institutional cultures (e.g., to reduce bureaucracy and introduce a performance orientation); and (v) adapt institutions to the ongoing decentralization effort. The broad thrust of reforms is in the right direction. It is, however, too early to determine if the implementation of these reforms will obtain a measurable increase in the performance of Government with regard to the social sectors. The unification of the Ministries of Health and Family Affairs into the Ministry of Health and Social Development, the expansion of the Ministry of Education to oversee Culture and Sports, the change of CORDIPLAN into the Ministry of Planning and Development and the creation of the Ministry of Science and Technology may simplify the Cabinet structure. This may, in turn, facilitate more focussed discussion on social sector strategy. Nonetheless, delegating the more administrative and operational aspects of policy to the sectoral Cabinets does not solve other fundamental problems, including the lack of participation of civil society. The measures taken to explicitly increase the accountability of government officials and institutions are in the right direction (e.g., citizens can claim damages for the actions or omissions of Government). However, abstracting from the issues of enforceability of this type of provision, without the introduction of a positive incentive system or at least an attempt to align more closely the interests of government officials with those of the public, the system will likely remain ineffective. While the Ministry of Planning and Development is responsible for program coordination across sectors and levels of government, including regional development initiatives, it is not clear what will be its role in practice in relation to the Decentralization Law, the role of the Social Cabinet and the role of the National Monitoring and Supervisory Council for Social Programns. The future role of the Armed Forces in the development process, and in the social sectors in particular (including at the local level), is also not sufficiently clear. Overall, Decree 253 can be regarded as a first step in the reform process, providing a general framework for a restructured Executive. Deeper, sustained and consistent actions need to be followed before improvements in performance become apparent. It is hoped that the regulations of the new Constitution will provide a more precise framework for the type of reform effort that Venezuela needs. 68. In the late 1980s and early 1990s, Venezuela adopted a decentralization program (in accordance with the 1990 Decentralization Law) that entailed a strengthening of sub- national governments in terms of resources, greater autonomy and an expansion of responsibilities, as well as increased demands on their capacity to manage, implement and deliver services. In particular, it sought to draw the boundaries of responsibilities between national and sub-national governments. With regard to social sector services, the law assigned joint responsibility of national and sub-national governments.20 Nevertheless, in practice, no clear delimitation of responsibilities has emerged. More importantly, it is unclear how the decentralization process will change with the adoption and implementation of the new Constitution. 20 The social services included here education, health and nutrition, employment services, childcare, development of the rural population and housing. - 36 - 69. The empowerment of sub-national governments has not increased their capacity to raise revenues, but rather has led to a growing reliance on resource transfers from the national government.21 In addition to transfers from the national government, lower levels of government (i.e., State and municipal governments) have some capacity to levy taxes and fees and to use the resulting revenues to finance social sector expenditures. 70. The decentralization efforts have resulted in an increase in expenditures by sub- national levels of government as a proportion of total public expenditure in the social sectors. It is less clear that the increase in resources spent by sub-nationial levels of government has led to a proportionate increase in the quality or quantity of their outputs. 71. The complexity of the task of coordinating social sector programs with the involvement of a gamut of national, regional and local actors, cannot be overstated. The situation has deteriorated as the transfer of responsibilities between the various levels of government has not necessarily occurred homogeneously across geographic areas due to a transfer mechanism consisting of individually negotiated agreements between the national government and the States. 72. Another exercise carried out by the States was the yearly Coordinated Investment Plan, which was agreed upon within the Association of State Governors and endorsed by the Central Government. The Coordinated Investment Plan influenced the overall allocation of expenditures among and within States and thus provoked externalities that affected social sector expenditures in unpredictable ways. State Governors did not have well defined incentives to ensure that their interests were aligned with those of the beneficiaries of social programs. Moreover, it was not clear how the public could systematically monitor the performance of social programs at the sub-national level and link outcomes to specific actions by regional or local officials. 73. The picture that emerges is one where multiple institutions and officials responsible for the design, implementation, monitoring and supervision of social sector policies and programs have not had clear and common objectives and responsibilities, and have lacked appropriate incentive-performance mechanisms.22 Furthermore, the public has (i) had limited mechanisms to influence the design of policies; (ii) not exercised sufficient participation in the monitoring and supervision of sociial programs; 21 There have been three mechanisms for resource transfer to sub-national governments (i) the old Constitution established a given percentage of the current income of Central Govermment to be distributed among the States in proportion to population; (ii) the Law on the Special Allocation of resources established that 25 percent of revenues generated by mining and hydrocarbon taxes were earmarked for transfer to the States; and (iii) the Intergovernmental Decentralization Fund has also been channeling resources to the States. 22 In addition, as a result of the December 1999 floods in Vargas State, the Goveniment created a Bolivar 2000 Plan which has many elements typical of a social development program. The project counts with the active involvement of the military in the implementation of social infrastructure reconstruction, employment generation schemes, and emergency hospital care in the Vargas State. Bolivar 2000 Plan is an "emergency" and "transitory" operation that has lasted over a year already. It is unclear how or when will the more permanent programs discussed above take over the tasks now under Bolivar 2000 Plan. - 37- and (iii) lacked the necessary means to regularly enforce accountability on elected officials or bureaucrats. This situation is not unique to Venezuela, as other countries have also experienced inter-institutional coordination problems, resulting in inefficient use of resources (see Box 5). 74. Various "social investment funds" have been used at different times to try to overcome the deficiencies of the various levels of government (see Box 6). The purpose of these funds was to coordinate the design, promotion, execution and financing of specific social programs. Initially, the funds were charged with the implementation of emergency programs designed to alleviate the temporary distress caused by economic shocks and adjustment programs. In many cases, it was found that coordination was required to ensure adequate maintenance and project sustainability after the social investment fund ceased. It has also been found that the activities of the funds require coordination with those of the NGOs (e.g., to facilitate community participation). Box 5. Social Policy Formulation in Peru Peru has had a multitude of social programs that operated independently from each other, used different service-delivery mechanisms and were not subject to stringent monitoring and evaluation. Although well-intentioned, many of these programs failed to reach the poor. For example, the Ministry of the Presidency had several programs in the education sector, in addition to those run by the Ministry of Education. The Ministries of Finance, Women and Human Development, Health, Education and Presidency all have nutrition programs. Policy coordination has been hindered by conflicting mandates. For example, an April 1997 decree made the Council of Ministers and the Social Policy Coordination Council "simultaneously" responsible for coordinating and improving the allocation of social expenditures among agencies and ministries. Another decree (June 1997) officially adopted a targeting and coordination strategy to be developed and managed by the Ministry of the Presidency, thereby making this Ministry "de facto" responsible for social policy coordination. Box 6. Some Social Funds in Venezuela In 1974, FUNDACOMUN was created to organize and assist in the improvement of conditions in urban slums. In 1983, the Social Development Fund was formed to carry out a Nutrition Bonus (Bono Alimentario) project. Subsequently, the National Fund for Child Care was created, and FUNDACOMUN's mandate was expanded to implement a program for social investment at the local level. In 1990, the Government created FONVIS with the general purpose of promoting and financing social programs and projects in urban and rural areas, in accordance with the priorities set out by the Executive. The Fund was expected to coordinate its activities closely with CORDIPLAN. Under the direction of the Minister for Family Affairs, the Board of Directors of FONVIS included representatives of the Ministry of the Interior, Ministry of Finance, and CORDIPLAN. Communities were able to submit projects for the consideration of FONVIS through local governments, thus strengthening the decentralization efforts. In an attempt to strengthen the decentralization process, FIDES (attached to the Ministry of the Interior but with financial and administrative autonomy) was established to provide resources to strengthen sub- national governments, including in their capacity to implement social sector programs. - 38 - 75. Given their structure and relative flexibility, the social funds could respond to the needs of the population more effectively than the more traditional institutions. Unfortunately, a multitude of funds has been superimposed on the pre-existing institutional framework, thereby making the overall institutional infrastructure even more complex. Rather than correcting the lack of participation of target populations in the design of social policy within pre-existing structures, the funds solidified a government strategy through the creation of parallel structures with overlapping mandates. The existence of these parallel structures has raised further issues of coordination, monitoring, supervision and accountability. 76. The design, style and operation of the social funds is that of temporary structures designed to address the most pressing needs of the population. They are patches, but not solutions to the more fundamental structural problems of the "mainstream" public institutions. Given the large number of multiple, overlapping social funds, the new Government has recognized the shortcomings. As part of the overall strategy to improve social sector performance, the Government has started to consolidate all the social funds into the FUS, which in turn may improve coordination and targeting of social programs. Given that the FUS is in the early stages of its implementation, it is Graph 15. Quality of Bureaucracy Index premature to make any assessment regarding its impact. 10 77. In sum, the social sector 7 e-__ _ -----. __.- institutional framework has not m5 facilitated the design of internally 3 - -''___:'________- - - - consistent policies, with clear 2 objectives and with strategic 0 priorities. Moreover, budgetary - $ S - T allocations, incentive mechanisms, Y .... - - Me> ----Aryenlina -Badi -- - CNle and monitoring systems have not ---Wonba ---Pe -Venezuela been used effectively to implement social sector policy. The Government has not significantly leveraged its resources by 10 partnering with NGOs and the ,, 8- private sector. 6 - JL4- 78. In this complex and C 2- overwhelming setting, with weak i 0- - 0) 0) 0) 0)0> 0 ) 0> . 0> 0) 0 administrative capacity, the system a X 0 0 0) 0) ) s gave way to "informality" and "discretionality", as insufficient -a>, - ... attention was given to the -.--CdTbia - --Pe Vaab strengthening of information gathering, accountability and mechanisms to improve policy design and monitoring performance, thereby increasing the possibilities for corruption. In fact, the International Country Risk Guide (ICRG) Quality of Bureaucracy Index (Graph 15) and the Corruption - 39 - in Government Index (Graph 16) both suggest larger problems that would necessitate a more comprehensive and coordinated reform effort that cuts across all sectors and levels of the bureaucracy.23 79. Over the last few years, these indexes in Venezuela have not improved, thereby supporting the hypothesis of public sector inefficiency. Without evidence to the contrary, the levels of corruption and the general quality of the bureaucracy associated with the social sectors are likely to be highly correlated with, and somewhat similar to, those exhibited by the government bureaucracy as a whole. Thus, the challenge to improve efficiency in the social sectors, including reducing corruption, is quite considerable. VI. Conclusions and Recommendations 80. The Chapter suggests that, before increasing social sector spending, the use of existing resources must be made more efficient and equitable, based on strategic priorities. To achieve this, a comprehensive institutional reform is warranted. Venezuela made some efforts in the 1 990s to strengthen its institutional framework for social sector policy-making, but the reforms did not succeed in improving the efficiency and the quantity/quality of services delivered. 81. Key areas to be addressed include: First and most importantly, the level of expenditures is not the most pressing problem. Social expenditures have been relatively protected vis-a-vis oil price fluctuation and fiscal adjustments. The problem lies with the efficiency with which resources have been allocated and used. Second, social sector institutions have lacked the capacity to formulate, coordinate and implement social sector policies. As a result, a large numbers of social funds emerged to overcome the deficiencies of the various levels of government. Unfortunately, these funds have been superimposed on the pre-existing institutions, thereby making the overall institutional infrastructure even more complex. Third, coordination, incentive, monitoring and evaluation mechanisms do not exist or simply do not work at the intra- and inter- institutional levels. Finally, the broader institutional and legal framework, and public sector culture (including the weak administrative capacity) have also contributed to the deterioration of social sector performance in Venezuela. 82. Some specific proposals to enhance social sector performance include (i) improving the efficiency in the allocation and use of existing resources through a clear definition of social policy objectives; (ii) strengthening the Social Ministries by prioritizing administrative capacity building, including reducing discretionality and thereby limiting possibilities for corruption; (iii) simplifying, streamlining, and increasing the administrative efficiency of social programs to improve their cost-effectiveness, including merging and simplifying social sector institutions and funds; and (iv) implementing a deep and comprehensive reform of the public sector using the opportunity afforded by the introduction of the new Constitution. 23 Although these indexes have embedded well-recognized and non-trivial elements of subjectivity and judgement, given their internal consistency over time, they are generally useful in detecting trends and turning points in the behavior of government entities. - 40 - Chapter IV. How to Improve Human Capital and Prevent its Deterioration?' 83. While social assistance programs can mitigate poverty in the short-term and protect the living standards of the population against major variations in real income, long-term poverty reduction will depend on rapid economic growth combined with substantial investments to accelerate the accumulation of human capital. Key types of human development investments are education, health and nutrition. This Chapter reviews the evidence on investments and results in these sectors. Based on the findings, the Chapter provides selected policy recommendations to accelerate human capital accumulation and to increase labor productivity. In addition, given the increase in violence and crime affecting not only the transaction costs of "doing business" in Venezuela but also social and human capital development, the Chapter explores some of the causes and factors affecting crime and violence. I. Investing in Education' 84. This Section reviews the evidence on Venezuela's educational attainrnent over time, demand for schooling, incidence analysis, internal and external efficiency and selected key policy decisions. Although educational attainment has improved over time, in general, the analysis reveals that its performance is inefficient and ineffective, and public investment in education continues to be regressive. These issues need to be addressed to accelerate human capital development in Venezuela. Educational Attainment Over Time 85. Since 1975 the average educational level of the labor force in Venezuela has almost doubled, from only 4.6 years, to 8.2 years in 1997 (see Graph 14). The proportion of the labor force with no schooling decreased significantly from 16 to about 5 percent between 1979 and 1997. By contrast, the proportion with university education increased from 6 to about 11 percent in the same period. The proportion of the labor force completing basic education-extended to nine years in 1980-continues to increase over time.2 * This Chapter was prepared by Vicente Fretes Cibils and Harry Anthony Patrinos. This Section draws on a background paper: "Education: Venezuela: Poverty Analysis-Final Report" (Bruni Celli, J., Arnas, V., and Gonzalez, M.; October 1998; Caracas). 2 A 1998 survey found that 6.7 percent of the population 10 year or older is illiterate. Only 12 percent of the population had attained a university education, and about 53 percent a basic education. Average schooling was 6 years in 1998. Source: ENSO 98, FUNINDES, Universidad Sim6n Bolivar. - 41 - Table 14. Educational Attainment of the Labor Force, Venezuela 1997 Variable Males Females All Years of schooling 7.7 9.2 8.2 Educational level (percent) No education 5.9 3.2 5.0 Primary 64.4 50.4 59.8 Secondary 18.3 26.2 20.9 Higher 3.2 4.5 3.6 University 8.2 15.7 10.7 Source: Direcci6n de Estadisticas de Poblaci6n, OCEI, 1997, Caracas. 86. The increase in educational coverage is consistent with important policy thrusts, as basic education in Venezuela is free, government-funded and compulsory. The 1999 Constitution reinforces these policy principles, extending the right of citizens and the responsibility of the State to ensure free access through undergraduate levels in the public system. Enrollments have increased over time, resulting in a relatively well-educated labor force. However, the relative success at the basic level has led to pressures on the secondary level, where increased enrollment is putting pressure on public resources. Nevertheless, the school system is still characterized by relatively high levels of grade repetition, dropouts, non-completion rates and low coverage-particularly at the secondary level. The quality of education remains below international standards and there are significant inefficiencies and inequities (see below). Demand for Education 87. The demand for education and its relationship with poverty is empirically analyzed by estimating the probability of school attendance and its determinants.3 The probability of school attendance of the population between the ages of 6 and 15 rose from about 92 percent in 1990 to about 95 percent in 1997. This suggests that interventions to promote school attendance have increased the probability of school attendance by as much as 3 percentage points. However, this result contradicts the Ministry of Education's statistics, which show that in the last five years school attendance has decreased by 9.6 percent in the first two phases of basic education and increased only by 0.4 percent in the third phase of basic education.4 Bearing this in mind, it is possible to The Encuesta de Hogares por Muestro (1990, 1995 and 1997), conducted by OCEI, was used. The analysis was limited to children between the ages of 6 and 15 (ages when school attendance was mandatory at the basic level). Basic education included three levels: the first two phases were in grade school (grades I to 6) followed by the third phase, junior high school (grades 7 to 9). While there are variations in the inter five-year period figures (for example, there as a net increase of 1.4 percent between five-year periods in the total coverage of basic education in 1989-1994), the enrollment figures of the Ministry of Education consistently indicate that coverage has dropped in net terms since 1991. It is possible that the contradiction in the results is due to problems with the Ministry of Education's data because, if there were a bias in the OCEI data (for example, parents over-reporting the school attendance of their children), that bias would not have changed over time. - 42 - cautiously conclude that interventions might have had a positive effect on school attendance. 88. Children in urban areas are more likely to attend school than are children in rural areas. However, this has been changing. In 1990 school attendance by children in urban areas exceeded that of children in rural areas by 10 percent; by mid-1990s, school attendance by children in urban areas exceeded that of children in rural areas by only 5 percent. Girls are more likely to attend school than boys. In 1990 the probability of an urban girl of average age (10.3 years) attending school exceeded that of an urban boy by 1 percentage point. This difference had not changed significantly by 1997. In rural areas the difference between attendance by boys and girls is greater, with the probability of girls attending school exceeding that for boys by 2 percentage points. This difference between attendance by boys and girls increases with age, particularly in rural areas. For example, in rural areas the probability of 14-year old girls attending school exceeded that for boys by 3 percentage points. As boys grow older, there is a smaller probability that they will be attending school. The negative impact of age on the probability of boys attending school intensified during the 1990s, due to an increase in the probability of school attendance by younger boys and a slight decrease in the probability of attendance by older boys. The higher the per capita income of the child's family, the greater the probability of the child attending school. However, the effect of this variable decreased during the 1990s both in cities and in rural areas, particularly among younger children. 89. In summary, being a young, urban girl increases the probability of school attendance. By contrast, being an older, rural boy decreases the probability of school attendance. During the 1990s, the populations less likely to attend school iolder, rural boys) showed the highest improvement in school attendance rates. Equity and Incidence Analysis 90. Public spending patterns by educational level, student fees by educational level in the public sector, and investment per student at each level are shown in Graphs 17 to 20. They indicate that: * In the 1980-1997 period, higher education received the same level of:resources as basic education; * Enrollment at the basic education level was, on average, six times higher than enrollment at the higher education level; * Per capita student expenditures on higher education were, on average, six times higher than for a student at the basic education level; * Actual investment per student at the higher education level was about 15,000 Bolivars at 1984 prices between 1984 and 1997, although with cyclical variations over time. However, the average per capita investment has been higher in the last five-year period; - 43 - 0 0 17 ........... CD CD CL r, 1384 ............ .......... ................. CD 13 p ................. S2. ............. .. z CD 13 CD CD ('D cn .....I...I.... n CD 0 0 ........... ............. 00 .................... w 4) - - I - I . ........ ....... . :rl i 133- CD CD CD CD CD 0 Cl 0 C, 0) C, C> C> CD C CD 0 Cl .......... .... ....................... '.,, 0 CD ..................... CL ..... .............. ............ ...................... CD CD . ................... 00 CD .................... CD 0 p CD -4 CL co ............ cn OD C, ....................... .................... kV LI tz El ...................... 0 co .............I ...... p r-- 00 cn No .................... un --r- Table 15. Use of Public Education by Income Class (percent) Income Class Basic Secondary Higher University Upper 2.7 2.3 9.6 18.0 Middle-Upper 8.6 8.8 21.9 26.6 Middle 35.4 40.4 39.0 36.3 Lower 53.3 48.4 29.4 19.2 Total 100.0 100.0 100.0 100.0 92. Together with the budgetary distribution by educational level, this makes it possible to calculate the distribution of the Venezuelan educational budget by income class (see Table 16). Table 16. Public Spending on Education is Regressive (percent) Low Lower Middle Middle Upper Middle and Upper Distribution of population by income class 38.2 42.3 14.2 5.3 Distribution of educational budget by income class 38.0 36.3 16.6 9.2 93. In considering the population distribution in the country according to income class it may be concluded that (i) the lower income class received public funds for education in a similar proportion to its representation in overall population; (ii) the lower middle income class received public funds for education in a proportion lower than its representation in the overall population; and (iii) the middle, upper middle and upper classes received public funds for education in a proportion higher than their representation in the overall population. This suggests that public investment in education in Venezuela is regressive. 94. Would a policy of paying tuition at the university level in public higher education institutions generate greater equality in investment in education? More than 55 percent of those attending official universities continue to be young people form the working and lower classes. Only 18 percent of enrollment in public universities corresponds to young people from the upper and upper middle classes. Unless a mechanism is developed that facilitates access for students from low income families, a strategy that calls only for charging tuition for university studies may not constitute a "progressive" policy. Venezuela, nevertheless, remains one of the few countries that does not charge tuition at public universities,5 and the 1999 Constitution ensures that this will continue at the undergraduate levels. There is, however, a large private university sector in Venezuela capturing about 30 percent of university enrollment. This suggests that there would be For example, university income derived from tuition charges represents about 10percent in Colombian universities and about 36 percent in Chilean institutions. The Venezuelani Constitution establishes the right of the citizens and the responsibility of the government to ensure free access through undergraduate levels in the public system. - 45 - scope for charging tuition fees at the higher education level to generate new resources for expanding the supply and improving the quality of higher education services, complemented by a policy of loans and scholarships to ensure access by the poor. Efficiency Issues 95. Basic and High School Education. The repetition rates observed during the 1980s and 1990s show that an average of 9.6 percent of students from the first to eleventh year repeats at least once. The highest percentages of repetition are in the first (15.9 percent) and seventh (14.6 percent) grades.6 96. An additional symptom of the inefficiency of the system is dropout rates. The seventh grade had the highest rates with an of average 22.6 percent during the 1980s and 1990s, reaching a peak of 27.7 percent during the 1994-1995 school year. The highest dropout rate has taken place between the seventh and ninth grade with 17.6 percent, while the rate from the first to the sixth grade has reached an average of 6.1 percent. The increase in dropouts after the sixth grade could be partially explained by the fact that Venezuelans, mainly the relatively poor, have not yet fully adapted to the nine-year basic education system, and their willingness to keep their children in school beyond the sixth grade may not yet be sufficiently strong. For poor families, keeping children in school beyond the sixth grade could be associated with high opportunity costs. The patterns nevertheless show that families enroll their children in the seventh grade after they have completed the sixth grade and that the dropouts occur after the child has started. Together with the high rate of repetition in the seventh grade, this indicates that inefficiencies in the basic education system must be addressed. 97. Another indicator of the system's efficiency is the proportion of students that begin and finish their education without repeating grades. In general, for a hypothetical 1,000 students who entered the first grade in the 1995-1996 school year, about 48 percent would complete the sixth grade without repeating, about 23 percent would repeat the ninth grade, and about 16 percent would repeat the eleventh grade. The fact that there are many repeaters means that the system inefficiently uses more resources for every graduate. Indeed, during the 1980s and 1990s, the educational system needed an average of 7.7 years of resources per student finishing the sixth grade, 17.3 years per student finishing the ninth grade and 29.3 per student finishing the eleventh grade. The worst rates were again obtained in the 1994-1995 school year when 8.2 years of resources were needed to complete the sixth grade, 21.7 to complete the ninth grade, and 40.7 to complete the eleventh grade. 98. Higher Education. In Venezuela, higher education institutions have differences in internal efficiency depending on whether they are public or private. Regardless of whether they are full-time universities or junior colleges and institutes that provide 6 During the 1995-1996 school year, the figures for the first and seventh grades were 16.2 percent and 15.4 percent. Also, the highest rate of repetition during the 1980s took place in the first grade, with 18.3 percent during the 1989-1990 school year, while the seventh grade followed with 17.3 percent during the 1987-1988 school year. - 46 - short-term three-year training courses, internal efficiency differences can be analyzed by examining the proportion of students graduating within five years. Of the students that began studies at public universities in 1989, 40.5 percent graduated within five years. In the case of private universities, 52 percent graduated on time. The results did not vary during the 1989-1994 period during which public universities graduated 53.5 percent of the initial students enrolled, whereas private universities graduated 69.6 percent. In the case of non-university institutes, of those enrolled as new students in 1991, after three years 35.8 percent graduated from public institutions, while 45.6 percent graduated from private institutions. 99. In the periods reviewed, universities showed greater efficiency than institutes. Possible explanations include (i) the students who enroll in the institutes are usually from families with fewer resources than those enrolling in universities, a condition which could increase the dropout rate because of the relatively higher opportunity costs (of the foregone income) resulting from remaining in school; and (ii) those who enroll in institutes take courses at night and work during the day, which could lengthlen the time needed to graduate. 100. The ratio of students to professors is lower in public institutions than in private ones.7 This is true of universities and non-university institutes, regardless of whether or not the total number of faculty considered distinguishes among categories, and whether or not only full-time, tenured faculty are considered. In addition, the difference with regard to how many faculty are full-time or part-time has no bearing on the basic result-that there are fewer students per faculty in public higher education institutions. The same can be said for the number of administrative employees and workers per faculty, which in the case of public universities is triple that of private universities. As a result, the unit costs per student and per graduate in the late 1980s and early 1990s were different in public and private universities. At public universities, the unit cost per student was triple that in private universities, while the unit cost per graduate was nine times higher.8 While more analysis is required, there is sufficient evidence to generally conclude that there are significant inefficiencies in public higher education institutions. 7 The total number of students enrolled in higher education in 1996/1997 was 733,294, which represents about 3.2 percent of the total population. Total enrollment increased by 60 percent from 1986 to 1996, slightly more in universities (63 percent) than in institutes and colleges (54 percent). The 1987/1996 student enrollment growth index in public universities was 74 percent compared with 33 percent for private institutions. However, the analysis of new first- year student enrollments shows a smaller difference: 62 percent for public and 38 percent for private institutions. If this trends continues, private higher education institutions will soon enroll one half of total student population. 8 Given the inter- and intra-institutional differences (e.g., purpose of institutions, careers, majors, and type of students) efficiency indicators must be considered as indicative. For additional analysis see "La Educaci6n Superior en Venezuela: Contexto, Problemas y Opciones de Reforma" (Navarro, J.C.; IDB; 1995). See also, "Characteristics of the Higher Education and Research System in Venezuela" (World Bank, in process, March 2000). - 47 - Rates of Return 101. The evolution of the rate of return to schooling in the 1980s and 1990s, including the impact of economic cycles, is presented in Graph 21.9 102. While the evolution in the rate of return to education appears to be sensitive in the short-term to economic cycles, in the long-term (over the last two decades), the rate of return to education shows a clear downward trend, dropping from about 12.5 percent at the beginning of the 1980s to about 10 percent at the end of the 1990s. While this is in part a consequence of the country's dismal economic performance in the last Graph 21. Rate of Return is Declining decades-negatively affecting Percent (1980-1996) labor demand and real wages- 13 it might also be interpreted as meaning that investment in 12 education is less profitable | today than 20 years ago. It also ll confirms that education has an impact on equity outcomes- 10 \ wage differentials tend to 9_,_,_,_l_,_-_r___E________ narrow as the supply of 76 78 80 82 84 86 88 90 92 94 96 98 educated labor increases and schooling equality increases, Year other things being equal. 103. Private rates of return to investment in education at different levels display clear patterns. To calculate the different rates, the premium to investing in education at a particular level is estimated relative to incomplete secondary schooling, which becomes the point of reference for analyzing the rate of return of the other educational levels. Thus, it is found that in early 1980s: * Those who did not complete basic education earn 57 percent less than high school dropouts; * Those who finished basic education (but did not continue) earn 1 percent less than high school dropouts; * Those who finished high school (but did not continue) earn 60 percent more than high school dropouts; * Those who took technical courses but did not complete them earn 62 percent more than high school dropouts; 9 See Annex C for a specification of the model used for this analysis. - 48 - * Those who have completed technical studies earn 96 percent more than high school dropouts; * Those who began university studies but did not finish earn 80 percent more than high school dropouts; * Those who completed their university studies earn 126 percent more than high school dropouts; * Those who took teaching courses but did not complete them earn 93 percent more than high school dropouts; and * Those who completed teacher training earn 120 percent more than high school dropouts. 104. This analysis was repeated for other years (1985, 1988, 1990, 1992, 1993, 1994, 1995 and 1996) and the results show that through time, high school and higher education have become increasingly less profitable with respect to the point of reference (incomplete secondary), while the profitability of having completed basic education remains more or less fixed in relation to the point of reference. Only basic education became increasingly more profitable with respect to the point of reference.'0 These general findings imply more emphasis needs to be placed on basic education. 105. Table 17 shows the results of calculating internal rates of return, both private and social, for all levels.'" Rates of return to primary education tend to be higher than those for university education, but not for every year. In 1982, 1988 and 1993, the private rate of return to basic education was below the corresponding rate of return to university education. The same occurred with the social rate of return in 1993. However, a constant throughout the period is the differential between private and social rates of return between grade school and higher education. This differential usually tends to be lower in basic education than in higher education, implying that higher educat:ion may be over-subsidized. 10 These results are consistent with general findings in other studies, which indicate that investments in basic education have become more profitable over time, while high school and higher studies have become less profitable. X l The social rates of return to investments in education are distinguished from private returns by the fact that they include not only the "opportunity cost" to the individual of schooling, but also the direct resource cost expended by the public sector for education. These results are based on the long or "full" method described in Annex B. - 49 - Table 17. Education is Still a Good Investment (percent) Private 1980 1982 1985 1988 1990 1993 1996 Primary(vs. none) 19.1 15.5 17.3 14.8 17.7 14.7 18.9 Secondary (vs. primary) 12.6 16.5 17.5 11.0 8.4 9.7 12.2 Technical (vs. secondary) -- -- 17.3 19.7 14.5 19.5 25.0 University (vs. secondary) 17.0 16.4 17.2 16.4 14.0 18.0 14.8 Social 1980 1982 1985 1988 1990 1993 1996 Primary(vs.none) 15.3 11.9 11.9 11.7 12.8 9.0 12.2 Secondary (vs. primary) 11.0 11.8 14.1 9.9 7.7 8.2 10.1 Technical (vs. secondary) -- -- 13.6 16.8 11.0 11.8 9.6 University (vs. secondary) 11.0 10.7 12.2 13.2 9.8 11.3 6.9 Note: -- indicates that the data are not available. 106. The lowest rate of return occurs at the senior high school level. Both social and private returns are low, with the private rate of return to secondary schooling being particularly low. The low rate of return appears to explain why there is so much pressure by secondary school graduates on the higher education system. 107. The private rates of return to technological education are very high, surpassing the rates for basic education in almost all years.'2 However, it should be noted that while the private rate of return is very high in technological areas, the social rates of return are not as high, again probably due to over-subsidization of technological education. 108. Although the returns have fallen over time with the increase in the overall schooling level (the supply of educated workers), returns have not fallen much-which is consistent with theory and evidence from other countries. In fact, while the overall rate of return went from about 13 to 10 percent between 1980 and 1996, the overall average rate of return is a high 11.1 percent. The average for the 1980s is 11.7 percent, and the average for the 1990s is 10.2 percent-still relatively high and making for good private and social returns on investments. Average years of schooling increased from 5 to 8 between 1980 and 1996-97. Thus an increase of 3 years of average schooling has led to decrease of less than 3 percentage points in the returns to schooling. To put it differently, a 60 percent increase in schooling has led to only a 22 percent decrease in the rate of return to schooling-implying that schooling could be expanded and still be profitable for some time in the future. 109. The rate of return evidence points to the need to spend more public resources at the basic education level and to target the poor in order to address questions of inequality. High private returns at the university and technological fields suggest the possibility of introducing some cost-recovery schemes for these levels while implementing adequate schemes to protect access by the poor. 12 This finding coincides with the results of calculations estimated with Mincerian earnings finctions. - 50 - Key Education Policies 110. Basic Education Policy. In the mid-1990, the Government announce,d an action plan to overcome the "crisis in education" borne out by high rates of repetition and dropouts, poor reading and mathematics skills, and deterioration of values with respect to work and society. Most of the attention regarding the negative results of the educational system was concentrated at the basic education level and as such the reforms focused on this level. The major objectives were to (i) improve the quality of education, (ii) improve the efficiency of delivering public education services; and (iii) correct social inequalities in the school system. To achieve these objectives, key measures were implemented during the 1997-1998 period, including (i) simplification of the first phase of basic education by reducing the number of subjects taught from an average of 11 to 6; (ii) concentration of the pedagogical effort on developing skills in which the performance of students in the basic education level was the poorest (e.g., reading and mathematics, work and ethical values); (iii) decentralization of responsibilities to the States, including of preparing 20 percent of the curriculum based on regional and local needs; and (iv) inclusion of classroom projects. 111. Moreover, in 1999 the Government launched a revolutionary integrated program (the Bolivarian School program) covering a comprehensive curriculum, including mathematics, social sciences, music, sports and theater, among other subjects, and providing daily school lunches and food to students. The main objectives are to (i) integrate the school system and community through the school and learning process; (ii) stimulate the transformation of education; (iii) provide general attention to all school- aged boys and girls; (iv) promote a participatory learning process; and (v) strengthen national and ethical values. While it is too soon to evaluate the impact of this program, the Government has remained committed to continue allocating resources towards the overall objectives of improving quality, efficiency, and promoting social equity. 112. To improve the quality of public education, teacher training and provision of educational materials (including books) have been key components of the strategy.13 As part of this effort, Regional Support Centers for Teachers were established. These Centers are institutions for ongoing training of teachers in active service, offering courses in language, mathematics, and social and natural sciences. The Centers also provide training in the classrooms and help the students themselves. By the end of 1997, five States had Centers and more were under construction in several other States. Another program has been the "3 x 1" program through which the Ministry of Eclucation has contributed 3 Bolivars for each Bolivar offered by regional governments to provide educational materials and refurbish school infrastructure. 113. To complement the above programs, the infrastructure program has focused on repairing schools. Under the system of double shifts, demographic pressure on schools 13 Teacher training programs and provision of educational materials, including books have been financed using resources of the IDB, World Bank, and the central and regional governments. By end of 1997, these program trained about 70,000 teachers and about 2,500 supervisors, and about 20 million books either had been purchased or were in the process of being purchased. - 51 - has not increased and therefore the need for new schools has not grown, given the lower rate of population growth of the last 15 years. The ultimate objective was to strengthen the institutional capacity of regional educational offices in light of the gradual decentralization of the administration of the national education system (see below). The maintenance of the refurbished facilities would be covered jointly by the regional offices and the schools that would be in charge of processing resources for financing such maintenance. 114. As part of the efforts to improve quality and efficiency, a decentralization strategy was completed in 1996, which included the "Budget for the Reorganization and Decentralization of the Educational System", a three-phase plan for decentralizing the management of the educational system. The objective of the first or preparatory phase was to gradually train administrative personnel to take charge of the education system at the State level. The second or co-managed phase would create a combined commission of national and State representatives to prepare the transfer of the administration of education to the States, which would require an audit of the regional educational systems. By end-1997, commissions of this type had been established in 17 States. An additional step in this phase was the appointment by mutual agreement of the Secretary of Education and the State Governor of a sole authority that will concentrate the decision- making capability for the educational zone as well as the education office or Secretariat. In 1997, 5 of the 17 States already had this authority. The last or transfer stage would have the Senate approve the final agreement for the transfer of administration of education to each State. The schedule for such transfer was to be established by a top level technical commission for decentralization, and the Ministry of Education would provide support for the new State educational administration office. By January 1998, the Senate had approved the transfer agreements for two States. 115. As it has evolved, the educational decentralization process has been slow given its political and administrative complexity. Since the decentralization of any dependency should be the product of an initiative of regional governments, the process has not been homogeneous. The differences in technical capacities of the State governments and in their preferences to assume one or another competency, influenced the rate of educational decentralization. It is noticeable that the poorer States are the ones that are the farthest behind in the process, which can be a reflection of the technical deficiencies in their education secretariats.14 Nevertheless, the decentralization strategy will have to be assessed in the context of the 1999 Constitution, the corresponding legislation to regulate and implement the 1999 Constitution and the strategic decision regarding management and provision of educational services envisaged by the Government. 14 One of the most difficult problems to solve in decentralization negotiations and which makes State governments reluctant to assume the transfer of the educational service is the problem of the accrued labor liability. In addition to the financial burden this problem poses, the fact renains that, since the capacity of the State governments to generate their own income is limited and since their financial dependence on constitutional subsidies is high, the financial commitments of the magnitude of such a labor liability would seriously compromise regional public finances. - 52 - 116. To improve social equity, particularly since mid-1996, social programs (Day Care, Food Subsidy, In-school Food, School Uniforms, and Breakfast, Snacks and School Cafeteria Programs) linked to the network of public schools or administered by the Ministry of Education have constituted an essential component of the overall Social Protection Program. While certainly not a completely homogeneous group these key programs have remarkable similarities with regard to the targeted population, that is, school age children enrolled in a formal school program (with some variation according to the specific program). However, the similarities of the programs may also indicate a significant degree of duplication. A major feature of these programs, with the exception of the food subsidy program, is that they are mainly in-kind transfers. This may be increasing unnecessarily administrative costs. Despite the efforts to include children in poor areas based on geographic targeting, poor children not enrolled in school are not covered by the majority of these programs (see Chapter III, Section IV for more details). 117. Higher Education Policy.'5 The Government's policy toward higher education in general and universities in particular was implemented in February 1997 in what was known as the "New Deal" between universities and the Government. A focal point of discussion revolves around the distribution of the education budget and its impact on society. Other points discussed were the burden of university budgets on public finance and the efficiency and efficacy of expenditures on higher education. 118. In the 1980s and 1990s, university education absorbed about 38 percent of the budget of the Ministry of Education, while only 22 percent has gone to preschool and basic education, although 70 percent of the students in the educational system are in these lower educational levels. In view of this situation and the burden that university education represents to the Treasury, the Ministry of Education has encouraged universities to develop other funding sources. The most controversial point is with regard to students contributing towards the cost of their education. 119. The Government's interest has not only been directed at sources of university financing, but also at the manner in which resources are distributed among and within the universities. Another important point of discussion has been the automatic raise in salaries of university professors officially implemented at the end of the 1960s. The authorities have established the goal of annulling this system and replacing it. with merit raises based on the number of papers published, courses taught or other inclicators. In order to accomplish this goal, in 1997 the Academic Performance Fund was created to establish mechanisms for financing and acknowledging academic excellence and performance. The National Commission for the Development of Higher Education was created to reform the sector and to create a national system to evaluate academic performance, as well as to carry out studies to examine the efficiency of expenditures on higher education. This Section does not include the Government strategy regarding science and technology, and research and development programs. It is, however, well-known that a key objective of the Government, including the newly created Ministry of Science and Technology, is to improve the science and technology system, both in quality and quantity, to contribute to accelerating human capital accumulation, with the concomitant impact on social and economic developmwent. - 53 - 120. While most of the above issues affecting higher education remain valid, the policies are being reassessed in the context of the 1999 Constitution and the strategic vision of the authorities regarding higher education. The Government is committed to improve the quality and the equity of the university system, both private and public, and is elaborating a comprehensive reform program. Specific objectives of the program include to (i) improve the quality and efficiency of the public university system; (ii) improve the equity and access to universities, and create the conditions to facilitate access by the poor; and (iii) increase institutional capacity to communicate and coordinate within the educational system and with other institution of higher education. Conclusions and Recommendations 121. The proportion of the labor force completing basic education continues to increase over time. The relative increase in school attendance between 1990 and 1995 was higher among those populations classified as having the lowest probability of attending (older rural boys) than among those with a higher probability of attending (younger urban girls). These findings suggest that, in general, education policy and programs reached their target populations. Despite efforts to invest in basic education, the quality of basic education remains below international standards. Significant inefficiencies and inequities remain in the school system, and the incidence of public investment in education continues to be regressive. The lower class continues to receive a smaller proportion of public spending in education than their representation in the total population. 122. Internal efficiency at the grade school and high school levels is very low and shows no signs of improving. The dropout rate in the first six grades has been on the rise since 1989, despite the targeting of this group of students by social programs. Although it may sound contradictory, it is possible that the dropout rate is positively correlated with increased attendance because those who attend school exclusively to benefit from the social programs are also those with a higher propensity to drop out. 123. The dropout rate has also been on the upswing. The increase in net coverage of enrollment corresponding to grades 7th to 9th may suggest that the increasing dropout rates may also have their origin in the growing presence at this educational level of non- traditional populations that in turn are more prone to fail in school. 124. The internal efficiency of higher education, measured as a proportion of the population that graduates in the prescribed time frame, is higher at universities than in institutes and junior colleges and higher in private than in public institutions. There is the hypothesis that low internal efficiency observed in institutes and junior colleges is related to the strong presence of young people with fewer resources. 125. Analyses of external efficiency demonstrate that the returns to schooling have been declining at all levels over the last 15 years, which partly reflects the country's dismal economic performance. The private rate of return to basic education does not - 54 - seem to differ much from the university rate of return. Nonetheless, the social rate of return of basic education usually although not always outstrips the social rate of return of university education. The rates of return for high school are generally lower. These factors could in part explain the strong pressure that persists by part of the population to continue to expand the higher education system. Nevertheless, rates of return on education remain relatively high and investment in education is economically and socially profitable. 126. Given the above conclusions, selected policy recommendations emerge, including: * Continue to implement the reform at the primary education level ancl monitor the impact on repetition and dropouts; * Accelerate the development of the new curriculum for the third phase of basic education in an effort to decrease the very high dropout rates observed in those grades for which the reason for dropping out appears to be failure by students; * Intensify the effort to increase investments in lower levels to improve the quality of basic education and make investment in education more progressive; * Given that the social programs appear to have contributed to increases in school attendance but have had little effect on reducing repetition and dropout rates, more attention should be paid to children from families with low incomes; * To increase internal efficiency in the public university system, new financing mechanisms with appropiate incentives that generate greater spending efficiencies should be considered; and * Given that university education still has high private rates of return and that there is a rapid expansion student enrollment and greater efficiency in private universities, the Government should consider further facilitating the expansion of private universities through strengthened scholarship programs for students from families with low incomes. II. Investing in Health and Nutrition Services' 127. This Section presents a brief analysis of the health sector, start.ing with an overview of the provision of services, financing and health status. It also examines aspects of equity in terms of provision and financing, analyzes efficiency in the sector and discusses key reforms over the last ten years. The analysis reveals that the performance of the health sector in Venezuela has suffered from (i) a large inequity in terms of access to health services and distribution of resources; and (ii) a low level of efficiency and lack of coordination among the institutions providing health services. This Section draws on a background paper prepared by Gabrielena AlcalA. - 55 - These issues need to be addressed to prevent further deterioration of health services, affecting negatively human capital accumulation, labor productivity and constraining social and economic development.'6 Sector Overview 128. Health care services in Venezuela are supplied by public, private and mixed social security institutions. Key institutions are the Ministry of Health and Social Development (MH), the Social Security Institute (IVSS), the Institute of Social Welfare of the Armed Forces and the General Direction of the Govermnent of the Federal District."7 The three models of health care provision are (i) the open "public" model of free access provided by the MH and financed by the Government; (ii) the "mixed" closed model in which health services are provided to affiliated workers and financed jointly by the employer and the worker, dominated by IVSS which covers about 25 percent of the population; and (iii) the "private provision" model provided by the private sector to those able to pay for the services. Overall, the health system is dominated by the public sector (e.g., 4,321 hospitals compared to 344 hospitals operated by the private sector), with private services serving high-income households in large population centers (see Annex D, Table 1). 129. In 1997 the average number of beds was estimated at 2.2 per 1,000-lower than the average for Latin America of 2.6 beds per 1,000 (see Annex D, Tables 2 and 3).l8 The MH has an important share of both the network of health care establishments and the installed capacity of health institutions. In terms of available beds, MH accounts for about 50 percent, compared to 12 percent for the IVSS, and 23 percent for the private sector. 130. Over the long-tern, total public spending on health as a percentage of the total national budget and as a percentage of GDP has varied considerably, with both declining over the last two decades (see Graph 22). Public spending on health as a percentage of GDP declined from about 3.1 to 1.5 percent between 1970 and 2000. Nevertheless, Venezuela's expenditures on health services rank relatively well when compared to other Latin American countries (see Box 7). 16 The extent and depth of this analysis is constrained by the lack of systematic and reliable data about the health sector in Venezuela. For example, there is only limited information on health financing, utilization, productivity and unit cost. Moreover, there is no information for a systematic demand analysis of health services. Despite the limitations, the Section uses available information from various sources to identify selected critical issues affecting the sector, to shed selected light on the issues, and to provide additional information for policy makers. The health system could greatly benefit from the collection and systematization of information that would help policy-makers evaluate the system and adopt better-informed decisions. 1 7 The Ministry of Health and Social Development resulted from the consolidation of the Ministry of Health and Social Welfare (MSAS) and the Ministry of Family Affairs (MF) as part of the public sector reforms implemented through the Ley Organica de la Administracion Central in 1999. 18 The 1999 World Development Indicators estimate the number of hospitals beds in Venezuela at 2.6 per 1,000. Source: "World Health Report", World Health Organization, 1999, Geneva. - 56 - Box 7. Health Expenditure in Venezuela Compared to Selected Latin American Countries Despite decreases in health expenditures, Venezuela's total expenditures on health are comparable to other Latin America countries of similar development levels. As a percentage of GDP, Venezuela spends around the same as Colombia, less than Costa Rica and Argentina, but more than Chile. In the case of per capita health expenditure, Venezuela spends only more than Colombia. Health Expenditures (1995) Total Public Sector Public Sector (% of GDP) (% of GDP) (% of total) The Americas 8.6 4.1 47 Argentina 9.7 4.3 44 Brazil 4.6 1.8 40 Chile 6.2 2.5 40 Colombia 7.4 2.9 40 Costa Rica 8.5 6.3 74 Venezuela 7.5 3.0 40 Source: World Health Health Expenditure, per capita, US$ 1990-1997 Colombia 140 Venezuela 213 Costa Rica 215 Chile 295 Brazil 351 Argentina 799 Source: WDI, 1999 Graph 22. Share of Public Ependitures in Health in the Budget andGDP (1970-2000) 20 4 15 --3 Z 102 0 - ~~~~~~~~~~~~~~0 CZ cz I)~ (N I I 00 I 0. N- CC CC 01 0~ O~ 0~ O(O 0~ a, a C> o~~~~~~~o o Shareas%ofNationalBudget - Shareas%ofGDP - 57 - 131. Over the past few years Table 18. Main Health Indicators, 1998 Venezuela has experienced both demographic and epidemiological Life expectancy at birth 72.9 changes, which have changed the Male 70.1 disease profile. '9 Some health Female 75.8 indicators have remained stable, Total fertility rate 3.0 Estimated crude birth rate 24.7 and other he improved ove Infant mortality rate 22.0 Under 5 mortality rate 25.0 has decreased and life expectancy Maternal morality rate 56.0 has increased (see Table 18 and Source: Core Health Data for the Americas Graph 23). Nevertheless, the (PAHO), 1998. Washington D.C. causes of death have changed- cardiovascular diseases, cancer and injuries due to accidents and violence, have become the leading causes of death. Moreover, other diseases have emerged such as malaria and HIV/AIDS. 132. Despite longer term improvements, some health indicators have slightly worsened in recent years. For example, although the maternal and infant mortality rates have decreased in G-4 h23 Evolutiofamneath Incators the last decades, they increased (1989-98) during the first half of the 1990s before again declining. There has also been little variation in 8Mta0 rnTtaityrawe the crude mortality rate, i.e., 4.4 700 per 1,000 in 1992 compared to 60 - 4.2 in 1995.20 The total fertility 50 rate has been declining, 40 gradually reaching 3 in 1998. 30 - _ _ __ Fertility rates, however, remain 20 high among women in their 20s 10 Lnt nrtalitv rate and illiterate rural women (with 10 an average of 8 children). In 00 Although some health indicators in Venezuela compare well with those of other countries in the region, the infant and maternal mortality rates remain higher than in some countries with lower per-capita income than Venezuela (see Graph 24). '9 Sources: MSAS, 1997 and "Core Health Data for the Americas" (Pan-American Health Organization, 1998). 20 The Latin American Demographic Center estimates under-reporting of deaths at 13.2 percent, which would make the actual average mortality rate 5.4 per 1,000 for the period 1990-1995. - 58 - Graph 24. Infant and Maternal Mortality Rates and GNP Per Capita for V'enezuela and Selected Latin American Countries, 1998 50- .Ecu ~ ~~--- -- --- ---- - -----._ ... _ } 300 45- Peru cs 250 * 40 _-_Hond____ * Hond DomRep _ 200 3025- laniaica Argentina8 5 *Eu *Anris IC 25-10riZ i E 20 - * *Venezuela l 1 5 * Umg e 10 * Costa Rica * Chile Chile - 0 - E: 50 r (R 5 O 0 2000 4000 6000 8000 10000 0 2,000 4,000 6,(00 8,000 10,000 GNP per capita 1998 (USS) GNP per capita 1998 (US$) 133. In indicators such as immunizations and maternal health coverage, Venezuela has performed below average (see Graphs 25 and 26).21 Graph 25: Immunization Coverage in Veneuela and Selected Graph 26: Maternal Health Coverage in Venezuela and Latin American Countries-Ranked by GNP per Capita, 1997 Selected Latin American Countries-Ranked by GNP per Capita, 1997 12012 - 1090943 90 f.U 1. 9477.5 95 88 995.5 97 96 60 -~~~~~~~~~~~~~~~~~~~~~~8 60 - ~~~~~ ~ ~~~~~~~~~60 --56 40 40 20 -lIII-. 20 - ~~~~ 4r~~~~~~~~~~ '~ ~ ~ ~ ~ ~ 0 I # ~ Incidence Analysis 134. There are, of course, differences in access to health services and status across the population.22 Public facilities and skilled personnel are distributed unequally throughout 21 These indexes are from "Evaluaci6n del Sistema Intergubernamental de Salud de Venezuela (1990-1996): Una Aproximaci6n Inicial" (Gonzalez R. M., 1997, Caracas). In the case of immunization, the index is an average of the percentage of children immunized against measles, polio and DPT. For maternal health coverage, the index is an average of the percentage of women with pre-natal care programs and the percentage of women attended during childbirth. 22 A desirable scenario to analyze equity issues would be to have information on the share of the poor and non-poor that benefit from public expenditures, and compare these with their shares in the population to see if expenditures are regressive or progressive. Unfortunately, there is neither - 59- the country, with a large proportion of the facilities and personnel concentrated in the Federal District metropolitan area. Particularly striking is the concentration of physicians, as the Federal District has more than 50 per 10,000 inhabitants while Miranda only has 8 per 10,000 inhabitants (see Annex D, Table 5). The distribution of the budget also shows signs of inequity, and public resources not being allocated based on poverty criteria. Graphs 27 and 28 compare the per capita State health budget, a poverty index and a geographical dispersion index.23 Several States with a high poverty level receive low levels of per capita budget, as do several States with high geographical dispersion. Graph 27. Distribution of the MH Budget to the Federal Entities Related to an Index of Poverty, 1996 High budget/low index High budget/high index ,8 0 0 0 Tu > 6000 M f Il 50 0 0 O >Me Suc ' 4 000 0 3 4 Lar 3i 3000 jtjZul 'V *DF- NE* X10 00 * DFA *rBa5p 0 0 0.1 0.2 0.3 0.4 0.5 Index of poverty Low budget/low index Low budget/high index Souirce: MSAS. 1997 sufficient nor reliable information on health financing to perform this type of incidence analysis. Analysis performed with the available data indicate that inequity problems exist. 23 The criteria for high/low index of poverty, geographical dispersion and budget distribution are relative and estimated based on the average for the 22 States. - 60 - Graph 28. Distribution of the MH Budget to Federal Entities Related to an Index of Geographical Dispersion, 1996 High budget/low index High budget/high index 8000 - 7000 - * T 6000 - ~5000-An esAra BoI ~Tac 1W4000 -*L Car ___ _ Fa1 0 0. 0. 0. .4 05 . m 3000 Mrul Co 200 Guar *+4ar Ana 1000 * Apu 0 0 0.1 0.2 0.3 0.4 0.5 0.6 Index of geographical dispersion Low budget/low index Low budget/high index Source: MSAS, 1997 Efficiency Issues25 135. In 1997, about 80 percent of the MH budget was allocated to salaries, about 15 percent to materials, services and repair services, and less than 1 percent to medical and non-medical equipment, construction and maintenance. In the case of the IVSS, in ambulatory care about 70 percent of the budget was allocated to salaries, about 12 percent to materials and inputs, about 10 percent to emergencies, about 5 percent to transfers, and about 3 percent to equipment and investments. Over the last few years these percentages have not changed significantly. Personnel expenditures crowd out the purchases of other inputs, leaving little for maintenance, equipment, drugs and other supplies. 136. The MH budget allocates about 70 percent to hospital care which is intrinsically costly, about 12 percent to arnbulatory care, about 6 percent to environmental sanitation, 24 The analysis of efficiency is also limited by the lack of information on utilization, productivity and unit cost. Nevertheless, several indicators highlight the low level of efficiency in the provision of health services. 25 The analysis of efficiency is also limited by the lack of information on utilization, productivity and unit cost. Nevertheless, several indicators highlight the low level of efficiency in the provision of health services. - 61 - about 11 percent to epidemiological research and the rest to social promotion. This has limited the allocation of resources to highly cost-effective intervention and preventive programs, such as immunizations, maternal-child care, and health education. 137. Type I and II hospitals of the MH have an occupancy rate of 30 percent, while Type III and IV hospitals have about 70 percent.26 This indicates that hospitals of greater complexity have more demand than less complex ones providing preventive medicine services, which are thus under-utilized. Also, because hours are short in both ambulatory facilities and hospital outpatient departments, patients tend to seek care from emergency rooms. This introduces an element of inefficiency and ineffectiveness which greatly Table 19. Number of Medical Visits in IVSS Institutions by Type of Services 1993 1997 Number (%) Number (%) General Medicine 1,617,290 22.1 666,029 20.5 Specialist 1,918,023 26.2 838,457 25.8 Pediatricians 763,620 10.4 290,962 9.0 Emergencies 2,246,171 30.7 1,167,405 36.0 Dentist 777,796 10.6 84,276 8.8 Total 7,322,796 100.0 3,247,129 100.1 Source: OCEI, 1999 increases unit costs and overloads hospital emergency services. In IVSS institutions, consultations for emergency care are greater than those that entail preventive care (see Table 19). Some other efficiency measures such as the average length of stay indicate low efficiency levels in MH hospitals.27 138. The current system for public budgeting and provider payment does not encourage efficient allocation of resources or internal efficiency in the delivery of services. Hospital and ambulatory services receive budgets based on prior year expenditures rather than on their performance in terms of volume, quality of services or impact on health status. While data do not permit a complete analysis of resource allocation, it can be inferred that many cost-effective health interventions (childhood immunizations, integrated programs of prenatal and delivery care, improved domestic 26 The public sector hospitals are classified as Type 1, II, II, and IV, based on a scale of increasing complexity. Type III and IV hospitals also carry out university and postgraduate training. The low occupancy rate of Type II hospitals is a matter of concern. For example, an 80 percent occupancy rate is often perceived as a threshold of concern in the US (see "Peru: Improving Health Care for the Poor" (World Bank, Report No. 18549-PE, May 1999). The occupancy rate in Type 11 hospitals in Chile is almost 70 percent and in Costa Rica about 90 percent (Pan American Health Organization, 1998). 27 Comparisons must be taken only as indicative because the size of these indicators are affected by the different definitions used by each country-e.g., some measures refer only to acute care facilities. For example, in 1999, the average length of stay was estimated at about 7 days. This can be compared to about 4 days in Type 11 hospitals in Costa Rica and to about 6 days in Chile. - 62 - hygiene and infant feeding practices) have been neglected. While many of these programs exist, epidemiological and program data indicate that they need to be expanded and strengthened. 139. With respect to human resources, there is a distortion in the health manpower pyramid. There is a high proportion of specialists relative to general physicians, a high proportion of physicians relative to nurses, and a high proportion of lower level, untrained manpower relative to nurses and other paramedical staff.28 The fact that there are insufficient numbers of middle-level personnel implies that physicians (who are more costly) perform technician or nursing level tasks.29 140. The low level of hospital autonomy is also a constraint. Hospital directors do not have the sufficient budgetary and administrative decision-making authority to establish or maintain quality and efficiency. Hospitals are not given the responsibility for managing expenses related to the purchase of equipment, utilities, and maintenance contracts. With few budgetary controls, hospital personnel have no incentives to manage resources efficiently. Furthermore, hospital directors generally only work part-time because of low salaries, dividing their time between managing the hospital arnd practicing medicine. Key Health and Nutrition Policies 141. Given the new epidemiological profile, and the inequity, low efficiency and low quality in the provision of health services, the country embarked on a series of reforms to improve the effectiveness and efficiency of the sector. In the mid 1990s, the Government introduced the Ninth National Plan, an overall plan which included key objectives for the health sector, such as: * Strengthen the right to health and equity, and to combat inequalities and social inequities with regard to health, disease, and access to services; * Improve the efficiency and effectiveness of the health services system and to give outpatient services units the authority to make decisions; * Assign special priority to activities for health promotion, and damage and risk prevention, thus strengthening primary care and the outpatient network; * Reassert the role of the Government in developing health services and to democratize the health structure, with broad societal participation; and 28 International standards suggest that an appropriate personnel distribution in the sector should be about 25 percent for managers/administrative support personnel and lower level workers, and 75 percent for professional and technicians. In the MH, about 55 percent of the budget is being spent on lower level workers. 29 In absolute number, the number of physicians per 100,000 is high compared to other LAC countries (see Table D4). The rate of physicians to nurses in Venezuela is similar to the ones in other LAC countries, but is about 4-6 times higher than in some OECD and Asian countries. See "Venezuela Health Hector Review" (World Bank, Report No. 10713-VE, Volume [, March 1993). - 63 - * Ensure the role of the MH in determining policies, managing, coordinating, and regulating the health sector. 142. To complement these objectives, as part the Government's Box 8. Main Lines of Action of the Health economic stabilization and social ReformProgram program implemented in 1996 (see * Implement a universal health system Chapter II), public finance reforms. guidted the, institutinal ce reformsf te * Give priority to preventive health care guided the institutional reform of the MH and the reduction of public * Coordinate budget distribution according to expenditures in the health sector. The demand reform affected the health sector a Promote civil society participation in the provision through actions taken to reduce the of health services bureaucratic structure of the MH and . Induce the development of the necessary human IVSS as well as the beginning of a resources discussion on cost-recovery issues. . Increase health sector investment Moreover, the reform tried to accelerate privatization, allowing some . Develop regulatory capacity at the central level forms of privatization in the health . Establish common systems of planning, evaluation sector (such as transferring the and information across the States administration and/or management of * Use common methods to approve fund transfers to some institutions to private providers). the States * Separate the financing role from the provision of 143. In the 1990s, the Government health services role also introduced a program (La Reforma * Establish methods for recording costs del Sector Salud) to improve the health system. It included health sector modernization and health services reform.30 The main objectives were to (i) improve the quality of the health system; and (ii) improve the equity, effectiveness and efficiency of the management of the system (see Box 8). 144. In addition to the decentralization of the sector, the introduction of draft legislation for the Comprehensive Social Security System (LOSSSI) in the late 1990s constituted an important initiative to complement the overall health sector reform (see Box 9). However, all reforms including decisions on the type and coverage of social security, remain on hold pending debates and issuing corresponding legislation and regulations to implement the general principles established in the 1999 Constitution (see Box 10). 30 These programs are contained in the Proyecto Salud, supported by The Inter-American Development Bank and the World Bank. - 64- Box 9. The Proposed Social Security System of 1997 A proposed law for the Comprehensive Social Security System (LOSSSI) was introduced in March 1997, covering benefits from medical care and the use of free time to pensions for retired persons. The aim of this proposed legislation was to integrate social assistance, health and social security under the management of one entity. This new system had five sub-systems: health, pensions, professional training and forced unemployment, housing, and recreation. LOSSSI established the basic principles of the social security system, leaving the specific aspects to the particular laws of each subsystem. The main characteristics of the health subsystem were contained in the Proyecto de Ley del Subsistema de Salud (PLSS). The main proposals of this project were: - Separation of the provision, financing and regulation functions of the IVSS - Establishment of a Health Fund to which employers and workers will contribute - Granting the insured person freedom to choose the institution that will provide him/her with health care - Creation of institutions to manage the new Health Fund (Administradoras de los Fondos de Salud) - Establishment of a basic health package for workers affiliated with the compulsory fund - Introduction of new forms of management for hospital and clinics of the IVSS The LOSSSI established a transition period until the creation of the new health subsystem. The proposed contribution rate was 6.25 percent of the average salary during the transition period and the treasuly would cover the administrative costs and any other expenditure necessary to guarantee the current health care coverage under the IVSS. A study reviewed the design and the fiscal impacts in the short run. The study estimated that the proposed reform could imply a significant increase in the rate of contributions from its participants or greater subsidies from the state to cover for potential deficits. For example, in 1999, the income to the sub-system, corresponding to the 6.25 percent contribution rate, would have been about Bs. 274,574 million (US$442 million) and the estimated expenditures Bs. 545,627 million (US$880 million). This would imply a deficit of about Bs. 271,053 million (US$437 million), or approximately 0.55 percent of GDP. A contribution rate of about 12.4 percent is required for the system to break even. In previous years, the deficits incurred by the IVSS were financed with cross subsidies from the pension system and from general subsidies from the central government. However, with the new proposed LOSSSI and PLSS, the accounts of the systems should be kept separated, which means that deficits in the health sub-system could not be financed with resources of other subsystems. Note: The expenditures above only refer to only recurrent costs, but there are also one-time costs, including the social benefits to be paid to those personnel that will be removed from their functions, and other non-labor costs. - 65 - 145. In 1989, Congress approved the Ley Organica de Box 10. Principles of the Health and Social Descentralizaci6n, Transferencia y Security System Delimitaci6n de Competencias del Poder Publico. This law transferred The 1999 Constitution includes four major principles the competencies and services of with regard to health services (i) free health provision; (ii) financial provision guaranteed; (iii) universality; and certain sectors (including health) to (iv) community participation. the States. With this process, the MH would have been responsible for The 1999 Constitution stipulates that every Venezuelan sector policies and regulations and has the right of access to the social security system and ceased to perform some operational that the Government has the obligation and responsibility of ensuring this right, by creating a system functions, which would have been that will be universal, integral, and of participatory transferred to the States and financing (with direct or indirect contribution). The lack municipalities.31 As such, public of contribution capacity will not be a reason for resources assigned to the MH and the excluding somebody from the system. Furthermore, the IVSS would have decreased, and the financial resources of the social security system would alloatio to Sabe administered only for social purposes under the allocation to State governments direction of the Government. increased. However, implementation of this law was uneven and with mixed results. 146. A study was recently carried out to analyze the effect of the decentralization process in the performance of the health sector.32 In one of the study's findings, there was no difference in mortality rates before and after the decentralization-mortality rates did not change as a result of the decentralization process. There were, however, four cases in which the mortality rates increased in the decentralization period: maternal mortality in the State of Bolivar and infant, neonatal and maternal mortality rates in the State of Nueva Esparta. While the study did not establish which factors affected the mortality rates in these States, some explanations are that (i) these States improved their registration procedures on the mortality rates and thus showed less under-reporting; (ii) health performance actually deteriorated; and (iii) other factors outside the health system negatively affected these indicators. 31 The first transfer of health services was approved in 1992 for 5 States (Anzoategui, Bolivar, Carabobo, Falc6n and Zulia), and thus far, 13 out of the 23 States have been given both the responsibility and resources for the provision of health services. The transfer of responsibilities include in 1993, Aragua; in 1994, Merida, Monagas, Sucre, Tachira and Trujillo; and in 1995, Lara and Trujillo. 32 In the study, maternal/child health care was analyzed in five decentralized States (Anzodtegui, Aragua, Bolivar, Carabobo, Falc6n) and compared with a control group of three non-decentralized States (Guarico, Nueva Esparta and Portuguesa). Two periods were analyzed: (i) the pre- decentralization period, 1960-1989; and (ii) the decentralization period, 1990-1995. This study is the first attempt to evaluate the impact of the decentralization of the health sector. One of the main conclusions is that in order to improve performance in the sector, it is necessary to improve data collection. See Gonzalez R. M., 1997. - 66 - 147. Moreover, to evaluate the performance of the decentralized States, the percentage variation between the years 1990 and 1996 was calculated for the following, indicators: prenatal, infant, pre-school, family planning, medical visits and immunization coverage. In the case of medical visits, the differences between decentralized and non-decentralized States were significant (see Table 20). Prenatal, infant and pre-school medical visits increased significantly in the decentralized States. The percentage variation in the non- decentralized States was negative in all the cases, indicating a decline in visits. With respect to immunization coverage, the decentralized States show a similar or greater increase in the coverage than the non-decentralized ones (except for BCG). Table 20. Percentage Variations in Coverage in Maternal-Child Health Care in Selected States, 1990-1996 Decentralized States Non-Decentralized States Anzoat.. ATagu| Bolivar Carab. FalcOn Aver j Giarico N. ERa Portuusa Avera Medical Visits Prenatal 28.8 23.9 -27.0 na na 8.6 2.0 -65.9 na -31.9 Infant 23.0 38.0 na na na 30.5 na -30.6 na -30.6 Preschool 9.8 42.7 na na na 26.2 na -20.7 na -20.7 Farnily -68.0 32.5 -15.7 na na -17.0 -16.0 -36.3 na -26.1 Planning _ Immunizations Polio na 3.3 na 24.1 na 13.7 -9.6 13.6 19.4 7.8 Measles na 6.2 na 6.7 na 6.4 -8.7 16.1 11.7 6.3 Triple na 1.1 na 7.0 na 4.1 -15.5 9.9 11.8 2.1 BCG na -13.4 I na 18.7 na 2.7 6.6 19.7 13.2 13.1 Notes: The percentage variations correspond to the differences between the years 1996 and 1990 except in the following cases (i) Anzoategui: prenatal coverage corresponds to 1992-96; (ii) Bolivar: prenatal coverage and family planning correspond to 1990-1995; (iii) Carabobo: BCG coverage corresponds to 1993-96; (iv) Nueva Esparta: prenatal coverage corresponds to 1990-95, infant and preschool coverage corresponds to 1990-95, infant and preschool coverage correspond to 1993-96 and BCG to 1991-96; and (v) Portuguesa: BCG coverage corresponds to 1991-95. Source: Gonzalez, R. M., 1997 148. While the data may suggest that the decentralization process has had a positive impact, one must be cautious in interpreting these results.33 Nevertheless, other evidence reinforces the hypothesis that decentralization in health has been somewhat successful (see Box 11). 33 The analysis did not include all States, and discrepancies were found in the data provided by the States and by the MH. Finally, the analysis does not explain the reason behind the changes (i.e., it is possible that the improvements were the results of factors other than the decentralization process). - 67 - Box 11. Some Signs of Success with the Decentralization Process Following decentralization the State of Lara created a public institution (Fundasalud) to channel Ministry of Health funds to eligible NGOs under contracts to manage public health services, including 60 ambulatory health centers, various hospitals, and milk distribution programs, among others. The NGOs administer public funds and serve to ensure quality of care and efficiency. An evaluation comparing Fundasalud ambulatory health centers with traditional State-managed ones showed (i) significantly better client perceptions of quality of medical, nursing, dental, laboratory and emergency services; (ii) significantly higher rating of the operation of the centers; and (iii) significantly greater demand for services resulting in lower costs per patient. Six health facilities Six health facilities Venezuela with civil participation without civil participation Average total cost per month Bs. 2,928,832 Bs. 2,336,833 Average number of patients per month 1,730 423 Average cost per patient Bs. 1,693 Bs. 5,524 Source: Improving Health Care for the Poor, 1999. Adapted from C. Mascareno (1997). Conclusions and Recommendations 149. To overcome key problems of the health sector, to improve the provision of health services and to facilitate human capital accumulation while increasing labor productivity, Venezuela must deepen its health and nutrition reforms, particularly regarding financing issues. The Government must (i) ensure sustainable mechanisms of financing at sufficient levels; and (ii) improve the allocation and efficiency of that financing. To achieve these objectives, the Government could: - Improve resource allocation to ensure equitable and efficient use of scarce resources; - Link budgetary transfers to the provision of minimum health services for the poor; * Include in the resource allocation incentives for performance, efficiency, and equity; * Assign priority to high-impact programs, such as maternal and childhood problems and control of endemic diseases; * Enlarge the participation of private providers in the health sector; * Emphasize cost-effective interventions grouped under a basic health service package. - 68 - 150. To improve equity in the sector, the Government could: * Expand health service coverage and reduce differences in access to health services between income groups, and improve targeting of programs to the poor. 151. To enhance the efficiency and effectiveness of service delivery, the Government could: * Develop institutional capacity for fulfilling new responsibilities at central and State levels, particularly in relation to policy-making, strategic planning, resource allocation, and monitoring of both resource use and results; * Develop new models of health care delivery to respond to changing health problems. III. Investing to Reduce Crime and Violence 152. A relatively new issue in Venezuela's social development agenda is crime and violence. The problem is, however, not new. Moreover, during the last fifteen years, there has been a significant increase in total crime and in violent crimes, particularly against people. Crime is significantly eroding the country's capital, and has negative consequences in the lives of the poor. To address the problem, this Section reviews the evidence regarding key determinants of crime and violence in Venezuela. Since young men commit most crime, particularly street crime, youth crime issues are given especial emphasis. Key public policies to reduce crime are then explored, and the Section ends with conclusions and recomnmendations. Crime and Poverty 153. One general assumption is that poverty is a determinant of crime (and a cause of violent crime2, but analytical research has shown the weakness of this cause-effect relationship.3 For example, in different periods of this century in which the poverty rate was notably higher in large cities, the rate of crime has been lower. On the contrary, the increase in crirne, particularly violent crime, has tended to increase beginning with the 1960s, an era in which most countries including Latin American countries had high sustained economic growth rates.35 The current debate has incorporated evidence This Section draws partly on a background paper prepared by Ana Maria Sanjuan. 34 See, for example, "The Myth of Social Class and Criminality: An Empirical Assessment of the Empirical Evidence" (Tittle, C., Villemez, W., and Smith, D.; American Sociological Review 43: 643-656; 1998). 35 In England, in 1914, there were 88 aggressions per 100,000 inhabitants, increasing to 850 per 100,000 inhabitants in 1989. Robberies went from 0.5 per 100,000 inhabitants in 1914 to 50 per 100,000 inhabitants in 1980 and homicides increased from 0.5 per 100,000 in 1914 to 1.1 per - 69- indicating that it is middle income countries that have experienced a jump in crime in the last few decades.36 This fact can be observed in Latin America where it is not the countries with the lowest income that have the highest crime rates, but rather those whose income distribution is highly uneven i.e., countries with high income inequality.37 154. The relationship between crime and poverty is perceived to be largely a result of the fact that the most over-represented groups in crime statistics have unstable occupations, low income, and are vulnerable to economic shocks. Moreover, they live in poor social and economic conditions that in turn expose them to more frequent arrest. The presumed relationship between crime and poverty could be based on a utilitarian definition of criminal behavior where the use of crime is considered a "survival strategy", as well as an extreme although illegitimate recourse for minimizing deprivation. 155. In line with this argument, Venezuela should have shown a significant increase in crimes against property at the beginning of an economic crisis. However, official figures show no appreciable variation in the proportion of such crimes. For example, in 1982- which marks the beginning of an economic downturn in Venezuela-70 percent of total crimes were crimes against property and 19 percent against persons. What the statistics do show, however, is an overall increase in violent crimes over time. While in 1982 approximately 22 percent of the crime against property involved the use of violence, in 1997 that proportion rose to 40 percent, with homicides rising by 100 percent. 156. Moreover, econometric analysis indicates that the relationship between the level of crime (including violent crime) and economic growth in Venezuela is not statistically significant over time.38 The relationship between violent crime and economic growth is not linear-in certain years with high economic growth homicide rates increased, while in other years with negative economic growth homicide rates decreased. For example, in 1990 and 1991, when Venezuela experienced economic growth, homicide rates did not decrease but rather increased. By contrast, in a year with negative economic growth (1996) homicide rates actually declined. 157. The analysis of the data partly supports the hypothesis that the rise in crime and violent crime over the last decades has resulted from higher inequality and social exclusion, particularly from poor young people. For most of the countries of the region including Venezuela, almost 60 percent of the victimns and perpetrators of crimes are 100,000 inhabitants in 1987. See "Drogas, Quadrilhas, Crimes, Exclusoes e Outros Esportes Nacionas " (Zaluar, A.; Revista Monitor Puiblico, N° 11; 1996; Rio de Janeiro). 36 See, "Firearm-related Deaths in the United States and 35 Other High- and Upper Middle-Income Countries" (Powell, K. and Dahlbert; International Journal of Epidemiology 27: 214-221; 1998). See, for example, "Inequality and Violent Crime" (Fajnzylber, P., Lederman, D. and Loayza, N. mimeo, World Bank, December 1999). 38 A statistically significant relationship between the crime rate and poverty over the last 17 years has also not been found. - 70 - young people between the ages of 15 and 29, living in poor neighborhoods, having a limited and precarious education, and facing restricted access to the formal labor market.39 Public schools in poor districts lack the appropriate infrastructure and are therefore unable to provide sufficient incentives and motivate young people to learn and accumulate human capital. 158. While poverty is not necessarily a determinant of crime, crime is eroding the country's capital, including social and human capital, and has particular consequences in the lives of the poor. The average cost of crime and violence in Venezuela has been estimated at about 11 percent of GDP per year in 1999-2000 and its direct and indirect costs affect the quality of the public services that the poor precariously receive.40 159. In sum, a comparison of crimes recorded in Venezuela in 1982 and 1]997 shows that: * There was a significant increase in total crime (see Annex E, Table 1). In 1982, only the Federal District of Caracas and the State of Carabobo exceeded the national crime rate, but by 1997 the national rate had increased and was surpassed by the Federal District plus four States (Carabobo, Aragua, Nueva E]sparta and Monagas); * There was an increase in the use of violence in crimes against people (the increase in homicide rates was the highest) and in crimes against property (robbery rates and auto theft also increased significantly). See Annex E, Tables 2-;41 * Crime is concentrated in Caracas and in the cities with the largest populations, especially those in the States of Carabobo and Aragua; 39 The 1997 unemployment rate in this age group was about 20 percent. See "Stylized Facts and the Characteristics of the Labor Supply in Venezuela" (World Bank, Report No. 17901-VE, June 1998). 40 In the health sector for example, the capacity of hospitals is stretched by the numerous intentional lesions that must be attended daily, which is detrimental to the attention given to other illnesses. Selective operations and routine procedures must be postponed because of the increase of violent traumas requiring immediate attention. Because of inadequately equipped hospitals and the fact that patients cannot pay for medical and surgical supplies, certain injured persons have to forego proper treatment. In the education sector, schools located in communities with high rates of violence have limited attendance. In a number of schools in high violence areas, children are not allowed to use backpacks to canry their books and school supplies for fear that they may conceal weapons. In addition, poorer people are often victims of violent assaults and robberies while on public transportation which, besides depriving them of their belongings, restricts their mobility. See "Magnitud de la Violencia Delictiva en Venezuela" (IESA, LACSO, 1997, Caracas). 41 A recent study commissioned by the Inter-American Development Bank found that 17 percent of adults living in Caracas were victims of robbery in the last year. Also, 4 percent reported that themselves or a close relative had been kidnapped recently. Moreover, 9.7 percent reported that a close relative had died in the last year as a result of violent crime. These figures show that Venezuela now has the highest or second highest incidence of violent crime in Latin America, at levels equal or worse than Colombia, El Salvador and other countries in the region (according to similar studies conducted elsewhere). - 71 - * Although a statistically robust correlation between crime and poverty cannot be established, crime is significantly eroding the country's physical, human and social capital, and has negative consequences in the lives of the poor. Violent Acts and Profiles 160. Homicide. Homicide, which in 1986 constituted about 32 percent of total violent deaths, increased to about 62 percent of such deaths in 1997. Homicide became the primary cause of death among men between the ages of 15 and 29, displacing traffic accidents.42 It does not appear that the variation in the homicide figures is due to any specific policy but rather to other factors, such as demographic factors. 161. In 1986, 75 percent of total homicides were comnmitted with firearms, while in 1997, 92 percent were the result of use of firearms. The increase in homicides with firearms could be explained by a combination of factors, including the presence of key situational precipitators like alcohol, drugs, and firearms. 162. About 70 percent of male victims were between 15 and 29 years old. This figure indicates the epidemic nature of homicide among young people because the mortality rate in this age bracket can reach up to 250 homicides per 100,000 inhabitants.43 If estimates were taken only with the young people from poor neighborhoods, the rate would double. Of the women victims of homicide, about 48 percent were between the ages of 15 and 29 years of age. The results of the toxicology reports on victims of violent deaths in 1986 and 1997 show that, on average about 30 percent of the tests were positive for alcohol while only 2 percent were positive for cocaine. However, since 1995 the number of victims with positive tests for cocaine has increased. 163. Intentional Lesions. Cases reported to the Medicina Legal show that in 1996, about 60 percent of men were injured in public quarrels, 20 percent in robberies, 10 percent in family quarrels, and about 10 percent in work-related confrontations. "Arguments" were the motive for about 75 percent of all lesions to men. In the case of women about 50 percent of injuries were sustained in family quarrels, about 40 percent in public quarrels, about 7 percent in robberies, and about 3 percent were the result of child abuse and rape. Quarrels and confrontation were the cause of 90 percent of all injuries attended. 164. Violence Against Women. Of the total number of violent accidents that occurred in Caracas between 1986 and 1997 (excluding traffic accidents) in which the victims 42 Although records show the sharp increase in homicides in Caracas between 1989 and 1994, they also show a sustained reduction in those rates between 1994 and 1997. Except for suicides, all violent deaths decreased by the same proportion. This does not take into consideration the fact that the total population of young people in the capital would have to exclude those residing in middle and upper income sectors of the city, who are less vulnerable to homicide than those living in depressed areas. - 72 - were women, about 40 percent were under the age of 4 and about 10 percent were between 5 and 9 years of age, indicating that about 50 percent of accidents ilivolve girls less than 9 years of age.44 For females, the probability of being involved in a violent accident between ages 0 and 9 increases by 100 percent, highlighting the fact that women's violence begins at a very early age whether through direct assault, negligence or lack of protection. Also, the home is dangerous for women as the number who were victims of homicide in their homes is four times the comparable figure for men, a finding that also holds in the case of intentional injuries. 165. One of the major problems relating to domestic violence is under-reporting to authorities. A study showed that between January and April 1998, of the total number of patients treated in emergency in one of the largest hospitals in Caracas, aboul. 64 percent were women, and of these about 62 had been subject to domestic violence .45 Among women with signs of mistreatment by their spouses (who did not report them as such but as "accidents"), about 42 percent had suffered traumatic injuries and about 43 percent had "fallen". The doubtful nature of these accidents becomes clear when information is cross checked against accident characteristics for age groups.46 This study strongly coincides with the results of interviews with victims of domestic violence who for the most part were within the 15 to 39 age bracket. Almost 70 percent of the injuries were sustained in the home, of which about 30 percent were caused by the spouse, 20 percent by the ex- spouse and about 10 percent by the boyfriend. As to the reasons leading to inijury, in the majority of cases (about 52 percent) the causal factor was alcohol. Also in about 50 percent of the cases, the violence had not occurred for the first time which corroborates the high rate of repetition among victims of domestic violence. 166. One of the compounding effects of domestic violence is the "second" victimization that women undergo in their homes if they report the aggression to authorities. Of the victims interviewed, about 60 percent reported that the situation worsened after they went to the police because the police do not take action unless the injury is serious. For the most part, the police assume that this is a problem of the couple (about 22 percent) and do not follow up on the reported aggression (about 22 percent). The Law Against Domestic Violence has been recently enacted but there are no specialized departments to receive reports of aggression and to provide professional attention in such cases. As such, female victims of domestic violence do not have the benefit of institutional protection. Although this "low intensity" violence in most cases is not fatal, it is nonetheless a serious form of discrimination against women. 44 The figures for men in the same age group show that the probability of having an accident between the ages of 0 and 4 is only about 16 percent and between 5 and 9 years cf age is about 5 percent. 45 See "Diagn6stico Situacional de los Servicios de Prevencion v Atencion a la Violencia Intrafamiliar Contra laMujer" (Sanjuan, A.M., mimeo, 1998). 46 There is an unusually high proportion of women with uncharacteristic injuries (such as "falls" to which women over 65 would be more prone). In the 15-34 age group, about 54 percent of injuries are the result of trauma and about 46 percent of "falls". - 73 - 167. Violence and Youth-Exclusion. The epidemic nature of homicide and its significance for poor young people between the ages of 15 and 29 is well recorded. Almost 70 percent of the victims of homicide are young and poor, and the aggressors have the same characteristics. A series of interviews with young people vulnerable to violence found very similar patterns to those found in research from other countries in the region: the greatest risk for those living in the "barrios" or shanty towns is not surviving adolescence. The high mortality rate due to homicides in these areas counteracts and outweighs the reductions in the infant mortality rate. 168. From a very young age there is exposure to violence in the cultural milieu, communities and families. This generates problems of affection and self-esteem and therefore requires a display of behaviors among these youth geared to reinforcing and preserving self-image. Based on a recent study of 500 children from around the country,47 violence was an everyday occurrence and was present (i) in homes, for example in the form of screaming and fighting (37 percent) and physical mistreatment or blows (36 percent); (ii) in school, for example in the form of gangs (28 percent), treatment by teachers (25 percent), and stalking by "delinquents" (23 percent), and (iii) in communities, for example in the form of conflicts between neighbors (33 percent), gang wars (27 percent), and "barrio hoodlums" (25 percent). New research has shown that some violent youths have very young mothers with scant education who work in precarious jobs with increased difficulty in order to survive, thus increasing potential household problems and tensions. As a result, the risks of violence in the family also increase. 169. Most Violent Youths Are School Dropouts.48 Violence experienced in the home destabilizes children affecting their performance in school, beginning a vicious cycle perpetuated by low performance which de-motivates them from studying.49 The high dropout rate in low income environments is one of the elements that leads to the creation of juvenile sub-cultures and the concomitant search for an identity. During adolescence, the phase in which humans develop diverse strategies for building individual and group reputations, these young people are constantly persecuted by their parents, their communities and the police. Such a conflictive environment increases the perception of attack on their self-image and social recognition, in the sense that they can lose prestige within the group if they do not resort to violence to defend themselves and their image. Consequently, for these youths danger is not only profitable to maintaining their image 47 See "Voices for Change" N° 2 and N° 9 (CECODAP, 1998, Caracas). 48 In a survey by Consultores 21 in 1997, among the explanations given by parents with school age children who do not attend school, 22 percent said that "the children did not want to attend school", 7 percent said their children were "lazy, scoundrels", 5 percent said their children were repeaters and 3 percent said that their children were treated badly in school. About 18 percent said that they did not have money to send their children to school. See "Education, Youth and Citizenship" (Bruni Celli, J., 1998); and "Ideas for an Educational Debate, National Assembly on Education" (National Education Council, Caracas). 49 See "A violencia urbana e osjovens" (Cardia, N., mimeo, 1997, Sao Paulo). Also, see "Urban Poor: Perceptions of Violence and Exclusion in Colombia" (Moser, C. and Mcllwaine, C., World Bank, 2000). - 74 - but also constitutes a survival mechanism in managing their conflicts. Thte types of violence in which they begin to be involved is strongly related to futile motives, conflicts to earn the respect of others, defense of honor, courage, and territorial supremacy-all of which are expressions of chaotic peripheral neighborhoods.50 170. Over the past few years, young people at risk have seen their expectations of upward mobility and access to social services seriously deflated because of the lack of jobs or increased demand for higher qualifications for professional training (i.e., higher level of human capital). This leads them to accept informal and transitory employment or alternatively, to engage in activities that guarantee high and quick income returns. In this context, any crime is highly viable because it is the closest and clearest option for quick promotion and social mobility. The inadequacy and limited nature of legitirmate means for achieving higher consumption patterns strengthens this passage to other illicit activities. The transgression that accompanies this passage is based on the belief that the risks are compensated by the ensuing gratification (e.g., higher income and consumption, success with women and the fear or respect of men and the community). 171. For young people living in poor neighborhoods, special programs to combat alcohol or drug addiction are limited or inaccessible. They have neither community networks to help reinsert them in society or neutralize the conflict,. nor do they have institutional or parental support to help them find alternative means for their constructive emergence as players in society. Moreover, young people at the greatest risk are the preferential targets of repression by police and justice. Usually, young people are killed before they even reach the criminal justice system by either the police or in fights in which authorities do not intervene. Their exclusion does not solely relate to economic variables, but also to the fact that an entire generation and the communities to which it belongs are excluded from the legal system. Key Public Policies for Reducing Crime 172. The Context. In a recent evaluation of the perception of the justice system, 85 percent of individuals interviewed said that they had little or no confidence in the system and about 76 percent felt that it was not functioning and should be changed.51 The two most serious problems cited were (i) corrupt judges and court officers; and (ii) slow processes. Of those interviewed, about 95 percent also stated that justice in Venezuela favors the rich and about 50 percent were in favor of using their own means to obtain justice; 61 percent have heard talk about reforming the system but 56 percent believe that even with reform there will be little or no improvement. Moreover, 60 percent of the interviewees expressed the opinion that corruption is the principal obstacle. See, for example, "Youth Gangs and Violence in Latin America and the Caribbean: Literature Survey" (Rodgers, D., World Bank, August 1999). 51 See "Diagnosis of Social Perception of Justice in the Caracas Metropolitan Area, Public Opinion Poll, Final Report" (United Nations Development Program, March 1998, Caracas). - 75 - 173. One of the problems associated with poor economic performance over the last decades and the perceived climate of insecurity which creates social pressure is the overall feeling of injustice derived from impunity. Notable is the lack of punishment for corruption, embezzlement, public crime and other offenses-Venezuela is perceived as being among the ten most corrupt countries in the world based on the 1998 list prepared by Transparency Intemational. Based on a survey in 1998 (Pan-American Survey by Wall Street Journal of the Americas), the main problem in Venezuela is corruption (26 percent), followed by the economy (15 percent), insecurity (13 percent) and inflation (13 percent). 174. Policies. The policies for maintaining public order, security and justice which have been put into practice in the last five years, have been based on increasing the budget for justice and security, updating the police force (increasing infrastructure and the number officers, purchasing police vehicles and equipment), changes in criminal legislation (adoption of the accusatory system instead of the investigative system) wherein the Public Prosecutor's Office acquires new competencies, and increasing prison capacity.52 175. Investment in Security and Justice. In the last five years, significant resources have been earmarked for justice and personal security (see Table D6). Even in circumstances of overall spending reductions, the resources for security remained stable or were increased. During the 1980s, public expenditures in the sector remained more or less constant (between 2.6 percent and 3.3 percent of total) at a time in which crime rates rose significantly. However, since 1994 public expenditures in security and public order have increased and were given more importance than administration of justice and the public ministry. This allocation of public resources, plus the restriction of recourses for crime prevention and the lack of integrated social programs including youth programs, underscores the implicit contradictions of the policies regarding the causes of the crisis in public order. 176. The Police. In a public poll in 1991, to the question "If you are robbed or assaulted, what is the probability in your opinion that the assailant will be a policeman?" 69 percent of respondents answered most probable/probable, while in 1997 that percentage rose to 82 percent. Venezuelans do not trust the police, particularly in Caracas (about 80 percent of the population rejects the police). Also, Venezuelans characterize the police as being discretionary in their actions and as proceeding arbitrarily without any effective control by the Government or citizens. 177. Caracas has seven police forces, which added to the investigation and security staff of the State, with a police force of 552 officers per 100,000 inhabitants in 1997 (up 52 Other proposals such as the Integral Security Plan, the Human Rights Commission and other more recent initiatives, because they are not far-reaching and permanent or have not had impact yet, were not incorporated into this analysis. - 76 - from 290 officers per 100,000 inhabitants in 1990).53 Police presence in low income communities is neither permanent nor subject to community control, and usually takes the form of violent, intermittent mass "operations". In the poor communities with the most serious violence problems, a deep-seated fear of police actions exists, making crime prevention more difficult. Those communities or demanders with higher resources do not trust public security and contract private security services that almost double the number of total police officers-since 1993 private security has increased in the country by 534 percent. 178. Administration of Justice.54 Impunity and inefficiencies in the administration of justice in criminal cases are major constraints in reducing crime in Venezuela.55 According to official figures, the ratio of judges per 100,000 inhabitants was about 6.5 in 1982, falling to about 5.5 in 1997. As the optimal proportion of judges to population in developing countries is 4.7 judges per 100,000, the problem does not appear to be the number of judges but the way in which justice is administered. This is visible in the relationship between crimes recorded and the final sentences in lower courts for criminal cases where it is evident that only about 5 percent of cases recorded receive a sentence.56 179. The Penitentiary System. Between 1982 and 1997, the incarceration rate in the country rose from about 111 to about 114. There are also severe delays in processing: in 1982, 70 percent of the total number of prisoners were tried, while in 1997 that number fell to only 63 percent. For the whole country, about 97 percent of inmates are poor and do not have sufficient means to afford their defense.57 In 1982, about 50 percent of all prisoners were arrested for committing crimes against property and about 30 percent for crimes against persons; in 1997, these rates were about 60 percent and about 25 percent respectively. There is a greater emphasis in punishing crimnes against property which affect the wealthier segments of society instead of punishing crimes against people in which the poor are more often the victims. 53 The 1997 figure is almost 100 percent higher than the number of police officers considered ideal-350 officers per 100,000 inhabitants-by the United Nations for cities in developing countries. See "Situaci6n del Delito y de la Seguridad de los Habitantes en los Paises de America Latina" (Carranza, E., Delito y Seguridad de los Habitantes, 23-60, Siglo Veintiuno Editores, 1997, Mexico). 54 This is an area that is the target of recent reforms and legislative changes. This Section is confined to a discussion of the relationship between justice and violence. 55 It should be kept in mind that many violent crimes, especially those involving repeat offenders, are resolved on the fringes of the law in a cycle of private revenge that perpetrates the cycle of violence. 56 In 1982, a year in which less crimes and homicides were reported and there were more judges, still only 5 percent of cases in lower courts received a sentence. 57 See "Estadistica Delictiva: 1982-1997" (Division of Prisons, Ministry of Justice., 1997, Caracas). - 77 - Conclusions and Recommendations 179. In Venezuela, the relationship between crime (including violent crime), economic growth and poverty is not statistically significant. However, crime and violence are negatively affecting the accumulation of the country's capital. More importantly, the deterioration of human capital and the cost in human lives is detrimental to the lives of low income families, both in terms of personal costs as well as costs to the sectors that are key to alleviate the impact of poverty, namely education and health. 180. There is some evidence that the rise in crime and violent crime over the last decades has resulted from higher inequality and social exclusion, particularly of poor young people. In Venezuela, almost 60 percent of the victims and perpetrators of crime are between the ages of 15 and 29, live in poor neighborhoods, have a limited education and restricted access to the formal labor market. High dropout rates, lack of stable job opportunities, chaotic relations with the police and the weakness of the boundary between the legal and the illegal, significantly influence the decision of young people to commit crimes. To tackle this problem, the Government could consider creating additional opportunities for this segment of society by enhancing both social and human capital in poor neighborhoods. To enhance social capital, the Government could consider (i) improving the provision of and access to public services, including safety in poor neighborhoods; (ii) providing incentives for private and voluntary sector provision of public services in poor neighborhoods; and (iii) promoting community-based care, family and youth services, and probation and adoption systems that permit youths to remain in a family and community environment. To enhance human capital, the Government could consider programs (i) facilitating access to education and retention of at-risk youth in school (e.g., developing surveys to identify youth at risk, increasing the use of demand-side mechanisms-such as vouchers and income transfers-adapting and piloting "full service community school models", and piloting learning modules to enrich curricula); and (ii) reaching youth outside the education system (e.g., creating and expanding youth development centers and supporting the development of alternative and remedial educational programs). 181. Domestic violence is a serious problem in the country and is a common occurrence that affects the capacity of poor women to earn an income, as well as having consequences for the minor children exposed to such violence. To reduce this type of violence, the Government could consider establishing more explicit judicial instruments towards penalizing domestic violence and creating special institutions for treating victims of domestic violence. 182. Venezuela has a weakly integrated policy of citizen security. The rise in crime and violence can be linked to the poor administration of public security and justice, and citizen control policies. They have an impact on the negative perception of all policies against crime, increasing the justification for looking for solutions on the fringes of the law and for increasing corruption. This "lack of justice" primarily for the poor, needs to be addressed to complement efforts to increase social capital (through better - 78 - neighborhoods, community actions, and youth services) and human capital (through better quality and quantity of education, and programs to keep youth in school). 183. Finally, the lack of data and the weakness of official statistics make it difficult to more precisely determine all the causes and factors behind the rise in violence and crimes in Venezuela. It is therefore necessary to improve the information system to allow decision-makers to have better and timely data for formulating policies for effectively combating the problem. - 79 - I I00 Annex A Summary of Economic Plan and Objectives for 2000 On February 2, 2000, President Chavez initiated a series of presentations to introduce the plans in the area of Economic Development that his Government will implement. In fact, on the I" anniversary of his Presidency, after presenting political and social reforms, the President presented the Macroeconomic Plan. The macroeconomic goals of the Government for the year 2000 were: * Expected inflation 15-17 percent (actual: 13 percent) * GDP 2.2 percent (actual: 3.2 percent) * Unemployment Reduce by 2 percent (fell from 14.9 to 14.2 percent) * International reserves by the end of the year US$16 billion (actual: US$17 billion ex. FIEM) 3 Deposit interest rate 17 percent (actual: 15 percent) * Fiscal deficit 1.5-2 percent (preliminary 2 percent of GDP) D Salary increase 10 percent (first semester) * Average exchange rate Bs.697 per US$ (actual: Bs.654 per US$) The most relevant measures to be taken are: * VAT reduction from 15.5 to 14.5 percent (reduction based on the collection of 6,000 to 4,000 Fiscal Units); 33 percent reduction of tax exemptions. The Minister of Finance is in charge of implementing these measures; * End of the Bank Debit Tax (BDT), beginning May 14; * Immediate elimination of the BDT for transactions in the Stock Market; * Regularization on the payments of Export Drawbacks, beginning May 15; * Maintenance of monetary and exchange rate policy, with exchange rate policy against devaluation; * Public sector salary increase of 10 percent beginning May 11; * Restructuring of Public Sector Expenditure in response to the tragedy of December 1999; * Export promotion through a Rotating Fund; * Tax incentives for the creation of new productive jobs, including exemptions on payments or taxes (income tax, VAT, assets, stamp and tariffs), for creaLtion of jobs in Special Free-Trade Zones, Industrial Parks; * SENIAT's reorganization (Revenue Services Office); * Government procurement and purchases will favor local producers/vendors to provide incentives for creating new jobs; - 81 - * Financial support for new investors, in particular small and medium enterprises (through government credit agencies.); * Industrial parks rehabilitation; * Creation of the Instituto Nacional de Desarrollo Industrial; * Rescue of the infrastructure (agricultural, telecommunications, etc.) by implementing the Zamora y Bolivar 2000 plans; * Acceleration of Construction of Public Works (Bs. 400 billion are committed) and implementation of Plan Vargas; * Fiscal incentives and rebates for production expansion of agricultural products such as palm oil, rice, coffee, cacao, sugar, cattle raising, fishing, yucca and key products such as corn, sorghum; and * Fiscal credits of 80, 70 and 10 percent for agricultural, tourism and industrial activities. Chronology of specific plans presented In February 2000: Plan to regain/recover the State of Vargas Plans for Expansion of Agriculture, Tourism, Industry and Commerce Aluminum Sector's Plan Oil Sector's Plan for 2000 - 2009 Zamora's Plan 2000 for Tierras y Hombres Libres Educational Project Health and Social Sector Development's Plan Revision and re-launch of the Bolivar 2000 Plan In March 2000: Plan for De-concentration and Creation of New Cities Electricity's and Natural Resources' Plans SENIAT's and Custom's Plans - 82 - Annex B The Link Over the Oil Cycle (Between Real Government Expenditures and Real Oil Prices) Econometric analysis of available data on real government expenditures and real oil prices indicates that that there is a robust link over the oil cycle between real total government expenditures and real oil prices. However, real social sector expenditures are relatively protected from developments in the oil market over the cycle. We estimated Vector Auto-Regressive (VAR) models using estimated cycles in real oil prices, real total government expenditures and real social sector expenditures.' The data set consisted of yearly observations from 1971 to 1998. For each variable, cycles were calculated by taking the difference between observed data and the corresponding Hodrick- Prescott trends. It is worth noting that these calculations in fact yield a cyclical component, plus white noise and any other stationary unobserved components in the series. We assumed that well behaved cyclical components do exist and that the differences between the underlying cyclical processes and the estimated cycles are pure noise. To obtain additional insights into how real oil prices and real expenditures are linked, the issue of cycle formation in the series under consideration is addressed. The VAR selection procedure took into account actual information, goodness-of- fit, estimated log-likelihood and F-statistics, as well as alternative lag lengths. To capture significant oil shocks over the time series, we introduced two dummy variables for the years 1974 and 1986, both of which were found to be highly significant. Short lag lengths proved to be most satisfactory. Two VAR models were selected taking first, the cycle in total government expenditure as one of the endogenous variables; and second, taking the cycle in social expenditures as one of the endogenous variables. In both cases, the cyclical component of the price of oil was also endogenous. The model selected for total government expenditures is:2 (1) OilPricet= 0.003+0.251*Dumnmy74-0.268*Dummy86+0.620*OiI Priceti10.250*Total Exp,.1+et (2) Total Expt=-0.008+0.281*Dummy74-0.OO*DUmmy86+0.216*OilPricet,+0.265*TotalExpt-+ut The endogenous variables were logs of the estimated cycle components in standardized indices (1987=100) of real oil prices, total real government expenditures and real social sector expenditures. 2 Equation (1) has: R2=0.65 and Equation (2) has: R7=0.56. This lag specification was superior to the alternatives in terms of F-statistics, log-likelihood and information criteria. The dummy variable for 1974 was highly significant in both Equations; the dummy variable for 1986 was significant only in Equation (1). - 83 - The model selected for social sector expenditures is:3 (3) Oil Pricer= 0.003+0.240*Dummy74-0.270*Dummys6+0.551*jil Price1.1-O.461*Social Exp,1.+v, (4) Social Expt=0.001+0.064*Dummy74.0.055*Dummy86+0.098*Oil Pricet-+0.364*Socia] Exp11+s, We also estimated a VAR model that simultaneously included oil price, total expenditure and social expenditure cycles as endogenous variables. The results were consistent with those derived from the models described by Equations (1) and (2), and (3) and (4).4 The most relevant impulse response functions are shown in Graphs 1 and 2. Graph 1 shows that total government expenditures immediately and positively responds to a shock in oil prices.' The response in total expenditures peaks in period 2 following the impulse disturbance in oil prices, and then tails off slowly over several periods. The relative response of social expenditures to a positive impulse in oil prices is smaller than the corresponding response of total expenditures at all periods, indicating that social expenditures are relatively protected from developments in oil markets (see Graph 2). The Graph 1 Graph 2 0.060 0.060 To 0.0° 0.j°- tal Ex 0.040 0.040 _ P- C 0 030 O'~~~~~~~~~~~~a 0.3 Cy 0.030 \ -- 03- \ Re 0.020 o. 0.020 on 0.010 ses 0.000 -.00 -0.010 -0.010 Period Periods Response to Standardized Oil Price Cycle Response to Standardized Oil Price Cycle Impulses |-Response to Standardized Total Expenditure CycleI -Response to Standardized Social Expenditure Cycle Impulses graphs also show that both types of expenditure respond positively and significantly to own-shocks in the first two to three periods of analysis; responses tail off relatively quickly, as we might expect given the cyclical nature of the processes under consideration. In the impulse response functions shown below, the estimated co-variation components of Equation (3) has: R2=0.71 and Equation (4) has: R2=0.20. This lag specification was superior to the alternatives in terms of F-statistics, log-likelihood and information criteria. The dummy variables for 1974 and 1986 were significant only in Equation (3). It is also interesting to note that the social expenditure cycle exhibited much higher responsiveness to shocks in total expenditures than to oil price shocks. By contrast, the response of total expenditure cycle was dominated by own shocks and oil price shocks, with only limited responsiveness to social expenditure innovations. The impulse response functions shown in Graph I are estimated from the model consisting of Equations I and 2; the impulse response functions shown in Graph 2 are estimated from the model consisting of Equations 3 and 4. - 84- the innovations in the VAR systems have been attributed to the corresponding estimates of the innovations of the underlying oil price process.6 To deepen our analysis of the relationship between the cyclical components of the price of oil and government expenditures, we also specified various state--space Auto Regressive Moving Averages (ARMA) models for the estimated "cycle" series; the most satisfactory being an ARMA(2,2) process.7 These state-space models allowed us to estimate series of cycle-generating disturbances that underlie the estimated "cycle" series for oil prices and government expenditures.8 The ARMA's results corroborated the results obtained through VAR analysis of the estimated cycle series. We also found that the disturbance driving the oil price cycle is Granger-causing the disturbance that generates the total expenditure cycle; by contrast, the latter disturbance does not Granger-cause the process driving the oil price cycle. These results confirm our earlier findings. In addition, we found that the disturbance driving the oil price cycle does not Granger-cause nor is it Granger-caused by the process underlying the social expenditures cycle.9 These results are consistent with the VAR analysis described above and indicate that that social expenditures are relatively protected from developments in oil markets. Finally, we obtained further evidence in support of the general characteristics of the impulse response functions described above by estimating additional VAR models for the ARMA's cycle-generating disturbances and estimating the corresponding impulse response functions. The results are: (i) the total expenditure disturbances respond positively to the corresponding oil price disturbances (with a peak in the second period); and (ii) the response of social expenditure disturbances peaks at a later period, suggesting a greater degree of inertia and inducing a more or less counter-cyclical response pattern. 6 In the process of trying alternative VAR specifications, we also estimated impulse response functions attributing the co-variation component of the innovations to the expenditure variables. We discarded these impulse response functions because they yielded results that were not supported by our analysis of cycle formation and direction of causality. This is to be expected, as it can be shown that the cycle-plus-noise model has an ARMA(2,2) representation ("Forecasting, structural time series models and the Kalman filter", Harvey, A., Cambridge 1989). Please note that these disturbances are not supposed to be white noise but are expected to have a more complex structure. Also, please note that the state-space variables that we have liberally chosen to describe as "cycle-generating disturbances" are not the same as the disturbances/shocks/innovations that appear in the VAR models described above. In addition, it is worth noting that the disturbance generating the total expenditure cycle does not Granger cause nor is it Granger caused by the corresponding social expenditure disturbance. - 85 - Annex C Education-Methodology for Estimating Rates of Returns The earnings function method, derived from Mincer (1974), involves the fitting of a function specified as: lnY, = a + /3S, + yi +EX y2EX2i + y3Z± + where lnY is natural logarithm of monthly earnings, S is the number of years of schooling of individual i, and EX and EX2 are years of experience and its square, and Z is a vector of control variables. Hours worked per week are added as an independent variable to this function as compensatory factors. In this semi-log specification the coefficient on S (,B) is interpreted as the private rate of return to one additional year of schooling, averaged across all levels of schooling and all individuals in the sample. The earnings function method is used to estimate returns to different levels of schooling by converting the continuous years of schooling variable (S) into a series of dummy variables representing the levels of schooling. After fitting the extended earnings function, InY, = a + /3jPRIM, + /2SEC, +,83 UNIV1 + y1EX, + r2EX21 + 3Z,+ £ where PRIM, SEC and UNIV refer to dummy variables for completing primary, secondary and post-secondary education, the private rate of return to different levels of schooling is derived from the formulas: r(pRIM) = /Ji / SPRjM r(SEC) (/32 - 81 ) / (SSEC - SPPJM) r(umII) = (/83 - /82) / (SUAIV- SSEC) where SPRIM, SSEC and SuNIv stand for the total number of years of schooling for each successive level of education. However, it is incorrect to assume that primary school graduates forego earnings for the entire duration of their studies. Therefore, only three years of foregone earnings for primary school graduates are assumed. A more complete method is the long, or net earnings profile, method. This method uses detailed age-earnings profiles by level of education are used and the discount rate that equates a stream of education benefits to a stream of educational costs at a given point in time is found: Y_ b,/ (Il+r)- = Lct/(1+r)' t=O0 t=_O - That discount rate is the internal rate of return on the investment. - 86 - Annex D Health and Nutrition-Statistical Appendix Table 1. Network of Public and Private Health Care Establishments, 1996 Rural Urban Hospitals Total Ambulatories Ambulatories MH 3,369 627 182 4178 IVSS 1 83 35 119 Defense Minister - - 9 9 Municipalities - - 15 15 Total public sector 3370 710 241 4321 Private sector - - 344 344 Total 3370 710 585 4665 Source: M, 1997. Table 2. Installed Capacity of Health Institutions in Venezuela, 1994 Institutions Architectonics beds Budgeted beds Public sector 40,510 35,009 MH 25,611 20,693 IVSS 6,744 6,264 Government of the Federal District 2,864 2,850 IPASME 40 40 Defense Minister 1,136 1,136 Autonomous Institutions 4,095 4,026 Private sector 11,790 11,790 Total 52,300 46,799 Source: MH, 1997. Table 3. Distribution of Human Resources in the Health Sector, 1993 Number % Number per 1,000 population Physicians 40492 31.28 19.4 Nurses medical forms 16013 12.37 7.7 Nurses aides 46757 36.12 22.3 Dentists 10361 8 5 Pharmacist 6356 4.9 3 Veterinaries 197 0.15 0..1 Engineers 280 0.22 0. I Health inspectors 1467 1.13 0.7 Food inspectors 478 0.37 0.2 Nutritionist 908 0.7 0.4 Nutritionists aides 2035 1.57 1.0 Responsible for 465 0.36 0.2 Medical forms aides 2665 2.06 1.3 Others 989 0.76 0.5 Total 129463 100 Source: Based on ihe data from "Anuano de Epidemiologfay Estodisticas Viales", MH, 1993. - 87- Table 4. Physicians and Nurses in Venezuela and Other Selected Countries, 1993 Physicians, Nurses, Rate of Rate per Midwives, Rate physycians to WHO Member State 100,000 per 100,000 nurses Latin America Argentina 268 54 5.0 Brazil 134 41 3.3 Chile 108 42 2.6 Colombia 105 49 2.1 Costa Rica 126 95 1.3 Cuba 518 752 0.7 Ecuador 111 34 3.3 Mexico 107 40 2.7 Peru 73 49 1.5 Uruguay 309 61 5.1 Venezuela 194 77 2.5 OECD countries France 280 392 0.7 Japan 177 641 0.3 Portugal 291 304 1.0 United States of America 245 878 0.3 Asia Malaysia 43 160 0.3 Philippines 11 43 0.3 Republic of Korea 127 232 0.5 Singapore 147 416 0.4 Thailand 24 99 0.2 Indonesia 12 67 0.2 Source: World Health Report, 1997. - 88 - Table 5. Distribution of Public Facilities and Physicians, 1995 Hospitals per Doctors per % of total % of total % of total 100,000 10,000 States population hospitals doctors population population Amazonas 0.4 0.2 0.6 1.1 29.4 Anzoategui 4.7 5.7 4.2 3.2 18.0 Apure 1.8 2.1 1.3 3.1 15.7 Aragua 6.2 4.5 4.6 1.9 15.4 Barinas 2.4 2.1 1.7 2.3 14.3 Bolivar 5.2 4.3 3.6 2.2 13.9 Carabobo 8.3 5.0 7.7 1.6 18.8 Cojedes 1.0 0.9 1.5 2.2 29.6 Delta Amacuro 0.5 0.5 0.4 2.6 15.5 Distrito Federal 10.4 23.3 26.3 6.0 51.5 Falcon 3.2 2.7 3.3 2.3 21.0 Guarico 2.7 3.9 2.3 3.9 17.5 Lara 6.6 5.1 4.9 2.1 15.3 Merida 3.1 4.8 4.4 4.1 29. 1 Miranda 10.5 5.3 4.1 1.3 8.0 Monagas 2.5 2.2 2.2 2.3 17.9 Nueva Esparta 1.5 0.7 1.6 1.2 21.6 Portuguesa 3.3 2.4 2.2 1.9 13.7 Sucre 3.6 3.1 2.7 2.3 15.2 Tachira 4.3 5.1 3.6 3.2 16.'9 Trujillo 2.6 3.1 2.7 3.2 21.4 Yaracuy 2.1 1.4 1.6 1.7 15.3 Zulia 12.9 11.7 12.6 2.4 20.0 Total 100 100 100 2.7 20.5 - 89 - Annex E Crime and Violence-Statistical Annex Table 1. Total Recorded Crimes* (National and sub-national levels) 1982 1990 1991 1992 1993 1994 1995 1996 1997 Venezuela 819 1255 1184 1210 1276 1270 1153 1173 1040 Caracas 1746 3013 2949 3037 3221 2774 2255 2190 2080 Amazonas 408 706 806 841 823 814 752 1107 734 Anzoategui 648 963 923 1004 1110 1161 1139 1055 844 Apure 654 553 544 575 606 716 636 665 525 Aragua 807 1443 1403 1297 1385 1501 1462 1405 1211 Barinas 452 684 661 731 678 732 759 858 816 Bolivar 775 945 966 971 1084 1105 1067 1032 935 Carabobo 953 1619 1540 1639 1693 1706 1485 1596 1429 Cojedes 520 727 633 689 698 866 914 874 803 Delta Amacuro 495 694 598 506 533 616 497 675 605 Falc6n 338 331 284 333 419 513 522 564 526 Guarico 746 1115 1005 928 957 938 909 906 911 Lara 450 652 620 627 629 806 830 949 736 Mrida 582 829 816 810 860 849 774 837 875 Miranda 415 749 702 743 785 849 809 795 679 Monagas 672 982 1144 1275 1489 1571 1462 1255 1156 NuevaEsparta 665 1107 1078 962 1008 1169 928 1379 1172 Portuguesa 602 654 608 674 648 691 688 813 678 Sucre 491 607 560 492 593 641 681 796 715 Tachira 513 840 763 687 752 794 732 733 735 Trujillo 343 507 529 483 546 582 518 546 504 Yaracuy 431 662 507 498 535 649 654 734 539 Zulia 657 799 727 805 783 856 829 855 657 * Rate per 100.000 inhabitants. Source: Estadistica Delictiva, Statistics Division, Ministry of Justice; own calculations. - 90 - Table 2. Recorded Lesions* (National and sub-national levels) 1982 1990 1991 1992 1993 1994 1995 1996 1997 Venezuela 138 152 167 174 163 152 141 138 137 Caracas 251 205 292 295 245 207 179 175 182 Amazonas 63 133 154 139 142 144 121 179 151 Anzoategui 132 151 170 177 181 160 147 139 135 Apure 159 147 173 171 150 180 157 153 127 Aragua 112 144 166 148 151 150 147 135 132 Barinas 102 134 161 183 179 152 165 168 172 Bolivar 134 144 156 154 147 145 137 137 149 Carabobo 138 161 158 177 180 175 163 160 157 Cojedes 117 165 129 154 145 138 145 126 134 Delta 230 181 162 163 141 127 146 142 Amacuro 162 Falc6n 85 66 65 82 87 96 90 101 84 Guarico 213 300 288 258 249 230 195 174 194 Lara 72 75 91 92 78 98 102 104 88 Merida 119 167 188 187 172 152 133 146 170 Miranda 75 97 106 122 109 95 85 82 82 Monagas 216 223 275 321 350 299 267 227 :242 Nueva Esparta 134 108 120 140 119 128 112 148 128 Portuguesa 115 121 128 157 149 123 114 119 117 Sucre 124 130 140 135 139 139 143 156 141 Tachira 139 175 163 157 159 167 162 145 145 Trujillo 92 112 114 119 142 120 94 104 88 Yaracuy 126 133 133 145 139 146 135 113 115 Zulia 88 104 97 106 95 94 93 91 84 * Rate per 100.000 inhabitants. Source: Estadistica Delictiva, Statistics Division, Ministry of Justice; own calculations. - 91 - Table 3. Recorded Homicides* (National and sub-national levels) 1982 1990 1991 1992 1993 1994 1995 1996 1997 Venezuela 11 13 13 16 21 22 21 22 19 Caracas 18 44 46 68 91 96 88 83 64 Amazonas 3 6 6 4 12 8 5 6 5 Anzoategui 9 8 8 9 10 12 12 12 12 Apure 12 10 9 13 10 11 9 14 13 Aragua 9 7 6 8 11 16 18 21 20 Barinas 10 8 8 11 7 10 11 11 10 Bolivar 9 12 13 14 16 17 18 20 19 Carabobo 11 12 12 15 19 21 20 29 24 Cojedes 10 8 6 6 6 10 8 9 9 Delta Amacuro 7 6 1 10 11 11 6 8 6 Falc6n 8 6 5 5 6 7 5 7 7 Guarico 11 12 14 16 15 16 12 16 18 Lara 7 5 5 5 6 6 7 8 9 Merida 12 7 7 7 8 7 3 7 6 Miranda 5 8 8 12 16 21 18 20 15 Monagas 7 7 8 11 13 9 9 10 12 Nueva Esparta 3 6 3 4 6 7 6 6 7 Portuguesa 12 9 8 7 10 10 12 16 16 Sucre 5 5 5 5 8 7 6 9 7 Tachira 6 9 9 9 10 8 9 9 7 Trujillo 14 10 10 11 9 9 11 13 11 Yaracuy 7 6 5 7 5 7 7 8 5 Zulia 18 11 10 12 16 17 15 17 14 * Rate per 100.000 inhabitants. Source: Estadistica Delictiva, Statistics Division, Ministry of Justice; own calculations. - 92 - Table 4. Recorded Thefts and Robberies* (National and sub-national levels) 1982 1990 1991 1992 1993 1994 1995 1996 1997 Venezuela 150 132 114 117 154 175 161 166 141 Caracas 414 367 343 353 462 447 358 341 329 Amazonas 16 31 49 52 42 36 33 45 40 Anzoategui 107 95 100 102 136 168 151 136 108 Apure 44 28 28 25 33 45 31 48 45 Aragua 168 118 97 100 166 192 222 248 207 Barinas 40 47 36 44 50 58 54 87 76 Bolivar 153 114 108 122 160 158 171 152 129 Carabobo 237 203 163 178 204 258 241 290 247 Cojedes 57 48 42 52 49 55 59 96 74 Delta 9 28 21 33 28 25 26 49 35 Amacuro Falc6n 23 13 12 11 21 34 37 43 46 Guarico 97 88 62 75 90 133 127 101 97 Lara 65 61 49 44 69 142 126 123 90 Merida 64 43 44 45 50 56 57 52 64 Miranda 88 99 80 316 106 136 116 114 84 Monagas 48 46 66 76 80 108 140 114 96 Nueva Esparta 49 56 50 33 48 112 69 III 81 Portuguesa 85 69 50 41 65 99 105 136 97 Sucre 53 50 46 33 63 68 79 96 85 TIchira 37 39 40 57 45 62 50 78 67 Trujillo 31 16 23 15 24 33 35 42 24 Yaracuy 65 41 27 22 34 49 56 71 58 Zulia 137 88 69 75 99 116 122 123 84 * Rate per 100.000 inhabitants. Source: Estadistica Delictiva, Statistics Division, Ministry of Justice; own calculations. - 93 - Table 5. Recorded Vehicle Theft* (National and sub-national levels)** 1982*** 1990 1991 1992 1993 1994 1995 1996 1997 Venezuela 51 41 47 69 82 85 99 70 Caracas 142 108 127 218 231 216 241 167 Amazonas 5 29 1 0 0 1 1 2 Anzoategui 28 3 41 59 76 73 72 63 Apure 2 1 1 1 2 11 4 3 Aragua 47 41 38 60 90 94 101 99 Barinas 8 2 5 7 7 7 9 10 Bolivar 48 49 56 68 76 84 79 61 Carabobo 100 79 89 110 138 137 177 127 Cojedes 7 7 10 13 19 26 29 3 Delta Amacuro 0 0 0 0 1 1 1 0 Falc6n 4 3 1 6 6 7 8 6 Guarico 20 17 16 26 43 32 33 31 Lara 25 23 16 22 46 51 65 47 Merida 5 4 3 4 9 11 7 5 Miranda 30 27 30 37 52 57 57 31 Monagas 8 7 10 15 17 23 42 44 Nueva Esparta 0 6 12 15 27 28 28 18 Portuguesa 36 19 15 28 36 51 83 47 Sucre 2 1 1 3 2 0 0 5 Tachira 3 2 3 7 8 8 13 8 Trujillo 4 6 6 3 5 10 8 8 Yaracuy 11 7 5 7 12 20 26 16 Zulia 58 48 62 80 94 117 143 88 * Motorcycle theft is included/ **Rate per 100,000 inhabitants/ ***There were no records for this type of crime in 1982. Source: Estadistica Delictiva, Statistics Division, Ministry of Justice; own calculations. - 94 -