Report No. 13993-TUN Republic of Tunisia Poverty Alleviation: Preserving Progress while Preparing for the Future (in Two Volumes) Volume I Main Report August 1995 Country Operations Division Country Departnient I Midcle East ancd North Africa Region Document of the World Bank Document of the World Bank 1I p Currency and Exchange Rates Currency Unit: Tunisian Dinar (TD) TD per US$ Period Averages 1980=0.4050 1981 = 0.4938 1982 =0.5907 1983 =0.6788 1984=0.7768 1985 =0.8345 1986=0.7940 1987=0.8287 1988 =0.8578 1989=0.9493 1990=0.8783 1991 =0.9246 1992 =0.8844 1993 = 1.0037 1994 = 1. 0116 Fiscal Year January 1st - December 31st Weights and Measures Metric System ABBREVIATIONS AND ACRONYMS AMG: Assistance Medicale Gratuite (Free or subsidized Health Care) BNA: Banque Nationale Agricole (National Agricultural Bank) CGC: Caisse Generale de Compensation (General Compensation Fund) CNRPS: Caisse Nationale de Retraite et de Prevoyance Sociale (National Pension and Social Insurance Fund) CNSS: Caisse Nationale de Securite Sociale (National Social Security Fund) CPI: Consumer Price Index (Indice des prix a la consommation) EU: European Union (Union Europeenne) FIAP: Fonds d'Insertion et d'Adaptation Professionnelle (Employment & Training Fund) FNAH: Fonds National d'Amelioration de l'Habitat (National Rehabilitation Housing Fund) FNRLR: Fonds National de Resorption des Logements Rudimentaires (National Fund for Slum Elimination) FNS: Fonds National de Solidarite (National Solidarity Fund) FODERI: Fonds Special du Developpement Rural Integre (Special Fund for Rural Development) FONAPRA: Fonds National pour la Promotion de l'Artisanat et des Petits Metiers (National Fund for Craftmanship Promotion) FOPROLOS: Fonds pour la Promotion des Logements Sociaux pour les Fonctionnaires a Bas Salaire (Housing Finance Fund for Low-Salary Civil Servants) INS: Institut National de la Statistique (National Institute of Statistics) LSMS: Living Standards Measurement Survey (Enquete sur le Niveau de Vie des Menages) MI: Middle-Income Countries (Pays A Revenus Moyens) MOA: Ministry of Agriculture (Ministere de l'Agriculture) PDR: Programme de Developpement Regional (Regional Development Program) PDRI: Programme de Developpement Regional Integre (Integrated Regional Development Program) PIC: Poverty Incidence Curve (Courbe d'Incidence de la Pauvrete) PNRPQ: Programme National de Rehabilitation des Quartiers Populaires (National Sanitation Program in Urban Areas) PDUI: Programme de Developpement Urbain Integre (Urban Development Program) SMAG: Salaire Minimum Agricole Garanti (Agricultural Minimum Wage) SMIG: Salaire Minimum Interprofessionnel Garanti (Non-Agricultural Minimum Wage) SNIT: Societe Nationale Immobiliere de Tunisie (Real Estate Company of Tunisia) SONEDE: Societe Nationale d'Exploitation et de Distribution des Eaux (National Water Supply Utility Company) SPROLS: Soci6te de Promotion des Logements Sociaux (Agency for the Promotion of Public Housing) STEG: Societe Tunisienne de l'Electricite et du Gaz (National Power and Gas Company) This report is based on the contributions of a team including Nawal Kamel, Stephen Mink, Martin Ravallion, and Dominique van de Walle of the Bank, and Fernando Clavijo, Juan Lopez, and Nicola Rossi, consultants. Valuable comments were received from, among others, Kathy Lindert and Miria Pigato. As task manager, Setareh Razmara was responsible for the overall preparation of the report. Lionel Demery and Jamil Salmi were peer reviewers for the report. At the time this report was prepared Mahmood Ayub was the Division Chief, John Underwood the Lead Economist, and Daniel Ritchie the Director of the department. When the Yellow Cover was discussed, Mr. Christian Delvoie was the Division Chief. Dominique Dietrich desk-topped the report. The cooperation of many Government departments in Tunisia is gratefully acknowledged. In particular, extensive assistance was received from the National Institute of Statistics, the Human Resource Division of the Ministry of Economic Development, Ministries of the Social Affairs, Public Health and Education. Without their help, preparation of this report would not have been possible. REPUBLIC OF TUNISIA POVERTY ALLEVIATION: PRESERVING PROGRESS WHILE PREPARING FOR THE FUTURE Contents Page No. EXECUTIVE SUMMARY .....................................i CHAPTER I - A SUCCESSFUL MACROECONOMIC FRAMEWORK FOR POVERTY ALLEVIATION ....................1 A - The Tunisian Stabilization Experience .................... 1 B - Public Finance in the Social Sectors ..................... 4 Need to Ensure Sustainability of Public Spending in Social Sectors 5 C - Growth, Employment, and Wages ...................... 5 Changes in Production Factors and Prices ............... 5 Employment Generation .......................... 6 Wage Policies ................................ 7 D - Adjustment Reduced Income Disparities ................... 7 CHAPTER II - POVERTY AND SOCIAL CONDITIONS ................. 10 A - Reduction in Poverty Incidence ....................... 10 Other Poverty Measures .......................... 12 Robustness of Poverty Reduction ..................... 13 B - Who Are the Poor? ............................... 14 Regional Profile of Poverty ........................ 14 Demographic and Socioeconomic Characteristics of the Poor ... . 14 Income and Employment Characteristics of the Poor ... ....... 15 C - Progress in Social Indicators .......... . ............... 16 D - Data Requirements for Future Poverty Monitoring ............ 18 CHAPTER III - URBAN AND RURAL POVERTY REDUCTION ............ 19 A - Structural Determinants of Poverty ...................... 19 Physical Assets ............................... 19 Income-Earning Opportunities ...................... 20 B - Policies and Factors Contributing to Poverty Reduction ... ...... 21 Agricultural Price Policies ........................... 21 Credit Policies to Promote Investment ........ .. ........ 22 Land Policies ....... .......... ................ 22 Private Transfers ...... ......... ............... 23 Public Transfers ................ ............... 23 Employment Programs ............ .. ............. 23 Growth in Temporary, Low-Wage, Low-Skill Employment ... . 25 Expansion of the Urban Informal Sector ...... .. ........ 27 C - Conclusions and Recommendations ...................... 27 Improving Agriculture-Based Incomes .................. 27 Increasing Access to Land Tenure .................... 28 Expanding Earning Opportunities in Rural Areas ........... 28 Increasing Labor Productivity while Improving Human Capital . . 28 Improving Access to Education and Training ............. 29 Reducing Labor Market Rigidities .................... 29 Improving Income Schemes for the Urban Unemployed ....... 29 Improving Information on Effectrive Wages and Income Sources in Urban and Rural Areas ....................... 30 CHAPTER IV - GOVERNMENT SPENDING AND BENEFICIARIES ... ...... 31 A - Education and Training ............................. 31 Progress in Education Indicators ..................... 31 The Distribution of Education Subsidies ................ 32 Education Pricing Policies and their Impact on the Poor ... .... 33 B - Health ....................................... 34 Progress in Health Indicators ....................... 34 Distribution of Health Subsidies ..................... 35 Health Care Pricing Policies and their Impact on the Poor ..... 37 C - Basic Physical Infrastructure ......................... 38 Safe Water Supply ............................. 38 Electricity ................................... 39 Housing .................................... 40 D - Social Assistance Programs .......................... 41 Direct Transfers ............................... 41 Consumer Subsidies ............................ 42 E - Conclusions and Recommendations ...................... 43 Education ................................... 43 Health Care ................................. 44 Safe Water Supply ............................. 46 Housing .................................... 46 Direct Transfers ............................... 47 Consumer Subsidies ............................ 47 Bibliographical References ...................................... 48 List of Tables Table I.1 : Selected Economic Indicators Before and During Reform, 1980-93 ..... 2 Table I.2 Composition of Disposable Household Income, Selected Years, 1985-92 . . 8 Table 1.3 : Per Capita Consumption Expenditures, 1985 and 1990 .... ........ 9 Table II.1 Poverty Lines and Poverty Rates, 1985-90 .................... 11 Table II.2 : Cumulative Percentage of Population by Expenditure Group, 1990 ..... 12 Table I1.3 : Other Measures of Poverty, 1990 ....... . . . . . . . . . . . . . . . . . . 13 Table II.4 Social and Economic Indicators of Development, Selected Years, 1975-92 17 Table III.1 : Population and Employment Indicators, 1984-89 . . . . . . . . . . .... . . 20 Table III.2 : Growth in Employment and Wages, 1985-90 .25 Table III.3 : Competitiveness of Manufacturing, 1985-90 .26 List of Figures Figure I. 1 Evolution of Real Government Expenditures per Capita .5 Figure 11.1: Poverty Incidence Curves for Urban and Rural Areas .13 List of Boxes Box 1.1 : Evolution of Selected Macroeconomic Indicators. 3 Box IV. 1 : Institutional Framework of the Health Care System .35 List of Annexes Annex A.1 : The Cost of Capital, 1983-1992 Annex A.2 : Future Prospects for Poverty Reduction in Tunisia Annex B. 1 : Poverty Lines in Tunisia: An Assessment of Old Methodology Annex B.2 : Alternative Method to Construct New Poverty Lines for Tunisia Annex B.3 Different Poverty and Inequality Measures Annex B.4 : Recommendations to Improve Social Data Annex C. 1 Employment Generation and Wage Policy Annex C.2 : Informal Sector in Tunisia Annex C.3 : Agricultural Value-Added and Distribution Policies Annex D. 1 : Education Annex D.2: Health Care System Annex D.3: Water Supply Annex D.4 : Housing Annex D.5: Direct Transfers Statistical Annex REPUBLIC OF TUNISIA POVERTY ALLEVIATION: PRESERVING PROGRESS WHILE PREPARING FOR THE FUTURE EXECUTIVE SUMMARY 1. A Poverty Alleviation Success Story. Tunisia, which has reduced the number of its poor from 40% of the population in the 1960s to about 7%, has pursued poverty alleviation policies that could well serve as a model for other countries seeking similarly dramatic results. Tunisia's success is the product of the Government's long-standing commitment to social development. When structural adjustment became necessary in the mid-1980s, Tunisia carried it off without compromising its commitment to poverty alleviation. While undertaking sustained fiscal adjustment, the Government cut social expenditures less than other kinds of expenditure. Moreover, Tunisia maintained growth during adjustment, primarily through expansion in a few labor-intensive exports activities (textiles, tourism). Thus, the adjustment program, coupled with the pro-poor policy bias, led to a two-edge impact on poverty: increasing mean incomes and reducing income inequality, thus reducing poverty. The rural population in particular has benefited from Tunisia's pro-poor policies. 2. Challenges Ahead. The policies and institutions responsible for these results need no drastic change, but they do need to continue to adapt to a changing economic environment. Despite Tunisia's success in reducing poverty, even through lean times and macroeconomic stabilization, about 13 % of the rural population were still poor in 1990. The challenge of reducing poverty will become greater in the future, as the living standard perceived as the minimum acceptable rises. To continue to improve the living standards of the poor and to further reduce poverty, Tunisia needs to preserve its pro-poor bias in social policy and to expand the access of the poor to basic services in the face of scarce public resources. 3. Recommendations. To further reduce poverty, the report suggests the following: 3 While prudent macroeconomic policies will remain vital for continued growth, the authorities also need to maintain incentive-neutral policies to promote sound, broadly based sustainable growth. 3 Because poverty remains mainly a rural phenomenon, improving the earning capacity of the rural poor must be a priority. To this end, the focus needs to be on increasing poor farmers' security of access to land and developing nonagricultural income-earning opportunities. * Public support for investment in the human capital of the poor needs to be continued. Efforts should concentrate on increasing retention and improving quality in basic education, improving the quality of basic health care services, and expanding infrastructure services to regions in which poverty is more acute. With public resources being more scarce, this report focuses its recommendations on reallocating public resources toward the poor while inducing the better-off to rely on privately provided services, and enhancing cost recovery in the supply of public services. * While direct income transfers contributed to the alleviation of poverty in the past, improved targeting is needed to enhance their effectiveness. * To support poverty monitoring, analysis of welfare changes, and better-informed policy choices, the informational base should be improved and coordination among all ministries concerned with social policies affecting the poor strengthened. - 11 - A. Evidence of Success in Alleviating Poverty 4. Tunisia's progress in reducing poverty has continued beyond the adjustment period of the mid- 1980s. The number of poor declined from about 40% of the population in the 1960s to 7.4% in 1990. Other measures of social welfare have also improved. Primary enrollment is now virtually universal. At about 117% (110% for girls), gross primary enrollment in Tunisia is 12.5% higher than the average in middle-income countries. Primary and secondary enrollment of girls has risen substantially. And over the past two decades, the illiteracy rate has declined by half, to about 32%. In health, progress has also been significant. Infant mortality has decreased from 175 deaths per thousand live births in 1956 to less than 50 in 1992. Birthrates are lower, the population growth rate is about 2% a year, and the total fertility rate has fallen from an average of 7 children per woman in 1956 to less than 3.5 in 1992. Progress in access to basic physical services has been similarly impressive. About 90% of the urban population and 70-80% of the urban poor now have direct access to piped water and electricity. In rural areas, about 62% of the population and 40-50% of the poor have access to potable water, and 40% and 30% access to electricity. 5. These results could be achieved only through sustained efforts and prudent macroeconomic policies. Moreover, it was largely thanks to the quality of public social services that Tunisia was able to continue improving social indicators while pursuing its comprehensive stabilization and structural reform program since the mid-1980s. B. How Economic Development during Adjustment Helped the Poor 6. Tunisia's stabilization and reform program has been successful on two counts: it resulted in sustained growth of about 5% a year; and because the growth has been sufficiently spread over the population, the program helped to reduce the incidence of poverty from 11.2% in 1985 to 7.4% in 1990 and to improve income distribution. These results are all the more remarkable in the light of the significant loss of earnings from oil exports during the period, reflecting both lower international prices and depletion of oil reserves. 7. With real consumption expenditure per capita increasing by 10% during 1985-90, roughly two- thirds of the decrease in poverty can be attributed to growth and the rest to improved income distribution. Further growth in real consumption per capita of about 6% during 1991-93 suggests that the incidence of poverty declined to about 6% in 1993. Several features of the economy during 1987-93 are particularly relevant to poverty: * While undertaking sustained fiscal adjustment, the Government cut social expenditures less than other expenditure categories. From 1987 to 1993/94, social spending rose from 47.5% to 52.5% of gross government expenditures while remaining a constant share of gross domestic product (GDP) at about 18%, and per capita social spending increased by 14% in real terms. Government social expenditures made a decisive difference in smoothing adjustment, reducing poverty, and improving income distribution. * During the period, productivity (value-added per employee) increased by 2.1 % a year compared with near-stagnation in real product wages (0.1 % annual growth). Greater economic openness stimulated growth in labor-demanding sectors (textiles, clothing, leather products, tourism, and simple off-shore activities). Although most of the new jobs generated by this growth were at low wages and for unskilled and temporary workers, they helped keep poverty under control. - iii - * Despite the growth in the economy and in jobs, the labor force grew faster than employment (2.4% a year in 1984-93, compared with 2.2%), and the formal sector unemployment rate rose (from 13.1 percent in 1984 to 15.3 percent in 1989). As part of this increase in formal sector unemployment is behavioral - "queuing" for higher paid jobs - and as many of the unemployed are not household heads, its effect on poverty is uncertain. * On the whole, wage policy -- including minimum wage policy -- has been designed to improve competitiveness and reduce macroeconomic imbalances. Because the policy has resulted in real average wages, in line with minimum wages, growing more slowly than labor productivity, it has not hurt employment. * Although real wages have fallen, disposable household income has increased as a result of increased income from self-employment (entreprises individuelles) and social transfers. The distribution of income also improved as shown by the fall in the Gini Coefficient from 43.4% in 1985 to 40.1 % in 1990. In 1985, the richest decile of the population accounted for about 34% of total consumption expenditures, and the bottom decile for 2.2%. By 1990, the gap had narrowed, with the shares at 30% and 2.3%. Moreover, in 1990, average expenditures in both the poorest and the wealthiest groups were closer to mean expenditures than in 1985. Inequality fell in both urban and rural areas, but remained higher in urban areas. C. Profile of the Poor in 1990 Who Are the Poor? 8. According to the 1990 Household Consumption Survey, about 600,000 people -- or 7% of the population -- have annual expenditures below the poverty line calculated for this study and prepared by the National Institute of Statistics and the World Bank, defined as TD 196 per person, and another 7% have per capita expenditures only 25 % above the poverty line. But these aggregate poverty figures mask important urban-rural and regional differences. Although poverty fell much more rapidly in rural than in urban areas in 1985-90, poverty remains primarily a rural phenomenon. About two-thirds of Tunisia's poor live in rural areas, and the incidence of poverty is appreciably higher in rural areas (13.1 %) than in urban areas (3.5%). There is also a marked disparity in poverty among regions, with the highest incidence in the north-west region, followed by the center-west. Because rural poverty fell more rapidly, urban poverty necessarily rose as a share of the total. Thus, although the incidence of poverty in urban areas remains lower, it is a potential cause for concern in the future and calls for close monitoring. 9. In both rural and urban areas, poor households tend to be above average in size, have a large proportion of children and a high dependency ratio, and tend not to be headed by young breadwinners. Two-thirds of the poor belong to households whose breadwinner has no formal education. School enrollment rates are lower for the poor, and for any expenditure group, they are lower in rural areas. 10. In rural areas, most poor household heads are wage earners, and many also own land and livestock. The rural poor derive their incomes primarily from agricultural activities (own-farm activities and agricultural wage labor), but commonly also earn income outside agriculture. The need to diversify their sources of income stems in part from the inability of small farms to provide a full-time occupation, and in part from the variability of agricultural incomes with fluctuations in rainfall. The incidence of rural poverty is highest among households whose head is employed in construction, followed by services and agriculture. In urban areas, more than 60% of the poor earn their living as wage earners, followed by self-employment in nonagricultural activities. By occupation of household head, the highest incidence of urban poverty is in construction. - iv - Why Are They Poor? 11. Data constraints make it difficult to gain a clear understanding of recent progress in poverty alleviation and to identify the likely determinants of poverty. Nevertheless, a strong association can be established between lack of human capital and poverty in both rural and urban areas. Because of the dynamics of urbanization in Tunisia, the structural causes of urban and rural poverty are closely related, and it seems that there has been a shift in poverty toward urban areas. The urban labor force has grown more rapidly than formal sector employment, and urban unemployment is increasing. How this growth in unemployment is affecting urban poverty is uncertain, however. Driving the growth of the urban labor force are rural-urban migration and the expanding working age population. Pushed out of rural areas by lack of earning opportunities, migrants are attracted to the cities by expectations of employment and the better economic and social services. Poor rural households engaged in productive activities typically have access to agricultural land. But their landholdings are small (no more than 2 hectares), rarely irrigated, and often exhibit low productivity, especially in rainfed areas. The small size of their landholdings limits poor farmers' ability to mitigate crop risks through farm management choices. And lack of land titles and registration constrain poor farmers' ability to obtain formal credit and thus to take advantage of new opportunities. The lack of title and land rights also hampers small farmers' ability to invest in their property. D. Future Challenges in Alleviating Poverty and Improving Living Standards Economic Policies for Sustained, Broadly Based Growth 12. Policies encouraging broadly based economic growth will continue to play an important role in reducing poverty and improving income distribution in Tunisia. The Government should continue its efforts to achieve sustained, labor-intensive growth while ensuring that the benefits of broadly based sectoral growth are distributed across all income groups and between the rural and urban areas. The best policies for achieving this outcome are sector-neutral policies that encourage growth in activities in which Tunisia has a comparative advantage. 13. During the adjustment and the stabilization period, agricultural, off-farm, and other sources of income, including public and private transfers, all contributed to the growth in rural household incomes and to the reduction in poverty. Thus, policy efforts to raise rural incomes and improve the earning capacity of the rural poor should focus on improving both agricultural and nonagricultural earnings. 14. To increase agricultural incomes and help poor farmers adjust to price changes resulting from the planned liberalization of producer prices, policy interventions should facilitate farmers' shifts to more profitable crops and more efficient production methods adapted to their land. Potentially important policy changes would aim at: * Increasing poor farmers' security of access to land by establishing a cheaper and simpler land titling system. This system would have two effects: it would provide poor farmers with incentives to invest in or improve their land, and it would allow those with uneconomically small land holdings to sell the asset. In addition, the Government could ensure that its management policy for the forests and collective lands does not preclude access to these land resources for the poor, while maintaining their environmental sustainability. * Improving poor farmers' access to credit (unsubsidized), to small-scale irrigation, and to information and services provided by extension agents (for livestock and tree crops). 15. Because of the riskiness of agricultural income in rainfed areas, the rural poor who do not migrate to urban areas will continue to devote some of their labor to income-earning activities outside agriculture. Thus, the Government's poverty alleviation policies should recognize that rural households will be unresponsive to programs that focus narrowly on agricultural activities, and give increased attention to developing nonagricultural income-earning opportunities. Existing regional development programs encouraging microenterprises and handicrafts, which remain modest in scale, should continue to be promoted, along with the necessary ingredients for the success of such activities -- expanded, higher quality education, improved rural roads to promote markets, easier access to credit, and essential services such as electricity and potable water. Generating employment and increasing the availability of basic infrastructure and social services in rural areas would also help slow migration to urban areas, and thus perhaps reduce the increase in urban labor force relative to the growth in employment. 16. In urban areas, the expansion of employment in a few labor intensive, mostly export-oriented sectors and the growth of the informal sector may have contributed most to the decline in poverty during adjustment. But the low productivity of these mostly low-skilled, low-wage jobs is a short-term solution to poverty alleviation and should be addressed if progress in poverty reduction is to be continue. 17. To increase the living standards of the population, the course of economic development should be that, once the labor market starts to tighten, the low-skilled jobs eventually are transformed into skilled jobs. Particularly if the progress in alleviating poverty is to be sustained, and the living standard of the population as a whole improved, high value-added activities need to be developed through sound sectoral growth and policies that do not discourage investment in industry and nontourist services. At the same time, investments in the human capital of the poor could continue to ensure that they can keep pace with economic and structural changes and benefit from the new skilled jobs. The informal sector (entreprises individuelles) has the potential to create jobs and to increase incomes, particularly for the low-skilled workers. Improving the informal sector's access to credit would help ensure that it can eventually create higher-value employment. Moreover, the Government could give particular attention to regulations that tend to prevent the growth of productivity of urban entrepreneurs and could provide greater incentives and improved services to meet their needs. 18. Generating employment may require relaxing labor regulations that constrain the functioning of the labor market, such as restrictions on hiring and laying off workers. Such reform of labor regulations might give rise to temporary unemployment. To prevent formal sector unemployment from causing poverty, it would be advisable to put in place a safety net -- introducing limited compulsory unemployment insurance benefits -- revise vocational training programs so that the skills of unemployed youth better match the needs of enterprises, and encourage the development of private employment agencies. In its minimum wage policy, it is advisable that the Government continues its practice of keeping increases below the rate of growth of labor productivity. As industry and agriculture become more competitive, productivity and wages would increase, rendering minimum wages unnecessary except to protect the poor. To make better informed policy choices for improving the living conditions of the needy groups, the Government efforts should primarily emphasize more on improving the required information on effective wages in rural and urban areas and understanding factors which affect the sources of income of the poor in these areas. 19. To reach the poorest among the unemployed, the public works programs in both rural and urban areas should be continued and expanded. Public works programs in construction are cost-effective and self-targeted, and provide an efficient vehicle for transferring income to the poor and an effective safety net, particularly in rural areas. The Government needs to explore the costs and benefits of further improving these programs. It could also explore possibilities for providing more basic, on-the-job training in technical skills and work team supervision through these programs, which would help young - vi - participants move to employment elsewhere in the construction sector. The prospects for expanding and improving the public works programs need further in-depth review. Reallocating Public Resources to Expand Access to Basic Services 20. Despite significant improvement in the living standards of the Tunisian population, including those of the poor, progress in social indicators has not been uniforn across the country, and significant disparities remain between urban and rural areas. In education, the high selectivity of the school system tends to have a social bias, with poorer children much more likely than better-off children to drop out beyond the first cycle of basic education. In health, the share of women receiving prenatal care is still 35% lower in rural areas than in urban areas, and infant mortality is higher in the country's poorest western region, in the south, and among lower-income groups. And in basic infrastructure (potable water and electricity), access to services is still much greater in urban areas despite impressive expansion in rural access. 21. With poverty becoming less widespread and more localized, assistance to the poor must be carefully targeted. To further improve social conditions for the poor and to ensure that social expenditures benefiting the poor can be sustained, subsidies to the nonpoor will need to be reduced. To this end, the Tunisian authorities could usefully reallocate resources to priority sectors and beneficiaries and improve targeting while preserving the sustainability of public spending in the basic social sectors. Sometimes this can be done with little risk of resistance, as when general food subsidies were replaced by targeting through quality differentiation. Other methods need not be so innovative. Encouraging private provision of services, such as schooling and health care, especially at higher levels, and increasing cost recovery for many public services could free resources to provide more basic services to the poor. Expanding basic services in rural areas is particularly important -- to reduce poverty, which is concentrated in rural areas, and to lessen disparities in access to social services between rural and urban areas. 22. Education. To increase the productivity of Tunisians and thus their capacity to contribute to economic growth, there is a need both to improve the quality and to expand basic education. This need is reflected in the slowing of progress in schooling indicators, the persistently high dropout rates at the basic education level, and the stagnation in the rate of transition from the sixth to the seventh year. Hence, the Government policies could aim at: * Intproving quality and reducing selectivity. To reduce the selectivity bias of the education system, it should be ensured that basic education is successfully completed by those between the age 6-15, the first nine years being compulsory for that age group. At the primary and secondary level, in addition to increasing capacity, it is advisable that measures also aim at reducing geographical disparities, improving the distribution of resources (particularly teachers), and promoting good quality private secondary education to reconcile enrollment ambitions with the constraints of the public budget. At the higher education level, measures could also aim at promoting cost-effective curricula and cost recovery, improving budget management, and removing the constraints to the establishment and operation of private institutions. * Providing equal opportunities in schooling. Under compulsory primary education, all children, aged between 6-15, regardless of their parents' social status, should complete successfully the basic education. To this end, the Government should make special efforts to reduce dropout rates at the basic education level among poor children, in remote areas, and among girls, in order to increase overall enrollment rates. The Government could consider several options: (i) offering vouchers or scholarships to cover all out-of-pocket costs associated with attending primary school - vii - (such as for books, writing materials, and uniforms) or even making books and learning materials available free of charge; (ii) offering a subsidy that reflects the opportunity costs of attending school if a child enrolls and stays enrolled in school for the year; and (iii) offering integrated child development programs targeting poor children in rural and urban areas. The possibility of household and geographic targeting could be explored for these subsidies and integrated programs. Producing detailed estimates of the cost and feasibility of introducing vouchers, subsidies, and integrated child development programs at the basic education level would require further analysis, based on the forthcoming data from the 1995 Living Standards Measurement Survey (LSMS). Careful, sustained targeting would be important to prevent any such program from becoming an entitlement for the nonpoor, as has happened in other countries. 0 Introducing partial cost recovery in higher education and promoting private sector involvement. Expanding enrollment in basic education and compensating parents for school-related costs and the opportunity cost of school attendance have budgetary implications that need careful consideration. One way to relieve the pressure on the public budget is to implement partial cost recovery in higher education. Another is to encourage greater involvement by the private sector in providing secondary and higher education. Both measures would result in releasing scarce public resources that now subsidize the better-off and that could be reallocated to benefit the poor. 23. Health Care. Although basic health care services are available to almost all the population, regardless of income, Tunisia's health sector -- particularly the public sector -- faces a serious challenge: to provide more and better-quality health care services more equitably while being able to rely on little increase in public resources. As in the other social sectors, this challenge translates into a need to reallocate public health expenditures to improve the quality of basic health care services. This goal could be achieved by improving the equity and efficiency of operations while reducing the financial burden on Government. Thus, Government policies should aim at: 3 Improving equity. From the distributional point of view, the objective could be to develop more equitable burden-sharing arrangements among the health sector's institutional financiers and, thus, among the population. Besides adjusting -- and combating evasion of -- contributions to benefits, this objective implies reexamining the current targeting system for free and subsidized health care vouchers (AMG I & II), which presently cover about four times as many households as are truly in need of partly or fully subsidized health care. To improve the targeting of the genuinely needy, the characteristics used to identify the poor and the criteria for eligibility for free and subsidized care should be thoroughly revised. And to improve the incentives for the better-off to enroll in contributory insurance schemes, the Government could promote the development of more attractive insurance schemes. * Adjusting the social security contributions to the health ministry's budget. The current practice of allowing insurance funds to underpay for public health services provided to their affiliates constitutes a direct subsidy to Tunisia's middle- and upper-income classes, who could afford to increase their contributions through higher premiums or copayments. This subsidy diverts scarce budgetary resources that could be targeted toward poorer groups for whom the marginal benefit of increased health care is higher. Adjusting the social security funds' contributions to the health ministry's budget is thus essential. * Updating the pricing structure. Whatever revamnping of the financing mechanism is undertaken, further increases in user fees or changes in the user fee structure seem likely in the near future - - and could be advocated. A thorough evaluation of the impact of fee increases requires estimates - viii - of price elasticities and willingness-to-pay measures not now possible with the available information. Further results from ongoing work in this area and from the analysis of the 1995 LSMS data should serve to improve target subsidies to services most in need by the poor and to improve the efficiency of facility use. 0 Promoting the development of private sector. For a more efficient and better quality health care system, the participation of the private sector could be encouraged. Moreover, in response to growing consumer demand for a greater range and choice of health care services, private voluntary insurance schemes could also be developed to help cover health expenditures not borne by public compulsory insurance schemes. 24. Basic Physical Infrastructure. In basic infrastructure, government regional programs, the main vehicle for expanding access to basic infrastructure (potable water and electricity) in rural areas, should continue to be emphasized. Safe water supply and housing are two areas where further expansion of access is needed, along with improved efficiency of provision. * Safe water supply. To support the rehabilitation and expansion of the water supply, in urban and particularly in rural Tunisia, a revision is needed in the pricing structure of the national water supplier, SONEDE. Among the alternatives that should be explored is using the long-run marginal cost of water to set prices, which would allow price discrimination by consumption range on the basis of consumption behavior of subscribers. The resulting increase in water charges would be concentrated in the middle consumption ranges and leave the lowest consumption range almost untouched, providing better targeting of cross-subsidies. A revision of the pricing structure would not be complete, however, if it did not take into account the difference in the cost of delivering water services between rural and urban areas (by extending the cross-subsidization scheme, if needed). * Housing. Current government housing subsidies benefit mostly middle-income Tunisians. To use public funds more efficiently and move toward a private, market-based housing strategy, the Government needs to assist the poor differently. It is advisable that the Government (i) reform the subsidized loan system (FOPROLOS) to better focus resources on the needy; (ii) continue to improve housing conditions in poor areas (both rural and urban) by expanding access to basic infrastructure (potable water, electricity, sanitation); and (iii) encourage the development of a rental market in urban areas by revising the tax treatment of rental income. 25. Social Assistance Programs. Tunisia has used two types of social assistance programs to alleviate poverty: direct transfers and consumer subsidies. In addition, the National Solidarity Fund, a program funded by the civil society and budgetary resources, has been created in 1993 to help areas without basic infrastructure and which do not benefit from regular government and municipal programs and projects. * Direct transfers. Direct transfer programs in Tunisia are pro-poor, and they will continue to play an important role in poverty alleviation. But the targeting of these programs could be improved. Better socioeconomic indicators at the household level are needed. The information provided by Tunisia's Household Consumption Survey could be exploited to further the understanding of the effects of socioeconomic characteristics (family size, age, education, and the like) on poverty. The wealth of information to be derived from the ongoing 1995 Living Standards Measurement Survey (LSMS), combined with that from the Household Consumption Surveys, could be used to identify the characteristics of poverty groups and analyze the impact of policy initiatives on the poor. The improved micro data resulting from this exercise could then be used to refine a - ix - set of indicators for monitoring living conditions, which could in turn form the basis for a poverty alleviation strategy. * Consumer subsidies. In 1990-91, the Government launched an innovative reform of its consumer food subsidy program designed to improve the targeting of subsidies and reduce expenditures with the least possible impact on lower-income groups. The program relies on a self-targeting mechanism based on quality differentiation -- "inferior" and "superior" goods. Although the results have been impressive, the effectiveness of the program could be sharpened in several ways (i) to reinforce self-targeting, the application of both the "inferior" and the "superior" goods approaches could be strengthened by further tailoring the selection of subsidized goods to the consumption pattern of the poor, and liberalizing government controls on unsubsidized goods; (ii) when retail price increases are effected, attempts could be made to ensure that prices increase faster on goods consumed disproportionately by higher-income groups; and (iii) the potential for further reducing costs and liberalizing prices could be exploited. But it must be recognized that any real price increases may have an adverse effect on the welfare and dietary intake of the poor, and the Government should pay close attention to the potential impact on the poor. Data Requirements for Future Poverty Monitoring 26. Targeting policies to the poor requires that the Government continuously identify the poor and understand how they would respond to changing public policies. Although Tunisia's National Household Budget and Consumption Survey is well managed and provides high-quality profiles of consumption and expenditure patterns, better data are needed on social and labor market conditions. In particular, more information is required on: (i) the economic characteristics of the poor and their living conditions; (ii) sources of household income, including transfers; (iii) labor market conditions and wages; (iv) access to social and physical infrastructure and safety net programs; and (v) spatial data on prices. 27. To support and coordinate efforts in monitoring poverty, analyzing welfare changes, and making better-informed policy choices, a "user working group" could usefully be established, drawing representatives from all relevant line ministries. This group could be charged with setting priorities, coordinating poverty related sectoral policies, supervising the timely reporting of poverty monitoring results, rationalizing data collection efforts, and organizing analytical capacity. Moreover, as living standards continue to rise, the perception of what is an acceptable minimum standard will also change. Thus, accordingly a higher poverty line could also evolve in Tunisia, as in many other industrial and higher-income developing countries. I CHAPTER I A SUCCESSFUL MACROECONOMIC FRAMEWORK FOR POVERTY ALLEVIATION In the second half of the 1l980s, the Tunisian authorities undertook a comprehensive stabilization and structural adjustment program. 7his reform program improved the economy's performance and, coupled with the country's persistent efforts in social development since independence, reduced the incidence of poverty from 11.2% in 1985 to 7.4% in 1990. Poverty fell in large part because of the strong economic growth, but also in part because of a shift in income distribution in favor of the poor. The pattern of economic growth produced an increase in employment and in the income of the self-employed and allowed public transfers to needy households to continue. The challenge for Tunisia in the future is to sustain the growth in employment and in per capita income and to ensure that the benefits of growth are distributed across all income groups. As poverty becomes less extensive and more localized, targeting will become increasingly important in the design and placement of antipoverty programs. This chapter provides the macroeconomic background to the poverty assessment in the following chapters. It analyzes the incentive structure and discusses the link between the structural adjustment policies and poverty reduction. A. The Tunisian Stabilization Experience 1.01 Improving social conditions and alleviating poverty have consistently been a focus of attention in Tunisia. Because of this emphasis on social development and the Government's commitment to continue improving social welfare, Tunisia has been able to reduce the incidence of poverty even while undertaking stabilization and structural adjustment. 1.02 Tunisia's emphasis on the social aspects of development since independer.ce has produced a relatively well-designed and robust social system. The authorities expanded the education system to allow almost every child to enter primary school, although it remains relatively selective even at that level. A social security system covers wage earners (in the private and public sector), the self-employed workers as well as the employees in government, and free health care is available to almost all citizens. In addition to these basic social services, there are a number of specific programs aimed at preventing and reducing poverty, including consumer subsidies, public job creation programs, direct monetary transfers, and regional development programs. 1.03 Stabilization and structural reforms became necessary as a result of growing macroeconomic imbalances. During the late 1970s, when oil export earnings were high, Tunisia experienced a period of high public sector investment, rapid increases in public sector wages--which set the pace for private sector wages--and generous subsidies on many basic consumer goods. These policies were continued after 1980, even as oil export earnings began to decline. Deficits on the budget and balance of payments current account grew (to 5.6% and 7.9% of GDP on average during 1980-86), and by 1986 it was clear that Tunisia was heading for a balance of payments crisis. 1.04 To avert a crisis, in 1986 the authorities initiated a comprehensive stabilization and structural adjustment program aimed at reducing the distortions in the economy, improving efficiency, and resuming growth. To increase competitiveness and reduce the current account deficit, the authorities relied on demand management through tighter fiscal and monetary policies and exchange rate adjustment (devaluing the dinar by 10% in August 1986). They reduced public sector investment, restrained consumption by keeping wage increases in both the public and the private sector below the inflation rate, and began cutting consumer subsidies. The devaluation helped to reduce nominal wages in dollar terms, stimulating the competitiveness of exports. Starting in 1986, these stabilization measures were accompanied by a series of structural reforms: liberalization of import restrictions, price controls, and most interest rates, and measures to improve the functioning of markets, including implementing tax reforms, abandoning requirements for investment and credit authorizations, introducing a unified investment code, and undertaking major financial sector reforms. - 2 - 1.05 The stabilization program was largely successful, and the economy recovered quickly (Table 1.1 and Box 1.1). Despite lower investment, GDP growth was higher in 1987-93, at 4.7% a year, than during 1980-86, at 3.6% a year. Supported by strong demand from the European Union (EU), export performance, notably in manufactures and tourism, was a crucial element of this growth, compensating for the loss of earnings from oil. The exchange rate policy and a fall in investment narrowed the current account deficit from 8% of GDP in 1986 to 4% in 1993. The lower current account deficits, coupled with steady export-led growth and a prudent external debt strategy, led to systematic improvement in debt indicators. Inflation averaged around 7%, and the current account convertibility began in 1993. Table I.1: Selected Economic Indicators Before and During Reform, 1980-93 (percent) Indicator 1980-86 1987-93 GDP growth 3.6 4.7 Agriculture 3.6 5.4 Manufacturing 6.4 6.6 Textiles 4.3 9.4 Construction 1.2 5.4 Services 4.0 4.6 Tourism 0.6 4.4 Population gtowil: 2.4 2.2 Growth iil real per a::pita private 1.7 2.1 consumption Investment/GDP 29.3 23.2 Gross domestic savings/GDP 21.0 19.7 Current account deficit/GDP -7.9 -3.2 Central government net budget deficit/GDP -5.6 -3.7 Real growth in manufacturing exports 5.8 10.0 Real growth in tourism exports -0.9 4.6 GDP deflator growth 8.8 6.8 CPI growth 9.2 7.1 Real effective exchange rate (1980=100) 95.3 67.9 Total debt/GDP 51.0 63.3 a Gazoduc and Miskar are not included. Source: World Bank data. 1.06 The Government reduced the net budget deficit from 5.3% of GDP in 1986 to 2.6% in 1993 by cutting budgetary investment and restraining the growth of recurrent expenditures through limits on government wage increases and reductions in consumer subsidies. As a result of these measures, government expenditures fell from 37.4% of GDP in 1986 to 30.1% in 1993. At the same time, government revenues, following the tax reform and the decline in oil revenues, fell from 32.1% of GDP in 1986 to 27.5% in 1993. The bulk of government revenues came from indirect taxes, and a quarter was from nontax sources, such as petroleum revenues; only a small percentage accrued from direct income taxes. - 3 - Box: 1:1 Evolution of Selected Macroeconomic Indicators 125 ~ ~ ROi GNP per rt consumption p.c. 110 105 100 95 105 85 75Releect 65 55 Avemge nominal exchange rate 45 35 s , , , g , , , , ' . . W ~~ ~ ~~~ cl C0 C: C iC 0.02 0 O i / \ . ~~~~~~~~~~~~~~~~~~~~Current account -0.02 defcitGP -0.04 -0.06 -0.08 Cetntrj ovemnment ne budget dettcft/GDP -0.12 -0.12 , , , , . 1soo~~~ ~~~~~~ - I I 1400Retaiutaeii ag 1300 1200 1100. 1000 . 900 - t t Real average 800 anaSMG48hu) 700 Sioo . i~ S00 500 Note: The Household Consumption Surveys years are 1985 and 1990. - 4 - 1.07 Several features of the changes in the economy during the adjustment period are relevant to poverty alleviation: * The Government achieved fiscal adjustment while protecting social sector spending. * The growth pattern was such that economic growth generated jobs at low wages and for unskilled labor, unemployment rose despite the growth, and average real and minimum wages declined during the second half of the 1980s. * Despite the fall in the real wage, disposable household income increased and income distribution improved. B. Public Finance in the Social Sectors 1.08 Social expenditure' choices have a direct impact on households' standard of living, and they also play an indirect role in poverty alleviation. The level and composition of public investment and recurrent public spending are important determinants of the quantity and the quality of the social and economic infrastructure. That infrastructure, in turn, affects poor people's opportunities to earn income and their ability to overcome some of the consequences of poverty, such as obstacles to developing human capital. 1.09 Through its social expenditure choices, Tunisia's Government helped smooth the adjustment process, reduce poverty, and improve income distribution. It did this by targeting assistance programs to the needy and increasing the efficiency of the public provision of social services that enhance the well-being of the population, particularly the poor. In addition, since 1993, the civil society is also playing an increasing role in providing targeted assistance. 1.10 From 1987 to 1993/94, per capita public social spending increased by 14% in real terms, and its share in gross government expenditures rose from 47.5 % to 52.5 % (Figure 1.1). So, while undertaking sustained fiscal adjustment, Tunisian authorities managed to protect the social sectors. That is not to say, however, that Tunisia's economic adjustment did not affect spending on social services: as a share of GDP, social expenditures rose from 18.3% to 20.9% during the most severe period of stabilization, 1987-89, but fell to 18.5% by 1993, mainly because of the reduction in consumer subsidies achieved through better targeting. 1.11 Among basic social expenditures, rural expenditures for potable water and electricity gained in importance, accounting for about 4% of the total social budget by 1993, compared with 2% in 1986.2 Health and social security spending also increased as a share of social expenditures, reaching about 35% by 1993, compared with 31 % in 1986. After 1989, assistance programs were better targeted to poor households in an effort to gradually reduce consumer subsidies of the Caisse Ggnerale de Compensation (CGC). Budgetary transfers to public enterprises (including operating subsidies and financing of losses) showed only a marginal reduction, despite improved cost recovery. 1. Social Expenditures include: education, health care, basic infrastructure, social security, consumer subsidies, direct cash transfers, regional development programs. 2. Rural expenditures include rural programs for potable water and electricity. These programs are: Programme de Developpement Regional (PDR), Programme de Dgveloppement Regional Integre (PDRI), Societe Nationale d'Exploitation et de Distribution des Eaux (SONEDE), and Fonds Special du Developpement Rural Integre (FODERI). - 5 - 1.12 Of course, public spending on social services will not improve the social conditions of the poor unless it actually reaches the poor. The allocation and distribution of selected government social expenditures, and their effectiveness in addressing the special needs of the poor, will be discussed in Chapter IV. Fig. 1.1: Evolution of Real Government Expenditures per capita (1986= 100) 115 110 105 100 95 90 I I I t I 1986 1987 1988 1989 1990 1991 1992 1993 | * Total gross government expenditures [ Total gross social expenditures Need to Ensure Sustainability of Public Spending in Social Sectors 1.13 In the near future, unless resources (including new funds) are reallocated to priority sectors and beneficiaries, the Tunisian authorities are likely to face a serious challenge in supporting the living standards of low-income groups while preserving the sustainability of public social sector spending. The demographic pressure (due to population growth and changes in the population's age composition) and structural changes affecting the economy's labor absorption call for at least maintaining social expenditures at recent levels if efforts to protect the welfare of low-income groups and reduce poverty are to be sustained. And the present macroeconomic constraints call for higher efficiency on the expenditure side. 1.14 Several trends affecting the need for social expenditures are cause for concern. If Tunisia's economy is to maintain its competitiveness, capital expenditures in the social sectors will need to grow. The increasing demand for secondary and higher education may require higher spending on education as a share of GDP. And user charges fall far short of providing reasonable cost recovery in many social services, and cost recovery may become increasingly necessary in order to maintain poverty-related social expenditures (see Chapter IV). C. Growth, Employment, and Wages Changes in Production Factors and Prices 1.15 Economic growth in 1987-93 was based on efficiency gains, prompted in large measure by the greater openness of the economy. Structural reforms, through changes in incentives, promoted growth in labor- - 6 - demanding sectors, especially those oriented to export markets, with low investment. The cost of capital, deflated by the average cost of labor (average wage in manufacturing), rose from an index of 100 in 1983 to 165 in 1993, increasing the incentive to hire workers.3 The fastest-growing sectors in the second half of the 1980s were relatively labor-intensive ones with low-skilled jobs--textiles, clothing, leather, and tourism. 1.16 Adjustment also shifted the relative price structure for traded and nontraded goods. After moving in favor of nontraded goods during 1983-86, relative prices moved in favor of tradables in 1987-93. The relative price of construction (measured by the implicit value added deflator in the national accounts) declined during 1983-93. Relative terms of trade in agriculture improved slightly during the entire period. The relative price index for manufactures increased from 89.8 in 1983 to 101 in 1993, reflecting the higher profitability of traded goods production. The relative price of services changed only slightly during the period. Employment Generation 1.17 During the adjustment period, most new jobs generated were at low wages and for unskilled and temporary workers. The expansion of low-wage jobs was driven by investment and growth in sectors requiring little skill and yielding low value added, such as tourism and simple off-shore activities.4 Employment creation also responded to relative factor prices. There was almost no net job creation in agriculture; most of the growth in employment was in manufacturing (about a third of the job creation, notably in textiles) and services (about half of the job creation, notably in tourism and trade). There was also some expansion in civil services. 1.18 Formal sector employment - the only sector for which information is available - grew more slowly than the economy, however, and unemployment in the formal sector continued to increase. As part of this increase in unemployment is behavioral - "queuing" for higher paid jobs - and many of the unemployed are not household heads, its effect on the overall poverty is uncertain. In 1987-93, the population increased on average by 2% a year, and the labor force by 2.4%, while total employment (wage earners and non-wage earners) grew on average by 2.2% a year. Based on the Census and the Employment Survey, unemployment grew from 13.1% in 1984 to 15.3 % in 1989.5 It increased most among the young because too few new jobs were being created to absorb the young workers entering the job market.6 1.19 Urban unemployment also increased. Although urbanization has led to a clear shift of both the labor force and employment to urban areas, the urban labor force grew faster than urban employment, driven by both rural-urban migration and rising urban working-age population. If urban fornal sector unemployment continues to increase, the urban unskilled poor, who are especially dependent on their labor for generating income, will bear the greatest burden. Trends in urban and rural poverty and the causes for not generating enough income are discussed in Chapters II and III.7 3. For the calculation of the cost of capital, see Annex A.1. 4. See World Bank, "Republic of Tunisia: Private Sector Assessment", Report # 12945-TUN, November 1994. 5. The results of the 1994 Census on unemployment rates will be available before the end of 1995. 6. 'I'he unemployment rate is calculated as the percentage of the active population age 18-59 who had not worked in the week before the survey but were looking for work. It is based on the 1984 Census and 1989 Employment Survey . The unemployment rate for the active population age 15 and above was 16.4% in 1984, and 16.1% in 1989. Because the criteria used to define the active population and the employed population have changed in each surveys, the unemployment results should be treated with caution . 7. Aniex C. I provides more detailed discussion on the evolution of employment and wage policies. -7 - Wage Policies 1.20 On the whole, wage policy (including minimum wage policy) has been designed to improve competitiveness and reduce macro imbalances. To undo the excessive wage growth of the late 1970s and early 1980s, the Government slowed wage increases in the public sector which also influenced the evolution of private sector wages. Average wages declined by 0.1 % per year in real terms over 1983-93, with the sharpest declines in 1986-89. The reduction in real average wages, in line with a sharper drop in minimum wages, and the reduction in the overvalued real exchange rate helped bring down inflation and slow the growth of per capita consumption.' Since 1990 real wages have started to climb back, and by 1993, reached the same level in real terms as in 1983. The ratio between minimum and average wages has also declined over the last decade. 1.21 Wages in all sectors are fixed in accordance with scales based on the non-agricultural minimum wage (SMIG). To minimize the negative effect of minimum wages on employment, the Government has kept the rate of increase below the rate of growth of labor productivity. Although labor policies may be causing unnecessary rigidities in labor markets by making dismissals difficult and preventing wage flexibility, they did not impede the adjustment of effective wages in the formal sector. In fact, the decline in the average and minimum wages, and in effective wages, helped to increase demand for labor: based on the findings of the Industrial Surveys, which should be treated with caution, employment grew most in sectors where real wages declined the most. Moreover, because effective market wages exceed the non-agricultural minimum wage, the latter does not seem to be binding, and its low level is irrelevant with respect to poverty alleviation (see Chapter III). D. Adjustment Reduced Inconie Disparities 1.22 During the adjustment, total employment increased without a rise in real wage earnings. In the formal sector, the share of labor (the wage bill) in total value added declined from 47% in 1983 to 41% in 1989 and 40% in 1993. Indeed, labor productivity increased by about 2.1 % a year in 1987-93 (compared with an average increase of 1.0% a year in 1983-86), while real product wages increased by only 0.1 % a year (compared with a decline of 1.2% a year in 1983-86). Therefore, the gains in value-added per worker have exceeded real wage increases, both in the aggregate and in most sectors (including manufacturing, construction, and tourism). 1.23 Despite the stagnation in real product wages, a fall in real wages, and limited growth in employment, disposable household income increased. In 1987-92, real disposable household income grew 4.4% a year on average, compared with 1.1 % a year in 1983-86. This higher growth rate can be attributed mostly to an increase in the number of self-employed (entreprises individuelles), particularly in urban areas, and an increase in social transfers (Table I.2).9 Because self-employment is an important source of income for poor households (see Chapter II), its growth has helped reduce poverty. 1.24 Absolute poverty is bound to decrease whenever there is distribution-neutral economic growth. Thus Tunisia, by ensuring that the gains from economic growth were spread throughout the population and that income distribution became more equitable, was able to reduce poverty despite the decline in labor's share in total value-added and limited growth in employment. 8. See Chapter III and Annex C. 1. 9. Although at the national level the share of wage earners in total employment increased slightly from 64.4% in 1984 to 65.9% in 1989 and stayed constant thereafter, in urban areas wage employment decreased tromil 75.3% to 74.5% of total employment. - 8 - 1.25 With real consumption expenditure per capita increasing by 10% during 1985-90, roughly two-thirds of the decrease in poverty is attributable to the growth in mean consumption, and the rest to improved distribution.'0 If income distribution had not changed over the period, the same rate of growth in consumption would have reduced the poverty headcount index to about 8.5% in 1990, compared with the 7.4% actually achieved." Moreover, with per capita private consumption increasing by about 6% during 1991-93, the incidence of poverty is likely to have fallen to only about 6% in 1993. Based on the current poverty line standards, it is estimated that a moderate increase in the rate of growth in mean consumption of about 2.5% a year, combined with continued improvement in equality consistent with recent trends, could eliminate absolute poverty in Tunisia by the year 2000. Table 1.2: Composition of Disposable Household Income, Selected Years, 1985-92 (percent) 1985 1986 1990 1992 Total disposable incomea 100.0 100.0 100.0 100.0 Wage income' 48.7 46.5 41.8 44.1 Gross operating surplus 44.1 44.3 46.2 45.1 Social transfersc 6.6 7.5 7.8 8.2 Other incomed 0.6 1.7 4.3 2.6 Memo: Total tax burden 11.7 12.2 11.5 12.6 of which: Direct taxes 3.5 3.7 2.8 3.4 Gross operating surplus Individual Enterprises/Gross 44.9 50.5 51.1 n.a. Production n.a. Not available. a Disposable income after taxes. b Wage income after contribution to social security (cotisations sociales). c Prestations sociales. d Interest and dividends and other income. Source: INS, "Les Comptes de la Nation - 1993". 1.26 The distribution of consumption (a proxy for income) improved between 1985 and 1990 in Tunisia, with the Gini coefficient falling from 0.43 in 1985 to 0.40 in 1990 (Table I.3). 12 In 1985, the richest decile of the population accounted for 33.9% of total expenditures, and the bottom decile for only 2.2%. In 1990, this disparity was reduced, with the richest decile accounting for 30.5% and the poorest for 2.3%. Moreover, in 1990, average expenditures for both groups were closer to mean expenditures than in 1985, and consumption 10. Total expenditures is used as a proxy for income. Inequality in Tunisia is analyzed through changes in expenditure shares by decile and measures of inequality in 1985 and 1990. Mean consumption in 1985 was 471 Dinars per person, or 651 Dinars per person in 1990 prices, using the specific CPI (consisting of a basket of food consumed mainly by the poor); mean consumption in 1990 was 716 Dinars per person. 11. For a discussion of future prospects for poverty reduction, see Annex A.2. 12. The Gini coefficient measures inequality. A value of zero represents perfect equality and a value of one perfect inequality (see Annex B.3). - 9 - distribution also improved for middle-income groups (see Table 1.3, and Chapter II, Figure 2.1). Inequality decreased in both urban and rural areas, but remained higher in urban areas.'3 Income distribution by sector, based on the household head's occupation, improved during 1985-90 in tourism, construction, and trade, but worsened in manufacturing. In agriculture and fishing, income was more evenly distributed and remained unchanged during the period.'4 Table I.3: Per Capita Consumption Expenditures, 1985 and 1990 (current Tunisian Dinars) 1985 1990 Decile Urban Rural Total Urban Rural Total 1 202 107 134 306 167 218 (32.6) (36.4) (28.5) (34.4) (36.3) (30.5) 2 269 137 181 407 220 295 (43.5) (46.6) (38.4) (45.7) (47.8) (41.2) 3 320 167 226 501 266 370 (51.7) (56.8) (48.0) (56.3) (57.8) (51.7) 4 378 196 277 591 310 454 (61.1) (66.7) (58.8) (66.4) (67.4) (63.4) 5 441 226 331 688 369 542 (71.2) (76.9) (70.3) (77.3) (80.2) (75.7) 6 523 269 399 812 436 644 (84.5) (91.5) (84.7) (91.2) (94.8) (89.9) 7 632 321 482 965 517 782 (102.1) (109.2) (102.3) (108.4) (112.4) (109.2) 8 799 404 624 1.194 626 976 (129.1) (137.4) (132.5) (134.2) (136.1) (136.3) 9 1.130 543 891 1.649 822 1.356 (182.6) (184.7) (189.2) (185.3) (178.7) (189.4) Means per person 619 294 471 890 460 716 (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) Note: Figures in brackets are consumption as a percentage of the mean. Source: INS Household Consumption Survey; Bank staff estimates. 13. In urban areas, the top decile of the population accounted for 29% of total expenditure in 1990, and the bottom decile for only 2.6%; the percentages were 27.2% and 2.8% in rural areas. The absolute value of expenditure in rural areas is, on average, less than half that in urban areas: rural expenditures averaged about 52% of urban expenditures in 1990, compared with 48% in 1985. 14. Sectoral Gini coefficients for 1990, based on the household head's occupation, are as follows: agriculture and fishing, 35.6% (35.8% in 1985); manufacturing, 37.7% (35.1 % in 1985); construction, 36.9% (38.9% in 1985); trade, 37.2% (40.5% in 1985); and tourism, 34.7% (42.9% in 1985). - 10 - CHAPTER II POVERTY AND SOCIAL CONDITIONS Seven percent of Tunisia's population have per capita consumrlption expenditures below the new poverty line and are considered poor. Anothler 7% have expenditures only 25% above the new poverty line. Poverty remnains primarily a rural phenomenon, with marked disparities among the r.gions. The highest incidence of poverty is in the north-west, followed by the center-west, regions characterized by hilly terrain and by their distance from the dvnamnic economies of the eastern coastal cities. Uniequal access to basic social services adds to the disparities between rural and urban areas. Informiationi on the economic activities of the poor is limited, and more complete and up-to-date knowledge is essential to assist the Government in designing effective policies for attacking poverty and improving targeting programs in the 1990s. This chapter constructs a profile of poverty in Tunisia and discusses recent trends in standards of living, based on the 1990 Household Consumption Survey. It also revieiws data needs for improved poverty monitoring in the fiuture. A. Reduction in Poverty Incidence 2.01 This report classifies people as poor if their consumption expenditures fall below a poverty line representing the cost of a basket of basic food and nonfood consumption needs. Based on Household Consumption Surveys, Tunisia's National Institute of Statistics (INS) has set its poverty lines using a version of the "food share method" since 1980. The INS poverty lines have been anchored to the consumption behavior of the twentieth percentile in 1980. The average food expenditure per calorie was calculated for this percentile, multiplied by the predetermined food energy requirement, and to calculate the poverty line the average food expenditure was then divided by the average budget share devoted to food of that percentile. Based on this method and adjusted by the consumer price index, in 1990, the urban and rural poverty lines were set at TD 278 and TD 139 per person per year and the incidence of poverty had fallen from 7.7% in 1985 to 6.7% in 1990. Based on this method, an important feature of the official poverty lines is that the incidence of poverty has remained higher in urban than rural areas (poverty rate of 5.7% in rural areas, versus 7.3% in urban areas in 1990) and only one third of the poor live in rural areas. The methodology used to construct these poverty lines and their drawbacks are discussed in Annex B.1 " 2.02 To ensure that the poverty lines reflect a similar standard of living across regions, and using the same household consumption data, the National Institute of Statistics (INS) and the World Bank have developed new poverty lines for 1990. In the new methodology, instead of fixing the twentieth percentile as the reference group, a fixed nominal expenditure level is used to define the reference group. Since the cost of food appears to be similar across regions, the reference group should have a similar standard of living across regions (at least in terms of command over food). This choice was guided by a value judgement that the composition of the diet used to define the poverty line should be consistent with the consumption behavior of households in a neighborhood of the poverty line. To define the basic nonfood good component of the poverty line a new criteria has also been used. Based on this method of the Bank the new poverty lines are TD 218 per person per year in urban areas, and TD 185 in rural areas.'6 The poverty indices and the poverty profile in this report are based on these new 15. All poverty line data in this report are in terms of annual per capita expenditures. 16. The new poverty lines may need to be revised once spatial price indices are available from the 1995 Living Standards Measurement Survey (LSMS). In addition, it should be noted that the methodology used in this study does not take into account household economies of scale. Because the methods for estimating equivalence scales are controversial, the per capita measures - 11 - poverty lines, and the methodology used to construct them is discussed in Annex B.2. Based on its willingness to mitigate the weaknesses of the old methodology, the Tunisian Government intends to adopt the coherent methodology of a poverty profile suitable for calculating poverty lines, once the information of the 1995 Household Consumption Survey is available. Table 11.1: Poverty Lines and Poverty Rates, 1985-90 Poverty line Population Poverty line Headcount Core poor share (ID/person/ index millions (%) year) (% poor) Large cities 1985 25.9 179 2.6 0.05 1990 26.6 248 2.8 0.06 Other urban 1985 28.3 151 6.4 0.12 1990 32.9 208 4.0 0.11 Total urban 1985 54.2 158 4.6 0.18 1990 59.5 218 3.5 0.17 Rural 1985 45.8 134 19.1 0.63 1990 40.6 185 13.1 0.43 National 1985 100 142 11.2 0.80 1990 100 196 7.4 0.60 Source: INS Household Consumption Survey; World Bank staff estimates. 2.03 Changes in poverty over time in Tunisia can be examined through a comparison of the Household Consumption Surveys of 1985 and 1990.17 Based on the new poverty lines, the head-count index shows that 11.2% of the population was poor in 1985, and 7.4% in 1990 (Table 11.1), indicating a sizable drop in the incidence of poverty during the period. Poverty fell most during that period in rural areas. 2.04 Based on the new poverty lines, despite the substantial reduction in rural poverty, it still remains appreciably higher than urban poverty (13.1%, compared with 3.5%), and more than 70% of Tunisia's poor live in rural areas. Yet the percentage of the poor living in urban areas increased during 1985-90, rising from 22% to 28%. This phenomenon reflects the dynamism between the rural and urban sectors (see Chapter 111). 2.05 A cumulative breakdown of the rural and urban population into 12 expenditure classes and the share of food in household budgets show how much households differ in the distribution of expenditures and the extent of inequality in expenditures between rural and urban areas (Table 11.2). For example, only 2% of the rural population spends more than TD 1,500 per capita per year, compared with more than 12.3% in urban areas are used in this report. But it seems likely that economies of scale could have an effect, so that the poverty line estimates for larger households could be lower. Thus, the upcoming LSMS could perhaps consider economies of scale in estimating poverty lines for 1995. 17. To estimate new poverty lines for 1985, the new poverty lines suggested for 1990 have been adjusted tor intlationi, using Tunisia's specific consumer price index. - 12 - (15.4% in large cities and 9.7% in other cities). Table 11.2 also shows that the share of food in the budget declines as total expenditure per person increases, though it does so quite slowly for the poor." 2.06 The distribution of consumption is steep near the poverty line. This implies that estimates of the number of poor will be quite sensitive to the choice of poverty line. For example, about half the poor have expenditures only 25 % below the poverty line. And raising the poverty line by about 25 % to about TD 250 per person per year would increase the share of the population in poverty from 7.4% to about 14.1 %. Table I1.2: Cumulative Percentage of Population by Expenditure Group, 1990 Share of Food in Budget Expenditure Groups Rural Urban National Rural Urban in Dinars Large Cities Other Total Urban 1. 0-125 4.1 0.1 0.5 0.3 1.9 61.2 53.1 2. 125-150 6.9 0.3 1.5 1.0 3.4 59.3 51.5 3. 150-200 15.6 1.0 3.3 2.3 7.7 60.4 53.2 4. 200-250 26.1 2.9 7.4 5.4 13.8 58.5 50.7 5. 250-300 37.0 5.7 12.4 9.4 20.6 57.9 51.9 6. 300-400 54.3 14.1 23.2 19.1 33.4 54.3 51.5 7. 400-500 68.2 23.2 35.1 29.8 45.4 51.7 49.2 8. 500-600 77.9 34.9 45.9 41.0 56.0 48.6 47.3 9. 600-750 87.3 49.1 59.8 55.0 68.1 43.5 44.8 10. 750-1,000 93.9 65.9 76.9 72.0 80.9 41.0 41.6 11. 1,000-1,500 98.0 84.6 90.3 87.7 91.9 35.8 36.8 12. > 1,500 100.0 100.0 100.0 100.0 100.0 31.2 27.0 Total/mean 100.0 100.0 100.0 100.0 100.0 46.9 37.5 Source: INS Household Consumption Survey. Other Poverty Measures 2.07 In addition to the incidence of poverty, it is also important to examine the depth and the severity of poverty. For these purposes, the poverty gap index and the poverty severity index are useful (Table 11.3). Poverty measures are discussed in Annex B.3. 18. Although at any given nominal expenditure level, food's share in the budget tends to be higher in rural areas, reflecting differences in consumption habits, relative prices, or activity levels, it is fairly constant among the poor. - 13 - 2.08 The poverty gap index measures the depth of Flgure 2.1: Poverty lncidence Curves for Urban and Rural poverty: it is the ratio of the minimum cost of eliminating Areas, 1985-90 poverty with perfect targeting to the maximum cost with no targeting."9 It is useful to policymakers because its Cunv dp p t> size indicates the potential gains from better targeting. In 9 0 1990, the estimated cost of raising the consumption of 7 - Urban (190O poor people to the poverty line was less than 2% of the -O RrWa (1990) cost of giving every person, regardless of how poor, a 40 ufl. (18) 30 ~~~~~~~~~~~~~RuNI (1905) transfer equal to the poverty line.0 But the poverty gap 20 index is insensitive to the distribution of consumption 1 v among the poor. If income were transferred from a poor so lO 150 20 25 30 NO 4W 45 5W person to someone who was poorer, neither the head s wow0s"13SW palk ty count index nor the poverty gap would change. 2.09 The poverty severity index is a weighted sum of poverty gaps in which the weights are given by the poverty gaps as a proportion of the poverty line. Thus, this index reflects the distribution of welfare among those below the poverty line by giving each a different weight based on his or her degree of deprivation. Because this index is sensitive to inequality among the poor, it is markedly higher for rural than for urban areas: while rural areas account for 41% of Tunisia's population, they account for 78% of the total severity of poverty. The poverty severity index's sensitivity to the distribution of income below the poverty line makes it useful in designing and evaluating the effect of alternative policies. Table I1.3: Other Measures of Poverty, 1990 Poverty measures Urban Rural National Large Cities Other Poverty gap index 0.51 0.86 3.25 1.74 (% of poverty line) Poverty severity index 0.15 0.29 1.21 0.63 (xlOO) Source: World Bank Staff estimates. Robustness of Poverty Reduction 2.10 The choice of a poverty line and of poverty measures is inevitably arbitrary. Thus, it is important to test the robustness of assessments of changes in poverty.21 Plotting the poverty incidence curves (PIC) (the percentage of the population below each possible poverty line) for 1985 and 1990 confirms that poverty is higher in rural 19. The minimum cost of eliminating poverty using targeted transfers is simply the sum of all the poverty gaps in the population - -that is, the cost of bringing the consumption of each poor person up to the poverty line. The maximum cost of eliminating poverty, if policymakers know nothing about who the poor are, would be the cost of giving an amount equal to the poverty line to everybody to make sure that nobody is poor. 20. Similarly, the estimated cost of raising the consumption of those who are just (25%) above the poverty line to the upper poverty line is less than 4%. 21. If the distribution of the poverty indicator for 1990 lies entirely below that for 1985, then poverty has unambiguously fallen. - 14 - than in urban areas and that poverty fell between 1985 and 1990 (Figure 2. 122).* No matter where the poverty line is drawn, then, the conclusions that poverty fell in both urban and rural areas, and that poverty is greater in rural areas are robust. The PICs also confirm that the distribution of consumption has improved for all expenditure groups, including the middle-income groups. B. Who Are the Poor? Regional Profile of Poverty 2.11 Tunisia's population is urbanized, with more than 60% living in areas classified as urban. It is also regionally concentrated, with about 40 % in the district of Tunis and in the governorates of Kairouan, Sousse, and Sfax. Although there are marked differences in poverty among the regions, in all regions, the incidence of poverty is higher in rural areas. Based on the new poverty lines, the overall incidence of poverty is highest in the north-west, followed by the center-west. About 40% of the urban and two-thirds of the rural poor live in these two regions, characterized by hilly terrain and by their distance from the dynamic economies of the eastern coastal cities. The incidence of poverty is lowest in the Tunis District and in the south (see Statistical Annex Tables). The low incidence of poverty in the south is mainly because much of the population in this region earns their livelihood from irrigated oasis agriculture. Demographic and Socioeconomic Characteristics of the Poor 2.12 The 1990 Household Consumption Survey indicates that poor Tunisian households, both urban and rural, tend to be above average in size, to have many children, and to have a high dependency ratio.24 In both rural and urban areas, more than three-quarters of the poor household heads are of working age (between 31 and 60 years old). Poor households tend not to be headed by young breadwinners: only 5-7% of the poor households are members of households with heads age 18-30. Most likely that is because people in that age-group from poor families remain in the homes of their parents and have not yet established independent households. Rural households headed by females are often considered to have a greater likelihood of being poor. But the National Household Consumption Surveys that have partially addressed this issue do not indicate that gender of household head is a significant determninant of poverty.5 Moreover, women in Tunisia have more formal rights (to inherit land, to acquire land in case of divorce, and to pursue education) than do women in neighboring countries (see Statistical Annex Tables). 2.13 Poverty is strongly associated with lack of schooling. Of the poor in both urban and rural areas, regardless of sex, about two-thirds belong to households whose breadwinner has no formal education. In urban areas, this proportion drops sharply as expenditures increase; this drop is less pronounced in rural areas. This 22. The figure gives the PICs in constant 1990 prices, expressed as a percentage of the poverty line for 1990. Thus, multiplying the horizontal axis by 1.96 (the poverty line divided by 100) gives the cumulative distribution functions of real consumption per person in 1990 prices. Multiplying by 1.42 gives the distributions in 1985 prices. 23. For this purpose the nominal distributions were normalized as "welfare ratios", given by expenditures per person divided by the appropriate poverty line. 24. This is a common finding; see M. Lipton and M. Ravallion, "Poverty and Policy," in J. Behrman and T.N. Srinivasan (eds), Handbook of Development Economics, volume 3 Amsterdam: North-Holland, 1993. Also see P. Lanjouw and M. Ravallion, "Are Larger Households Really Poorer?" World Bank, mimeo, Policy Research Department, 1993. 25. Women may be underrepresented as heads of poor households because respondents in the Consumption Surveys often cite a male relative or absent husband as the head of household, even when that individual is effectively absent from household decisionmaking. - 15 - is mainly because in rural areas lack of schooling is much more important for everybody in this area, although the poor are more likely to suffer. In urban Tunisia, the big difference between the poor and the rich is in the incidence of upper secondary and higher education. Nationally, school enrollment rates tend to be lower for the poor and to increase with expenditure levels. The school enrollment rate of the poorest two groups in rural areas is barely half that of the richest urban group. And for any expenditure group, enrollment rates tend to be lower in rural areas than in urban areas. 2.14 The 1990 Household Consumption Survey shows that a large majority of Tunisia's population (about 90% in rural areas), as well as most of the poor in both urban and rural areas, live predominantly in maisons arabes.26 In urban areas, however, this type of housing declines with increases in per capita expenditures. About 15% of the rural poor also live in mud and straw houses (called Gourbi). The poor and rural inhabitants generally live in fewer rooms than wealthier and urban Tunisians. Residence in a house with a bathroom also is highly correlated with expenditures and with urban location: about 30% of urban homes have a bathroom (defined as a room containing both a bathtub and a toilet), compared with less than 2 % in rural areas (see Statistical Annex Tables). Income and Employment Characteristics of the Poor 2.15 Poverty also is correlated with sector of employment and region of residence. The poor belong primarily to households whose main source of iivelihood is wage labor (both agricultural and nonagricultural). The highest incidence of poverty is in the rural construction sector, followed by rural services and agriculture. Construction also shows the highest poverty incidence in urban areas. 2.16 In rural areas, most poor hous,l-Told heads rely on wage labor (both in agriculture and in nonagricultural sectors), though they often hold small quantities of land and livestock assets. The sectoral breakdown of the principal activity of household heads shows the predominance of agriculture as a source of income for the rural poor: for roughly two-thirds of poor rural households, the primary occupation of the main breadwinner is in agriculture; for 14%, it is in construction; and for 12% it is in services. Although the rural poor derive their income primarily from agricultural activities (own-farm activities and agricultural wage labor), most also depend on off-farm income sources, in part because agriculture is not a full-time occupation on small farms and in part because agricultural incomes vary with fluctuations in rainfall. Recent agricultural surveys found that, even though agriculture was the principal activity for about two-thirds of farmers (own-farm and off-own farm activities), for all sizes of farms combined, nonagricultural sources contributed about half as much income as agricultural sources in a year of normal rainfall.27 Although information on sources of income are not available, off-farm income sources are probably a mix of seasonal and informal sector jobs, including self-employment outside agriculture. 2.17 Poor farmers have small parcels of land and show a different emphasis in their crop choice than nonpoor farmers with large parcels. They devote smaller shares of their land to cereals (staples for own consumption) and a larger share to tree crops (cash crops). This pattern is especially evident in the center and south, where on predominantly rainfed farms of less than 2 hectares (ha), tree crops account for four-fifths of the area, barley for less than 10%, and wheat for an insignificant share. Livestock activities, particularly raising small ruminants, 26. "Maisons Arabes" have large, open-air courtyards lined by rooms. 27. Ministry of Agriculture, Enquete Agricole de Base, 1989 and 1991. - 16 - also are important for poor farmers.2" Livestock often is the most important source of agricultural income on small rainfed farms. And on the smallest farms, livestock is a form of saving for difficult periods. Moreover, small ruminants can be cared for by less active members of the household, such as youths and the elderly. Thus, for farmers with small parcels of land, raising livestock is a means to manage the riskiness of agricultural income strearns while making good use of household labor. 2.18 The urban poor in Tunisia are concentrated in peri-urban areas: close to two-thirds live outside large cities. The most important source of income for the urban poor is wage earnings, followed by self-employment. According to classifications based on occupation of household head, more than 60% of the urban population in the lowest expenditure groups live in households whose head relies primarily on nonagricultural wage labor. Self- employment in industry, commerce, and services also is an important source of employment for poor household heads in urban areas. 2.19 Although the correlation between poverty and formal sector unemployment is not high, in both rural and urban areas, unemployment tends to be higher among poor household heads and a larger proportion of houselholds in the lowest expenditure groups are headed by an unemployed person. The 1990 Household Consumption Survey showed that the incidence of unemployment drops as expenditure levels increase. In the lowest expenditure groups more than 5 % of households are headed by an unemployed person, compared with less than 1% in the highest expenditure groups. The surprisingly low incidence of unemployment among the poor household heads can be explained in part by the fact that unemployment hits young job seekers particularly hard (about 70% of the unemployed are under age 30), and young unemployed people who remain in the homes of their parents are not reported.29 The incidence of poverty among unemployed household heads, however, is high: overall, about 22% of unemployed household heads are identified as poor. 2.20 Lack of education is an important characteristic of poor unemployed household heads. Some 70% of unemployed household heads have not attained education beyond the primary level, and 98% have not attained more than secondary school. Among the poor, more than 90% of breadwinners have not attained more than primary education. Thus, it can be assumed that most poor unemployed household heads in rural and urban areas have not gone beyond primary education. C. Progress in Social Iindicators 2.21 The consumption-based poverty measures in the foregoing sections do not adequately account for other dimensions of welfare. For example, per capita consumption expenditure, because it excludes the monetary value of free or subsidized social services, provides only a rough approximation of the distribution of welfare. Welfare is clearly related both to the amount of private goods consumed and to the benefits from publicly provided goods - cducation, health, and basic social infrastructure such as safe water supply and electricity. 2.22 In Tunisia, progress in consumption-based poverty measures has been accompanied by improvements in other measures of living standards. Social indicators and non-income measures of welfare in Tunisia compare favorably with those of countries at higher levels of development (such as Greece, Ireland, Portugal, and Spain). But there are disparities in access to social services between urban and rural areas and among regions. Although the rural poor share many characteristics with the urban poor, all rural Tunisians, particularly the poor, have 28. Livestock are concentrated on small farms. Farms of 0-5 ha have the highest stocking rates, with 11.5 livestock units per hectare, compared with the national average of 3.4 units per hectare. 29. Definition of the unemployment at the time of the Household Consumption Survey of 1990 is different from the one in the Employment Survey of 1989. - 17 - lower access to social services than urban Tunisians. One cause of the poor access is the dispersion and isolation of poor rural households, which lead, for example, to higher costs for connecting them to the electricity grid and for reaching them with water systems. Progress in social indicators and the urban-rural imbalance in access to public services are discussed in Chapter IV. Table 11.4: Social and Economic Indicators of Development, Selected Years, 1975-92 Indicator 1975 1980 1987 1990 1992 Middle-Income Countries, 1992 Income and population GNP per capita (US$) 710 1,280 1,180 1,430 1,720 2,490 Population growth rate (%) n.a. n.a. 2.48 1.91 1.9 n.a. Urban population (% of total) 48 53 54 59 60 62 Health Life expectancy at birth (years) n.a. 63 66 67 68 68 - Female advantage n.a. 1 1.47 n.a. 1.80 6 Maternal mortality rate n.a. 1,000 n.a. 127 n.a. n.a. (per 100,000 live births) Infant mortality rate 121 62 49 n.a. 43 43 (per thousand live births) Population per physician 5,900 3,694 2,162 1,872 1,530 2,020 Population per hospital bed 400 470 472 516 532 n.a. Child malnutrition(% age-group) 21 n.a. 10 8 8 n.a. Immunization DPT n.a. 36 75 93 90 n.a. Measles n.a. n.a. 74 82 80 n.a. Education Adult illiteracy rate (%) -Total 62 54 42 35 32 21 - Female n.a. n.a. 53 44 41 26 Primary School Gross enrollment rate (%) 97 103 116 116 117 104 - Male 116 118 126 123 123 n.a. - Female 78 88 107 109 110 99 Pupil-teacher ratio 40 39 31 28 26 25 Repeater rate (%) 19 19 23 20 21 n.a. Secondary School Gross enrollment rate (%) 21 27 40 45 46 54 - Female 15 20 34 40 42 55 n.a. Not available. Note: For some indicators, data for 1970, 1985, and 1991 are used for the years 1975, 1987 and 1992. Source: World Bank data. 2.23 Tunisia has made continuous progress in social indicators over the past decade, with impressive results. Its primary school enrollment rate is 12.5 % higher than the average for middle-income countries (MI) (Table 11.4). And it has made important improvements in female primary and secondary enrollment. The illiteracy rate has declined substantially, though it remains higher than the average for middle-income countries. This lag is probably due to "catching up" --- older adults remain illiterate, while most of the young are literate. The health status of the Tunisian population also has improved, with the development of both preventive and - 18 - curative health care services and the decline in birthrates. Infant mortality has been reduced, life expectancy has increased, and the crude death rate has fallen at a rate comparable to that of other middle-income countries. 2.24 In housing conditions and access to basic public services, enormous progress has been made overall and in reducing urban-rural imbalances. The National Employment Survey (Enquete Nationale Population-Emploi) estimated in 1989 that nonsuitable accommodations (oukala, gourbis, tents, and the like) accounted for about 2 % of the total in urban areas and more than 9% in rural areas, compared with 3 % and 21 % in 1980.30 In 1992 about 73 % of the total population had access to safe piped water supply. This included the urban population, more than 90% of which -- and about 70-80% of the poor -- had direct access to piped water and electricity. Access to potable water and electricity has also been improving in rural areas. About 65 % of the rural population and about 40-50% of the rural poor had access to potable water, and about 40% and 30% had access to electricity." D. Data Requirements for Future Poverty Monitoring 2.25 Pursuing and fine-tuning the targeting policies for reducing poverty require that the Government continuously identify the poor and understand how they would respond to changing public policies. Although Tunisia's National Household Consumption Survey is well managed and provides high-quality profiles of consumption and expenditure patterns, better data on social and labor market conditions are needed. This report notes many important questions that can only be answered with better data on sources of income and on access to social services and their effectiveness. In particular, more information is required on: (i) the economic characteristics of the poor and their living conditions; (ii) sources of household income, including public and private transfers; (iii) labor market conditions and wages; (iv) access to social and physical infrastructure and the efficiency of public services, including safety net programs; and (v) spatial data on prices. In addition, the experience of other countries has shown that allowing liberal access to unit record data, by increasing the opportunity to study a wide range of economic and social policy issues, leads to better poverty analysis and policy prescription. 2.26 In order to make better-informed policy choices, Tunisia's Government intends to upgrade the Household Consumption Survey instrument, especially in light of the upcoming 1995 Living Standards Measurement Survey (LSMS). Recommendations for improving the social data and poverty monitoring are provided in Annex B.4. 2.27 To support and coordinate efforts in monitoring poverty, analyzing welfare changes, and making better- informed policy choices, a "user working group" could be set up. Drawing representatives from all relevant line ministries and from nongovernmental institutions (academics), this group could coordinate poverty-related sectoral policies and priorities, supervise the timely reporting of poverty monitoring results, rationalize the data collection efforts, organize analytical capacity, and advise on the choice of modules to be added to future surveys. 2.28 Moreover, experience elsewhere indicates that as living standards rise, the perception of what is an acceptable minimum standard also will change. Tunisian authorities may want to consider phasing in a higher poverty line in the future, to supplement - but not yet replace - the current poverty line. Higher poverty lines should be introduce gradually, so that there is a transitional period in which both the lower and the higher poverty lines are used. 30. The proportion of nonsuitable accommodation was about 44% in 1956. 31. The proportion of rural population with access to safe piped water was about 50% in 1986. - 19 - CHAPTER III URBAN AND RURAL POVERTY REDUCTION During the adjustment period in Tunisia, incomes improved for poor households in both rural and urban areas. In rural areas, where the poor have multiple sources of income, agricultural activities contributed only a small part to the income gains. Thus, the recent household income gains and the reduction in rural poverty can be attributed mostly to nonagricultural activities, to emigration, and to other sources of income, such as public and private transfers. In urban areas, the expansion of employment opportunities -- mainly unskilled temporary work and at declining real wages -- in a few labor-intensive sectors and the growth of the informal sector probably contributed most to the decline in the incidence of poverty. Although absolute poverty in Tunisia remains largely rural, the faster decline in rural poverty has resulted necessarily in an increase in the share of urban poverty. Urban poverty is therefore a potential cause for concern in the future and calls for close monitoring. The country's high degree of urbanization means that the causes of rural and urban poverty, as well as the factors contributing to poverty reduction, are closely related. Emigrants are pushed out of the rural areas by lack of earning opportunities in agriculture and attracted to cities by expectations of employment and the better economic and social services. Rural-urban migration, along with rising working-age population, have been the main driving forces behind the rapid growth of the urban labor force during the past decade relative to the growth informal sector employment opportunities. The Government can help slow rural-urban migration by promoting job creation, improving income opportunities outside agriculture, and increasing the availability of basic infrastructure and social services in rural areas. Moreover, it can prevent a deepening of urban poverty and raise the living standards of the urban poor through measures to improve their productivity and by adopting more flexible labor market regulations to help increase the demand for labor. After discussing structural determinants of poverty, this chapter focuses on factors that have boosted the incomes of poor rural and urban households during adjustment. And it explores whetherfuture poverty alleviation efforts can continue to rely on the same factors or whether the Government's strategy needs to be redesigned. A. Structural Determinants of Poverty 3.01 Although the details differ, the structural causes of rural and urban poverty are broadly the same: limited physical assets and limited human capital assets. In turn, the lack of human capital contributes to a lack of income-earning opportunities. Clearly, the causes of poverty are complex, and various factors come into play, but data constraints hamper a more precise understanding of other potential determinants of poverty. Physical Assets 3.02 The urban poor have only their labor as an asset, but poor rural households often have small landholdings. But this land is rarely irrigated, often has poor productivity even when it is rainfed, and is typically less than 2 ha (see Statistical Annex Tables).32 Moreover, although agricultural surveys show that almost all 32. About 85,000 farmers operate farms under 2 ha. If data were available to allow the deduction from this total of the small farms that are irrigated and thus profitable, the remaining number of poor farm households in all likelihood would not differ significantly from the number of core poor in rural areas. - 20 - farms under 2 ha are owned by their operators, property rights are rarely backed by titles and registration.33 Land rental is uncommon in this farm size class, as are sharecropping and other forms of land management. 3.03 Small landholdings limit poor farmers' ability to mitigate crop risk through farm management choices. Farms of 0-2 hectares consist of about two parcels on average, compared with four for farms of more than 10 ha. Although this fragmentation of land can impose time and other costs. on farmers, possessing a number of parcels also permits crop diversification across different soil types and growing conditions and therefore helps reduce risk. Income-Earning Opportunities 3.04 Urbanization has led to a clear shift in both the labor force and employment to urban areas. But the urban labor force has grown more rapidly than formal sector employment, and formal urban sector unemployment is increasing. The available information and sources of data preclude drawing a direct relation between urban formal sector unemployment and urban poverty. Moreover, underemployment is likely to be a more significant problem for the poor, especially in rural areas, where the need to diversify income sources is greatest. Table III.1: Population and Employment Indicators, 1984-89 (average annual percentage rate of growth) Indicator Urban Rural Population growth 3.2 1.7 Natural growth 2.1 2.5 Growth from rural-urban migration 1.1 -0.8 Growth in participation rates 1.5 -3.4 Growth in employment rates -0.3 0.5 Total employment growth 4.4 -1.2 Source: 1984 Census and 1989 Employment Survey; INS. 3.05 Because of the differences in coverage among data sources and the insufficient information, caution should be used in discussing the relation between poverty and unemployment. The Employment Survey is the best source of data on formal sector employment and the labor force. It shows that during 1984-89, the urban labor force grew by 4.7% a year, while the rural labor force declined by 1.7% a year. The growth in the urban labor force can be explained in part by the growth of the urban population -- due in turn to migration from rural areas -- and in part by the increase in the urban working age population (see Annex C.1). During the same period, new formal sector jobs were created in urban areas at the rate of 4.4 % a year (Table III.1), and there was a shift in employment to the urban industrial and services sectors, at the expense of agriculture. As a result, the share of urban employment increased from 55.5 % in 1984 to 62.3 % in 1989. Meanwhile, the share of the poor living in urban areas rose from 22% to 28% during 1985-90. Urban formal sector unemployment increased from 127,200 in 1984 to 207,700 in 1989, while rural unemployment fell from 118,000 to 108,900. Thus, in 1989, 33. Only 14% of farms of 0-2 hectares are properly titled, and 41% have no titles at all. - 21 - the rural unemployment rate (14.8%) was marginally lower than the urban rate (15.6%), and total rural unemployment accounted for about a third of national unemployment.34 3.06 The Household Consumption Survey captures more effectively the impact of informal sector activities. It shows that, although the incidence of unemployment among poor household heads, both urban and rural, is low, about 22 % of unemployed household heads are identified as poor. This survey also shows that the incidence of unemployment drops as expenditure levels increase (see Chapter II). 3.07 Data constraints are most binding for the analysis of underemployment. The available information indicates that underemployment is particularly high in agriculture, textiles, and construction, where work is typically seasonal and temporary and low-skill. In 1993, roughly 1.1 million adults were employed in agriculture --about 380,000 farm heads, 140,000 salaried workers, and 600,000 family workers. Of the salaried and family workers, about two-thirds are engaged for less than half the year. And of the farm heads, more than 40% also derive income from sources other than their farm. As a result workers engaged in agriculture work an average of only 134 days a year in the sector, compared with the full-employment standard of 250 days a year. B. Policies and Factors Contributing to Poverty Reduction Agricultural Price Policies 3.08 The agricultural price policies between 1986 and 1992 could have affected the rural poor both as asset holders and as agricultural wage earners. As asset holders, poor rural households experienced only marginally positive income effects from the changes in agricultural trade and input subsidy policies under the reform program. Agricultural price policies reduced input subsidies and adjusted producer prices to be more in line with world prices. With little access to irrigation, the poorest farmers are affected mostly by prices for rainfed crops and for livestock. Substantial cuts in subsidies for fertilizers, pesticides, animal feed, and irrigation water also had little impact on the incomes of poor farmers because of their low use of these inputs.35 Anecdotal evidence suggests that poor farm households in some areas are improving their agricultural incomes through intensification of livestock activities and better management of tree crops, except for the deep cuts in olive oil producer prices in late 1993. As agricultural wage earners, poor rural households benefited marginally from a small increase in agricultural employment. These effects indicating that the increase in agricultural incomes of rural poor households was only minor, while poverty incidence was reduced in rural areas, suggest that the income improvement is mainly due to nonagricultural activities and incomes. 3.09 The rural poor are net consumers of agricultural products, but since the consumer prices of subsidized goods are set administratively and not directly linked to producer prices, the changes in agricultural price policies have had a moderate effect on the consumption of the poor. For a more detailed discussion of consumer food subsidies, see Chapter IV. 34. In 1989, total national unemployment (among those age 18-59) was 316,000, of which 207,100 were in urban areas and 108,900 in rural areas. See Annex C.l. 35. As discussed in Annex C.3, analyses of simple farm models, representative of poor farm households, support this conclusion. This farm budget analysis is only partial, taking into account the direct effects of price trends on revenues, but not the subsequent behavioral adjustments relating to input use and activity focus. - 22 - Credit Policies to Promote Investment 3.10 Access to credit could be an important instrument for enabling poor farmers to adopt new inputs, improved technology, and new crops. But the formal banking sector has generally considered poor rural households to be poor credit risks and has been reluctant to lend to them. The poor repayment performance and existing indebtedness of many poor rural households are further constraints to lending to them.36 3.11 To improve access to credit for poor rural inhabitants, the Governmnent has introduced many programs and funds to provide credit to poor, small farmers and rural entrepreneurs.37 These targeted government credit programs, managed through the Ministry of Agriculture (MOA), are administered by the National Agricultural Bank (BNA) and funded by donor -- or Government -- financed projects. As in other countries, however, the Government's rural credit programs, when not combined with other development measures (basic rural infrastructure and technical packages to make fanns more viable, such as technology, and extension), have demonstrated limited effects on agricultural development and poverty reduction. Among the existing programs, most effective has been the credit component of the Rural Integrated Development Program (PDRJ), funded by the Fonds Special du Developpement Rural lntegre (FODERI) program.3" This credit program has directed about two-thirds of its support to agriculture, a quarter to microenterprises and handicrafts, and the rest to fisheries. The program has had some impact on improving access to credit for the rural poor: evaluations indicated that of the 37,300 recipients of credit in the first phase of PDRI, some 32,000 had never before access to formal credit. There have also been efforts to promote investments by poor rural households, principally regional projects financed by the Government and donors and targeted at poor communities. In practice, the poor seek access to credit through PDRI programs, since these provide a subsidized rate, and then apply for an investment grant. 3.12 The urban informal (small enterprise) sector also has had limited access to formal credit, but the available information does not allow an assessment of how this constraint might have hampered the growth in self- employment. Partial evidence suggests that informal sector enterprises rely mainly on their own funds for financing. There has been a government program to provide investment financing to small enterprises, but it has proved ineffective. Fonds National pour la Promotion de I 'Artisanat et des petits Metiers (FONAPRA) created to give technical and financial support to independent workers in small enterprises, ended up playing a very limited role in financing investment in the informal sector mainly because of administrative complications. Land Policies 3.13 The unequal distribution of land, the lack of title and land rights, and the fragmentation of the land contribute to the persistence of rural poverty. Lack of title and land rights hampers small farmers' ability to invest in or sell their assets. Land policies could be used to reduce rural poverty in two ways: by improving the security of farmers' existing rights to land and by making more land accessible to poor farmers. Tunisia has 36. Tunisian farmers are primarily small-scale and low-income, and few could be considered bankable. Of the 380,000 farm holdings, 86% are under 20 ha, covering about 41% of agricultural land; most of these are part-time or subsistence operations. The larger farms (above 50 ha) represent only 4% of farmers but cover about 34% of the land. Almost all formal credits are received by landholders with farms over 50 ha. 37. By the 1990s, there were 48 Iines of credit, each with specific agricultural investment objectives and targeted mostly to small and medium-scale farmers. But they are small in scale and the participation of poor households is minor. In 1993, the Government reduced its lines of credit to four by reforming its credit assistance to the households in the poorest rural areas. For more details, see World Bank "Tunisia: National Rural Finance Project", yellow cover staff appraisal report 13922-TUN, February 1995; and Annex C.3. 38. This Government credit program had been managed through the National Agricultural Bank (BNA). - 23 - chosen to pursue efforts to improve tenure security, but with apparently modest results." Land registration and titling remain cumbersome and costly. The inefficiency of this process constrains access to formal credit, since title to land is often required as collateral for bank loans. 3.14 The Government has pursued a policy of privatizing communal land development, but the effects of this policy for the poor are at best neutral, and probably negative. Although no analysis exists on how Tunisia's rural poor have fared under the policy, they often use communal land as a kind of insurance policy, relying on it periodically to supplement their income when they face setbacks in their primary income-earning activities. Thus, privatization of communal land likely means the loss of a means of income stabilization for the rural poor. Private Transfers 3.15 Rural families commonly receive monetary transfers from their family members, mostly male, working in cities in Tunisia or overseas on a seasonal basis or longer. Although there are no reliable estimates of the actual values of such transfers, or of their importance in the incomes of poor rural households, they are thought to be considerable. 3.16 Because of the instability and inadequacy of income from agriculture activities, private transfers are likely to be an important means for stabilizing and diversifying the incomes of poor rural families. Poor rural households relying in part on private transfers are choosing to place family members in labor markets with risks largely unrelated to the risks facing the household's agricultural income stream. A corollary is that policy efforts to improve the incomes of the rural poor through intensification of agricultural activity will face practical constraints. Poor households will choose to devote family labor to such intensification only if the additional income it will bring compares favorably with both the income and the risk diversification benefits to be gained by having family members work outside the agricultural sector. Public Transfers 3.17 The main direct public income transfers to the poor are made through the needy family program (Familles Ncessiteuses), as discussed in Chapter IV. About 60% of the needy families benefiting from this program reside in rural areas and the rest in urban areas. Although data constraints make it difficult to measure the contribution of the program on alleviating poverty, clearly the poor have benefited from the income transfers. In addition, the Government often launches other initiatives during drought years to mitigate income declines among small farmers. For example, in response to the drought of 1993/94, the MOA implemented temporary subsidies, at a cost of more than TD 21 million, affecting, among other things, the distribution costs of imported animal feed concentrates, dry-season irrigation tariffs for cereals, in-kind distribution of inputs for the next growing season, and selected animal health services. Although these interventions undoubtedly help the rural poor stabilize their incomes in drought years, aspects are open to criticism. For example, the provision of livestock feed encourages farmers to maintain herds at current sizes despite the reduced carrying capacity of pasture in drought years, and thus puts at risk environmental resources and the long-run sustainability of farming practices. Employment Programs 3.18 An important source of employment for the poor in Tunisia is the public works programs Chantier (in urban areas mostly road maintenance, sewer cleaning and installation, removal of wastewater, and cleaning of 39. For more details, see Annex C.3. - 24 - public roads; and in rural areas, road work, soil conservation, and forestry activities).' These programs, which provide short-term jobs in an attempt to reduce underemployment and unemployment are a key vehicle for transferring income to the poor. The programs, the largest and most effective employment programs for unskilled workers in Tunisia, have helped to alleviate poverty in both urban and rural areas. The public works programs have been effective in targeting the poor through self-targeting mechanisms (setting wages below the minimum wage and locating in predominantly poor areas).4' They provide an important source of seasonal employment, and often are scaled up in drought years to supplement the declining agricultural incomes of the rural poor. The workers in the public work programs are primarily from agriculture (66%) and construction (28%). Overall during 1987-91, the programs employed on average 75,000 workers and spent about TD 16 million a year, one- third in urban areas and two-thirds in rural areas (about TD 170, roughly US$170, a year per unskilled employee).4 3.19 About 6% of the participants are youths under age 18 and 30% are over 60. Because the programs include almost no training, they provide little opportunity for young participants to transfer into more permanent employment. Therefore, other programs with training opportunities might be more suitable for the young. And for the elderly, direct assistance might be more appropriate if it could be inexpensively targeted. 3.20 Government .rrigation development projects have also created some jobs, thought they contribute a smaller share of rural employment than in the past. During the Seventh Plan (1987-91), projects to establish new irrigation areas and to intensify irrigation brought an additional 61,000 ha into use, but contributed only 1.6% of the total workdays in the agricultural sector. 3.21 The Government also operates programs to promote microenterprises and handicrafts through credit, marketing assistance, and modest amounts of training for small business ventures. These programs, part of the Integrated Rural Development Program (PDRI) and a complement to the public works programs, seek to create sustainable, nonagricultural income-earning opportunities for the rural poor by improving their skills and assets. But the small number of beneficiaries limits the potential impact of these programs. Moreover, because eligibility often requires official registration of businesses and licensing of artisans, the programs have effectively excluded small handicraft and cottage entrepreneurs and landless workers and women in rural areas. Simplifying the eligibility requirements-and rationalizing eligibility and selection criteria could help broaden the programs' impact. Experimental steps in this direction are being taken under the Northwest Mountainous Areas Development Project; these should be monitored and replicated if successful and cost-effective. 3.22 To alleviate poverty and improve the living conditions in the most vulnerable areas, the Government has introduced, since 1993, a specific national solidarity program, Fonds de Solidaritn Nationale (FSN 2626). This program, covering seven years (1993-2000), is financed through central government budget and the civil society (private and NGO). The FSN provides regional development programs and seeks to create jobs in 1102 localities, of which 51 % are located in the north-west and center-west of the country. Given this program is new, the existing information does not allow to assess its efficiency and its contribution to poverty alleviation. 40. There are several other ongoing employment programs, such as employment services to job seekers, a wide range of training programs and insertion programs (FIAP) and a newjob creation program, Programme deDeveloppement Urbain Integre (PDUI). Although these programs are socially valuable, they are not directly targeted toward the poor, and their contribution to poverty alleviation is negligible. For more details, see Ministry of Planning and Regional Development, "Evaluation des Programmes d'Emploi", February 1993; and World Bank,"Tunisia: The Social Protection System", report 11376-TUN, April 1993. 41. Surveys indicated that 74% of the participants in rural areas own land, but on average only 5.8 ha. 42. About 80% of the program costs go to wages for unskilled workers, and the rest to skilled workers and administrative costs. - 25 - Growth in Temporary, Low-Wage, Low-Skill Employment 3.23 During 1986-93, agriculture contributed only marginally to the creation of new wage-earning employment. Nonagricultural activities increased in rural areas, but data constraints do not permit an assessment of rural employment growth outside agriculture. The available information shows that, during this period, most new jobs originated in urban areas. These were primarily low-productivity, low-wage, low-skill, and temporary jobs. In fact, during 1985-90, temporary employment in manufacturing and services increased faster than permanent employment, and the share of skilled labor declined slightly relative to unskilled workers (see Table III.2).43 But it was precisely this growth in employment that has helped to alleviate poverty in urban areas. The rural poor may also have benefited from the growth of such opportunities. One benefit of the availability of more unskilled jobs for poor households is the greater access to varied sources of income. Table 111.2: Growth in Employment and Wages, 1985-90 (average annual percentage rate of growth) Employment Real wages Manufacturing Permanent 1.42 0.07 Temporary 6.14 -3.03 Export-oriented manufacturing Permanent 9.63 0.26 Temporary 13.24 0.27 Textiles Permanent 2.18 -1.01 Temporary 9.44 -4.52 Services Permanent 1.30 -1.83 Temporary 16.44 -1.06 Tourism Permanent 3.37 -1.06 Temporary 19.95 -7.72 Industrial minimum wage (48 h) -0.55 Source: Annual Industrial Surveys; and World Bank staff estimates. 3.24 Despite rigidities in the labor laws, changes in sectoral employment appear to be negatively correlated with changes in real wages, illustrating the growing flexibility in the labor market. The fall in the real wage resulted in an increased total wage bill because more workers were hired at lower average wages and in 43. According to annual Industrial Surveys conducted by the INS in the manufacturing sector, firms increased their hiring of temporary workers by about 25%, but increased the number of permanent workers by only 6%. In services, this phenomenon was even more marked: permanent workers increased by only 5% during the period, while the hiring of temporary workers almost doubled. Because of some sampling and data problems, however, findings based on this source of information should be treated with caution, and only the trends should be used. - 26 - temporary jobs. A large share of new jobs in urban areas arose in textiles and tourism,44 both sectors in which wages for temporary workers fell more than wages for permanent workers.45 Growth in low-wage jobs resulted in lower unit labor costs that, together with a real devaluation of the exchange rate (7% devaluation per year in the real effective exchange rate in 1985-90), helped these sectors to compete internationally (Table III.3). 3.25 Firms' increased hiring of temporary workers, reflecting a concentration in low-skill activities, provided flexibility in adjusting to changing economic conditions. It is also a reflection of labor market rigidities. For example, formal sector hiring and firing regulations and the required severance packages for dismissed workers make it difficult and costly for firms to reduce the number of permanent workers even as they face a deteriorating business environment.' Table III.3: Competitiveness of Manufacturing, 1985-90 (average annual percentage rate of growth) Employment Real Wages Real Output Unit Labor per Worker Cost Total Manufacturing -0.17 0.38 1.00 -8.10 Export-oriented Manufacturing 10.04 -0.01 1.55 -9.95 Textiles and Clothing 3.43 -2.52 3.31 -12.14 Source: Annual Industrial Surveys, and World Bank staff estimates. 3.26 Official minimum wages have played virtually no role in poverty creation or reduction. If the average number (1.2) of working age members of an average-size poor household earned minimum wages (SMIG or SMAG) on a full-time basis, the household would still fall below the poverty line.47 But the effective (or actual) wage for unskilled and temporary workers in urban areas exceeds the industrial minimum wage (SMIG). And in rural areas, the agricultural minimum wage (SMAG) effectively applies to only a small proportion of salaried agricultural workers and does not appear to serve as a benchmark for wages in the rest of the labor market." Anecdotal evidence for recent years suggests that the effective wage for unskilled male agricultural workers has 44. About half the temporary jobs created in manufacturing and services were in textiles and tourism. 45. The increase in temporary and unskilled jobs was driven mainly by the larger drop in wages for these jobs relative to wages in permanent and skilled jobs. Average real wages in manufacturing have been falling by 1.4% a year for permanent and skilled workers, and by 2.5% a year for temporary and unskilled workers. In services, wages for both permanent and temporary workers have declined by 1% a year, and those for unskilled and skilled workers by 6.7% and 3.5% a year. 46. Issues of labor legislation and labor market functioning, although important, are not the concern of this study. However, it should be noted that firms are required to provide severance payments equal to up to six months' wages for workers dismissed after the restructuring of public and private enterprises; in addition to the payroll taxes, imposed by social security, the current labor legislation imposes supplementary charges that raise social charges to as much as 40% of the wage bill for permanent workers in some industries, particularly in construction. 47. The SMIG is around 6TD a day, while the SMAG is about 4.6 TD a day. 48. The agricultural minimum wage (SMAG) is most consistently respected for full-time salaried agricultural workers, who number about 50,000, or less than 5% of the agricultural labor force. The employers' association (UTICA) estimates that only 17% of the urban workers earn the industrial minimum wage. - 27 - consistently exceeded the SMAG, not just during peak demand periods in the cropping cycle. But, based on anecdotal evidence it appears that unskilled temporary female laborers often are paid less than the SMIG in the textiles sector and less than the SMAG in agriculture. Expansion of the Urban Informal Sector 3.27 Partial evidence for 1984-89 suggests that the informal sector may have contributed to the creation of employment for the poor.49 The drop in the share of formal wage employment in total urban employment from 75.3% in 1984 to 74.5% in 1989, coupled with the analysis of disposable household income in Chapter I, indicates expansion in the urban informal sector (entreprises individuelles). Small enterprises not only might have helped absorb the inflow of migrants in urban areas, they also may have grown in rural areas. Partial evidence, discussed in Annex C.2, indicates that small enterprises employ more than a third of urban wage earners. However, data constraints make it difficult to quantitatively assess the role of the informal sector in alleviating poverty. C. Conclusions and Recommendations Improving Agriculture-Based Incomes 3.28 Because most rural poor families own some agricultural land and are involved to some degree in agricultural activities, agricultural policies will continue to have a major impact on poverty, even for those without land. With input subsidies affecting the poor largely eliminated, policies which influence producer prices are of most interest. Over the medium term, such policy will entail further reductions in protection, primarily of cereals, in accordance with GATT agreements, and this will affect poor and better-off farmers alike. 3.29 Helping poor farmers to adjust to the expected real and relative price changes will thus be a critical part of government policy. Poor farmers are less able to adjust flexibly to shifting prices than better-off farmers. Yet even poor farmers have adjusted to recent price policies and are shifting to more profitable crops, although these are mostly tree crops that have not yet begun to produce. Thus, there should be some scope for further improving agriculture-based incomes by facilitating farmers' shifts to more profitable crops and more efficient production. 3.30 Policy improvements in a number of areas would help poor farmers take better advantage of new opportunities. First, access to nonsubsidized credit needs to be improved. Measures to improve credit access should focus on: (i) improving collateral for credit by updating registration of uncontested land titles and by simplifying procedures for registering changes in land title; and (ii) testing and implementing a climatic risk insurance scheme where production is economically viable (this scheme is under development with the Ministries of Agriculture and Finance). Second, access of poor farmers to information and services provided by extension agents needs to be improved, particularly for livestock and tree crop activities. For example the Ministry of Agriculture's livestock vaccination programs rarely reach poor farmers, even though these farmers account for a significant share of the national herd. The costs of reaching poor farmers are high, and innovative institutional approaches will be needed to reach groups of poor farmers at lower costs. Third, some public investments targeting poor farmers are needed, such as in small-scale irrigation infrastructure. 49. The informal sector is defined as small family enterprises with no regular salaried employees, or microenterprises (Entreprises Individuelles). See Annex C.2 for more details. - 28 - Increasing Access to Land Tenure 3.31 The ability of poor farmers to significantly improve their welfare on the basis of agricultural activities will continue to be severely constrained by the small size of their farms. There is some potential for increasing the access of the poor to land through the distribution of state lands, but that would require a reversal of current policy and raise the practical problem of relocating the poor because the land is not located where the poor are concentrated. More practical, and potentially as beneficial, is improving poor people's access to land tenure that they have traditionally used. There is a need for cheaper, simpler, and updated titling of private land. Forest land and collective lands are also arias in which greater attention to the se patterns of the poor is warranted. The poor, particularly in the northwest, rely on state forestland as an important secondary source of income. As the ongoing Forest Development Project has done, the state and forest management need to recognize and manage these traditional uses and ensure that sound environmental management of the forest resources incorporates social benefits for the poor. For collective land, particularly pastureland, the Government needs to reassess its policy of promoting division into individual holdings to ensure that it does not marginalize the poor by reducing their access. Expanding Earning Opportunities in Rural Areas 3.32 Income-earning opportunities in rural areas should be promoted by working to raise the productivity of both agricultural and other economic activities (on-farm and off-farm). The Govermnent's poverty alleviation policies need to recognize that the rural poor will continue to devote some of their labor to income-earning activities in sectors outside agriculture. Rural households will be unresponsive to programs that focus narrowly on agricultural activities, without addressing the riskiness of agricultural income, if the labor requirements make it difficult for the households to maintain other sources of income. 3.33 Tie development of rural income-earning opportunities in sectors other than agriculture should receive increased ; :verniae ut attention. The main vehicle for ehis effort, the PDRI cornponents promoting microenterprises and handicrafts, remains modest in scale. The Government is taking ade4uaie steps to promote rural income-earning opportunities -- increasing the length and quality of education, promoting markets through improved rural roads, and providing such essential services as electricity and potable water. But the difficulty of gaining access to credit remains an important constraint. 3.34 Government public works programs have been an efficient vehicle for transferring income to the poor and an effective safety net, particularly in rural areas. These programs should continue to be used and even expanded. The Government needs to explore the costs and benefits of further improving these programs. It could also explore possibilities for providing more basic, on-the-job training in technical skills and work team supervision through these programs, which would help young participants move to employment elsewhere in the construction sector. The prospects for expanding and improving the public works programs need further in-depth review. Increasing Labor Productivity while Improving Human Capital 3.35 Poverty reduction in rural and urban areas has been driven by the growth of employment in manufacturing and services, especially in export-oriented sectors and, to some extent, in the informal sector. The new jobs generated -- low-value added jobs for unskilled temporary workers at low wages -- have solved the immnediate poverty problems of low-skilled people displaced by the restructuring of the economy or migrating from rural areas. The course of development should be that, once the labor market starts to tighten, these jobs should eventually get converted into skilled jobs. Particularly if the progress in alleviating poverty is to be sustained, and the living standard of the population as a whole improved, high value-added activities need to be developed. - 29 - To this end, any tax and investment incentives should be sector-neutral to encourage sound cross-sectoral growth. At the same time, investment in the human capital of the poor should continue in order to ensure that they can keep pace with economic and structural changes and have the skills necessary to benefit from the new jobs created. The info-mal sector ('ntrepri5es individuelles) has the potential to create jobs, particularly for low-skilled workers. Improving small enterprises' access to credit is a key to ensuring their potential to eventually create higher-value employment. The Government also should give attention to regulations that hamper the growth of productivity among urban entrepreneurs and should provide greater incentives and improved services to meet the needs of these entrepreneurs. Improving Access to Education and Training 3.36 The Household Consumption Survey indicates that poverty is strongly associated with lack of education. Most poor household heads have not gone beyond primary education, and many are functionally illiterate. Expanding primary education can thus play an important part in increasing the opportunities of the poor (see Chapter IV). Moreover, with rising unemployment for unskilled labor, the poor in both rural and urban areas bear the greatest risk, since they are especially dependent on their labor. Thus, besides improving the access of the poor to basic education, vocational training and apprenticeship schemes to improve the skills of poor and unemployed youth (potential future poor household heads) need to be made more relevant to the needs of enterprises.50 A higher skilled labor force would increase labor productivity. Hence investing in the human capital of the poor will help enable them to keep up with economic and structural changes and adapt to the new, higher-skilled jobs. Reducing Labor Market Rigidities 3.37 To ensure that new employment continues to be generated may require relaxation of labor regulations that constrain the functioning of the labor market, such as legal constraints on hiring-and laying off workers and labor policies preventing wage flexibility. The increase in recent years in the hiring of temporary workers, most at low wages, has been in part a response to a rigid labor code that prevents firms from hiring and firing as business conditions change. The hiring and firing constraints have the unwanted effect of reducing the demand for permanent skilled workers. But they are also understandable, because of the absence of unemployment insurance benefits in Tunisia. These restrictions could be gradually eliminated at the same time that a safety net is devised for laid-off workers who might fall into poverty. The Government could consider introducing protection for the unemployed through a limited compulsory unemployment insurance benefit, revising vocational training programs to improve the skills of unemployed youth to match the needs of enterprises, and encouraging the development of private employment agencies. With respect to the minimum wages, it is recommended that the Government continue its practice of keeping increases below the rate of growth of labor productivity. As industry and agriculture become more competitive, productivity and wages should increase, rendering minimum wages unnecessary except to protect the poor. Improving Income Schemes for the Urban Unemployed 3.38 A limited compulsory unemployment insurance benefit to address unemployment in urban areas would not alleviate poverty, but it would support the unemployed and prevent future poverty resulting from unemployment.5' Participation in the scheme should be restricted to employed contributors in the formal sector, as it is in most industrial countries, and the benefits could be for a limited period. But unemployment insurance 50. See World Bank "Tunisia: Private Sector Assessment", report 12945-TUN, November 1994. 51. See World Bank, "Tunisia: The Social Protection System", Report 11376-TUN, April 1993. - 30 - cannot be relied on to provide all support for the unemployed and to alleviate the poverty caused by unemployment. Instead, unemployment insurance for a limited period (to help the unemployed in the formal sector find jobs) should be combined with public works employment (for those who are unsuccessful for a time) and job training (for youths). Because unemployment insurance schemes are complicated to administer, detailed estimates of the cost and feasibility of a contributive unemployment insurance are needed. In addition, the net gain of introducing such a scheme should be compared with the costs of the existing income security mechanisms, notably severance pay. Improving Information on Effective Wages and Income Sources in Urban and Rural areas 3.39 No information is available on effective wages and on the rural-urban wage differential. To make better informed policy choices for improving the living conditions of the needy groups, the Government needs to understand the composition of incomes of the poor and the effective wage trends for rural and urban unskilled labor. Thus, efforts should emphasize on improving information on sources of income and effective wages and understanding factors which affect these for the poor. - 31 - CHAPTIER IV GOVERNMENT SPENDING AND BENEFICIARIES This chapter analyzes the effectiveness of selected government programs providing basic services (education, health care, and basic infrastructure) and social assistance in improving the distribution of resources and addressing the special needs of the poor in Tunisia. From this analysis it draws conclusions about where and how targeting efficiency and cost recovery can be improved in existing programs. The analysis of education finds that prinmary schooling is pro- poor, though dropouts rates are high even at that level. Although disparities exist in health services, almost all of the population, regardless of income, has access to basic health care. But public health expenditures go beyond an apparently strict pro-poor bias, resulting in a system that provides fairly uniform subsidies across all income groups. Despite recent progress in expanding rural infrastructure, basic infrastructure services retain a significant urban bias. The potential for cost recovery could be exploited to provide a sustainable source of financing for services benefiting the poor. In direct transfer programs, there is need for improved targeting. A. Education and Training 4.01 In Tunisia, about 95% of students are enrolled in public schools, which fall under the purview of the Ministry of Education. Since the reform of basic education in July 1991, Tunisia's education system has provided nine years of basic education, instead of six years, and four years of general secondary education, instead of seven years.52 The private sector accounts for the remaining 5% of enrollment. Its role is negligible at the primary level (0.6% of enrollment), where it responds to the needs of the urban upper classes. But at the secondary level, it accounts for 12% of enrollment, serving primarily those who drop out of the public sector. In addition, a highly diversified vocational training system at the basic (second cycle) and secondary levels accommodates large numbers of basic or secondary education dropouts. The capacity of private sector vocational training centers at these levels is equivalent to that of the public sector and is developing rapidly (primarily in services). Progress in Education Indicators 4.02 Since independence, Tunisia has given high priority to education. Education is in principle accessible and free of charge to all individuals at all levels of learning.3 Primary enrollment is virtually universal, and the illiteracy rate (for the age-group 10 years and plus) is about 35%. The gross primary enrollment rate is about 117% (110% for girls).54 In 1990, the illiteracy rate of the 15-19 age group, an indicator of the effectiveness of primary education, was 13% (5% for boys and 22% for girls) in Tunisia, compared with 36% in Egypt, 21% in Morocco, and 12% in Algeria. Illiteracy rates in these four countries were about the same in 1956. 52. Tunisia's education system provides 13 years of schooling before higher education. Basic education consists of a six-year first cycle and a three-year second cycle. The four years of general secondary education consist of a two-year core curriculum and a two-year program of specialization in the arts, experimental science, mathematics, technical studies, and economics and management. General secondary education leads to the baccalaureate examination, which provides entree to higher education. 53. Tunisia is embarking on a wide-ranging Programme National d'Alphabitisation targeting illiterate people between the ages of 15 and 29, especially women in rural areas. 54. In 1990, the mean duration of school enrollment in Tunisia was 8.3 years, compared with 9.6 years in Egypt, 5.1 in Morocco, and 9.1 in Algeria. - 32 - 4.03 In 1993/94, about 90% of children age 6-9 were enrolled in primary school. Enrollment of this age group was higher among boys and in urban areas (91.9% for boys compared with 87.8% for girls, and about 90% in urban areas compared with about 80% in rural areas). Beyond this age-group, children begin to drop out of school: in 1989, net enrollment rates average about 80% for the age-group 10-14 and 45% for ages 15-19. Dropout rates are consistently higher for girls in the first grades (1 to 4) of primary education, but they are higher for boys in grades 7 to 12.55 Grade repetition also is high, resulting in over-aged children in the classroom. Schooling of the group age 6-9 and above has substantially improved since basic education was instituted and the drop-out rate decreased. The overall drop-out rate fell from 7.6% in 1986/87 to 6.9% in 1990/91, to reach 4.6% in 1993/94. Several actions and measures have been taken to further decrease this drop-out rate which, in average, affects more boys than girls: in 1993/94, the rate was 4.8% for boys and 4.3% for girls. 4.04 The high dropout and repetition rates reflect the selectivity of Tunisia's education system.56 This selectivity tends to be socially biased, and poor children are much more likely to drop out than better-off pupils. For example, in 1989, of a hundred poor pupils entering primary education (grade 1), only 20 enter the first year of general secondary education (grade 10), 8 reach grade 13, and only about 3 obt2.n the geneal bazcalaureate. By contrast, of a hundred better-off pupils, 25 reach grade 10, 14 enter grade 13, and 8 obtain their baccalaureate. Moreover, children from disadvantaged backgrounds predominate in the less prestigious nonscience courses of study.57 4.05 Also leading to early drop out are the indirect costs of education (clothing, transportation, and the like) and the foregone income (actual or imputed), given the student's ability to perform at school.58 The quality of teaching and the distribution of education services in the country also contribute to early drop out, though to a lesser extent. The Distribution of Education Subsidies 4.06 Public spending on educatic.. in Tunisia (by the Ministries of Education and Higher Education) has historically been high. It was equivalent to about 6% of GDP in 1992, with almost 90% for recurrent expenditures. Primary education account for 41 % of recurrent expenditures, secondary education for 33 %, and higher education and training for about 18% and 5%. The residual includes administrative overhead (see Statistical Annex Tables). 4.07 BeLween 1986 and 1992, real costs per student increased by less than 1 % in primary education, while they declined sharply in secondary and higher education.59 These trends in unit costs reflect growing enrollment levels in secondary and higher education more than they do changing wage differentials or decisions about teaching 55. In 1989, about 7% of children who attend grades I to 6 drop out before completing the primary level. The dropout rate rises to more than 10% for grades 7 to 13 and reaches 6% in higher education. 56. Of a hundred pupils, only 78 reach grade 6 and only 62 successfully complete primary education, 32 enter grade 7, 23 enter grade 10 (first year of secondary), 12 complete grade 13 (end of secondary), and only 5 obtain the baccalaureate and may enter university. 57. See World Bank, "Republic of Tunisia: Post-Basic Education Study", report 12670-TUN, December 1993. 58. See Ministere de I 'Education et des Sciences, "Rapport relatif it la convention sur l 'elimination de toutesformes de discrimination lI'egard desfemmes", Tunis, 1993. This report cites two surveys (in 1979/80 and 1990/91), in which more than 60% of the students surveyed reported that the direct and indirect costs of education were the main reasons for their leaving school. 59. In principle, recurrent costs should include an estimate of depreciation of fixed capital. Correcting, however roughly, the figures in the text for depreciation does not change the conclusions. - 33 - personnel. And they show that the Government is placing priority on primary education and that it has not increased the allocation of public resources to secondary and higher education in response to the relative increase in enrollment at these levels. In 1992, the cost per student enrolled in higher education was about four times as much as the cost per student in secondary education, which was, in turn, more than twice as much as the cost at the primary level. 4.08 In absolute terms, education subsidies increase with total per capita consumption. According to the 1990 Household Consumption Survey, the per capita subsidy was about twice as large for the richest 10% of the population than for the poorest 10%. As a share of per capita consumption, it is about 30% for the poorest decile and about 6 % for the richest decile. The per capita subsidy averaged TD 52 per year for rural people, about two- thirds of the average subsidy in urban areas (TD 84). On average, benefits accrued most to the district of Tunis, to the south and to the center-east region.' 4.09 Two very different patterns characterize education subsidies: at the primary level, the incidence of per capita subsidies is pro-poor, and at the secondary and higher levels, it is pro-rich. On average, the poorest 10 percent receive more than twice as much subsidy at the primary level as the richest 10% -- and, as a share of per capita consumption, about seven times as much. The key factors driving the pro-poor bias of primary education are the high enrollment rates and the fact that poor families tend to be younger and to have more children. Dropout rates, highest at the end of the primary cycle, do not offset the bias. 4.10 The pattern is reversed in secondary and higher education: the richest 10% of the population receives three times as much subsidy than the poorest 10% at the secondary level, and 20 times as much in higher education. Public spending on secondary schooling and higher education favors the better-off both in absolute terms and as a share of per capita consumption. The pro-rich bias of secondary and higher education is driven mainly by the rapid rise in enrollment rates as incomes rise. Education Pricing Policies and their Impact on the Poor 4.11 Although the Tunisian Government is the major provider and funder of schooling at all levels, the 1990 Household Consumption Survey confirms that families incur nonnegligible costs (for books, writing materials, food in school, private courses) even in subsidized public schools. In fact, the hidden costs of schooling (nonfee, out-of-pocket expenditures by the family and the opportunity cost of time in school) are often large enough to keep poor youths out of school, as the dropout rates show. Moreover, the private costs of attending school are far from evenly distributed.6' Pupils in public schools in urban areas spend, on average, twice as much as rural pupils (i.e., TD 19 compared with TD 8 per capita per year). Books and writing materials account for 49% of total private costs in urban areas and for 58% in rural areas. But the opportunity cost of time spent on 60. Annex D. 1 provides more detailed analysis of who benefits from public spending on education. 61. The full price of education to the family includes the share of school costs that is passed on in the form of tuition and other fees, plus the cost of learning materials, uniforms, and transportation and the opportunity cost of time associated with school attendance. - 34 - school-related activities (travel time to and from school and time devoted to study) is higher in rural areas and for the poor.62 4.12 Whether access to education is equal depends on whether people face the same prices relative to their income. The 1990 Household Consumption Survey indicates that the out-of-pocket monetary costs of schooling vary negatively with income. In urban areas these costs account for 4-5 % of total consumption expenditure at low total expenditure levels and for 1-2% in the higher expenditure brackets. Similarly, in rural areas, the costs average about 4% in the lower expenditure brackets and less than 1 % in the higher bracket. To a large extent, this pattern reflects expenditures on books and writing material; food in school accounts for a small share, especially in urban areas.' B. Health 4.13 Over the past three decades Tunisia has developed a health care system that covers almost all of the population and compares very favorably with those of most countries at comparable levels of development. Although the system has inefficiencies and gaps, its breadth and quality are impressive. It covers almost 90% of the population with reasonably good services and has avoided excessive investment in expensive equipment or procedures. But it needs to be protected from the risks arising from its heavy dependence on the Government budget and to be improved in line with the evolution of demand. The organization and operation of the system's funding mechanisms exhibit features that, in the medium term, may result in a decline in the quality of services and inequitable access to health care. 4.14 The public sector, through the Ministry of Public Health, is the main provider of health care services. It supplies 90% of all hospital beds and 50% of the outpatient services. The social security fund (Caisse Nationale de Securite Sociale, CNSS) operates a parapublic health network whose clinics provide general outpatient services, specialized care, and diagnostic services to private sector workers affiliated with the CNSS and their dependents. The private sector also offers health care services. Though still modest in scale, the private health sector is expanding rapidly (see Box IV. 1). Progress in Health Indicators 4.15 As discussed in Chapter II, the health status of the Tunisian population has improved significantly in the past three decades. The population currently grows at a rate of 1.9 % a year, the total fertility rate has fallen from 62. A recent survey analyzing female dropout rates in primary education provides evidence that in rural areas the distance to the nearest school exceeded 2 miles for half the enrolled population. The same survey indicates that about one-third of the (female) students interviewed would take part in their household's productive activity (mostly agricultural) or earn a salary in local farms. See UNICEF, "Etude sur les causes de l'abandon scolaire primaire de la fille en milieu rural en Tunisie", 1990. 63. These indicators of the affordability of public education underestimate the potential hardship for poor households as long as they take into account the existing special programs designed to assist students from disadvantaged socioeconomic environments. These programs provide (i) free food in schools, (ii) books and writing materials, and (iii) scholarships. In 1993/94, scholarships amounted to TD 6.5 million, almost entirely for secondary education. Full-time and part-time grants cover almost the entire cost of full and halfboarding for 215 days a year (240 days for Lyckes Pilotes). These grants amount to TD 135 (165 for Lycees Pilotes) and TD 68 (83 for Lycees Pilotes) a year. Free food in (primary) schools accounted for TD 2.2 million, mostly from the Programme Alimentaire Mondial. About 300,000 pupils (some 20% of the total), largely in the neediest areas of the country, benefited from that program, which distributes a daily ration for 120 days to pupils selected on the basis of their nutritional status, their families' economic situation, and the distance of their house from school. Eligible children are identified by school teachers, rural health centers, and family planning centers. - 35 - an average of seven children per woman in 1956 to less than 3.5 in 1992,4 and the infant mortality rate has decreased from 175 deaths per thousand live births in 1956 to less than 50 in 1992. But the improvements have not been uniform across the country, and significant disparities remain between urban and rural areas. In 1988, the Enquete Demographique de la Sante, the latest available source of information, reported that although the infant mortality rate for Tunisia averaged about 57 per thousand live births in 1988, it was 50 in urban areas but nearly 64 in rural areas, where 35% fewer women receive prenatal care. Infant mortality was higher in the poorest western region of the country and in the south and among the lower strata of the population.65 Urban- rural disparities are evident in the availability of health care services, which is greater in Tunis and o,her large cities.'6 But it is likely that the recent efforts of the Government to reduce the disparities in the coverage of services, especially in basic health care services, will translate into improvements in the population's health status. The ongoing health survey will provide more up-to-date information on the progress in reducing disparities. Box IV.1: The Health Care Delivery System The public health care sector, under the Ministry of Public Health, provides development guidelines, coordinates sector agencies, and supervises sector activities. Public health facilities are strategically located and classified according to the level of care provided. At the primnary level, the 1,504 basic health centers, 100 district hospitals, and 6 free-standing matemities (maternites autonomes) provide preventive services, basic outpatient care and primary care hospital services. At the secondaty level, 25 regional hospitals, located primarily in urban areas, provide some specialized care and hospital services. The tertiary level encompasses university hospitals (CHU) and specialized care institutes. The CHU and the specialized care institutes located in the largest urban centers (Tunis, Kairouan, Sousse, Sfax, and Ariana) offer high-level services. With the public sector the main provider of health care facilities and programs, the majority of health workers are public employees--60% of medical staff, more than 90% of paramedics, and more than 70% of administrative personnel. In 1991, the public health sector handled about 12 million outpnarent visits (56% to basic health centers, 18% to district hospitals, 11 7% to regional hospitals, and 15% to university hospitals) and about half a n1ilion inpatient admissions (22% by district hospitals, 33 % by regional hospitals, and 45 % by university hospitals). With the average length of hospitalization about four days in district hospitals, six days in regional hospitals, and nine days in university hospitals, public hospitals provided more than 3.5 million days of hospitalization. The public sector is also a provider of services related to the health in the work place. The parapublic health network of the CNSS (Caisse Nationale de S&curitd Sociale) consists of 6 polyclinics and 3 hemodialysis centers located in the major cities. The clinics provide general outpatient care, specialized care, and diagnostic services to private sector workers, affiliated with the CNSS and their dependents. The parapublic network was developed in the mid-1980s in response to the perceived low quality of public health services. The private sector facilities, concentrated in urban centers, account for less than 10% of the beds but a larger share of admissions and hospital days, and they play an increasingly important role in outpatient care. In 1992, the private sector facilities consisted of 1,959 medical offices, 642 dental offices, 968 phartnacies, 366 nursing health centers, 106 laboratories, and 30 hemodialysis centers. There also are 45 private hospitals, of which 22 are in Tunis and 5 in Sfax. The private sector is also a provider of services related to the health in the medium-large companies for their employees. Distribution of Health Subsidies 4.16 Health expenditures in Tunisia were estimated at 5.2% of GDP in 1992, compared with 5-8% in other middle-income countries. In 1992, public sector health care expenditures represented 52% of total health 64. Contributing to these improvements were a higher marriage age and increased education and family planning efforts. 65. The definition of this strata is based on the household head's education, taken as a reasonable proxy for a household's earning potential. 66. For detailed analysis of the geographical distribution of the health care system in Tunisia, see Annex D.2. - 36 - expenditures, with compulsory health insurance funds accounting for about a fourth of that, and private sector financing accounted for the remaining 48%.6' 4.17 The most salient characteristic of the distribution of Ministry of Public Health expenditures is the high share (68% in 1992) for staff remuneration -- some 20% higher than levels considered appropriate in well-run systems elsewhere. In addition, about 50% of operating expenditures are incurred by tertiary-level hospitals, as a result of their high share of hospital beds and their higher-than-average unit operating costs (see Statistical Annex Tables). 4.18 Basic health care services are available to almost all of Tunisia's population regardless of income. The state provides free or subsidized health care to the lowest-income groups through two public schemes: (i) the Assistance Medicale Gratuite, Type I (AMG I), provides free health care to about 117,000 families classified as needy (familles necessiteuses; see section on direct transfers below) by the Ministry of Social Affairs; and (ii) the Assistance Medicale Gratuite, Type II (AMG II), provides subsidized health care to roughly 660,000 families whose annual income does not exceed the industrial minimum wage (SMIG). These families are issued a passbook (carnet d'indigence) for a small annual fee (TD 10 in 1992) and pay reduced fees (ticket moderateur) for care averaging about 12% of the regular fees for outpatient visits and a lump sum for each hospital visit that depends on the hospital and level of health care and averages about 60 % of the regular one-day inpatient charge. Altogether, about 800,000 families, more than 50% of the population, enjoy free or heavily subsidized health care, with 40% of the population benefiting from AMG II alone.68 4.19 The availability of subsidized health care schemes, AMG II, to many who are not really poor, creates multiple coverage for the wage earners, provides strong incentive for the self-employed to evade affiliation with CNSS, and constitutes the major cause of budgetary outlays for public health care. In fact, according to the law, many of the 660,000 households enrolled in AMGII would be eligible for various CNSS schemes (for example, the scheme for the self-employed).' Moreover, since the AMG II system permits beneficiaries to join the program at any time during the year, people tend to enroll and pay contributions only after they become ill. 4.20 For basic health services, AMG I and II participants appear to be the main users of district hospitals and basic health centers, while CNSS affiliates tend to rely on the CNSS out-patient clinics in areas where they are available. For other more costly services in tertiary (university) hospitals, AMG I and II participants account for about 35% of outpatient visits and about 30% of inpatient visits, while the CNSS affiliates account for 40% and 50% of outpatient and inpatient visits. The actual distribution of subsidies among income groups might be substantially more uniform -- and less pro-poor -- than expected. Regular users and affiliates to the Compulsory social insurance schemes enjoy substantial subsidies, and enrollment in the AMG II scheme is, in all likelihood, wider than warranted by the present income distribution.70 Thus, not only the AMG I and AMG II beneficiaries, 67. The predominance of the public sector derives from the system of compulsory insurance. Health insurance is compulsory for all employees in the public and private sectors, and it reimburses mainly for publicly provided services. See World Bank "Republic of Tunisia: The Social Protection System", report 11376-TUN, April 1993. 68. See World Bank "Republic of Tunisia: Hospital Restructuring Support Project", report 9157-TUN, February 1991. 69. In fact, the current situation, meaning the availability of budgetary subsidies, encourages both under-enrollment in the CNSS health insurance schemes and under-reporting of income (and hence partial evasion of contributions). See World Bank "Republic of Tunisia: The Social Protection System", report 11376-TUN, April 1993. 70. This assessment is based on information on the financing of the public health care system and on the distribution of health care activities. For a detailed analysis of the distribution of subsidies for health care see Annex D.2. - 37 - but also the health insurance funds and their affiliates, benefit substantially from the present system of financing public health care.7' 4.21 Further doubts about whether the present public health care system is pro-poor arise from an analysis of private out-of-pocket health expenditures. An analysis of household health budgets by expenditure groups, based on the Household Consumption Survey, shows inappropriate targeting of the needy population and a strong pro-urban bias in the health care system. Although the share of the household budget devoted to health care increases with income, this share is positive even at very low consumption levels and greater in rural areas at all income levels. For the poorest 5% of the urban population, health expenditures account for about 1.2% of total expenditure (2.2% of nonfood expenditures), while for the poorest 5% of the rural population, the expenditure share increases to about 2.1 % (3.5% of nonfood expenditures). For the poorest 15% of the population, health expenditures account for 2.5% of the total budget in urban areas and 2.1% in rural areas. Health Care Pricing Policies and their Impact on the Poor 4.22 Improving both the accessibility and the quality of the health sector continues to be an important goal in Tunisia. Policies to achieve that goal need to be carefully tailored to ensure that more of the benefits of good- quality, affordable health care accrue to the poor. The distributional impact of two policy tools needs to be thoroughly assessed: (i) increasing direct charges to users of public facilities; and (ii) expanding health insurance. 4.23 To assess the impact of changing direct charges to users of health care facilities, the present structure of health care prices needs to be examined in the light of three key principles of pricing policy: (i) cost recovery, (ii) efficiency incentives, and (iii) equity. Cost recovery shows how much user fees contribute toward financing the current system, and thus how much subsidization is needed. It may also suggest appropriate targeting of subsidies and margins for cross-subsidization. For example, revenues from curative activities could be used to subsidize preventive activities, and lower fees could be charged at facilities that primarily serve the poor. Efficiency incentives indicate the ability of a pricing structure to provide incentives for patients to seek care at the appropriate facility. Equity refers mainly to equity of access across income groups. 4.24 An analysis of subsidies for outpatient and inpatient visits shows that the present pricing structure is not necessarily cost-effective, efficient, or equitable (see Annex D.2)."2 Even regular user fees fall far short of recovering costs. The implied subsidy structure often provides no incentives for patients to enter the system through primary health care facilities or lower-level hospitals rather than going directly to higher-level hospitals, which are scarcer and more costly than primary care facilities, and should focus on treating the more severely ill. And the apparent progressivity of the pricing structure is severely limited by the fact that even better-off individuals willing to pay more tend to enjoy substantial subsidies when entering outpatient clinics. 4.25 This subsidy for even better-off people provides further evidence that the Tunisian public health care system is not strictly pro-poor, but instead a badly targeted system providing more or less uniform subsidies to the entire population. Current fees, applied fairly uniformly in health centers and subcenters, imply larger subsidies for lower-income groups, but fees are not the only cost of obtaining medical care. People across all income groups face different travel times, and thus different transportation costs, and different quality of medical 71. In fact, the social insurance funds (CNRPS and CNSS) are: (i) paying less than the full cost of health services provided by the public sector; (ii) subsidizing, in relative terms, parapublic and private providers; and (iii) using extra resources generated through their health insurance schemes to compensate for deficits incurred through other benefit schemes (pensions). 72. Relative subsidies for outpatient visits are nearly constant for AMG 11 beneficiaries, but they increase with the level of health care for direct payers. But relative subsidies for inpatient visits show little variation with the level of health care. - 38 - care. Although no data on travel time to facilities are available, the geographical distribution of health infrastructures and personnel suggests that there is scope to improve the targeting of the public health care system toward the poor. The upcoming 1995 LSMS will provide information on this subject. C. Basic Physical Infrastructure 4.26 Government programs to improve Tunisia's stock of basic infrastructure accounted in 1993 for about TD 175 million (1.2% of GDP, see Statistical Annex Tables). About 42% of capital expenditures under these programs went to rural potable water and electrification programs, 24% to transport subsidies, and 34% to housing programs. Potable water expenditures include subsidies to Societe Nationale d'Exploitation et de Distribution des Eaux (SONEDE), expenses of the Direction Ggnrale du Genie Rural, and support for general programs that include water supply components."3 Electricity expenditures include subsidies to Societe Tunisienne de l'Electricite et du Gaz (STEG) and support to specific programs to enhance rural electrification. This section assesses the effectiveness of public infrastructure expenditures in alleviating poverty, focusing on rural water supply, electricity, and housing policies.74 Safe Water Supply 4.27 As discussed in Chapter II, access to potable water (including piped water and safe water supply) has been improving in both urban and rural areas, including for the poor. According to the 1989 Enquete Nationale Population-Emploi, access to piped water provided by SONEDE increases with living standards. More than 90% of the urban population and about 70-80% of the urban poor have direct access to piped water. In rural areas, about 65% have access to potable water, compared with 51% in 1986. In 1989, more than 54% of the rural population had no direct access to piped water and resorted to public tanks and wells, public fountains or oued. Among those in rural areas with access to piped water, only 12% enjoyed a regular SONEDE connection, more than 27 % fetched water from private tanks and wells, and about 7 % regularly bought water. Among poor rural households, a quarter still rely on unsafe sources (such as springs), a quarter have their own wells, and another quarter rely on public standpipes. The rest rely on SONEDE pipe systems or water purchased from vendors. Rural water conditions are particularly worrisome in the governorates of Zaghouan, Beja, and Siliana, where more than a third of households live far from a water source.'5 4.28 During 1987-92, rural access to potable water has been expanded mainly through investments by SONEDE and the Rural Engineering Service of the Ministry of Agriculture (Direction Generale du Genie Rural). SONEDE services about 32% of the rural population with connections to communal services, and Rural Engineering, which focuses on reaching dispersed households, provides safe water to about 33% of the rural population. The target for the end of 1996 is to reach about 75% of the rural population, with SONEDE servicing about 36% and Rural Engineering about 39%. Most rural households have a strong preference for connection to the SONEDE network, because of its higher quality of water and lower tariffs. But the dispersion and isolation of poor rural households mean that the costs of reaching them with water systems are higher, and the delivery of service to them has been delayed until after the public investment programs provide lower-cost 73. Programs intended to enhance rural water supply and electrification include: Programme de Developpement Regional (PDR), Programme de Developpement Regional Integre (PDRI), Programme Presidentiel, and Fonds National de Solidarite (FNS). 74. Further analysis is also needed for transportation programs, but the available information is too limited for closer scrutiny of these programs. 75. The above evidence is reinforced by 1990 Household Consumption Survey showing a strong positive correlation between household expenditure levels and access to safe water, in both urban and rural areas. - 39 - systems for rural agglomerations. For SONEDE, maintenance and operation costs in rural areas are about two to three times higher than those in urban areas. 4.29 Although SONEDE's expansion in rural areas has a large redistributive income effect among consumners, its current pricing policy may not be sustainable. The average price of water has always fallen short of the long-term marginal cost of production -- by about 40% in the 1980s and 30% in the early 1990s76- requiring an increase in government subsidies to finance investment projects (see Annex D.3 for a more detailed analysis of SONEDE's pricing policy). SONEDE's pricing structure discriminates between consumption ranges and types of users (domestic, industrial, tourism-related and collective), leading to substantial cross-subsidization, but is uniform across the country. Tariffs by consumption ranges are highly progressive, so, in theory, substantial transfers should take place from the highest-income groups (and from productive activities) to the lowest-income groups. The first consumption block (0-20m3 per quarter) was designed for the poorest groups, yet applies to more than 40 percent of consumers, indicating the difficulty of precisely targeting the poor. About 90% of subscribers fall under consumption ranges 1 through 3 (0-70 ni3 per quarter), which are the most heavily subsidized.7 Only one subscriber out of a hundred belongs to the highest consumption range (150 m3 and above). The effect of SONEDE's pricing policy is that productive activities end up heavily subsidizing low-, middle-, and possibly high-income consumers. The progressively higher prices for the higher consumption ranges has cut demand relatively more in the higher brackets and shifted the distribution of subscribers toward the lower consumption ranges (in fact, between 1985 and 1992, the number of subscribers in the highest consumption range fell by half). 4.30 The very low actual progressivity of the water pricing structure is manifested in the differences in the share of water in household budgets among income groups and between urban and rural areas. In urban areas water's share of the budget ranges from 2.5% in the lowest consumption expenditure groups (except the lowest one) to about 1% in the highest expenditure group. In rural areas, water consumption always accounts for less than 0.5% of the budget and the share slowly declines as incomes rise. Clearly, water consumption figures reflect more than households' behavior; they also reflect quantity constraints (lack of piped water systems) in rural areas and in some parts of urban Tunisia. Electricity 4.31 Responsibility for planning, designing, constructing, and operating of power supply systems in urban and rural areas is assigned to STEG, an autonomous public entity overseen by the Ministry of Industry. The 1989 Enquete Nationale Population-Emploi indicates that, although nearly 100% of the urban population has access to electricity, more than 60% of the rural population has no direct access to STEG electricity and rely mostly on oil lamps and other traditional means for lighting. The share of the population lacking direct access to electricity is highest in the poorer groups and regions: it was 80% and above in the governorates of Zaghouan and Madhia, and in Kairouan and Kasserine in the center-western region.79 76. See SONEDE, "Etude Economique sur l'Eau Potable en Tunisie, Phases 1-3', 1993. 77. The water tariff for the third consumption range is 60% higher than the tariff for the first one, and the tariff for the fourth consumption range (covering 7% of subscribers) is 200% higher than that for the third. 78. A preliminary analysis of consumption data suggests that safe (SONEDE) water is a luxury good for the lowest-income strata in urban areas and a necessity for middle and upper classes. In rural areas, the opposite pattern shows up, with income elasticities rising with income. 79. The above evidence is reinforced by the 1990 Household Consumption Survey showing a strong positive correlation between household expenditure levels and access to electricity, in both urban and rural areas. - 40 - 4.32 STEG's pricing structure differentiates between high-, medium-, and low-tension uses. Among low- tension uses, prices are set to discriminate among consumption ranges (economique, intermediaire, and normale) and types of users (domestic, public, agricultural). Tariffs for medium-tension uses also discriminate among types of users, with the aim of subsidizing agricultural uses. Like water tariffs, electricity tariffs are set at highly progressive rates across consumption ranges, leading to some cross-subsidization, and are homogeneous across the country.'t The share of electricity in household budgets declines with income in urban areas but shows a flat profile in rural areas. Thus while subsidies decrease with income in absolute terms, they might not decrease with income in relative terms, leaving distribution unchanged or even worsened. Housing 4.33 According to the 1990 Household Consumption Survey, about 81 % of the population own their dwellings.8' Among renters in urban areas, the poor devote a higher share of their expenditures to rent than the nonpoor. 4.34 To improve the quality of housing and Tunisians' access to housing, the Government operates several public housing programs. These include housing rehabilitation programs (Fonds National d'Amelioration de I'Habitat,` FNAH; and Fonds National de Resorption des Logements Rudimentaires, FNR?LR), and housing finance subsidies to low-income salaried families (Fonds pour la Promotion des Logements Sociaux pour les Fonctionnaires a Bas Salaire, FOPROLOS). The slum rehabilitation programs absorb about 25 % of budgetary housing expenditures and are targeted to poor rural (FNRLR) and urban (FNAH) areas (see Statistical Annex Tables). Although these programs are efficient vehicles for improving the quality of housing, particularly in urban areas, they are completed and their share in social housing expenditures has declined sharply in recent years. 4.35 The rest of the housing funds are allocated to housing finance, through subsidized loans of FOPROLOS, to provide salaried families with modest incomes (less than twice the industrial minimum wage -- SMIG) easier access to home ownership, mainly in urban areas. The program is carried out by Tunisia's major public housing suppliers -- Societe Nationale Immobiliere de Tunisie (SNITS) and Societe de Promotion des Logements Sociaux (SPROLS), private sector developers, and the Banque de I 'Habitat, with budgetary support through FOPROLOS. Loans are provided at subsidized interest rates (5 % a year for terms of 25 years) (see Annex D.4). 4.36 Despite the use of incomes to determine eligibility for housing finance subsidies (FOPROLOS), the pro-poor bias of the present policy is more apparent than real. The distribution of mortgage payments by income brackets, derived from the 1990 Household Consumption Survey, suggests that mortgages (subsidized or unsubsidized) are confined largely, by definition, to middle- to high-income salaried people in urban areas. Recent simulation results confirm that eligibility conditions favor those earning between one and two times the SMIG; subsidies decline sharply for those with incomes above and below that range.83 Moreover, the 1990 Household Consumption Survey indicates that the number of participants in housing purchase schemes is almost negligible and that most are professionals and white collars workers. 80. The basic tariff (economique) in 1993 was TD 0.059 per kwh and applied to about 580,000 household users (out of 1.5 million subscribers). The high tariff (normale) was TD 0.079 per kwh and applied to some 405,000 household users. 81. The limited development of a rental market can be explained largely by the high taxes on rents (24%). 82. The FNAH includes the sanitation programs in urban areas (Programme National de Rehabilitation des Quartiers Populaires, PNRQP). 83. See World Bank "Tunisia: Housing Sector Strategy Note", report 13563-TUN, September 1994. - 41 - D. Social Assistance Programs 4.37 The Tunisian authorities have attempted to alleviate poverty through two types of social assistance programs. One is a system of subsidies on basic items of consumption. The other is a system of direct transfers, in kind and in cash, to the needy (the elderly, the handicapped, schoolchildren, and needy families). Expenditures on these programs have fallen considerably in recent years, mainly because of improved targeting of consumer subsidies on food (from about 4% of GDP in the second half of the 1980s to less than 3% of GDP in 1993). Four-fifths of government expenditures on social assistance programs are devoted to consumer subsidies on food, and about a tenth go to direct transfers to needy families. The rest covers remuneration of relevant government personnel (see Annex D.5). Direct Transfers 4.38 The direct transfer programs provide protection to the hard core poor unable to support themselves.' Targeted on the basis of indicators, these programs include (i) food aid through school cafeteria and food rations for preschoolers (about 300,000 children, see section on education above); and (ii) financial aid to the handicapped (about 5,000 people) and the elderly poor (about 4,700 people). Direct transfers also provide cash assistance to poor families who need income support (about 101,000families necessiteuses in 1992, see section on distribution of health subsidies above)'. The civil society (private sector, national non-governmental organizations (NGO)), also provides additional transfers in kind (food, clothes, school materials, etc.) and in cash to the needy families at religious feasts and school reopenings.86 4.39 Although the number of households benefiting from direct cash transfers has been increasing, coverage is still inadequate, including in the poorest regions -- the north-west and center-west (see Annex D.5). In its first year, 1986, the program covered about 65,000 families, about 81 % of the eligible population. In 1994, coverage had reached 107,000 families, about 72% of the eligible population. If the program had attained full coverage, providing transfers to all those meeting its eligibility criteria, coverage would have reached almost 700,000 people, nearly 100,000 more than the 1990 household survey estimated to be below the new poverty line. Partial comparison of the number of households benefiting from cash transfers and the number of households below the poverty line (as defined at the national level by the 1990 Household Consumption Survey) confirms that coverage under the cash transfer program is not fully satisfactory and that there is margins for improvement.7 4.40 The program direct cash transfers to needy families suffers from the difficulties that typically arise in targeting assistance to the poor using means tests and specific eligibility criteria (see Annex D.5). Administration is complex, and eligibility lists are rarely updated. Even when the lists are updated, coverage is not always extended to those newly identified as eligible. And because of lack of information on the characteristics and determinants of poverty, the eligibility criteria are very general, and entitlement to benefits is not always fully defined. As a result, some of the truly needy do not benefit from the program, while others receive multiple 84. Further details on direct transfers are reported in World Bank, "Republic of Tunisia: The Social Protection System"; report 11376-TUN, April 1993. 85. Along with permanent programs, support to needy families also takes the form of occasional assistance. See Annex D.5. 86. In 1994 the contribution of the civil society was MTD 12.5, representing about half the budgetary transfers. 87. Recent work on a large sample of beneficiaries sheds additional light on the characteristics offamilles necessiteuses and on the design of eligibility criteria to identify them. But the sample (not designed to be statistically significant) consists of only about 31,000 families (30% of the beneficiary families), two-thirds of them living in urban areas. See Social Affairs Ministry, "Etude Preliminaire sur les Families Necessiteuses dans le Gouvernorat de Tunis, Beja, El Kef, Sousse, Sfax, Kebili', May 1994. - 42 - benefits from various social assistance and insurance programs. The lack of distinction between the group-based and need-based approaches also leads to overlap between programs. Direct transfers have been provided to defined groups, without distinguishing between benefits awarded to individuals and those to families. Moreover, the screening process bypasses the transitional poor and the newly poor (poor unemployed and those between jobs). Because of lack of information, the impact of these programs on poverty is not known. To update the eligibility lists and improve the eligibility criteria for targeted transfers, the Ministry of Social Affairs has conducted a survey on all needy families considered eligible in 1993.88 The analysis of the results of this survey, coupled with the results of the 1995 household consumption survey and living standard survey, will permit in the future a better orientation and coordination of the programs to alleviate poverty (such as direct transfers, AMG, consumer subsidies, regional development programs, etc.). It should also be noted that financial constraints have kept cash transfers to needy families consistently below subsistence. Although the average transfer had increased in real terms by about threefold by 1992, rough calculations indicate that it was still below the rural and urban absolute poverty lines.89 Consumer Subsidies 4.41 Consumer food subsidies were introduced in Tunisia in 1970 through the Caisse Generale de Compensation (CGC).90 The subsidies were not explicitly targeted until after 1990. Up to 1990, basic staples - - cereals (durum wheat, bread wheat), edible oils, milk, and sugar -- were available in unlimited quantities at below-market prices to anyone who chose to buy them, and the Government paid the difference between the producer prices and the subsidized consumer prices. This aggregate market-level approach made the subsidies cheap to administer, but poorly targeted and therefore excessively costly. 4.42 The consumer subsidy program has aimed at redistributing income toward the poor and at protecting their purchasing power and nutritional status. To some extent, the universal subsidy program was successful in meeting these goals. It was progressive in relative terms, contributing over five times more to the purchasing power of the poor than to that of the rich as a share of total expenditures.9" In absolute terms, however, the program conferred greater benefits on the rich than the poor -- two times more in 1985 and 1.2 times more in 1990. The subsidies nevertheless made a significant difference for the poor. In 1990, they contributed roughly 9 % to the total expenditures of the lowest income quintile. The subsidies also contributed substantially to caloric and protein intake across income groups. On average, subsidized foodstuffs provided more than 70% of the total caloric intake and close to 80% of the total protein consumption of the poor in 1990.9 88. The 1993 survey of the Ministry of Social Affairs includes 149,000 needy families of which 97,000 families are beneficiaries of the assistance and 52,000 are considered to be eligible. The final results of this survey are not yet available. 89. Care should be taken in determining family-based poverty lines (those that do not equally weight) all the family members as a reference for determining the social assistance cash transfers. 90. For a detailed analysis of consumer subsidies, see World Bank, "Republic of Tunisia: From Universal Food Subsidies to a Self-Targeted Program", Green Cover report 11946-TUN, November 1993. This report is being revised to incorporate the results of the impact of consumer subsidies reform program based on the INS, "1993 Enquete Restreinte sur la Consommation des produits subventionnes". The latest version of the report is published in June 1995. 91. In 1985, food subsidies contributed 5 times more to the total expenditures of the lowest-income group than to those of the highest; in 1990, they contributed 5.4 times more, and in 1993, 7.4 times more. 92. Intake derived from purchased subsidized products only (excludes on-farm consumption). INS, 1990 Household Consumption Survey. - 43 - 4.43 But the high and rising costs of the program, combined with inefficiencies and the substantial leakage to the nonpoor, made an overhaul of the universal system an urgent priority. In 1990-91, the Government launched a medium-term program (supported by the Bank's Structural Adjustment Loan of 1988) to: (i) raise retail prices, (ii) reduce the costs of production of subsidized commodities, and (iii) introduce a self-targeting mechanism for consumer subsidies to minimize the hardship of price increases for the poor. A particularly innovative aspect of the Tunisian reform program is its reliance on self-selection mechanism to improve the targeting of subsidies. Self-selection mechanisms use quality differentiation as means to target subsidies to the poor. "Inferior" products that are available to everyone but tend to be purchased by low-income consumers are subsidized, while "superior" quality products more likely to appeal to higher-income consumers are marketed at unsubsidized prices. 4.44 Introduction of the self-targeting mechanism has helped reduce expenditures on the subsidy program from 3 % of GDP and more than 7% of government expenditures in 1990 to 2% and 6% in 1993. And reforms transformed the program from one that transferred more absolute benefits to the rich than to the poor to one that disproportionately benefited the poor -- 1.2 times more than the rich in 1993. E. Conclusions and Recommendations Education 4.45 Tunisia has invested heavily in human resources, particularly through education and training, and progress is visible in many areas -- from literacy rates to school attendance. Despite remarkable quantitative achievements, noteworthy shortcomings remain. The slowing of progress in primary education indicators, the persistently high dropout rates at that level, and the stagnation in the transition rate from the sixth to the seventh year indicate a need for both quantitative and qualitative improvement of basic education. In particular, the high selectivity in the first cycle of basic education causes several problems. Young people wishing to join the labor force experience difficulties because of their low level of schooling. The development of more expensive and seemingly less effective technical or vocational training to meet the needs of dropouts leads to increased budgetary outlays. And the resulting disparities in human capital lead to higher inequality. 4.46 Improving quality and reducing selectivity at the basic education level not only would enhance equity, but would also increase the economic efficiency of the education system.93 One way to reduce the selectivity bias of the education system is to ensure that the age group 6-15, for which basic education (the first nine years) is compulsory, completes this cycle. At the general education level (primary and secondary), additional measures should also be taken to reduce geographical disparities, improve the distribution of resources (particularly teaching staff), and promote good quality private secondary education to reconcile enrollment goals with the constraints of the public budget. At the higher education level, the imbalance between output and the needs of the labor market, the low internal efficiency of the system, and the inefficient distribution and use of available resources call for more cost-effective curricula, cost-recovery measures, improved budgetary management, more effective governance of universities, and removal of the constraints on the establishment and operation of private institutions.' Although the reform of basic education in 1991 has improved the quality of education and reduced the selectivity of the system, it is necessary to further reduce the selectivity bias of basic education. 4.47 A more efficient and cost-effective education system alleviates poverty only if the relative costs of schooling -- and thus opportunities -- are equal. To provide equal opportunities and ensure that all children successfully complete basic education requires expanding access among those in remote areas, particularly girls, 93. See World Bank, "Republic of Tunisia: Post-Basic Education Study",report 12670-TUN, December 1993. 94. See World Bank, "Republic of Tunisia: Higher Education Restructuring Project", report 9795-TUN, February 1992. - 44 - and reducing dropout rates in order to increase enrollment rates. Possible ways to achieve these goals include: (i) offering vouchers or scholarships to cover all out-of-pocket costs associated with attending primary school (such as the cost of books, writing materials, and uniforms) or even making books and learning materials available free of charge; (ii) offering a subsidy that reflects the opportunity costs of attending school for the needy; and (iii) offering integrated child development programs targeting poor children in rural and urban areas. Two possible mechanisms could be used to deliver these benefits, particularly for primary education of the girls in rural areas. Through household targeting, the costs of schooling could be covered by allowing direct cash transfers to needy families to increase with the number of pupils in the family, with the increment available only if a child enrolls and stays enrolled in school for the year (see the section on direct transfers). Alternatively, geographic targeting could be used to cover entire communities or regions. Geographic targeting is simpler to implement and could be administered and monitored through schools, but it inevitably allows leakage. In addition, in Tunisia's education system, within-group variance is more critical than between-group variance. Thus, the better mechanism would be household targeting. Further analysis, based on the upcoming 1995 LSMS data, would be required to produce detailed estimates of the cost and feasibility of introducing vouchers, subsidies, and integrated child development programs at the basic education level. Careful, sustained targeting would be important to prevent any such program from becoming an entitlement for the nonpoor, as has happened in other countries. 4.48 Subsidizing all school-related expenditures and compensating parents for the opportunity cost of school attendance have budgetary implications that need to be carefully considered and analyzed. From the 1990 Household Consumption Survey it can be inferred that subsidizing nonfee costs for the poorest 5% of the population would require an additional per capita subsidy of about TD 5-10 a year and total additional budgetary outlays of about TD 2 million (at 1993 prices). Extending the subsidy to the entire first decile of the population would more than triple the cost, to about TD 6 million. Although the available data do not allow a full assessment of the budgetary implications of a job-compensating subsidy (opportunity costs of attending school), it seems fair to assume that introducing such a subsidy could easily double the estimates above. How could these subsidies for the poor be financed? One approach would be to implement partial cost recovery in higher education and encourage the development of private secondary and higher education. These measures would release public resources that currently subsidize the better-off and that could be reallocated to the poor. Any additional public funds should be used to expand basic schooling and improve its quality, ensuring that the improvements particularly benefit girls. At current enrollments in higher education, introducing fees to partially recover costs for students not eligible for scholarships or subsidized loans would certainly cover the costs of subsidized programs for the poor.95 Further analysis is required to produce detailed estimates of the cost and feasibility of introducing vouchers and subsidies at the basic education level. Health Care 4.49 Basic health care services are available to almost all of Tunisia's population, regardless of income. Yet Tunisia's health sector, particularly the public health sector, faces a serious challenge: to provide more and better- quality basic health care services more equitably while being able to rely less extensively on public resources. As in the other social sectors, this means improving the equity and efficiency of operations, reexamining the financial burden sharing arrangements to reduce the Government's financial responsibilities, and concentrating public resources on the poorest segments of the population. 95. University students pay almost no tuition fees; all financing of universities come from the government budget. Scholarships are available for students from families with an annual income of TD 1,450 or less. About 50% of students in higher education are eligible for scholarships or subsidized loans. Scholarships amount to TD 40 a month and are reduced for repeaters. Operating costs (library books, classroom supplies, utilities, room and board) account for about 26% of the recurrent budget for higher education. If, say, the richest third of the unsupported students (about 12,000 students) could pay the operating costs, about TD 6 million could be recovered and used for subsidies programs for basic education of the poor. - 45 - 4.50 From the distributional point of view, the objective to be pursued is clear: to allocate the financial burden more equitably among the health sector's institutional financiers and thus among the different segments of the population. Rather than develop a new approach, the Government's strategy is to rationalize and update existing mechanisms. Under this strategy, health care could continue to be financed from three sources: (i) Government resources, supporting the provision of free or subsidized basic health care to the genuinely poor; (ii) social health insurance system funds, covering the health care costs of contributors and their dependents and serving as the main source of health sector resources; and (iii) direct payments for services by nonpoor, uninsured people. 4.51 Besides combating evasion and adjusting contributions to match benefits, rationalizing the health financing system requires several actions. The system for targeting free and subsidized health care vouchers (AMG I and II) could be examined to account for the fact that, of 800,000 beneficiary households, only 202,000 are presently estimated to actually need partly or fully subsidized health care based on the new poverty lines.'6 To improve targeting of the genuinely needy under AMG I, the definition of poverty needs to be thoroughly revised (see section on direct transfers). The eligibility criteria for AMG II also need revision. And developing more attractive insurance schemes would strengthen the incentive for people to participate in the insurance system. 4.52 Also needed is an adjustment of the contributions by the Caisses (CNSS and CNRPS) to the Health Ministry's budget for services provided to its contributors. Subsidizing public health insurance funds by allowing them to pay less than the full cost of public health services constitutes direct financial support to Tunisia's middle and upper classes, which could afford to increase their contributions through higher health care premiums or copayments. This practice diverts scarce budgetary resources that could be targeted to poorer groups, for whom the marginal benefit of increased health care is higher. Preliminary experiments to adjust these contributions are under way; the experiments are based on the available utilization and unit cost data for health care services and involve the restructuring of the public hospitals. 4.53 In addition to revamping the present financing system, increases in both subsidized and regular user fees or changes in the user fee structure seem likely -- and could be advocated -- in the near future. A large-scale evaluation of the impact of fee increases requires estimates of price elasticities and willingness-to-pay measures -- estimates not now possible given the available information. Pending further results and analysis of the 1995 LSMS data, the current price discrimination by level of health care, used since the introduction of the reduced fees (ticket moderateur), could be revised to improve target subsidies to the poor. The existing price discrimination by level of health care, charging lower fees (or higher subsidies provided) at health subcenters and relatively higher fees at health centers and hospital outpatient clinics, needs to be adjusted in such a way that regular user fees could better recover costs. At present the apparent progressivity of the pricing structure is severely limited by the fact that even better-off individuals willing to pay more tend to enjoy substantial subsidies. Under more suitable price discrimination by level of health care, much higher regular fees could be charged at health centers and hospital outpatient clinics, while the existing lower fees would continue to be charged at health subcenters. This revised pricing structure could provide an affordable entry into the health care system through the health subcenter and allow those willing to pay to bypass the health subcenter to go directly to higher levels of care (for example, incentives to enter the health care system through primary health care facilities rather than hospitals, and incentives for those willing to pay more to use private sector facilities). 4.54 Finally, for a more efficient and better quality health care system, the efficient participation of the private sector in the provision of quality services should be encouraged. This would: (i) mobilize additional resources and relieve some of the burden on the public resources for health care that could be better targeted toward the 96. It should be noted that currently 800.000 health care vouchers are distributed to households. Given that in practice one household can have several vouchers, it is not necessarily clear that the number of health care vouchers will correspond to the number of households. - 46 - poor; (ii) introduce competition into the system which would improve efficiency and quality of care, and potentially prevent cost escalation; and (iii) respond to growing consumer demand for a greater range and choice of services (such as encouraging the development of complementary private voluntary schemes).9' Safe Water Supply 4.55 Supporting the rehabilitation and expansion of the water supply, treatment, and distribution systems needed in urban and rural Tunisia calls for a revision of the water pricing structure. Arnong other alternatives, consideration could be given to using the long-run marginal cost as a reference point for the pricing structure, which allows price discrimination by consumption range on the basis of information on the consumption behavior of SONEDE subscribers." Preliminary research indicates that the resulting increase in water charges would be concentrated in the middle consumption ranges, leaving the lowest consumption range almost untouched and thus improving the targeting of cross-subsidies.' A revision of the present pricing structure also needs to take into account the higher costs of delivering water services in rural areas than in urban areas (by extending the cross-subsidization scheme, if needed).'"u Housing 4.56 Government housing subsidy programs disproportionately benefit the middle-income class. To achieve more efficient use of public funds and move toward a private, market-based housing strategy, the Government needs to assist the poor through a different approach. It could: (i) revise the present subsidized loan system (FOPROLOS) to focus these resources on the needy; (ii) continue improving housing conditions in poor areas (rural and urban) by increasing access to basic infrastructure (potable water, electricity, sanitation); and (iii) encourage the development of the rental market in urban areas by reducing tax rates on rental income. 4.57 To promote smooth functioning of the housing, land, and housing finance markets, and the development of the private sector, a long-term strategy for structural improvement and growth in the housing sector could be developed. This strategy could focus on: (i) expanding the primary mortgage market and creating a secondary mortgage market to develop housing finance and deepen the financial market, reforming contractual savings schemes and FOPROLOS; and (ii) adopting measures to facilitate the development of a private land market (in particularly, by reforming and expanding land registration). 97. See World Bank, "Republic of Tunisia Hospital Restructuring Support Project", report 9157-TUN, 1991; "Republic of Tunisia: The Social Protection System", report 11376-TUN, 1993; and "Republic of Tunisia: Health Sector Development Strategy: Towards an Expanded Role for the Private Sector", White Cover report, January 1995. 98. The long-run marginal cost (LRMC) is the soundest basis for a rational pricing policy. It is defined as the increment in capital and current cost needed to meet the related projected in increase in demand. The practice of LRMC pricing is supposed to be conducive to a more efficient use of existing capacity, as well as to an optimum capacity expansion in the future. The LRMC applies only to the cubic meter of water purchased. Estimation of the LRMC is important to arrive at a single tariff base for the whole country. The discrepancy between tariffs charged and the respective marginal costs makes possible to estimate the implicit cross-subsidization between users for urban and rural ares. 99. SONEDE, "Etude Economique sur l'Eau Potable en Tunisie, Phase 2", 1993. See World Bank, "Republic of Tunisia: Water Supply and Sewerage Project", staff appraisal report 12939-TUN, July 1994. 100. It is estimated that, in 1992, the long-run marginal costs of water production in rural areas were more than twice as high as in the urban areas. - 47 - Direct Transfers 4.58 Although Tunisia's direct transfer programs are pro-poor, they are not relatively well targeted, mainly because the target population is not well-defined. To improve the targeting of assistance, it is necessary to identify a well-focused set of objectives. This helps refine the scope of programs and improve their administration. The goal of Tunisia's direct cash transfer programs is to reduce poverty by providing assistance to low-income groups through an integrated approach that combines and coordinates different mechanisms, tailoring them as closely as possible to the households' circumstances. Thus, a clear understanding of the economic and social characteristics of the poor is needed to develop adequate household socioeconomic indicators for identifying poverty groups, monitoring living conditions, and analyzing the impact of policy initiatives on the poor and other vulnerable groups.""' An important source of such information in Tunisia is the Household Consumption Surveys, and these surveys will soon be richly supplemented by information from the ongoing 1995 Living Standards Measurement Survey (LSMS). The information from the Household Consumption Survey could be more fully analyzed to further the understanding of poverty in Tunisia, particularly the effects of socioeconomic characteristics (family size, age, education, and the like) on poverty (see Annex B.4). The results of this analysis would help in refining a set of monitoring indicators that would, in turn, form the basis for a more focused strategy to alleviate poverty. In all likelihood, direct transfers would continue to play an important role in poverty alleviation efforts. Consumer Subsidies 4.59 The Government's three-pronged reform of consumer subsidies has proved to be an institutionally feasible and politically acceptable way of cutting budgetary expenditures while protecting the consumption of the poor. But several steps could be taken to sharpen the effectiveness of the reform: (i) to improve targeting performance, the application of both the "inferior goods" and the "superior goods" approaches could be strengthened by further tailoring the selection of subsidized goods to the consumption pattern of the poor, and liberalizing government controls on unsubsidized goods; (ii) to control budgetary costs, retail price increases need to keep pace with inflation, but care should be taken to ensure that prices increase faster on goods consumed disproportionately by higher-income groups; (iii) the large scope for further cost reduction and liberalization could be exploited; and (iv) the information needed to define the targeted group and monitor the progress of the reform could be regularly collected. 4.60 It must be recognized that real price increases can have an adverse effect on the welfare and dietary intake of the poor. The current "targeting" of price adjustments, reducing or eliminating subsidies on products consumed disproportionately by the rich, dampens these effects. But even these adjustments hurt poor people if they consume the goods affected. In cutting subsidies, therefore, the Government should take steps to cushion the impact of price increases on the poor and to maintain the existing level of transfers to lower-income groups. Beyond a careful design of further self-targeting reforms, the Government should introduce, where feasible, other complementary and more cost-effective measures or increases in subsidies on well-targeted products as it attempts to further reduce poverty in the coming years. m:\setareh\poverty\report\main. gr 8.17.95 101. It is expected that the criteria for qualifying as a needy family will soon take into account official poverty lines derived from the Household Consumption Survey. - 48 - BIBLIOGRAPEHCAL REFERENCES M. Lipton and M. Ravallion, "Poverty and Policy," in J. Behrman and T.N. Srinivasan (eds), Handbook of Development Economics, volume 3 Amsterdam: North-Holland, 1993. P. Lanjouw and M. Ravallion, "Are Larger Households Really Poorer?" World Bank, mimeo, Policy Research Department, 1993. Atkinson, Anthony B., "On the Measurement of Poverty", Econometrica 55: 749-764, 1987. Bouis, Howarth E., and Lawrence J. Haddad, "Are Estimates of Calorie-Income Elasticities too High? A Recalibration of the Plausible Range", Journal of Development Economics 39: 333-364, 1992. Chen, Shaohua, Gaurav Datt, and Martin Ravallion, "POVCAL: A User-Friendly Computer Program for Poverty Analysis Using Grouped Data", Policy Research Department, World Bank, 1992. , "Is Poverty Increasing in the Developing World?", WPS 1146, Policy Research Department, World Bank, 1993. Lipton, Michael and Martin Ravallion, "Poverty and Policy" in Jere Behrman and T.N. Srinivasan (eds) Handbook of Development Economics Volume 3 Amsterdam: North-Holland, 1993. Ravallion, Martin, "Poverty Comparisons", Fundamentals in Pure and Applied Economics Volume 56, New York: Harwood Academic Press, 1993. Ravallion, Martin and Benu Bidani, "How Robust is a Poverty Profile?", World Bank Economic Review, January 1994. Summers, Robert and Alan Heston, "The Penn World Table (Mark 5): an Extended Set of International Comparisons, 1950-1988", Quarterly Journal of Economics 106:327-368, 1991. van de Walle, Dominique and Martin Ravallion, "A profile of Poverty in Morocco", mimeo, Policy Research Department, World Bank, 1992. Tunisian Studies and Surveys Ministry of Agriculture, Enque^te Agricole de Base, 1989 and 1991. Ministry of Planning and Regional Development, "Evaluation des Programmes d'Emploi", February 1993. Ministry of Education and Sciences, "Rapport relatif a la convention sur I 'elimination de toutes formes de discrimination a l'egard des femmes", Tunis, 1993. - 49 - Ministry of Social Affairs, "Etude Preliminaire sur les Familles Necessiteuses dans le Gouvernorat de Tunis, Beja, El Kef, Sousse, Sfax, Kebili", May 1994. SONEDE, "Etude Economique sur 1 'Eau Potable en Tunisie, Phases 1-3", 1993. Oueslati, Abdelmajid, "Determination d'un Seuil de Pauvrete en Tunisie", Bulletin de Stateco, Paris, France: Institut National de la Statistique et des Etudes Economiques, June 1987. Institut National de la Statistique (INS), "Enquete Nationale sur le Budget et la Consommation des Menages 1985 et 1990", Volumes A and B. INS, "1984 Census". INS, "National Survey of Population and Employment 1989". INS, "Industrial Surveys", 1985-90. INS, "1993 Enquete Restreinte sur la Consonmmation des produits subventionnes", June 1994. Ministry of Regional Development, CGDR, "Evaluation des programmes d 'emploi: approche globale," Rapport final, fevrier 1993. CNEA, "Etude du systeme de credit agricole en Tunisie : etablissement de l 'eligibilite des exploitations agricoles au credit bancaire," July 1987. Ministry of Agriculture, DG/FE, D/IA, "Budget Equipment: 1993. Opgrationsfinanci&res". Matoussi, Mohamed Salah; Ayadi, Mohamed; Baccouche, Rafik et Goaied, Mohamed "L'etude de comportement de la demande de consommation en Tunisie", 1994. Matoussi, Mohamed Salah; Ayadi, Mohamed; Baccouche, Rafik et Goaied, Mohamed "Projet taxes et subventions" 1993 et revise en 1995, contrat CRDI-FNRS, Groupe de Recherche en Finance Quantitative, Faculte des Sciences Economiques et de Gestion de Tunis. World Bank Reports "World Development Report 1990. Poverty", New York: Oxford University Press for the World Bank, 1990. "Implementing the World Bank's Strategy to Reduce Poverty. Progress and Challenges", Washington DC: World Bank, 1993. "Republic of Tunisia: Private Sector Assessment", report 12945-TUN, November 1994. - 50 - "Republic of Tunisia: National Rural Finance Project", yellow cover staff appraisal report 13922-TUN, February 1995. "Republic of Tunisia: The Social Protection System", report 11376-TUN, April 1993. "Republic of Tunisia: Post-Basic Education Study", report 12670-TUN, December 1993. "Republic of Tunisia: Housing Sector Strategy Note", report 13563-TUN, September 1994. "Republic of Tunisia: From Universal Food Subsidies to a Self-Targeted Program", Green Cover report 11946-TUN, November 1993. "Republic of Tunisia: Higher Education Restructuring Project", report 9795-TUN, February 1992. "Republic of Tunisia: Hospital Restructuring Support Project", report 9157-TUN, 1991. "Republic of Tunisia: Health Sector Development Strategy: Towards an Expanded Role for the Private Sector", White Cover report, January 1995. "Republic of Tunisia: Water Supply and Sewerage Project", staff appraisal report 12939-TUN, July 1994. "Republic of Tunisia: Small Farmers: Potential and Prospects", report 9323-TUN, May 1991. Other Studies UNICEF, "Etude sur les causes de l'abandon scolaire primaire de la flUe en milieu rural en Tunisie", 1990. Charnes, Jacques, "Deux etudes sur l'emploi dans le monde arabe", Paris. Centre francais sur la population et le developpement. 57p. CEPED dossiers no. 11. FAO/CP, "Mise en place d'un systeme de promotion, suivi et evaluation permanente du developpement agricole," Documents de travail 6 a 8. TCP/TUN/0053. June 2, 1992. FAO/CP, "Evaluation du programme de developpement rural integre (PDRI)," Rapport d'evaluation, Vols 1 et 2. July 1993. Kristjanson P., Roth M., Taher G., Mosbah B.H., Ayda M. and Larbi H. ; "Rapport preliminaire sur la creation d'une base de donnees agricoles". Tunis: Ministry of Agriculutre; USAID. Abt Associates, University of Wisconsin, April 1989. Lachaud, Jean-Pierre, "Le marche du travail en Tunisie : structure, desequilibres et ajustements". Geneva. Institut international d'6tudes sociales. 57p. Mohamed, Elloumi, "Travail et revenus exterieurs dans la dynamique des exploitations agricoles", Afrique et D6veloppement (Senegal) 12, no. 4; 41-70. 1987. - 51 - Morrison, Christian, "Evolution of the Tunisian Labor Market". Development Research Dept. The World Bank. 51p. Radwan, Samin Muhammad at al, "Rural labor and Structural Transformation in Tunisia", League of Arab States, Dept. of Social & Cultural Affairs, March 1987. Trabelsi, Mohsen; "Les exploitants agricoles en Tunisie: disparites regionales de l'emploi non salarie", Revue tunisienne de sciences sociales (Tunisie) 27 no. 103 : 9-52. 1990. Triki S. and Maklouf E., "Rural Women's Activities and Management of Natural Resources in Tunisia", Febraury 1993. m: \setareh\poverty\report\bibliog I .l e / d, & r r o n e a n a S ~~e a onz uto us 3~~~~~~~~~~~~~~~~~~~~~~~
Группа Всемирного банка · Pre-2003 Economic or Sector Report
Tunisia - Poverty alleviation : preserving progress while preparing for the future (Vol. 1 of 2) : Main report
Открыть оригинал документа
Полный текст размещён на сайте публикующей организации. lawenc.com индексирует метаданные и ведёт на официальный источник.
Полный текст
Основные сведения
Организация
Группа Всемирного банка
Тип документа
Pre-2003 Economic or Sector Report
Страна
Тунис
Источник
Всемирный банк