Africa Region Human Development Working Paper Series Cost and Financing of Education Opportunities and Obstacles for Expanding and Improving Education in Mozambique (Sumário em português incluido) Africa Region The World Bank © July 2003 Human Development Sector Africa Region The World Bank The views expressed herein are those of the authors and do not necessarily reflect the opinions or policies of the World Bank or any of its affiliated organizations. Cover photo by Curt Carnemark. Cover design by Word Express Interior design by Word Design, Inc. Contents Foreword vii Acknowledgements ix Definition of Terms and Abbreviations and Acronyms x Executive Summary 1 Resumo Executivo 6 1 Macroeconomic and Demographic Context 12 Evolution in gross domestic product and fiscal resources 12 Evolution in education spending, absorptive capacity, and highly indebted poor countries 14 The likely impact of HIV/AIDS on children and teachers 16 2 Enrollment Pattern 18 Structure of the educational system 18 Enrollment trends 19 Primary education (EP1 and EP2) 19 Secondary education (ESG 1 and ESG2) 19 Teacher training 20 Technical education and vocational training 21 Higher education 21 Adult education 21 Conclusions on educational enrollment 21 Overall index of educational development and a global measure of efficiency 22 Pattern of student flow over the system 23 Demand- and supply-side perspectives on the development of education 24 iii iv MOZAMBIQUE: COST AND FINANCING OF EDUCATION 3 Financial Aspects of Education 27 Recent trends in budget allocation by economic purpose and level of education 27 A closer look at recurrent budget data for 1998 28 Unit cost estimates and analysis 29 The macroperspective 29 The microperspective 31 High costs of school buildings and construction 37 Unit costs and economies of scale 37 4 Internal Efficiency and Management Issues 40 Efficiency in student flow 40 Allocation of teachers to individual schools 40 Lower primary education (EP1) 40 Upper primary (EP2) and lower (ESG1) and upper secondary (ESG2) education 45 Some elements regarding the quality of primary education 46 Student outcomes and school and student characteristics 46 The social and cultural context of learning in Mozambique 47 5 Analyzing Equity 50 Enrollment patterns by gender and by geographic location 50 An overall perspective 50 Gender and geographic disparities in schooling profiles 52 Inequity in secondary enrollment 55 Some factors that may affect gender equity 56 Equity in the distribution of public resources for education among cohorts of students 56 Structural inequity in the distribution of public resources in education 56 Social inequity in the distribution of public resources in education 57 6 A Reminder on External Efficiency 60 General social benefits and the impact of education in the traditional sector of the economy 60 Education and training for the modern sector of the economy 61 Annexes 64 Figures Figure 1 Access to the educational system in Cabo Delgado, grades 1­8, 1998 1 Figure 2 School life expectancy and spending on education in select Sub-Saharan African countries, around 1995 3 Figure 3 Teacher allocation in public EP1 schools, 1998 4 Figura 4 Acesso ao sistema educativo em Cabo Delgado, graus 1 a 8, 1998 6 TABLE OF CONTENTS v Figura 5 Esperança de vida escolar e gastos com a educação em países seleccionados da África Subsariana, cerca de 1995 8 Figura 6 Repartição dos professores nas escolas públicas do EP1, 1998 9 Figure 1.1 Overspending and underspending in proportion to total budget, 1990­98 16 Figure 2.1 Gross enrollment rates in primary education, 1993­99 20 Figure 2.2 Gross enrollment rates in secondary education, 1993­99 20 Figure 2.3 School life expectancy and spending on education in select Sub-Saharan countries, around 1995 22 Figure 2.4 Admission rate in primary education, 1998 23 Figure 2.5 Admission rate in secondary education, 1998 23 Figure 3.1 Unit recurrent cost as a function of enrollment size in public EP2, 1998 38 Figure 3.2 Unit recurrent cost as a function of enrollment size in public ESG1, 1998 38 Figure 4.1 Allocation of teachers in public EP1 schools, 1998 43 Figure 4.2a Allocation of teachers in public EP1 schools in rural areas, 1998 43 Figure 4.2b Allocation of teachers in public EP1 schools in urban areas, 1998 43 Figure 4.3 Allocation of teachers in public EP1 schools in Cabo Delgado, 1998 44 Figure 4.4 Allocation of teachers in public ESG1 schools, 1998 45 Figure 4.5 Endowment of teachers in EP1 and EP2 schools, by province, 1998 45 Figure 5.1 Relative gender gap and GER in different provinces, 1998 51 Figure 5.2 Enrollment rate in EP1 through ESG2, by gender and grade level, 1998 53 Figure 5.3 Enrollment rate in EP1 and EP2, by geographic location and grade level, 1998 53 Figure 5.4 Admission rate in EP1 and EP2 in Cabo Delgado, by gender, 1998 54 Figure 5.5 Admission rate in EP1 and EP2 in Gaza, by gender, 1998 54 Figure 5.6 Lorenz curve of the distribution of public resources, 1998 57 Tables Table 1.1 Population, GDP, and total public expenditures, 1990­99 13 Table 1.2 Output and employment, 1998 14 Table 1.3 Public spending on education as a share of GDP and total government spending, 1990­99 15 Table 2.1 Distribution of enrollment, by level of schooling, 1998 19 Table 2.2 Enrollment and gross enrollment rates, by level and type of education, select years, 1993­99 20 Table 2.3 Pattern of school access and survival, 1998 23 Table 2.4 Survival and transition rates, 1998 24 Table 2.5 EP1 students in complete and incomplete public schools, 1998 25 Table 3.1 Distribution of public spending on education, by economic purpose, 1996­99 28 Table 3.2 Distribution of public spending on capital, by level and type of education, 1996­99 29 Table 3.3 Public recurrent spending on education, by level and type of education, 1998 30 vi MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 3.4 Recurrent unit cost, by level and type of education, 1998 31 Table 3.5 International comparison of recurrent unit cost, as percent of per capita income by level of schooling, 1998 31 Table 3.6 Per student spending at the school level, by level of education, with and without adjustment for repetitions and dropouts within the cycle of schooling, 1998 32 Table 3.7 Pupil to teacher ratio and pupil to nonteacher ratio, by level of education at the school level, 1998 33 Table 3.8 Distribution of teachers, by level of qualification and level of schooling, and the average teacher and nonteacher salary, by level and type of education, 1998 34 Table 3.9 Reconstructing the unit cost of education, by level and type of schooling, 1998 35 Table 3.10 Change in per pupil spending as a result of changes in input variables, 1998 36 Table 4.1 Index of internal efficiency, by cycle of study, 1998 41 Table 4.2 Survival rate in EP1, 1994­99 42 Table 4.3 Regression analysis of student outcomes, 1999 46 Table 4.4 Teacher experience for sampled schools, by province, 1999 47 Table 5.1 Enrollment, by gender and level of education, 1997 51 Table 5.2 Gross enrollment rate in EP1, by province, 1998 51 Table 5.3 Access to grades 1 and 5, by province, 1998 52 Table 5.4 Enrollment in absolute figures in ESG1 and ESG2, 1998 55 Table 5.5 Distribution of public resources in education, by gender, 1998 57 Foreword T his study is part of a series of education formance over the past decade and the significant country status reports (CSRs) that are being human development challenges that persist, have prepared by World Bank staff in collabora- made an even stronger case for such an approach. tion with national teams from various Sub-Saharan In Mozambique, as in other low-income countries, countries. The immediate objective is to enhance the two specific goals in education stand out: ensuring knowledge base for policy development in the edu- that all children complete basic schooling of adequate cation sector. More broadly, CSRs create a basis for quality; and managing the production of graduates at engaging a diverse audience in dialogue on educa- post-basic levels, in terms of quantity and skill level tion sector policies and for developing a shared and mix, to match the demand for skilled labor. To vision for the future. These processes have become fulfill these objectives, an infusion of resources, from increasingly important as governments renew their both domestic and international sources, may well be commitment to reduce poverty, and the interna- needed. However, an equally important issue is to tional donor community pledge to provide the ensure that currently available resources are used effi- needed financial assistance. ciently and equitably. The evidence suggests that in In this new dynamic of international develop- many Sub-Saharan countries, substantial scope for ment assistance, the World Bank has oriented its progress exists in this regard. Some countries have corporate mission toward supporting governments indeed already embarked on reforms for improving in the fight against poverty, making the task its the performance of their education systems. In the defining priority. In practice the strategy is articu- 1990s, for example, policies to encourage more effec- lated through a two-prong approach: support for tive use of teachers through multi-grade teaching and policies that accelerate economic growth combined double-shifting were put in place to boost coverage. with explicit measures to promote a more equitable In the post-war period in Mozambique this even led distribution of the benefits from growth. Human to triple-shifts in many schools. More remains to be development, especially through investments in done, particularly to tackle the structural constraints education, plays a central role in both processes. that encumber the education system. The challenges This fact receives recognition in the emphasis include defining an appropriate division of responsi- placed on education in the debt relief arrangements bility between the public and private sectors for under the Heavily Indebted Poor Countries (HIPC) financing and delivering education services, setting Initiative. Governments have also invariably sustainable levels of teacher remuneration, and creat- treated education as a priority sector in their ing institutional arrangements for effective system Poverty Reduction Strategy Papers (PRSPs). In the management. Reform in these areas are key to case of Mozambique the impressive economic per- improving the quality of learning outcomes, which in vii viii MOZAMBIQUE: COST AND FINANCING OF EDUCATION turn will be crucial for progress toward the Education Regarding the process for preparing the CSRs, the for All targets. main feature is that it is a product of a partnership In order to identify appropriate measures and between the World Bank and national teams from the address the structural issues, a first step is to develop various countries. Because participation in the analyti- a country-specific knowledge base that sheds light cal work is essential for developing a deep under- on the key weaknesses in the education system. In a standing of policy issues, an integral part of the process large number of Sub-Saharan countries, however, for preparing a CSR involves training and capacity that knowledge base remains sparse, reflecting the building where these are needed. It is important to systematic neglect of analytical work in the past. In note that CSRs are diagnostic documents whose pur- order to re-capitalize it, the World Bank has initiated pose is to help identify the policy questions rather than the preparation of CSRs, using a standard format to to offer solutions and make recommendations. The consolidate the available information in a policy-rel- process of policy development is more appropriately evant manner. led by the national team, and the availability of a CSR Two features characterize the CSRs, one pertaining provides a good basis for disseminating the findings to its technical content, the other to the process by and stimulating a broad national dialogue on the way which it is accomplished. With regard to content, five forward. The document also serves as a basis for aspects are worth mentioning. First the CSRs pay preparing a rational and defensible plan for sector close attention to issues of equity and the distribution development that informs the country's medium-term of public resources for education, given the impor- budget planning exercise. These processes are already tance of these topics in the HIPC and PRSP context. well underway in the Mozambican context. The Second, the analysis relies on commonly available national team has used the CSR extensively for policy administrative data as well as household surveys, an dialogue within the country and to target interventions approach that has helped to improve the consistency under the current Education Sector Strategic Program and robustness of the statistical results. Third, the (ESSP). The CSR will play a pivotal guiding role in the CSRs put a sharper focus on outcomes by emphasiz- Government's current plans to update its ten-year ing indicators beyond the usual gross and net enroll- strategic plan as part of the country's poverty reduc- ment ratios. In particular, it documents the schooling tion strategy and Country Assistance Strategy to be careers of children as reflected in the shares of the prepared. For the World Bank's part, the document is population that enter grade 1 and attain the various helping to facilitate ongoing discussions to improve grades in the educational ladder. The more detailed the internal and operational efficiency of the system, approach has helped to sharpen the socioeconomic, and most important to improve quality. gender and geographical disparities in schooling, as The publication of the CSR for Mozambique is well as clarified the sources of the disparities. In the intended to institutionalize our collective knowledge case of Mozambique, some unique results emerge about the country's education sector, including from this analysis: widespread grade repetition was broader macro economic and social issues like identified as an important factor in producing the HIV/AIDS and the nature of the policy challenges, observed socioeconomic disparities in education; and to share that knowledge as widely as possible. It is also important is that a large number of schools offer my hope that as new knowledge emerges in the course instruction only in grades 1 and 2, an arrangement of implementing the country's poverty reduction strat- that reflects severe teacher shortages and poorly- egy, the CSR would be updated to track progress in managed deployment of teachers across schools. A overcoming the constraints in the education sector that fourth example of the technical content of CSRs is the currently impede poverty reduction in Mozambique use of school-level data to assess the scope for and which may in the long run slow economic growth. improving service delivery to the poor. Finally, the CSRs also make use of data on student learning-- Birger Fredriksen where they are available--to gauge the education Senior Education Advisor system's performance in this important domain, and Africa Region to identify cost-effective measures for progress. The World Bank Acknowledgements T his study is the product of close collabora- Statistics, Ministry of Higher Education, Science tion between the World Bank and the Gov- and Technology, the Pedagogical University and the ernment of the Republic of Mozambique. Eduardo Mondlane University. The Bank team and study was conducted under the The study was initiated by Donald Hamilton. general supervision of Soren Nellemann who was The work also benefited from inputs from others also the main coordinator of the report. Aminata including Noel Kulemeka and Alexandria Valerio. Maiga was the main contributor on equity issues. Elizabeth Forsyth, Patricia Tierney and Donald Rau Keiichi Ogawa and Ramahatra Rakotomalala pro- provided expert editorial advice. Julia Anderson vided inputs in many areas, while Alain Mingat and managed the production process. Jee-Peng Tan served as overall technical advisers. Helpful comments and information were The Mozambican technical working group con- received from a number of donors in Mozambique. sisted of senior policy makers and technical staff The report was financed by the World Bank, with from the Ministry of Education, including Virgilio contributions from the governments of Mozam- Juvane, Manuel Rego, Antonio Ilidio, Antonio bique, France and Norway, the latter through the Chambal, Zacarias Mazembe, and Kauxique Maga- Norwegian Education Trust Fund managed by the nlal, with contributions made by representatives World Bank. from other parts of government, including Ministry An informal version of the report in Portuguese of Planning and Finance, Ministry of State Admin- has been published by the Ministry of Education. istration, Ministry of Labor, National Institute for ix Definition of Terms/Abbreviations and Acronyms Abbreviations and Acronyms GDP Gross Domestic Product ADPP A Teacher Training Nongovernmental GNP Gross National Product Organization HIPC Highly Indebted Poor Countries CSR Country Status Report INE National Institute for Statistics EP1 Lower Primary Education, Grades 1­5 MAE Ministry of State Administration EP2 Upper Primary Education, Grades 6­7 MINED Ministry of Education ESG1 Lower Secondary Education, Grades MTEF Medium Term Expenditure Framework 8­10 MPF Ministry of Planning and Finance ESG2 Upper Secondary Education, Grades NER Net Enrollment Rate 11­12 PRSP Poverty Reduction Strategy Paper ESSP Education Sector Strategic Program SIDA Swedish International Development FDI Foreign Direct Investment Agency GAR Gross Admission Rate UP Pedagogical University GER Gross Enrollment Rate UEM Eduardo Mondlane University Definition of Terms Currency Equivalents (as of May 2000) Currency Unit: Metical (Mt); plural: Meticáis US$1.00 = Mt12,000 (Year 2000, approximately) Fiscal Year January 1 to December 31 Executive Summary Enrollment, equity, and children's chances of Figure 1 Access to the educational system in Cabo completing an education Delgado, grades 1­8 (1998) Enrollment rates 100.0% M 90.0% ozambique has made great strides in 80.0% expanding the total number of students 70.0% enrolled in primary education. From 1993 60.0% to 1999, the national gross enrollment rate1 increased 50.0% from 60 to 92 percent at the lower primary level 40.0% (EP1). By the year 2000 more than 2.1 million chil- 30.0% dren were enrolled in EP1. Nevertheless, repetition 20.0% and dropout rates are high throughout the system, 10.0% and access is limited beyond the EP1 level; this com- 0.0% 1 2 3 4 5 6 7 8 bination has led to uneven enrollment by level of education. The Mozambican educational system Source: Mozambique school census, 1998. resembles a rough L shape, with most students enrolled in the beginning grades and very few enrolled beyond grade 5. In other words, there is a rate, to grade 1 was about 94 percent nationally. steep decline in the gross grade-specific enrollment Figure 1 shows the access rate in Cabo Delgado, rate (which can be taken as a measure of the access which reflects the rate in Mozambique as a whole. or gross admission rate2 to the system at each grade) As can be seen from the figure, the country has suc- as shown in figure 1. ceeded, or almost succeeded, in universalizing access to grade 1. However, high repetition and Survival rates dropout rates (around 25 and 10 percent for EP1, respectively) make the system inefficient and shows Another way to analyze educational opportunities that Mozambique is still many steps away from pro- in the system is to assess a student's likelihood of viding universal basic education for all. Thus, while completing a certain grade level and becoming lit- almost all Mozambican children now enter grade 1, erate. In 1998 the gross admission rate, or access 85 percent of the entrants are unlikely to make it 1 2 MOZAMBIQUE: COST AND FINANCING OF EDUCATION past grade 7. In fact, most pupils are likely to drop across localities and population groups. Thus, out before they attain numeracy and literacy skills. increasing the country's overall enrollment rate Mozambique has a very low average retention inevitably implies better chances for schooling rate of 45 percent (52 percent for boys and 38 per- among those who are currently marginalized. cent for girls) at primary level. Incomplete schools are part of the problem. About 25 percent of pri- Macroeconomic and financial aspects mary schools are considered "incomplete," mean- ing they do not offer the full primary cycle of five Macroeconomic and demographic context grades. In this case it is the system failing to provide the students with the educational opportunities The development of education is determined largely they seek. Making existing schools complete should by macroeconomic and demographic conditions. thus be given high priority because it would However, the capacity of the system to absorb improve student retention and lower repetition and resources efficiently and produce good-quality edu- dropout rates. cation varies across countries with similar per capita Geographic location of schools presents another income. Thus countries with similar gross domestic challenge. According to the 1997 household survey, product (GDP) per capita show widely different the average child travels as far as 45 kilometers one levels of education development. way to reach his or her school, often taking more Throughout the 1990s, Mozambique enjoyed a than one hour to make the journey. To make matters period of unprecedented economic growth during worse, up to 32 percent of rural villages do not have which per capita income rose sharply and the popu- a primary school. Supply-side instruments (imple- lation grew much more slowly than expected. Both menting complete schools reasonably close to where factors were favorable for improving and expanding children live, possibly using multigrade teaching education coverage. The GDP almost doubled when applicable) are thus likely to help address between 1992 and 1999 (from Mt17.4 trillion to Mt30.9 what probably will be the most important issue in trillion in constant prices of 1995), equivalent to a 77 Mozambican education for years to come. However, percent growth in constant prices over that period.3 the ability to address some of these issues also At the same time, domestic revenues rose 8 times,4 depends on the overall resources available to the and government expenditures increased about 10 sector, as well as an understanding of the different times.5 Despite relatively high growth, external fund- reasons behind low retention as they vary from loca- ing still contributed about half of the government's tion to location. budget, a situation unlikely to change in the future. Equity in enrollment Education expenditures The gross admission rate (or access) to grade 1 is Education spending increased about five times in fairly high across localities and population groups nominal terms during the 1990s.6 The share of recur- but wide disparities in admission rates emerge by rent domestic spending on education in total recur- grade 5, reflecting the impact of grade repetition and rent government spending remained stable at about dropping out. The disparities are particularly strik- 16 percent. Similar to overall government spending, ing between the southern and northern provinces, about half of total education spending (mostly for between urban and rural areas, and between girls capital expenditure) was financed from external and boys. Girls' gross admission rate to grade 5 is sources. The share of total spending on education particularly low in some places, falling below 20 declined from about 5.5 percent of GDP in 1991­93 percent in such provinces as Cabo Delgado (12 per- to about 3.5 percent in the late 1990s. This, however, cent); Nampula (13 percent); Niassa (16 percent); was largely due to the rapid growth in the economy Tete (18 percent); and Zambezia (15 percent). and not as a measure of improved efficiency on One positive note is that the higher the overall education spending due to a lowering of repetition gross enrollment rate, the smaller the disparities and drop out rates. EXECUTIVE SUMMARY 3 The share of domestic spending on education fol- Figure 2 lowed a more or less similar downward trend, aver- School life expectancy and spending on aging 2.3 percent of GDP in 1991­93 and only 1.7 education in select Sub-Saharan African countries, around 1995 percent in the late 1990s. Even if we take into account total spending on education (including School life expectancy external financing), the share of spending on the 14 sector was only 3.5 percent of GDP, which is very low by international standards. There clearly is 12 scope for Mozambique to increase its domestic 10 funding of the educational system. 8 Educational development and 6 efficiency gains 4 Mozambique Another way to look at this is to consider the value 2 obtained for the money spent. How many years of 0 education can a Mozambican child expect to receive 0 2 4 6 8 10 given current resource allocation? Between 1995 Overall expenditure on education as % of GNP and 1999, the number of years a child could expect Source: Educational and financial statistics for Mozambique and UNESCO. to attend school in Mozambique improved from about 3.4 years to 4.6 years on average, but this greater than administration spending for all of the remained comparatively low for the region; around country's EP2 and ESG1 schools. Not counting the same time (1993), the school life expectancy was spending on administration, primary schools estimated to be about 5 years in Tanzania, 6 years in received about 44 percent of total spending for the Malawi, and 11 years in Zimbabwe.7 While the situ- sector, or about 52 percent if we include spending on ation has changed since then in many of these coun- administration. These proportions were relatively tries, figure 2 shows there is a clear need in Mozam- stable over the period 1995­99. Secondary education bique not only to improve efficiency in the use of received less than 10 percent of the total, while resources but also to increase the overall amount of higher education received about twice that amount. resources going to the sector. Overall, these figures raise a question about the bal- ance of spending between the secondary and higher Budget resources, distribution, and construction levels of schooling in the country and indicate that costs. In 1999, the lion's share of external funding for there is also clearly room for improving efficiency at the education budget was used for school construc- the central level with a view to enhance service tion. This is not surprising for the period following delivery and reduce overhead costs. the civil war; what is unique, however, is that the cost of construction of a primary school in Mozam- Aspects of efficiency, quality, and management bique is about 5 to 20 times the cost of construction in neighboring countries. Although construction in Efficiency in per pupil spending general is more expensive in Mozambique, there and outcomes clearly is room for improvement, balancing recur- rent costs with capital investment costs. At the primary school level, Mozambique's unit cost From 1995 to 1999, the cost of delivering services is, on average, similar to that of its neighbors, but (i.e. the amount spent on administration) repre- lower than the average unit cost in Anglophone sented close to 20 percent of recurrent spending for African countries and to a greater extent, Franco- the education sector, of which the central ministry phone countries. A relatively similar pattern holds (MINED) accounted for more than half. This figure for secondary education. For the post-primary is very high, particularly when considering it is levels, especially higher education, spending per 4 MOZAMBIQUE: COST AND FINANCING OF EDUCATION student is on the high side compared with neigh- numbers of pupils.12 This fact suggests that there is boring countries, but more or less in line with the significant inconsistency in the allocation of teach- average of all African countries. Because repetition ers among the provinces. More significant is that and dropping out raise the unit costs, additional about 80 percent of the variance in teacher deploy- costs (and resources) are needed to produce a grad- ment occurs within the provinces themselves. uate--implying an inefficient use of resources.8 The Therefore, the allocation of teacher resources index of overall efficiency in EP1 is only 58 percent, appears to be a function of how teacher resources indicating that 42 percent (1 ­ 0.583) of the public are managed, particularly at the provincial level. resources mobilized in EP1 are de facto wasted9-- The distribution of available teachers is an issue resources that otherwise could have been used to that will require attention in the years to come, improve the system or enroll new students. The especially as HIV/AIDS increases the attrition rate majority of wasted resources were lost in EP1. of teachers. Finally, the budgetary mechanisms and procedures in the sector clearly should be Allocation of teachers across schools. Overall in addressed, as this will be one way to ensure the ade- Mozambique, the allocation of teacher resources is quate distribution of resources. inconsistent10 across the estimated 6,000 lower pri- mary schools that existed in 1998, both in terms of Quality of education teacher qualifications as well as the teacher to pupil ratio. Figure 3 shows that even though schools tend Education in Mozambique is characterized by sub- to benefit from a larger number of teachers when stantial disparities in learning across schools. With- the size of enrollment is larger, there also is wide out taking into account the characteristics of the stu- variability in the number of teachers for schools dent body, the variability in the use of resources had enrolling a similar number of pupils. The variance a much greater impact on quality of education, and in the allocation of teachers is wider in rural areas11 thus also on outcomes, than did variability in the than in the country as a whole but is similar to the amount of resources available. The pedagogical variance in the allocation of teachers in urban areas. management of schools in Mozambique appears to be quite weak. Any mechanism established to Teacher deployment across provinces and teacher improve student learning and school quality will management. In Mozambique, some provinces with need to (a) measure student learning on a regular a similar number of teachers service very different Figure 3 Teacher allocation in public EP1 schools, 1998 Number of teachers 40 A A A A A AA A A A B A ACABB 30 AAA A A C AC A AA A A A BAA B A AA A A A A AA B BA C AB A A A A ABA C B AA A A A A A A A A ACABA ABAAA ABA A B A A 20 A ABA A BBC C D AA ABA AA A A A A AA A AAA AAAB BBBAACB GBBAAA BCD A A B A AA AB ABBBFD DCD DBEBAABBA B A AAAA A A A AAA A A A A A C BAAAFCJFCDBBCCBAB A A A DA B AA A A AAA B EDCEDGIEHDFAECE CE ACC BA BA C B A 10 A ABADHELLKOEFFFFJCGEBEEFBDCDCBDCABA BAAAA A AA AA A ADFRTSZWZVXQSKOGKODJMIKHBFDDCBBAA AA A CBFFIPZZZZZZZZZZZZZZWZNTGCBDC A A A AA CHTZZZZZZZZZZZZZZZZVJHFBB CA A GZZZZZZZZZZZZLCGACABAAA 0 0 200 400 600 800 1000 1200 1400 1600 Number of students Legend: A = 1 observation, B = 2 observation, etc. Source: Mozambique school census, 1998. EXECUTIVE SUMMARY 5 basis at the school level; and (b) help low-achieving A basic element to consider is that the Mozambican schools to do better. economy is, and will remain in the next decades, charac- Factors exogenous to the school also are impor- terized by a dualistic structure with a dominant tradi- tant, such as the parents' mother tongue, availabil- tional sector and a relatively small, but growing, modern ity of books at home, and so forth. School inputs, sector of employment. Any strategy for human capital such as school furniture, learning materials, and formation in the country should take this structural classroom materials (chalk, blackboard) clearly do context into account. The basic objective is to pro- have some bearing. Finally, teacher education and vide the type of human capital that best fits the the number of years of teaching have a significant demand of the two subsectors, while helping the influence on learning, which is not surprising. economy modernize and provide livelihoods that Addressing the issue of learning and the individ- alleviate proverty in the long run. ual's ability to learn after he or she has left school and apply that knowledge is critical for improving 1. The gross enrollment rate is the total number of students the impact of education on society. This impact enrolled, regardless of age, over the age-specific population group (for example, pupils ages 6 to 10 for EP1); therefore, it relates to external efficiency, which is concerned is sometimes higher than 100 percent. with the performance of graduates after they have 2. The gross admission rate (or gross enrollment rate by exited the schooling system and have entered their grade) is the total number of students enrolled in a specific grade, regardless of age, over the age-specific population working life. group (for example in figure 1, the gross admission rate or gross enrollment rate in grade 4 in the province of Cabo Del- External efficiency gado was 48 percent of the total number of 9 year old age cohort. 3. GDP grew by a factor of 10 in nominal terms between To what extent does the knowledge imparted to a 1992 and 1999. cohort of youngsters while they are in school 4. From Mt661 billion to Mt6.3 trillion. 5. From Mt1.5 trillion to Mt13.5 trillion. (whether for one year or 20) correspond to the best 6. By 540 percent, from Mt286 billion in 1992 to Mt1,834 bil- that can be achieved both for the individuals and lion in 1998. for society? 7. With an average of 4.85 years in Francophone African countries and 7.48 years in Anglophone countries. Primary education is an all-purpose investment 8. With the current rates of repetition and dropout, only 42 whose results (basic reading and writing, basic out of 100 students will complete grade 5. To produce these arithmetic and problem solving, basic life skills) 42 students in grade 5, it would be necessary to finance 210 (42 x 5) years of schooling, not counting repeaters and contribute to improved productivity in the tradi- dropouts, that is. The consequence is that to finally produce tional sector. Moreover, the economies in a number the 42 pupils in grade 5, the system would have to finance 125 (the 100 students in grade 5 plus the 25.4 percent of of developing countries have difficulty absorbing repeaters in that grade), instead of the 42 that would have the graduates from secondary or even higher edu- been strictly necessary. The same applies to the next grades. cation, making unemployment of graduates wide- 9. We also can use the efficiency ratio to get a sense of the average number of years the budget needs to finance to help spread in many African countries. Under these cir- one child on average get to grade 5: it takes 8.6 years for all cumstances, overinvesting in higher levels of the pupils (including repeaters and dropouts) to pass education is more a burden, rather than an engine through EP1; in other words, it costs the government an additional 3.6 years of per pupil spending to allow a child to for, economic growth. However, it should be reach the fifth grade. emphasized that education is a necessary, but not 10. As far as the quantitative distribution of teachers is con- sufficient condition for economic growth and social cerned, we should aim at providing schools of similar size of enrollment with a similar number of teachers. development and often the poor returns to invest- 11. For example, among all the schools with at least 10 teach- ment in higher education is a result of poor quality ers, enrollments vary between 100 and 1,200 pupils; similarly, courses that are irrelevant to the economy. Finally, schools enrolling 600 pupils may have between 5 and 15 teachers. several research findings show that primary school- 12. For example, Cabo Delgado and Inhambane both have ing has a smaller impact on economic growth in about 2,500 teachers, but the enrollment in Cabo Delgado is African countries than in low-income countries sit- about 134,000 pupils, 41,000 pupils fewer than in Inhambane, which has 175,000 pupils. Similarly, both Tete and Maputo uated in other regions of the world. provinces enroll about 140,000 pupils each, but Tete has 3,000 teachers, while Maputo has fewer than 2,000. RESUMO EXECUTIVO As matrículas, a equidade e a possibilidade de Figura 4 as crianças completarem a sua educação Acesso ao sistema educativo em Cabo Delgado, graus 1 a 8, 1998 As taxas de matrícula 100.0% M 90.0% oçambique fez grandes progressos quanto 80.0% ao número total de alunos matriculados no 70.0% ensino primário. De 1993 a 1999, a taxa 60.0% bruta nacional de matrícula1 aumentou de 60 a 92 50.0% por cento no primeiro nível primário. Até ao ano 40.0% 2000, mais de 2,1 milhões de crianças foram matri- 30.0% culadas no EP1. No entanto, as taxas de repetência 20.0% e de desistência são elevadas em todo o sistema, e o 10.0% acesso é limitado para além do nível de EP1; esta 0.0% combinação de circunstâncias levou a uma situação 1 2 3 4 5 6 7 8 desigual por nível de ensino. Um gráfico das matrí- Fonte: Recenseamento das escolas de Moçambique, 1998. culas do sistema educativo de Moçambique tem sensivelmente a aparência de uma curva em forma de L, com a maioria dos alunos matriculados nos de ser alfabetizado. Em 1998 a taxa bruta de matrí- primeiros graus e muito poucos matriculados para cula, ou taxa de acesso, no 1º grau foi cerca de 94 além de 5º grau. Por outras palavras, verifica-se por cento em todo o país. A Figura 1 mostra a taxa uma forte queda da taxa bruta de matrícula por de acesso em Cabo Delgado, o que reflecte a taxa de nível de graus (usando a taxa de matrícula como Moçambique no seu conjunto. Como se pode ver na um substituto para o acesso ou a taxa bruta de Figura 1, Moçambique conseguiu no início, ou admissão2 no sistema em cada nível), como se pode quase conseguiu, proporcionar acesso a todos. ver na Figura 1. Contudo, as elevadas taxas de repetência e de desistência (cerca de 25 e 10 por cento para o EP1, Taxas de sobrevivência respectivamente) indicam que o sistema é inefi- ciente, e mostram que Moçambique ainda não con- Outra maneira de analisar as oportunidades educa- seguiu proporcionar o acesso universal a todas as tivas do sistema é avaliar a probabilidade de um pessoas. Por conseguinte, se bem que a probabili- aluno completar o ano de um determinado nível e dade de uma criança moçambicana entrar para o 1º 6 RESUMO EXECUTIVO 7 grau tenha aumentado consideravelmente nos últi- províncias, o que reflecte as taxas elevadas de mos anos, 85 por cento dos alunos provavelmente repetência e de desistência. É de notar que, para as não irão para além do 7º grau. De facto, a maioria raparigas, a taxa de admissão ao 5º grau é inferior a dos alunos provavelmente desistirão antes de sabe- 20 por cento em várias províncias (Cabo Delgado, rem ler e contar. 12 por cento; Nampula, 13 por cento; Niassa, 16 por Moçambique tem uma taxa média de retenção cento; Tete, 18 por cento; e Zambézia, 15 por cento). muito baixa, 45 por cento no nível primário (52 por Contudo, um aspecto positivo é evidentemente o cento para os rapazes e 38 por cento para as facto que, quanto mais alta for a taxa bruta global raparigas). O problema de as escolas "serem de matrícula numa província, mais numerosas são incompletas" leva a uma retenção ainda mais baixa. as matrículas de raparigas. Portanto, aumentar a Cerca de 25 por cento das escolas primárias são taxa global de matrícula provavelmente fará subir a consideradas "incompletas", o que significa que não taxa de matrícula das raparigas. As taxas de oferecem um ciclo primário completo de 5 graus. matrícula variam entre as províncias do sul e do Neste caso, é o sistema que não proporciona aos norte, e a disparidade é evidente nas taxas de alunos as oportunidades de ensino que eles buscam. sobrevivência escolar nas zonas urbanas e rurais. As Deveria ser dada a mais alta prioridade a fazer com crianças matriculadas em escolas urbanas ou semi- que as escolas existentes sejam "completas", porque urbanas têm uma maior probabilidade de comple- isso levaria a uma melhor retenção dos alunos e a tar a sua educação do que as crianças que frequen- baixar as taxas de repetência e de desistência. tam uma escola rural. A situação geográfica das escolas coloca mais outro problema. Sabemos que, segundo o levanta- Aspectos macro-económicos e financeiros mento de 1997 às famílias, a distância média entre a casa e a escola se eleva a 4,5 quilómetros, e que o O contexto macro-económico e demográfico período de tempo médio que um aluno do EP1 leva a chegar à escola é frequentemente de mais de uma O desenvolvimento da educação é, em grande hora. Trinta e dois por cento das aldeias nas zonas medida, determinado pelas condições macro- rurais não têm uma escola primária. Melhorar os económicas e demográficas do país. Contudo, a instrumentos da oferta (pôr em funcionamento capacidade do sistema de absorver recursos eficien- escolas "completas" a uma distância razoável do temente e de produzir um ensino de boa qualidade lugar onde vivem as crianças, em classes que varia entre países com um rendimento per capita abrangem vários graus, quando isso for praticável) semelhante. Assim, entre os países que têm um é uma boa maneira de abordar o que provavel- produto interno bruto (PIB) per capita semelhante mente será o problema mais importante do ensino verificam-se amplas diferenças no nível de desen- em Moçambique nos anos vindouros. Todavia, a volvimento da educação. capacidade de abordar alguns destes problemas Durante toda a década de 1990, Moçambique também depende dos recursos globais disponíveis a desfrutou de um período de crescimento económico este sector, assim como da compreensão das sem precedentes, tendo o seu rendimento per capita diferentes razões subjacentes à baixa retenção, pois passado de USD 179 em 1990 para cerca de USD 260 elas variam de lugar para lugar. em 1999, e a população crescido muito mais lenta- mente do que se esperava. Ambos esses factores A equidade nas matrículas foram favoráveis para melhorar e ampliar a cobertura da educação. O PIB quase duplicou entre A taxa bruta global de admissão (ou de acesso) ao 1992 e 1999 (tendo passado de Mt 7,4 trilhões para EP1 varia conforme as províncias e é nitidamente Mt 30,9 trilhões em preços constantes de 1995), o inferior para as raparigas do que para os rapazes. A equivalente a um crescimento de 77 por cento em taxa de acesso ou taxa bruta de admissão que é preços constantes durante esse período.3 bastante alta no 1º grau é consideravelmente Simultaneamente, o rendimento interno aumen- inferior quando se chega ao 5º grau em algumas tou 8 vezes,4 e as despesas governamentais multi- 8 MOZAMBIQUE: COST AND FINANCING OF EDUCATION plicaram-se cerca de 10 vezes.5 Apesar de o cresci- média de cerca de 3,4 anos para 4,6 anos, mas isso mento ser relativamente alto, o financiamento exter- continua a ser comparativamente baixo para a no contribuiu ainda com metade do orçamento do região: aproximadamente na mesma altura (1993), a governo, situação que provavelmente não mudará esperança de vida escolar foi calculada em aproxi- no futuro. madamente 5 anos na Tanzânia, 6 anos no Malawi, e 11 anos no Zimbabwe. 7 Se bem que a situação As despesas com a educação tenha mudado desde então em muitos desses países, a Figura 2 mostra que existe uma necessi- As despesas com a educação também subiram certa dade evidente em Moçambique não só de melhorar de cinco vezes em termos nominais durante a a eficiência na utilização dos recursos, mas também década de 1990.6 A participação das despesas de aumentar o montante global de recursos afecta- internas recorrentes com a educação nas despesas dos a esse sector. recorrentes totais do governo permaneceu estável em cerca de 16 por cento. De maneira semelhante às Recursos orçamentais, distribuição e custos de despesas governamentais globais, aproximada- construção. Em 1999 a parte mais importante do mente metade das despesas totais com a educação financiamento externo para o orçamento de (a maioria para despesas de equipamento) foi finan- educação foi utilizada na construção de escolas. Isto ciada por fontes externas. A participação das despe- não é surpreendente no período subsequente à sas totais na educação decresceu, tendo passado de guerra civil; porém, o que é singular é que o custo de cerca de 5,5 por cento do PIB em 1991-93 para construção de uma escola primária em Moçambique aproximadamente 3,5 por cento em fins da década é cerca de 5 a 20 vezes o custo de construção nos de 1990. Isto, contudo, foi devido em grande medi- países vizinhos. Embora a construção seja em geral da ao célere crescimento da economia, e não a uma mais cara em Moçambique, é evidente que esta situa- maior eficiência nos gastos educacionais devida a ção pode ser melhorada, ao equilibrar os custos taxas de repetência e de desistência mais baixas. recorrentes com os custos de investimento do capital. A participação das despesas internas com a Entre 1995 e 1999, o custo da prestação de ser- educação seguiu uma tendência descendente mais viços ou as despesas de administração representa- ou menos semelhante, sendo em média de 2,3 por cento do PIB em 1991-93, e apenas 1,7 por cento em Figura 5 fins da década de 1990. Mesmo se levarmos em Esperança de vida escolar e gastos com conta os gastos totais com a educação (e inclusiva- a educação em países seleccionados da mente o financiamento externo), a participação das África Subsariana, cerca de1995 despesas no sector foi de apenas 3,5 por cento do Esperança de vida escolar PIB, o que é muito baixo segundo os padrões inter- nacionais. É claro que está dentro das possibili- 14 dades de Moçambique aumentar o financiamento 12 nacional do seu sistema de educação. 10 O desenvolvimento educacional e os ganhos 8 em matéria de eficiência 6 Outra maneira de ver a questão é ponderar o valor 4 Mozambique obtido pelo dinheiro gasto. Quantos anos de 2 educação pode esperar receber uma criança moçambicana de acordo com a actual distribuição 0 0 2 4 6 8 10 de recursos? Entre 1995 e 1999 o número de anos Despesas globais com a educação em percentagem do PNB que uma criança pode esperar frequentar a escola em Moçambique melhorou, tendo passado de uma Fonte: Estatísticas educacionais e financeiras relativas a Moçambique e à UNESCO. RESUMO EXECUTIVO 9 ram quase 20 por cento das despesas recorrentes no medida, nos países de expressão francesa. Para o sector da educação, das quais o ministério central, o ensino secundário, a situação é sensivelmente MINED, usou mais de metade. Este montante é semelhante. Para os níveis de grau posteriores aos muito elevado, especialmente se considerarmos que do ensino primário, especialmente do ensino supe- ele é superior às despesas de administração de todas rior, as despesas por aluno são relativamente altas as escolas de EP2 e ESG1 do país. Sem contar as comparadas com as dos países vizinhos, mas despesas de administração, as escolas primárias sensivelmente semelhantes à média em todos os receberam em torno de 44 por cento das despesas países africanos. Contudo, os gastos por aluno pare- totais para o sector, ou seja, aproximadamente 52 cem ser relativamente baixos na escola primária e por cento se incluirmos as despesas de administra- relativamente altos no ensino superior. Visto que as ção. Estas proporções foram relativamente estáveis repetências e desistências fazem aumentar o custo durante o período de 1995 a 1999. O ensino secundá- unitário, são necessários custos (e recursos) adicio- rio recebeu menos de 10 por cento do total, enquanto nais para produzir um graduado (diplomado), o que o ensino superior recebeu cerca do duplo desse que subentende uma utilização ineficiente de montante. Em termos globais, estes números recursos.8 O índice de eficiência global no EP1 é de suscitam a questão do equilíbrio das despesas entre apenas 58 por cento, o que indica que 42 por cento os níveis secundário e superior de ensino no país, e (1 ­ 0,583) dos recursos públicos mobilizados no indicam que existe uma possibilidade evidente de EP1 são de facto desperdiçados9-- recursos que de melhorar a eficiência a nível central com vista a outra maneira poderiam ter sido utilizados para melhorar a prestação de serviços e os custos gerais. melhorar o sistema ou matricular novos alunos. A maioria dos recursos desperdiçados foram gastos Os aspectos de eficiência, qualidade e gestão no EP1. A eficiência dos gastos por aluno A repartição de professores nas escolas. Em e os resultados termos globais, a repartição dos recursos docentes em Moçambique é desproporcionada10 entre as A nível da escola primária, o custo unitário de escolas primárias do primeiro ciclo, calculadas em Moçambique é em média semelhante ao dos países 6.000, que existiam em 1998, tanto em termos de vizinhos, mas inferior ao custo unitário médio nos qualificações dos professores como do rácio países africanos de expressão inglesa e, em maior professor-aluno. A Figura 3 mostra que, apesar da Figura 6 Repartição dos professores nas escolas públicas do EP1, 1998 Número de professores 40 A A A A A AA A A A B A ACABB 30 AAA A A C AC A AA A A A BAA B A AA A A A A AA B BA C AB A A A A ABA C B AA A A A A A A A A ACABA ABAAA ABA A B A A 20 A ABA A BBC C D AA ABA AA A A A A AA A AAA AAAB BBBAACB GBBAAA BCD A A B A AA AB ABBBFD DCD DBEBAABBA B A AAAA A A A AAA A A A A A C BAAAFCJFCDBBCCBAB A A A DA B AA A A AAA B EDCEDGIEHDFAECE CE ACC BA BA C B A 10 A ABADHELLKOEFFFFJCGEBEEFBDCDCBDCABA BAAAA A AA AA A ADFRTSZWZVXQSKOGKODJMIKHBFDDCBBAA AA A CBFFIPZZZZZZZZZZZZZZWZNTGCBDC A A A AA CHTZZZZZZZZZZZZZZZZVJHFBB CA A GZZZZZZZZZZZZLCGACABAAA 0 0 200 400 600 800 1000 1200 1400 1600 Número de alunos Legenda: A = 1 observação, B = 2 observação, etc. Fonte: Recenseamento das escolas de Moçambique, 1998. 10 MOZAMBIQUE: COST AND FINANCING OF EDUCATION tendência de as escolas beneficiarem de um grande Os factores exógenos à escola também são número de professores quando o número de matrí- importantes, tais como a língua materna dos proge- culas é maior, o número de professores das escolas nitores, a disponibilidade de livros em casa, etc. Os com um volume semelhante de matrículas também insumos escolares, tais como, mobiliário escolar, varia imensamente. A variância na distribuição de materiais para o ensino e as salas de aula (giz, professores é maior nas zonas rurais11 do que no quadro negro, etc.) têm evidentemente uma inci- conjunto do país, mas é semelhante à variância na dência sobre os resultados. Finalmente, a educação distribuição de professores nas zonas urbanas. dos docentes e o número de anos de prática do ensino exercem uma grande influência no apro- A distribuição de professores por província e a veitamento dos alunos, o que não é surpreendente. gestão dos docentes. Em Moçambique, algumas A abordagem da questão do aproveitamento e a províncias com um número semelhante de capacidade de a pessoa aprender depois de sair da professores oferecem serviços educacionais a um escola, e aplicar os conhecimentos adquiridos, é número bastante diferente de alunos.12 Este facto crucial para melhorar o impacto da educação na sugere que existe uma grande discordância na sociedade. Esse efeito está relacionado com a distribuição de professores através das províncias. eficiência externa. Ainda mais significativo é o facto que cerca de 80 por cento da variância na repartição dos professores A eficiência externa ocorre dentro das próprias províncias. Por conse- guinte, a repartição de recursos docentes parece ser A eficiência externa diz respeito ao desempenho dos feita em função da maneira como os recursos diplomados depois de estes terem saído do sistema docentes são geridos, especialmente a nível provin- escolar e entrado na vida de trabalho. Em que cial. A distribuição dos professores disponíveis é medida os conhecimentos proporcionados a um uma questão à qual terá que ser prestada atenção grupo de jovens quando frequentavam a escola (seja nos próximos anos, especialmente à medida que o durante 1 ano, seja durante 20 anos) corresponde ao VIH/SIDA aumenta a taxa de eliminação natural melhor que se pode alcançar para essas pessoas e dos professores. Finalmente, os mecanismos e pro- para a sociedade? cedimentos orçamentais no sector deveriam clara- O ensino primário é um investimento com mente ser abordados, por ser essa a maneira de múltiplas finalidades, cujos resultados (capacidade assegurar uma distribuição adequada dos recursos. básica de ler e escrever, aritmética básica e solução de problemas, e competências básicas para a vida) A qualidade da educação contribuem para melhorar a produtividade do sector tradicional. Adicionalmente, as economias A educação em Moçambique caracteriza-se por em alguns países em desenvolvimento têm dificul- uma grande disparidade na aprendizagem em dade em absorver os diplomados do ensino secun- todas as escolas. Sem levar em conta as característi- dário e mesmo superior, o que faz com que o cas do corpo discente, a variabilidade no uso de desemprego de diplomados seja generalizado em recursos produziu um efeito muito maior na quali- muitos países africanos. Nestas circunstâncias, o dade do ensino, e portanto também nos resultados, investimento excessivo nos níveis superiores de do que a variabilidade no montante de recursos ensino é mais um peso do que um motor de cresci- disponíveis. A gestão pedagógica das escolas em mento económico. Contudo, deve-se frisar que isso Moçambique parece ser bastante débil. Qualquer é uma condição necessária, mas não suficiente, para mecanismo que seja instituído para melhorar a o crescimento económico e o desenvolvimento aprendizagem dos alunos e a qualidade do ensino social, e a fraca rentabilidade do investimento no terá que (a) medir a intervalos regulares o ensino superior é o resultado da fraca qualidade aproveitamento dos alunos a nível escolar, e (b) dos cursos e da pouca relevância destes para a eco- ajudar as escolas pouco eficientes a funcionarem nomia. Finalmente, alguns trabalhos de investiga- melhor. ção mostram que o ensino primário produz um RESUMO EXECUTIVO 11 4. De Mt 661 biliões para Mt 6,3 triliões. efeito menor no crescimento económico dos países 5. De Mt 1,5 triliões para Mt 13,5 triliões. africanos do que em países de baixos rendimentos 6. Em 540 por cento, tendo passado de Mt 286 biliões para situados noutras regiões do mundo. Mt 1.834 biliões em 1998. O elemento básico a considerar é que a economia 7. Com uma média de 4,85 anos nos países africanos de de Moçambique é, e continuará a ser nas próximas expressão francesa e 7,48 nos países de expressão inglesa. décadas, caracterizada por uma estrutura dualista, com 8. Com as taxas actuais de repetência e de desistência, apenas 42 de 100 alunos irão completar o 5º grau. Para pro- um sector tradicional dominante e um sector moderno, duzir esses 42 alunos do 5º grau seria necessário financiar 210 relativamente pequeno, e crescente em termos de (42­5) anos de escolaridade, isto é, sem contar os repetentes emprego. Qualquer estratégia para a formação do e os desistentes. A consequência é que, para finalmente pro- duzir os 4 alunos do 5º grau, o sistema teria de financiar 125 capital humano no país deverá levar em conta este (os 100 alunos do 5º grau mais os 25,4 repetentes desse grau), contexto estrutural. O objectivo fundamental é o de em vez dos 42 que seriam estritamente necessários. O mesmo proporcionar o tipo de capital humano que melhor se aplica aos graus seguintes. se coaduna com a procura dos dois sub-sectores, 9. Também podemos usar o coeficiente de eficiência para ter uma ideia do número médio de anos que o orçamente precisa ajudando simultaneamente a economia a desen- de financiar para ajudar uma criança, em média, a chegar ao volver-se, passando do tradicional ao moderno e, a 5º grau: leva 8,6 anos para que todos os alunos (inclusiva- longo prazo, aliviando a pobreza. mente os repetentes e os desistentes) passem por todo o EP1; por outras palavras, custa ao governo mais 3,6 anos de gastos por aluno para que uma criança possa chegar ao 5º grau. 10. No que respeita a distribuição quantitativa de professo- 1. A taxa bruta de matrícula é o número total de alunos res, o objectivo é o de fornecer às escolas com o mesmo nú- matriculados, irrespectivamente da idade, de todo o grupo mero de matrículas um número semelhante de professores. populacional de uma idade específica (por exemplo, os 11. Por exemplo, entre as escolas que têm pelo menos 10 alunos entre os 6 e os 10 anos no caso do EP1); por professores, as matrículas variam entre 100 e 1.200 alunos; de conseguinte, ela é por vezes superior a 100 por cento. maneira semelhante, as escolas nas quais estão matriculados 2. A taxa bruta de admissão (ou taxa bruta de matrícula por 600 alunos poderão ter entre 5 e 15 professores. grau) é o número total de alunos matriculados num grau 12. Por exemplo, Cabo Delgado e Inhambane têm ambos específico, irrespectivamente da sua idade, num grupo cerca de 2.500 professores, mas o número de matrículas em populacional específico de idade (por exemplo, na Figura 1 a Cabo Delgado é em torno de 134.000 alunos, isto é, 41.000 taxa bruta de admissão ou a taxa bruta de matrícula no 4º menos do que em Inhambane que tem 175.000 alunos. grau na província de Cabo Delgado foi de 48 por cento do Paralelamente, nas províncias tanto de Tete como de Maputo número total de coortes do grupo etário de 9 anos. foram matriculados cerca de 140.000 alunos em cada uma, 3. O PIB cresceu num factor de 10 em termos nominais entre mas Tete tem 3.000 alunos, enquanto que Maputo tem menos 1992 e 1999. de 2.000. 1 Macroeconomic and Demographic Context T he characteristics of an educational system Mozambique was one of the poorest countries in the and the development of education are deter- world. However, since the government's adjustment mined largely by macroeconomic and demo- program began in 1987 and peace was declared in graphic conditions and by the capacity of the 1992, Mozambique's economy has experienced system to absorb existing and increasing resources strong growth. Economic liberalization and large and to use them efficiently. Other important aspects inflows of external aid have contributed to signifi- relate to the social, cultural, and religious contexts cant economic recovery, albeit from very depressed within which the system operates. Although these levels. The economic liberalization policies were constraints and conditions set the framework under made even more effective by the signing of the Gen- which an education system can develop, the level of eral Peace Agreement in 1992, which ended the state education, in turn, is a determinant of economic of civil war, and by the return of democratic rule fol- development and, depending on the choices made, lowing the 1994 elections. The combined effects of has an impact on poverty alleviation in the country. these events have exceeded the expectations of both In this chapter, we focus on the broad characteristics national policymakers and the international donor of the macroeconomic and demographic context in community, and Mozambique is now one of the Mozambique and the capacity of the system. fastest-growing economies in the world. At the same time, annual inflation fell from around 50 percent in Evolution in gross domestic product 1995­96 to around 5 percent in 1998­99. and fiscal resources As shown in table 1.1, GDP almost doubled between 1992 and 1999 (from Mt17.4 trillion to For almost 20 years after it became an independent Mt30.9 trillion in constant prices of 1995), equiva- country in 1975, Mozambique stagnated at lent to a 77 percent growth in constant prices over extremely low levels of per capita income as a result that period. Measured against the growth in pop- of both the weak physical and human capital infra- ulation, estimated at an annual rate of 2 percent, structure inherited from the former colonial power GDP per capita increased by more than two and the centralized model of economic management thirds: from Mt1.2 million per capita (1995 con- adopted by the socialist government. In addition, stant prices) in 1992 to about Mt1.9 million in 1999 the protracted civil war between 1976 and 1992 dec- (this translates into an increase from US$144 per imated the scant capital infrastructure that existed or capita in 1992 to about US$260 in 1999 at current was being created in the country. With an average prices and exchange rates) implying a real annual per capita gross domestic product (GDP) increase in per capita GDP of about 10 percent a hovering close to US$100 through the 1980s, year. From this perspective, Mozambique has 12 MACROECONOMIC AND DEMOGRAPHIC CONTEXT 13 Table 1.1 Population, GDP, and total public expenditures, 1990­99 Indicator 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Population (millions)a 14.1 14.2 14.4 14.6 14.7 14.9 15.1 15.3 15.7 16.1 GDP (current trillions of meticáis) 2.3 3.6 4.8 7.8 13.1 21.3 32.1 397 46.2 52.9 GDP (constant 95 trillions of meticáis) 18.1 19.0 17.4 19.0 20.4 21.3 22.8 25.4 28.4 30.9 GDP per capita, US$ (new population census) 179 175 136 144 151 160 188 225 249 259 Real growth rate per year (percent) -- 5 ­8 9 7 4 7 11 12 9 Ratio of total government expenditures to GDP (percent) 29.6 26.8 31.2 29.4 31.2 24.2 21.1 23.9 22.1 25.4 Ratio of revenue to GDP (percent) 12.8 12.5 13.9 14.0 11.6 11.3 10.8 11.6 11.5 11.8 Exchange rate (metacal to U.S. dollar) 929 1,435 2,433 3,723 5,918 8,890 11,294 11,546 11,850 12,711 a. Population growth from 1990 to 1997 is based on a growth rate of 2 percent per year. After 1997 the growth rate is estimated at 2.3 percent. Source: GDP, total government expenditures: International Monetary Fund; public education expenditures: Ministry of Education; GDP per capita in U.S. dollars: estimates based on 1997 population census figures and end-of-year exchange rate. been doing quite well. This is confirmed by the In absolute terms, the resource envelope, meas- fact that domestic revenues (in current prices) ured as both per capita income and total public increased almost nine times (847 percent), from spending, has grown substantially since 1992. This Mt661 billion to Mt6.3 trillion, over the same implies that increasing investments in education are period. Although impressive, this growth in possible. One important caveat is the high and domestic revenues was slower than the growth in increasing dependency on external grant funding, a GDP, which may reflect a lag in the government's reflection of the government's difficulty in raising ability to mobilize public funding. revenues at the same pace. As shown in table 1.1 and Annex 2, domestic rev- To further consider the prospects for educational enue as a share of GDP declined from 14 to about 12 development, it is important to assess the outlook for percent between 1992 and 1998. Moreover, during economic growth and an increase in revenues. The this period Mozambique's external grant financing Mozambican economy is expected to continue grow- grew by a factor of nine (when external assistance is ing at 7 to 8 percent a year over the next five years. To expressed in meticáis terms). External funding still estimate the impact of this growth on revenues, we finances about half of the government's expendi- look at the composition of the labor force and the tures, the largest share being for investment items. economic sector's contribution to GDP (formal sector At the same time, as shown in table 1.3 total public contribution to revenue), as shown in table 1.2. expenditures grew about eight times, from Mt1.5 The National Statistical Institute (INE) estimates trillion to Mt13.5 trillion between 1992 and 1999; in that the labor force was 5.6 million in 1980, 6.0 mil- other words, total expenditures grew more slowly lion in 1990, and 9.7 million in 2000, an increase of than revenue. This means that dependence on exter- about 60 percent between 1980 and 2000.1 However, nal financing has remained unchanged, and the although the relative share of people working in government's budget will continue to rely on exter- agriculture has declined slightly, agriculture still nal support to remain sustainable. These character- employs about 79 percent of the population. Using istics are further exemplified by the share of total the figures presented in table 1.2, only about 9 per- public expenditures in GDP, which fell from about cent of the workforce, or about 800,000 people, are 31 to 25 percent during the same period. employed in the formal sector,2 of which about 14 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 1.2 will increase as a result of HIV/AIDS (as in the Output and employment, 1998 short run output remains the same, but workers die).4 However, the issue is whether HIV/AIDS will Value added Share of formal affect the average output per worker. Judging by Millions of Share of labor force the case of Uganda, HIV/AIDS will have a negative U.S. dollars total (percent) (percent) effect on average productivity, productivity that Agriculture 1,224 34.5 79.0 would normally be an alternative means to increase Industry 742 20.9 6.0 the value added in the formal sector and thereby increase revenues. In addition to the long-term Services 1,578 44.5 15.0 effects on average productivity per worker, produc- Total 3,544 100.0 100.0 tivity levels in Mozambique will continue to be depressed by the already low level of education and Source: World Bank, "Growth Performance and Reform Agenda," Report No. 20601-M2, June 30, 2000. the resulting low productivity. As a whole, in the immediate future Mozam- 120,000 are employed in the public sector (exclud- bique will continue to be dependent on the growth ing the armed forces). The formal sector accounts of the formal economy and a small formal sector for more than one-third of gross domestic product (tax-paying) labor force in addition to revenues and about 60 percent of domestic revenues. While generated from taxes on primary products and on Mozambique's favorable prospects for long-term value added. growth are strengthened by the country's untapped economic potential in natural resources,3 future rev- Evolution in education spending, absorptive enues will be generated largely from the formal capacity, and highly indebted poor countries sector (agricultural exports, industry, and the serv- ice sector). Therefore, the fiscal burden is likely to be As can be seen from table 1.3, total spending for the borne by a small fraction of the labor force in the education sector, including higher education, grew, near future. in nominal terms, about four times between 1992 With the introduction of the value added tax, the and 1998, from Mt286 billion to almost Mt1.8 tril- country's fiscal base may expand to cover more tax- lion. However, a more precise picture emerges if the payers. But fiscal pressure is only one measure of evolution in constant 1995 prices is considered. As the prospects for increasing revenues. The other is can be seen from table 1.3 total education expendi- the size of the labor force compared with the size of tures fluctuated significantly over the period with the nonworking population, which is known as drastic decline and increases during the periods demographic pressure or the dependency ratio. The 1993-1995 and after 1996. Overall from 1992 to 1998 dependency ratio, which provides a proxy for the total education expenditures grew 9.4% over the fiscal burden borne by the working population, period or about 1.5 percent per annum, while the declined until 1995, easing slightly the fiscal burden domestically financed share of the expenditures as the labor force expanded and life expectancy grew by twice as fast or about 2.9 percent per rose. However the ratio has been consistently on the annum. The volatility in the level of education rise since 1997. In order to forecast the ratio in the spending demonstrate the dependency of external future, we have to take into account the spread of financing and difficulty in predicting flows of HIV/AIDS. resources to the sector. External funding (mostly for If HIV/AIDS in Mozambique follows the pattern capital expenditure) represents a large share of total of its neighboring countries, it will surely reduce spending: about half of total spending on education life expectancy and have a severe impact on the until 1997 and 45 percent in 1998. The absolute population's number of active working years and increase in funding of education, especially domes- productivity. Although the dependency ratio in the tically financed, is a positive development. How- short run will remain stable or may even improve ever, as will be discussed, the share of education slightly, the marginal output per worker probably spending of GDP has shown a slightly downward MACROECONOMIC AND DEMOGRAPHIC CONTEXT 15 Table 1.3 Public Spending on education as a share of GDP and total government spending, 1990­1999a 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 GDP (billions of meticáis) 2,334 3,580 4,757 7,829 13,145 21,267 32,093 39,693 46,134 52,913 Total government spending (billions of meticáis) 690 958 1,483 2,305 4,097 5,157 6,773 9,498 10,207 13,460 Total government of spending as a percentage of GDP 29.6 26.8 31.2 29.4 31.2 24.2 21.1 23.9 21.8 25.2 Ministry of Education spending 109 171 263 380 653 799 1,300 1,322 1,468 1,645 Total domestic expenditures (billions of meticáis) 50 75 125 171 203 268 486 648 875 1,005 Total external expenditures (billions of meticáis) 59 96 138 209 450 531 814 674 593 640 Ministry of Higher Education spending -- -- -- -- -- -- -- 237 400 476 Total domestic expenditures (billions of meticáis) 7 12 22 34 58 65 109 123 163 21 Total external expenditures (billions of meticáis) -- -- -- -- -- -- -- 114 237 262 Total domestic expenditure on education (billions of meticáis) 57 87 148 206 260 333 595 771 1,038 1,219 Total external expenditure on education (billions of meticáis) (59) (96) (138) (209) (450) (531) (814) 788 830 902 Total expenditure on education (domestic + external) (116) (183) (286) (415) (710) (864) (1,409) 1,559 1,868 2,121 Total expenditures on education (domestic + external in constant 95 billions of meticáis 900 971 1,049 1,005 1,101 864 1,000 996 1,148 1,239 Total domestic expenditures on education (in constant 95 billions of meticáis) 442 461 543 499 403 333 422 493 638 712 Annual growth rate in total domestic education expenditures on education in constant 95 prices 4% 18% ­8% ­19% ­17% 27% 17% 30% 12% Total expenditure on education as a percentage of GDP) 5.0 5.1 6.0 5.3 5.4 4.1 4.4 3.9 4.0 4.0 Domestic spending on education as percentage of GDP 2.4 2.4 3.1 2.6 2.0 1.6 1.9 1.9 2.2 2.3 Exchange rate (meticáis per US dollar) 929 1,435 2,433 3,723 5,918 8,890 11,294 11,546 11,850 12,711 --- Not available. a. Figures denote actual expenditures, except for 1999, wich denotes the amount budgeted. Source: Ministry of Education and World Bank, 1997 Mozambique Population Census. (Washington, D.C.: World Bank, 1997 trend. This combined with the increase enrollments as a whole, the share was reasonably stable over the over the same period point to a concern about the decade, around 16 percent (with a low of 10 percent overall funding of education. in 1994). When spending on education is contrasted with Measured against GDP, the share of education government spending, the share of the education spending declined from about 5 percent in 1992 to sector in total public expenditures declined from about 4 percent in 1998. The share of domestic around 19 percent in 1991 to 18 percent in 1998. If spending on education in GDP followed more or we focus on domestic spending and on recurrent less a similar trend, with figures averaging 2.6 per- expenditure for both the sector and the government cent in 1990­93 and about 2 percent in the late 1990s. 16 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Figure 1.1 cial and district levels, as well as to difficulties in Overspending and underspending in preparing annual budgets. proportion to total budget, 1990­98 Despite weakness in budgetary mechanisms and in the country's capacity to implement budgeted Ratio of effective to budgeted spending (percent) actions, arguments exist for expanding the system 130 in both effective coverage and quality. This will 120 undoubtedly require the mobilization of additional Overspending resources. In the context of HIPC and debt relief, 110 conditions should be favorable to a move in that direction. 100 Balanced budget In 1998, the domestic education budget was 90 about US$85 million, and it is expected that debt 1990 1991 1992 1993 1994 1995 1996 1997 1998 cancellation will release, on average and on a yearly Year basis, about US$120 million for the coming years. This amount is expected to grow gradually, reach- Note: Overspending or underspending is defined as the ratio between the amount effectively spent at the end of the year and the approved budget for that year. A ing about US$167 million in 2017. Even if only one- balanced budget is indicated by 100 percent, while any block above 100 percent represents spending above the budget (and vice versa). quarter of these annual amounts is allocated to the Source: Ministry of Education, Department of Administration and Finance. education sector, this would still amount to an addi- tional US$30 million in 2001 and about US$40 mil- lion in 2017. Though the increase would help to Domestic spending only produces the even lower bring the share of domestically funded education in figure of 1.7 percent of GDP allocated to education. GDP from 1.7 percent to something below 2.5 per- These figures are very low indeed, since domes- cent, this figure remains low. Even as the HIPC tic spending comprises about 4 percent of GDP, on resources can help, the development of the sector average, in African highly indebted poor countries will require additional resources and more effective (HIPCs). Although the fiscal capacity of Mozam- use of said resources. bique is on the low side (around 12 percent of GDP), it is not worse than that of Niger or Burkina Faso, The likely impact of HIV/AIDS which spend around 3 percent of their national on children and teachers product on education. Mozambique allocates a much smaller share of total government spending Developing the system of education in Mozam- to education; the figure was around 14.5 percent in bique entails increasing the number of children 1997­99, compared with around 25 percent in a enrolled in school and the number of teachers number of African countries. recruited. However, the prevalence of HIV/AIDS There clearly is scope for the country to channel undoubtedly will affect the educational system and more finances toward funding of its education its prospects for both children and teachers. Clearly, system, even while external funding may continue the supply of and demand for teachers will be to play a significant role. The absorptive capacity of affected: sick teachers likely will miss classes, and the country and the budgetary mechanisms and the attrition rate from an increase in adult deaths procedures in the sector need to be improved in likely will rise, necessitating a larger recruitment of order for these resources to be effective. Figure 1.1 teachers every year. Based on some reasonable shows that the education sector was overspending assumptions, Annex 3 shows the development of in the early 1990s, but since 1995 has not been able total population, school-age population, teacher to spend all of the resources allocated to it. This is stock, and the dependency ratio over a 10-year worrisome in light of persistent claims that more period with and without HIV/AIDS. The growth resources are needed to run the sector effectively. rates for the total population and school-age popu- Underspending in the sector could be due to a lation are assumed to be 2.3 and 3.0 percent, respec- shortage of liquidity or cash flow transfer to provin- tively, in the absence of HIV/AIDS.5 MACROECONOMIC AND DEMOGRAPHIC CONTEXT 17 The attrition rate under the scenario without clearly has important implications for the develop- AIDS reflects current attrition rates due to retire- ment of the pre- and in-service training system, as is ment, death, or change in employment. This sce- discussed later in this report. nario assumes that, with the current growth in the Another critical issue that has not been touched school population (3 percent in 1998), the number of on is the impact that HIV/AIDS will have on the teachers should increase from 40,000 in 1997 to conditions under which education will be deliv- 58,741 in 2010. Furthermore, given the number of ered. With a larger number of orphans and a higher teachers who would retire, die, or change jobs, as level of teacher and student absenteeism due to ill- reflected by the annual attrition rate of 4 percent, in ness, delivering quality education will become addition to the new teachers required by 2010, an increasingly difficult. Alternative and flexible deliv- estimated 2,350 teachers would need to be replaced ery modes have to be considered, and putting due to attrition. schools closer to the communities may become Based on evidence from the HIV/AIDS pan- increasingly important. demic in Uganda, Malawi, and Zambia, the sce- nario with AIDS assumes that the growth rates in total population and school-age population will 1. This is based on population projections that proved to be overestimated, as illustrated by the head count of the 1998 decline, respectively, by 0.8 and 0.2 percent, which population census; in fact, the labor force in 1999 was proba- would bring them down to 1.5 and 2.8 percent; both bly around 7.5 million, rather than the 9.7 million of the INE population groups would grow at a slower pace, projection (suggesting an increase of 33 percent instead of 60 percent in the size of the labor force between 1990 and 1999). although growth of the school population would 2. Employment in industry includes one-third of services (5 remain almost unchanged. It is assumed that the percent of total formal sector employment), including public attrition rate due to death from AIDS would administration, and 10 percent of the agricultural sector increase 3 percentage points to 7 percent. employment (the lion's share of employment in the agricul- tural sector is small-scale or subsistence farming); this figure HIV/AIDS will impact the fertility and mortality is consistent with the proportion directly observed in the rate of women, resulting in a lower growth rate in 1996 household survey. the population of school age children. Therefore, 3. For example, although agriculture accounts for about one- third of GDP and 80 percent of exports (representing 6 per- fewer teachers will be needed to maintain the ratio cent of GDP, or US$233 million, in 1998), less than 10 percent of pupils per teacher. Instead of requiring 58,741 of arable land is currently under cultivation. teachers by 2010, only 57,276 teachers will be 4. Jill Armstrong, Uganda's AIDS Crisis: Its Implications for required because of the slightly smaller school-age Development, World Bank Discussion Paper 298 (Washington, D.C.: World Bank, 1995). population. However, because HIV/AIDS also will 5. The scenario in Annex 3 builds on the assumption that affect the mortality rate of teachers, the replacement there will be no change in coverage (no change in enrollment of teachers due to deaths caused by AIDS will rates) or in the ratio of the number of pupils to the number of increase 70 percent in 2010, from 2,350 teachers due teachers. Thus it describes the system as it was in 1998 and its evolution with or without the impact of HIV/AIDS. The 1997 to normal attrition to 4,009 teachers due to normal figures for total population and school-age population are attrition and the impact of AIDS. based on the 1997 Mozambique census. The school-age pop- On average, calculated over the period from 1997 ulation covers the age group 6 to 15 years old, or grades 1 to 10 (lower and upper primary education and lower secondary to 2010, the government will have to increase the education); the teacher stock is based on 34,000 lower pri- annual replacement of teachers by 50 percent as a mary teachers, 4,000 upper primary teachers, and 2,000 lower secondary teachers, as registered in 1997. result of deaths from AIDS. This phenomenon 2 Enrollment Pattern C hapter 1 described education in the context The government also provides technical educa- of macroeconomic trends in government tion and vocational training, teacher training, and revenue and expenditures, education adult education. Several public and private spending, and issues related to demographic providers offer technical education and vocational changes and the impact of AIDS in Mozambique. training.2 The Ministry of Education offers three This chapter describes the overall education system levels: (a) elementary technical education, (b) basic and changes in enrollment patterns by level and technical education, and (c) middle technical edu- type of education. It then examines the link between cation. The basic and middle levels offer three areas attainment of education and public education of specialization: agriculture, industry, and com- spending and investigates student flow by present- merce. In relation to general education, the elemen- ing access, survival, and transition rates. Finally, it tary level lasts for two years and is equivalent to provides possible explanations for factors related to EP2; the basic and middle levels last for three and the demand for or supply of education. four years, respectively, and are equivalent to ESG1 and ESG2. Structure of the educational system Teacher training institutes are divided into three categories: (a) basic level (primary teacher training Today, formal education in Mozambique comprises institutes, before 1998), (b) middle level, and (c) an a 7-3-2-4 system: seven years of primary education, additional middle level after 1998, known as IMAP. three years of lower secondary education, two years To qualify for basic-level entry, a student needs to of upper secondary education, and four years of complete grade 7, while a tenth-grade graduate is higher education. In addition, preprimary educa- qualified to enter a middle-level teacher training tion consists of kindergarten and nursery schools institute. Graduates from teacher training institutes for children under the age of 6. Primary education are qualified to teach only in primary schools. In is compulsory and targets children between the order to become a qualified secondary school ages of 6 and 13. It is divided into the lower level teacher, a student has to graduate from the Peda- (grades 1­5), known as ensino primario 1 (EP1), and gogical University or Eduardo Mondlane Univer- the upper levels (grades 6­7), called ensino primario sity. The requirement for entrance into these institu- 2 (EP2); these, however, will be gradually merged. tions is ESG2 (grade 12). Lower and upper secondary education is known as Adult education, which is targeted to persons 15 ensino secundario geral, ESG1 (grades 8­10) and years of age or older, includes primary, general, and ESG2 (grades 11­12). Higher education consists of middle levels of instruction. These three levels are three public and three private institutions.1 equivalent to EP2, ESG1, and ESG2, respectively. 18 ENROLLMENT PATTERN 19 After Mozambique's independence in 1975, to put the situation observed in 1998 in a time per- evening classes were created for adults to partici- spective. Table 2.2 shows the evolution in the pate in literacy campaigns or in technical and gen- number of students enrolled as well as the gross eral education. enrollment rate (GER)3 across the different levels and types of education between 1993 and 1999. Enrollment trends Primary education (EP1 and EP2) The distribution of enrollment by level of education in 1998 is presented in table 2.1. The Mozambican As seen in table 2.2, the enrollment in EP1 grew by education system could be described figuratively as 871,000 between 1993 and 1999, from 1.2 million an "L-shape", with the bulk of students enrolled in pupils to 2.1 million pupils. This is equivalent to a lower primary education. Less than 8 and 3 percent 32 percentage-point increase in the gross enrollment of the total student population are enrolled in upper rate, from 60 to 92 percent during the same period. primary (EP2) and lower secondary education Figure 2.1 depicts graphically the dramatic changes (ESG1), respectively, while 87 percent of total enroll- in the gross enrollment rates in EP1. The increase in ment is in lower primary education (EP1). This sug- enrollment is partly a result of the rapid expansion gests that the overall structure of education in the in the number of primary school classrooms, which country is not in balance, leading to questions about nearly doubled between 1993 and 1999, and which both internal efficiency and issues of supply and were built mostly in rural areas and in areas previ- demand. However, the actual structure of enroll- ously inaccessible due to the war. ment is affected by two facts: the system is rela- During the same period, the number of students tively new, and primary education has expanded so enrolled in EP2 also increased nearly 65 percent, recently. The younger and larger cohorts have not from 119,000 students in 1993 to 193,000 students in yet reached the higher levels of schooling. 1999. However, the change in the gross enrollment To assess whether the current shape of the rate was not as large as that in EP1. The gross enroll- Mozambican education system is a result of the ment rate in EP2 increased only 8 percentage points: recent expansion in primary education, it is useful from 17 percent in 1993 to 24 percent in 1999. During the same period, the number of classrooms in EP2 increased by 1,500: from 2,600 in 1993 to Table 2.1 4,100 in 1999. One of the possible reasons for the Distribution of enrollment, by level of schooling, 1998 low enrollment in EP2 could be the lack of class- rooms (schools). Number Level of schooling of students Percent Secondary education (ESG1 and ESG2) Preschool 18,000 0.8 Access to secondary education is very limited, and Lower primary (EP1) 1,880,949 86.7 few Mozambicans proceed to this level. Both lower Upper primary (EP2) 168,777 7.8 and upper secondary education are important because they produce a professionally trained and Lower secondary (ESG1) 53,693 2.5 qualified workforce and prepare people for higher Upper secondary (ESG2) 7,352 0.3 education and eventual positions of leadership in Technical and vocational education 27,068 1.2 the public and private sectors. Graduation from ESG1, or the tenth grade, is particularly important, Teacher training 4,855 0.2 as this level of education is required to enter teacher Higher education 8,536 0.4 training institutes. Table 2.2 shows that the absolute number of pupils in secondary education increased Total 2,169,230 100.0 rapidly between 1993 and 1999. However, the gross Source: Mozambique school survey data and census, 1998. enrollment rate remains very low at 6.6 percent in 20 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 2.2 Enrollment and gross enrollment rates, by level and type of education, select years, 1993­99 1993 1995 1997 1999 Number of GER Number of GER Number of GER number. of GER students (percent) students (percent) students (percent) students (percent) Primary EP1 1,237,063 60.3 1,436,831 68.3 1,780,881 82.7 2,108,790 92.3 EP2 118,909 16.7 136,464 18.7 164,078 22.0 193,523 24.4 Secondary ESG1 31,761 3.2 40,588 4.0 51,821 4.9 72,914 6.6 ESG2 1,612 0.3 4,927 0.8 4,253 0.6 9,142 1.3 Technical or vocational 17,637 20,810 23,119 20,996 -- Not available. Source: School survey data, 1993, 1995, 1997, and 1999; census data, 1997. ESG1 and 1.3 percent in ESG2, again indicating a shown, include a low internal efficiency and low very low admission rate. Figure 2.2 presents these survival rates at each grade level, and finally the changes graphically. number of students and admission to secondary As the GER for lower primary education before education. 1993 remained at the same level, the pupils who were enrolled in the first grade in 1993 or before Teacher training should have proceeded to secondary education, which they did not. Therefore, the recent expansion The overall number of students enrolled in teacher does not seem to have resulted in an increased training institutions for primary teachers increased admission and enrollment into secondary educa- by 1,500: from 4,116 in 1993 to 5,683 in 1999 (see tion in related terms. The reasons for this, as will be Annex 4), utilizing the capacity of the institutions to Figure 2.1 Gross enrollment rates in primary education, Figure 2.2 1993­99 Gross enrollment rates in secondary education, 1993­99 Percent 100 Percent 7 90 80 6 70 5 60 EP1 4 50 EP2 ESG1 ESG2 40 3 30 2 20 1 10 0 0 1993 1994 1995 1996 1997 1998 1999 1993 1994 1995 1996 1997 1998 1999 Source: Mozambique school survey data and census, 1993­99. Source: Mozambique school survey data and census, 1993­99. ENROLLMENT PATTERN 21 their limits. Every year nearly 1,000 students gradu- in 1999. But progress is being made and enrollment ate from the three levels of teacher training insti- in higher education institutions was gradually tutes. In addition to this, the number of graduates increased in the past six years. from the ADPP teacher institute was expected to rise Of Mozambique's three universities and four from 180 graduates in 1998 to about 400 by 2001.4 higher education institutions, the Eduardo Mond- Secondary level (grades 8­12) teacher training is lane University is the largest, with an enrollment of undertaken by the Pedagogical University and the nearly 7,000 students in 1999. While enrollment in Eduardo Mondlane University education faculty Eduardo Mondlane University grew by nearly 3,000 with different entry requirements for teachers at students over the past six years, enrollment in the different grades. Altogether, about 150 students Pedagogical University also increased by about 800 graduate each year from the Pedagogical University students (see Annex 4), to about 2,000. The Higher and Eduardo Mondlane University. Institute of International Relations, the other public Attrition among teachers and growth in the institution, had an enrollment of 234 students in school-age population outweigh the slow increase 1999. (The Naval Academy is not discussed in this in the number of graduates from the teacher train- study.) Enrollment in private institutions has ing institutions in recent years. To attract more stu- remained low, however, they are taking on an dents to the teaching profession and retain them, increasingly critical role. It is estimated that, by the government adopted new salary scales for the 2002, more than one-third of students will be public sector, including teachers, in 1999. enrolled into a private higher education institution. Technical education and vocational training Adult education Enrollment in technical education and vocational Although adult education is not discussed in detail training increased by about 3,000 students between in this study (other than the brief discussion in 1993 and 1999 (see Annex 4).5 The shares of enroll- chapter 6 on the issue of external efficiency), it ment in basic and middle levels for 1998 were 10 should be noted that the adult literacy rate (within percent for agriculture, 40 percent for commerce, the age group of 18 to 65 years) in Mozambique was and 50 percent for industry, with industrial and estimated at 40 percent in 1997, and the rate for commercial programs clearly favored over agricul- females was even lower (24 percent). Both rates tural programs. were lower than the Sub-Saharan average of nearly 60 percent (almost 50 percent for females). Higher education Conclusions on educational enrollment Many empirical studies mention the importance of science and technology for economic growth, and In conclusion, regarding the pattern of educational the need for a skilled workforce in that sector to effi- enrollment in Mozambique, the system's character- ciently use available technologies. In Mozambique, istic "L-shape" of enrollment by grade is a conse- the importance of this area is well recognized. quence of the clear focus that was placed on pro- UNESCO indicators show that the share of natural viding physical access to education, especially sciences, engineering, and agriculture in Mozam- primary education, in the years after the war. This bique's tertiary education was above that of other focus succeeded in increasing enrollment at all Sub-Saharan African countries in 1998. Mozam- levels. However, it does not appear to have changed bique had the largest share, estimated at 50 percent, the internal efficiency of the system and therefore among the Sub-Saharan African countries, followed the attainment levels. by Nigeria (41 percent), Tanzania (39 percent), and Another way to view this pattern is to look at Ethiopia (36 percent). However, the higher educa- school life expectancy, which is the sum of the age- tion system in Mozambique is relatively small, and specific enrollment rates for primary, secondary, the gross enrollment rate was only about 1 percent and tertiary education. As previously shown, 22 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Figure 2.3 resources that are mobilized by the country for the School life expectancy and Spending on sector. Total public spending on education repre- education in select Sub-Saharan African sented 4.1 percent of the country's gross national countries, around 1995 product (GNP) in 1995 and only 3.4 percent in 1999. School life expectancy Efficiency improved from 1995 to 1999, not as a 14 result of informal spending, but because the econ- 12 omy grew faster than government spending, as explained earlier. However, Mozambique remains 10 low compared to Francophone and Anglophone 8 countries, as well as low-income Asian and Latin 6 American countries.6 4 To illustrate the level of efficiency of resource Mozambique 2 use for education coverage, Figure 2.3 plots school life expectancy on the y axis and spending on edu- 0 0 2 4 6 8 10 cation as measured by the share of GNP on the x Overall expenditure on education as a percentage of GNP axis. Figure 2.3 shows the location of select Sub- Saharan African countries in these two dimensions Source: Educational and financial statistics for Mozambique and UNESCO. around 1995. Mozambique's position in this graph, mapped as the large dot, is significant. The arrow indicates the move taken by the country between because both overall resources and enrollments 1995 and 1999. increased, a logical expectation is for the number of In 1995, Mozambique spent 4.1 percent of public years of formal education to increase as well; this is expenditure on education as a percentage of GNP, discussed next. and school life expectancy was estimated as 3.4 years. As can be seen in figure 2.3, the large dot rep- Overall index of educational development and a resenting Mozambique was way below the frontier global measure of efficiency line linking the most efficient countries (those whose coverage was maximum given what they Between 1995 and 1999, school life expectancy in spent). For example, Cameroon spent a similar Mozambique improved from 3.4 to 4.6 years. How- amount of public resources on education (3.7 per- ever, in spite of the progress made, the figure cent of GNP), but its education outcome was more remains relatively low by international standards. than double that of Mozambique (8.1 years). One For 1993, the year for which international compari- reason is that in Cameroon private contributions to son data are available, the average school life education are bigger. Zambia also managed public expectancy was about 5 years in Francophone resources effectively, spending only 2.6 percent of African countries and about 8 in Anglophone GNP on education, but attaining an outcome of 8 African countries; it was about 8 in low-income years of school life. If Mozambique would use its countries of Asia and 9 in those of Latin America. If resources more efficiently, it could improve school we focus on the neighboring countries of Mozam- life expectancy with the same amount of resources. bique, the school life expectancy was estimated to In 1995, there was substantial potential for be about 8 in Zambia, 5 in Tanzania, 5 in Malawi, improvement in the country's school system. and 11 in Zimbabwe. These figures suggest that the Between 1995 and 1999, school life expectancy overall quantitative coverage of education in increased (from 3.4 to 4.6 years), while spending on Mozambique is still lagging behind that of most education, as a proportion of GNP, declined from 4.1 comparable countries in the region and elsewhere. to 3.4 percent. By definition, efficiency improved, as To get a sense of efficiency in resource use, at documented in figure 2.3 by the arrow and by a least as far as coverage is concerned, it is useful to move toward the pseudo-efficiency frontier. Never- contrast these statistics with the amount of theless, there is still room for improvement. ENROLLMENT PATTERN 23 Pattern of student flow over the system and the inefficiency of the system of education in Mozambique. More students are able to enroll in formal education Figures 2.4 and 2.5 illustrate the continual than ever before, and the increases in the gross decline from one grade to the other, which is most enrollment rates and school life expectancy have significant between grades 5 and 6 (the transition begun to reflect this improvement in access, partic- from EP1 to EP2). This continual decline between ularly the rapid growth in the gross enrollment rate each grade in EP1 and EP2 reflects the large number in EP1. However, the enrollment rates in EP2 and of dropouts in each grade; the system also has a higher levels of education remain relatively low. fairly high rate of grade repetition (24 percent in This section investigates the efficiency of student EP1, 25 percent in EP2, and 24 percent in ESG1 on flow in education, particularly in EP1 and EP2, by average). This high level of repetition generates a presenting the rates of school access, survival, and high gross enrollment rate in EP1 and disguises transition. high dropout that should be a major concern for As table 2.3 shows, the admission rate into the education policy for the years to come. first grade was estimated at 94 percent in 1998, a strong indication that Mozambique has succeeded, Figure 2.4 or is gradually succeeding, in providing educa- Admission rate in primary education, 1998 tional access for all. The picture becomes less rosy as 100.0% pupils progress through the system. The high access 90.0% rate in grade 1 declines dramatically to a low 35 per- 80.0% cent in grade 5 and continues to decline to 14 per- 70.0% cent in grade 7. When pupils reach grade 12, they 60.0% represent only 1 percent of their cohort. In other 50.0% words, hypothetically, if 100 pupils were to enter Percentage 40.0% grade 1, only 37 pupils would survive to grade 5. By 30.0% grade 7, only 15 pupils would survive. After that, 20.0% the number of pupils would continue to decline 10.0% until only 1 pupil would enter grade 12. These fig- 0.0% 1 2 3 4 5 6 7 8 ures underline the low level of survival (retention) Grade Source: Mozambique school survey data, 1998. Table 2.3 Pattern of school access and survival, 1998 (percent) Figure 2.5 Grade Access rates Survival rates Admission rate in secondary education, 1998 Student in grade 1 93.7 100.0 16.0% Student in grade 5 34.9 37.3 14.0% 12.0% Student in grade 6 20.8 22.2 10.0% Student in grade 7 14.3 15.3 8.0% Student in grade 8 6.4 6.8 Percentage 6.0% 4.0% Student in grade 10 3.1 3.4 2.0% Student in grade 11 1.6 1.7 0.0% 7 8 9 10 11 12 Student in grade 12 1.1 1.2 Grade Source: Mozambique school survey data, 1998. Source: Mozambique school survey data, 1998. 24 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Using access and survival rates by each grade, more times a student repeats, the higher chance that this study also examines how the transition rates student will drop out). This cycle generally may be vary from one level of education to the other when explained by demand-side or supply-side factors. combined with survival rates. Hypothetically, Two specific cases are considered here. although 37 out of 100 pupils survive by the end of The first case occurs when a student who gets to EP1, only 22, or 60 percent of these 37 pupils, move grade j during a certain year t cannot move to grade up to EP2 (see table 2.4). Similarly, nearly 70 percent j + 1 the next year because his or her school is of these 22 pupils survive EP2, but fewer than half "incomplete"; in other words, it does not offer that of the surviving pupils then move on to ESG1. Since grade. Therefore, the pupil cannot be said to drop the survival rates in ESG1, as well as the transition out of school, but rather is pushed out due to rates from ESG1 to ESG2, are extremely low, very incomplete service (absence of school building, lack few pupils continue until grade 12. of teachers). This is clearly a supply-side issue that Based on the figures provided in table 2.4, the can be addressed by providing educational continu- dropout rate is very high throughout the system: on ity at the local level (possibly using a multigrade average, across every consecutive grade in any formula if the school-age population in the commu- cycle, from EP1 to ESG1, about 13 percent of stu- nity is small). Supplying a standard number of dents drop out or do not make it to the next grade available grades, however, may not produce a 100 (13 percent in EP1, 15 percent in EP2, and another percent retention rate since some pupils may decide 13 percent in ESG1); in ESG2, the pattern is even to drop out anyway. more pronounced. Such a pattern is surprising The second case occurs when the school offers the because a high dropout rate in one cycle normally next grade, j + 1, but the children end their studies ensures that the remaining students have the rele- anyway. Two factors may account for this behavior: vant characteristics (cognitive and social skills, for example) to succeed in further studies. (a) School characteristics do not fit the expectations of the children or their parents. The school Demand- and supply-side perspectives on the may be too far from home, or the "quality" development of education (language of instruction, curriculum con- tent, dedication or behavior of teachers) The reasons for the low internal efficiency (drop-out may not be perceived as relevant enough to and repition) relate mostly to the dropout cycle, and the needs of the locality. The school calen- are compounded by the high repetition rate (the dar may not match that of agricultural activities in the community. In circum- Table 2.4 stances where demand for the type of Survival and transition rates, 1998 schooling offered is weak, adjustments in (percent) the supply could be made to encourage children to remain in school. Indicator, by grade level Rate (b) Schooling is not affordable. Parents may be Survival rate between grade 1 and grade 5 37.3 unable to afford the direct costs of education (the payment of fees, the purchase of books Transition, EP1­EP2 59.6 or uniforms required by the school) or may Survival rate in EP2 68.8 need their children's labor (for productive Transition, EP2­ESG1 44.8 activities--agricultural or pastoral--or for housekeeping activities--fetching water or Survival rate in ESG1 49.1 fuel wood, taking care of infants). Transition, ESG1­ESG2 49.9 From an education policy perspective, it is cru- Survival rate in ESG2 70.5 cial to determine the role that these two factors play Source: Mozambique school survey data, 1998. in the high dropout rate in primary grades. For ENROLLMENT PATTERN 25 example, it would not be relevant to provide a The data in table 2.5 reveal that out of the 175,659 school with more of the same educational services pupils who entered schools that offered all five currently offered if the parents are not sending their grades in 1998, 69,454 pupils were enrolled in newly children to school now. Table 2.5, established from a established schools, while 106,205 were enrolled in secondary analysis of the individual school data set schools that had been operating for a long time. of the Ministry of Education, provides some inter- Therefore, at least 23 percent of entrants to grade 1 esting pieces of information. had limited chances to progress through grade 5 There were 6,031 EP1 schools in the 1998 school because the school they entered was incomplete, survey; of these, 3,271 did not have students although the figure may be slightly higher based on enrolled in all five grades of the cycle, while only grade 1 enrollment in incomplete schools.. 2,760 (or 46 percent of the total number of EP1 With 25 percent as a conservative estimate, schools) were offering all five grades during that assuming for the moment that students would school year. However, the complete schools remain enrolled until grade 5 if exposed to a com- accounted for 73 percent of total enrollment in EP1 plete school, it follows that out of the 63 (1­0.37) and 62 percent of the new entrants in grade 1, which percent of dropouts between grades 1 and 5, about leaves 27 percent of new students enrolling in a 40 percent would be accounted for by the incom- school that did not offer all the grades during that pleteness of some of the schools in the country. Fol- year. One possibility is that some of the schools that lowing this line of argument, the majority of low- did not offer the five grades in 1998 had been retention problems in Mozambique would not be recently established and perhaps would have resolved simply by extending to five grades those offered the full cycle in due course. EP1 schools that are currently incomplete. (Obvi- ously, this is not to say that it should not be done; Table 2.5 EP1 students in complete and incomplete public schools, 1998 Number of nonrepeated students, by grade Type of Number of public schools schools Grade 1 Grade 2 Grade 3 Grade 4 Grade 5 Incomplete 3,271 Recently established 1,289 69,454 44,322 23,920 8,127 150 12,178 325 15,544 12,418 462 22,935 16,950 12,639 352 18,797 14,954 11,281 8,127 Long established 1,982 106,205 70,635 43,886 18,521 71 6,234 341 16,717 12,186 699 35,891 24,473 18,863 871 47,363 33,976 25,023 18,521 Complete Schools 2,760 285,436 240,348 203,775 161,846 130,789 Total 6,031 461,095 355,305 271,581 188,494 130,789 Source: Mozambique school survey data, 1998. 26 MOZAMBIQUE: COST AND FINANCING OF EDUCATION there is no doubt that all EP1 schools should be complete.) The point is that, by itself, making 1. Public higher education institutes are the Eduardo Mond- lane University, the Pedagogical University (teacher training incomplete schools complete would be not be suffi- institute for secondary school teachers), and the Higher Insti- cient to attain a high retention rate in EP1. tute of International Relations (an institute for diplomacy Analysis leads to the same conclusion: making and political sciences). In addition, the government runs a number of technical or higher education institutions like the existing schools complete will help, but much more Naval Academy and some agricultural training centers. Pri- is needed to improve student retention. Two factors vate higher education institutes are the Catholic University, may account for the problem: either the cost (direct with campuses in Beira, Nampula, and Niassa, and the Higher Institute of Science and Technology and the Poly- as well as opportunity cost) of schooling is too high technic University in Maputo. for some parents (a demand-side issue) or some of 2. Public providers include the Ministry of Education, Min- the characteristics of existing schools are not attrac- istry of Higher Education and Science Development, Min- istry of Labor, Ministry of Agriculture, and Ministry of State tive for parents (a supply-side issue) or both. Administration. It is tempting to conclude that the problem lies 3. The GER is defined as the total number of students mostly on the demand side and that educational enrolled not counting their age over the age-specific popula- policies are not applicable. However, the results at tion group (for example, for EP1, this means pupils ages 6­10); therefore, it is sometimes higher than 100 percent. In hand call for further and more rigorous analysis of this study, the GER is calculated based on school survey data the factors accounting for low retention in various (1993, 1995, 1997, 1999) and census data (1997). School survey cycles of study. Information from the 1997 house- data are collected at the beginning (March) and the end (December) of each school year; this study only uses the data hold survey indicates that progress also can be collected at the end of each school year. made on the supply side; for example, the average 4. ADPP is a teacher training institute run by a nongovern- distance to school is as high as 4.5 kilometers, and mental organization and the only nonpublic teacher training the average time needed to reach an EP1 school is institute recognized by the government. 5. This only includes the technical education and vocational often longer than one hour. The survey data also training run by the Ministry of Education. show that as much as 32 percent of rural villages do 6. A straightforward way to assess efficiency is to determine not have a primary school. This suggests that the ratio of coverage to spending. The figure obtained is 0.76 supply-side instruments (establishing complete (3.1/4.1) years per percentage of GNP in 1995 and 1.35 (4.6/3.4) years per percentage of GNP in 1999, demonstrating schools reasonably close to where children live, pos- substantial efficiency gains over the four-year period. There sibly using multigrade teaching when applicable) is a slight discrepancy in the numbers as spending in some may be well suited to improve on what is probably years in Mozambique in this study are measured against GDP and not GNP. For comparative analysis above they are the most important issue in Mozambican education not signigicant. for the years to come. 3 Financial Aspects of Education T he preceding chapters raised concerns about Mozambique has a relatively high level of external the efficiency and effectiveness of the educa- financing of its social sector in general and of edu- tion system. In this chapter, we quantify cation in particular. To a large extent, external these inefficiencies in financial terms. We examine financing corresponds to capital expenditure the present trend of education finance, distinguish- financing, but the borderline is fuzzy between ing it by level and type of education as well as by recurrent and capital expenditure in the social type of expenditure. Beyond the global financial sector.1 Moreover, external financing tends to be amounts, we propose estimates of unit cost at the more volatile than ordinary national budget spend- different levels of education and analyze the factors ing, in particular for recurrent expenditure. As a affecting their variations, in particular, the level of whole in 1999, external financing represented about teacher salaries and the pupil to teacher ratio. On 30 percent of total spending (on and off budget) on this basis, we estimate the cost of some of the ineffi- education, or 43 percent of government spending ciencies discussed in chapters 1 and 2. for the same year. These figures used to be even higher in previous years; for example, in fiscal years Recent trends in budget allocation by economic 1996 and 1997, off-budget figures exceeded those of purpose and level of education the government budget. The high level of external financing for education The Ministry of Education (MINED) is responsible in Mozambique means that relatively little of the for all educational institutions, except for Eduardo state budget is spent on capital expenditure, which Mondlane University, the Research Institute for receives only about 10 percent of public education International Relations, and the Pedagogical Uni- outlays. Within recurrent expenditure for education versity, whose budgets are presented to the Min- (excluding higher education, for which the break- istry of Planning and Finance. Regarding technical down by type of expenditure has not been made), education and vocational training, besides MINED, 76.5 percent was devoted to wages and salaries in the Ministry of Labor, Ministry of Agriculture, and 1999, while 23.5 percent was for goods and services. Ministry of State Administration also run their own The share of salaries within recurrent expenditures technical education and vocational training pro- for MINED increased slightly from 72.5 percent in grams. This chapter reviews only the expenditure of 1996 to 76.5 percent in 1999--a consequence of both MINED and that of the three above mentioned an increase in the number of teachers and the adop- main higher education institutions. tion of a new teachers' salary scale in 1998. Table 3.1 shows the education budget allocation Assessing the allocation of public expenditure by by economic purpose between 1996 and 1999. level and type of education is key to understanding 27 28 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 3.1 Distribution of public spending on education, by economic purpose, 1996­99 (billions of meticáis) Purpose 1996 1997 1998 1999 Ministry of Education (on and off budget) 1,299,700 1,321,755 1,468,482 1,645,814 Budget 485,770 647,755 875,482 1,005,814 Recurrent 441,089 583,219 798,039 907,711 Wages and salaries 319,317 422,699 591,893 694,942 Goods and services 121,773 160,520 206,146 212,768 Capital 44,680 64,536 77,443 98,104 Construction 35,897 49,369 58,583 71,668 Equipment and repair 8,783 15,167 18,861 26,436 Off budget 814,000 674,000 593,000 640,000 Higher Education (under Ministry of Planning and Finance budget) 109,000 123,000 163,000 214,000 Total 1,408,770 1,434,755 1,631,482 1,859,814 Note: For the years 1996 to 1998, figures refer to actual spending, while those for 1999 are budget figures. Source: Ministry of Education and Ministry of Planning and Finance. the government's true expenditure priorities and its particularly given the fact that spending for admin- ability to sustain policy objectives. By looking at the istration in the central ministry was larger than total recurrent budget of the sector between 1996 spending for running all the country's EP2 and and 1999 and presents its distribution by level of ESG1 schools. There is no doubt that this high level education. Two types of comments can be made of spending needs to be investigated further. from these data (see Annex 5) Annex 5 suggests that teacher training and, to a The first concerns budget distribution across lesser extent, secondary education were relatively levels of education. Aside from spending on admin- favored over the 1996­99 period, as table 3.3 sug- istration, primary schools received about 44 percent gests. However, it is important to remember that of total spending for the sector, or about 52 percent national spending on capital items represents only a if we include spending on administration. These small fraction of effective spending, given the mag- proportions were relatively stable over this four- nitude of external financing.2 year period. Secondary education got less than 10 percent of the total, and higher education received A closer look at recurrent budget data for 1998 about twice that amount. These figures do not address efficiency or equity, but they do raise a ques- The data provided so far are either actual or bud- tion about the balance of spending between the sec- geted (1999) spending. However, it appears that the ondary and higher levels of schooling in the country. criteria used for distributing resources across the The second comment concerns spending on levels and types of education may not be com- administration. In each of the four years, spending pletely consistent with what is seen from the sur- on administration represented about 20 percent of veys conducted, at the school level in particular. recurrent spending for the education sector. This We reconstructed spending, in part the cost of figure is very high by international comparisons, wages, to produce an aggregate figure that is fairly FINANCIAL ASPECTS OF EDUCATION 29 Table 3.2 Distribution of public spending on capital, by level and type of education, 1996­99 (billions of meticáis) 1996 1997 1998 1999 Share of Share of Share of Share of Share of Share of Share of Share of Level of capital recurrent capital recurrent capital recurrent capital recurrent education Amount spending spending Amount spending spending Amount spending spending Amount spending spending Primary 12,675 28.4 72.8 28,044 43.5 69.3 33,653 43.5 69.6 45,555 46.4 72.9 Secondary 14,372 32.2 14,4 22,492 34.9 15.1 26,990 34.9 16.5 32,389 33.0 15.4 Technical 4,866 10.9 7.5 5,000 7.7 7.7 6,000 7.7 9.4 7,200 7.3 6.7 Teacher training 12,767 28.6 5.3 9,000 13.9 7.8 10,800 13.9 4.5 12,960 13.2 5.0 Total 44,680 100.0 100.0 64,536 100.0 100.0 77,443 100.0 100.0 98,104 100.0 100.0 Source: Bank calculation based on Ministry of Education and Miinistry of Planning and Finance data. close to the national global budgetary figure (Mt773 for higher education. The table also contrasts billion instead of MT798 billion), which increases Mozambican figures with data from neighboring our confidence in these estimates.3 Table 3.3 pro- countries and some country groupings. vides the estimated figures for 1998. At the primary level, Mozambique has a level of unit cost that is, on average, similar to that of its Unit cost estimates and analysis neighbors, lower than that in Anglophone African countries, and lower still than that in Francophone The macroperspective countries. A relatively similar pattern holds for sec- ondary education. For the higher level, spending From the aggregate financial estimates, it is possible per pupil is globally on the high side compared to propose figures for the level of spending per with that of neighboring countries, but more or less pupil in the different levels of education. Table 3.4 average for African countries (which is very high if provides the details, using 1998 meticáis and per compared with the unit cost of low-income coun- capita GNP units. tries in either Asia or Latin America). Focusing on unit cost at the school level (not As a whole, there is some scope for efficiency including administration costs), we give unit cost in gains in the Mozambican system of education by tap- EP1 the value 1. Per pupil spending is estimated to ping the benefits of a return to scale in upper primary be 3.1 in EP2, 5.5 in ESG1, 13.1 in ESG2, 9.2 in techni- and secondary education especially in urban areas. cal education, 22.6 in teacher training, and 120 and Nevertheless, with a view to issues like distance to 72, respectively, at Eduardo Mondlane University school, safety of girls, and lack of boarding facilities, and Pedagogical University.4 The unit cost is three education goals need to be carefully weighed against times higher in EP2 than in EP1, and the unit cost in the economy. However, when compared with a ESG2 is more than twice that in ESG1. This obviously number of other African countries, the magnitude of raises questions about the allocation of spending. the returns to scale in Mozambican education Estimates of spending per pupil in per capita appears to be less. This being said, there is neverthe- GNP indicate whether the cost of education in the less scope for increasing efficiency in EP2 and ESG1 country is low or high, especially by international schools in order to avoid school enrollments of fewer standards. Table 3.5 presents estimated unit costs than 200 students. This would obviously entail some for primary level (EP1 and EP2 combined) and sec- changes in school mapping, or the distribution of ondary level (ESG1 and ESG2 combined), as well as students to schools offering grades 6­7 and 8­10. 30 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 3.3 Public recurrent spending on education, by level and type of education, 1998 Personnel (number) Spending (millions of meticáis) Non- Non- Total Goods and Total Level of education Teachers teachers Total Teachers teachers personnel services budget Ministry of Education 44,931 10,318 55,249 530,851 70,385 601,236 172,076 773,312 Administration 2,949 2,949 39,807 39,807 95,313 135,120 Central (MINED) 502 502 11,494 11,494 64,515 76,009 Provincial Directorates of Education 951 951 13,714 13,714 22,010 35,724 District Directores of Education 1,496 1,496 14,599 14,599 8,788 23,387 EP1 34,563 2,831 37,394 322,526 7,150 329,676 13,170 342,846 EP2 4,997 1,438 6,435 81,950 3,284 85,234 11,516 96,750 ESG1 2,077 991 3,068 42,578 2,035 44,613 9,421 54,034 ESG2 378 266 644 13,230 1,067 14,297 3,311 17,608 Technical education 1,158 738 1,896 31,862 1,941 33,803 11,786 45,589 Teacher training 328 456 784 11,906 969 12,875 7,193 20,068 Adult education 710 20 730 5.758 33 5.791 316 6.107 Other 720 629 1,349 21,041 14,099 35,140 20,049 55,189 Higher education (under Ministry of Planning and Finance) -- -- -- -- -- -- -- 163 Eduardo Mondlane University 458 1,665 2,123 -- -- -- -- 145 Pedagogical University -- -- -- -- -- -- -- 17 Total for the sector 926,312 n.a. Not applicable. -- Not available. Source: Ministry of Education, Ministry of Planning and Finance, Eduardo Mondlane University, Bank calculation. The unit costs provided in table 3.4 correspond to system of education have a substantial bearing on the actual level of spending per student, on the one the real cost of providing a complete cycle of school- hand, and per year, on the other. This does not mean, ing in the country, as shown in table 3.6. for example, that the cost for one pupil to complete Taking into account the fact that repetitions and a given cycle of schooling is equal to the product of dropouts lead to an increase in unit cost, the that unit cost per year and the number of years in increase in spending per pupil for completed cycles the cycle. Grade repetition and dropping out may varies between about 23 percent in upper secondary occur in the course of studies. Consequently, the cost and 45 percent in lower primary levels. In all cycles of studies for those completing the cycle is larger of education, the proportion of public resources than what would be strictly necessary without these wasted is relatively large, amounting to about disruptions in the flow of students during the cycle Mt219 billion in 1998 (or about US$15 million) of study. These disruptions in the Mozambican within MINED; the larger share of the wasted FINANCIAL ASPECTS OF EDUCATION 31 Table 3.4 Recurrent unit cost, by level and type of education, 1998 Per-pupil spending (meticáis) Public spending Per capita GNP (billions of Enrollment Without With (without Leel of education meticáis) (public) (Mt) administrationa administration)b EP1 342,846 1,880,949 182,300 231,400 0.06 EP2 96,750 168,777 573,200 696,900 0.19 ESG1 54,034 53,693 1,006,400 1,208,200 0.34 ESG2 17,608 7,352 2,395,000 2,752,100 0.81 Technical education 45,589 27,068 1,684,000 1,879,600 0.57 Teacher training 20,068 4,855 4,133,500 4,550,300 1.40 Eduardo Mondlane University Pedagogical University 17,000 1,564 10,870,000 -- 3.69 -- Not available. a. Figures for spending on administration (inside the Ministry of Education) were based on the total cost of salaries for each level or type of schooling (spending by District Directorates of Education and Provincial Directorates of Education has been distributed over EP1 to ESG2, while spending by the Ministry of Education has been distributed over EP1 to ESG2 as well as technical education and teacher training). b. The value used for the country's per capita GNP in 1998 is Mt2.946 million. Source: Ministry of Education, Ministry of Planning and Finance, authors' calculations. resources is in EP1, with a figure of Mt166 billion (or The microperspective about US$11 million). However, the phenomenon is not limited to primary and secondary education; So far, we have estimated unit cost from a efficiency in resource use is also quite low in higher macroperspective, that is, by relating aggregate education. The university is characterized by both a spending to the number of students enrolled at each high rate of repetition (35 percent the first year, 30 level of schooling. In this way, we arrive at figures percent the second year, and down to 5 percent the that are valid for the average national schooling fifth year) and a relatively low capacity to produce conditions and that derive from the various ele- graduates (fewer than half of the students entering ments of the educational policy followed in the the university finish). These factors have a signifi- country. Because these elements are subject to cant impact on unit cost (the loading factor is 1.84) change, it is important to assess how the level of of a graduate, amounting to almost Mt200 million. unit cost is linked with the schooling conditions and Table 3.5 International comparison of recurrent unit cost as percent of per capita income, by level of schooling, 1998 Level of Anglophone Francophone schooling Mozambique Kenya Zambia Madagascar Malawi Africa Africa Asia Primary 0.07 0.12 0.04 0.06 0.06 0.10 0.15 0.08 Secondary 0.40 0.34 0.37 0.20 1.01 0.66 0.49 0.19 Higher 6.41a 4.0 2.3 2.6 8.4 6.3 5.6 0.9 Spending on higher education as a ratio of primary 91 33 57 43 140 63 32 11 a. Higher education: (Mt145,000 + Mt17,000 billion)/(Mt6,772 + Mt1,564)/Mt2,946 million per capita = 6.41 spending per pupil of per capita GNP. Source: World Bank Sector Studies. 32 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 3.6 Per student spending at the school level, by level of education, with and without adjustment for repetitions and dropouts within the cycle of schooling, 1998 (meticáis) Unit cost per year Unit cost per cycle Level of Without adjustment for With adjustment for Without adjustment for With adjustment for schooling repetitions and dropouts repetitions and dropouts repetitions and dropouts repetitions and dropouts EP1 182,300 350,700 911,500 1,753,600 EP2 573,200 798,800 1,146,400 1,597,600 ESG1 1,006,400 1,585,300 3,019,200 4,755,900 ESG2 2,395,000 3,097,900 4,790,000 6,195,800 Higher education (Eduardo Mondlane University) 21,417,000 39,407,000 107,085,000 197,036,000 Source: Ministry of Education, Ministry of Planning and Finance, authors' calculations. elements of educational policy. Moreover, at any pedagogical materials, and (e) the average spending given moment in time, variations in schooling con- per pupil on administration (at the national and ditions exist across geographic locations (provinces, decentralized levels) (see Annex 6 for model). districts) or, more frequently, across individual Following are the relevant pieces of information schools. This implies some variation in unit cost regarding these components. Table 3.7 provides around the national average. To assess how varia- data on the pupil to teacher ratio and the pupil to tions in educational policies affect the unit cost of nonteacher ratio by level of education, while table schooling and to understand how differences in 3.8 provides data on personnel salaries. schooling conditions across locations and schools In Mozambique, the ratio of pupil to teacher (as help to explain variations in unit cost around the well as the ratio of pupil to nonteaching staff) tends national average, we need to understand why unit to be lower the more advanced is the level of edu- costs vary with regard to schooling conditions. cation (from 54 in EP1 down to 12 in higher educa- Analysis shows that policy interventions can tion for teachers and from 664 to as low as 3 for non- improve cost efficiency. For example, the recurrent teaching staff). This pattern is relatively standard. unit cost (of a level of schooling as a whole, in a par- Within EP1, in particular, the same tendency also ticular province, or in a specific individual school) exists, with classes being larger, on average, in the varies depending on multiple policy-related deci- first grades than in the last grades of the cycle. sions: (a) the average salary of the teachers (which This pattern may contribute to the low retention of depends on the distribution of teacher qualifications pupils in EP1, as what pupils learn during their ini- and on the average salary at each skill level), (b) the tial years of schooling depends, more than at higher pupil to teacher ratio (which is close to class size levels in their schooling career, on the attention they when a given group of students is taught full time receive from their teacher. In contrast, the proportion by a single teacher, but is smaller than class size of nonteaching staff at the school level (notwith- when students have specialized teachers and when standing those in administrative or support posi- the teaching duty of a teacher is less than the time of tions) is relatively high. This is the case in teacher students' instruction, as in secondary schooling), training and even more so in higher education, (c) the average salary of nonteaching staff and the where the support staff outnumber the teachers. student to nonteaching staff ratio at the school level, From an international perspective, Mozam- (d) the average amount of resources per student for bique's pupil to teacher ratio is, on average, on the FINANCIAL ASPECTS OF EDUCATION 33 Table 3.7 Pupil to teacher ratio and pupil to nonteacher ratio, by level of education at the school level, 1998 Pupil to teacher ratio Mozambique in low-income countries Pupil to Pupil to Teacher to Anglophone Francophone Level of schooling teacher ratio nonteacher ratio nonteacher ratio Africa Africa Asia EP1 54.4 664 12.2 -- -- -- EP2 33.8 117 3.5 -- -- -- EP1 and EP2 51.8 -- -- 39 53 38 ESG1 25.9 54.2 2.1 -- -- -- ESG2 19.4 27.6 1.4 -- -- -- ESG1 and ESG2 24.9 -- -- 22 : 31 23 Teacher training 14.8 10.6 0.7 -- -- -- Technical education 23.4 36.6 1.6 -- -- -- Higher education 12.0 3.3 0.3 15 18 15 -- Not available. Source: Ministry of Education, World Bank Sector Studies. Bank calculation. high side in primary education (in EP1, not in EP2), replaced teacher training colleges as the basis for but on the low side in higher education. As a whole, training primary school teachers. Training has taken the distribution of teachers by level of schooling is various forms, from a one-year to a three-year pro- skewed for the benefit of students in higher educa- gram with a sixth- and seventh-grade exam as an tion (and to the detriment of pupils in the lower pri- entry requirement, respectively. Today they only mary cycle). offer the 7 + 3 years program and have produced, on The fact that the pupil to teacher and pupil to average, about 950 teachers per year since the mid- nonteacher ratios vary substantially across levels 1990s; this figure is expected to grow slowly in the and types of education, as well as across schools in future. The ADPP teacher institutes (which are run the different levels of schooling (see chapter 4), by a teacher training nongovernmental organiza- partly explains the large variance observed in unit tion) are the only nonpublic teacher training insti- cost. A second potentially important explanation is tutes recognized by the government. In 1997, the linked to the level of qualification and pay of staff government initiated a new teacher training pro- employed at these different levels of schooling. gram. The program hires tenth-grade graduates and Table 3.8 provides basic data in this regard. provides two years of training. Due to the relatively Table 3.8 shows that the professional qualifica- low output of teacher training institutions, tions of teachers (particularly at the primary level) untrained teachers represent a substantial share of vary as a result of the diversity of training models the teaching force, especially in the lower primary used to confront the shortage of teachers. During the cycle, where about one-third of teachers in 1998 had pre-independence period, most teachers attended not received any kind of preservice training. teacher training colleges, which comprised four Table 3.8 also shows, in the last column, wide years of primary school and four years of preservice variation in the average teacher salary across the training; in 1998, few teachers received this training. different categories. Using a private rate of return to After independence, teacher training institutes education of, say, 10 percent a year would imply 34 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 3.8 Distribution of teachers by level of qualification and level of schooling (in percent), and the average teacher and nonteacher salary, by level and type of education, 1998 Average salary Percent Teacher Technical (thousands Teacher categories EP1 EP2 ESG1 ESG2 training education of meticáis) Categories 1, 11 0.3 1.4 1.6 4.9 -- -- 40,800 Categories 2, 4 (6+3; 7+3) 27.3 4.6 0.2 0.0 -- -- 12,200 Categories 3, 15 43.0 7.7 8.8 18.0 -- -- 8,200 Category 5 (8+2) 0.4 10.1 0.9 0.0 -- -- 16,700 Categories 6, 7, 8, 9, 10, 12 (10+2; 9+3) 0.7 50.4 55.6 8.9 -- -- 21,100 Category 13 (bachereis) 0.0 0.6 5.2 11.1 -- -- 42,300 Category 14 (licenciados) 0.0 0.6 5.6 46.6 -- -- 53,500 "Nonqualified" 28.4 24.5 22.0 10.5 -- -- 7,500 Total 100.0 100.0 100.0 100.0 -- -- Annual average salary Teachers In thousands of meticáis 9,332 16,488 20,456 35,754 36,299 27,515 -- In per capita GNP units 3.2 5.6 6.9 12.1 12.3 9.3 -- Non-teaching staff (thousands of meticáis) 2,526 2,284 2,053 4,011 2,125 2,630 -- n.a. Not applicable. Source: Ministry of Education and Ministry of Planning and Finance. Bank calculation. that licenciados, who teach in upper secondary education, the average annual salary for an EP2 schools, earn about 2.2 times as much as category 3 teacher is 1.77 times that of an EP1 teacher. teachers (recruited on the formula "6 or 7+1). Such Expressed in per capita GNP of the country in 1998, a range is neither economically justified nor sus- the figures for teacher pay range from 3.2 in EP1 to tainable in the medium run (it will be a major 12.1 in upper secondary schooling. impediment to the expansion of coverage for sec- For primary education as a whole (amalgamat- ondary education in the future), The question, ing EP1 and EP2), the average teacher salary is esti- therefore, is the extent to which the benefit in terms mated at 3.5 times the per capita GNP of the coun- of effective student learning matches the difference try; however, if the incidence of low retention and in cost. A similar case can be observed for teachers the flow of recruitment of new teachers is taken into beyond EP1, and the same question should be consideration, primary teacher pay is actually raised for higher levels of schooling. around 5.0 times the per capita GNP of the country. The structure of teacher pay by category, To assess the relevance of the level of teacher together with the distribution of teachers by level of remuneration in the country, we can compare the education, translates into wide variations in the level of teacher salaries (expressed in per capita average level of teacher remuneration (and ulti- GNP units) with the pay of other workers holding mately into variations in spending per pupil) across similar credentials and experrience. For primary the different levels of schooling. In lower primary schooling, the average figure is 3.6 times the per FINANCIAL ASPECTS OF EDUCATION 35 capita GNP in English-speaking Sub-Saharan used to simulate the impact on the unit cost of African countries, 6.3 times that in French-speaking changes in the variables that characterize the logis- Sub-Saharan African countries, and 2.6 times that in tical context of schooling and together account for non-African low-income countries. With these fig- the unit cost of education. ures as references, actual primary teacher pay in The total unit cost per primary pupil is Mt230,550, Mozambique (3.2 or 3.8 in the definitions above) or about US$18, excluding external financing (which, appears relatively close to that in Anglophone for example, covers about 75 percent of the cost of African countries. However, if we use the average textbooks and learning materials).5 pay associated with the credentials of new recruits Accordingly, the unit cost per primary pupil (about 5.0 times the per capita GNP), teacher depends on the weight of each variable. It is clear salaries in primary education are more on the high that the average salary cost of teachers and the than on the low side in Mozambique. This is likely pupil to teacher ratio have more weight than some to be even more the case as we consider secondary of the other parameters. However, some of these education, where teachers receive a relatively high inputs have less influence on student outcomes-- premium compared with primary school teachers. for example, school buildings and class size within Nevertheless, it should be pointed out that GNP per a certain margin. Yet the availability of learning capita in Mozambique is below the Sub-Saharan materials, the relevance of curriculum, the language Africa average and therefore the pay of teachers of instruction, and the pedagogical management appears on the high side. methods do affect student outcomes. For now, we It is now possible to reconstruct the unit cost and simply assess which variables cause the largest provide the following breakdown, reconciling the changes in per pupil spending. Table 3.10 shows the macro and micro perspectives in unit cost analysis. sensitivity of per pupil spending as a result of Table 3.9 provides the results. changes in the equation's input variables. As tables 3.4 and 3.9 show, the macro and micro To simplify the analysis, each variable is calculations of unit costs are consistent, with only increased or reduced by 50 percent,6 while the very small variations. The identity of the unit cost, remaining variables are held constant (unchanged). on which this microanalysis is based, can also be As can be seen from table 3.9, an increase in the Table 3.9 Reconstructing the unit cost of education by level and type of schooling, 1998 Technical Teacher EP1 EP2 ESG1 ESG2 education training Pupil to teacher ratio 54.4 33.8 25.9 19.4 23.4 14.8 Average teacher salary (thousands of meticáis) 9,332 16,488 20,456 35,574 27,515 36,299 Per pupil spending on teacher (meticáis) 171,544 487,811 789,807 1,833,711 1,175,855 2,452,635 Pupil to nonteacher ratio 664 117 54.2 27.6 36.6 10.6 Average nonteacher salary (thousands of meticáis) 2,526 2,284 2,053 4,011 2,630 2,125 Per pupil spending on nonteacher (meticáis) 3,804 19,521 37,878 145,326 71,858 200,472 Per pupil spending on goods and services (meticáis) 7,002 68,232 175,460 450,353 435,422 1,481,566 Per pupil spending on administration (meticáis) 48,200 123,700 201,800 357,100 195,600 416,800 Total per pupil spending (meticáis) 230,550 699,264 1,204,995 2,786,490 1,878,735 4,551,473 Source: Ministry of Education and Ministry of Planning and Finance. Bank calculation. 36 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 3.10 Change in per pupil spending as a result of changes in input variables, 1998 Average salary Spending on Average cost of pedagogical Spending on salary cost nonteaching Pupil to materials administration Recurrent Change of teachers Pupil to staff teaching per pupil per pupil unit cost from Variable in (millions of teacher (millions staff (thousands (thousands (thousands actual sensitivity analysis meticáis) ratio of meticáis) ratio of meticáis) of meticáis) of meticáis) (percent Actual figures 9.3 54.4 2.5 664 7.0 48.2 230.550 n.a. 50 percent increase in average salary cost of teachers 14.0 54.4 2.5 664 7.0 48.2 316.322 + 37.2 50 percent reduction in pupil-teacher ratio 9.3 27.2 2.5 664 7.0 48.2 402.094 + 74.4 50 percent increase in average cost of nonteaching staff 9.3 54.4 3.8 664 7.0 48.2 232.452 + 0.8 50 percent reduction in pupil-teaching staff ratio 9.3 54.4 2.5 332 7.0 48.2 234.355 + 1.7 50 percent increase in spending on pedagogical materials per pupil 9.3 54.4 2.5 664 10.5 48.2 234.052 + 1.5 50 percent reduction in spending on administration per pupil 9.3 54.4 2.5 664 7.0 24.1 206.450 ­ 10.5 n.a. Not applicable. Source: Ministry of Education and Ministry of Planning and Finance. Bank calculation. average annual primary teacher salary from Mt9.3 with that in other Sub-Saharan African countries, million to Mt14.0 million will increase per pupil Mozambique has certain characteristics that will spending 37 percent. Such an increase could be the influence this ratio: (a) the pupil to teacher ratio result of, for example, a decompression of the wide varies widely across schools, districts, and range in salaries or an increase in the qualification provinces, (b) double and triple shifts are the rule of teachers, from the 7 + 1 formula to the 7 + 3 for- rather than the exception in urban areas, and (c) as mula for the recruitment of EP1 teachers. A salary access to schooling expands into more remote areas, level of Mt14.0 million represents about five times this ratio will be difficult to maintain. the per capita GNP of the country and is above the As can be seen from table 3.10, the remaining average for English-speaking countries, but below variables do not affect the spending per pupil sig- the average for French-speaking countries, in Sub- nificantly. However, one of the variables that is Saharan Africa. Although such an increase may or known to have a positive impact on student out- may not produce better student outcomes, it would comes--spending on pedagogical materials per help to attract and retain teachers. pupil--could be doubled with a 50 percent reduc- Similarly, reducing the pupil to teacher ratio by tion in administrative costs per student. Further- half would increase per pupil spending 74 percent, more, because the provision of learning materials is since the number of teachers would have to be dou- so small, it is likely that an increase in this compo- bled. Again, as the salary bill is the largest posting nent would have a more far-reaching impact on stu- in the budget, this is not surprising. However, while dent outcomes than a general increase in salary and the pupil to teacher ratio is adequate compared at a lesser cost. FINANCIAL ASPECTS OF EDUCATION 37 High costs of school buildings and finance 82,200 teacher-years (12,000 × 50 percent × construction 13.7 years), or 4,110 new teachers for the next 20 years (82,200 teacher-years/20 years of work). In Mozambique, there are about 6,500 primary The annualized value of classroom usage is schools, half of which are built of traditional mate- about Mt16.6 million and Mt11.5 million, depend- rials. Since the peace agreement in 1992, the gov- ing on the design, using a 20-year lifespan and an ernment has placed a priority on rebuilding the opportunity cost for capital of 5 percent. The annu- school infrastructure and improving the standards alized value represents, respectively, 133 percent of existing schools. By almost doubling the number and 92 percent of the recurrent unit cost of primary of primary schools since 1992, the government has education, here again with huge differences been successful in this endeavor. However, there is between Madagascar (where the annualized cost of a dire need to improve the school environment building represents 29 percent of the recurrent unit beyond the physical infrastructure with such items cost) and Senegal (where the cost of building repre- as training, materials, and teacher incentives. There- sents only 16 percent of the unit cost). In the Africa fore, the costs of construction relative to the amount region, the annualized value of classroom building of resources spent on these softer components--and is between 15­25 percent of the recurrent cost; in in comparison with the level of teacher salaries-- Asia, this statistic is generally even lower. are important measurements. Although the border- Although there is a minimum construction cost line between capital and recurrent spending is for a durable, good-quality school, it is clear from imprecise, since 1992 a significant proportion of Annex 7 that, in comparison with the level of external financing has been targeted at construc- teacher salaries and with other countries like Mada- tion. In some years, as much as US$20 million has gascar and Senegal, the cost of construction in been spent on construction, equivalent to about Mozambique is too high. The high cost of construc- one-third of all education spending. This has a sig- tion has two implications: (a) it limits the total nificant impact on both the total resources available number of classrooms that can be constructed or for components such as textbooks and on the costs rehabilitated, and (b) it limits the amount of avail- of construction, which determines how many class- able resources for softer components like learning rooms can be built. materials, recruitment, and teacher pay. The high In Mozambique, the cost of constructing one pri- investment in physical infrastructure gives little or mary classroom is equivalent to 20 years, or 13.7 no return, yet there remains a need both to provide years, of the annual salary of a primary school students with facilities that allow a complete cycle teacher, depending on the design used. In other of study in a number of localities and to improve words, the costs of constructing one classroom existing facilities. could finance a full-time teacher for 20 years or 20 full-time teachers for one year (or 13.7 years, Unit costs and economies of scale depending on the design). Construction costs are very high in Mozambique compared with the costs In this final section, we analyze the efficiency in Madagascar or Senegal (only 3.9 and 2.2 years of opportunities that lie in the economies of scale in the annual salary of a primary teacher, respec- education production at the school level. The most tively). Even if we take Madagascar (3.9 years) and straightforward way to do this is to calculate the the least costly of the two designs in Mozambique level of per pupil spending in each individual (13.7 years), the costs of constructing a school in school and relate it to the size of enrollment. The Mozambique are more than three times what they different schools are somewhat scattered in figure are in neighboring countries (see Annex 7). 3.1, which is not surprising given the pattern of In light of the current target under the Education teacher deployment in the country. However, an Sector Strategic Program of building 12,000 new average pattern in the relation between unit cost classrooms over the next five years, reducing the and size of enrollment is also apparent. This pattern costs of construction by half, for example, could (analyzed with the help of a regression of the unit 38 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Figure 3.1 Figure 3.2 Unit recurrent costs as a function of Unit recurrent costs as a function of enrollment size in public EP2, 1998 enrollment size in public ESG1, 1998 Per pupil spending (thousands of meticáis) Per pupil spending (thousands of meticáis) 520 750 500 700 480 650 460 440 600 420 550 400 0 200 400 600 800 1000 1200 1400 1600 500 Number of students 0 200 400 600 800 1000 1200 1400 1600 Number of students Source: Mozambique school census, 1998, and microunit cost function. Source: Mozambique school census, 1998, and microunit cost function. cost against the inverse of the size of enrollment at the school level) implies that per pupil spending dents. Here, again, the shape of the curve suggests tends to decline as the size of enrollment of the that when enrollment is below 250 students, unit school increases. If average per pupil spending costs rise significantly (18 of the 78 ESG1 schools of declines, marginal costs are lower than average the country are in this category). Ten ESG1 schools costs. This pattern, which is generally very strong have even fewer than 180 students, implying a when the size of enrollment is low, tends to fade as teacher cost per pupil larger than Mt600,000. enrollment increases. There is a point beyond which As a whole, there is some scope for efficiency the decline in unit cost becomes minimal; this level gains in the Mozambican system of education by of enrollment is said to be economical. tapping the benefits of a return to scale in upper pri- In EP1, we do not observe a decline in per pupil mary and secondary education especially in urban spending as the number of pupils enrolled increases. areas. Nevertheless, with a view to issues like dis- Returns to scale, however, are apparent in EP2, and tance to school, safety for girls, and lack of boarding even more in secondary education, as can be seen facilities, education goals need to be carefully from figures 3.1 and 3.2, which are constructed from weighed against the savings from the economy of individual school data for EP2 and ESG1. scale: if larger schools imply a reduced admission in In EP2 (figure 3.1), the average unit teacher cost particularly rural areas it would not be appropriate. plateaus around Mt420,000 to Mt430,000 when the In general, the cost of providing education would size of enrollment in the school is larger than about be expected to rise with the inclusion of the most 500 pupils. However, in schools with enrollments remote and marginalized population. However, below 250 pupils (137 schools out of the 363 EP2 when compared with a number of other African schools of the country, about 38 percent, have fewer countries, the magnitude of the returns to scale in than 250 students), the increase in unit cost becomes Mozambican education appears to be less. This significant. Schools enrolling fewer than 100 pupils being said, there is nevertheless scope for increasing have unit costs that are, on average, larger than efficiency in EP2 and ESG1 schools in order to avoid Mt500,000; of the 363 EP2 schools in the country, 31 school enrollments of fewer than 200 students. This can be so characterized. would obviously entail some changes in school A similar, although slightly more pronounced, mapping, or the distribution of students to schools pattern is evident for ESG1 schools (figure 3.2): the offering grades 6­7 and 8­10. plateau in per pupil spending (around Mt530,000) is reached in schools enrolling more than 600 stu- FINANCIAL ASPECTS OF EDUCATION 39 new recruits, and (b) the salary scales prevailing in the coun- 1. For example, the textbooks used by students over a three- try for the different categories of teachers and nonteaching year period may be considered a capital expenditure because staff. The distribution of spending by level of education dif- the good acquired is not consumed over a single calendar fers somehow from that shown in the previous tables, but period. However, it may be necessary to distribute one-third we are reasonably confident that these estimates are more of the stock of textbooks every year, making the spending on accurate. this item regular on a yearly basis; in other words, it would 4. Because the range of variation in unit cost is very wide be quite similar to a recurrent expenditure. The same princi- across the different levels and types of education, we focus ple applies to a variety of the goods and services paid for by on unit cost at the school level (not including spending on external assistance. administration) and give the value 1 to unit cost in EP1. 2. This study examines data on capital spending only for the 5. The following equation reflects the current unit cost of a Ministry of Education. primary pupil in EP1: 3. For various reasons, caution is advisable when using RUC = AVT/PTR + AVNT/PNTR + PEDPP + ADMU. Thus financial data, whether budgeted or actual spending figures. RUC = Mt9,332,000/54.4 + Mt2,526,000/664 + Mt7,002 Therefore, we reconstruct spending, in particular the cost of +Mt48,200 = Mt230,550. wages, by estimating (a) the number of personnel employed 6. This does not impact on the variability of any policy inter- in the different levels of education and services, adjusted for vention, but its cost-effectiveness impact. 4 Internal Efficiency and Management Issues I n chapters 1 to 3 we explored some of the prob- complete the first cycle of primary education (EP1) lems and inefficiencies of the Mozambican edu- and reach grade 5. Table 4.1 summarizes the index of cation system and the cost implications. These internal efficiency in education by different grades problems are related to the internal efficiency of the and levels of education. The gross enrollment rate in system, the variance across schools, and the rela- EP1 was 92 percent in 1998, but 24 percent of the tionship between inputs into the education system pupils enrolled were repeaters. The proportion of and student achievement. This chapter investigates repeaters is highest (25 percent) in grade 1, since this these areas further. is where the bulk of students are enrolled. The term efficiency describes the relationship What is surprising is the fact that this share of between inputs and outputs, and this relationship repeaters remains almost unchanged for grades 2 to can be analyzed from several perspectives. From 5 and onward (about 13 percent in each grade), in the public sector's perspective, internal efficiency is spite of the normal assumption that, as the cohort particularly concerned with the relationship progresses, more of the better qualified students between inputs and outputs, within the education remain (yet the assumption does hold true for girls; system or within individual institutions, and how see chapter 5). The repetition may not be related public resources are used and how they influence directly to the performance of the students; instead, student outcomes. The chapter begins with a review it may be related to structure and culture, whereby of the efficiency in student flow, which was dis- teachers are expected to ask a number of students to cussed briefly in chapter 2. Following is an exami- repeat or pupils repeat for some other reason. This nation of how well public resources are distributed, argument, of course, is based largely on the anec- in particular the allocation of teachers and their dotal information that some parents prefer their salaries. This is important because a substantial por- children to repeat because they believe that repeti- tion of public resources is spent on teacher salaries, tion is necessary to learning. In the last part of the and an even distribution of teacher resources con- chapter, we investigate this point further. tributes to student outcomes. The chapter ends with Due to the high dropout rate in EP1 and low a discussion of the quality of education and factors transition from EP1 to EP2 (grades 5 to 6), the gross affecting that quality. enrollment rate in EP2 is only 24 percent. The aver- age proportion of repeaters in EP2 is also high (25 Efficiency in student flow percent). The proportion of repeaters is 24 percent in ESG1, but it is even higher (35 percent) in grade The education system in Mozambique is not very 10. The increase in the repetition rate at grade 10 efficient because the majority of students do not may be explained by the fact that (a) a large number 40 INTERNAL EFFICIENCY AND MANAGEMENT ISSUES 41 Table 4.1 seek formal employment and not return to tradi- Index of internal efficiency, by cycle of study, tional activities, such as farming. 1998 (percent unless otherwise indicated) These reasons for the low survival rates hold true for every grade throughout the education system in Rate EP1 EP2 ESG1 ESG2 Mozambique, particularly the crowding out of stu- dents being promoted. Obviously, if a grade level is Gross enrollment rate 92.3 24.4 6.6 1.2 populated by a large number of repeaters, these Net access to first grade 93.7 20.8 6.4 1.6 repeaters occupy the places that would otherwise Net access to last grade 39.4 14.3 3.1 1.1 accommodate new students; many pupils therefore are prevented from moving up to the next level or Proportion of repeaters from entering school at all. This situation implies Average 23.7 24.7 24.2 10.1 some waste of public resources. Using a cross-sectional perspective, the survival By grade leve rate in EP1 is estimated at 42 percent in 1994, 40 per- 1 / 6 / 8 / 11 25.4 24.3 22.3 7.4 cent in 1995, 37 percent in 1998, and 38 percent in 2 / 7 / 9 / 12 24.5 25.2 19.4 13.2 1999, indicating deterioration in retention in EP1.1 However, using the quasi-time-series perspective, 3 / / 10 / 24.4 34.6 which we believe provides better estimates, the sur- 4 / 20.2 vival rate is estimated at 43 percent between 1994 and 1995 and 44 percent between 1998 and 1999, 5 / 18.8 indicating an improvement of 1.4 percentage points Index of overall efficiency between the periods.2 Finally, by taking the full Percent 58.3 71.8 63.5 77.3 time-series perspective (which may provide esti- mates that are subject to bias), we get a retention rate Number of years necessary of 47 percent over the 1994­99 period (see Table 4.2). to complete 8.6 2.8 4.7 2.6 Considering these figures, as well as the confi- Index of efficiency with dropouts dence we place in the different methods, the 1998 Perdent 73.1 90.8 81.1 84.6 survival rate in EP1 was around 42 percent, and some improvement took place over the 1994­99 Number of years necessary period. Nevertheless, the figure for 1998 (42 per- to complete 6.8 2.2 3.7 2.4 cent) is very low by any standard, and the improve- Index of efficiency with repeaters ment over the five-year period was very modest. Percent 79.7 79.0 78.2 91.4 This indicates that out of 100 pupils entering grade 1, approximately 42 will attain grade 5. Number of years necessary To produce these 42 students in grade 5, it would to complete 6.3 2.5 3.8 2.2 be necessary to finance 210 (42 × 5) years of school- Source: Mozambique school survey data. ing in the absence of repetition and dropouts. How- ever, given the high number of repeaters, the system has a larger number of pupils in the early grades and of students fail the grade-10 examination, cannot be progressively fewer in the higher grades. Therefore, promoted to ESG2, and therefore repeat, (b) a in order to produce the 42 pupils in grade 5, the number of the students who pass the examination system has to finance 125 students (the 100 students cannot enter ESG2 because the existing spaces are in the grade loaded by the 25.4 percent of repeaters taken up by repeaters, and (c) students who pass in that grade), instead of the 42 that would have the examination, but are unable to find a job, decide been strictly necessary. The same applies to the next to remain in the same grade. The final point relates grades. The outcome is that the system is financing to the notion that students who have gone through many more years than the 210 necessary to produce some sort of formal education are most likely to 42 grade-5 students in five years. 42 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 4.2 Survival rate in EP1, 1994­99 (percent) Method of estimation 1994 1995 1994­95 1998 1999 1998­99 1994­99 Cross-sectional estimate 41.5 39.5 37.3 37.6 Quasi-time-series estimate 47.0 42.6 44.0 Full time-series perspective 47.0 Source: Mozambique school survey data, 1994­99. Dividing by 210 the number of school-years the teacher resources. One is qualitative and concerns system effectively finances to produce these 42 the characteristics of the teachers (their qualifica- grade-5 students, we get the efficiency ratio, in tions, years of experience, and so forth). The second which the value of 1 describes a system without is quantitative, with two complementary dimen- dropout or repetition. As seen in table 4.1, the index sions: (a) Are the teachers who are currently avail- of overall efficiency in EP1 is only 58 percent, indi- able deployed in an efficient or--better--in a consis- cating, by complement to 1, that about 42 percent (1 tent way? (b) Are there enough or too many teachers - 0.583) of the public resources mobilized in EP1 are in the country given the need to provide a reason- wasted. We can also use the efficiency ratio to get a able quality of education to the country's children? sense of the average number of years the budget To gauge whether the current pattern of teacher must finance to help one child, on average, reach allocation is efficient and consistent, we need a gen- grade 5. It takes 8.6 years for all the pupils (includ- eral criterion for assessment. As far as the quantita- ing repeaters and dropouts) to pass through EP1,3 tive distribution of teachers is concerned, a clear and it costs the government an additional 3.6 years yardstick is balance: providing schools of similar of spending per pupil to allow a child to reach the size of enrollment with a similar number of teach- fifth grade. ers. The task is therefore to document the extent to Similarly, the index of efficiency at the higher which the current allocation of teachers in Mozam- levels of education is low: 72 percent in EP2, 64 per- bique is consistent with this principle. cent in ESG1, and 77 percent in ESG2, indicating that it takes students a number of additional years Lower primary education (EP1) to pass through each level of education and the standard curriculum. Table 4.1 also presents the Overall in Mozambique, the allocation of teacher index of efficiency without repeaters and dropouts, resources is inconsistent across the roughly 6,000 and these indexes become slightly better than that lower primary schools that existed in 1998. Figure of overall efficiency (which includes repeaters and 4.1 shows that, even though there is a general ten- dropouts), since fewer years are required for all dency for a school to benefit from a larger number pupils to pass through each level of education. of teachers when the size of enrollment is larger (R2 = 85.5 percent), there is wide variability in the Allocation of teachers to individual schools number of teachers for schools enrolling a similar number of pupils. Efficiency also concerns the way resources are To what extent does this inconsistency stem from distributed and used. The distribution of teachers is differences across provinces or from differences one of the most critical resources in relation to cost, within each (or some) of the different provinces? equity, and student outcomes. We now explore how Similarly, does the urban or rural location of the teachers are deployed in Mozambique, that is, how schools affect teacher allocation? We first investi- they are allocated to both provinces and individual gate whether there is any disparity between urban schools. Two factors are related to the allocation of and rural areas. INTERNAL EFFICIENCY AND MANAGEMENT ISSUES 43 Figure 4.1 Figure 4.2a Allocation of teachers in public EP1 Schools, Allocation of teachers in public EP1 schools 1998 in rural areas, 1998 Number of teachers Number of teachers B A 25 A 40 A A A A A A A A A A A A A AA A A A A A A A A B A ACABB 20 AAA AA A A A 30 AAA A A C AC A AA AA A A AA AA A A A A BAA B A AA A A AAA AA A B A A A A AA B BA C AB A A A A BAAA AA B AAA A B A A A ABA C B AA A A A A AA AA AAA B A A AA A A A A A A A ACABA ABAAA ABA A B A A 15 AA ABABA A A A AAAAAB AA A 20 A ABA A BBC C D AA ABA AA A A A A AA A A BBCBABAABBA A A A AAA AAAB BBBAACB GBBAAA BCD A A B A AA A A A AAAABADCAB AA AA A AA AA AB ABBBFD DCD DBEBAABBA B A AAAA A A A AAA A A A A AAAB ADABBCABAA AA AAA A A A A A A C BAAAFCJFCDBBCCBAB A A A DA B AA A A CABACCDBA C CBB AC BB A BA A A A A A AAA B EDCEDGIEHDFAECE CE ACC BA BA C B A 10 A CBBABF BCDBCBAAABBCAABBC BBAA A AA AA 10 A ABADHELLKOEFFFFJCGEBEEFBDCDCBDCABA BAAAA A AA A ABCCHIGHEACBCEAFDABCBAC B BBA AA B A A AA A ADFRTSZWZVXQSKOGKODJMIKHBFDDCBBAA AA A A BDEFELHKLIJEFBBEBCGDFABECCBAAAA AA A CBFFIPZZZZZZZZZZZZZZWZNTGCBDC A A A AA ACDJLKSHPHKHIEIEHIBFEEEG AAAAAA CHTZZZZZZZZZZZZZZZZVJHFBB CA A A AGCJMXZZTPXKUMVMNSGPDBACB A A GZZZZZZZZZZZZLCGACABAAA 5 CBCDHIYZZZZZZZZYQSPQIJGDBAAAA A 0 BDPZZZZZZZZZZZZZYRHGFAA C BHRZZZZZZZZZZZRQHEDDBA AA A AGZZZZZZZZZZZLCFACABA A 0 200 400 600 800 1,000 1,200 1,400 1,600 1 FZZZZZZZNNCBA A A Number of students 0 200 400 600 800 1,000 1,200 1,400 1,600 Legend: A = 1 obs, B = 2 obs., etc. Number of students L d A 1 b i B 2 b i Source: Mozambique school survey data, 1998. Legend: A = 1 obs., B = 2 obs., etc. Source: Mozambique school survey data. As can be seen from figure 4.2a, even though the positive relationship between the number of pupils Figure 4.2b and teachers in rural areas holds, the magnitude of Allocation of teachers in public EP1 schools the disparities is wider than for the country as a in urban areas, 1998 whole. In the rural areas, 77 percent of the dispari- ties of teacher allocation can be explained by the Number of teachers rural location of the schools, versus 85 percent for A A A A AA A A A B ACABB the whole sample.4 For example, among all the 30 AAA A A C B A AA A A AAA A A AA A A A AA BA B AB A A A A AA C B A A schools providing 10 teachers, we find that enroll- A A A AAABA AB AA ABA A A 20 A AA A A B C C AA A A AA A AA A ABA DAA ABA A A AA ACB BBB CBBA A A A A AA A ments vary between 100 and 1,200 pupils; similarly, A A A B CAA A A B C B AA AB BACABEB B AA A A A 10 B BB BBBA C AB A A A AA A DCBCCBCA A AA AA schools enrolling 600 pupils may have between 5 BB ABEEBAECBA B A A A A BFBBD BACB AB A A DDDCBA AA 0 and 15 teachers. Figure 4.2b depicts a similar case 0 200 400 600 800 1,000 1,200 1,400 1,600 for urban areas, where 78 percent of the variance Number of students can be explained by the school location.5 In a sense, L d A 1 b ti B 2 b ti t Notes: Legend: A = 1 obs, B = 2 obs, etc. this is surprising because it is often thought that Source: Mozambique school survey data. organizing the deployment of teachers is easier in urban than in rural settings. If the inconsistency in teacher allocation cannot that some provinces with a similar number of be explained by differences between urban and teachers are servicing quite different numbers of rural areas, perhaps it is the result of how teacher pupils. For example, Cabo Delgado and Inhambane resources are managed. If this is a management both had about 2,500 teachers in 1998, but the issue, it is important to determine the extent to enrollment in Cabo Delgado was about 134,000 which the variance (or inconsistency) is similar pupils, or 41,000 fewer than the 175,000 enrolled in across provinces, which reflects how the budget is Inhambane. Similarly, both Tete and Maputo distributed across provinces. Province enrolled about 140,000 pupils each, but Statistical modeling can be used to assess the Tete had almost 3,000 teachers, while Maputo extent of the variance between provinces or within Province had fewer than 2,000. Maputo City and a province.6 Annex 8 shows the average variations Tete both had relatively generous endowments of between provinces in the allocation of teachers at teachers given the number of pupils enrolled. These the EP1 level, and Annex 9 shows the interprovince figures suggest that there is indeed a significant variations. level of inconsistency in the allocation of teachers When we first consider the total number of between the provinces and that some provinces pupils and teachers at the provincial level, it is clear benefit more than others. 44 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Allowing for variations between the provinces ever, a substantial degree of inconsistency in reduces the residual variability of the model (the R2 teacher allocation to individual schools also exists goes from 85.5 to 87.2 percent). Of the 14.5 percent- within the different provinces. If we use the value age points of total inconsistency (or variance in the [1 ­ R2] as a measure of inconsistency in teacher distribution of teachers), 12.8 percentage points cor- deployment, Maputo Province, Nampula, Sofala, respond to the inconsistency of distribution of and Tete show reasonably good performance on resources within the provinces. In relative terms, this account: R2 exceeds 90 percent (less than 10 per- this means that 87 percent of total inconsistency is cent of inconsistency). In contrast, Gaza, Niassa, found within the provinces in their allocation of Zambezia, and Maputo City have a much higher teachers to individual schools and that only 13 per- degree of randomness in teacher deployment, cent stems from variations between provinces. demonstrating relatively poor management of the These results do not imply that the allocation of public resources with which they are endowed. To resources between the provinces is adequate and illustrate the existence of intraprovince disparities efficient or that all provinces have an equal degree in the allocation of teachers, figure 4.3 shows the of inconsistency in allocating teachers to individual case of Cabo Delgado, in which the randomness is schools. According to this method, Gaza, Inham- moderate (R2 = 0.89); Zambezia or, particularly, bane, and Maputo Province are underendowed in Maputo City would show a more scattered picture. teachers given the number of pupils enrolled, while There is ample room for improvement in the allo- Tete, Niassa, Nampula, and Maputo City are rela- cation of teachers in EP1 (and probably generally in tively overendowed. A second and preferred analy- the allocation of resources) to the different provinces sis (as it provides a better description of the reality) and to the different EP1 schools within most of the estimates the relationship between teachers and provinces. Adopting criteria for both dimensions pupils at the school level separately for each of the and finding ways to implement them progressively 11 provinces. over time are clearly the direction in which to pro- Looking at simulations of the average number of ceed. Across provinces, the government should seek teachers in schools of 300 pupils, which is close to to define criteria so that the budget allocation to the average size of EP1 schools in the country (see provinces is in line with the needs of each province Annex 9). Interprovince disparities are quite sub- in an equitable manner. Within provinces, a similar stantial: Gaza, Inhambane, and Maputo Province mechanism should be established so that the distri- are clearly underendowed in relative terms, with bution of personnel and resources to different schools of 300 pupils having, on average, around schools is consistent and the conditions of schooling four teachers. In contrast, the provinces of Niassa are reasonably equitable across locations. To do that and Tete, and especially Maputo City, have, on average, more than six teachers for the same size of enrollment. This means that it would be necessary Figure 4.3 to increase the number of teachers in the three Allocation of teachers in public EP1 schools underendowed provinces by more than 50 percent in Cabo Delgado, 1998 to reach equality. The case of Maputo City is spe- Number of teachers cial: a school in Maputo City enrolling 300 pupils A A A has 9.7 teachers, which is 2.6 times more than an 20 A A A A average school of the same size in Gaza Province. 15 A A A AA A A Maputo City is a special case due to the concentra- A A A BAA A A A 10 A A A A A A A A A tion of teacher training colleges and activities C A A AA A AA BCABB A AAAA A AABAEEEH CEACA B A 5 A CCBDMJFFDD CA during the war. BDEHHCFQFDCB A ABIQZZINGDGAAA IZZQZLECD C 1 BJFKCAA The global allocation of teachers varies widely 0 200 400 600 800 1,000 1,200 1,400 1,600 across the different provinces, and this variance Number of students should be addressed (perhaps by improving the Legend: A = 1 obs., B = 2 obs., etc. procedure by which the budget is prepared). How- Source: Mozambique school survey data. INTERNAL EFFICIENCY AND MANAGEMENT ISSUES 45 would require a plan implemented over time, since are relatively under- or overendowed in teachers in it might be difficult to redeploy the teaching force at EP1 also under- or overendowed in the other levels the provincial and national level overnight. The of schooling? and (b) to what extent are provinces attrition and recruitment of teachers can be used to that deploy teachers efficiently in EP1 able to per- adjust the allocation of teachers, but creating incen- form better than other provinces in EP2? tives for teachers to move to other schools may also As in EP1, provinces do not receive resources be necessary. This policy may be pertinent to the that match the number of students they enroll in a idea of merging EP1 and EP2 schools. totally consistent and equitable way. For example, schools with about 300 students enrolled (the aver- Upper primary (EP2) and lower (ESG1) and age size of enrollment at the country level in EP2 upper (ESG2) secondary education schools) have, on average, 8.6 teachers, but the figure varies from below 7 in Gaza, Inhambane, and The allocation of teachers in upper primary and sec- Sofala to around or over 11 in Niassa, Tete, and ondary education shows characteristics similar to Maputo City. This is a wide range for provincial dif- those analyzed for the lower primary schools. For ferences; it would be necessary, on average, to example, for upper primary schools, which have increase the number of EP2 teachers in Sofala by a about 15 teachers, enrollment ranges between 200 factor larger than two to get to an endowment sim- and 1,200 students, indicating extensive variation in ilar to that of Niassa. The degree of consistency in the ratio of teachers to students from school to teacher allocation across EP2 schools also varies school. The same pattern applies to schools at the across provinces: a relatively high level of random- lower secondary level, as figure 4.4 shows. Some ness exists in Cabo Delgado, Gaza, Maputo lower secondary schools have between 400 and 1,000 Province, and Zambezia (R2 below 0.80), while students enrolled, but these schools, which differ so Manica, Sofala, and Tete perform better in the man- widely in terms of enrollment, have the same agement of teacher deployment (R2 around 0.90). number of teachers on average, namely, about 20. Contrasting the endowment of EP1 and EP2 In addition to the wide disparities in teacher allo- teachers by province (estimated as the average cation at all levels of education, does some similar- number of teachers in schools of average size) ity exist in the pattern of teacher allocation across shows that the provinces that are overendowed in the different levels of schooling? This question obvi- EP1 tend to be at an advantage in EP2, as figure 4.5 ously pertains only to the pattern across provinces. shows. Only Maputo is relatively underendowed in In other words, (a) to what extent are provinces that EP1 but overendowed in EP2. Provinces such as Figure 4.4 Figure 4.5 Allocation of teachers in public ESG1 Endowment of teachers in EP1 and EP2 schools, 1998 schools, by province, 1998 Number of teachers 14 A A A school 60 12 A EP2 50 A 10 A 40 A A A A A 30 A 8 A A 300-student AA A A A in 20 A AA BA AB A A BA A A 6 AAAAA B AA B 10 AAAA A A A A A A A A A Teachers 4 0 3 4 5 6 7 8 9 10 0 500 1,000 1,500 2,000 2,500 3,000 Number of students Teachers in 300-student EP1 school Legend: A = 1 obs., B = 2 obs., etc. Source: Mozambique school survey data. Source: Ministry of Education, school survey data, 1998. 46 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Gaza, Inhambane, and Sofala suffer in teacher Table 4.3 endowment in both EP1 and EP2, while Niassa, Regression analysis of student outcomes, Tete, and Maputo City benefit from a relatively gen- 1999 erous allocation of teachers in both cycles. Indicator Result What are some of the reasons for this pattern? To what extent is it the result of a historical develop- Student female/male ­ 2.75* ment or of budgetary procedures? Regardless of the Student age -- reasons, there clearly is substantial room for Urban/rural ­ 4.59* improvement and policy change. Suburban/rural ­ 3.91** Some elements regarding the quality Share of teachers with more than of primary education 10 years of education + 0.029* Share of teachers with no formal training ­ 0.039* Inputs, like the endowment of teachers, often relate to the quality of education. Many factors might Maputo City Reference affect the quality of education: teachers who are not Cabo Delgado ­ 9.20* trained (unqualified), classes that are overcrowded, a dearth of pedagogical material, or a static curricu- Maputo Province ­ 0.85 lum. Obviously, these considerations do have some Zambezia ­ 6.85* bearing. However, a school's quality may also be -- Not available. * Significant at 1 percent. ** Significant at 5 percent. judged by student outcomes--in other words, the Source: Project assessment data, 1999. average level of student learning, schooling careers, or the disparities in learning across students. We can analyze these conditions by (a) relating student available on students (only age and gender are outcomes to school inputs (mobilized at the school known) and the schooling context (only the propor- or class level), (b) costing the inputs, and (c) design- tion of teachers with more than 10 years of formal ing a cost-efficient strategy to maximize outcomes education and the proportion without teacher train- for a given unit cost or, conversely, to minimize ing are known at the school level). spending per pupil for a given level of student out- The results of multivariate regression show that come (identifying which mix of school inputs is the the share of variance of student test scores optimal combination). In the following section, we accounted for by these variables is relatively small investigate these issues further as well as factors (10.7 percent), but some results are significant: girls that may affect these variables. tend to perform less well than boys (-2.75 points in the test scale), and rural students tend to perform Student outcomes and school and student better than urban ones; within cities, students tend characteristics to do better in the suburbs than in the inner city. However, other things being equal, the level of We have not been able to locate data to conduct a learning achievement is considerably lower in full analysis in the Mozambican context. However, Zambezia (-6.85 points) and lower still in Cabo Del- we have been able to use some of the data collected gado (-9.20 points) than in Maputo, with no signif- by the National Institute of Educational Develop- icant difference between Maputo Province and ment for the Assessment Project, also referred to as Maputo City.7 PASE (see table 4.3). The PASE survey collected data Having more educated and better trained teach- from 90 EP1 schools located in Cabo Delgado, ers has a positive impact on student learning (see Maputo City, Maputo Province, and Zambezia. table 4.3 and 4.4). In the sampled schools, 44 percent More than 4,200 students were tested in Portuguese of teachers had more than 10 years of formal school- and mathematics at the end of grades 2 and 3. ing, with values at the school level ranging from 0 Unfortunately, very few pieces of information are to 100 percent; similarly, 30 percent of teachers had INTERNAL EFFICIENCY AND MANAGEMENT ISSUES 47 Table 4.4 able resources into effective student learning. This Teacher experience for sampled schools, pattern also applies to Mozambique. by province, 1999 (percent) A marginal gain of 9.1 points would account for interschool variations in both student characteris- Share of teachers Share of teachers tics and school inputs, and 18 points would be asso- with more than with no ciated with interschool variations in the capacity of 10 years of schooling formal training schools to produce learning in the students they Cabo Delgado 47.7 40.3 enroll out of the resources they receive.10 Even though the figures provide only orders of Maputo City 80.1 7.8 magnitude, the variation in resource use at the school Maputo Province 44.9 33.9 level probably makes more difference to outcomes than Zambezia 33.5 28.8 the amount of the resources themselves. It follows that the pedagogical management of schools in Source: Project assessment data, 1999. Mozambique is weak and that any strategy aimed at improving student learning and the effective quality of schooling should not focus exclusively no formal training (ranging from 0 to 100 percent at on the input side (however important inputs are). the school level).8 It should also pay attention to improving the Between a school where 20 percent of the teach- pedagogical management of the system at the ers have 10 or more years of education and another school level, which includes relevance of the school where the percentage is 60 (these are not curriculum, teaching methods, and so forth. Thus extreme cases), the variation in student test scores to achieve significant improvements, it is impor- was estimated at 1.2 points for formal teacher edu- tant to (a) measure student learning on a regular cation and 1.6 points for teacher training. These fig- basis at the school level and to (b) design a system ures are sizable, but the magnitude remains rela- by which low-achieving schools are led to do tively modest. better. Toward this end, the role of inspectors and Differences in student outcomes across schools school heads could be clarified, and a firmer struc- can stem from three sources: (a) variations in the ture of responsibility and incentives could be set characteristics of the student body, (b) variations in in place. the amount and distribution of resources made Beyond these logistical factors, we now examine available, and (c) variations in the capacity of indi- the influence of other factors that are important, but vidual schools to transform the resources they have exogenous, to the school, such as the parents' into student learning. Student learning tends to mother tongue, availability of books at home, and differ widely across the different schools of the the like. In addition, we elaborate on the cultural sample (and probably of the country). In fact, the aspect of teaching and learning from a study by share of variance of test scores is as high as 38 per- Mikael Palme that addresses the quality and culture cent.9 For reference, it is useful to compare this in primary teaching on the basis of recent research figure (38 percent) with that found in similar stud- findings on Mozambique.11 ies conducted elsewhere in the Africa region: 23 percent in Togo, 28 percent in Mali, 28 percent in The social and cultural context of learning Senegal, and 31 percent in Burkina Faso. Only Mau- in Mozambique ritania has a figure higher than that of Mozambique (47 percent). In addition to this analysis, the PASE study assessed Mozambique therefore has substantial dispari- the impact of a number of factors on the pupils' ties in learning across different schools. In Burkina achievement in mathematics and Portuguese: those Faso, Mali, Mauritania, Senegal, and Togo, variabil- affecting the student's characteristics and inputs, ity in the use of inputs had a much larger impact such as socioeconomic background, the language of than the capacity of schools to transform the avail- instruction versus the mother tongue spoken at 48 MOZAMBIQUE: COST AND FINANCING OF EDUCATION home, availability of school supplies and classroom schooling and its lack of relevance to the local com- equipment, and repetition of students. munity. In particular, the study found the following. It is not surprising that factors such as exposure First, in the sampled Mozambican schools, teach- to Portuguese at home and the availability of books ing was characterized by strong ritualization and a at home have a positive effect on student outcomes. high level of formal, functional interaction between Hearing Portuguese spoken at home has a positive the students and the teacher. This promoted student correlation with both Portuguese and mathematics passivity. Teachers had poor mastery of their sub- test scores, but the correlation is much more impor- jects and few teaching skills; in fact, teachers tended tant in relation to the language scores than to the to rely on explanations in the textbooks, providing mathematics scores. In addition, the study found a few opportunities for students to ask exploratory positive correlation between the presence of books questions. As a consequence, the teachers did not in the child's home and performance in Portuguese; know whether the students had attained adequate however, this factor did not affect the mathematics skills. The language of instruction further exacer- grade. Another factor was the education of the par- bated this problem, since most pupils, and most ents or other family members: students whose par- teachers as well, did not command a high level of ents or siblings knew how to read and write in Por- proficiency in Portuguese. Since so much of what tuguese scored higher than those whose parents did was supposed to be taught and learned in early not. Moreover, students who had repeated a grade grades was not totally learned, teachers constantly actually scored lower than students who had not demanded skills and knowledge that the majority repeated: test scores for a large sample of students of students did not have. As a result, students indicated that those who had never failed had aver- repeated or dropped out. This finding appears to be age grades of 12.71 and 10.60 in mathematics and somewhat consistent with that of learning out- Portuguese, respectively, as opposed to 12.2 and comes and teachers' experience and qualifications. 9.67 for those who had repeated once and 11.9 and Thus teachers with a better education and longer 9.51 for those who had repeated more than once. experience are assumed to have mastered their sub- This finding is particularly interesting because jects more completely. many parents believe that if their children do not Second, the interviews with 30 Makua families in repeat, they will not learn. Of course, this analysis Nampula revealed a strong view of schooling as does not take into account the characteristics of the belonging to a closed universe to which ordinary students who do not repeat. people have no access. In this context, schooling School inputs, such as school furniture, learning was ascribed a purely functional value, one that materials, and classroom materials (chalk, black- could open up the door to modern society, but was board) do have some bearing. For example, Cabo not relevant to the local community. More impor- Delgado and Zambezia, which have the lowest tant, the school and teacher, although highly gross enrollment rates in the country, have the least respected, were considered to be outside the norms amount of classroom furniture (around 70 percent and values of the community. In this context, the of their students sit on the floor). This tends to con- proximity of the school became quite important firm the high variance between provinces and since continued education beyond EP1, and espe- between schools. cially after grade 6, required arrangements through On the issue of the schools' capacity to manage the extended family network. the inputs, we return to the study conducted by Besides being considered a costly endeavor Mikael Palme.12 The study draws on a national (from lost income and cost associated with sending evaluation of primary education textbooks under- food or money), school also was seen as a contra- taken by a joint national and international team dictory one: those who managed to "succeed" usu- between 1990 and 199213 and on a 1993 study of 30 ally changed, rejecting the traditional community Makua peasant families in Nampula. In this study, (and farming), but many also were excluded from two factors that influenced effective teaching the "modern" society. This had special conse- related to the culture of teaching and the nature of quences for girls, as a man could not marry a girl INTERNAL EFFICIENCY AND MANAGEMENT ISSUES 49 two years (the mean was set to 100, and the standard devia- with more education than himself and girls played tion was set to 15). This score was then regressed against the a key economic role in peasant Mozambique. In age and gender of students, the geographic location of the accordance with the matrilineal society, the labor school (urban, suburban, or rural setting), and the two avail- able characteristics of teachers at the school level (the pro- and products of the men who married into the portion of teachers with more than 10 years of education and family remained the rightful property of the the proportion of teachers with no formal training). mother-in-law. On the one hand, this argues 8. The figures for the proportion of teachers with more than strongly for placing the school in the community 10 years of education and the proportion of untrained teach- ers in the sample are very close to the national average, sug- and expanding grade levels, rather than forcing the gesting that the structure of the sample is not distorted. children to move. On the other hand, if matrilineal Beyond this global structure, the two teacher characteristics are not distributed evenly across the different provinces, as tradition is commonplace in rural Mozambique and documented in Table 4.5. Teachers in Maputo City are both placing schools in the community is necessary, it is better educated (80 percent of teachers have more than 10 still not a sufficient condition to ensure the retention years of formal education) and better trained (only 8 percent have no formal training) than in the central-northern of girls. provinces. Only 33 percent of teachers in the Zambezia sample have more than 10 years of education, while 40 per- cent of teachers in Cabo Delgado have no formal training. 1. To analyze the survival rate in a cross-sectional manner, 9. Using a dummy variable for each school is a simple way the survival rate is calculated as the ratio between the net to categorize each school as unique, while estimating the enrollment rate (calculated as the ratio between the number global impact of the three sources of variation. of nonrepeaters and the number of children of the relevant 10. We used the average level of student learning in grade 3 age in the population) in the last and the first year of the (grade 2) as a proxy for both the characteristics of the stu- cycle. dents enrolled in the school and the resources made available 2. A "compact" or reduced form of a time-series analysis to it, as a complementary variable to account for the variabil- looks at the survival rate between two consecutive grades ity in individual test scores in grade 2 (grade 3). Inclusion of over two consecutive years, and the estimates for each grade this variable increased the share of variance in test score are then assembled to get a figure for the cycle as a whole. accounted for by the model from 10.7 to 19.8 percent. The These estimates can be made at different points in time, marginal gain in R2 is 9.1 points, while the difference allowing one to assess whether things are improving or dete- between 19.8 percent and the 37.8 percent of the model, using riorating. the set of school dummies, is 18 points. 3. 5 years/0.583 = 8.6 years. 11. Mikael Palme, "Cultural Ambiguity and the Primary 4. This is documented by the lower value of the R2 (77 per- School Teacher," in Education, Cultures, and Economies, chap. cent against 85 percent for the whole sample). 14 (Upsala University, 1997). 5. This is documented by an R2 that is very much like that 12. Palme, "Cultural Ambiguity and the Primary School observed in rural areas (78 percent). Teacher." 6. Model 1 indicates that inconsistency in teacher alloca- 13. "Cadernos de pesquisa," National Institute of Educa- tions to individual schools represents 14.5 percent (100-85.5) tional Development, 1997, funded by Swedish International of total variability in the number of teachers at the school Development Agency (SIDA); the research team involved level. Model 2 indicates that, allowing for disparities in researchers from National Institute of Educational Develop- teacher allocations between provinces, the inconsistency (or ment, the Stockholm Institute of Education, and the Center variance) within each province accounts for 12.8 percent for Research on Bilingualism at Stockholm University. The (100-87.2). study comprised 150 lessons in grades 1 to 5 in eight schools in two regions; more elaborate observations of 22 7. We constructed a standardized measure of student learn- lessons in four schools focused on the qualitative aspects of ing, aggregating mathematics and language for each of the language use. 5 Analyzing Equity A s we saw in the previous chapters, many of policy interventions needed to address them, we the issues facing the government relate to analyze the gender-disaggregated gross enrollment the allocation and distribution of resources trends in Mozambique as a whole, by province, to education. Equitable distribution of educational and by geographic area (rural and urban). This resources is important because the distribution of analysis will indicate whether there is gender or income, wealth, opportunities, and power in society geographic inequity and the magnitude of that is a broad measure of equity and a reflection of gov- inequity; it also will document whether the ernment policy choices. Education, in particular, inequity resides more in access, retention, or both. forms the basis from which many individuals Second, we examine the extent to which the public obtain access to these services in the future and is an resources for education are evenly distributed important tool for poverty reduction. In the previ- within a cohort of youngsters. ous chapters, we discussed some of the inequities in resource allocation, school enrollment, student to Enrollment patterns by gender and by teacher ratio, and unit cost; these are supply-driven geographic location inequities. In this chapter, we further analyze the extent to which the increase in the gross enrollment An overall perspective rate between 1993 and 1999 has helped to reduce gender and geographic gaps (or inequities) as well We start by analyzing disparities in enrollment by as disparities in public resource distribution. level of education. On this count, Mozambique has Since 1992 the government has sought to ensure made encouraging progress in the past few years. more equitable distribution of educational As discussed in earlier chapters, the overall gross resources. This chapter explores these equity- enrollment rate grew from 60 percent in 1993 to 92 related issues from the perspective of the benefici- percent in 1998. However, table 5.1 shows fewer aries. In the educational equity literature, there are girls than boys were enrolled in the various grade various ways to assess equity in education: (a) eval- levels from primary to secondary in 1997. Out of uate differences in access to specific levels or types 1,780,881 students enrolled in EP1 in 1997, 59 per- of education, (b) compare the distribution of bene- cent were boys, and 41 percent were girls. In the fits among people with different education and same year, 60 percent of those enrolled in EP2 were socioeconomic backgrounds, and (c) assess who boys, and 40 percent were girls. In ESG1, boys made pays for and who benefits from education. up 58 percent of enrollment (girls, 42 percent), while To identify the extent and nature of existing in ESG2 boys comprised 60 percent (girls, 40 per- inequities in the distribution of education and the cent) of enrollment. Thus in 1997, we find more or 50 ANALYZING EQUITY 51 Table 5.1 Table 5.2 Enrollment, by gender and level of education, Gross enrollment rate in EP1, by province, 1997 1998 Girls as a Province Boys Girls Total a share Level of Number Number Total of total Cabo Delgado 88 58 77 education of boys of girls enrollment enrollment Gaza 130 117 126 EP1 1,041,958 738,923 1,780,881 41.5 Inhambane 119 103 116 EP2 97,724 66,353 164,078 40.4 Manica 93 63 82 ESG1 30,237 21,583 51,821 41.6 Maputo Province 130 121 128 ESG2 5,441 3,701 9,142 40.5 Nampula 82 52 71 Technical educationa Niassa 95 68 86 18,937 8,131 27,067 30.0 Teacher traininga Sofala 87 53 74 2,289 2,566 4,855 52.0 Higher educationa Tete 90 65 82 7,328 2,994 10,322 29.0 Zambezia 103 65 89 a. Figures refer to 1998. Source: Ministry of Education statistics. Maputo City 121 114 118 Total 101 74 92 less the same proportion of boys (60 percent) and Source: Ministry of Education, school survey data, 1998. girls (40 percent) from EP1 to ESG2. When the distribution of education is analyzed from a provincial perspective, an interesting picture emerges, as shown in table 5.2. Whereas the overall jeopardy for girls living in low-enrollment provinces. gross enrollment rate in EP1 was 92 percent in 1998, In addressing these inequities, the government the provincial GER ranged from 71 percent in Nam- might consider both regional and gender disparities pula to around 130 percent in Gaza and Maputo jointly. To understand the possible reasons for these Province. The GER takes into account the repeaters disparities, we look at access, repetition, and reten- as well as under- and overage children, and the tion trends, by province and by gender. magnitude of interprovincial disparities highlights the existence of real inequities in the chances offered to youth in various provinces. Figure 5.1 In all provinces, girls lagged behind boys. On Relative gender gap and GER in different average, the gender gap was 27 percentage points, provinces, 1998 ranging from a low of 7 percentage points in Relative gender gap Maputo City or 9 percentage points in Maputo Province, to 30 percentage points in Nampula, and 80% a high of 38 percentage points in Zambezia. Obvi- 60% ously, the gender gap tends to narrow as the overall 40% gross enrollment rate increases, as seen in figure 5.1.1 Such a finding confirms the assertion that 20% broader educational provisions in the system as a 0% whole are key to achieving higher female school 70 80 90 100 110 120 130 140 participation.2 Global GER at the provincial level This pattern indicates that regional and gender differences tend to compound, thus creating a double Source: Ministry of Education, school survey data, 1998. 52 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Table 5.3 schooling; thus girls' retention lags behind that of Access to grades 1 and 5, by province, 1998 boys. From an analysis of disparities across provinces, the access rate of boys was generally Access to Access to grade 1 grade 5 high, but universal access may not be achieved in (percent) (percent) Sofala and Nampula or even in Maputo City, unless Ratio Ratio Retention, the supply of schools keeps pace with population of boys of boys grade 1 growth. The figures in table 5.3 also show that girls Province to girls to girls to grade 5 progressed less than boys in the province of Sofala Cabo Delgado 1.33 2.33 0.18 and, to a lesser extent, in Cabo Delgado, Manica, Nampula, and Zambezia. In these provinces, the Gaza 1.05 1.23 0.42 demand for girls' schooling is relatively low; special Inhambane 1.06 1.29 0.46 attention ought to be paid to this finding. Manica 1.30 2.00 0.32 Regarding access to grade 5, only 35 percent of the eligible age group gets to the last grade of lower Maputo Province 1.05 1.06 0.42 primary education, meaning that nearly two-thirds Nampula 1.31 2.31 0.26 of entrants in school drop out during the first four years of primary school. To a certain extent, this pat- Niassa 1.22 2.06 0.21 tern holds true in all provinces. The best case in the Sofala 1.56 1.77 0.46 country is Maputo City, where only 20 percent of new entrants failed to reach grade 5. In all other Tete 1.21 2.00 0.25 provinces, less than half of grade-1 entrants reached Zambezia 1.31 2.40 0.27 grade 5. The worst cases are Cabo Delgado, Nam- Maputo City 1.04 1.06 0.79 pula, Niassa, Tete, and Zambezia, where only one- fourth or fewer reached grade 5. In these provinces, Total 1.24 1.56 0.37 both boys and girls suffered, but the situation was Source: Ministry of Education, school survey data, 1998. especially acute for girls. In Cabo Delgado, only 12 percent of girls of the relevant age group enrolled in grade 5 in 1998; a similar figure was observed in Gender and geographic disparities in Nampula (13 percent) and in Zambezia (15 percent). schooling profiles In addition to a between-province comparison, this type of analysis can inform calculations for spe- Enrollment rates can vary between boys and girls as cific groups such as boys and girls, on the one hand, well as across provinces. However, as underscored or for urban and rural settings, on the other. in chapter 2, the enrollment rate may not be suffi- Although the enrollment rates of rural girls are cient to describe the system's effective coverage likely to be lower than those for either girls in gen- given the high level of repetition and the low level eral or urban dwellers as a whole. Annex 10 pro- of within-cycle (or intracycle) retention in Mozam- vides the basic data used.. As a whole, differences bique. Gender and geographic location must be by sex are important, but geographic disparities are considered. Table 5.3 provides data, by province, on even stronger, as can be seen in figures 5.2 and 5.3. access rates in grade 1 as well as on enrollment rates Boys have better access to primary grade 1, as in grade 5 (the end of the lower primary cycle). their enrollment rate at this level is 20 points above In access to both grade 1 and grade 5, gender dis- that of girls. In the course of EP1, the gap between parities are striking. The boy to girl ratio is esti- boys and girls narrows in absolute terms but mated at 1.24 (104:84) in grade 1 and 1.56 (42:27) in widens in relative terms because boys outnumber grade 5. These figures indicate that (a) probably all girls by more than 50 percent at that level. After this boys have access to grade 1, while only 84 percent point in the schooling process, disparities between of girls have access and that (b) gender disparities boys and girls remain on the same order of magni- widen in the course of the lower primary cycle of tude until grade 11. The main source of gender dis- ANALYZING EQUITY 53 Figure 5.2 Figure 5.3 Enrollment rate in EP1 through ESG2, by Enrollment rate in EP1 and EP2, by gender and grade level, 1998 geographic location and grade level, 1998 120.0 Boys 120.0 Girls Urban Total 100.0 Semiurban 100.0 Rura l Total 80.0 80.0 (percent) 60.0 rate 60.0 40.0 40.0 Enrollment 20.0 20.0 0.0 0.0 G1 G2 G3 G4 G5 G6 G7 G8 G9 G10 G11 G12 G1 G2 G3 G4 G5 G6 G7 Grade level Grade level Source: Ministry of Education, Table 5.4. Source: Ministry of Education, Table 5.4. parities lies in retention in the lower primary cycle, terms of access to and retention in secondary followed by access to that cycle. These patterns pro- schooling. It would be important to understand vide some clue as to where interventions are likely what explains the differences in performance in to make a greater difference. order to address them effectively. Geographic disparities are much more alarming. A complementary way to illustrate the blend of However, the pattern is different from that by gender and geographic disparities is to construct which gender disparities occur since, as far as data schooling profiles at the provincial level. For that indicate, there may be no geographic differences purpose, we present the cases of Cabo Delgado and between rural and urban areas in terms of access to Gaza. The situation in Cabo Delgado illustrates the grade 1. magnitude of both weak performance and wide However, an urban, semiurban, and rural com- gender disparities in the distribution of education. parison shows three features of the problem. First, Here, there is a high overenrollment in grade 1 for the semiurban trend is a steep slope that starts in boys, while girls are within the national limits (94 EP1 above the national trend (110 percent against 93 percent). By grade 5, the overall enrollment falls to percent total access to grade 1). By grade 2, the 20 percent, much lower than the national access figure becomes the same for semiurban as for urban rate, with a 12 percent enrollment rate for girls. In (about 92 percent), both being above the national this regard, Cabo Delgado is comparable to other average. By grade 3, however, the semiurban trend provinces such as Nampula (13 percent) and Zam- falls below urban performance to become nearly the bezia (15 percent), but it represents the worst case. same as the national average. Starting in grade 5, Throughout EP1, there is a constant gap between the semiurban trend starts gradually falling behind boys and girls. By grade 6, in EP2, this gap between the national trend, and, by grade 7, in EP2, the the sexes becomes much wider, leading to even less semiurban trend becomes very close to the national female access in EP2 than occurred in EP1. average again. Second, as shown in figure 5.3, The province of Gaza, in contrast, has a grade-1 urban children start at the same level as rural chil- overenrollment in common with many other dren but survive in much higher rates along the provinces but also displays a more equitable per- way. Urban children clearly are privileged in the formance. The GER in this province is above the system. Third, while the semiurban and urban national averages (115 percent). The gender gap trends follow these paths, the rural trend clearly lies remains among the lowest in both EP1 and EP2 (boy at the bottom after grade 1, where it starts at the to girl ratio is 1.05 and 1.23 in EP1 and EP2, respec- same level as both urban and national averages. An tively). In EP2, the gender gap is wider in grade 6 important shift occurs at this transitory point in than in grade 7. Girls survive at higher rates relative 54 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Figure 5.4. Admission rate in EP1 and EP2 in Cabo Delgado, by gender, 1998 Gross admission rates in EP1 in Cabo Delgado Gross admission rates in EP2 in Cabo Delgado Percentage Percentage 25 120 20 100 Male 80 15 Male Female Female Total Total 60 National 10 National 40 5 20 0 0 6 7 1 2 3 4 5 Grade Grade Source: Ministry of Education, school survey data, 1998. Figure 5.5 Admission rate in EP1 and EP2 in Gaza, by gender, 1998 Gross admission rates in EP1 in Gaza Gross admission rates in EP1 in Gaza Percentage Percentage 140 30 120 25 100 20 80 Male Male Female Female 15 Total 60 Total National National 10 40 20 5 0 0 1 2 3 4 5 6 7 Grade Grade Source: Ministry of Education, school survey data, 1998. ANALYZING EQUITY 55 to their number than do boys. Thus there is a need in Maputo City in ESG2. In the provinces of Cabo to understand what factors contribute to this rela- Delgado, Nampula, Niassa, and Manica, girls con- tive gender balance in order to replicate it in other stituted about 25 percent of the student population provinces. in ESG2, while in the other provinces, female par- The results presented for Cabo Delgado and ticipation was between 30 and 49 percent, as Gaza confirm the importance of provincial dispari- shown in table 5.4. ties and the fact that the equity dimension of edu- It is worrisome that the provinces with the cation in Mozambique has a different magnitude in widest gender gaps at the secondary level, such as different provinces. Targeted actions may be war- Cabo Delgado, Manica, Nampula, and Niassa, are ranted to redress provincial imbalances. the ones where females were already lagging in pri- mary education. This suggests that the source of Inequity in secondary enrollment gender disparities at the secondary level emanates from a disadvantage at the primary level. Regard- While these inequities tend to limit the chances of ing access to the secondary level, analyses of world having significant female participation beyond the trends by Alain Mingat3 show that, in the beginning lower primary level, in 1998, girls represented 42 of the 1990s, girls' chances to enroll in secondary and 47 percent of the total enrollment in ESG1 and school were lower than boys' in most countries. ESG2, respectively, which is low, particularly in Although girls are more advantaged in countries ESG1. Across provinces, the disparities are strik- such as Nicaragua and Honduras, they are at or ing, with girls' representation ranging from 21 per- near parity in countries like the Philippines, Zim- cent in Niassa to 55 percent in Maputo City in babwe, Colombia, and Madagascar. In Sub-Saharan ESG1 and from 21 percent in Manica to 53 percent Africa, girls remain very disadvantaged in most Table 5.4 Enrollment in absolute figures in ESG1 and ESG2, 1998 ESG1 ESG2 Girls Girls Province Number Percent Total Number Percent Total Cabo Delgado 680 22 3,116 84 25 332 Gaza 2,203 46 4,739 226 37 604 Inhambane 1,961 43 4,571 128 38 335 Manica 892 33 2,710 76 21 366 Maputo City 9,891 55 18,096 2,154 53 4,068 Maputo Province 2,707 49 5,525 202 33 612 Nampula 2,036 32 6,312 212 27 771 Niassa 554 21 2,633 104 23 455 Sofala 2,888 39 7,348 455 38 1,191 Tete 1,520 35 4,308 148 32 465 Zambezia 1,455 33 4,429 221 38 587 Total 26,787 42 63,787 4,010 47 8,786 Source: Ministry of Education; authors' calculation. 56 MOZAMBIQUE: COST AND FINANCING OF EDUCATION countries. In an effort to reverse these trends in EP1 girls' school survival.4 Reassessing teacher deploy- and EP2, the Mozambican government proposes to ment may also be needed. increase the number of female teachers and improve the image of females in general. Equity in the distribution of public resources for education among cohorts of students Some factors that may affect gender equity In order to assess structural inequity in education, Having established that, beyond provincial dif- this report examinesthe overall distribution of ferences, girls are generally at a disadvantage, it resources among members of a given cohort, irre- may be useful to identify which of the supply and spective of the social characteristics of those who demand factors may account for the persistent benefit from the resources. Secondly, the report gender disparities. In general, girls are more sensi- explores how social characteristics (such as gender, tive than boys to the organization of schooling. geographic location, and income), impact equity. Among these elements, availability of school and distance between home and school seem to play a Structural inequity in the distribution of significant role, as do the availability and character- public resources in education istics of teachers. Individuals appropriate public resources in educa- Availability of schools. The provinces with the tion through their schooling. Those who do not lowest retention rates are those with around a 30 access the education system do not benefit from percent gender gap in enrollment in EP1. Analyses these resources, while the longer individuals conducted in 1998 indicated that northern remain in the schooling system, the more resources provinces had the lowest percentage of complete they appropriate. Consequently, the distribution of schools; less than 60 percent of their children were public resources derives from (a) the distribution of in complete schools. A closer look at distribution the terminal level of schooling of the members of during the year in which the schools were estab- the cohort and (b) the structure of public unit cost lished shows that about 60 percent of the incom- across the various levels of education. For example, plete schools were not newly established schools for EP2, the enrollment ratio is 22 percent, but, as 7 and thus corresponded to a structural scenario. percent of the cohort moves up to ESG1, 15 percent Therefore, although other explanations such as of the cohort has EP2 as their terminal level of family conditions might contribute to low female schooling. Examining the cumulative resources retention, the distance from home to school seems associated with each terminal level shows that to affect access and retention for both boys and those whose terminal level is EP2 will have appro- girls, but it has a much stronger impact on girls than priated Mt1.2 million in EP1 and another Mt1.4 mil- it does on boys. lion in EP2, amounting to Mt2.6 million. Annex 11 displays the data used to compute the distribution Availability and distribution of teachers. The avail- of public resources in education among a hypothet- ability and qualifications of teachers are without ical cohort of 100 children. doubt among the key determinants of school qual- Using cumulative information on the terminal ity and student survival. In the last several years, level of schooling, on the one hand, and the more attention has been paid not only to teacher resources appropriated, on the other hand, we con- training but also to the gender consciousness of structed the Lorenz curve (see figure 5.6). The first teachers and the need to increase the number of diagonal in the graph indicates equitable distribu- female teachers. This advocacy for more female tion of public resources, while the Lorenz curve teachers was first governed by a concern over the shows by how much the reality of the country devi- lack of role models for female students. Today, the ates from the equitable line. evidence is even more compelling that increasing The Gini coefficient is useful in quantifying the the number of female teachers substantially boosts degree of structural inequity.5 By definition, the ANALYZING EQUITY 57 Figure 5.6 lized for education in Mozambique in 1998; this rep- Lorenz curve of the distribution of public resents a case that is not close to equity. However, to resources in education, 1998 put the analysis in a comparative perspective over time, the Gini coefficient was estimated to be 0.66 in Percent of cumulative spending 1992, suggesting that there was some improvement 100 B in the structural distribution of public education 90 resources in the country. 80 70 Social inequity in the distribution of public 60 resources in education S 50 40 The second method by which to assess equity in an 30 educational system is to examine the social charac- 20 A teristics of targeted groups in order to determine 10 who benefits more from the public resources made 0 available to the system. Based on school survey 0 10 20 30 40 50 60 70 80 90 100 data, we conducted analysis of equity by gender. Cumulative percentage of cohort The data on the distribution, by terminal level of Source: Ministry of Education; school census and population census data; author's calculation. schooling, of a cohort of 100 children (50 boys and 50 girls) and on the distribution of public resources in education by sex are given in table 5.5. value of the Gini coefficient varies between 0 and 1; The table provides a breakdown by gender of the smaller values indicate less inequity in the struc- total amount of public resources spent on education tural distribution of public resources. In Mozam- for a cohort of 100 children. Out of the Mt197.1 mil- bique, the value of the Gini was 0.53 for 1998. For a lion that represent the total spending for the cohort more directly interpretable figure, it was estimated (table 5.5), Mt77.7 million was appropriated by girls that the best educated 10 percent of a cohort appro- and Mt119.4 million by boys, representing, respec- priated about 63 percent of the total resources mobi- tively, 39 and 61 percent. Thus, in a population of 50 Table 5.5 Distribution of public resources in education, by gender, 1998 Aggregate Enrollment Terminal level Cumulative per cumulative spending ratio (percent) of schooling pupil spending (millions of meticáis) (millions of Level of schooling Girls Boys Girls Boys meticáis) Girls Boys No schooling 15.0 2.0 0 0 0 EP1 70.0 96.0 26.1 34.9 1.2 31.3 41.9 EP2 17.8 26.2 6.1 9.2 2.6 15.9 23.9 ESG1 5.6 7.8 2.2 3.0 6.8 15.0 20.4 ESG2 1.2 1.8 0.45 0.45 14.9 6.7 6.7 Higher education 0.3 0.9 0.15 0.45 58.9 8.8 26.5 Total -- -- 50.0 50.0 -- 77.7 119.4 -- Not Available. Source: Ministry of Education, school survey data, 1998. 58 MOZAMBIQUE: COST AND FINANCING OF EDUCATION boys and 50 girls, boys get more than 50 percent (61/39) more resources than girls do. 1. To document this pattern, we computed the relative gender gap (the ratio of the difference between the enroll- From this equity analysis, we can conclude that ment rate of boys and girls to the GER of girls) and plotted (a) the very few who persist in the system absorb a that statistic against the global GER at the provincial level. significant proportion of the resources (the 10 per- Figure 5.1 shows that, on average, the wider is the coverage, the narrower is the gap. cent best educated appropriate as much as 63 per- 2. A. Mingat and Suchaut, B. "Les systèmes éducatifs cent of the public resources used in the sector and africains: Une analyse économique comparative." DeBoeck that (b) boys appropriate about 50 percent more of Université. Brussels, Belgium these resources than girls. Both structurally and 3. A. Mingat and V.M. Kenghe (forthcoming). "La scolarisa- tion des filles dans le contexte de l'Afrique noire."The World socially, the serious inequities in the Mozambican Bank. Washington, D.C. educational system ought to be addressed. 4. V.M. Kenghe and Mingat, A. 2000. Intgernational Compara- tive Analysis of Female Teachers; Assessing the Impact of the Gender of the Teacher upon the Performance of Primary Education in Africa. The World Bank. Washington, D.C. 5. The Gini coefficient is determined as the ratio of the area between the Lorenz curve and the first diagonal (S) and the area of the triangle OAB. 6 A Reminder on External Efficiency C ontrary to the analysis of internal efficiency, the level of the individual (better educated people which assessed the operation of the school- command higher earnings than less educated ones) ing system using results observed while the and at the level of society (better educated societies students were still enrolled (student learning, enjoy higher levels of economic growth or achieve a schooling career, or the level of disparities in out- better level of health indicators). comes among different groups), external efficiency To be sure, the collective effects of education gen- is concerned with the performance of graduates erally go beyond what is appropriated privately by after they exit the schooling system and enter their individuals, and education carries what economists working life. The relevance of education and train- call positive externalities. To illustrate the concept ing activities in a country is therefore found in the of externality, one can take the example of the edu- social and productive life of individuals. To what cated farmer who improves his agricultural tech- extent does what a cohort of youngsters learns niques (choice of new crops, use of better seeds, while in school (even though for quite different adequate use of fertilizers). This change results for amounts of time, from 0 to 20 years) correspond to him in higher earnings (private impact), but it may the best that can be achieved both for the individu- also lead his less educated neighbors to imitate als and for society given the amount of resources what he did and result, for them too, in a higher made available to the sector? level of productivity (the individual who originated From this perspective, one can distinguish eco- the change in agricultural techniques did not cap- nomic and social effects, on the one hand, from indi- ture the positive externality). vidual and collective effects, on the other. Social To analyze a system of education and training in effects may concern dimensions such as health (for terms of external efficiency, we address two com- example, more educated mothers tend to be more plementary objectives. First, how can a government effective vis-à-vis the health of their children), civic allocate the available public resources across the life (more educated people are likely to contribute different levels and types of education and training more to community relations and make informed in order to get the maximum amount of social ben- political choices), or population (more educated efits from its investments in human capital? Second, societies are more likely to control demographic given the importance of externalities, as well as the growth). Concerning the impact of education in the dual nature (private and social) of the investments economic sphere, the relationships with both in human capital, what mechanisms (institutional employment and economic growth are obviously of and financing arrangements, in particular) would primary importance. These impacts in the economic help to align the behavior of individuals with what or social domains may themselves be seen both at would be optimal from a social point of view? 59 60 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Providing answers to these two questions human capital to satisfy, as efficiently as possible, requires consideration of country-specific condi- the demands from the two sectors of the economy. tions, such as the characteristics and status of the If this is indeed the case, what is the implication for labor market. Despite a dearth of labor market doc- the traditional sector and for the modern sector of umentation in Mozambique, existing data does the economy? indicate an education and training strategy that could help the country to marshal and distribute General social benefits and the impact the type of human capital likely to maximize social of education in the traditional sector and economic development. of the economy A basic element to consider is that the Mozambi- can economy is, and will remain in the next Education helps to improve the living conditions of decades, characterized by a dualistic structure with individuals by improving their earnings (this is a dominant traditional sector and a relatively small, observed in all poverty studies, the 1996 survey in but growing, modern sector of employment. Any Mozambique being no exception), their health strategy for human capital formation will have to status, and those of their children, and it contributes provide the type of human capital that best fits the to reducing fertility rates. In Mozambique as else- demands of the two subsectors and will have to where, basic education has a larger impact on these help the economy move from its traditional to its outcomes for females than for males. To a large modern segment. extent, the impact of education on these results may In 1998, Mozambique's per capita GDP was esti- go beyond what is directly appropriated by the mated at about US$250 following a six-year period individuals. The positive effects of basic education of relatively strong growth (7 percent per year on on health are large enough in themselves to justify average between 1992 and 1998). If the economy investments in basic education (in fact, the health were to grow 8 percent per year for the next 10 impact of one dollar spent on girls' education may years, per capita GDP could increase from the be larger than that of the same dollar spent on stan- US$250 in 1998 to about US$400 in constant terms in dard health activities).2 2010.1 However, even experiencing such impressive A recent analysis of all low-income countries of economic growth, Mozambique would be consid- the world (with per capita GDP below US$2,000, in ered a low-income country. 1993 U.S. dollars) showed that countries that demon- In 1998, employment in the modern sector of the strated the best economic performance were those economy represented something like 9 percent of that had previously invested more in education.3 the workforce, leaving 91 percent for the traditional However, two complementary and more specific sector, which is predominantly agricultural. This results are of even greater importance. First, for structure is typical of countries sharing similar countries in the low-income group, only primary levels of economic development, in particular in the education had a positive and significant impact on African context. International comparative analysis economic growth; investments in both secondary of trends in the distribution of employment and higher education did not have a consistent or between the traditional and the modern sector, significant positive impact on economic performance along with economic development (estimated from in the countries under consideration. Second, the either a time-series or a cross-sectional perspective), impact of primary education on economic growth, demonstrates a consistent pattern. The traditional even though positive, was weaker, on average, in the sector and agriculture will still be major contribu- African context than in other regions of the world. tors to overall employment in 2010. The share of What do these two aspects suggest for Mozambique? employment in agriculture may diminish, but only The fact that primary education had a positive by about 5 percent, while the share of employment impact on economic growth was expected; the fact in industry will remain minimal; employment in that no such impact was found for secondary and services and construction likely will develop. Based higher education was less predictable. This leads to on these estimates, the country will have to provide two simple ideas. The first point is that primary A REMINDER ON EXTERNAL EFFICIENCY 61 education is an all-purpose investment whose con- This being said, even though the impact of pri- tent (basic reading and writing, basic arithmetic and mary education is not as large in Africa as it is in problem solving, basic life skills) fits the variety of other regions of the world, its estimated global activities concerned with improving productivity in social rate of return (from a growth model, which the traditional sector, however simple those activi- includes the influence of social benefits on eco- ties are. The second point is that, in a number of nomic outcomes) would still be around 16 percent developing countries (African countries being no per year, a substantial figure.6 Micro studies con- exception), secondary education and even higher firm that primary education does positively impact education have been expanded to such an extent the informal sector by enhancing the ability of pri- that their economies are incapable of using the mary students to capitalize on work experience in graduates in an efficient way (unemployment of that sector, and to command higher earnings from graduates is widespread in a number of African these informal activities. countries). Under these circumstances, over-invest- In general, social and economic external effi- ing in these levels of education becomes more a ciency clearly supports the development of primary burden on economic growth than an engine for it. education. Mozambique therefore should design a Nevertheless, it should be pointed out that often it strategy that would lead to effective and full cover- is not the investment in post-primary education, age of a reasonably good quality of basic education. that constitute the problem, but rather the quality As a minimum, all children in a given cohort and relevance of education offered and its linkages should, in the near future, complete the last grade of to the economy and labor market.4 It is self-evident EP1 (with five years of education validated) and, that no country can develop socially or economi- probably later on, continue through the full EP1 and cally without investment in secondary and higher EP2 cycles of study. education. However, such investment needs to be Such a policy also would be supported by equity balanced with the rest of the education system on in considerations, since those who currently do not get the context of the social and economic development to the last grade of EP1 come disproportionately of the country.The results of a number of research from deprived households. In the context of studies show that the impact of primary schooling poverty reduction and the country's debt reduction on economic growth is less in African countries initiative, there is no doubt that improvements in than in low-income countries in other regions of the coverage and quality of primary education will be a world. Two types of explanations can account for priority for Mozambique in the years to come. This this result. First, African pupils tend not to work in does not mean that other levels of education should agriculture or in traditional trades, as schooling is be left out of the overall strategy for the develop- meant to enable them to get a job in the civil service. ment of human capital in the country. Second, the impact of primary education on the labor productivity of farmers is especially signifi- Education and training for the modern sector cant when agricultural activities are supported by a of the economy facilitating environment (for example, a land policy, a price policy both for inputs and outputs, active If primary education is to be considered the all-pur- extension workers). When these conditions are not pose human capital investment that fosters produc- met (less in the African than in the Asian context), tivity in the traditional sector, the growth path for primary education has a smaller impact on produc- the country undoubtedly rests on development of tivity in agriculture (as well as on productivity in the modern sector of the economy, particularly as general).5 Besides, the impact of primary education this often forms the fiscal basis for revenue collec- on economic development (or on success in World tion. Since the productivity per worker is much Bank projects) is much larger when the proportion higher in the modern than in the traditional sector, of literate workers exceeds approximately half of Mozambique should do as much as it can to foster the labor force; Mozambique is still below this the expansion of its modern productive sector. From threshold. this perspective, human capital plays an obvious 62 MOZAMBIQUE: COST AND FINANCING OF EDUCATION role, and this probably will be even more the case in a system is not sufficient per se to enable effective light of the technological changes currently under regulation of the flow of graduates in relation to way and anticipated. labor market demands. However, the experience accumulated in the Mechanisms are needed for taking effective developing countries over the last three decades action, and these need to address both supply and clearly indicates that (a) although the expansion of demand. The first mechanism is administrative in the modern sector can be hampered by a lack of nature. The state monitors the development of the qualified manpower, the sector's dynamics are system (controls supply) through (a) the admission determined by macroeconomic policies that are of students (number and characteristics) in different exogenous to the sector, and (b) the supply of edu- segments, trades, and fields of specialization of cated manpower must be driven by the demands technical education, vocational training, and higher for quantity and quality. In other words, the avail- education and through (b) control of the budget and ability of human capital is a condition that is neces- operation of the institutions delivering the services. sary, but not sufficient, for development of the The state establishes a system of data gathering (for modern sector. Even though some possibilities of example, general data on employment, direct and substitution exist in the labor market (substitution reverse tracer studies) and uses this information to between labor and capital, on the one hand, and inform its strategy for monitoring the development between different types of human capital, on the of the system over time. Sometimes this type of reg- other), there is a point beyond which a further ulation performs poorly. One reason is that it rests increase in the supply of graduates would result in on confidence in labor market studies, and these unemployed graduates and wasted resources.7 This studies sometimes are not conducted in a rigorous concerns, in particular, technical education, voca- enough manner. A second, and probably more tional training, and higher education. We do not important, reason is that the method depends sub- have well-established, factual documentation of the stantially on the capacity of the policymaker to case for Mozambique, but anecdotal observations make difficult decisions when they are justified on suggest that some employment problem may exist well-documented efficiency and equity grounds. for the graduates in technical education and voca- The second type of mechanism relies on market tional training. The contrary may be true for higher forces to regulate the output of graduates and the education, where students in certain fields--like operation of the education and training institutions engineering and information technology--leave (controls demand) so that the supply of graduates before completing their studies. reasonably matches the demands of the economy. Tracer studies and recruitment studies do not Such a mechanism relies on the invisible hand of the exist in Mozambique, thereby presenting difficul- market, and must first identify the interests of the ties in assessing the extent to which higher educa- actual or potential individuals implied in the game. tion and technical education and vocational train- In addition, the mechanism must determine how to ing are in line with demands from the economy set up a structure in which there is convergence (quantity and distribution of trades and fields of between private and public interests. Along this specialization, relevance of the curriculum content). line, investment in human capital is partly a public Also lacking is a basic system of data collection on good and partly a private good. It is private in that employment in the various segments of the modern individuals privately appropriate a substantial sector (for example, civil service, state enterprises share of the benefits accruing from the investment and parastatals, private sector, cooperatives). in the form of higher earnings and often contribute Mozambique needs a system of data gathering in to its financing. order to document the relationship between the In higher education, three players are involved: output of education and the labor market, as well as the students, the providers of education and training to assess the degree of external efficiency of techni- services, and the employers. The first two agents-- cal and higher education and its evolution over students and school management--may play the time. However, experience demonstrates that such major role in higher education, while employers and A REMINDER ON EXTERNAL EFFICIENCY 63 the final users may have a more central role in tech- of mechanisms, elements can be borrowed from nical education and, especially, vocational training each to correspond to the specific social and eco- or applied research. In higher education, the system nomic circumstances of the country. In other words, is all the more likely to operate on its own if students there is clearly a wide array of choices to consider in are strongly inclined to choose trades and fields of implementing an instrument that would help to specialization that are in high demand in the labor align human capital formation at the uppermost market (they are even more likely to do so if they levels with the evolving demands of the economy. contribute to the financing of their studies). The However, keeping the status quo--the absence of same is true when the providers have an interest in such a mechanism--should not be considered a rea- developing these types of studies and in delivering sonable option. the services at a low cost to attract fee-paying cus- tomers and reap some benefit from their activities. This does not imply that the students pay the full 1. Eight percent growth and 3 percent population growth yields approximately 5 percent real growth. cost; if a school pays even a small fraction of the cost, 2. Lawrence H. Summers, "Investing in all the people", students behave in an efficient way. It also does not (World Bank, Washington D.C., 1992). imply that all students should pay the same or that 3. Alain Mingat and Bruno Suchaut, "Les systèmes éducatifs the contribution should be either now or in mone- africains: Une analyse économique comparative" (World tary terms--payment can be deferred to a time Bank, Washington, D.C., 2000). 4. Constructing Knowledge Societies: New Challenges for when the graduate has started to earn a living, and Tertiary Education (World Bank, Washington D.C., 2002) students might be asked to work as primary teach- 5. This was demonstrated by Mark Rosenzweig, who com- ers for two years before becoming eligible for admis- pared green revolution zones with areas where no such activ- sion to higher education. On the institutional side, ities were undertaken in the Indian context. Mark Rosen- zweig, "Where Are There Returns to Schooling?" American structural and financing arrangements might be dis- Economic Review, Papers and Proceedings 85 (May 1995): cussed. The institutions might be private and be 153­48. partly subsidized; they might also be fee-paying 6. Mingat and Suchaut, "Les systèmes éducatifs africains: public institutions. In both cases, some of their Une analyse économique comparative" (World Bank, Wash- ington, D.C., 2000). resources might eventually be linked to the per- 7. The fact that higher education does not present a positive formance of their students in the labor market. marginal impact in the growth models discussed in the pre- For technical education and vocational training, vious section illustrates the case. similar mechanisms might be discussed, but it might be useful to allow employers to enter the pic- ture. In particular, instruments such as matching grants have proven useful for matching the output of training institutions with demand from the econ- omy; requiring one or two years of employment in the trade of specialization before graduation also might be discussed. In conclusion, although virtues and flaws can be found in both the administrative and market types Annexes Annex 1 Population, GDP, and total public expenditures, 1990­99 Indicator 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Population (millions)a 14.1 14.2 14.4 14.6 14.7 14.9 15.1 15.3 15.7 16.1 GDP (current trillions of meticáis) 2.3 3.6 4.8 7.8 13.1 21.3 32.1 397 46.2 52.9 GDP per capita, (US dollars, new population census) 179 175 136 144 151 160 188 225 249 259 GDP (trillions of constant 1995 meticáis) 18.1 19.0 17.4 19.0 20.4 21.3 22.8 25.4 28.4 30.9 Real growth rate per year (percent) -- 5 ­8 9 7 4 7 11 12 9 GDP per capita (thousands of constant 1995 meticáis) 1,287 1,334 1,211 1,301 1,383 1,425 1,509 1,660 1,811 1,926 Total revenue domestic (billions of meticáis) 298 447 661 1,093 1,526 2,413 3,479 4,623 5,311 6,262 Total external grants, before loans (billions of meticáis) 226 397 652 932 1,857 2,090 2,291 3,705 3,818 6,381 Total revenues (billions of meticáis) 524 844 1,313 2,025 3,383 4,503 5,770 8,328 9,129 12,643 Total government expenditures (billions of meticáis) 690 958 1,483 2,305 4,097 5,157 6,773 9,498 10,207 13,460 Recurrent spending 343 458 757 1,167 1,978 2,188 3,077 4,272 5,268 6,606 Capital spending 347 501 726 1,137 2,119 2,969 3,696 5,226 4,939 6,855 Overall balance after grants (billions of meticáis) ­166 ­114 ­169 ­280 ­714 ­654 ­1,003 ­1,170 ­1,078 ­817 Ratio of total government expenditures to GDP (percent) 29.6 26.8 31.2 29.4 31.2 24.2 21.1 23.9 22.1 25.4 Ratio of revenue to GDP (percent) 12.8 12.5 13.9 14.0 11.6 11.3 10.8 11.6 11.5 11.8 Exchange rate (metacal to U.S. dollar) 929 1,435 2,433 3,723 5,918 8,890 11,294 11,546 11,850 12,711 a. Population growth from 1990 to 1997 is based on a growth rate of 2 percent per year. After 1997 the growth rate is estimated at 2.3 percent. Source: GDP, total government expenditures: International Monetary Fund; public education expenditures: Ministry of Education; GDP per capita in U.S. dollars: estimates based on 1997 population census figures and end-of-year exchange rate. 64 ANNEXES 65 Annex 2 Public Spending on education as a share of GDP and total government spending, 1990­1999a 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 GDP (billions of meticáis) 2,334 3,580 4,757 7,829 13,145 21,267 32,093 39,693 46,134 52,913 Total government spending (billions of meticáis) 690 958 1,483 2,305 4,097 5,157 6,773 9,498 10,207 13,460 Recurren spending 343 458 757 1,167 1,978 2,188 3,077 4,272 5,268 6,606 Capital (domestic + external) spending 347 501 726 1,137 2,119 2,969 3,696 5,226 4,939 6,855 Total government of spending as a percentage of GDP 29.6 26.8 31.2 29.4 31.2 24.2 21.1 23.9 21.8 25.2 Ministry of Education spending 109 171 263 380 653 799 1,300 1,322 1,468 1,645 Total domestic expenditures (billions of meticáis) 50 75 125 171 203 268 486 648 875 1,005 Recurrent domestic spending 45 65 104 145 142 249 441 583 798 907 Capital domestic spending 5 10 21 26 61 19 45 65 77 98 Total external expenditures (billions of meticáis) 59 96 138 209 450 531 814 674 593 640 Ministry of Higher Education spending -- -- -- -- -- -- -- 237 400 476 Total domestic expenditures (billions of meticáis) 7 12 22 34 58 65 109 123 163 21 UNESCO 0 0 0 0 0 1 1 1 1 1 Eduardo Mondlane University 5 10 19 29 49 52 92 106 145 195 Pedagogical University 1 2 3 5 8 12 16 16 17 18 Research Institute for International Relations 0 0 0 0 0 0 0 0 0 0 Total external expenditures (billions of meticáis) -- -- -- -- -- -- -- 114 237 262 Total domestic expenditure on education (billions of meticáis) 57 87 148 206 260 333 595 771 1,038 1,219 Total external expenditure on education (billions of meticáis) (59) (96) (138) (209) (450) (531) (814) 788 830 902 Total expenditure on education (domestic + external) (116) (183) (286) (415) (710) (864) (1,409) 1,559 1,868 2,121 Share of total expenditure on education in total government expenditure (percent) 16.8 19.1 19.3 18.0 17.3 16.8 20.8 16.4 18.3 15.8 Share of recurrent domestic spending on education in total government recurrent spending (percent) 15.2 16.8 16.6 15.3 10.1 14.4 16.4 15.4 17.6 15.8 Total expenditure on education as a percentage of GDP) 5.0 5.1 6.0 5.3 5.4 4.1 4.4 3.9 4.0 4.0 Domestic spending on education as percentage of GDP 2.4 2.4 3.1 2.6 2.0 1.6 1.9 1.9 2.2 2.3 Exchange rate (meticáis per US dollar) 929 1,435 2,433 3,723 5,918 8,890 11,294 11,546 11,850 12,711 --- Not available. Source: Ministry of Education and World Bank, 1997 Mozambique Population Census. (Washington, D.C.: World Bank, 1997 a. Figures denote actual expenditures, except for 1999, wich denotes the amount budgeted. 66 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Annex 3 Estimated population, school-age population, and teacher requirements with and without incidence of HIV/AIDS, select years, 1997­2010 Scenario and indicator 1997 2000 2005 2010 Without AIDS Total population (millions) 15.3 16.4 18.3 20.5 School-age population (millions) 3.2 3.5 4.0 4.7 Stock of teachers 40,000 43,709 50,671 58,741 Replacement of teachers due to attrition (attrition rate of 4 percent per year) 1,600 1,748 2,027 2,350 Number of new teachers required per year 2,836 3,140 3,641 4,150 With AIDS Total population (millions) 15.3 16.0 17.2 18.5 School-age population (millions) 3.2 3.4 3.8 4.2 Stock of teachers 40,000 43,455 49,889 57,276 Replacement of teachers due to attrition (attrition rate of 7 percent per year) 2,800 3,042 3,492 4,009 Number of new teachers required per year 3,951 4,329 4,969 5,609 Note: Total population growth rates: 2.3 percent without AIDS and 1.5 percent with AIDS; school age population growth rates: 3 percent without AIDS, and 2.8 percent with AIDS. Source: World Bank scenario. ANNEXES 67 Annex 4 Enrollment and gross enrollment rates, by level and type of education, select years, 1993­99 1993 1995 1997 1999 Number of GER Number of GER Number of GER number. of GER students (percent) students (percent) students (percent) students (percent) Primary EP1 1,237,063 60.3 1,436,831 68.3 1,780,881 82.7 2,108,790 92.3 EP2 118,909 16.7 136,464 18.7 164,078 22.0 193,523 24.4 Secondary ESG1 31,761 3.2 40,588 4.0 51,821 4.9 72,914 6.6 ESG2 1,612 0.3 4,927 0.8 4,253 0.6 9,142 1.3 Technical or vocational Elementary 100 158 253 499 Basic 15,669 16,907 18,529 20,497 Middle 1,868 3,745 4,337 n.a.y. Teacher traininga Basic 3,950 3,598 3,836 4,300 Middle level 166 240 515 114 Additional middle level -- -- 658 1269 Higher education 5,283 6,844 8,650 11,619 Public Eduardo Mondlane University 4,036 5,200 6,200 6,800 Pedagogical University 1,214 1,489 1,520 1,987 Research Institute for International Relations 33 155 155 234 Private Higher Institute of Science and Technologyb 201 644 Polytechnic Universityc 371 919 Catholic Universityc 203 1,035 -- Not available. a. Technical or vocational education run by the Ministry of Education as presented in this table includes both day and evening classes. Includes elementary, basic and middle levels. Enrollment figures for students in the middle level of technical/vocational education in 1999 is not available b. Opened 1997. c. Opened 1996. Source: School survey data, 1993, 1995, 1997, and 1999; census data, 1997. 68 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Annex 5 Distribution of recurrent government spending on education, by level of education, 1996­99 (billions of meticáis) 1996 1997 1998 1999 Type of spending Amount Percent Amount Percent Amount Percent Amount Percent Ministry of Education 441,089 583,218 798,039 907,711 Administration 87,146 16.2 109,862 15.4 172,074 17.9 170,754 15.2 Central 51,654 57,782 88,783 98,702 Provincial directorates 23,217 38,832 53,859 45,559 District directorates 12,274 13,248 29,432 26,493 Lower primary (EP1) 195,569 36.4 261,922 36.6 311,536 32.4 412,408 36.8 Upper primary (EP2) 45,442 8.5 55,865 7.8 72,464 7.5 90,485 8.1 Lower secondary (ESG1) 35,366 6.6 43,467 6.1 65,675 6.8 72,066 6.4 Upper secondary (ESG2) 12,804 2.4 25,187 3.5 25,192 2.6 34,283 3.1 Technical education 24,921 4.6 34,966 4.9 51,831 5.4 46,008 4.1 Technical (elementary) 500 1,145 2,756 2,898 Technical (basic) 14,258 19,513 29,243 28,870 Technical (middle) 10,163 14,308 19,833 14,240 Teacher training 17,620 3.3 15,922 2.2 24,894 2.6 34,325 3.1 Teacher training (basic) 9,463 9,188 13,426 11,800 Teacher training (middle) 8,157 6,735 11,467 6,119 Miscellaneous 22,221 4.1 36,027 5.0 74,374 7.7 47,382 4.2 Adult education 2,598 3,582 5,635 1,830 Teacher housing and student dormitories 13,752 14,616 29,054 20,394 Student-teacher dormitory 9,831 9,833 15,400 11,601 Student dormitory 3,921 4,783 13,655 8,793 Sports department 434 369 404 470 Short-term training 960 5,041 16,104 -- Others 4,477 12,419 23,177 24,688 Higher education (under Ministry of Planning and Finance) 96,000 17.9 132,000 18.4 163,000 17.0 214,000 19.1 Total 537,089 100.0 715,218 100.0 961,039 100.0 1,121,711 100.0 -- Not available. Source: Ministry of Educatiion and Minisry of Planning and Finance. ANNEXES 69 Annex 6 To understand why unit costs vary with regard to schooling conditions, it is useful to consider the following analytical framework: B Budget SBT Salary bill of teachers SBNT Salary bill of nonteaching staff PEDOP Aggregate spending for pedagogical materials and operation ADM Aggregate spending on administration AVT Average teacher salary NT Number of teachers AVNT Average salary of nonteaching staff NNT Number of nonteaching staff at the school level RUC Recurrent unit cost NSTU Number of students PTR Pupil to teacher ratio PNTR Pupil to nonteaching staff ratio PEDPP Spending on pedagogical materials per pupil ADMU Average spending on administration per pupil B = SBT + SBNT + PEDOP + ADM B = AVT * NT + AVNT * NNT + PEDOP + ADM RUC = B/NSTU = AVT * NT/NSTU + AVNT * NNT/NSTU + PEDOP/NSTU + ADM/NSTU. This yields the following expression: RUC = AVT/PTR + AVNT/PNTR + PEDPP + ADMU. 70 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Annex 7 Classroom construction costs in Mozambique, Madagascar, and Senegal, 1998 Country and indicator Rural primary schoola Community schoolb Mozambique Construction cost of one classroom Mt200,000,000 Mt137,500,000 Annualized value of one classroomc Mt16,600,000 Mt11,500,000 Annual teacher salary, EP1 Mt10,000,000 Mt10,000,000 Cost of construction Number of years of annual teacher salary 20.0 13.7 Annualized value as a percentage of recurrent unit cost 133 92 Madagascar Construction cost of one classroom Fmg13,000,000 Annualized valuea Fmg1,100,000 Annual teacher salary Fmg3,360,000 Costs of construction Number of years of annual teacher salary 3.9 Annualized value as a percentage of recurrent unit cost 29 Senegal Construction cost of one classroom Fcfa4,000,000 Annualized valuec Fcfa325,000 Annual teacher salary Fcfa1,800,000 Costs of construction Number of years of annual teacher salary 2.2 Annualized value as a percentage of recurrent unit cost 16 a. The cost of one classroom is based on the current Education Sector Strategic Program cost of a classroom, latrine, and provision for supervision, calculated as follows: EPR = 1 classroom.(US$14,000) + 1 latrine (US$1,000) + supervision (US$500) = US$16,000 = Mt200 million. b. The cost of one classroom is based on Lutheran World Federation (LWF) and PRONES (UNICEF) standards = 1 classroom (US$ 9,500) + 1 latrine (US$1,000) + supervision (US$500) = US$11,000 = Mt137.5 million. c. Based on an opportunity cost of capital of 5 percent and a lifetime of 20 years. Source: World Bank (Ministry of Education) Project Appraisal Document, Annex 13; World Bank estimates. ANNEXES 71 Annex 8 Relation between the number of teachers and pupils in public EP1 schools, 1998 Model 1 Model 2 Number of Number of Province pupils teachers Coefficient t Coefficient t Number of pupils n.a. n.a. 0.016 189 0.0165 175.6 Cabo Delgado 134,112 2,421 Reference n.a. Gaza 184,131 2,382 ­1.60 11.7 Inhambane 175,071 2,499 ­1.18 8.4 Manica 113,176 1,824 ­0.51 3.2 Maputo Provincia 141,229 1,807 ­2.38 13.9 Nampula 295,988 5,559 0.27 2.3 Niassa 92,108 1,949 0.39 2.9 Sofala 121,393 1,966 ­0.48 3.1 Tete 138,970 2,882 0.76 5.5 Zambezia 340,845 4,984 ­0.85 7.5 Cidade de Maputo 143,926 2,399 ­0.03 0.1 Intercept n.a. n.a. 0.04 0.39 R2 (percent) n.a. n.a. 85.5 87.2 n.a. Not applicable. Source: Ministry of Education. Bank calculation. 72 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Annex 9 Relationship between the number of teachers and pupils in EP1 at the school level, by province, 1998 Number R2 Province of schools Intercept Coefficienta (percent) N300b Total 6,035 0.04 0.0162 85.5 4.9 Cabo Delgado 555 0.36 0.0166 89.2 5.3 Gaza 534 ­0.74 0.0151 82.9 3.8 Inhambane 477 ­0.86 0.0166 87.7 4.1 Manica 306 ­0.38 0.0171 87.7 4.8 Maputo Province 258 0.97 0.0110 94.6 4.3 Nampula 1,044 ­0.65 0.0211 93.0 5.7 Niassa 558 0.27 0.0195 83.1 6.1 Sofala 343 ­1.09 0.0193 91.1 4.7 Tete 521 ­0.19 0.0215 93.3 6.3 Zambezia 1,344 ­1.20 0.0194 80.4 4.6 Maputo City 85 5.75 0.0132 76.4 9.7 a. All coefficients significant at the 0.001 level. b. N300 is the estimated number of teachers in a 300-pupil school (country average). Source: Ministry of Education. World Bank calculation. ANNEXES 73 Annex 10 Basic data for schooling profiles, by gender and rural or urban location, 1998 (percent) Gender Geographical location Ratio of Ratio of Boys Girls boys to girls Total Urban Semiurban Rural urban to rural Access to G1 104.0 84.0 1.24 93.7 91.0 110.0 94.0 0.97 Enrollment rate in G5 42.0 27.0 1.56 34.9 59.5 33.0 22.1 2.69 Retention rate, G1-G5 40.4 32.1 37.2 65.3 30.0 23.6 Transition rate, G5-G6 59.6 63.7 55.9 75.0 52.4 46.6 Enrollment rate in G6 24.1 17.2 1.40 20.8 44.6 17.3 11.0 4.05 Enrollment rate in G7 16.8 11.6 1.45 14.3 38.9 14.5 8.0 4.86 Retention rate, G6-G7 69.7 67.4 68.8 87.2 83.8 72.7 Transition rate, G7-G8 43.5 43.4 44.7 -- -- -- Enrollment rate in G8 7.3 5.4 1.35 6.4 -- -- -- Enrollment rate in G10 3.9 2.4 1.63 3.1 -- -- -- Retention rate, G8-G10 53.4 44.4 48.4 -- -- -- Transition rate, G10-G11 51.2 50.0 51.6 -- -- -- Enrollment rate in G11 2.0 1.2 1.66 1.6 -- -- -- Enrollment rate in G12 1.5 0.8 1.88 1.1 -- -- -- Retention rate, G11-G12 75.0 66.7 71.0 -- -- -- -- Not available. Source: Ministry of Education, school survey data, 1998. 73 74 MOZAMBIQUE: COST AND FINANCING OF EDUCATION Annex 11 Structural distribution of public resources, 1998 1 2 3 4 5 Enrollment Terminal Per-pupil spending Cumulative per-pupil Aggregate cumulative spending ratio level of per cycle (millions spending (millions Amount (millions Level of schooling (percent) schooling of meticáis) of meticáis) of meticáis) Percent No schooling ­ 17 0 0 0 0 EP1 83 61 1.2 1.2 73.2 37.1 EP2 22 15.3 1.4 2.6 39.8 20.2 ESG1 6.7 5.2 4.2 6.8 35.4 18.0 ESG2 1.5 0.9 8.1 14.9 13.4 6.8 Higher education 0.6 0.6 44 58.9 35.3 17.9 Total ­ 100.0 ­ ­ 197.1 100 -- Not available. Source: Ministry of Education. World Bank calculation. 74 References Armstrong, Jill. 1995. Uganda's AIDS Crisis: Its Implications Mingat, Alain, and V.M. Kenghe (forthcoming). "La sco- for Development. World Bank Discussion Paper 298. larisation des filles dans le contexte de l'Afrique Washington D.C. noire."The World Bank. Washington, D.C. Fry, Peter and Rogèrio Utui (ADEA Working Group on Nellemann, Soren, and Virgilio Juvane, 1998, Financing Higher Education). 1999. Promoting Access, Quality and Plan, Education Sector Strategic Program, 1998­2002, Capacity Building in African Higher Education: The Ministry of Education, Mozambique. Strategic Planning Experience at the Eduardo Mondlane Palme, Mikael. 1997. Cultural Ambiguity and the Primary University. World Bank, Washington D.C. School Teacher, in "Education, Cultures and IBE (International Bureau of Education). 1997. 1998. 1999. Economies," Chapter 14. Upsala University, Sweden. World Data on Education. Geneva. Rosenzweig, Mark. 1995. Where are There Returns to INDE, National Institute of Educational Development. Schooling? American Economic Review, Papers and 1997. Cadernos de pesquisa. Ministry of Education, Proceedings. Mozambique. Summers, Lawrence H. 1992. Investing in All the People. Lockheed, Marlaine E, Adriaan Verspoor, and associates. World Bank. Washington D.C. 1991. Improving Primary Education in Developing Tan, Jee-Peng, and Alain Mingat. 1992. Education in Asia. Countries. New York: Oxford University Press. A Comparative Study of Cost and Financing. World Bank Ministry of Education, 1998. Benefits of Learning. Ministry Regional and Sectoral Studies. Washington D.C. of Education, Mozambique. TFHES (Task Force on Higher Education and Society). Ministry of Education. 1998. Education Sector Strategic 2000. Higher Education in Developing Countries: Promise Plan, 1997­2001. Ministry of Education, Mozambique. and Peril. Washington D.C.: World Bank for the TFHES. Ministry of Education. 1990-2000. Annual Statistics. Min- United Nations Development Program (UNDP). 1996, istry of Education, Mozambique. 1997. 1998. Cooperation for Development Report. Ministry of Planning and Finance. 1990-1999a. Annual UNDP, Mozambique. Budgets. Mozambique World Bank, 1995a. Priorities and Strategies for Education, A ------1998b. Despesas do Quadro Orçamental de Médio World Bank Review. World Bank, Washington D.C. Prazo por Sectores e sua Comparação com Orçamento de ------1998b. Report and Recommendation under the HIPC 1999. Maputo, Mozambique. Debt Initiative. Report No. P7231-MOZ. Washington Ministry of Planning and Finance, Eduardo Mondlane D.C. University, International Food Policy Research Insti- ------1999c. Education Sector Strategic Program. Project tute. 1998. Understanding Poverty and Well-Being in Appraisal Document. Washington D.C. Mozambique: The First National Assessment (1996-97). ------2000d. Can Africa Claim the 21st Century? Washing- Maputo, Mozambique. ton D.C. Mingat, Alain, and Bruno Suchaut. 2000. "Les Systèmes ------2000e. A Chance to Learn: Knowledge and Financing éducatifs africains: Une analyse économique comparative." for Education in Sub-Saharan Africa. Washington D.C. De Boeck Université, Brussels. 75 76 MOZAMBIQUE: COST AND FINANCING OF EDUCATION ------1998, 1999f. World Development Indicators 1998, 1999. Washington D.C. ------2000g. Mozambique Growth Performance and Reform Agenda. Report No 20601-MOZ. Washington D.C. _____2002h. Constructing Knowledge Societies: New Chal- lenges for Tertiary Education. Washington D.C. 76