32802 Africa Region Human Development Working Paper Series Cost, Financing and School Effectiveness of Education in Malawi A future of limited choices and endless opportunities Development Research Group The World Bank Copyright © August 2004 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 design by Word Express Typography by Word Design, Inc. Cover photo by Rae Galloway ii Contents Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .iii Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .vii Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .x List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xi Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 1. Macroeconomic and Demographic Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12 Evolution in gross domestic product, per capita income and fiscal resources . . . . . . .13 Evolution in inter-sectoral allocation of public expenditure, in education spending, and HIPC resource availability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 HIPC (Highly Indebted Poor Countries) resources situation . . . . . . . . . . . . . . . . .18 2. Enrollment Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21 Structure of Malawi's education system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21 Enrollment trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22 Pre-primary education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23 Basic or primary education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23 Enrollment by age in primary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24 Disparities in enrollment data and sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25 Pattern of student flow through the system, survival, repetition rates and efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26 Completion rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29 Primary School Leaving Certificate (PSLCE) . . . . . . . . . . . . . . . . . . . . . . . . . . . .30 Junior primary schools and distance to schools . . . . . . . . . . . . . . . . . . . . . . . . . .30 Special education provisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31 Transition rates from primary into secondary school . . . . . . . . . . . . . . . . . . . . . .31 Secondary education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31 Enrollment in secondary schools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32 Secondary repetition rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33 Survival/completion in secondary and transition to tertiary . . . . . . . . . . . . . . . . .33 Teacher training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35 Primary teacher training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36 Secondary teacher training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36 Technical and vocational training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37 iii 3. Education Expenditures and Financing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41 The budgetary system in Malawi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41 Overall budgetary expenditure trends in the education sector . . . . . . . . . . . . . . . . . .42 Sub-sectoral allocations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .44 Composition of recurrent expenditures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45 Trends in unit public expenditures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .46 Decomposition of public unit costs in 2000/01 . . . . . . . . . . . . . . . . . . . . . . . . . . . .48 Trends in sub-sectoral emoluments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49 Development expenditures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50 Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50 Donor financing and extra budgetary grants . . . . . . . . . . . . . . . . . . . . . . . . . . . .50 Household financing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51 4. Internal Efficiency and Management Effectiveness in Resource Utilization . . . . . . . . . .55 Efficiency in student flow in primary and secondary education . . . . . . . . . . . . . . . . .55 Index of efficiency and schooling years to produce a graduate . . . . . . . . . . . . . . .55 Teacher deployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .57 Primary school level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .57 Deployment process in primary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .61 Secondary schools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .61 5. School Effectiveness, Malnutrition and Learning Outcomes . . . . . . . . . . . . . . . . . . . . .65 Level of learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .65 The learning environment in primary schools . . . . . . . . . . . . . . . . . . . . . . . . . . . . .67 School and home environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .67 Other factors influencing school effectiveness . . . . . . . . . . . . . . . . . . . . . . . . . . .68 Curriculum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .69 Impact on pupils and teachers of HIV/AIDS . . . . . . . . . . . . . . . . . . . . . . . . . . . .70 Prevalence and consequences of malnutrition affecting education attendance and outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .70 Correlation between schooling characteristics and learning outcomes . . . . . . . . . . . .73 6. Equity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .78 Enrollment patterns by gender and geographic location . . . . . . . . . . . . . . . . . . . . . .78 An overall perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .79 Gender and geographical disparities in schooling profiles . . . . . . . . . . . . . . . . . . .81 Completion rates in primary and secondary . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83 Public expenditure breakdown by quintiles and gender at different education levels .84 7. University Sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .87 Access and equity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .88 Financing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .89 Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .92 Relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .93 iv 8. Critical Challenges and Policy Choices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .96 Macroeconomic context and education expenditures . . . . . . . . . . . . . . . . . . . . . . . .96 Enrollment, repetition and survival . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .97 Equity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .97 Quality of education and externalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .98 Management of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .99 Budget process, supporting systems and harmonization of procedures . . . . . . . . . . .100 University education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .100 Realigning policy targets to resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .101 Changes in repetition and dropout rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .101 Secondary education and secondary teacher training . . . . . . . . . . . . . . . . . . . . . . .102 Recurrent cost implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .103 Possible priority areas of policy intervention . . . . . . . . . . . . . . . . . . . . . . . . . . .103 Annexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .179 List of Tables Table 1: Comparative information on government recurrent education spending as a percentage of government current revenue, excluding grants, and as a percentage of GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Table 2: Comparison of unit recurrent public spending per pupil . . . . . . . . . . . . . . . . . . .3 Table 3: Fourth-graders' scores on Monitoring Learning Achievement (MLA), 1999 (percentage correct) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Table 4: Students per textbook in primary, by standard . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Table 5: Pupil teacher ratios, standard 1 ­ 8, 2000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Table 6: Primary school gross and net enrollment ratios, by sex, according to background characteristics, 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Table 7: Enrollment by type of secondary school, Pupil Teacher Ratio and percentage of qualified teachers, 2000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 Table 8: Possible areas of policy intervention--outcomes (objective) . . . . . . . . . . . . . . . .10 Table 1.1: Social and economic indicators (latest year available) . . . . . . . . . . . . . . . . . . . . .14 Table 1.2: Evolution in population, GDP, and total government revenue and expenditures . .14 Table 1.3: Output and employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15 Table 1.4: Evolution in the dependency ratio, 1990­2000 . . . . . . . . . . . . . . . . . . . . . . . . . .16 Table 1.5: Trends in public spending on education, and comparison to total government spending, 1993/94­2001/02, in percentages . . . . . . . . . . . . . . . . . . . . . . . . . .18 Table 1.6: Comparative information on government recurrent spending on education as a percentage of government current revenue, excluding grants, and as a percentage of GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19 Table 2.1: Distribution of enrollment, by level of schooling, 2000 . . . . . . . . . . . . . . . . . . . .22 Table 2.2: Absolute enrollment numbers and gross enrollment ratios over time . . . . . . . . . .24 Table 2.3: Repetition and dropout rates, EMIS 2000 and DHS 2000 . . . . . . . . . . . . . . . . .27 Table 2.4: PSLCE results and transition to different types of secondary schools . . . . . . . . . .30 Table 2.5: Students with disabilities integrated in primary schools . . . . . . . . . . . . . . . . . . .31 v Table 2.6: Absolute enrollment and gross enrollment ratios in secondary18 . . . . . . . . . . . .32 Table 2.7: MSCE pass rates prior years . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34 Table 2.8: Secondary school transition rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35 Table 2.9: Primary teacher training graduates over time . . . . . . . . . . . . . . . . . . . . . . . . . . .36 Table 2.10: Secondary education teacher graduates, Unima and Domasi . . . . . . . . . . . . . . . .37 Table 3.1: Share of personal emoluments in recurrent expenditures . . . . . . . . . . . . . . . . . . .46 Table 3.2: Per pupil annual recurrent expenditure (in 1993/94 kwacha) . . . . . . . . . . . . . . .47 Table 3.3: Unit costs for a year of education, by level of education, 2000/01 . . . . . . . . . . .47 Table 3.4: Decomposition of public unit costs, 2000/01 . . . . . . . . . . . . . . . . . . . . . . . . . . .48 Table 3.5: Average emoluments per staff (in 1993/94 kwacha) . . . . . . . . . . . . . . . . . . . . . .49 Table 3.6: Growth of teaching staff and pupil­teacher ratios . . . . . . . . . . . . . . . . . . . . . . .50 Table 3.7: Annual household expenditures on primary education, 2001 . . . . . . . . . . . . . . .52 Table 4.1: Index of efficiency in the different cycles of study, 1999/00 . . . . . . . . . . . . . . . . .56 Table 4.2: The relationship between the number of teachers and pupils in public or subvented primary schools, 1999/00 . . . . . . . . . . . . . . . . . . . . . .59 Table 4.3: Regression results with and without using district dummy variables within each of the division subsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .59 Table 4.4: Efficiency ratios by rural and urban districts, 1999 . . . . . . . . . . . . . . . . . . . . . .60 Table 4.5: Pupil teacher ratios, standard 1­8, 2000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .60 Table 5.1: Fourth-graders' scores on Monitoring Learning Achievement (MLA), 1999 (percentage correct) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .66 Table 5.2: Students per textbook in primary, by standard . . . . . . . . . . . . . . . . . . . . . . . . . .67 Table 5.3: Time spent on school and homework in primary by grade . . . . . . . . . . . . . . . . .68 Table 5.4: Number of school days missed by primary school children during the 2001 school year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .68 Table 5.5: Percent of children 0­3 years of age with stunting in the Eastern and Southern Africa region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71 Table 5.6: Correlates of school variables with retention and exam pass rates . . . . . . . . . . .74 Table 6.1: Primary school enrollment ratios according to background characteristics, 200280 Table 6.2: Gross enrollment rates in std. 1­4 versus std. 5­8 . . . . . . . . . . . . . . . . . . . . . . .80 Table 6.3: Secondary school enrollment ratios, by gender, according to background characteristics, 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .81 Table 6.4: Retention standards 1­8 and 1­4 by division and by gender, 2000 . . . . . . . . . . .82 Table 6.5: MSCE and JCE results, 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .84 Table 6.6: Incidence of public education expenditure by level and quintile . . . . . . . . . . . . .84 Table 7.1: Unit cost in USD by college, UNIMA and Mzuzu University, 2000/01 . . . . . . . .90 Table 7.2: University enrollment and lecturer numbers by university and college, 2001 . . . .90 Table 8.1: Primary projections: enrollment, repetition, and PqTR, 2001­2015 . . . . . . . . .101 Table 8.2: Secondary projections: enrollment, PCR, repetition, GER and PqTR, 2001­2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .103 Table 8.3: Possible areas of policy intervention, including assessment of Cost (C), Feasibility (F), Political Risk (P) and Success Probability (S) (low, medium or high) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .104 List of Figures Figure 1: Enrollment by sex and level of education, 2000 . . . . . . . . . . . . . . . . . . . . . . . . . .6 vi Figure 2: MSCE pass rates by type of school, 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 Figure 3: Teachers allocation in public primary schools . . . . . . . . . . . . . . . . . . . . . . . . . . .8 Figure 1.1: Recurrent expenditure by functional classification, in percentages of total . . . . . .17 Figure 1.2: Development expenditure by functional classification, in percentages of total . . .17 Figure 2.1: Structure of Malawi's education system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22 Figure 2.2: Proportion of children who ever enrolled in primary by age . . . . . . . . . . . . . . . .24 Figure 2.3: Enrollment rate by age, different sources, 2000 . . . . . . . . . . . . . . . . . . . . . . . . .25 Figure 2.4: Gross enrollment rates by level of schooling, 2000 . . . . . . . . . . . . . . . . . . . . . . .26 Figure 2.5: Enrollment trends by standard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26 Figure 2.6: Estimated survival rate primary and secondary, 2000 (cross-sectional method) . .28 Figure 2.7: Primary school completion rate over time, two population estimates . . . . . . . . . .29 Figure 2.8: Enrollment in secondary education by type of school . . . . . . . . . . . . . . . . . . . . .33 Figure 2.9: MSCE pass rates by type of school, 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34 Figure 3.1: Real recurrent and development expenditures, 1993-2003 (millions of kwacha) .43 Figure 3.2: Real recurrent expenditures: MOEST and subvented organizations . . . . . . . . . . .43 Figure 3.3: MOEST recurrent expenditures in real terms . . . . . . . . . . . . . . . . . . . . . . . . . . .44 Figure 3.4: Real recurrent expenditures of subvented organizations . . . . . . . . . . . . . . . . . . .44 Figure 3.5: Real development expenditures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45 Figure 4.1: Teachers allocation in publicly funded primary schools . . . . . . . . . . . . . . . . . . . .57 Figure 4.2: Comparison of degree of randomness in teachers allocation to individual primary schools for selected Sub-Saharan African countries . . . . . . . . . . . . . .58 Figure 4.3: Teacher allocation in all CSS, CDSS, and grant-aided secondary schools . . . . . .62 Figure 4.4: Percentage of qualified teachers in different types of secondary schools . . . . . . . .62 Figure 5.1: SACMEQ test score, Southern and Eastern Africa . . . . . . . . . . . . . . . . . . . . . . .66 Figure 6.1: Enrollment by gender and level of education, 2000 . . . . . . . . . . . . . . . . . . . . . .79 Figure 6.2: Survival in primary, urban9 vs. rural (cross-sectional method) . . . . . . . . . . . . . .82 Figure 6.3: Proxy std. 4 and std. 8 completion rates by gender . . . . . . . . . . . . . . . . . . . . . . .83 Figure 7.1: Enrollment in tertiary per 100,000 inhabitants, Sub-Saharan Africa . . . . . . . . . .88 Figure 7.2: University enrollment by subject area, 2001 . . . . . . . . . . . . . . . . . . . . . . . . . . . .88 Figure 7.3: Student to lecturer ratios at tertiary level in Sub-Saharan Africa, 2000/01 . . . . . .92 vii Foreword T his study is part of a series of educa- arrangements under the Heavily Indebted Poor tion country status reports (CSRs) Countries (HIPC) Initiative. Governments have being prepared by World Bank staff also invariably treated education as a priority in collaboration with national teams sector in their Poverty Reduction Strategy from various Sub-Saharan countries. The Papers (PRSPs). The case for placing addition- immediate objective is to enhance the knowl- al emphasis on the education sector is particu- edge base for policy development. More larly strong in Malawi, where prospects for broadly, CSRs create a basis for engaging a economic growth depend on the ability to diverse audience in dialogue on education sec- diversify the economy in a country with a lim- tor policies and for developing a shared vision ited natural resource base and declining terms for the future. These processes have become of trade affecting traditional sectors. There is increasingly important as governments renew now a need to focus on learning outcomes and their commitment to reduce poverty, and the quality, which have not kept pace with the con- international donor community pledges to pro- siderable improvement of access to education vide the needed financial assistance. achieved following the 1994 Free Primary Edu- In this new dynamic of international devel- cation Policy that increased enrollment from 2 opment assistance, the World Bank has orient- million to 3 million pupils overnight. ed its corporate mission toward supporting In Malawi, as in other low-income coun- governments in the fight against poverty, mak- tries, two specific goals in education stand out: ing the task its defining priority. In practice the ensuring that all children complete basic strategy is articulated through a two-prong schooling of adequate quality; and managing approach: support for policies that accelerate the production of graduates at post-basic lev- economic growth combined with explicit els, in terms of quantity and skill level and mix, measures to promote a more equitable distri- to match the demand for skilled labor. To ful- bution of the benefits from growth. Human fill these objectives, an infusion of resources, development, especially through investments in from both domestic and international sources, education, plays a central role in both process- may well be needed. However, an equally es. This fact receives recognition in the empha- important requirement is to ensure that cur- sis placed on education in the debt relief rently available resources are used efficiently ix and equitably. The evidence suggests that in ness of the statistical results. Third, the CSRs many Sub-Saharan countries, substantial scope put a sharper focus on outcomes by emphasiz- for progress exists in this regard. Some coun- ing indicators beyond the usual gross and net tries have indeed already embarked on reforms enrollment ratios. In particular, it documents for improving the performance of their educa- the schooling careers of children as reflected in tion systems. In the 1990s, for example, poli- the shares of the population that enter grade 1 cies to encourage more effective use of teachers and attain the various grades in the education- through multi-grade teaching and double shift- al ladder. The more detailed approach has ing were put in place to boost coverage. More helped to sharpen the socio-economic, gender remains to be done, particularly to tackle the and geographical disparities in education, as structural constraints that encumber the edu- well as to clarify the sources of the disparities. cation system. The challenges include defining A fourth example of the technical content of an appropriate division of responsibility CSRs is the use of school-level data to assess between the public and private sectors for the scope for improving service delivery to the financing and delivering education services, poor. Finally, the CSRs also make use of data setting sustainable levels of teacher remunera- on student learning--where they are avail- tion, and creating institutional arrangements able--to gauge the education system's per- for effective systems management. Reform in formance in this important domain, and to these areas are key to improving the quality of identify cost-effective measures for progress. learning outcomes, which in turn will be cru- Regarding the process for preparing the cial for progress toward the Education for All CSRs, the main feature is that they are prod- targets. ucts of a partnership between the World Bank In order to identify appropriate measures and national teams from the various countries. and address the structural issues, a first step is The form of this partnership varies across to develop a country-specific knowledge base countries depending on the specific circum- that sheds light on the key weaknesses in the stances. In the case of Malawi, the CSR was education system. In a large number of the developed in close collaboration with the Min- Sub-Saharan countries, however, that knowl- istry of Education and Human Resources (for- edge base remains sparse, reflecting the sys- merly the Ministry of Education, Science and tematic neglect of analytical work in the past. Technology) and other Government agencies. In order to re-capitalize it, the World Bank has It is important to note that CSRs are diagnos- initiated the preparation of CSRs, using a stan- tic documents whose purpose is to help identi- dard format to consolidate the available infor- fy the policy questions rather than to offer mation in a policy-relevant manner. solutions and make recommendations. The Two features characterize the CSRs, one process of policy development is more appro- concerning their technical content, the other priately led by the national team, and the avail- the process by which they are accomplished. ability of a CSR provides a good basis for dis- With regard to content, five aspects are worth seminating the findings and stimulating a mentioning. First, the CSRs pay close attention broad national dialogue that informs the coun- to issued of equity and the distribution of pub- try's medium-term budget planning exercise. In lic resources for education, given the impor- the case of Malawi, the CSR has been exten- tance of these topics in the HIPC and PRSP sively used to target interventions under the context. Second, the analysis relies on com- Education Sector Support Project 1 (ESSUP1) monly available administrative data as well as and to further a broad based dialogue with all household surveys, an approach that has donors and stakeholders. Further, the CSR can helped to improve the consistency and robust- play a useful guiding role in the Government's x preparation of an education sector plan for It is my hope that as new knowledge emerges Malawi, which will set the directions for in the course of implementing the country's reforms and investments in the sector in com- development strategy, the CSR will be updated ing years. For the World Bank's part, the docu- to track progress in overcoming those con- ment is helping to facilitate ongoing discus- straints in the education sector that currently sions to improve the management of the impede poverty reduction in Malawi and system, its internal efficiency, and most impor- which may in the long run slow economic tantly, to improve quality. growth. The publication of the CSR for Malawi is intended to institutionalize our collective Birger Fredriksen knowledge about the country's education sec- Senior Education Advisor tor and the nature of the policy challenges, and Africa Region to share that knowledge as widely as possible. The World Bank xi Acknowledgements T his report has been prepared Human Resources (formerly the Ministry of through a process of close collabora- Education, Science and Technology), led by tion between the World Bank and Honorable Minister Y. Mwawa and included the Government of the Republic of among others Z.D. Chikhosi, Principal Secre- Malawi to collect and analyze information tary, K. Mwale, Director of Planning, Augus- about the status of the education system in tine Kamlongera, Grace Banda, Ken Ndala, Malawi. A large number of people have con- and Sutapa Choudhury. The report further tributed to the study, which was conducted benefited from important contributions by under the general supervision of Søren Nelle- Marelize Görgens from the EDMU, Esme mann. The Bank team included also Bart Van Katzamira from CERT, Simeon Hand and staff Uythem, Sajitha Bashir, Cathal Higgins, from MIE, and Wilfred Görtler from GTZ, Michael Mambo, Rae Galloway, Kirsten Maj- and from representatives of other parts of gov- gaard and Yves Tencalla. Technical guidance ernment, including Ministry of Finance and was provided by Alain Mingat, Ramahatra Economic Planning, Ministry of Labor and Rakotomalala, and Jee-Peng Tan. Comments Vocational Training, TEVET, National Statis- and support were provided by Suhdir Chitale, tics Office, the University of Malawi and Antonio Nucifora, Dandan Chen, and Birger Mzuzu University. Fredriksen. Peer reviewers were Vincent Gre- Helpful comments and information were aney, Harry Patrinos, and Lianqin Wang. Trish received from a number of donors in Malawi, Tierney has provided editorial assistance and including Dfid, UNICEF, JICA, UNDP, CIDA, Elsie Lauretta Maka has overseen the produc- USAID and WFP. The report was financed by tion of the report. the World Bank, with contributions from the The Malawian technical working group governments of Malawi and Norway through consisted of senior policy makers and technical the Norwegian Education Trust Fund. staff from the Ministry of Education and xii List of Abbreviations CDSS Community Day Secondary School CE Central East Division CIDA Canadian International Development Agency CSS Conventional Secondary School CW Central West Division DANIDA Danish International Development Assistance DEC Distance Education Centre DFID Department for International Development DSTD Department for School and Teacher Development (at MIE) DTED Department of Teacher Education and Development EFA (­ FTI) Education For All (- Fast Track Initiative) EMAS Education Methods Advisory Service EMIS Education Management Information System FPE Free Primary Education GDP Gross Domestic Product GER Gross Enrollment Ratio GNP Gross National Product GoM Government of Malawi HRMD Human Resource Management and Development IHS Integrated Household Survey JCE Junior Certificate of Education JICA Japan International Cooperation Agency KFW Kreditanstalt fur Wiederaufbau (Germany) MANEB Malawi National Examination Board MASAF Malawi Social Action Fund MCDE Malawi College of Distance Education MIE Malawi Institute of Education MIITEP Malawi Integrated In-service Teacher Education Program MK Malawi Kwacha xiii MLA Monitoring Learning Achievement MOEST Ministry of Education, Science and Technology (now Ministry of Education and Human Resources) MOF Ministry of Finance MSCE Malawi School Certificate of Education MTEF Medium Term Expenditure Framework MU Mzuzu University NER Net Enrollment Ratio NLS National Library Service ORT Other Recurrent Transactions PE Personal Emoluments PEP Primary Education Project PER Public Expenditure Review PIF Policy and Investment Framework (strategic document for the education sector in Malawi, developed by the Ministry of Education, outlining the goals by 2012) Poly Board Polytechnic Board of Governors PSLCE Primary School Leaving Certificate PRISAM Private Schools Association of Malawi PTA Parent Teacher Association PTR Pupil Teacher Ratio PqTR Pupil Qualified Teacher Ratio SACMEQ Southern African Consortium for Monitoring Educational Quality SE South East Division SEP Secondary Education Project SH Shire Highlands Division SSTEP Secondary School Teacher Education Project SW South West Division TDC Teacher Development Centre TSC Teacher Service Commission TTC Teacher Training College T&L Teaching and Learning (materials) UOM / Unima University of Malawi USAID United States Agency for International Development GOVERNMENT FISCAL YEAR July 1--June 30 CURRENCY EQUIVALENTS (Exchange Rate effective October 29, 2003) Currency Unit: Malawi Kwacha (MK) US$ 1.00 = 105.5 MK xiv Africa Region Africa Region Human Development Working Paper Series Cost, Financing and School Effectiveness of Education in Malawi A future of limited choices and endless opportunities xv Executive Summary Introduction chronic malnourishment has resulted in very high repetition and dropout rates. Despite the E ducation in Malawi is experiencing a fact that more than 95 percent of all Malawian serious and debilitating crisis of low children eventually begin school, only about quality, low efficiency, limited capaci- 60 percent complete standard 6 and some 4 ty and a shortage and inequitable dis- out of 10 pupils (40 percent) complete stan- tribution of human and financial resources. dard 8. The education environment is further under- As a successful outcome of the 1994 imple- mined by a very high pupil qualified teacher mentation of Free Primary Education (FPE)-- ratio (PqTR) in primary as a result of a short- which increased enrollment from 2 million to 3 age of trained teachers and the limited capaci- million -- most Malawian children, including ty of the teacher training system--approxi- the poorest, now enroll in school. While the mately 40 percent of primary and 70 percent of government has maintained spending per secondary teachers are untrained; inadequate pupil, regrettably the quality of education has learning materials and school infrastructure; been increasingly eroded. Learning outcomes overcrowded classrooms; and a teacher are negligible: the Southern Africa Consortium deployment system whereby teacher numbers for the Measurement of Educational Quality vary significantly across schools and grade lev- (SACMEQ) study revealed that less than 1 per- els and between urban and rural areas. To cent of pupils in standard 6 reach a desirable compound the problem, the HIV/AIDS pan- level in reading (20 percent reach a minimum demic is killing teachers and parents, raising level). Given the limited learning outcomes and absenteeism rates of both teachers and pupils the low grade attainment, the wastage of because of illnesses and the need for more help resources is evident and the room for efficien- in the household. With more than half the cy improvements enormous. population chronically malnourished, especial- Against this picture a stark reality confronts ly during the hunger period from January to Malawi: it is a small country with only 10 mil- April, absentee rates are further raised and lion people and with a natural resource base learning outcomes reduced. The poor learning that is limited to agriculture but highly depend- conditions, the HIV/AIDS pandemic and ent on the export of tobacco. The reality is that 1 2 Cost, Financing and School Effectiveness of Education in Malawi in the next decades the Malawian economy is One notable feature in Malawi is the fact likely to remain characterized by a dualistic that the share of recurrent expenditures allo- structure of employment, with a dominant tra- cated to higher education (15­20 percent) is ditional sector and a relatively small modern higher than the share allocated to secondary sector. The basic objective is to provide the education (10­15 percent). This is significant type of human capital that best fits the because approximately 275,000 pupils are demands of the two sub-sectors, while simulta- enrolled in secondary education while only neously helping the economy to move from its about 4,000 students are enrolled in higher traditional to its modern sector and alleviating education institutions. poverty, in the long run. Per Pupil Spending. The per-student costs of secondary and higher education are more expensive than primary education in most coun- Macroeconomic Context and tries. In Malawi, however, the discrepancies in Education Expenditures spending per pupil are so substantial that they raise concerns about the equitable distribution The development of education is determined of resources. As can be seen in Table 2, below, largely by macroeconomic and demographic the unit cost at the university level is between conditions. Throughout the 1990s, Malawi's US$2,500­3,000, while it is around US$10 in economy grew by approximately 3 percent--or primary compared to around US$35 in other slightly less than 1 percent above population SSA countries, and roughly US$40 in secondary growth. GDP per capita was approximately as a whole. The latter is affected by the very low US$175 in year 2001 and Malawi remains one unit costs (approximately US$20) and under of the poorest countries in the world. Total pub- funded Community Day Secondary Schools lic expenditures as a share of GDP rose from (CDSS) where the bulk of secondary students about 25 to 33 percent over the same period. are enrolled. In comparison, the spending per Total education expenditure in real terms has student in the traditional Conventional Sec- remained largely the same since 1994/95. As ondary Schools (CSS) is approximately US$100. Table 1 shows, by the end of the nineties, recur- As Table 2 shows, compared to other countries rent education expenditure as a proportion of in the region the unit cost for secondary and ter- GDP in Malawi was somewhat below countries tiary education appear very unbalanced. like Uganda (3.2%) and Kenya (6.3%)--coun- Household Financing. Despite the abolish- tries that are also expanding enrollment as a ment of primary school tuition fees in 1994, result of a free primary education policy. How- households still bear a considerable share of ever, in the past couple of years, recurrent education financing. Nearly all households in spending on educations has recovered in Malawi paid for one or more types of school Malawi, reaching 3.6% of GDP in 2001/02, the supplies during the year 2001. The majority of last year for which official data are available. families reported paying for textbooks and Table 1: Comparative information on government recurrent education spending as a percentage of government current revenue, excluding grants, and as a percentage of GDP Malawi Mozamb Kenya Rwanda Uganda Zambia Zimbabwe African 1998/99 2000/01 (1998) (1999) (2000) (2000) (1998) (1997) Countries Educ. Recurr. exp. as % of gov. curr. Rev. 17.3 15.0 18.1 26.2 32.6 30.1 12.3 28.3 17.8 Educ. recurr. exp. as % of GDP 2.5 2.8 2.0 6.3 3.2 3.2 2.3 7.1 3.04 Source: Bruns, B. et al. 2002. For Malasi, authors' calculations are used. Executive Summary 3 Table 2: Comparison of unit recurrent public spending per pupil Ratio to Primary As proportion of GDP/capita Mozam- Anglophone Mozam- Anglophone Malawi bique Kenya Zambia Africa* Malawi bique Kenya Zambia Africa* Primary 1 1 1 1 1 0.04 0.07 0.12 0.04 0.10 Secondary 5 6 3 9 7 0.21 0.40 0.34 0.37 0.66 University 348 92 33 58 63 14.9 6.4 4.0 2.3 6.3 * Group of 7 countries. Source: World Bank Sector Studies. uniforms and approximately half reported con- head quarters, the six education division tributing to the school development fund. In offices and the 33 district education offices. As primary, mean household expenditures on edu- a result of the FPE policy, the total number of cation average approximately 80 percent of students enrolled in primary education grew per pupil public expenditures. These household almost overnight from less than 2.0 million in expenditures are regressive and the poorest 1993/94 to almost 3.0 million in 1994/95 and households contribute more of their income to to about 3.2 million in 2001. About 70 percent education than the wealthiest Malawians. Fur- or 2.3 million are enrolled in lower primary thermore, household spending on education (standards 1 to 4). There were an estimated increases by grade level. A student in grade 3 275,000 students in secondary in 2000, 12,000 incurs 50 percent more costs than a student in students at primary teacher training colleges, standard 1 and one in standard 6 spends 4.7 4,000 students at universities, and about 1,500 times the expenditure of a standard 1 student. students were attending technical and voca- Given the high levels of poverty in the country, tional training institutes. and the susceptibility of agricultural and rural Concurrently, the national gross enrollment incomes to climatic shocks, this relatively high rate (GER) increased from 88 percent in dependence on household financing introduces 1991/92 to 120 percent in 1994/95 and has greater instability--especially in the financing remained at about that level since. Enrollment of critical learning materials, repairs and con- in secondary multiplied by almost four between struction costs. The high levels of household 1991/92 and 2000, with the GER now at about expenditures are not only regressive but are 27 percent. Malawi has 6 primary Teacher possibly important determinants in forcing stu- Training Colleges (TTC) and approximately dents to drop out in the early stages of primary 50,000 primary teachers, although some may education. be ghost teachers. However, because of the expansion there continues to be an acute short- age of qualified primary and secondary teach- Enrollment Rates and Impact of ers. In secondary, even though the student to HIV/AIDS teacher ratio (PTR) is at about the desired 40:1, the PqTR is 120:1. The total stock of secondary The formal education system in Malawi fol- teachers is about 7,500, of which less than a lows an 8­4­4 structure, i.e. 8 years of primary third are qualified secondary teachers. (Standard 1­8), 4 years of secondary (Form Impact of HIV/AIDS. In 2001 the official 1­4) and 4 years, or more, of university level number of primary teachers that died was 900. education. The primary and secondary levels In addition about 700 left the primary educa- are administered by the Ministry of Education, tion system owing to normal attrition. Because Science and Technology (MoEST)1 through its HIV/AIDS will continue to affect the mortality 4 Cost, Financing and School Effectiveness of Education in Malawi rate of teachers and the system will continue to the weakest. Performance in Life Skills test is expand, the replacement of both primary and best of all countries and this balances the weak secondary teachers as a result of deaths caused results in the other two areas. by AIDS will likely increase to approximately The low literacy results are confirmed by the 3,700 teachers a year. HIV/AIDS will also SACMEQ study testing the reading skills in increasingly affect attendance and retention of English of standard 6 pupils. The acquisition both teachers and students. The adverse effects of minimum and desirable mastery levels was on the annual training and replacement of extremely poor, with less than one percent of teachers, as well as on efforts to improve reten- all measured pupils reaching the desirable mas- tion and access of growing numbers of children tery level while about 22 percent reached the -- who will be orphaned or required to take minimum mastery level. The study also found care of their parents -- has significant implica- that Malawi had the lowest percentage of tions for achieving universal primary education pupils who met the minimum and desirable in Malawi. mastery levels of reading literacy when com- pared to six other countries in the region. The implication for Malawi is the need to address Quality of Education some of the underlying structural problems to improve learning outcomes. Some of these The combination of relatively low public structural problems relate to the curriculum spending on education per pupil, inadequate being taught, teachers qualification and years numbers of qualified teachers, poor learning of experience, school practice and the subject conditions and externalities affecting both content knowledge and pedagogic content teachers, parents and children has resulted in a knowledge being taught under the current in- deterioration in learning outcomes. Table 3 service teacher training program. Factors shows the performance of Malawi's fourth exogenous to the school are also important, graders in a test of numeracy, literacy and Life such as mother tongue, availability of books at Skills administered in a number of African home, homework, and nutrition. More than countries. Malawi's performance is slightly half the population is chronically malnour- below the average of all the sampled countries. ished. School inputs, such as teaching and However, for literacy its score is the lowest of learning materials, classroom materials, and all countries and for numeracy it ranks close to school furniture are inadequate or dilapidated Table 3: Fourth-graders' scores on Monitoring Learning Achievement (MLA), 1999 (percentage correct) Country Life skills Literacy Numeracy Average score Botswana 56.0 48 0 51.0 51.7 Madagascar 72.1 54.7 43.7 56.8 Malawi 77.0 35.0 43.0 51.7 Mali 56.9 51.8 43.6 50.8 Mauritius 58.0 61.0 58.5 59.2 Morocco 62.3 67.6 56.4 62.1 Niger 47.7 41.1 37.3 42.0 Senegal 46.7 48.9 39.7 45.1 Tunisia 74.7 77.9 60.4 71.0 Uganda 66.8 58.0 49.3 58.0 Zambia 51.0 43.0 36.0 43.3 Average 60.8 53.4 47.2 53.8 Source: UNESCO (1996). Executive Summary 5 Table 4: Students per textbook in primary, by standard Subject Std. 1 Std. 2 Std. 3 Std. 4 Std. 5 Std. 6 Std. 7 Std. 8 English 2 2 3 3 2 2 1 1 Math 2 2 3 3 2 2 1 1 Chichewa 2 3 3 3 2 2 2 1 Source: Draft Basic Education Statistics 2001. adversely affecting the conditions for learning. remains elusive as a result of the low survival In many schools the lower standards are taught throughout primary. Although more than 95 under the open sky, which often leads to the percent of all children enter primary at some cancellation of classes during the rainy season. point most repeat several grades. The level of Over one third of standard 6 pupils do not repetition is high throughout primary, and is have an English textbook. Overall, MoEST significantly higher in lower primary than in records the pupil to textbook ratio for three upper primary, with the exception of standard major subjects as shown in Table 4. 8. The average is about 25 percent, while 40 Finally, as shown in Table 5, the discrepan- percent of all children repeat standard 1. The cy in the PTR between lower grades and high- Monitoring Learning Achievement (MLA, er grades further complicates the conditions 1999) study for 11 African countries recorded for effective teaching and learning especially in the highest repetition rates of all countries in the lower grades. As can be seen the average Malawi in the first four grades of primary. PTR in standard one is 100, but many schools The high number of repeaters and over-age and classes exceed this number. There are clear pupils has increased the GER. The GER for implications for effective teaching. In addition, standard 1­4 is 167 percent (1999/2000), but the high PTR increases the number of repeaters is only 70 percent for standard 5­8. The low as pupils are not being helped, and this in turn GER in upper primary reflects the fact that a leads to a continuing high number of students large proportion of the children eventually enrolled in the lower grades. The high PTR drop out. While repetition levels are high in the becomes both a cause and effect of poor learn- early school years, dropout rates are modest in ing outcomes and low retention. the lower grades, but by the end of standard 6, about 40% of the pupils have dropped out, and at the end of standard 8, about 60%. Repetition, Dropout, Survival and The survival rate, or probability that a grade Completion 1 student will still be in school by standard 8, has actually diminished since the beginning of The poor learning environment leads to low the nineties prior to the introduction of FPE. levels of retention, which in turn adversely However, due to a higher intake of students to affect achievement. While Malawi, despite its primary school than before, more children in weak school infrastructure and lack of suffi- absolute numbers and as a percentage of school- cient classrooms, has almost succeeded in pro- age children, do complete primary. Thus, viding access for all children with the abolition Malawi is on the right track by having hugely of fees, universal primary education for all improved primary coverage and to a lesser Table 5: Pupil teacher ratios, standard 1 ­ 8, 2000 Std 1 Std 2 Std 3 Std 4 Std 5 Std 6 Std 7 Std 8 100 76 64 52 45 39 33 28 6 Cost, Financing and School Effectiveness of Education in Malawi Figure 1: Enrollment by gender and level of education, 2000 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% Primary Standard Standards Secondary Primary Higher 1-4 5-8 Teacher education Boys Girls Training Source: Authors' calculations, using data from NANEB database. extent completion, but needs now to focus on gap widens as students progress. In primary reducing dropout and repetition in primary. the percentage of girls was 48 percent in year In secondary education approximately 60 2000, 40 percent for secondary, 35 percent for percent of students enrolled in Form 1 proceed primary teacher training and 26 percent for to Form 4 (grade 12). Repetition, and early higher education. While the survival rates for and late entry, further impact on pedagogical both boys and girls are about the same, there is processes because of the broad range of skills a significant discrepancy of about 20 to 30 per- and maturity in each class. By addressing the centage points between the survival of an wide age-range of pupils, the system could urban versus a rural pupil.While there are improve efficiency, lower repetition and many reasons for this gap--including a better dropout rates and thus increase survival and pupil to teacher ratio in urban areas--it is also completion rates. However, these are not the related to a higher degree of poverty in the only determinants affecting attainment in rural areas. In general, as shown in Table 6, Malawi. Other issues like gender, geographic poverty is a strong predictor of enrollment and location and type of school determines the lev- survival--with the poorest segment of the pop- els of achievement and attainment. ulation having close to a 20-percentage point lower net enrollment rate than the wealthiest. At the secondary level, urban youth, aged Equity 14­17, are three times more likely than their peers in rural areas to attend secondary school Enrollment and access to education in the (NER 21 versus 7 percent). Regional differ- lower grades in Malawi is fairly gender-equi- ences in the secondary NER are also consider- table. However, as Figure 1 shows the gender able, and follow a pattern similar to that at the Table 6: Primary school gross and net enrollment ratios, by sex, according to background characteristics, 2002 Residence Region Wealth index quintile Urban Rural North Central South Lowest 2nd 3rd 4th Highest Total NER 90 80 93 84 76 73 74 81 85 91 81 GER 127 113 129 119 108 106 106 116 118.5 126 114.5 Results are for the de jure household population age 5­24. Executive Summary 7 Figure 2: MSCE pass rates by type of school, 2002 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% e d Day Day Aide Total Privat School Boarding Gov Night Grant Gov Community Source: Authors' calculations, using data from NANEB database. primary level. Differences in access by socio- schools, it is clear that the majority of students economic background at the secondary level in CDSS do not have access to sufficient number are large, with a GER of 62 percent among the of qualified teachers. wealthiest youth and 11 percent among the poorest youth. In secondary education a particular inequity Internal Efficiency arises from the type of school in which a pupil enrolls. Figure 2 shows the results of the Malawi Given the high repetition and dropout rates in School Certificate of Education (MSCE) final primary, 60 percent of the public resources at exam at the end of secondary in 2002 by type of this level are used either on children who drop enrollment. While the figure may indicate that out before finishing primary or on children with the exception of the Community Day Sec- repeating a year. Put another way, Government ondary Schools (CDSS) outcomes were fairly finances 20 school years to generate a single even, the fact is that close to 75 percent of stu- graduate from primary school. Thus the cost to dents are enrolled in CDSSs. The low pass rate government is an additional 12 years of per in the CDSSs is not surprising since, in general, pupil spending relative to a scenario with no the CDSSs are under-funded and staffed by dropout or repetition (8 years). This inefficient unqualified teachers (and the expenditures per use of scarce resources should be a matter of student are about one-fifth of the expenditures serious concern. For secondary, the situation is in CSS). The inequitable access to good second- significantly better: a little less than one quar- ary education and the acute shortage of trained ter of the public resources are used on repeti- secondary teachers become evident when the tion years or on children who drop out: it enrollment by type of school, PTR and percent- requires 5.2 schooling years instead of 4 years age of qualified teachers are considered as to produce a graduate. shown in Table 7 below. While there is no major Teacher deployment and teacher manage- difference in the PTR ratio across the types of ment. The quality of an education system is Table 7: Enrollment by type of secondary school, Pupil Teacher Ratio and percentage of qualified teachers, 2000 Type of schools Number of students Pupil­teacher ratio Percentage Qualified Teachers CDSS 186,415 46 45% CSS 57,624 49 88% Private 27,793 41 86% Night schools 3,117 20 54% Source: Basic Education Statistics, 2000. 8 Cost, Financing and School Effectiveness of Education in Malawi Figure 3: Teachers allocation in public primary ratios between urban and rural districts. In schools urban districts, the PqTR is between 60 and 50 70, in rural districts it is on average twice as 40 high. Inconsistency in teacher deployment is also evident within individual districts. In sec- 30 teachers of ondary, teacher deployment shows an even 20 lower consistency than primary; the degree of 10 Number randomness is 59 percent. When reviewing 0 pupil to teacher ratios and pupil-to-qualified 0 500 1000 1500 2000 2500 teacher ratios, strong differences are also Number of pupils apparent. In urban areas, PTRs and PqTRs are Source: Authors' calculations, using EMIS 2000. respectively 28:1 and 59:1 whereas the same indicators for rural areas are 36:1 and 131:1. generally associated with the provision of qual- By type of school, two thirds of CDSS teachers ified teachers to that system. Therefore, effi- were unqualified to teach secondary but this ciency concerns are related to the way this crit- was only true for less than 10 percent in CSSs. ical resource is distributed and managed. Since Among the countries in Sub-Saharan Africa teachers' salaries are the largest education Malawi is one of countries with the least con- expenditure item, their deployment also has an sistent deployment of primary teachers. A impact on unit cost and on the ensuing equi- more equitable deployment of teachers will be table distribution of resources. critical to improving attainment and achieve- A crucial aspect of teacher deployment has ment and more efficient use of scare resources. to do with clear criteria to provide schools of It is clear that the Government's ability to similar enrollments with comparable numbers address some of the internal constraints and of teachers. This feature can be represented by efficiency issues will determine how well a straight line graph (policy), as shown in Fig- improvement to the attainment and achieve- ure 3, and in the case of Malawi, a primary ment can be made. However, there are other school of 500 pupils would always have 8 and serious issues affecting the demand for teachers and a school with 1,000 pupils would education and the delivery of education always have 16 teachers. Figure 3 indicates services. that in Malawi there is a wide variability in the number of teachers available to schools that enroll a similar number of pupils. The data Externalities indicate that schools with 1,000 pupils can have anywhere between 5 to 40 teachers. Sim- In Malawi externalities plays a significant role ilarly, in schools with 40 teachers enrollments in the demand and supply of education servic- vary between 500 and 3,800. These data cor- es. Absenteeism is a particular contributing responds to 34 percent `randomness'. The factor to low learning. Ninety-seven percent of inconsistency in teacher deployment can partly primary pupils were absent one or more days be explained by differences in allocations during the 2001 school year. On average, among the districts, especially between urban pupils who were absent from school missed a and rural districts and inadequate monitoring total of 17 days during the year. The main rea- and management tools to ensure adequate and sons for absenteeism during the previous fair deployment. school year were: illness--86 percent; funer- This situation is further aggravated by sig- als--61 percent; did not want to go--39 per- nificant differences in pupil to qualified teacher cent; school clothes dirty--34 percent; too Executive Summary 9 hungry--28 percent; any type of work--12.5 students per 100,000 inhabitants in the entire percent; no money for fees--10 percent. Sub-Saharan region. Public universities in Teacher absenteeism in Malawi is reported as Malawi enroll fewer than 4,000 students. The chronic but the frequency varies widely. In first private university plans an initial intake of some schools, a teacher's time-on-task was as 50 students but it is yet to open. Further, there low as 1­2 hours each day. The main reasons are significant inefficiencies at the universi- suggested for this level of absenteeism were ties--some of these derive from the small scale attendance at funerals of colleagues, errands, of operation. Both the student to lecturer ratio collecting salaries or just "out for a moment". (7:1), as well as the student to support staff Geographic location of schools also presents a ratio (about 9:1) are the lowest in SSA with the challenge. While 64 percent of children in exception of Equatorial Guinea. Other clear urban areas live within 15 minutes of the near- inefficiencies stem from the large proportion of est school, this is only true for 33 percent of budget allocated to non-core functions, such as rural children. In more than one-third of rural boarding and feeding. In fact, the current Edu- areas children are required to walk for at least cation Act still limits total enrollment to avail- half an hour to school. able bed space at the colleges although some Prevalence and consequences of malnutri- colleges are beginning to ignore this require- tion affect the opportunity costs of education. ment. A very minor portion of total enrollment Malawi has one of the highest levels of stunt- includes non-residential students in parallel ing in children in Eastern and Southern Africa; programs that charge higher fees. Apart from almost half the children under 3 years of age gender inequity, a more significant but not sur- are stunted. From the Demographic and prising inequity is the socio-economic back- Health Survey (DHS) 2000, it is apparent that ground of students. Students from the wealthi- many children enter standard 1 either short or est quintile consume more than half of the underweight for their age--57 percent of chil- public expenditure at tertiary level; students dren 4­5 years of age are stunted and an from the poorest quintile have practically no alarming 28 percent of children are severely access whatsoever. stunted. Stunting and micronutrient malnutri- tion are associated with reduced capacity to learn, delayed school enrollment, decreased Policy Issues and Choices school attendance, and dropout--especially during the hungry period from November to The development of the education system in April. Families keep children out of school dur- Malawi will require major investments at all ing this period so children can earn money to levels and in all the critical areas. The imple- buy food or look for food. In addition, hunger mentation of FPE without sufficient public and fatigue are so severe that many children do funding has undermined the quality of the sys- not have sufficient energy to walk to school. tem beyond what could have been expected as a result of the focus on access. Malawi's recur- rent expenditure on education has been rela- University Education tively low compared to other countries that have undertaken FPE policies, such as Kenya The expansion in primary education and the and Uganda, but has been increasing the last shortage of trained teachers is further con- few years. Given the limited prospects for eco- strained by the limi0ted number of students nomic growth, improvement of efficiency will enrolled in higher education. Malawi has close be the key for the overall sector development. to the lowest proportion of tertiary enrolled A few policy issues derived from the study are 10 Cost, Financing and School Effectiveness of Education in Malawi summarized in the Table 8 below. A more com- a more realistic scenario), will require increas- prehensive list of policy issues, their timing, ing spending on education from a share of cost, feasibility, political risk and probability about 21% of the discretionary budget to for successful implementation are listed in 27%; and furthermore, require increasing the Chapter 8. share towards primary education from approx- Finally, in Chapter 8, this report analyzes imately 50 percent to 70 percent of total edu- the fiscal implications of the required upgrad- cation spending. This result strongly implies a ing of primary and secondary teachers needed need to focus now on improving efficiency in to improve the pupil to qualified teacher ratio, the use of resources, and recalibrate current on of the principal targets set in the govern- policy targets to available resources. The ment's Policy and Investment Framework for scarcity of resources must be taken into Education (PIF). This analysis is based on a account in the development of a comprehen- simulation model developed in cooperation sive and sustainable teacher development strat- with government; the model includes assump- egy: the backbone for improving the education tions on the gradual improvement in repetition system in Malawi. and dropout as a result of having more and better teachers. As can be seen in Chapter 8, a scenario for teacher training that will lead to Note the accomplishment of the EFA and PIF goals a 1. From 2004, the Ministry of Education few years after the targets set in the PIF (thus, and Human Resources. Table 8: Possible areas of policy intervention--outcomes (objective) Outcome (Objective) Policy Options/Interventions Re-balanced education · Recalibrate PIF (EFA) goals to financial and human resources, especially system established teacher development · Take a strategic long term view and establish % caps on public funding for secondary and tertiary education Accountability improved: Re-establish accounta- · Establish accountability at school level though social contracting between bility, and establish communities and schools and transparency systems incentives for teachers · Ensure teachers accept strategies/public consensus building and headmasters to · Restructure teacher salary provisions/ Provide career development and rotation, perform their duties especially for rural assignments/ Empower headmasters (role in promotions and in reprimanding teachers) · Improve supervision and inspection Equitable expansion of · Lower repetition and dropout rates to shorten average of school completion enrollment improved (automatic promotion/cycles of learning standard 1­4) · Target districts, areas or groups with inequitable access or survival, using incentives. e.g. Expand food for education of girls during the hungry season (November to April) · Use teachers efficiently, re-deploy teachers; evaluate and adjust accordingly the teacher resource requirements under a streamlined curriculum at secondary level · Improve incentives for private provision of education, particularly at secondary level (continued on next page) Executive Summary 11 Table 8 (continued) Outcome (Objective) Policy Options/Interventions Quality and internal · Redefine a coherent and sustainable teacher development model and strategy/ efficiency improvement Strengthen in-service teacher training and other teacher professional support · Lowering of PTR in standards 1­4 and redeployment of teachers as an integral part of the decentralization plans · Curriculum Reform: Continue the Primary Curriculum and Assessment (PCAR) process and complete the proposed review of secondary curriculum; · Improvements in the classroom conditions, other school facilities and learning materi- als through a direct support to school program (DSS) financed under the development budget/Improve community involvement and through the DSS program lower indirect cost of education/opportunity costs · Distribute iron supplements and support de-worming · Continue supervision of teachers, targeted provision of teaching and learning materi- als at secondary level, incl. science & technology · Expand enrollment at teacher training colleges in balance with qualified teacher needs, and at university level · Implement national policy on mother tongue instruction Efficiency enhancement · Appropriate distribution of teachers according to enrollment of public resource · Voluntary re-deployment of existing teachers with an incentive package or transfer of allocation and utilization pupils where feasible · Budget and expenditure framework for all resources/common procedures using GoM systems · Adapt EMIS and funding formulae to meet enrollment changes at district level · Adjust cross sub-sector resource allocation for future balanced sector development Strengthening · Improve data; EMIS, codification of all schools and teachers, merging of MoF and institutional MoEST salary and financial databases arrangements and · Strengthen district-level capacity of local system management, including school capacity for service inspection, teacher development, monitoring and reporting delivery · Strengthen management capacity at the school level, including the capacity of School Management Committee at primary level CHAPTER 1 Macroeconomic and Demographic Context A country's economic and social devel- nal resources. Unfortunately, the wastage in the opment is dependent on the natural Malawi education system is substantial and, as resources and the human capital such, the opportunities to increase efficiencies available to that society. In Malawi, are enormous. This is particularly the case in the natural resource base is limited to an agri- the current situation where institutional struc- culture sector heavily reliant on the export of tures are weak and where accountability and tobacco and on prevailing tobacco prices. This incentives to perform are largely non-existent. reliance on one economic sector, allied to a While the first two determinants, demographic recent decline in the terms of trade, makes the and macro-economic contexts, are often diffi- development of human resources--and the use cult to influence in the short term, the improved of knowledge--both an obstacle and an oppor- management of the system and the efficient use tunity for Malawi if the country is to diversify of resources are within the control of schools its economy and increase overall levels of pro- and the Ministry of Education, Science and ductivity. However, without an investment in Technology (MOEST).1 education and knowledge Malawi will have Other important aspects of education few, if any, choices and the prospective for relate to the micro-economic conditions, economic growth will remain bleak. including poverty and high opportunity The level of investment in education is deter- costs, and other important conditions affect- mined largely by the strength and size of the ing students and teachers, such as chronic economy (the macroeconomic conditions); by malnutrition, HIV/AIDS, social and cultural the demographic characteristics of the popula- settings--including language of instruction-- tion; by decisions about expenditure on educa- and the religious contexts within which the tion versus other sectors; and by the capacity of system operates. These constraints and con- the education system to absorb existing and ditions set the framework within which an increasing resources. As will be highlighted education system can develop, but the levels later, while the absolute amount of resources is of education, in turn, determine economic imperative for the development of a system, still development, and the ensuing choices have a more important is the ability of the government direct impact on poverty alleviation in the to efficiently use its existing internal and exter- country. 12 Macroeconomic and Demographic Context 13 The Malawi government is faced with some to primary education for all, it also forced the tough but undeniable questions and chal- government to recruit numerous teachers that lenges. For example, how much should the had no formal teacher training. In primary, the government invest in education? How much share of untrained teachers in the total stock of should it invest in the sub-sectors of education, teachers rose from 20% to 50% upon the when some adjustments to issues of quality, introduction of FPE and has largely stayed that access and equity need to be addressed? The way thereafter. present study will seek answers to questions The combination of these factors--external about limited choices and will attempt to environment (macro-economic, famine, mal- delineate a number of options and difficult nutrition, HIV/AIDS), declining per pupil choices that need to be addressed across the spending (especially in secondary education primary, secondary and higher education lev- and teacher training2), and increasing numbers els. The study also recognizes that without an of untrained teachers, especially in primary intervention, the Malawian education system and secondary--has resulted in the education will continue its downward spiral. In this first system becoming unbalanced and with partic- chapter, the focus is on general characteristics ularly deleterious effects on quality. In many of the macroeconomic, demographic and social ways one can speak of a collapse in the educa- context in the country, and the overall capaci- tion system. This has been further exacerbated ty of the system, including the impact of by a soaring debt burden and the collapse of HIV/AIDS. fiscal management and accountability follow- ing the introduction of multi-party elections and FPE. Malawi is one of the poorest non- Evolution in gross domestic conflict countries in the world today as reflect- product, per capita income and ed by its sobering human development indica- fiscal resources tors, as shown in Table 1.1. Table 1.2, below, traces the evolution in the Malawi is a small landlocked country situated population and economy. As shown (see also in southeastern Africa. Its land area is approx- Annex 1.1), in constant 1994 kwacha prices, imately 119,140 square kilometers and it has a GDP increased by 32 percent between 1991/92 population of about 10 million. In the 40 years and 2001/02, i.e. an average annual growth of since it became an independent country in 3.2 percent. Per capita income in current US 1964, Malawi has stagnated at extremely low dollars stood at US$177 in 2002. As the popu- levels of per capita income. In recent years, the lation grew by 21 percent over the last decade, prevalence rate of the HIV/AIDS pandemic has per capita income rose slightly at an average of increased to almost 20 percent and is responsi- roughly one percent per year. ble for the deaths of thousands of Malawians, Grants (including HIPC but excluding including teachers, civil servants and parents; it loans) as a share of GDP have been volatile is also leaving numerous children orphaned. over time but the overall trend over the past When the government introduced the Free Pri- decade is one of a strong increase, from levels mary Education Policy (FPE) in 1994, enroll- of 3 percent of GDP to 6­8 percent of GDP (or ment soared by approximately 1 million pupils as share of government revenue from levels of (from 1.9 million to 2.9 million). However, the 10­15 percent to levels of 25­35 percent). This policy has been implemented in a largely indicates a strong and increasing dependency unplanned manner and without adequate on foreign assistance in the public sector. financing. While it resulted in enormous Total public expenditures as a percentage of progress towards the goal of providing access GDP rose from about 25 to 35 percent during 14 Cost, Financing and School Effectiveness of Education in Malawi Table 1.1: Social and economic indicators (latest year available) Botswana Kenya Malawi Mauritius Mozambique S. Africa Tanzania Uganda Zambia Zimbabwe GDP Per Capita (constant 1995 US$) 3951 328 157 4429 195 3985 193 355 392 656 GNI Per Capita, PPP (current US$) 7190 1010 600 9940 820 9180 530 1230 750 2590 Growth of Per Capita GDP 3.2 0 3.5 4.5 5.2 0.7 1.6 4.2 -0.6 0 1995-2000 (% p.a.) Life Expectancy at Birth 39 48 39 71 43 48 45 42 38 40 Under-5 Mortality Rate (per 1,000 live births) 95 118 227 23 203 76 152 162 187 118 Adult Illiteracy (% of people ages 15 and above) 23 18 40 15 56 15 24 33 22 11 Gini Coefficient .. 0.45 0.40 .. 0.40 0.59 0.38 0.39 0.50 0.57 Percent Share of Income/Consumption of the Lowest 10% of Po pulation .. 1.80 2.50 .. 2.50 1.10 2.80 2.60 1.60 1.80 Percent Share of Income/Consumption of the Lowest 20% of Population .. 5.00 6.30 .. 6.50 2.90 6.80 6.60 4.20 4.00 Survey Year .. 1994 1997/98 .. 1996/97 1993/94 1993 1992/93 1996 1990/91 Note: The Per Capita GDP growth in Malawi is distorted by the 14.9% increase in 1995 after an 11.6% decline in 1994 (which is not included in the calculation). The picture from 1996­2000 for instance is of 1.7% annual growth, and from 1981­2000 of only 0.4% annual growth. Source: World Development Indicators. the last decade; in real terms, the public expen- revenues, partly through the above-mentioned diture grew by 94% from 1991/92 to 2002/03. increase in grants, and partly through increas- In general, both development and recurrent ing the public deficit reaching as much as 7.7% expenditures as a share of GDP have increased. by 2002/03. The higher level of public expenditure in In considering the prospects for educational 2002/03 compared to 1991/92 is partly development, it is important to assess the out- financed through slightly higher government look for economic growth and possible Table 1.2: Evolution in population, GDP, and total government revenue and expenditures 1991/92 1992/93 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 2002/03 Population (in million) 8.71 8.85 8.99 9.14 9.33 9.55 9.77 9.99 10.20 10.41 10.63 10.87 GDP in const. 1994 mill. Kw 10874 11057 10846 11438 12593 13188 13671 14197 14514 14936 14672 15141 GDP per cap. in const. 1994 Kw 1248 1249 1206 1252 1350 1381 1399 1421 1422 1434 1380 1393 Real growth rates n.a. 1.7% ­1.9% 5.5% 10.1% 4.7% 3.7% 3.8% 2.2% 2.9% ­1.8% 3.2% Inflation n.a. n.a. n.a. 58.9% 60.4% 23.4% 19.5% 37.3% 37.2% 29.6%3 n.a. n.a. Exchange Rate (MK/US$, avg) 3.20 4.00 6.57 12.01 15.30 15.88 23.76 37.58 51.82 65.87 74.45 76.74 Total Govt. Rev. (including grants) as % of GDP 22.0% 19.3% 19.7% 23.3% 21.7% 19.4% 22.1% 24.4% 24.4% 26.7% 23.0% 27.1% ­ Revenue (excluding grants) as % of GDP 18.7% 17.1% 16.8% 13.9% 15.0% 15.1% 18.0% 18.1% 17.5% 17.9% 16.4% 20.2% ­ Grants as % of GDP 3.3% 2.2% 2.9% 9.4% 6.7% 4.3% 4.1% 6.3% 7.0% 8.9% 6.6% 6.9% Total govt. Expenditures as % of GDP (%) 24.9% 30.5% 25.0% 38.1% 26.6% 22.1% 29.0% 29.1% 30.1% 31.9% 30.5% 34.8% ­ Recurrent expend. 19.9% 24.5% 20.2% 31.4% 21.1% 17.7% 22.9% 20.9% 19.5% 22.0% 23.4% 27.1% ­ Development expend. 5.0% 6.0% 4.8% 6.6% 5.5% 4.4% 6.1% 8.3% 10.6% 9.9% 7.0% 7.6% Deficit, counting Grants as a deficit-determining item5 ­3.0% ­11.2% ­5.3% ­14.8% ­4.9% ­2.7% ­6.9% ­4.7% ­5.7% ­5.2% ­7.5% ­7.7% Deficit, counting Grants as a deficit-financing item6 ­6.2% ­13.4% ­8.2% ­24.2% ­11.6% ­7.0% ­11.0% ­11.1% ­12.6% ­14.0% ­14.1% ­14.5% Sources: IMF "Selected Issues and Statistical Appendix. August 2002."; WB SIMA database; and authors' calculations. Government expenditure data are `actuals' with the exception of the most recent data for 2001/02 and 2002/03, which are estimates from IMF "First Review under the Poverty and Growth Facility and a Request for Waivers of Performance Criteria. August 2003." Population data are from WB-HNP Stats. Inflation and exchange rate are from IMF as referenced in PER. Macroeconomic and Demographic Context 15 increases in revenues. But the economy has force. In contrast, the majority of Malawians been struggling and its expected growth for the (84 percent), most of whom are subsistence coming years is not predicted to be much dif- farmers, contribute only 40 percent to the GDP ferent from the roughly 3 percent yearly (or about $159 per worker against $1,307 per growth experienced during the previous worker in services and industry). decade.7 The estimated growth is anticipated This clearly constitutes a dilemma. An to exceed the population growth, but per capi- expansion in industry and services is likely to ta incomes, and consequently per capita gov- contribute proportionally more to GDP and ernment revenues, are not expected to rise growth and fiscal revenue, but would only ben- much. efit a small segment of the population. In turn, In summary, the economy cannot be expect- while investment in agriculture has the poten- ed to generate much additional resources in the tial to increase, to some degree, productivity short to medium term. The structure of the and outputs (and thus fiscal revenue), the economy, with its high reliance on tobacco majority of farmers hold less than 1 hectare of exports, raises serious questions about the land10 and, as such, are unlikely to either con- long-term growth prospects for Malawi's edu- tribute or benefit much from any interventions. cation system: What are the goals for educa- Furthermore, opportunities for these small tion in Malawi? Can Malawi afford a further farmers are limited in terms of excess produc- major increase of its education system given tion and productivity gains, and thus for an the expected limited resources available? increased tax base. In other words, the majori- The determining factors when estimating the ty of Malawians are caught in a poverty trap impact of economic growth on government with extremely limited assets (land) and high revenues (and on the potential of generating illiteracy rates. additional resources to invest in education), In the long term, any per capita growth will are the composition of the labor force and the have to rely on increasing the land size of farm- contribution of each economic sector to GDP, ers and consequently gradually moving a high as shown in Table 1.3, below. number of farmers out of agriculture. This, in The estimated Malawi labor force in 2000 essence, is Malawi's dilemma: To increase pro- was 4.5 million9 with a US$1.5 billion econo- ductivity and provide new opportunities for my. The relative share of people working in the majority of rural Malawians, the govern- agriculture is still high at more than 80 percent ment will have to invest more in quality pri- of the population. As Table 1.3 indicates, mary education in the short term. From an industry and services contribute to approxi- educational perspective, however, a parameter mately 60 percent of the measured GDP, but for increasing economic growth is to support only benefits about 16 percent of the work the industry and service sectors and the expan- Table 1.3: Output and employment Share of total labor force GDP 2000 (in million USD)8 GDP sector share of total (in thousands and %) Agriculture 596 38% 3,766 (84%) Industry 271 17% 200 (4%) Services 699 45% 542 (12%) Total 1,566 100% 4,508 (100%) Sources: IMF Statistical appendix, July 2002. Uses 2000 GDP in constant 1994 factor cost Kwacha, and the average 1994 exchange rate of 8.74 Kw/USD Labor force from 1998 population census, and includes formal and non-formal sector employees. 16 Cost, Financing and School Effectiveness of Education in Malawi sion of these sectors to absorb increasing num- they do increase opportunity costs as a result bers of workers. This, in turn, requires a larg- of the weaker income bases in families, the er proportion of high quality secondary and need of an economic structure to support tertiary graduates, which necessitates further orphans, and because the characteristics of investments in post-primary education. HIV/AIDS contribute to a higher rate of illness While Malawi's favorable prospects for in families. These factors (indirect or direct long-term growth are strengthened by the costs) need to be addressed by government if country's intention to improve education and efforts to improve the human capital base are to diversify the economy, government revenues to be successful. Furthermore, the reduction in in the near and medium future will be generat- life expectancy has a severe impact on workers' ed largely by the current formal sector (agri- total number of active working years, their cultural tobacco exports, industry and the productivity and, consequently, their total net service sector). As a result, the fiscal burden is contribution to the economy. The marginal likely to be borne by a small fraction of the output per worker as a result of HIV/AIDS is labor force for the foreseeable future. probably inflated in the short run, as initially, On the other hand, a growing labor force output remains largely the same while workers and declining dependency ratio (see Table 1.4) die.11 However, in the case of some countries, will allow the fiscal burden to be shared by a Uganda, for example, HIV/AIDS is likely to relatively higher number of workers. The argu- have a negative effect on average productivity ment here is that if there are fewer people out- in the longer run--productivity gains that side the labor force than actively participating would normally be an alternative means of in it, the spending per beneficiary for public increasing the value-added in the formal sector subsidies can be increased. and, thereby, an increase in revenues. This The dependency ratio has declined slightly effect compounds the already depressed pro- over the last decade, indicating an ease in the ductivity levels in Malawi that result from the fiscal burden on workers. However, this slight- current low levels of education. ly positive trend in Malawi is hiding a fright- ening reality: HIV/AIDS. While the declining ratio suggests an easing in the fiscal pressure, Evolution in inter-sectoral allocation the reality is that the number of female-headed of public expenditure, in education households (30 percent), and the number of spending, and HIPC resource orphans, has increased. While these groups do availability not directly claim resources from government, Figure 1.1 indicates that prior to 1994, recur- Table 1.4: Evolution in the dependency ratio, 1990­2000 rent spending on social (and community) serv- Year Dependency ratio. ices was lower than spending on public affairs, 1990 0.99 unallocated expenses and economic services. 1992 0.98 This situation changed abrubtly from 1994/95 1994 0.98 on when social services recurrent expenditure 1996 0.98 increased and surpassed spending on general 1997 0.98 public affairs and economic services. Here, a 1998 0.97 little less than half of the increase in the pro- 1999 0.97 2000 0.97 portion of spending to social services went to the education sector (see Annex 1.2)--this was Source: HNP Stats, World Bank Note: Dependency ratio is calculated as the population aged 0­14 mostly linked to the introduction of FPE and plus >65 over the working force population aged 15­64 Macroeconomic and Demographic Context 17 Figure 1.1: Recurrent expenditure by functional classification, in percentages of total 50.0 45.0 40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 1990/91 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 General Administration Social Services Economic Services Unallocable Services Source: Malawi Economic Report, various issues Note: All percentages are based on `actuals', except for 2001/02, which is based on the `estimate'. the resultant increase in expenditures needed (general public service, defense, public order to pay the additional teachers recruited. and safety) and on unallocated services (debt Over time, the increase in the proportion of service, pensions and gratuities and others12). social services expenditure has largely been at The proportions of general administration and the expense of the proportion of economic unallocated services have been very volatile services expenditure, more than at the expense over time, but overall, no significant changes of the proportion on general administration have occurred during the past decade. Figure 1.2: Development expenditure by functional classification, in percentages of total 60.00 50.00 40.00 30.00 20.00 10.00 0.00 1990/91 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 General Administration Social Services Economic Services Unallocable Services Source: Malawi Economic Report, various issues Note: All percentages are based on `actuals', except for 2001/02, which is based on the `estimate'. 18 Cost, Financing and School Effectiveness of Education in Malawi Figure 1.2, below, shows breakdown by sec- bilateral off-budget development spending was tor of development expenditure and indicates estimated at 65 and 70 percent, respectively. that the shares of development expenditure are External funding was recently estimated to be even more volatile than recurrent expenditures. around one third of total education spending.13 This derives from the nature of large invest- The evolution in education expenditures, as ment projects and those that tend to be one- a share of GDP, shows a slightly negative trend offs typically dependent on donor support. in the nineties but started recovering in Nevertheless, the share of total development 2000/01. This ratio, however, may actually be budget that went to the social and community higher if off-budget development expenditure services sector in the post-1994 period was were included. clearly higher than in the pre-1994 era. The Table 1.6 indicates that Malawi, by the end share of development expenditure allocated to of the nineties, was somewhat below average education also fluctuated greatly, with a peak on education expenditure as a proportion of in 1994/95, but thereafter stabilized at around domestic revenue relative to other African 10 to 15 percent. countries. In particular, when compared to As can be seen from Table 1.5, below, total Uganda--another country that is going education expenditures (TEE) as a share of total through an expansion in enrollment as a result government expenditures (TGE) rose initially of a free primary education policy--the educa- (after the introduction of FPE) from 13 percent tion spending in Malawi was at the lower end; in 1993/94 to 18 percent two years later, but spending on education has however recovered declined after 1998 to pre-FPE levels. However, in the last few years. in 2001/02, spending on education rose again to 16% and is expected to have remained at this HIPC (Highly Indebted Poor higher level the past couple of years, even Countries) resources situation though official data are not yet available. Main- taining a high level of spending on the social sec- Malawi is eligible for assistance under the tors is a requirement under the HIPC initiative, Enhanced HIPC Initiative. HIPC receipts and as reviewed later in this Chapter. projected receipts were initially estimated in a The development expenditure on education World Bank report (World Bank. 2000) at seems to have been very low in some years. US$91.4 million for the fiscal years However, one explanation could be that most 2000/01­2002/03. An indicative allocation of education development expenditures are off the bulk of these resources by government budget and not included in public spending shows that education would be allocated a data. In 1999/00 and 2000/01, the share of substantial proportion of this amount estimat- Table 1.5: Trends in public spending on education, and comparison to total government spending, 1993/94­2001/02, in percentages 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 TEE in Total Government Expenditure (TGE) (%) 13 14 18 17 20 14 13 12 16 TEE (Recurrent) in TGE (R) (%) 13 10 16 19 21 14 16 12 16 TEE (Development) in TGE (D) (%) 15 28 23 8 15 13 9 12 15 TEE in GDP (%) 3.3 5.7 5.4 3.9 5.5 3.7 3.7 4.0 4.7 TEE (Recur.) in GDP (%) 2.6 3.6 4.0 3.5 4.7 2.5 2.8 2.8 3.6 TEE (Develop.) in GDP (%) 0.7 2.1 1.5 0.4 0.8 1.2 0.9 1.2 1.1 Sources: Malawi Economic Report, various issues Macroeconomic and Demographic Context 19 Table 1.6: Comparative information on government recurrent spending on education as a percentage of government current revenue, excluding grants, and as a percentage of GDP Malawi Ghana Mozamb. Nigeria Rwanda Uganda Zambia Zimbabwe * African 1998/99 2000/01 (1999) (1998) (2000) (2000) (2000) (1998) (1997) Countries Educ. recurrent exp. as % of gov. curr. rev excl grant 17.3 15 17.6 18.1 9.9 32.6 30.1 12.3 28.3 17.8 Educ. recurr. exp. as % of GDP 2.5 2.8 3.8 2 4.6 3.2 3.2 2.3 7.1 3.04 Source: Bruns, B. et al. 2002. For Malawi, authors' calculations are used. * Refers to 33 African countries ed at US$25.2 million over the full period.14 pre-packaging donor-supplied primary text- Of this amount, approximately two thirds books for each school with delivery directly would be allocated to teaching materials while from the supplier to the schools. another third would be for teacher training. As The first of these `completion point triggers' the bulk of this assistance was provided after also provides an additional means to examine the last year for which actual expenditures the current and expected resources made avail- data are available (2000/01), it is not clear able for the education sector; it will be possible whether these additional resources have mate- to verify how far the target of `at least 23 per- rialized and been allocated to the education cent' of education expenditure in the discre- sector. tionary recurrent budget has been achieved. After 2002/03 it is possible to expect a fur- However, the picture for recent years for which ther increase in the yearly debt relief, thus `actuals' are available and the most recent ensuring additional resources for the education `revised estimates' (i.e. for the most recent sectors. This is estimated at approximately 10 years no `actuals' are available) indicate a great percent above the 2002/03 level that was esti- deal of volatility and some discrepancies, mated at US$44.2 million in total and at depending on the data sources that are used US$12.8 million for education.15 (see Annex 1.3). This makes it difficult to draw The debt relief is contingent on the govern- any firm conclusions. However, the proportion ment's commitment to pursuing a limited set of of education sector expenditure in the discre- `completion point triggers' and actions recog- tionary recurrent budget appears on average to nized as important for poverty reduction and be relatively close to the target of 23 percent. growth. In the education sector, the focus is on Depending on the data source, the average for improving the quality of education by: (i) the past six years (mixture of `actual expendi- increasing to at least 23 percent the share of tures' for the first four years and `estimated education sector (including universities and expenditures' for the last two years) has been other subvented organizations) expenditure in 23.7 percent or 24.6 percent and, as such, is the discretionary recurrent budget; (ii) ensuring within the set target. yearly enrollment of 6,000 students for teacher training, and establishment of in-service train- ing for primary teachers; (iii) reallocating of Notes budgetary resources from secondary school 1. From 2004, the Ministry of Education boarding (except for `special needs education') and Human Resources. to teaching and learning materials; and (iv) 20 Cost, Financing and School Effectiveness of Education in Malawi 2. In primary there was a huge dip in 11. Jill Armstrong, Uganda's AIDS Crisis: 1994/95, upon the introduction of FPE; main- Its Implications for Development, World Bank ly due to hiring large numbers of untrained Discussion Paper 298 (Washington, D.C.: teachers, the unit cost started adjusting again World Bank, 1995). and by 1997/98 it reached more or less pre- 12. The travel budget as a share of GDP is FPE levels again and has largely remained at one of the highest in Africa. that level thereafter. 13. In 1999/00 recurrent and development 3. This is the 2000 value instead of 2000/01. spending together were 4,657 million kwacha, 4. This is the 2002 value instead of 2002/03. the on-budget donor spending was 516 million 5. Deficit as the difference between Revenue kwacha and off-budget donor spending 1,065 plus Grants, and Expenditure. million kwacha. For 2000/01 these were 6. Deficit as the difference between Revenue respectively 4,987, 492.9 and 1,295 million (excluding Grants), and Expenditure. kwacha. 7. IMF/World Bank CAS/PRGF projections. 14. 2000/01 US$2.4 million, 2001/02 US$ 8. Industry includes mining & quarrying, 9.9 million, 2002/03 US$ 12.8 million. manufacturing, electricity & water, construc- 15. However, the Reserve Bank (in "Reserve tion. Services include distribution, transport & Bank of Malawi: Foreign Exchange Budget, communications, financial & professional 1997­2002" on 22. October 2003) published services, ownership of dwellings, private com- projected total HIPC relief by calendar year munity services, and government services. (not fiscal year), where there is still a signifi- 9. Of course most of them work in the infor- cant expected increase from 2003 to 2004. mal economy. Projections for total HIPC debt relief by calen- 10. Average land size in the EU is about 40 dar year (not fiscal year), where there is still a hectare: despite high levels of productivity and significant expected increase from 2003 to use of technology EU farmers are dependent on 2004. Projections for total HIPC debt relief are high levels of subsidies (about 20 percent of as follows: 2001 US$27.4 million, 2002 income). US$28.8 million, 2003 US$51.6 million, and 2004 US$78.0 million. CHAPTER 2 Enrollment Patterns C hapter 1 described Malawi's educa- (CSSs). After two years of secondary, pupils sit tion system in the context of macro- for the national Junior Certificate of Secondary economic trends, government rev- Education (JCE); after two more years, for the enue and spending on education, Malawi School Certificate Examination and issues related to demographic changes and (MSCE). Tertiary education is provided by an the impact of HIV/AIDS. This chapter array of education institutions including pri- describes the overall education system. In addi- mary and secondary teacher training colleges, tion to common indicators of coverage, such as technical and vocational training schools, and gross enrollment ratios and entry rates, university colleges. For university entrance and detailed information is presented on patterns for the secondary teacher training college of cohort survival, grade-specific enrollment, (Domasi), an MSCE certificate is required. For transition rates, completion rates, dropout and primary teacher training the policy is to take repetition. MSCE graduates, but JCE passers can also be accepted. Technical and vocational training can start either after JCE or MSCE. Structure of Malawi's education Primary education, preschool education and system non-formal education initiatives such as adult literacy fall in the category of basic education, The formal education system in Malawi, as of with the purpose to equip people with basic 2003, follows an 8­4­4 structure: 8 years of knowledge and skills to allow them to function primary, 4 years of secondary and typically 4 as competent and productive citizens in a free years of university level education1 (see Figure society. Primary and secondary education is 2.1). At the end of primary, pupils sit for the administered by the MOEST through its head- Primary School Leaving Certificate Examina- quarters, the six education divisions and 33 tion (PSLCE), which determines their eligibili- district education offices. University level edu- ty for entry into secondary school. Public sec- cation is sub-vented by government, and uni- ondary students attend either Community Day versity institutions are autonomous. The pri- Secondary Schools (CDSSs) (previously vate sector participates in the provision of all MCDE), or Conventional Secondary Schools levels of education but this is most pronounced 21 22 Cost, Financing and School Effectiveness of Education in Malawi Figure 2.1: Structure of Malawi's education system Source: Basic Education Statistics. at the secondary level where private sector con- Enrollment trends tributions have been increasing substantially. The current school calendar was introduced Data on enrollment at the different levels is in 1997; the year is divided into three semes- presented in Table 2.1.2 The Malawi education ters, or terms, and commences in January and system is characterized by permitting only a ends in November. The end of each term is fol- small share of primary school graduates access lowed by a holiday of three to four weeks. The to post-primary levels. This situation was fur- longest school holiday commences in Novem- ther exacerbated by the introduction of Free ber when the schools close for almost two Primary Education, which led to an increase in months. primary enrollment of about 50 percent; or a Table 2.1: Distribution of enrollment, by level of schooling, 2000 Level of Schooling Number of students Percent Preschool 90,000 2.7 Lower primary (Std. 1­4) 2,200,000 65.1 Upper primary (Std. 5­8) 800,000 23.7 Secondary (estimated3) 275,000 8.0 Primary Teacher Training Colleges 12,000 0.4 Technical and Vocational 1,500 0.0 Universities 4,000 0.1 Total 3,382,500 100 Source: Basic Education Statistics 2000. Enrollment Patterns 23 total of one million new students entering pri- A troubling feature of the Malawi system is mary. Less than 9 percent of the total student the significant number of underage children population is currently enrolled in post-pri- enrolled in primary; for instance, in 2002, 28 mary education, while 91 percent of total percent of 5 year-old children were already enrollment is in primary education and of these attending primary school.5 To address this, 65 percent (or 2.2 million) are enrolled in according to the new primary school curricu- lower primary (Standards 1 to 4). lum (2002), the government is considering reducing the primary cycle from eight to seven years while introducing a pre-school year. Pre-primary education However, such a change may draw away resources from primary and secondary levels There is considerable evidence worldwide to already struggling to be financially sustainable. indicate that attending nursery school helps a Future investment needs to be based on a care- child prepare for primary school; it also pro- ful cost-benefit analysis, one that weighs gov- vides a foundation for lifelong learning. In ernment investment in pre-school education Malawi, the Ministry of Gender, Youth, and against improvements in the currently under- Community Services organizes public Early funded lower grades of primary. Childhood Development (ECD). In 2002 there were 1,618 childcare centers registered with the Ministry, with 4,666 pre-school teachers Basic or primary education responsible for approximately 80,000 children, aged between 3 and 6 years.4 Annex 2.1 Malawi has two types of primary schools: pub- includes background characteristics of children lic and private. The latest school census for between the age of 6 and 14 who attended 2001 established that there were 4,688 pub- nursery schools, whether public or private. licly funded primary schools with a total These data indicate that about one in ten chil- enrollment of 3,143,427 (EMIS database dren (9.4 percent) had attended nursery school 2001). These include schools where the propri- for an average of 1.7 years. Two factors are etor is the central government, the Local Edu- important predictors for nursery school enroll- cation Authority, or a religious agency that ment: geographic location and income group. receives public funding. The total number of In urban areas, 39 percent of children attended private schools is difficult to determine because nursery school compared with just 6 percent in government only recently started to register rural areas. Furthermore, 27 percent of chil- private schools. Currently, the number of gov- dren from the highest wealth quintile attended ernment registered private schools is 49 and nursery school--approximately 10 times more they enroll 18,500 pupils. The Private Schools frequently than children from the lowest Association of Malawi, PRISAM, estimates income group. This difference is compounded that there are about 40,000 pupils in 250 by the fact that years of attendance are longer schools. However, the Ed Data 2002 survey for both urban and wealthy children. Younger indicates that enrollment in private primary is children are also more likely to have gone to much higher; most likely around 5 percent of nursery school than older children; only 5 per- total primary enrollment--about 2 percent in cent of the 11­14 age group compared to 15 private non-religious schools, and 3 percent in percent of those aged 6­7 years had attended private religious schools. nursery school. This suggests that more par- Since the abolition of all primary school fees ents than heretofore are sending their children in 1994, access to primary has increased dra- to nursery schools. matically allowing practically all children to 24 Cost, Financing and School Effectiveness of Education in Malawi Figure 2.2: Proportion of children who ever enrolled in primary, by age 100% 80% 60% 40% 20% 0% 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0 Age DHS 2000 IHS 1998 Source: Authors' calculation using data from the DHS 2000 survey and IHS 1998. attend school. Figure 2.2, below, illustrates and that gradually a higher number of children that more than 95 percent of children attend would enroll within an age appropriate cohort primary school at some point in their lives. -- in other words, that the GER would con- One way to measure the increase in enroll- verge towards the Net Enrollment Ratio ment is to consider the changes in the gross (NER). However, as Figure 2.2 shows, timely enrollment rate (GER) over time. Table 2.2 enrollment remains an issue: at age 7 only illustrates that the GER for primary has about 70 to 80 percent of children have start- increased substantially over the past ten years-- ed school. Late or early entry is common par- from 88 percent prior to the introduction of ticularly in rural areas where children's ages FPE (in 1991/92) to roughly 120 percent at the are difficult to determine in the absence of doc- end of the decade -- an indication of the suc- umentation. The problem of early entry is cess of the FPE in improving coverage. common when the school is perceived as pro- viding free childcare or when younger children Enrollment by age in primary join their older siblings. Late entry is common when children have to walk far to get to When FPE was introduced about 10 years ago, school, and is then compounded through repe- many of the children that enrolled for the first tition of grades. The persistently high number time were over-age. It was anticipated that this of over- and under-age children points to both situation would level out in subsequent years demand and supply side issues. On one hand, Table 2.2: Absolute enrollment numbers and gross enrollment ratios over time 1991/92 1993/94 1995/96 1997 1999 2001 No. of GER No. of GER No. of GER No. of GER No. of GER No. of Students (%) Students (%) Students (%) Students (%) Students (%) Students Std. 1­8 1,671,583 88* 1,695,423 93* 2,887,007 120 2,805,785 115.1 3,016,972 120.8 3,187,835 Std. 1­4 1,239,535 1,466,928 2,203,663 2,090,728 160.2 2,178,080 167.6 2,296,039 Std. 5­8 432,048 428,495 683,444 715,057 63.2 838,892 70.1 891,796 Source: Basic Education Statistics and UN Population data used for EFA-FTI. * Taken from Basic education statistics. ** From DRAFT Basic Education Statistics 2001. Enrollment Patterns 25 Figure 2.3: Enrollment rate by age, different sources, 2000 140.00% 120.00% 100.00% 80.00% 60.00% 40.00% 20.00% 0.00% 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 ag e "DHS 2000" 2000 Min Educ enrollm and NSO smoothed est. based on orig. pop. census 2000 Min Educ enrollm and Unesco pop Source: Authors' calculation using data from the DHS 2000 survey, from EMIS 2000, and from the National Statistics Office and UNESCO for population estimates. the opportunity cost associated with having a factory learning outcomes and simultaneously child in school affects enrollment and atten- progress more rapidly through primary dance. On the other hand, the inadequate school.6 learning environment leads to low outcomes and high repetition rates. Disparities in enrollment data and The various sources of data on enrollment sources provide conflicting information about the extent of over- and under-age children in pri- The analysis in Annex 2.2 indicates that the mary (Figure 2.3 and Annex 2.3). The house- GER in primary has, in recent years, fluctuat- hold survey data indicate that the number of ed between roughly 110 and 133 percent.7 In over-age children is much higher than recorded light of the strengths and weaknesses of each by the official school census statistics records. data source, for the purpose of the ensuing cal- According to the household survey, close to 30 culations, the gross enrollment ratio is estimat- percent of 5 year-olds are already enrolled ed to be about 120 percent. (representing close to 100,000 children) If the GER for lower versus upper primary, whereas school census data puts this estimate as shown in Figure 2.4, is examined it provides at 15 percent. a more accurate picture of the pressures on the In addition to needing more reliable statis- lower standards as a result of repetition and tics, the more significant challenge is how to early/late entry. The gross enrollment ratio is as reduce the age disparity of primary school chil- high as 170 percent in the first four grades but dren. The existing wide age range makes teach- shrinks to a mere 70 percent in the last four ing and learning even more difficult in an envi- grades. The high GER in lower primary con- ronment that already has many additional firms that a high proportion of children repeat factors undermining the quality of education. standards 1 to 4. To address the problem of over-age children, Figure 2.5, below, indicates the evolution in the most important measure by far is to enrollment by standard during the nineties. improve the quality of education in order to The sharp rise in Standard 1 enrollment reduce repetition rates so children attain satis- between 1993/94 and 1994/95, and the more 26 Cost, Financing and School Effectiveness of Education in Malawi Figure 2.4: Gross enrollment rates by level of schooling, 2000 primary stand 1-4 stand 5-8 0 5 0 100 150 200 Source: Basic Education Statistics, and UN population. Note: For primary the corresponding population is children 6 to 13, with population 6 to 9 for Std. 1-4 calculations and 10-13 for Std. 5-8 calculations. moderate but still significant rise in enrollment subsequent years, new entrants were primarily in other standards, is attributable to the decla- children of normal entry age (age 6). ration of FPE. In fact, in 1991/92 when fees for Standard 1 pupils were abolished, enrollment Pattern of student flow through the grew by 19 percent over the previous year, and system, survival, repetition rates and in 1994/95, when the policy was extended to efficiency all primary school pupils, enrollment grew by 51 percent compared to the previous year. The gross enrollment ratio is a useful measure The much lower enrollment increases in the of the capacity of the education system in terms higher standards during subsequent years sug- of the number of places available. However, in gest that many children that enrolled in Malawi, grade repetition and drop out are com- 1994/95 did not progress to the next level. The mon, and consequently the gross enrollment decline in Standard 1 enrollment after the peak ratio is a poor measure of actual coverage.8 of 1994/95 is attributable to the fact that, in This next section investigates the efficiency of Figure 2.5: Enrollment trends by standard 1000000 900000 800000 Std 1 700000 Std 2 600000 Std 3 500000 Std 4 400000 Std 5 300000 Std 6 200000 Std 7 100000 Std 8 0 1992- 1993- 1994- 1995- 1997 1998 1999 2000 93 94 95 96 Source: Basic Education Statistics, various years. Enrollment Patterns 27 the education system, by examining pupil out to 1.39 for 2000. For the purposes of this admission, survival and transition rates. study, and to enable government to derive Comparing Data Sources to calculate Entry helpful conclusions and identify possible policy Rate, Population and Repetition. The `Cohort interventions, the repetition rate from the Rate of Entry' (or gross admission rate) is cal- household surveys is used in the following-- culated by dividing new standard 1 entrants unless otherwise indicated.9 Likewise the study (or non-repeaters) by the total population of will use the UN-EFA population estimate. school entry age. In 1987 this indicator was Repetition and Dropout Rates. A striking 0.87 (87 percent) but in 1998 it had increased characteristic of the primary education system dramatically to 2.29 (229 percent) when using is the widespread prevalence of repetition and the standard school census and population dropout. However, discrepancies in data make census statistics. It is obvious that under nor- it difficult to establish the current levels of rep- mal circumstances the `cohort rate of entry' etition and dropout. Table 2.3 summarizes rep- should not be more than 1, as there should not etition and dropout rates by grade from three be more new children enrolling into primary as different data sources, and shows averages for there are children of the corresponding age lower/upper primary and for primary as a group. As analyzed in Annex 2.4, a non-stan- whole. The average repetition rate from the dard use of the repetition rate concept, togeth- two household surveys (25.8 and 28.4 percent) er with a probable undercount of the popula- is almost twice as high as the recorded repeti- tion in the 1998 census explains part of the tion rate of 15.5 percent from the school cen- discrepancy. sus with the largest differences noted in Stan- It is thus likely that the school census statis- dard 1 and Standard 8. According to all the tics underestimate the level of repetition sources, the repetition rate generally declines (Annex 2.4). The household surveys available as children progress through the grades, with for 2000 and 2003 provide an alternative the exception of standard 8 where repetition is method to assess the extent of repetition. Fur- very high. In general, boys are more likely to thermore, if the high-end population estimates, repeat than girls, particularly in Standard 8.10 like those from UN-EFA, are used, the incon- This trend may be due to parents' greater con- sistency is further reduced. Using the best cern to see their sons, versus daughters, available data, the cohort rate of entry comes progress into secondary school. Table 2.3: Repetition and dropout rates, EMIS 2000 and DHS 2000 Avg. Avg. Avg. Std. 1 Std. 2 Std. 3 Std. 4 Std. 5 Std. 6 Std. 7 Std. 8 primary Stds 1­4 Stds 5­8 Repetition Ed Data 40.6 24.2 27.7 19.1 10.3 9.8 12.2 19.5 25.8 n.a. n.a. DHS 45.3 24.3 26.9 17.2 15.4 12.7 11.2 38.0 28.4 31.9* 18.6* EMIS 18.8 16.6 16.6 13.9 11.7 11.3 10.1 13.9 15.5 16.9 11.7 Dropout Ed Data 8.5 5.9 4.8 6.3 10.3 11.8 9.9 20.1 8.2 n.a. n.a. DHS 2.7 3.3 4.3 5.5 6.7 6.9 7.7 11.5 4.8 3.7 7.4 EMIS 13.9 11.1 9.3 8.9 7.7 8.3 8.3 8.2 10.4 11.3 8.0 Sources: Basic Education Statistics 1999, 2000. The repetition rate is calculated here as the repeaters in grade x in 2000 divided by the total enrollment in grade x in 1999. The dropout rate here is calculated as the dropouts from grade x in 2000 divided by the total enrollment in grade x in 1999. For DHS 2000 and Ed Data 2002 the results stem from the final reports of the studies. Repetition rates are calculated as the percentage of students in a given standard who are repeating that standard. Dropout rates are calculated as the percentage of students, in a given standard, in the previous school year (irrespective of whether or not they finished the previous year or left during that year) who did not attend school at any point during the school year. *Authors' calculations using dataset. 28 Cost, Financing and School Effectiveness of Education in Malawi The dropout rates reported by the different the present unsatisfactory Primary Completion sources vary considerably, from estimates Rates (PCR). On the positive side there is a around 4­5% per grade to as much as 10­11% firm basis for moving forward since at the per grade. Two data sources imply that moment 95 percent, or more, of all children dropout is more prevalent in the higher grades enroll: as such, further improvements will only than lower grades, while another source shows be achieved when a resolute effort to reduce the reverse pattern. Annex 2.6 shows in further repetition and dropout rates is implemented. detail the causes for children dropping out Survival Rates. Figure 2.6 shows an estima- from school. tion of the survival rate by grade in primary The Monitoring Learning Achievement and secondary school and indicates a steep (MLA, 1999) study revealed that out of 11 downward trend throughout primary and a African countries, Malawi has the highest rep- much flatter trend for secondary (see also etition rate in the first four grades of primary Annex 2.7). These data imply that dropout (see Annex 2.5). Further, the difference, rela- could be even higher than the 5­10% indicat- tive to other countries, is quite significant, par- ed in Table 2.3. For example, according to Fig- ticularly for the first three grades. Not only is ure 2.6, in the academic year 1999­2000, for the percentage of children passing from one every 100 non-repeating students that entered grade to the next by far the lowest in Malawi, Standard 1 only about 70 were still in school substantial numbers of children repeat more by Standard 4, corresponding to an annual than once--a phenomenon rare in most other dropout rate of more than 11%; by standard 8 SSA countries. The more recent household sur- only a quarter of the potential student popula- vey in 2002 recorded a higher dropout rate tion had persevered, implying that on average than the 2000 household survey. Here, also, as much as 18% of the class had dropped out dropout rates are much higher in the rural each year. These high dropout rates, however, areas relative to urban areas.11 are not consistent with the Primary Comple- In summary, primary school is characterized tion Rates presented in the following, and can by extremely high levels of repetition and therefore be regarded as higher end estimates. dropout, irrespective of the source of informa- Annex 2.7 also suggests a real deterioration tion. The challenge in Malawi is to improve on in survival rates between the time just before Figure 2.6: Estimated survival rate primary and secondary, 2000 (cross-sectional method) 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 Source: Basic Education Stats 2000 for enrollment, DHS2000 for repetition, UN-EFA population. Enrollment Patterns 29 the introduction of FPE until today. Results responds to average dropout around 10% of from the quasi-time series survival estimates the student body each year). indicate that a student's chances of "survival" Despite the clear positive trend, however, it throughout primary went down, possibly as is not likely that Malawi will reach a PCR of much as 20 percentage points. However, due to 100% by 2015. As such, it may prove more the greater intake into primary during the same prudent for the government to develop its own period, the number of children finishing pri- targets and goals, ones based on a credible and mary has gone up in absolute terms, as reflect- sustainable plan that will enforce and acceler- ed in the improved completion rates discussed ate the positive trend in improving the PCR. in the following. Annex 2.9 contains proxy completion rate calculations in which the non-repeaters at a Completion rates certain grade level are divided by the popula- tion of the corresponding age group. Com- The Primary Completion Rate (PCR) is the pared to the other data sources analyzed, these main indicator for Education for All (EFA) and estimates however seem to be in the higher for the Millennium Development Goal for edu- end, possibly underestimating real dropout: the cation. Both share the goal of achieving by year `proxy completion rate' for Standard 4 shows 2015 a completion rate of 100% for six grades the remarkable increase in completion rates of of primary. However, as primary school in approximately 45 percentage points over the Malawi has eight grades, and as there is no period in question--it reached about 90 per- exam after standard 6, the corresponding indi- cent in 2000. For the Standard 6 `proxy com- cator is the PCR based on the primary school pletion rate' (which would coincide with the leaving exam at the end of standard 8. Figure EFA/MDG minimal objective) there has been 2.7 shows the PCR for each year between 1992 an approximate 25­30 percentage point and 2000.12 It increased by about 15 percent- increase--it reached 60 to 70 percent in 2000. age points: from 20­25 percent to 35­40 per- For Standard 8, the `proxy completion rate' is cent between 1992 and 2000 and this repre- approximately 10 percentage points higher sents a significant and important improvement than the real PCR, and has not improved con- in the completion rate (a PCR of 35­40% cor- siderably since 1995. Figure 2.7: Primary school completion rate over time, two population estimates 50.0 40.0 30.0 20.0 10.0 0.0 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year Source: Basic Education Stats for PLSCE passers, NSO population. Highest level is standard NSO population. Lowest level is using NSO population plus 15%. (UN-EFA population is not available this far back). 30 Cost, Financing and School Effectiveness of Education in Malawi Being already very close to all children start- straint for pupils restricting their access to ing school, the challenge now is to improve the complete the whole cycle of primary education. PCR by reducing the number of dropouts. A To estimate the full extent of the junior comparison of the primary school completion school phenomenon, it is necessary to distin- rates with those of other countries in the region guish between two categories of schools. First, indicates that Malawi's PCR is lower than there are the recently established schools that most. While it is still at about average for 33 may eventually offer the full complement of sampled African countries, this is not a great standards, and second those schools that have consolation (see Annex 2.9). been established for a prolonged period but have not expanded to cater for all standards. Primary School Leaving Certificate Regardless, children attending these schools (PSLCE) are more at risk of being `pushed out' before completing the primary standards.13 Annex At the end of the 8-year primary cycle, pupils sit 2.10 indicates that approximately 27 percent public examinations, which lead to the Primary of all new Standard 1 entrants (650,829 School Leaving Certificate (PSLCE). The PSLCE pupils) enroll in schools that do not cater for is a pre-requisite to enroll in the secondary sub- all standards. Approximately two-thirds attend sector and it is used to select pupils for this level schools that were recently established and, as by ranking them and deciding what type of sec- such, may include additional grades in the ondary school they qualify to attend. The pass future or are already acknowledged as `feeder rates in the PSLCE are quite high as can be seen schools'. Overall, the extent of the problem of from Table 2.4. Performance between students long established school offering less than six in rural and urban areas is quite similar (rural grades appears to be affecting only about 4 79 percent; urban 77 percent, in 2000); so stu- percent of pupils but the key issue is the prox- dents who survive primary in rural areas do not imity of the `complete' school. appear to be at a disadvantage in their exami- Distance to school is one explanatory factor nation performance. to indicate why some children do not enter pri- mary or why a significant proportion of stu- Junior primary schools and distance dents start attending for the first time at ages to schools higher than the official entry age. Annex 2.10 provides data on distance and walking times to The existence of Junior Primary schools-- schools in both rural and urban areas. A total schools that provide less than the full eight of 64 percent of children in urban areas live grades of education--may be a supply-side con- within 15 minutes of the nearest school; for Table 2.4: PSLCE results and transition to different types of secondary schools YEAR 1994­95 1995­96 1997 1998 1999 2000 2001 2002 Pass Rate % 78 62 77 75 81 79 60 64 Candidates 116,875 120,994 128,272 126,515 131,444 139,036 161,786 157,835 Transition to CSS 9% 10% 15% 13% 16% 13%* ­ ­ Trans. to CDSS/Mcde ­ 36% 38% 49% 61% 55%* ­ ­ Trans. to private** ­ ­ 12% 8% 5% 7%* ­ ­ Source: Exam data is taken from MANEB. Enrollments are from EMIS. Transition rate is calculated as form 1 enrollment in year x / number of PSLCE passes in year x­1. * Based upon estimated enrollment numbers for 2000. **This may be underestimated because a significant number of private schools are not registered with the government and are not included in the statistics. Enrollment Patterns 31 rural children the figure is 33 percent--indeed, majority of graduating primary students. In more than one-third of rural children are 2000, the estimated Form 1 enrollment at required to walk more than half an hour. How- CDSSs was 58,101 out of a total secondary ever, the extent of the problem of access to enrollment of 81,443. complete schools requires further investiga- It is important to note that the combination tion, particularly as the use of feeder, or junior of high transition rates and the introduction of schools, may be a transient functional mecha- the CDSSs resulted in a significant intake into nism that caters for the limited number of chil- secondary education. Similarly to the impact of dren progressing to higher levels and, as such, the 1994 FPE policy, this expansion was may make good economic sense if it entails unplanned and lacked essential funding. As a more efficient teacher utilization. result, quality has been seriously undermined. Special education provisions Secondary education Government policy on the education of chil- dren with special needs emphasizes the impor- Secondary education is divided into two sec- tance of mainstream integration. Here, a key tions: junior secondary comprising Forms 1 issue is to raise community awareness of the and 2 and senior secondary, comprising Forms importance of facilitating the enrollment of 3 and 4. At the end of junior secondary, stu- these children into regular schools. Currently, dents sit public examinations. Successful can- there are a number of primary schools that didates are awarded a Junior Certificate of offer boarding facilities to children with severe Education (JCE) which is the basis for enroll- handicaps; the government supports these ment into senior secondary education. In Form through maintenance grants. There are cur- 4 students take the Malawi School Certificate rently 28,711 pupils attending normal primary of Education (MSCE) and the results deter- schools who are defined as having different mine access to various tertiary institutions, types of disabilities (Table 2.5). such as university education, and teacher train- ing14 institutions. Transition rates from primary into There are several types of Secondary secondary school Schools: CSSs, which include Government Day and Boarding Schools, as well as Grant Access to secondary education in Malawi is Aided Schools; CDSSs; Private Schools; and limited. However, transition rates for those Malawi College of Distance Education who pass the PSLCE have increased dramati- (MCDE). Annex 2.11 includes a description of cally; the average transition rate to secondary each. The CDSS schools only officially came education was about 75 percent of PSLCE into existence in 1998 when nearly all Dis- passers in 2000 (see Table 2.4, above). This tance Education Centers (DEC)15 were con- increase has been greatly facilitated by the verted into CDSSs. Most of the teachers at introduction of CDSSs, which now enroll the these centers, however, were primary school Table 2.5: Students with disabilities integrated in primary schools Physical Learning Emotional Visually Hearing Disabilities Disorders Disorders Impaired Impaired Total Type of Disorder 2999 9180 5476 4665 6391 28,711 Source: Basic Education Statistics, 2000 32 Cost, Financing and School Effectiveness of Education in Malawi teachers and were ill-prepared to teach sec- over-aged students enrolled in secondary ondary students. education. In fact, two thirds of all students As a result of the above weaknesses, a large are outside the official age range (14­17 number of students enrolled in poorly run years) for secondary. DECs whereas only a small elite group was Enrollment growth in absolute terms has able to access the well funded CSS schools. In been concentrated in the MCDEs and CDSSs; 1996­1997, DANIDA agreed to assist in the they registered an increment of well over upgrading of all DECs to standard secondary 100,000 between 1993/94 and 2000. Figure school status, under the title CDSS. However, 2.8 shows the distribution of enrollment by when DANIDA withdrew from the education type of school between 1993/94 and 2000 and sector, the government lacked the necessary highlights how the CDSSs have come to repre- fiscal resources to proceed with the plan: in sent the lion's share of secondary enrollment, effect the change from DECs to CDSSs up from approximately 40 percent in 1994/95 occurred in name only. As a result, while a to 70 percent in 1999/00. CDSS is expected to provide an equivalent The Private Schools Association of Malawi standard of education as the better endowed (PRISAM) is aware of the existence of a total CSS, the reality is that they are extremely of 300 private secondary schools in the coun- under-funded and the quality, as measured by try as a whole but, of these, only 171 are offi- examination pass rates, for example, is much cially registered. Furthermore, PRISAM con- lower. The majority of secondary students are cedes that there surely are a significant enrolled in CDSSs, and therefore this issue number of private schools not even known to constitutes one of the major challenges facing them. As such, private school enrollment sta- the government. tistics are not reliable. Overall, PRISAM esti- mates that there are close to 90,000 pupils Enrollment in secondary schools enrolled in the 300 private secondary schools--this figure is significantly higher Enrollment in secondary education increased than MoEST estimates.19 Either way, it is dramatically during the last decade; it almost clear that private schools form a significant quadrupled from less than 70,000 students in portion of overall secondary education provi- 1991/92 to almost 275,000 in 2000 (see sion. In fact, according to the Policy Invest- Table 2.6). The GER increased from 10 per- ment Framework (PIF), particularly for sec- cent in 1993/94 to 27 percent in 200016 (the ondary schools, the private schools are GER is 30 percent in Forms 1­2 and 23 per- considered crucial in expanding supply of sec- cent in Forms 3­417). The NER is estimated ondary education in order to meet demand at about 7­9 percent. The difference between from a growing number of primary school GER and NER highlights the problem of graduates.20 Table 2.6: Absolute enrollment and gross enrollment ratios in secondary18 1991/92 1993/94 1995/96 1998 2000 No. of Students No. of Students No. of Students No. of Students No. of Students Secondary Total 68,428 88,134 138,770 192,272 274,949 Form 1­2 44,888 51,225 87,930 127,336 164,004 Form 3­4 23,540 36,909 50,840 64,936 110,945 Source: Basic Education Statistics. Enrollment Patterns 33 Figure 2.8: Enrollment in secondary education by type of school 200,000 150,000 Gov+grant aided 100,000 CDSS +night school private 50,000 0 93/94 95/96 1998 2000 Source: Basic Education Statistics. Secondary repetition rates Survival/completion in secondary and transition to tertiary Secondary repetition rates are low in Malawi, although repetition does exist in As previously noted, survival rates indicate the both private and public schools. Interesting- share of students who enrolled at the beginning ly, stated government policy does not permit of primary, who survive to the end of second- repetition in publicly funded schools. How- ary (see also Figure 2.6 and Annex 2.8). The ever, repetition does occur and the average survival rate in secondary, as a percent of new repetition rate in all schools, based on the entrants, in Form 1 is 58 percent.22 This num- DHS 2000 data, was estimated as 3.7 per- ber reflects survival through to Form 4; it does cent. In Form 1, the incidence was less than not, however, take into consideration the one percent but it was much higher (19.7 numerous students who fail the MSCE exam percent) in Form 4 due to repetition by stu- and are forced to leave the system without any dents who have failed or are not ready to formal proof of having completed the full sec- take their MSCE exam (average pass rates ondary cycle. are generally low).21 Some students may also Table 2.7 shows that MSCE pass rates repeat in order to improve MSCE grades and declined dramatically between 1993/94 and thereby improve their chance of entering 1999: down from 48 percent to 14 percent, higher education. which affected all types of schools. They sub- When students fail the JCE or MSCE, in sequently increased in 2002 to reach 30 per- some instances they change to a different cent. This poor outcome is attributed--at least school; for example, they may change from a partly--to the increasing numbers of students CSS to a CDSS or a private school. This prac- attending the low-achieving CDSSs. There is a tice runs counter to official policy and so the significant difference between the pass rates at phenomenon is not captured in EMIS statis- the CDSS schools and at the CSS schools (Gov tics. It may be advisable to adjust the EMIS Boarding, Gov Day, Grant Aided), as can be measurements to reflect the prevailing reali- seen in Figure 2.9. ties of Malawian secondary education, name- The meager learning outcomes achieved by ly that repetition does exist and has the CDSS students probably reflects two factors. potential to increase unless the quality is Firstly, the best primary school graduates tend improved. to enroll in the better-resourced CSSs, thereby 34 Cost, Financing and School Effectiveness of Education in Malawi Table 2.7: MSCE pass rates prior years Year 1993­94 1994­95 1995­96 1997 1998 1999 2000 2000* 2002 Pass Rate at MSCE 48% 32% 33% 23% 16% 14% 20% 22% 30% CSS (Gov Board, Gov Day, Grant Aided) 65%+ 55%+ 52%+ 36% 32% 30% 33% 37% ** CDSS + DEC 19% 37% 12% 8% 5% 4% 9% 8% 17% Private 34% 26% 27% 29% 35% 34.5% Pass Rate at JCE 78% 85% 89% 81% 81% 91% n.a n.a n.a CSS + Private 89% 91% 88% 87% 76% 93% n.a n.a n.a CDSS 40% 56% 85% 85% 83% 68% n.a n.a n.a Source: Pass taken from Basic Education Statistics and MANEB and includes external candidates. * Only internal candidates; Average for both the CSS and private schools together ** Gov. boarding 46%, Gov. day 48%, Grant Aided 59% leaving the CDSSs with weaker students who percentage points between 1993/94 and will require more concentrated and skilled 1995/96 at a time when the MSCE pass rate, instruction to achieve MSCE passing level. Sec- for the same cohort of students two years later, ondly, there is very significant under-funding of dropped from 33 percent to 16 percent. Finally, CDSSs relative to other public schools, includ- in some years JCE pass rates have not been con- ing the caliber and qualifications of teachers sistently better at CSSs than at CDSSs, contra- (see also Chapter 4, Figure 4.4), the availabili- dicting the trend for the MSCE pass rate. This ty of teaching and learning materials, and lack of consistency suggests that the JCE fails to available operational budget. provide students with realistic feedback on JCE pass rates present a mixed picture. First- where they stand midway through secondary.23 ly, the pass rate is substantially higher than the A number of explanations have been offered MSCE pass rate. Further, the JCE pass rate has to explain the low MSCE pass rates. The Final remained consistently high even when the Report from the Presidential Commission of MSCE pass rate dropped to its lowest level Inquiry into the MSCE results (2000) pointed (between 1997 and 1999). Indeed, it is quite to: (i) insufficient capacity at MANEB to han- surprising that the JCE pass rate went up by 12 dle the increased number of candidates from Figure 2.9: MSCE pass rates by type of school, 2002 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% Community Private Night Gov Gov Day Grant Total Day School Boarding Aided Source: Pass rates from MANEB and include external candidates. Note: pass rate for night schools is quite high. However, this may be a reflection that in night schools a lot of the students in Form 4 do not enroll to participate in the MSCE exam in the first place, so that those who do enroll are probably the ones better prepared for it. In other types of schools a much higher percentage of those in form 4 participate in the exam. Enrollment Patterns 35 Table 2.8: Secondary school transition rates Year 1993­94 1995­96 1998 1999 2000 2001* 2002 Form 4 Enrollment 18, 910 24, 685 28,984 42,759 50,747** 43,852** n.a. Number of MSCE Passes 7, 871 8, 032 6, 906 6, 180 10, 424 10,998 15,012 Univ. Year 1 Enrollment 921 883 795 833 n.a n.a n.a. Transition to University 7% 12% 10% 12% n.a n.a n.a. Prim Teacher Educ. Enrollment 2, 082 1, 747 + 1,923++ 2,000++ 2,000++ 7,000++ Transition to Teacher Education 16% 24% n.a. 31% 19% 18% 47% Source: Basic Education Statistics and MANEB database. * Data from DRAFT Basic Education Statistics 2001 ** Based upon estimated enrollment, not actually counted since many schools failed to provide data. + Between 1996 and 1998 the numbers changed dramatically due to the introduction of the MIITEP training program, with in total over this period 5 cohorts being trained, with a total of a little over 10,000 students having enrolled. ++ Between 1999 and 2001 MIITEP enrolled only one cohort each year due to resource constraints. In 2002 and 2003 however two cohorts each of 3000 to 3500 students enrolled, and from 2004 on this is expected to increase to two cohorts of 4,250 each, or 8,500 students in total. If the number of MSCE passers remain at the level of 2002 then in 2004 the transition rate to Teacher Education will be 57%. Note: No data on new entrants to first year of university is available after 1999. However, with the total enrollment in university at 3955 in 1999, 3977 in 2000 and 3654 in 2001, it is unlikely that the number of new entrants to university has changed substantially over the past years. Therefore, the transition rate to university will have changed only as a result of the increasing number of MSCE passers. As such, it is likely to have declined. CDSSs and private secondary schools; (ii) the for the next 10­15 years, so the completion problematic examination process with its secu- and pass rates at secondary level need to rity weaknesses; (iii) the need to prepare stu- improve to provide enough candidates for dents better for exams by a) improving their both university education and teacher train- attitude toward education and b) ensuring ing. Although there has been a significant appropriate levels of discipline are not misin- increase in the MSCE graduates since 2000, it terpreted because of increased democracy and is too early to determine if this will be main- human rights awareness; (iv) low teacher tained. morale; (v) lack of teaching and learning mate- rials and poor infrastructure; (vi) lack of school inspection systems; and (vii) an outdat- Teacher training ed curriculum, overly focused on academic skills and preparation for `white collar' jobs In general, the teacher training system is not but too little concerned with creative skills and producing sufficient numbers of new teachers preparation for vocational employment. for primary and secondary schools. When Free Despite an increase in Form 4 enrollment, Primary Education was introduced it was nec- the absolute number of students who passed essary to recruit large numbers of untrained MSCE declined between 19993/94 and 1999. teachers24; now, ten years later, there is still a This trend is particularly worrisome because high proportion of untrained teachers in the of an increasing need for secondary school system adversely affecting quality and learning graduates to enroll in teacher education in outcomes. Similarly, as a result of the transfer order to improve the education system as a of the DECs into full-fledged secondary whole. In 1998, the supply of MSCE gradu- schools (CDSSs), the government has had to ates was close to the need for teacher educa- hire untrained teachers--mostly primary tion entrants (due to increased teacher training teachers--to teach in the CDSSs and this has levels as a result of the MIITEP training pro- resulted in an additional drain on the pool of gram), and the spaces at first year of universi- qualified primary teachers while simultaneous- ty. The need for both primary and secondary ly increasing the number of under trained sec- education trained teachers will remain high ondary teachers in the system.25 36 Cost, Financing and School Effectiveness of Education in Malawi Primary teacher training Table 2.9 shows that the MIITEP cohorts graduated during the years 1997 to 2000 at an In the period following the introduction of FPE, average rate of slightly over 2,500 per year, enrollment in primary rose from 1.8 million (in and from 2001 to 2003 only one cohort grad- 1992/93) to 2.9 million (in 1994/95). As a uated each year because of funding shortfalls. result, demand for additional primary teachers This completely negated the rationale of increased inexorably. At the time, however, MIITEP since it trained the same number of annual output from the six existing teacher teachers as pre-MIITEP. In addition, it resulted training colleges was limited to approximately in teachers receiving significantly less formal 2,700 teachers. Consequently, MOEST recruit- training, left TTCs empty for substantial peri- ed 18,000 school leavers, and the ensuing chal- ods of the year--while tutors continued to lenge of providing some training for these receive their salaries. untrained recruits resulted in the birth of the The intake from 2002 onward has been 2 Malawi Integrated In-Service Teacher Educa- cohorts per year and so from 2004 on approx- tion Program (MIITEP).26 It is anticipated that imately 4,800 graduates can be expected to this high volume program of teacher education, graduate.27 It is further anticipated that the combining face to face instruction and distance total training capacity of the combined TTCs education, will need to continue for at least will shortly expand from 3,000 to 4,250 teach- another five to ten years in order to train the ers through GTZ funding and that total output unqualified teachers currently in the system, will increase to 6,800 students per year by increase the teacher stock to cater for primary 2006. enrollment increases, and replace teachers lost due to retirement, death or job change (attri- Secondary teacher training tion). The annual attrition rate, currently esti- mated at approximately 6 percent (3 percent Secondary teacher training is provided at death related, 3 percent retirement and dropout Domasi College of Education, at the Universi- related), results in the loss of 3,000 teachers per ty of Malawi (Unima), at Mzuzu University year--there are currently approximately and at the private institution, the African Bible 50,000 teachers in the system. College. Interestingly, the recommended PIF There are six publicly funded primary goal for a secondary PTR of 40:1 is currently Teacher-Training Colleges (TTCs) in Malawi. being achieved. However, the PqTR is 120:1 Two of these colleges belong to churches and (EMIS 2001). As previously mentioned, the receive a block grant. The total enrollment majority of the unqualified secondary teachers capacity is approximately 3,000 and this has are trained primary teachers, teaching at the remained largely static during most of the CDSSs. Total output from the combined train- nineties. The restricted output of teachers ing institutions remains low; it is currently esti- explains the rationale for introducing the high mated at approximately 350 to 400 teachers volume MIITEP crash program. per year.28 Domasi College also has a parallel Table 2.9: Primary teacher training graduates over time Year 1991/92 1993/94 1995/96 1997 to 2000 2001 2002 2003* 2004* 2005* 2006* Number of 10379 1923 2706 2000 4800 4800 6800 Graduates 1866 2666 2528 (cohorts (cohort (cohort (cohort (cohorts (cohorts (cohorts 1 to 5) 6) 7) 8) 9 & 10) 11 & 12) 13 & 14) Source: Basic Education Statistics till 1995/96, DTED thereafter. * These are expected numbers. Enrollment Patterns 37 Table 2.10: Secondary education teacher graduates, Unima and Domasi 1995 1996 1997 1998 1999 2000 2001 Unima B.A. Education 73 91 230 122 92 154 104 Domasi 229 224 225 None 157 161 127 Source: Basic Education Statistics for Unima. Document "Domasi--Capacity Building August 14, 2002" for Domasi. Note: Domasi had no graduates in 1998 because of the switch from a 2 year system to a 3 year system. Therefore the number of yearly graduates reduced starting 1999, since the enrollment capacity remained the same. The program was modified because there was "a lack of adequate content knowledge on behalf of the graduates ... therefore lacking confidence." There was also a switch from enrolling primary school teachers with at least 2 years teaching experience to enrolling MSCE graduates with good grades but no teaching experience. teacher training system in place, the Secondary only enrolled 1,500 students up to 2000, and School Teacher Education Project (SSTEP) pro- between 2000­2003 a total of 1,723 males and gram. It upgrades trained primary teachers 883 females graduated. The typical duration of who are already teaching at CDSSs.29 a formal program is four years. Private training providers are active in a variety of areas, particularly in commercial but Technical and vocational training also in technical fields, and the training pro- grams consist of short and long-term, full-time The technical and vocational training sector in and part-time courses. Trainees pay full fees so Malawi has been characterized by limited courses are primarily attended by the more capacity, a supply driven approach, minimal affluent school leavers or by employees in the involvement of the private sector, a narrow formal sector whose fees may be covered in focus on the formal sector, fragmentation of full, or partially, by the employers. The PRSP providers and training schemes and a diversity estimated a training capacity of 18,000 places of assessment/certification systems. Various within the private training providers. public and parastatal institutions--mainly There are other non-formal training institu- under the control of the Ministry of Labour & tions; churches and welfare organizations Vocational Training, the Ministry of Educa- manage many of these and they invariably tar- tion, the Ministry of Commerce and the Min- get school leavers. These institutions are fre- istry of Gender, Youth, & Community Ser- quently located in rural regions where students vices--provide training in conjunction with a have few opportunities to enter further formal number of private providers, for example, reli- training. The traditional apprenticeship train- gious mission training centers, social institu- ing, referred to as `master-craft training', is tions and local and international NGOs.30 In possibly the most important sub-system in 1999 an independent secretariat called Techni- terms of the numbers of trainee places they cal, Entrepreneurial, Vocational Education and provide. The courses are completely self-organ- Training Authority (TEVETA) was established. ized through on-the-job training and financing The main aim of TEVETA is to link public sec- arrangements, and course duration varies from tor initiatives more closely with the private sec- trade to trade and from employer to employer. tor. More details about Technical and Voca- tional Training programs can be found also in Annex 2.12. Traditional public technical and vocational Notes training in Malawi consist mainly of two types: 1. University level education provides certifi- an apprenticeship training system; or full-time cate, diploma, degree and Master's level cours- institutional training. The training programs es of 2 to 7 years duration. 38 Cost, Financing and School Effectiveness of Education in Malawi 2. At the end of 2003 the latest available 8. First, the ratio is inflated where repetition school statistics was the draft version of the is common. Second, due to dropout enrollment 2001 Basic Education Statistics. However, due levels tend to be higher in the lower grades and to difficulties with the data collection for sec- therefore the ratio, by averaging enrollment ondary and uncertainties with the subsequent across all grades, over-emphasizes enrollment estimation of enrollment in secondary, it was in the higher grades. Third, the same ratio can decided to use the 2000 enrollment in most describe two very different situations; one of cases. high initial entry and high dropout, and anoth- 3. For secondary enrollment, the number is er of lower initial entry and lower dropout. the estimated enrollment (as found in Basic 9. School census data are used for analyses Education Statistics 2000) and is again calcu- of trends over time, because these are available lated by imputing average enrollment numbers for a longer period than the household surveys. (averages calculated per type of school) for the 10. DHS 2000: Boys 40.2%; Girls 35.9%. schools that did not provide statistics--only 77 Ed Data 2002: Boys 22.1%; Girls14.2%. percent of registered secondary schools provid- 11. DHS 2000 registered an average ed the required statistics. dropout rate for rural areas of 5 percent versus 4. Source: Economic Report, 2002. This 2 percent for urban areas while Ed Data 2002 Ministry also provides assistance to communi- reported 9 percent rural versus 3 percent ty childcare programs and approximately urban. While dropout in lower primary among 118,000 children participate in these pro- girls and boys is approximately the same grams. (3.6% girls, 3.9% boys), the picture is quite 5. Noted in both the DHS 2000 and Ed Data different for upper primary (9.2% girls, 6.8% 2002 surveys. boys)--this suggests that the pressure in much 6. The significant variance between the age greater on girls to drop out. estimates of children in school from the 12. The PCR is calculated as the number of school census and household surveys also children successfully completing a cycle make the Net Enrollment Ratio (NER) an expressed as a percentage of the number of unreliable indicator if based solely on the children of the corresponding age cohort. For school census. The household surveys indicate the entire Malawi primary cycle, this is best an NER of 81 percent in 2002 (Ed Data), up calculated by the PSLCE passers divided by the from 78 percent in 2000 (DHS). When com- population aged 13 years. paring this to the GER level found (115% by 13. A school is considered `long established' Ed Data), it indicates that for every 81 pupils if the year of establishment was before "2000 age 6­13 in primary, there are 34 pupils who minus number of available grades minus 5"; or are either younger than age 6 or older than if the year of establishment was not given in age 13. the dataset (here 291 schools did not have a 7. By using enrollment numbers from the year of establishment). school census and the 1998 population census 14. In reality, a number of primary school data, the GER is 132.6 percent and significant- teacher training entrants enroll with only the ly higher than the results from any of the other JCE as their highest level of qualification even possible sources. For instance, using household though official government policy insists the surveys over the last five years (IHS 1998, DHS MSCE is a prerequisite. 2000, EdData 2002) gives a GER between 107 15. The DECs originally existed to provide percent and 120 percent. secondary education using distance education Enrollment Patterns 39 modalities, i.e. with learners' reading materi- 21. It is 3.1% in Form 2 and 1.5% in Form als printed by the MCDE center and with a 3. However, Form 2 ends with a formal exam- small number of teachers at these centers guid- ination to decide who progresses to Form 3. ing learners and occasionally grading their 22. Based on school census enrollment, DHS work. 2000 repetition, UN-EFA population. Using 16. The household surveys recorded similar school census enrollment and repetition, and levels; the IHS 1998 finding reported GER at NSO population, survival is 67%, but that fig- approximately 27% and the NER at 7%, while ure disregards the practice of secondary repeti- the Ed Data 2002 found respectively 27% and tion. However, if only DHS 2000 data is used 9%. the figure is 54%. 17. For GER calculations the corresponding 23. Many Malawians anecdotally refer to population is 14­17 (Form 1­2 with age the JCE as the "Joker Certificate". This indi- 14­15, Form 3­4 with age 16­17). cates the decreasing value of the qualification 18. In 2000 only 77% of the schools and its inability to provide the level of skills the responded to the data gathering questionnaire market demands. However, its existence raises and many publicly funded schools did not pro- an additional question: if conducting this vide any data. As such, the actually counted examination is expensive and disruptive to the total enrollment was only 164,459. The num- system as a whole, is there any merit in contin- ber used, here, is the one Basic Education Sta- uing with it? tistics 2000 estimates as the enrollment by 24. Teacher:pupil ratio increased from 1:84 imputing enrollment numbers for the missing in 1991/92 and 1:78 in 1992/93 to 1:108 in schools based upon the average enrollment for 1994/95. Source: Basic Education Statistics, that type of school. various years. 19. MoEST only includes registered schools. 25. MoEST increasingly prefers the term 20. The rationale for establishing PRISAM `under trained' since the majority are trained in 2000 was "to promote and improve private primary teachers. education services in Malawi and safeguard 26. MIITEP is a 2-year course, which com- the interests of all those concerned with the bines college and field based support and provision." The key objectives are: (i) allows multiple cohorts of 3,000 to 3,500 per improve the quality of education service in annum use the facilities of the six colleges. private schools (e.g. assist in teacher training, Theoretically up to three cohorts could gradu- conduct inspections, etc.); (ii) act as a coordi- ate from the program each year but, to date, it nating body for the efficient and effective has not succeeded in generating this volume for operation of private schools; (iii) control and a myriad of reasons. regulate the establishment and operations of 27. Based on an estimated pass rate of 80 private schools (e.g. facilitate registration of percent. schools with MoEST and the Malawi Nation- 28. Domasi supplies roughly 150, Chancel- al Examination Board to ensure minimum lor College, 100­150, Mzuzu University 20 standards are adhered to); (iv) assist in (but expected to increase to 100) and the acquiring donor and government finance (e.g. African Bible College 15). The Polytechnic and `soft' loans for infrastructure improvements); the Banda Agriculture Faculty at Unima do not (v) safeguard private schools against counter- specifically train teachers about 20 and 15 productive government legislation and regula- graduates respectively join the teaching force tions. each year from these two institutions. 40 Cost, Financing and School Effectiveness of Education in Malawi 29. It is a distance learning program of three program is to enroll approximately 300 MSCE years duration and presently three cohorts of graduates each year to train as secondary 300 are enrolled. There are also plans to con- teachers. tinue with some additional form of distance 30. See Promotion of non-formal Technical learning program at Domasi in the future but Entrepreneurial Vocational Education Training once the current untrained backlog at CDSSs is components in Malawi.by TEVETA (no refer- eliminated. The stated objective of the SSTEP ence year). CHAPTER 3 Education Expenditures and Financing T he preceding chapters raised concerns as such, a complete picture of education about the efficiency and effectiveness financing requires data from all three sources. of the education system. This chapter At the moment, data from the three sources is updates and supplements the data either incomplete or not up to date. Accurate and analyses that have been presented in prior data on budgetary expenditures are available reports on education financing in Malawi. The but there is a delay of two or three years. A main ones referred to are: (i) the education large part of the financing provided by external chapter of the Public Expenditure Review of the donors is off budget and hence not readily Government of Malawi (MOFEP 2001); and (ii) available. Details of household education the Policy and Investment Framework for the expenditures are available but only for the pri- Education Sector (MOE 2000). The chapter mary level for the year 2001. Therefore, the begins with a brief review of the budgetary sys- analysis of education financing and expendi- tem and reporting concepts and then provides tures has to be pieced together from different an analysis of the trends in budgetary expendi- sources of variable quality and some issues pre- tures on education and broad sub-sectoral allo- sented in this chapter might require further cations. This is followed by a detailed analysis investigation. of the components of recurrent expenditures, the trends in unit costs and decomposition of the structure of unit costs. The economic com- The budgetary system in Malawi position of developmental expenditures and the volume and composition of external donor The government budget captures public expen- financing, both through and outside the budget, ditures for the education sector under the are presented next, together with a discussion of recurrent budget and the development budget. the implications of these trends for the sustain- The recurrent budget comprises expenditures ability of investments in the sector. Finally, that are financed by tax revenues, non-tax rev- household level expenditures at the primary enues and grants, while the development budg- level are discussed. et comprises spending under multilateral loans Education in Malawi is financed by the gov- and government's local contribution. Expendi- ernment, external donors and households and, tures under the development budget are most- 41 42 Cost, Financing and School Effectiveness of Education in Malawi ly capital expenditures but some recurrent The budget provides public resources to the expenditure are also included in external proj- MOEST, which oversees primary and second- ects, which may need to be subsumed in the ary education, as well as non-university regular recurrent budget after the project peri- teacher training colleges.3 The subvented od (for instance, salaries of project administra- organizations receive funds directly from the tive staff or the cost of textbooks). Therefore, Ministry of Finance (MOF).4 unit recurrent costs based on the recurrent budget may result in an underestimate of the true costs of maintaining the education system. Overall budgetary expenditure trends More importantly, the development budget in the education sector does not capture grants from bilateral donors. In 2000/01, these off-budget grants accounted As stated in Chapter 1, real recurrent expendi- for about 70 percent of total development tures in the education sector have risen gradu- expenditures. A detailed breakdown of these ally between 1993/94 and 2001/02, while real expenditures, by economic categories, is not annual development expenditures approxi- readily available. mately halved over the same period, perhaps The data on budgetary expenditures must due to more off-budget development expendi- also be interpreted with caution because of the ture. Total real educational expenditures, as differences between approved, revised budgets shown in Figure 3.1 below, have remained and actual expenditures that spans over a 2­3 largely at the same level. year period for the same fiscal year. Budgetary However, there have been considerable instability, associated with high rates of infla- annual fluctuations in real recurrent expendi- tion and depreciation of the Malawi Kwacha, tures. From 2001/02, recurrent expenditures together with limited budget management have increased significantly, and though offi- capacity, contribute to substantial discrepan- cial figures are not available yet, it is expected cies between these estimates. that the higher level of recurrent spending on Finally, aggregate annual expenditures do education has been maintained the last couple not indicate the real difficulties of providing of years. The share of MOEST in total recur- stable financing for effective service delivery rent education expenditures has been in the and financing of critical inputs during the range of 72­80 percent. As can be seen from school year. Budgetary instability in allocations Figure 3.2, the growth in real recurrent expen- and in actual releases to the Ministry is partly ditures has been mostly attributable to the associated with periodic natural shocks expenditures of MOEST; those of subvented (droughts, floods), which reduce agricultural organizations have shown very modest growth. production and hence revenues, while raising The real growth in recurrent expenditures demands for expenditure for relief.1 for MOEST has largely been driven by the External agencies contribute about 90 per- increase in personal emoluments as a result of cent of the resources for the development the increase in the number of teachers after the budget (about 95 percent of total on- and off- introduction of FPE. The increase in real budget development expenditure) and about expenditure on non-salary expenditures had one third of the education budget as a whole. been modest up to 1999­2000 (and thereafter While budgetary instability has been a chronic only unreliable estimates are available). Expen- feature, the suspension of foreign grant money ditures on other inputs required to improve and loans (in 1992 and 20002) have further education quality have not kept pace.As shown added to the volatility of expenditures in the in Figure 3.4 below, recurrent expenditures by sector. subvented organizations (mainly universities Education Expenditures and Financing 43 Figure 3.1: Real recurrent and development expenditures, 1993-2003 (millions of kwacha) 700 600 500 400 Millions 300 200 100 0 Recurrent Development Total 1993-94 1994-95 1995-96 1996-97 1997-98 1998-99* 1999-00 2000-01 2001/02 2002/03 Source: Authors' calculations, using PER and GOMM budget data. and colleges) have, on average, increased over expenditures on other subvented organizations the past five years, despite a decline during two have remained very low in absolute terms. years. In the preceding period, between Development expenditures of the MOEST, 1993/94 and 1996/97, their funding had as shown in Figure 3.5 below, comprising the declined steadily. Not surprisingly, the overall overwhelming share of the development growth over the last decade has been compar- budget for education, have declined sharply atively lower for the subvented institutions between 1993/94 and 1999/2000--mainly as (universities) than for basic education sector. a result of the decline in external funding. As will be discussed in more depth in chapter Development expenditures for subvented 7, real expenditures on tertiary institutions organizations have stagnated at very low have driven this growth. The actual (real) levels. Figure 3.2: Real recurrent expenditures: MOEST and subvented organizations 700 600 500 400 Millions 300 200 100 0 Recurrent Development 1993-94 1994-95 1995-96 1996-97 1997-981998-99* 1999-00 2000-01 2001/02 2002/03 Total Source: Authors' calculations, using PER and GOPM budget data. 44 Cost, Financing and School Effectiveness of Education in Malawi Figure 3.3: MOEST recurrent expenditures in real terms 450 400 350 300 250 200 150 100 50 MOEST 0 recurrent MOEST 1993- 1994- 1995- 1996- 1997- 1998- 1999- 2000- Salary 94 95 96 97 98 99* 00 01 Non-salary Source: Authors' calculations, using PER and GOPM budget data. Sub-sectoral allocations cations to secondary education, which fluctu- ated around 10 percent between 1993/94 and Allocations for primary education have com- 1998/99, have increased to about 15 percent. prised the largest share of recurrent expendi- Three other features of the sub-sectoral allo- tures on education, and, until the last few cations of recurrent expenditures need to be years, the overwhelming share of development highlighted. First, the allocation for tertiary expenditures (See Annex 3.2). The share allo- education is substantially higher than that for cated to primary education of total recurrent secondary education, although the latter serves expenditures rose from 50 percent in 1993/94 a much larger number of students (about 70 to approximately 60 percent in 1995/96. Allo- times more students). Second, the overall sub- Figure 3.4: Real recurrent expenditures of subvented organizations 140 120 Millions 100 80 60 40 20 0 1993-94 1994-95 1995-96 1996-97 1997-98 1998-99* 1999-00 2000-01 2001/02 2002/03 Subvented recurrent Tertiary Subvented other Source: Authors' calculations, using PER and GOPM budget data. Education Expenditures and Financing 45 Figure 3.5: Real development expenditures 300 250 200 150 Millions 100 50 0 1993-94 1994-95 1995-96 1996-97 1997-98 1998-99* 1999-00 2000-01 2001/02 2002/03 MOEST Development Subvented - develpment Source: Authors' calculations, using PER and GOPM budget data. sectoral share for secondary disguises the very shown later, changes in on-budget spending are low share for CDSSs compared to the relative- a result of complementarities with the sub-sec- ly high share for CSSs--as mentioned in the toral off-budget spending by external agencies' previous chapter. The result is a severe under- spending, as well as a result of changing prior- funding of general secondary education. Third, ities over time (both for the donors and the the share for teacher education is 3 percent or government). less, raising the question of whether allocations to teacher education are consistent with the government priority of improving the quality Composition of recurrent expenditures of teachers. This is particularly relevant because it has been claimed that budget short- The breakdown, or functional composition of falls were the main reason for recruiting only recurrent expenditures, indicates what these one cohort of teachers to the TTCs each year expenditures financed. The share of expendi- between 1999 and 2001, when MIITEP was tures allocated to personal emoluments is high- originally intended to facilitate at least two est in primary by a wide margin, fluctuating cohorts per year. between 86­91 percent between 1993/94 and The analysis of development expenditure is 1999/00. This high share of emoluments has strongly influenced by the separation of on- fueled the growth in overall MOEST salary budget and off-budget spending since the share expenditures. In the other sub-sectors, the of off budget is substantially larger--although share of emoluments is surprisingly low; it is it is difficult to obtain complete information on generally less than 50 percent of total recurrent this. However, a review of the on-budget devel- expenditures. The share of non-personnel opment expenditure by sub-sector indicates expenditure has been rising in all sub-sectors that allocations, within the developmental on- other than primary. budget expenditures, were concentrated up to While overall non-personnel expenditures 1998/99 on primary education. In recent years, (ORT) have risen only modestly in real terms, however, the level has dramatically shifted the sub-sectoral trends in these expenditures towards secondary education. As will be require further analysis. First, in primary edu- 46 Cost, Financing and School Effectiveness of Education in Malawi Table 3.1: Share of personal emoluments in recurrent expenditures 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 Primary 91% 83% 90% 86% 87% 86% 83% Secondary 60% 56% 40% 30% 58% 40% 21% Teacher Ed 39% 33% 31% 28% 23% 41% 38% Tech &Voc Ed 34% 22% 17% 14% 9% Not appl Not appl Tertiary 42% 44% 35% 32% 25% 28% 29% Source: Authors' calculations, using PER and GOPM budget data. Notes: Secondary Education is very difficult to estimate because of the duality of the CSS system versus the MCDE/CDSS system. The P.E. calculations, here, exclude the expenditure for MCDE (since it was easily separated). However, from the time nearly all of MCDE was transferred into CDSSs in 1997/98) the expenditure on CDSSs has been captured directly within the standard secondary education expenditure and this is the expenditure reflected here. However, to further complicate matters, it may be that salaries for at least some of the teachers at CDSSs ­ since they are trained primary and not secondary school teachers ­ are captured under primary expenditure and not under secondary expenditure. cation the share of ORT seems too low, and spending per pupil has more or less returned within this category, the share of expenditures to pre-FPE levels.8 for teaching and learning (T&L) materials has In secondary education as a whole, per pupil been 5 percent or less.5 The scarce non-salary expenditure has decreased dramatically. The expenditures also cover items such as funeral sub-sector has seen enormous increases in costs and the travel of civil servants. In some enrollment at CDSSs, but with little additional reported cases, the entire ORT has been used resources allocated. Spending per pupil has to cover funeral costs that arise from also declined slightly for CSSs as enrollment HIV/AIDS related deaths. Second, in all the has increased. Additionally, in recent years other sub-sectors, the share of ORT is relative- there has been a gradual shift from boarding to ly high and constitutes a rising share of total day schooling, and cost sharing was intro- recurrent expenditures. A detailed breakdown duced at the remaining boarding schools. of these expenditures shows that, historically, Overall real spending per pupil in 1999/00 was the largest share accrues to either boarding6 or less than half of what it was in 1993/94. For an ambiguous "others" category. The share for TTCs the drastic fall in real unit cost is at least teaching and learning materials ranged partially a result of a change in the delivery sys- between 5­7 percent in most sub-sectors, tem from the traditional pre-service system to except in tertiary education where it reached the MIITEP program.9 As such, this is an 11 percent. The "others" category includes example of improving resource efficiency. mainly travel allowances (the government has At university level, the real unit cost has one of the highest travel budgets as a share of increased over the same period. A comparison GDP in the world).7 of the averages for the past three years (98/99­00/01) versus that of the first three years (93/94­95/96) shows per pupil spending Trends in unit public expenditures has increased more than 20 percent. The over- all trend of rising real unit costs at the tertiary Following the enormous increase in enroll- level is in marked contrast to the trends in ment after 1994, real unit expenditures other sub-sectors and will be discussed in more declined in primary education due to the detail later in the study. insufficient number of teachers and the subse- The per pupil spending as a multiple of the quent rise in PTR. With the hiring of many unit cost of a primary student in Table 3.3, untrained teachers, this situation was rapidly reflects the vast disparities in the level of serv- altered and since 1997/98 real recurrent ice provision and funding. For instance, each Education Expenditures and Financing 47 Table 3.2: Per pupil annual recurrent expenditure (in 1993/94 kwacha) 93/94 94/95 95/96 96/97 97/98 98/99 99/00 2000/01 Primary 94 58 83 82 108 93 95 61 Secondary 589 356 322 252 266 181 276 n.a. Teacher Ed 3546 3946 4482 1739 1051 418 n.a. 1205 Tertiary 19056 18457 20049 19471 28364 22488 26572 21355 Sources: PER for data until 1998/99. For 1999/00 primarily from Basic Education Statistics. For 2000/01 from `Consolidated annual appropriation accounts'. year students in CSSs receive about 10 times 3.3, below. At the primary level, the annual the public aid that primary students do. Includ- current public expenditure on each student ing the CDSSs, however, this ratio is closer to represents a mere 5 percent of annual per capi- 5, indicating the low level of resource provi- ta income, which would undoubtedly need to sion in CDSSs. The discrepancy in the level of be raised if quality is to improve. Average funding of CDSSs and CSSs needs to be expenditure per secondary student is about 21 addressed if the quality is to improve. These percent of per capita income11 and a universi- wide differentials in quality represent ty student receives about 10 to 15 times the per inequitable use of resources--especially as a capita income. higher proportion of the poorer students Annex 3.3 shows the ratios in comparison attend CDSSs. to other developing countries and indicates On a per student basis, secondary and terti- that while Anglophone countries have higher ary education are undoubtedly always more ratios than other regions, none approach the expensive than primary. However, the ratios disparity in per pupil expenditures apparent in found in Malawi for CSS, and especially for Malawi. In Malawi the absolute level of per universities, are well beyond any rational basis pupil expenditures is actually low at all levels in terms of quality norms, as can also be seen except for tertiary (see Annex 3.4). At the pri- from comparative data in Annex 3.3. At the mary level the annual recurrent expenditure tertiary level, a student receives 300 times was only US$14 in 1999/2000, compared to more resources than those allocated to a pri- US$3712 in Sub-Saharan African countries. By mary student (2000/01).10 contrast, the annual expenditure on tertiary Another measurement is the unit cost, or education has been largely within the spending per pupil, as a share of annual per US$2,500­3,000 per student range and with capita income of Malawi, as shown in Table even higher levels during peak years. Table 3.3: Unit costs for a year of education, by level of education, 2000/01 Recur. Expend. Unit cost As multiple As prop of 2000/01 (MK `000) Public enroll in MK of GDP/cap spending in prim Primary 1,511,514 3,143,427 481 0,04 1.0 Secondary public 354,954 153,119 2318 0.21 5 Techn/voc 67,521,205 1,504 44,894 3.55 83 Prim. teacher training 56,619 6,000 9,437 0.84 20 Domasi (non-univ sec teach training) 29,174 540 54,026 4.81 112 Univ 625,930 3,743 167,227 14.89 348 Source: Authors' calculations, using 2000/01 actual expenditure data. 48 Cost, Financing and School Effectiveness of Education in Malawi Decomposition of public unit costs mous. Spending on teaching and learning in 2000/01 materials and other expenses per pupil is approximately 50 and 540 times higher, Expenditure per student is a measure of the respectively, at the tertiary level than at the public resources that, on average, a country primary level. devotes annually to each student's education. At the primary level, unit public recurrent To understand when an increase or decrease is costs represented only 4 percent of per capita an improvement in efficiency, it is important to income. Average teachers' salaries are about understand the composition, or breakdown, of 4.5 times per capita GDP, slightly higher than the unit cost. In particular, the variations in the the 3.5­3.8 multiples of per capita GDP in relative price of the main inputs (e.g. teachers' many other Sub-Saharan countries. Most of salaries) and the quantity of inputs devoted to these other Sub-Saharan countries, however, education (teachers and non-teaching staff per have higher GDPs per capita, so the current pupil or learning materials per pupil) affect the salary level for primary teachers' may well be per student expenditures.13 appropriate for Malawi14 The low public unit In Table 3.4, the factors explaining the sig- cost results from the very high pupil­teacher nificant difference between primary and uni- ratios--relative to international standards-- versity unit costs are: (i) the very low and from the negligible spending on recurrent pupil­teacher ratio at tertiary level; (ii) a simi- items such as teaching and learning materials. larly low pupil­other staff ratio; and (iii) a high In summary, the differences in teachers' level of spending on "other recurrent" costs. remuneration across sub-sectors do not The differential in average teacher salaries account for the huge disparities in unit costs. across the sub-sectors is not a significant con- The high differentials are more directly linked tributing factor in explaining the differentials to the differences in organization of service in unit costs. The primary student equivalent delivery and spending on non-salary expendi- of tertiary education non-salary expenditures tures. In secondary education, an increase in in 2000/01 was 347 times that of non-salary public spending on CDSSs has to be consid- spending in primary education. In contrast, the ered. At the tertiary level, the fixed costs of the ratio of average teacher salary to the average system, relative to the limited total student university salary was only 4 times. The population, need to be addressed. While some pupil­teacher ratio at the university level is restructuring of PTRs to eliminate gross ineffi- almost one-tenth of that in the primary sub- ciencies (for example, courses with low sectors. The differential in the ratio of pupils to demand) is required, the overall issue is to raise other staff between the two sub-sectors is enor- the scale of operation. A strategic question for Table 3.4: Decomposition of public unit costs, 2000/01 Secondary- Teacher Primary conventional education University Pupil­Teacher ratio 66 23 41 7 Teacher average salary (USD) 872 1522 2511 3843 Pupils per Other staff 1943 65 38 6 Per pupil spending on T&L materials (USD) 1.9 6.0 0.0 93.8 Per pupil spending on Other (USD) 2.7 29.2 682.1 1439.5 Memo: Teachers' salary as a multiple of per capita income 5.95 10.39 17.15 24.71 Source: Authors' calculations using 2000/01 actual expenditures. Education Expenditures and Financing 49 Malawi is how to reorient higher education age emoluments per staff, data on the number and expand access, thereby reducing unit cost, of staff are required. In the secondary sector it while keeping total recurrent public cost from is difficult to estimate staff numbers due to rising significantly. As will be shown in Chap- lack of transparency in the data on staff levels, ter 7, there should be scope for increasing the particularly at the CDSSs. A lot of teachers number of university graduates without over- teaching at CDSSs are primary school teachers supplying the job market. A recent tracer study who are often still registered, and paid, as if indicated that university graduates quickly find they were teaching in primary schools. Obvi- formal employment jobs, thus suggesting that ously this distorts staff numbers at both pri- demand still outruns supply. mary and secondary levels. There has not been sufficient clarity by the central administration toward the divisions and lower levels on how Trends in sub-sectoral emoluments exactly to handle this consistently. Table 3.5 indicates that per person staff This section examines possible reasons for the emoluments in the primary and tertiary sectors decline, or stagnation, of real unit recurrent declined in real terms during the nineties. In costs in primary, secondary and teacher train- the primary sector, average salary expenses ing--interestingly, these costs have tended to were 20 percent lower in 1999/00 relative to rise for tertiary education. The focus is on 1993/94. This decline in real average emolu- emoluments, or salary expenditures. Non- ments does not necessarily indicate that the salary expenditures, or ORT expenditures, are teachers' salary levels have declined; it is due in classified under the broad and somewhat part to the hiring of a large number of ambiguous category of "other". Present untrained teachers but at lower salary levels. accounting systems and the lack of consistency As such, when these teachers receive training in data entry make further analysis of this cat- and achieve the same salary scales as their egory difficult. However, as ORT expenditures trained counterparts, average emoluments per constitute more than 50 percent of total recur- teacher are expected to rise (discussed further rent expenditure at post-primary levels, the in Chapter 8). earlier analysis of the impact of ORT spending In the tertiary sector the decline in average is a better predictor to explain differences real emoluments per staff has been even more between the sectors. Here, too, further analysis significant: between 1993/94 and 1999/00 it of these expenditures would be beneficial and fell by approximately 50 percent. As before, possibly contribute to improved resource uti- these figures do not imply that the salary level lization. of existing tertiary level teachers have neces- In recent years, total real emoluments have sarily declined; rather they indicate that there increased in only the primary sub-sector; in all has been an increase in the number of staff at other sub-sectors real emoluments have lower salary levels, including both teaching remained largely unchanged. To estimate aver- and non-teaching staff. Table 3.5: Average emoluments per staff (in 1993/94 kwacha) 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 Primary 5,822 2,986 4,379 4,299 5,628 5,105 4,786 Tertiary 91,477 64,158 50,262 47,951 44,940 40,441 46,345 Source: PER and authors' calculations. Note: In Primary nearly all staff are teachers, whereas in tertiary this is not the case. 50 Cost, Financing and School Effectiveness of Education in Malawi Table 3.6: Growth of teaching staff and pupil­teacher ratios 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 Teachers Primary 27948 45775 49138 47370 46577 45784 47870 Tertiary 321 459 486 440 509 526 560 P­T ratio Primary 68 62 59 61 60 63 61 Tertiary 12 8 7 8 6 6 6 Source: PER and authors' calculations Table 3.6 shows that it is mainly the growth Donor financing and extra budgetary in teacher numbers at both the primary and grants tertiary level that has contributed to the absolute growth in personnel costs; it is not Donors play a critical role in the education sec- due to an increase in salary levels. In the case tor. They contribute in the region of 80­90 per- of the primary sector, there was an addition of cent of funds in the development budget. They 20,000 teachers between 1993/94 and also contribute extra-budgetary grants that 1999/2000, but since enrollment has increased amount to approximately double of the devel- rapidly due to FPE, the PTR declined only opment budget for 1999/00 and 2000/01 (see from 68 to 61. In the tertiary sector, there was Annex 3.5). In total, roughly 95 percent of a similar percentage increase in the teaching total sector development expenditures are staff (75 percent) but since enrollments actual- financed externally while the remaining 5 per- ly declined over the period, the PTR went from cent of domestic financing comprises govern- 12:1 to 6:1. This trend in the tertiary sector ment's counterpart funding. As such, donor indicates a decline in efficiency. projects and programs determine the nature of educational sector development and influence costs, efficiency and future recurrent expendi- Development expenditures ture requirements. Since data on extra-budgetary grants are Composition not regularly compiled, or universal in their coverage, it is not possible to say whether On-budget development expenditures consist total, or sub-sectoral allocations, are in line primarily of investment, or capital, expendi- with policy priorities. However, the partial tures. Between 1993/94 and 1999/00, con- data available do clearly indicate that sub- struction or civil works has typically absorbed sectoral allocations for off-budget grant approximately 60 percent of all development assistance still strongly favor primary educa- spending; furniture another 5 percent; rehabil- tion, in contrast to the high share of second- itation about 4 percent, pedagogical materials ary in the development budget. Therefore, accounted for 9 percent (but varied between 1 the estimate is that about 40 percent of total to 25 percent), training allocations required developmental expenditures (including off- about 6 percent (but varied between 3 to 11 budget) go to primary, compared with 35 percent), and `other' absorbed 19 percent.15 percent to secondary. This is considerably Overall, the share of development expenditures different to the distribution by sub-sector, as in the total education budget is approximately discussed earlier, of total expenditures for 25­30 percent annually. education. Education Expenditures and Financing 51 Although precise estimates of extra-budget- terms. This is because access to the upper lev- ary donor financing cannot be derived, some els of education is highly inequitable and large- indication of the magnitude in specific areas ly benefits the higher income quintiles. In the can be gauged from the following data: total following section the characteristics of house- annual average donor expenditure (both on- hold spending at the primary level are and off-budget) was approximately US$37 mil- explored. lion at the end of the nineties. Projected expen- As much as 83% of parents of primary ditures by the main donors during the period pupils in public school paid for one or more 2001­2007 are given in Annex 3.6. types of school supplies during the year 2001. Government is reliant on a wide range of The majority of families reported paying for donors and organizations, and this makes sec- textbooks and uniforms and approximately tor planning, increased efficiency and equitable half reported contributing to the school devel- allocation of resources difficult. First, donor opment fund, which finances maintenance. financing of education has been, and remains, Surprisingly, the average household expendi- volatile. Second, external financing is the pri- ture for public primary school was almost as mary determinant of the level and composition high (80%) as the per pupil public expendi- of non-salary operating expenditures (such as ture. As Table 3.7 shows, the household expen- textbooks and teaching-learning materials) and ditures for primary education are also regres- these would normally be financed under the sive with the lowest quintile paying more in public recurrent budget. The volatility and absolute terms than the richer income groups unpredictability of these funds not only affects (with the exception of the wealthiest quintile); planning but also impinges on measures to hence they were paying disproportionately improve quality. Third, since much of the exter- more as a percentage of their total household nal financing is off-budget, the recurrent cost expenditures. Equally striking is the rapid rise implications of these projects are not explicitly in annual expenses within each standard: rela- taken into account during the design phase and tive to Standard 1, students in Standard 3 incur this, in turn, affects the sustainability of project 50 percent more expenses, students in Stan- interventions if, and when, external assistance dard 4 spend 2.4 times more and Standard 6 ceases. Fourth, although donor coordination is pupils spend 4.7 times more than Standard 1 amicable it remains minimal and there is a need pupils.16 for greater coordination and coherence to This high level of household expenditure for ensure that overall sub-sectoral goals will be primary school is a possible explanatory factor achieved in a cost-effective manner. Two exam- of student dropout in primary. A comparison ples of the problems arising from extra-budget- of household and public expenditures in Stan- ary funding and uncoordinated donor financing dard 1 is particularly instructive. The extraor- are highlighted in Annex 3.7. dinarily high PTR of about 100 to 1 effective- ly means that per pupil public expenditure is about 40­50 percent below the average for the Household financing primary cycle. At this grade level, household expenditures actually exceed public expendi- Despite low household incomes, Malawian tures. This finding is inconsistent with the prin- households continue to bear a considerable ciple of pro-poor spending which ideally share of both the recurrent and capital costs of should ensure that the largest subsidies benefit primary education. At the higher levels of edu- the poorest groups. cation, household contributions are likely to be A comparison of household monetary con- greater in absolute terms but lower in relative tributions for primary schools shows that both 52 Cost, Financing and School Effectiveness of Education in Malawi Table 3.7: Annual household expenditures on primary education, 2001 Mean annual per pupil hh expenditure Kwacha US$ As % of per pupil public exp Public schools 761 10.1 80 Private schools 3600 48.0 Mean annual per pupil hh expenditure--public schools Lowest quintile 721 9.6 76 Second 589 7.9 62 Middle 445 5.9 47 Fourth 698 9.3 74 Highest 1259 133 Rural 648 68 Urban 1636 172 Mean annual per pupil expenditure--public plus private schools As% of std 1 exp. Standard 1 559 7.5 1 Standard 2 659 8.8 1.2 Standard 3 829 11.1 1.5 Standard 4 1359 18.1 2.4 Standard 5 1488 19.8 2.7 Standard 6 2652 35.4 4.7 Average for all grades 913 12.2 Source: Ed Data of 2002. Note: Exchange Rate 72.1 Kwacha to 1 USD. the poorer and the rural Malawian household The current high levels of poverty, the sus- contribute a larger share of their incomes than ceptibility of agricultural and rural incomes to in some of the neighboring countries (Annex climatic shocks and poor terms of trade, and 3.8). For example, in Malawi, the poorest the relatively high dependence on household quintile of households contributes 5­6 percent financing all result in greater instability. This of its income for education and the richest particularly impacts the financing of critical quintile contributes only 1 percent whereas in teaching and learning materials, school repairs countries like Uganda, Zambia and Kenya the and construction costs. Instability in the poorest only contribute half the above amount. financing of the education system impacts neg- These direct monetary contributions should atively on quality and on children's progress be added to the indirect, or non-monetary/in- through the system. It also compounds the kind contributions (such as materials and later problem of inefficiency in public resource labor) made by the poorer households--espe- use through high repetition and excessive cially in the rural areas. Within the last year, 72 dropout rates. percent of rural households, in comparison to 25 percent of urban households, contributed labor to primary schools. In addition, 20 per- Notes cent of rural households also contributed 1. A particular issue at present (2003) is the materials relative to only 6 percent in urban cash budget system, introduced in 1996 as a areas. Again, the poorest sector carried the short-term stabilization measure to restore fis- largest burden with 77 percent of the poorer cal balance. This has resulted in unplanned and households contributing labor versus 40 per- exogenous cuts in discretionary spending. cent for the highest quintile. Monthly expenditure allocations to line min- Education Expenditures and Financing 53 istries depend on actual cash revenues and for- ferent divisions (cost centers). The last clear eign loan inflows during the month--after tak- ORT breakdown in T&L and other categories ing into account non-discretionary expenditure therefore also dates from 1998/99. on interest payments. Because of delays in the 6. It is stated GOM policy to increase cost release of funds by external agencies (as have sharing with students on items such as board- been the case over the past years in the absence ing and feeding at secondary level and particu- of an IMF program) there are often cash man- larly at university level. In practice, no drastic agement problems and delays in releases to changes have occurred at the university level, implementing Ministries. mostly because it is the Education Act that 2. After about a two-year delay a new IMF foresees boarding provision at tertiary level, PRGF program was approved in October thereby contributing to the very high public 2003. unit cost at this level and at the same time pro- 3. The principal subvented organizations are hibiting real chances to significantly increase the tertiary institutions, notably University of enrollment. At secondary level, however, Malawi and Mzuzu University. Other subvent- boarding expenditure should have diminished ed organizations comprise the Malawi Nation- drastically in the last year or so, as one of the al Examination Board (MANEB), Malawi HIPC triggers is to shift the burden of board- Institute of Education (MIE), Malawi Nation- ing to the households and free up much need- al Library Services (MNLS) and Malawi ed resources, for the CDSSs, for example. The National Commission for UNESCO actual expenditure for the most recent year 4. The normal budgetary process is that the should therefore begin to reflect this change. Ministry of Education, Science and Technology 7. The question is whether such a high and educational subvented organizations pres- expenditure is justified. One possible explana- ent forward estimates to the Ministry of tion is that these expenditures are partly a Finance (MoF), which then approves the budg- means to supplement staff emoluments. How- et based on estimates of the total resource ever, without a more detailed breakdown of envelope. Revised estimates of expenditure are this category; one that would require piecing presented at the end of the fiscal year, but actu- together this type of expenditure from the bot- al expenditures are only available a couple of tom up (since it is not immediately available in years later, after it has been reviewed by the MOEST statistics, it is not possible to explain auditor general. Hence, data for the most the increases in expenditure. The current recent years are usually approved or revised accounting system should theoretically allow estimates, which may be above or below the detailed analysis, but the low level of consis- actual expenditures. While revised estimates tency in the use of the different categories by tend to be higher than the approved amounts, inadequately trained staff at the divisions and actual expenditures usually lie (but not always) headquarters, have rather achieved the oppo- in-between the forward estimate and the site in making even basic analysis more diffi- revised estimate. cult. Therefore this kind of exercise is also 5. It is important also to mention that after beyond the scope of the present study. 1998/99 with the new cost center based 8. While average PTRs were maintained, accounting structure it has become almost despite increased enrollment, this hides the fact impossible to calculate breakdowns to more that PTRs are very different between urban detail than simply P.E. versus ORT for all of and rural areas and between lower and higher the education levels. This is partially due to the grades. Also hidden is the fact that the rapid fact that the more elaborate budgeting tools expansion of the primary enrollment resulted are not always used consistently among the dif- in the employment of many unqualified teach- 54 Cost, Financing and School Effectiveness of Education in Malawi ers, so that the PqTR increased dramatically. 13. In analyzing the breakdown of unit cost These issues are addressed in more detail else- in general the following key components were where in the study. considered: (i) average teacher salary, (ii) 9. Although it is clear that the unit cost for pupil-teacher ratio, (iii) average other staff teacher training has been significantly reduced, salary, (iv) pupil-other staff ratio, (v) per the exact extent is difficult to establish because pupil spending on teaching-learning materi- in recent years the MIITEP program (with the- als, (vi) per pupil spending on other operating oretically 2 cohorts per year graduating where- expenses (including for example pedagogical as in the pre-service system half of one cohort support and travel); and (vii) per pupil spend- would graduate per year) has not been running ing on social welfare and boarding. In up to its maximum capacity. As such, there are Malawi, however, the data for the last two quite wide fluctuations in number of graduates components are not disaggregated and so the and in the ensuing unit cost by year. last two items covering travel, grants and 10. This disparity may be explained by sub- boarding will constitute "other recurrent stantial inefficiencies at the tertiary level. expenditures". 11. These ratios would need further adjust- 14. Bruns et all. 2002. ment because the expenditure figures exclude 15. Clearly, some expenditures in the administration at the division level. Adminis- development budget could be classified as tration disproportionately benefits the conven- recurrent expenditures, but they are included tional secondary schools. in the former because they are financed by 12. Bruns et al. 2002. This study finds that donors and form part of the public invest- the non-weighted average of non-teacher ment plan--these include project staff salary salary recurrent spending in 33 African coun- expenditures, textbooks and rehabilitation of tries is US$9 and represents 24.1% of the total schools. recurrent spending. Thus the average total 16. However, since these data include some recurrent spending would be 9*100/24.1, or private school children, they should be inter- US$37.4 preted with caution. CHAPTER 4 Internal Efficiency and Management Effectiveness in Resource Utilization T his chapter further explores efficien- Efficiency in student flow in primary cy and management issues in the and secondary education delivery of education. In a country like Malawi, where resources are Index of efficiency and schooling years scarce, it is a priority to ensure that available to produce a graduate resources are used optimally if the desired objectives are to be achieved. As such, the The sudden expansion in the system in 1994 chapter examines the relationship between allowed most children to gain access to educa- inputs into the education system and student tion. This expansion, however, was not outcomes as measured by the level of grade matched with the additional trained teachers attainment. It will also identify resource or the essential teaching and learning materials wastage resulting from the high repetition and needed to assure quality. As a result, the lower pupil dropout rates and inequities in the allo- standards are characterized by large class sizes cation of teachers. taught by mostly untrained teachers teaching The efficiency analysis will reflect different in bleak environments, often under a tree.1 perspectives and include analysis at the nation- This has contributed to high repetition and al, division, district and school levels. The dis- dropout rates and has resulted in the wastage cussion will examine how effectively inputs are of a significant share of the resources allocated translated into pupils who complete the course to the education sector. cycle, or pass the final exam, and examine if The magnitude of the waste and inefficiency the same amount of inputs could produce addi- is reflected in Table 4.1. It shows that with the tional graduates. The issue of learning out- current level of repetition and dropout in the comes will be explored in Chapter 5. system, the government pays for 20 school 55 56 Cost, Financing and School Effectiveness of Education in Malawi years in order to get one child to graduate from student drops out, the index of efficiency primary instead of the normal 8 school years. would be as high as 81%; thus wastage less This is also reflected in the so-called `index of than 20%. In the same way, assuming no stu- efficiency', which for Malawi is 39.7 percent. dent needs to repeat and the current dropout This figure indicates that 60.3 percent of the rates are maintained, then half (50 percent) of public resources are used either on children the resources would be wasted. Clearly, high who drop out before finishing primary,2 or on dropout rates have the greatest adverse affect repetition years (both of those who eventually on the efficiency of the system and, as such, finish and those who do not3 and are also lost any planned policy interventions would main- to the system). In comparison, similar data for ly need to target a reduction in dropout rates. Mozambique4 indicate that 30 percent are However, since repetition and dropout are pos- used either on children who dropout or on rep- itively correlated, reductions in repetition will etition years, which is only half the level of have a positive effect on dropout rates (see Malawi. Paradoxically, this wastage may be Annex 4.1 for a more detailed discussion). due to insufficient spending on primary educa- Internal efficiency in secondary is signifi- tion (e.g. qualified teacher, appropriate PTR, cantly better than in primary but, nevertheless, sufficient T&L materials, adequate infrastruc- inefficiencies still exist. At the secondary level, ture, etc.), as reflected by the low unit cost for a little less than one-quarter of public resources primary school in Malawi. Put another way, are wasted due to repetition and dropout: it quality education probably requires a certain requires 5.2 school years instead of 4 to pro- unit cost per child per year. duce a graduate. Also, inefficiencies resulting The table also shows that if the current level from repetition alone are low, consuming of repetition is maintained, and assuming no about 5 percent of the public resources. Ineffi- ciencies are substantially higher as a result of Table 4.1: Index of efficiency in the different cycles of dropout only and consume close to one fifth of study, 1999/00 the available public resources.7 This positive Primary Secondary picture probably results from the rigorous selection process. However, as will be made Total Total clear, there are significant differences between Index of overall efficiency5 the various types of secondary schools. Percent 39.7 76.8 At first glance, repetition may appear to not Numbers of years necessary to have 1 graduate from primary 20.1 5.2 make a difference from government's perspec- Index of efficiency if the system tive since the cost of one teacher is the same, would not have dropouts, i.e. whether s(he) teaches ten or a hundred pupils. the only source of inefficiency However, if a genuine effort is made to deploy are repeaters teachers to roughly achieve the prescribed 60:1 Percent 80 95 PTR, then the total number of teachers required Numbers of years necessary to complete 10.0 4.2 (and thus their cost) is much higher if there are Index of efficiency without more repeaters. Alternatively, if teachers are not repeaters, i.e. the only source deployed based upon numbers of children in a of inefficiency are Dropouts class due to repetition, then repetition obvious- Percent 50 81 ly will have an impact on pedagogical practices Numbers of years necessary and student learning due to overcrowding. This to complete 16.06 4.9 creates a vicious circle: students repeat partially Source: School Survey Data in Basic Education Stats 1999 and 2000, as a result of very high PTRs, and the PTRs and repetition rates from DHS2000 final report for primary and from DHS2000 dataset with authors' calculations for secondary. remain high because of the repetition. To lower Internal Efficiency and Management Effectiveness in Resource Utilization 57 the PTRs to the recommended PIF levels, and effective for teaching and learning. A further without reducing the repetition rate, more important aspect is the qualifications and teachers would be required than without any, experience of teachers: unequal deployment, or or with limited, repetition. For instance, if it is the deployment of untrained and inexperienced assumed that in primary there is a total enroll- teachers, can lead to poor student outcomes, ment of 3 million pupils and a repetition rate of which in turn contribute to both higher 25 percent and MK 1000 (about US$10) is dropout and repetition rates and to lower sur- spent on a child annually then total spending of vival rates. MK750 million extra is required to achieve the An important aspect of teacher distribution same PTR as in a scenario without repetition. is associated with the formulation of clear cri- In other words, a reduction of one percentage teria for providing schools of similar enroll- point in the repetition rate will generate savings ment size with a comparable number of teach- equivalent MK30 million (about US$300,000) ers.8 The task, therefore, is to document to per year. what extent current teacher allocations in Malawi are consistent with this principle.9 As such, the following analysis examines the Teacher deployment extent that this principle holds true by formal- ly estimating the average relationship between The quality of an education system is generally the number of teachers and the number of associated with the supply of teachers in the pupils at the school level by using a regression schools, particularly qualified teachers. Since analysis. The result indicates the degree that teachers' salaries in primary school constitute schools deviate from the average relation- the lion's share of expenses--both as a share of ship.10 This technique is applied for the coun- unit cost and total public expenditures--teacher try as a whole, but separate data on the differ- deployment is closely associated with unit cost ent divisions/districts and urban/rural areas are per pupil or school, and therefore with the equi- provided. table distribution of funding. It is important, therefore, to explore how teachers are allocated Primary school level at the division, district and school level. The distribution of teachers is generally Figure 4.1 visualizes the regression results for based on the targeted PTR, one that is deemed all public primary schools in Malawi. It clear- Figure 4.1: Teachers allocation in publicly funded primary schools 50 40 30 teachers of 20 10 Number 0 0 500 1000 1500 2000 2500 Number of pupils Source: Authors' calculations, using EMIS 200011 58 Cost, Financing and School Effectiveness of Education in Malawi Figure 4.2: Comparison of degree of randomness in teachers allocation to individual primary schools for selected Sub-Saharan African countries 60 50 40 30 20 Randomness % 10 0 Tome Chad Faso Mali Guinea Niger Benin Togo Namibia Senegal ZambiaRwanda Gabon d'IvoireMalawiUganda Ethiopia Sao Mauritania Cameroon Mozambique Madagascar Côte Guinea-Bissau Burkina Countries Source: Mingat, 2003. ly indicates that not all schools are on the result of variance, either between the divisions straight line (each dot represents a school); (i.e. some divisions over endowed in compari- rather there is a significant spread that corre- son to others), or somewhere at a lower level in sponds to an R-square value of 66 percent, or the system. For this, a set of dummy variables to 34 percent `randomness'--this means that for the different divisions in the regression this part of the variation in teacher deployment model is used. By doing this, differences is not explained simply by difference in the between the new `randomness in teacher number of pupils in the schools but by other deployment' (33.6 percent=100­66.4) and the factors.12 For instance, schools with 1000 one noted earlier (34.1 percent), provides an pupils can have anywhere between 5 and 40 indication of the variability that is explained teachers and, similarly, schools with 40 teach- by differences in allocations to different divi- ers have enrollments varying between 500 and sions (see Table 4.2). Since this difference is 3800. There are a number of schools that are quite small, it is possible to state that teacher extremely far removed from the norm, where deployment between divisions is equitable immediate action is most urgent.13 since none appears strongly favored, or disfa- Figure 4.2 puts this 34 percent `randomness' vored. This suggests that allocation of teachers in perspective by comparing Malawi to a num- by the Ministry Headquarters to divisions is ber of African countries based on this indica- fairly equitable, but that allocations within tor. The average for the sample of countries is division and districts may not be. Table 4.2 25 percent; as such, Malawi is clearly among also shows the average school size and the the countries with a more arbitrary teacher average number of teachers in each division. deployment system. This indicator also indi- The national average school size is 671 pupils, cates that there are significant opportunities with slightly less than 11 teachers.14 for immediate improvements. To further analyze the degree of inconsisten- To further identify where management prob- cy in allocating teachers within the divisions, a lems might exist, it is possible to assess the regression was run for various subsets of extent to which the degree of randomness is a schools across each of the six divisions. In Internal Efficiency and Management Effectiveness in Resource Utilization 59 Table 4.2: The relationship between the number of teachers and pupils in public or subvented primary schools, 1999/00 Pupils Teachers Divisions Total / Avg. per school Total / Avg. per school Coefficient Number of pupils ­ ­ 0.016 Central West 760,015 / 761 12134 / 12 Reference Central East 509,080 / 594 7958 / 9.5 (­0.2) Northern 469,864 / 458 8146 / 8 0.6 South Eastern 389,243 / 777 5435 / 11 ­1.5 Shire Highlands 418,741 / 882 5960 / 12.8 ­1.6 South Western 388,745 / 739 6476 / 12.3 (0.3) Intercept ­ ­ 0.2 R2 (%) ­ ­ 66.4 Source: Authors' calculations, using EMIS 2000.15 Regression values between brackets are not significant at the 5% level. Table 4.3, below, the column `Degree of ran- respective districts. In those divisions that lack domness' indicates the variability established a main urban area (Central East and Shire by the regression for each of the division sub- Highlands) it is clear that no district appears to sets. The degree of inconsistency differs sub- be strongly favored, or disadvantaged. Howev- stantially from one division to another. er, divisions with a clear urban center present a It is also possible to simulate what would be radically different picture. The current teacher the average number of teachers in schools of a human resource management system signifi- given size in each of the divisions. In Table 4.3, cantly favors urban areas in districts where the column one provides the results of these simu- division has a major urban center. lations for schools of 671 pupils, which is the Further analysis for the subset of schools in average size of the primary schools in Malawi. each district provides additional information The average number of teachers in an average on teacher deployment to individual schools size school ranges from 9.4 teachers (South within individual districts. The results of this Eastern division) to 11.9 teachers (Northern analysis (see Annex 4.2) point to a wide varia- division). tion in teacher deployment among the districts, The difference between both sets of ran- and indicate a need to evaluate teacher deploy- domness values in Table 4.3, above, provides ment at the district level. an indication of the consistency of teacher Large differences in PTRs exist between deployment in each division across their urban and rural districts (see Annex 4.2). The Table 4.3: Regression results with and without using district dummy variables within each of the division subsets Number of teachers Degree of randomness Percent difference in average school Degree of with dummy variables between degree of Division (671 pupils) randomness for districts randomness Central East 10.6 49.4 49.2 0.4% Central West (with Lilongwe) 10.9 39.5 29.7 24.8% Northern (with Mzuzu) 11.9 30.7 24.8 19.2% South East (with Zomba) 9.4 36.1 30.7 15.0% South West (with Blantyre) 11.1 17.3 13.8 20.2% Shire Highlands 9.8 41.6 40.2 3.4% Source: Authors' calculations using EMIS 2000 data. 60 Cost, Financing and School Effectiveness of Education in Malawi Table 4.4: Efficiency ratios by rural and urban districts, varies in the rural districts, ranging from 39 1999 percent to 53 percent. In addition, the propor- Efficiency Ratio Urban Rural Total tion of female teachers in urban centers (73­87 Pupils per teacher 45 66 63 percent) is much higher than in rural areas Pupils per qualified teacher 66 129 118 (19­48 percent). Pupils per stream 86 70 72 Table 4.4 illustrates that schools in urban dis- Pupils per stream (std 1­2) 100 93 n.a. tricts are appreciably larger (about 3 times) than Pupils per stream (std 3­7) 82 60 n.a. Pupils per stream (std 8) 74 44 n.a. in rural districts. The urban institutions also Pupils per school 1,865 606 646 have larger class size than their rural counter- Pupils per classroom 144 92 96 parts, even though they have much lower PTRs. This results from the dubious practice of urban Teachers per stream 1.94 1.06 1.13 schools employing almost two teachers for each Teachers per school 42 9 10 stream of pupils--this practice is inappropriate- Teachers per classroom 3.24 1.40 1.51 ly referred to as `team teaching'. In effect, it Source: PER (MOFEP 2001) invariably results in one teacher being present with all the entire group of children while average PTR in 2000 was 1:63, a figure close his/her colleague is involved in some other activ- to the PIF objective of 1:60. However, in all ity, or is absent altogether.16 So although urban urban centers, the PTR falls below this target. areas have a liberal supply of teachers and are On the other hand, the ratios in rural districts, well below the target PTR, these deployment such as Mangochi (82:1), Phalombe, Chik- practices result in classes being even larger, on wawa, Machinga (75:1), Lilongwe Rural-East, average, than in rural areas. Dedza (72:1) and Ntcheu, Thyolo, Mulanje There are also notable differences in the (71:1) are far removed from the policy target. PTRs by grade,17 ranging from extremely high A review of the deployment of qualified levels in the lowest grades to quite low levels in teachers clarifies the situation even more. The the highest grades (Table 4.5). One possible average pupil-to-qualified teacher ratio (PqTR) explanation for this may be that following the is enormously high at 123:1. The geographic introduction of FPE, the redeployment of discrepancies are clear: urban areas maintain a teachers from the higher standards to the lower 60:1 ratio of qualified teachers, but rural areas standards has been limited, even though it is in are severely deprived of sufficient qualified these lower grades that the new influx of chil- teachers (PqTR is close to 200:1 for Lilongwe dren has occurred. The higher grades have Rural East, Mangochi, and Phalombe). If lower PTRs because a high percentage of the expressed as a percentage of qualified teachers children drop out before reaching the higher (of the total teaching staff) urban areas have a grades. Another contributing factor is the con- relatively high proportion of qualified teachers: tinuing use of `specialized' teachers for the it is 70 percent in Mzuzu; 87 percent in higher standards; they frequently teach only Zomba; 70 percent in Blantyre, and 58 percent certain subjects and consequently these indi- in Lilongwe. The percent of qualified teachers viduals having abundant `free time'. Table 4.5: Pupil teacher ratios, standard 1 -- 8, 2000 Std 1 Std 2 Std 3 Std 4 Std 5 Std 6 Std 7 Std 8 100 76 64 52 45 39 33 28 Source: EMIS 2000 Internal Efficiency and Management Effectiveness in Resource Utilization 61 Deployment process in primary remediate the present anomalies. However, the recently established decentralization policy The current deployment process (discussed in provides a great opportunity to improve the depth in Annex 4.3) is clearly flawed. A more situation, by giving districts greater freedom to equitable distribution of teachers requires hire and pay teachers, and by supplying them immediate consideration but this will necessar- with financial resources based upon a neutral ily take time because of the difficulties associ- formula using the number of pupils with some ated with redeploying teachers who are in -- adjustment criteria for poverty level or popula- for them -- comfortable positions, but which tion density. As a result, rural districts should are not equitable nor efficient for the education be able to advertise open posts as soon as a system as a whole. However, the large numbers teacher leaves. of teachers that need to be recruited into the system over the next few years provides a win- Secondary schools dow of opportunity to bring about some of the required changes. Figure 4.3 shows the distribution of teachers in A more effective budget mechanism can public secondary schools18. In these schools, deter unplanned teacher movement. However, average size is 233 pupils while the average to encourage teachers to move to areas where number of teachers is 8.9. The teacher deploy- shortages are endemic, such a budget system ment is even more unequal than in primary, as needs to be complemented by incentives, and the R-square in the regression analysis explain- by insurances for teachers to not be `stuck' ing the number of teachers in a school by the their whole careers in undesirable locations. number of students is only 41%, leaving as Possible incentives might include a combina- much as 59% of the variation unexplained. tion of the following: more speedy promotion There are also some secondary schools with opportunities for teachers working in less extreme differences to the norm, which may desired areas; higher rates of leave accumula- need immediate action to adjust the number of tion; larger and more regular annual incre- teachers. ments; hardship allowances that reflect the dif- When examining the PTRs and PqTRs at the ficulties of the place of employment; free secondary level, it is clear there are strong dif- housing and utilities; higher transport ferences between rural and urban areas. For allowances to facilitate more frequent travel to example, in 199919 the PTR and PqTR in centers; etc. At the same time extreme caution urban areas were respectively 28:1 and 59:1 is required to establish mechanisms whereby whereas these indicators for rural areas were the additional benefits are only given to the 36:1 and 131:1. teacher as long as (s)he serves in a hardship In the publicly funded secondary school cat- area. In other words, a hardship bonus should egory, three main types of schools are found. not follow the teacher once (s)he relocates to a The CDSSs are acknowledged to be perform- non-hardship posting. Another option would ing under difficult circumstances, relative to be to require new graduates to serve a number CSSs and grant aided schools. For an average of years in a rural area, as some sort of reim- size secondary school (approximately 233), a bursement for the training received. CDSS has roughly 4 teachers less, while grant The myriad of inconsistencies in teacher aided schools have about 1.8 teachers more deployment practices in many Malawian than government day and boarding schools.21 schools make it clear that the establishment of Teachers' qualifications are equally impor- an equitable policy on teacher utilization will tant in determining pupil performance. The not, in itself, provide a magical panacea to vast majority of CDSSs are staffed with trained 62 Cost, Financing and School Effectiveness of Education in Malawi Figure 4.3: Teacher allocation in all CSS, CDSS, and grant-aided secondary schools 30 25 20 teachers 15 of 10 5 Number 0 0 100 200 300 400 500 600 700 800 Number of pupils Source: Authors' calculations using EMIS 2000.20 primary teachers who are generally lacking in assumption here is that the goal is to produce knowledge of subject content, but they are also children with at least primary education com- often uncomfortable teaching at secondary pletion. Another more valid argument might level. Figure 4.4 clearly illustrates the radical be that completing 8 grades is more than is difference between the various types of second- required to achieve full literacy--and it can be ary schools. seen a self-standing achievement of the educa- tion system. However, typically 5­6 years of quality education are perceived necessary to Notes reach this level and more than half of the 1. In general, the higher standards in dropouts do not achieve either. Malawi are allocated classrooms when 3. Note: because repetition rates are used in demand does not match supply. standard 8, the effect of having to pass the pri- 2. Pupils may of course have received some mary leaving exam to graduate from primary is benefit from their time at school. However, the taken into consideration. Figure 4.4: Percentage of qualified teachers in different types of secondary schools 100% 80% 60% 40% 20% 0% 1999 2000 2001 CSS (including grant aided) %unqual Priv %unqual CDSS %unqual night %unqua Source: Basic Education Statistics, 1999-2001 Internal Efficiency and Management Effectiveness in Resource Utilization 63 4. Mozambique: Cost and Financing of Edu- tion). If instead, R square is 100%, it indicates cation, World Bank, July 2003. that there is a perfect consistency and that the 5. Calculated with the Quasi-time-series number of teachers in any school is exactly estimate for the survival rate, using years what would be anticipated on the basis of the 1998/99 and 1999/00. number of students enrolled. This can be pre- 6. This means that even if there was no rep- sented by a straight line graph, as shown in fig- etition, on average 16 years of schooling are ure 4.1. In the case of primary schools in required to produce 1 primary school graduate Malawi, a school of 500 pupils would always because for each child that completes primary, have 8 teachers and a school with 1000 pupils there are 2 who completed 4 grades only and would always have 16 teachers. then dropped out. 13. Some examples of these are: (i) 20 7. This includes those pupils who sit schools have a PTR of less than 40, with some through the 4 forms, but who never pass the less than 10, while on the other end of the MSCE exam. Even though they finished the spectrum, there are 10 schools with a PTR of same number of grades as MSCE passers, their more than 80--it actually goes as high as a employment opportunities and especially PTR of 200; (ii) 15 very large schools having expected salary levels are substantially lower more than 4,000 pupils, up to a maximum of (Kadzamira E.C. 2003). 8,600 pupils; (iii) schools with more than 100 8. An exception to this rule might be in the teachers, combined with a low PTR of 25 or case of very small schools in remote areas 37; etc. However, these schools do not repre- where the student catchment area does not sent the average teacher deployment situation have enough children to achieve the PTR poli- in the country and so, to clarify the situation, cy aim--even when incorporating multigrade schools with more than 60 teachers and/or teaching. more than 2,500 pupils are not included in Fig- 9. For example, an average PTR of 60:1--as ure 4.1. prescribed in PIF for primary schools--can be 14. The fact that the Northern region has on achieved in an equitable manner and with the average fewer teachers per school than Shire same PTR nearly everywhere throughout the Highlands, does not automatically mean they system. However, it may also result from very are under endowed with teachers since they different PTRs, with some far above and oth- also have the smallest average schools. These ers far below the average. average school sizes and average numbers of 10. Schools for which the number of teach- teachers in each school, of themselves, do not ers exceed the expected value based on the size give an indication on the equity of teacher of their enrollment are over-endowed, and deployment! vice-versa 15. Based on 4383 out of a total of 4647 11. Subset of 4384 schools with valid com- listed schools with valid data. binations of teachers and pupils, and not being 16. The practice is also frequently accompa- private schools, are used out of a total of 4647 nied by subject specialization. schools in EMIS 2000. 17. If teachers are registered as teaching in 12. It is appropriate to use the [100%--R more than 1 grade, then they are added to the square], to obtain an aggregate quantitative group of teachers for each category. indicator of the degree of inconsistency or ran- 18. It is important to note that in secondary domness in teacher deployment. If the value of schools a significant number of students attend R square is 0, it implies the absence of any sta- private schools--official statistics indicate tistical relationship between the two variables approximately 10%. However, there are also a (i.e. complete inconsistency in teachers' alloca- number of non-registered private schools for 64 Cost, Financing and School Effectiveness of Education in Malawi which no statistics are available so, here, the 21. When running a regression that percentage might be higher. However, since the includes dummy variables for the type of sec- government has no direct influence on teacher ondary school, the degree of randomness deployment practices, these are not included in decreases from 59 percent to 52 percent this analysis. The grant-aided schools, howev- (100­48.2)--or by about 12.2 percent. This er, are included because their teachers are fully indicates that a part of the variability is funded by government. explained by the over-endowment of certain 19. MOFEP, 2001. types of schools versus others (see Table 4.7). 20. 593 schools with valid data were used The coefficients show that in comparison to and this represents 138152 students. However, the reference government day and boarding in the Basic Educ. Stats the total ESTIMATED school, the grant aided schools fair favorably, enrollment is 274949, i.e. including those whereas the CDSSs have fewer teachers for a schools that did not provide data and the pri- given number of students. vate schools that are not included here. The 22. For 2001, the Draft Basic Education Sta- sample does include 31 schools that do not tistics is used. provide data on student numbers but have between 1 and 20 teachers. CHAPTER 5 School Effectiveness, Malnutrition and Learning Outcomes E arlier chapters examined aspects of the its pupils. For example, SACMEQ1 examined Malawian education sub-sectors and levels of English literacy achievement in Stan- critiqued the degrees of efficiency with dard 6 (or its equivalent grades) in several which they succeed in progressing stu- countries in the region.2 In Malawi, less than dents through the system. However, it is also one percent of all measured pupils reached the necessary to review how effectively students `desirable' mastery level in literacy, while 21.6 learn. This section provides this information percent reached the `minimum' mastery level. and is based, mainly, on EMIS data, national The results show further that the most sig- examination data and internationally compa- nificant differences are between children in rable standardized surveys of levels of achieve- large towns and the rest of the country, and ment in basic skills in primary standards across socio-economic levels (see Annex 5.1). (Monitoring Learning Achievement (MLA) Some small differences also exist between (i) and Southern African Consortium for Moni- the wealthiest children versus children of all toring Educational Quality (SACMEQ). The other socio-economic categories and (ii) chil- section uses a cross-country perspective to dren from rural areas/small towns versus those detail student achievement levels and discusses from urban areas.3 There is also a small gender the correlation between schooling characteris- gap--the girls performed less well than the tics and learning outcomes. The chapter boys. Poor rural girls not only constitute the includes a discussion of malnutrition, which group most at risk of dropping out before has been consistently documented as a factor achieving primary completion (see equity chap- affecting children's attendance and ability to ter), but they are also least likely to achieve sat- learn effectively. isfactory learning outcomes if they do manage to stay in school. Compared with the SACMEQ test scores of Level of learning six other countries from Eastern and Southern Africa, Malawi had the lowest percentage of The many challenges facing the primary educa- pupils meeting the minimum and desirable tion system in Malawi combine to produce a mastery levels of reading literacy (see Figure less than conducive learning environment for 5.1). 65 66 Cost, Financing and School Effectiveness of Education in Malawi Figure 5.1: SACMEQ test score, Southern and Eastern Africa 70 64.8 60 56.4 52.8 50 46.1 % 40 37 Minimum 30 26.7 Desirable 25.9 25.8 Scores 23.4 21.6 20 10 7.6 5.2 2.3 0.6 0 Kenya Zambia Malawi Zimbabwe Mauritius Zanzibar Namibia Source: SACMEQ. A separate UNESCO study, Monitoring improve health and reduce the prevalence of Learning Achievement (MLA, 1999), meas- HIV/AIDS. ured fourth-graders' scores in life skills, litera- The level of English used in the literacy and cy and numeracy and compared a number of numeracy tests may have contributed to countries (See Table 5.1). Malawi's low scores. In contrast, the Life Skills Table 5.1 reveals that Malawi's overall per- test uses more simple English statements and formance is roughly comparable to the other also includes drawings. Another contributing countries surveyed, although it has the lowest factor could be that the Life Skills test assesses literacy rate. In numeracy, Malawi also ranked more general knowledge that children acquire at the lower end. Interestingly, Malawi's per- outside school, while the numeracy and litera- formance in Life Skills is superior to all the cy tests measure skills specifically achievable countries in the study and this may prove through effective teaching in schools. Finally, important in trying to implement strategies to the above results add fuel to the current debate Table 5.1: Fourth-graders' scores on Monitoring Learning Achievement (MLA), 1999 (percentage correct) Country Life skills Literacy Numeracy Average score Botswana 56.0 48 0 51.0 51.7 Madagascar 72.1 54.7 43.7 56.8 Malawi 77.0 35.0 43.0 51.7 Mali 56.9 51.8 43.6 50.8 Mauritius 58.0 61.0 58.5 59.2 Morocco 62.3 67.6 56.4 62.1 Niger 47.7 41.1 37.3 42.0 Senegal 46.7 48.9 39.7 45.1 Tunisia 74.7 77.9 60.4 71.0 Uganda 66.8 58.0 49.3 58.0 Zambia 51.0 43.0 36.0 43.3 Average 60.8 53.4 47.2 53.8 Source: UNESCO (1999b). School Effectiveness, Malnutrition and Learning Outcomes 67 on the important role of local language in ini- the expected age range would be 11 to 12 tial learning achievement and suggests that full years; there were also wide age ranges which implementation of the existing national lan- would make effective instruction difficult. guage policy4 is likely to contribute to The Standard 6 classroom environments improved learning outcomes. were generally exceedingly bleak. The most The MLA study also provides disaggregated basic piece of equipment, such as a usable data by gender, urban/rural, and type of school chalkboard, was missing in 14 percent of the (public/private). The average test score for all classrooms. Only 35 percent of pupils had a areas (literacy, numeracy, Life Skills) shows a desk to sit at and a mere 29 percent of pupils less than significant 3.3 percentage point dif- had writing places; less than half of the class- ference in favor of boys. Differences between rooms had a teacher's chair and table; only half urban and rural students are also modest; there of the classes had an English dictionary; but is a 2.2 percentage point difference in favor of other essential learning materials, such as urban schools and this is one of the lowest for maps, were even more infrequent. In addition, the countries sampled. However, the differ- a mere 24 percent of Standard 6 pupils were in ences between private schools and public classrooms that contained library books while schools are more pronounced with 6.2 per- only 26 percent of schools had a school library. centage points in favor of private schools (8 Textbooks were also in short supply. Over percentage points for literacy alone).5 This dif- one-third of Standard 6 pupils did not have an ference is approximately in the mid-range, rel- English textbook. Overall, EMIS 2001 finds ative to other countries. the following pupil to textbook ratio for three major subjects (Table 5.2). The situation should have changed drastical- The learning environment in primary ly since then, as CIDA provided a total of 11 schools million textbooks distributed to all primary schools between the end of 2001 and June In addition to looking at test scores, it is also 2002. Unfortunately, no EMIS statistics after necessary to understand characteristics of the 2001 were available at the time of this report. learning environment. This section highlights a Going forward, the government will need to number of issues that are considered germane ensure regular textbook provision as part of to the learning process.6 the primary education sector budget, or through long-term donor commitment. School and home environment About one-quarter of teachers reported never having met the parents/guardians of a Children in Standard 6 were greatly over-aged, pupil during an academic year, varying some- a finding that confirms issues raised in the pre- what across divisions. Sixty-three percent of vious analysis on primary enrollment. The parents/guardians noted the existence of a PTA average age was slightly over 15 years whereas at the school their children were attending, Table 5.2: Students per textbook in primary, by standard Subject Std. 1 Std. 2 Std. 3 Std. 4 Std. 5 Std. 6 Std. 7 Std. 8 English 2 2 3 3 2 2 1 1 Math 2 2 3 3 2 2 1 1 Chichewa 2 3 3 3 2 2 2 1 Source: Draft Basic Education Statistics 2001. 68 Cost, Financing and School Effectiveness of Education in Malawi Table 5.3: Time spent on school and homework in primary by grade Std. 1 Std. 2 Std. 3 Std. 4 Std. 5 Std. 6 Std. 7 Std. 8 Percentage Max. 5 hours on school-related activities 73.5 65.5 28 26.5 7 3.5 1 (2) Average number of hours on school-related activities 4.8 5.0 5.8 5.9 6.7 6.9 7.1 (7.1) Percentage who do not have homework during the week 92 84 66 52 33 28 15 (5) Average hours spent on homework per week outside school 1.7 1.9 1.6 2.2 2.4 2.8 3.0 (3.2) Source: Ed Data 2002. Note: For Std. 8 these results are based on a sample of less than 50 pupils and, as such, are less reliable. while 96 percent of respondents said that there after-class study sessions; and time spent on was a school committee.7 Eighty percent of extracurricular activities. parents who stated that there was a school It is apparent that the time spent on school committee indicated that it was doing a good activities increases through the standards. job; 11 percent said it was not; and 8 percent Children in Standards 1 and 2 spend 5 hours either had no opinion or did not answer the per day on school activities; those in standards question. Ninety-five percent of respondents 3 and 4 spend 6 hours; and those in Standards indicated their belief that schools should teach 5 through 7 spend 7 hours on school-related more practical skills than done at present. activities. Only 2 percent of primary school The Ed Data survey revealed that only pupils spend more than 8 hours on school- slightly more than half of the children sur- related activities. There is virtually no differ- veyed, aged 6­14 years, live with both of their ence in the time spent on schooling by sex, by biological parents. Twenty-two percent lived urban-rural residence, by region, or relative to with their mother (but not with their father), 2 wealth. percent lived with their father (but not with While the percentage of pupils doing home- their mother), and 22 percent lived with nei- work outside of school increases through the ther of their natural parents, although not all standards, the amount of time spent per week of these were orphans. In brief, around one in on homework outside school varies little; it five children have lost one or both parents. ranges from approximately 2 hours in the lower standards to 3 hours in the higher stan- Other factors influencing school dards. Most striking are the differences by effectiveness wealth, with 38 percent of pupils in the highest quintile and 24 percent of pupils in the lowest Table 5.3 summarizes time children spend on quintile doing homework outside of school school-related activities in an average school hours. day.8 This time includes time spent traveling to On average, pupils who were absent from and from school; time spent in classes and school missed a total of 17 days during the Table 5.4: Number of school days missed by primary school children during the 2001 school year Lowest Highest Urban Rural Public Private quintile quintile Total Days absent 12 17 17 12 19 12 16.7 Source: Ed Data 2002. Note: only asked to pupils who were absent at least one day during the year. However these represent 97 percent of total. School Effectiveness, Malnutrition and Learning Outcomes 69 year. Pupils in rural areas were slightly more year. This disturbing trend is at variance with likely than those in urban areas to have missed the continuity children need if they are to school at least once, and they missed more acquire basic literacy and numeracy skills. days of school (17 rural versus 12 urban days missed). Children from households in the high- Curriculum est quintile missed fewer days than those from households in the lowest quintile (12 days ver- The nature, scope and relevance of the school sus 19 days). The SACMEQ study shows that curriculum also significantly impacts school one quarter of the Standard 6 pupils were effectiveness. In the following section, the pri- absent on the day of testing. mary and secondary curricula are briefly In the Ed Data study, the reasons given by reviewed. parents for the absenteeism of their children Primary Curriculum. Recently, a Primary during the previous school year were: illness-- Curriculum and Assessment Reform (PCAR) 86 percent, funeral--61 percent,9 did not want document has been drafted and is expected to to go--39 percent, school clothes dirty--34 be presented to Cabinet for formal approval percent, too hungry--28 percent (29 percent shortly. This revised curriculum will: (i) rural, 18 percent urban), any type of work-- address the `crowded' subject range; (ii) high- 12.5 percent, no money for fees--10 percent. light child-centered, more effective teaching Teacher absenteeism in Malawi is reported strategies; (iii) increase awareness of as chronic. The main reasons suggested for this HIV/AIDS; (iv) promote heightened gender high level of teacher absenteeism was atten- consciousness; and (v) provide an increased, dance at the funerals of colleagues or being and necessary, emphasis on `mother' tongue busy on errands, such as collecting salaries or instruction.11 The draft curriculum advocates examination correction payments. In some three groupings for the entire primary stan- schools, teacher time-on-task was as low as dards: Infant, Middle Primary and Upper Pri- 1­2 hours each day. Anecdotal evidence sug- mary, and suggests organizing core subjects gests that absenteeism is high also among dis- around these groupings. The curriculum trict, division and Ministry staff. In summary, reform also intends to develop resource mate- it is apparent that there is a breakdown in insti- rials to support teaching within each group. tutional accountability and that this is linked Finally, an Assessment and Training Module, to the failure to provide appropriate incentives already available, will assist teachers to imple- to perform assigned duties. ment the recommended continuous assessment Teacher Mobility. While teacher absenteeism practices. contributes significantly to interruptions in the A detailed plan for the introduction and teaching and learning processes, a related diffi- evaluation of the new primary curriculum is culty is causing additional disruptions, namely, now required. This plan should: (i) indicate the the degree of teacher mobility. For example, a likely training required for existing TTC tutors study in 2000 by the Improving Educational and for practicing teachers; (ii) outline the Quality (IEQ) Project (in partnership with the manner in which newly qualified teachers will Malawi Institute of Education)10 indicated that be aligned with the new curriculum; (iii) review many teachers do not remain with the same deployment practices to ensure they are consis- standard for a reasonable length of time. For tent with local language instructional require- example, by October of the school year ments; and (iv) agree community and teacher approximately 50 percent of the 188 teachers sensitization programs to clarify the impor- surveyed were no longer teaching the class they tance of local language to achieve positive had commenced with in February of the same learning outcomes. 70 Cost, Financing and School Effectiveness of Education in Malawi Secondary Curriculum. A revised secondary 50% more teachers every year than in a situa- curriculum was introduced to Forms 1 and 2 in tion without death from AIDS. Assuming that 2000, and to Forms 3 and 4 in 2002. This revi- this evidence is applicable to Malawi, this has sion reduced the excessive number of subjects important implications for the development of to seven core subjects and approximately 14 the pre- and in-service teacher training sys- possible elective subjects. This `new' curricu- tems, and for the costs of teacher training. In lum, however, has been challenged, principally addition, keeping the increasing number of by the universities. It is now necessary for children who are orphaned in school may MOEST to: (i) facilitate a review of the exist- require additional investment and support ing document; (ii) decide on subsequent structure. Furthermore, the higher level of teacher in-service efforts; and (iii) to indicate teacher and student absenteeism because of ill- plans for revising teacher training in the con- nesses and increasing opportunity costs will text of previous, and intended, changes to the undoubtedly adversely affect teaching and document. learning conditions for all students, and there Teacher education programs have not been is evidence showing that absenteeism from fully responsive to changes in secondary school HIV/AIDS is a greater problem to the educa- curricula. Therefore, the National Strategy for tion system than the cost of additional teacher Teacher Education (2002) advocates the need training. to review both curricula concurrently. Prevalence and consequences of Impact on pupils and teachers of malnutrition affecting education HIV/AIDS attendance and outcomes The HIV prevalence rate is estimated to be Malawi has one of the highest levels of stunt- around 20 percent in Malawi with severe con- ing in children for all countries in eastern and sequences for the education system. Numerous southern SSA. The problem is associated with studies on the impact of HIV/AIDS have been periodic food shortages combined with poor prepared and one study "The Impact of child caring practices and has led to almost HIV/AIDS on Primary and Secondary School- half of the children under 3 years of age being ing in Malawi: Developing a Comprehensive stunted (see Table 5.5). These high rates have Strategic Response"12 as described in Annex persisted in the last decade--even during times 5.2, summarizes the results and impact on when food supply has been adequate. Stunting teachers over a 10-year period; both the direct and micronutrient malnutrition are associated effect from HIV/AIDS illness and loss of life, with delayed school enrollment; decreased and its indirect negative impact on repetition school attendance,13 especially during the and dropout rates. Evidence from the hunger period from November to April; and HIV/AIDS pandemic in Uganda and Zambia reduced capacity for adults to work and suggests that the population growth will school-age children to learn satisfactorily. The diminish due to many adults dying at an early recent food crisis has been extreme, and one age, but that the growth of school age popula- that affected most of the population. It is evi- tion will decline only slightly. In those coun- dent that Malawi remains chronically food tries, it is anticipated that the attrition rate due insecure because of food shortages in some to death from AIDS may increase as much as 3 areas and seasonal food shortages in most of percentage points to around 8 percent in total. the country. In this situation, young children In order to make up for this higher attrition, are particularly at risk because they have the government will need to train approximately highest nutritional requirements of any age School Effectiveness, Malnutrition and Learning Outcomes 71 Table 5.5: Percent of children 0­3 years of age with stunting in the Eastern and Southern Africa region Box 5.1 The Impact of Stunting and Country % Stunted Year of Study Undernutrition on Education and Devel- Malawi 47.5 2000 opment: Angola 47.5 2001 Lesotho 44.6 2000 In young children, malnutrition is associ- Tanzania 41.3 1999 ated with half of all child deaths (Pelletier, Zambia 44.0 2001/2 Uganda 35.5 2000/1 1994), i.e. half of children who currently Mozambique 35.9 1997 die from a childhood illness would not if Kenya 35.3 1998 they were well nourished. Stunted children Zimbabwe 26.5 1999 are less likely to enroll in school at the Botswana 25.3 2000 appropriate age than children without Source: National data from countries with available recent stunting and have lower test scores in information for this age group. school, particularly in mathematics Note: Stunting means <­2 S.D. below the NCHS standard for height for age. (Drake, et. al., 2002). Physiologically, stunted individuals have less capacity to group and are dependent on others for their work than taller individuals. As a result of food and health care. their reduced ability to work and impaired While rates of malnutrition in young chil- cognitive function, short individuals have dren have been documented, less is known lower earning potential than their taller about the national prevalence of malnutrition counterparts. in school-age children. From the Demographic and Health Survey (DHS) 2000, it is apparent that many children are entering Standard 1 enrollment, school attendance and dropout either short or underweight for their age: 57.4 rates. Families keep children out of school dur- percent of children 4­5 years of age are stunt- ing this period so children can earn money to ed and an alarming 28 percent are severely buy food or look for food. In addition, hunger stunted. These high levels of malnutrition and fatigue are so severe that many children do reflect cumulative nutritional insult and a dire not have enough energy to walk to school. nutritional situation, most of which occurs Anemia, caused by iron deficiency, is preva- before 3 years of age.14 lent throughout the country, as is Vitamin A National figures on malnutrition in adoles- deficiency (VAD). AIDS is also a cause of mal- cence (usually defined as ages 10­19 years) are nutrition, anemia and other micronutrient defi- limited. In one study,15 prevalence of malnutri- ciencies. Because diet is poor (both in quantity tion was high in 10­14 year-olds. The 2000 and quality), it is likely that school children are household survey shows that 17 percent of affected by these deficiencies, and also defi- women 15­19 years of age have low body cient in a number of other micronutrients that mass index (BMI<18.5); this is 2­3 times high- affect health and capacity to learn. er than any other age group of reproductive Malnutrition and increasing education age women (20­49 years of age). However, on opportunity costs. From the above, it is clear the positive side, there is some possibility of that the overall nutritional status of children in catch-up growth in children who are stunted Malawi has an adverse effect on learning and during early childhood. raises opportunity costs during crisis periods. In Malawi, the beginning of the school year The nutritional status of school age children is (January) occurs during the hunger season, a combination of the quality and quantity of which has a profoundly negative effect on their current food intake and also of their 72 Cost, Financing and School Effectiveness of Education in Malawi Box 5.2: The Impact of Anemia, of Iodine Deficiency, and of Selected Micronutrient Deficiencies on Education and Development: Iron deficiency anemia (IDA) in children (particularly <2 years) has irreversible negative con- sequences on learning in older children, even when IDA is corrected later in life (Booth and Aukett, 1997). Lower test scores are associated with anemia (Drake, et al., 2002). A recent study found that U.S. children with IDA had poorer math scores than children without IDA (Halterman, 2001). Giving iron supplements to young children increases growth (Drake, et. al., 2002) and may decrease levels of stunting which also affects school enrollment and other indicators for coverage and quality of education. In Malawi, Shrestha (1994 as reported by Latham, et al., 1995) found that by giving iron supplements to children 6­8 years of age in Ntcheu District, eye-hand coordination improved and IQ increased by 7 points. Vitamin A deficiency increases mortality in young children, is a leading cause of blindness in young chil- dren worldwide and affects iron metabolism, increasing iron deficiency anemia. Zinc is anoth- er micronutrient, which is important for the growth of children under 5 years of age and pre- pubertal children (Brown, et al., 2002). Zinc deficient children who receive zinc supplementation have decreased severity of diarrheal disease and pneumonia (Bhutta, et al., 1999). Children living in iodine-deficient areas have lower IQs and poorer cognitive and motor function than children living in non-iodine deficient areas. Most of the damage to IQ occurs before birth but several studies show that improvement can occur in abstract reasoning and verbal fluency when urinary iodine status improves (Drake, et al., 2002). The Shretha study (1994 as reported by Latham, et al., 1995) found that children 6­8 years of age performed bet- ter on tests for stamina and certain physiological tests when they received iodine supplements. Improvement in mental development translated to a 21-point increase in IQ. When iron was also given, IQ increased by 26 points. intake before they enter school, particularly recurrent problem in school children because before the age of two years. Lack of food, children are normally not fed breakfast and either seasonally or year round, and poor child often walk long distances to school and this feeding practices are the major causes of mal- results in their being unable to benefit from a nutrition in the country. As such, ensuring day of learning. A 1995 report stated that good nutrition and health in very young chil- "throughout Malawi's three regions, teachers dren is an investment in school children. Fur- reported that between 50­90 percent of chil- thermore, programs that promote good health dren come to school hungry" (Latham, et al., and nutrition in school children are also need- 1995). ed and essential to maximize children's ability During periods of food shortage, attendance to learn while they are in school. at primary school dropped dramatically (to Lack of food has devastating effects on the zero in some schools) because children were energy intakes of school children. According to kept home to look for food or they were too one study, in non-crisis years energy intakes in weak to attend. A report for UNICEF (Gal- school children were only 74 percent of the lagher, et al., 2002) indicates that attendance recommended amounts (from Latham, et al., rates dropped off seriously between November 1995), and presumably in crisis years this and April -- the months prior to harvest, often would decrease further. Morning hunger is a referred to as the hunger period. In Malawi, School Effectiveness, Malnutrition and Learning Outcomes 73 the academic year begins in January, which has totality of student learning. Using the EMIS a negative impact on student enrollment and 2000 dataset, a regression analysis is per- dropout. While this drop in attendance is a formed to examine the correlation between the chronic annual problem, in 1999 the problem mentioned outcome changes and changes in persisted during most of the school year.16 schooling input characteristics--for example During the most recent food crisis, school the percentage of qualified teachers, unit costs, meals were the only food some children were repetition rate, PTR, proportion of female receiving and these helped stave off starvation teachers, etc. in families that were without grain reserves or Table 5.6, below, summarizes the results of money to purchase food. There is also some four regression models based on retention and anecdotal information to suggest that during pass rates of individual schools. The first three this period, younger children (<6 years of age) models use retention throughout primary as a were enrolling in school to obtain at least one measure of student achievement; the last one meal per day. uses the pass rate in the PSLCE.17 The first two However, access to education is different models measure the variability in retention from a demand for education. As such, food rates from Standard 1 to 818 based upon vari- can prove to be either a barrier or an incentive ability in the input variables. The models differ to attend school. Meanwhile, Malawian par- in that they use slightly different input vari- ents are faced with the stark reality of increas- ables. Retention is calculated with a simplified ing opportunity costs at a time when the per- formula: in each school the non-repeating stu- ceived benefits from education are, at best, dents in Standard 8 are divided by the non- minimal. repeating students in Standard 1. This process provides an average retention of 31 percent. The third model measures the variability in Correlation between schooling retention rates from Standards 1 to 4,19 and characteristics and learning outcomes the average retention measured in this manner is 49 percent. To appreciate further those characteristics that It is important to note that the EMIS school influence learning, this section investigates the census only reports school-related variables. elasticity between education outcomes and The most important finding is that these select- input variables--in other words, the extent ed variables only clarify a very minor part of the that learning outcomes improve with changes differences in retention and pass rates among in inputs. The strength of this relationship will schools. In other words, there are factors which also provide insight into the overall manage- are not included in the regression model that are ment of the system; the expectation is that much more powerful contributors to the differ- changes in the typical policy-related inputs ences in the selected outcome variables. For the should have a direct impact on learning out- regressions on retention in Standards 1 to 8, comes. This analysis is especially important only about 10 percent of the differences can be because of an emerging paradigm shift from explained by the differences among schools in inputs to outcomes. respect to the selected input variables The review measures changes in outcomes (urban/rural, proprietor, repetition rate, double by examining changes in retention throughout shifting, proportion female teachers, text books primary and by investigating changes in the per student, unit cost). In the same way the pass rates in the national PSLCE. Such an regression for retention in Standard 1 to 4 can analysis only provides a partial picture of the explain only 5 percent of the differences in the education system, and it does not capture the retention among different schools. In the case of 74 Cost, Financing and School Effectiveness of Education in Malawi Table 5.6: Correlates of school variables with retention and exam pass rates (1) Retention (2) Retention 1 to 8 Model 1 to 8 Model (3) Retention Variable with unit cost without unit cost 1 to 4 (4) Pass rate Urban/rural20 (1.5) (2.3) 9.4 (­1.1) Central Gov21 (­2.0) (­1.1) ­4.2 (2.3) Religious22 ­2.4 ­2.4 (1.2) (­.33) Private23 (­3.7) (­.5) ­1.4 (3.6) Pupil:teach ratio ­.14 0.06 Repetition rate (0.05) (.011) (­.19) ­.2024 Double shift or overlap25 3.4 3.2 (5.6) ­4.3 Proportion female teachers 0.0526 (0.02) (0.08) 0.06 Percentage unqual. Teachers ­.17 (0.04) Total books per student (.17) (.23) (­.21) (­0.08) Unit cost 7.9 (3.1) Intercept 21.3 49.4 46.1 73.0 R-square .090 .099 0.047 .020 No. schools 2585 2574 3722 2503 Source: Authors' calculations using EMIS 2000. Note: Values in brackets are not statistically significant at the 5% level. the regression using pass rates, it can only analyses have demonstrated--and confirmed explain about 2 percent of the differences in the through similar analyses in other countries--is PSLCE pass rates among schools. that absolute resources (unit costs), the PTR, In the regressions on retention in Standard 1 and the qualification of teachers are predictors to 8, the significant variables27 are the PTR, for improvement in learning outcomes. Obvi- the percentage of unqualified teachers and the ously the weak correlation between the vari- unit cost. In regression (2) the interpretation ables means that there are important non- for the PTR variable is that if a school with a explained (by this model) factors to the national average of 70 pupils per teacher could learning process in schools, like for instance reduce this number to 60, retention would the limited time-on-task or the high teacher improve by nearly 1.5 percentage points. Simi- and student absenteeism. In summary, this larly, for unqualified teachers, if a school with review suggests that more efficient manage- the national average of 52 percent unqualified ment of the system, greater accountability and teachers could lower this by 10 percentage improvements in the existing incentives are points, then retention would improve by essential precursors for a successful interven- approximately 1.7 percentage points. tion in Malawi's education system. Additional- In regression (1) the interpretation for the ly, factors external to the system, like for unit cost is that if a school has a unit cost of instance malnutrition, can also have an impor- about 100 MK more (as a result of a lower tant effect, and addressing such should be part PTR and/or more qualified, thus better paid of an overall strategy to improve children's teachers) than the national average of about learning opportunities. 1,115 MK, then retention is likely to be higher by about 0.8 percentage points.28 In a well-managed education system, a sig- Notes nificant relation is expected between inputs 1. In Malawi the SACMEQ study only and outcomes. In Malawi, however, there is included the English literacy test, not the very little relation shown when using these tra- mathematics test, as it did in many of the other ditional school-based variables. What these participating countries. School Effectiveness, Malnutrition and Learning Outcomes 75 2. In the case of Malawi, approximately one 9. Attending funerals or assisting family third of the pupil data were missing--there members with funeral-related responsibilities. were only 1983 valid responses out of planned 10. See "American Institutes for Research. sample of 3100. Several reasons were suggest- 2000." ed for this; the major ones were a lack of trans- 11. The reconfirmation of local language port for data collectors and pupil absenteeism instruction for Standards 1 to 3 during a recent on the day of testing. In most other countries language symposium was a positive develop- the response rate was 90 percent or more. This ment. means that the Malawi results have to be treat- 12. Esme Chipo Kadzamira, Dixie Maluwa- ed with caution but the gaps are unlikely to Banda, Augustine Kamlongera, Nicola Swain- invalidate the general finding of the study. son, September 2001, Center for Educational 3. A degree of caution is required when Research and Training (CERT), Malawi. interpreting these results because many of the 13. DHS--Ed Data 2002 found that 29% of pupils who live in towns and from the higher rural children and 18% of urban children were socio-economic levels are also the ones who absent some time during the previous school may regularly come in contact with English year because of being too hungry. outside school, which may have an effect on 14. Studies in smaller areas of the country their performance on English reading tests. confirm high rates of malnutrition in all 4. The draft Malawian language policy was school-age children. Surveys conducted by reaffirmed in 1996 and stated that in the lower Save the Children/U.S. (SCF/US) in selected standards, local languages be used as the medi- schools of Mangochi district show high preva- um of instruction. But for a number of reasons, lence of stunting (39 percent), underweight (38 the policy is not fully implemented--these percent) and wasting (5 percent) in all school include: ambivalent attitudes about the value children (ages 6 years to over 14 years of age) of this policy, lack of the required teaching and (Save the Children. 2002). Stunting rates were learning materials, language mismatching highest in the 10­14 year age group (42 per- between various region and teacher posting cent). Higher rates of stunting in older age decisions, time tabling difficulties, confusions groups have been corroborated by another about the distinction between teaching a lan- group, which found that 43 percent of 11-year- guage and developing literacy, etc. As such, old school children were stunted and by a 5- many pupils frequently fail to acquire basic lit- country study (Partnership for Child Develop- eracy skills because they are forced to learn in ment, 1998). This 5-country study found that a language other than their L1 (MLA, 1999). stunting was higher in boys than girls--a find- 5. The number of private schools sampled is ing also of the SCF/US study. too small to allow for reliable generalizations. 15. Save the Children. 2002. 6. The main sources are the 2002 household 16. In order to increase enrollment and survey and the 1998 SACMEQ study. attendance at schools, WFP, in partnership 7. PTAs are voluntary organizations run by with MOEST and UNICEF, is implementing a parents and teachers and focus on the welfare pilot school-feeding program in selected of students and teachers and the overall devel- schools in the country. In 1999 Dedza East dis- opment of the school. The school committees trict was selected as the pilot area for the are the statutory representatives of the MOEST school feeding program with the target of at the primary school level and are charged 23,500 primary school students attending 24 with overall development of the school. schools. The objectives of the program were to: 8. The data is confined to children attending improve child enrollment and attendance in all day schools; boarding pupils are excluded. standards; improve pupils' capacity to concen- 76 Cost, Financing and School Effectiveness of Education in Malawi trate and assimilate information by reducing is not effected. Only the absolute level of reten- short-term hunger; increase attendance and tion might be underestimated. enrollment and reduce dropout rates of girls in 19. The large influx as a result of the Free Standards 5­8. The program in turn increased Primary Education policy in 1994 may distort enrollment by 20 percent, stabilized class the results somewhat. However, by the year attendance patterns, reduced dropout rate of 2000 (for which data exists), a major part of girls and boys in all standards, improved edu- those entering primary in 1994 have either cational results for both girls and boys (accord- dropped out or are presently in standard 4-- ing to interviews with parents, teachers and this means that the standard 4 enrollment has students) and attained a more equitable ratio already incorporated the effect of FPE. of boys and girls in Standard 5­8. 20. Dummy variable for urban versus rural It should be noted that school feeding pro- schools. A positive value (if significant) means grams elsewhere do not translate into better that urban schools perform better, given that nutritional status of students because parents, all other variables in the model are the same. A knowing that their children receive a meal at potential reason why this variable is not signif- school, reduce the food they provide to chil- icant in 3 models, may be because the main dif- dren at home. However, committees were ferences between rural and urban areas may be encouraged to provide food more strategically captured elsewhere in the model under the in the day, when it would have the greatest pupil: teacher ratio and pupil: qualified teacher impact on attentiveness and learning and ratio--which are much higher in rural areas-- would maximize its effect on educational out- or under the unit cost which is much lower in comes. Students in Standards 1 and 2 attend the rural areas. school from 7 to 11 am and receive Likuni 21. This is a dummy variable for central phala at about 9 am, which helps reduce government schools. It reflects the difference in hunger students experience at school and, in outcomes between a school where the central theory (there has been no evaluation on the government is the proprietor versus the refer- program's impact on learning) increases atten- ence type of school which is one where LEA is tiveness, attention span and ability to learn the proprietor, considering all else the same during the subsequent school hours. Because of about the schools. the logistics of feeding all children at the same 22. This is a dummy variable for central time, older students, who have a longer school government schools. It reflects the difference in day, usually receive their meal later in the day. outcomes between a school where a religious 17. It is important to remember that it is the entity is the proprietor versus the reference variability in the pass rates of schools that is type of school which is one where LEA is pro- analyzed. Children reaching the level sitting the prietor, considering all else the same about the PSLCE are already a minority group within the schools. school system, irrespective of how the condi- 23. Same as previous, but between private tions of their school have been throughout the school and LEA school. eight grades and irrespective of any other non- 24. Interpretation is as follows: a 1 percent- school related conditions experienced. age point decrease in repetition in the whole of 18. The Free Primary Education policy may primary (below its average of about 16% as distort the results somewhat. However, for the measured by EMIS) corresponds to a 0.2 per- most part the extra enrollment as a result of centage point increase in pass rate of the grade FPE will be relatively proportional over the dif- 8 students who take the exam. ferent schools, so that the variance in retention 25. Dummy variable for schools with double among schools (which is being examined here) shift system or schools with an overlapping School Effectiveness, Malnutrition and Learning Outcomes 77 system of classes with a single shift and classes percentage point increase in the retention in with a double shift. The reference here are sin- primary. gle shift schools. A positive value therefore 27. Those that had a very high T-statistic as means the result is better in schools with dou- a measure of significance. ble shifting or overlapping. 28. The unit cost is calculated using the 26. Interpretation is as follows: a 10-per- number of teachers for a given number of stu- centage point increase in the proportion of dents and their assumed income given their women in the whole of primary (above its grade levels. average of about 24%) corresponds to a 0.5 CHAPTER 6 Equity T he present chapter discusses educa- ly, this section examines who benefits from tion equity and focuses on differ- public spending on education and other equi- ences across gender, geographical ty issues. location/region, household income The government has implemented a number groups and consumption quintiles. The under- of initiatives, explicitly aimed at ensuring equi- lying premise is that an equitable distribution table access to education. Here, FPE, in partic- of education resources plays an important role ular, has improved the access of the poorer seg- in reducing poverty and in improving the dis- ments of society to primary education. At tribution of opportunities, of income and secondary and tertiary levels, also, a number of wealth and of participation in society. In addi- initiatives have been initiated. For example, the tion, improving educational access by the government has created extra secondary school poor, by the marginalized and by the most places by changing Distance Education Centers vulnerable groups leads, eventually, to fairer into Community Day Secondary Schools; it has access to a number of public services. Further- provided scholarships to girls and orphans; more, equitable access to education con- and has established education institutions in tributes to the re-distribution of income and disadvantaged areas. At tertiary level, a bur- this, in turn, benefits the poorest groups in sary scheme to support female students study- society. This is particularly true in the educa- ing non-traditional courses, such as engineer- tion sector. It constitutes the largest govern- ing, is in place; there is also a loan scheme to ment expenditure item and it has the widest assist needy students `whose parents or outreach because almost all families in guardians fail to pay the obligatory fees'.1 Malawi have children, or are guardians for Although these policy interventions are target- children and, as such, have the potential to ed, in part, to improve inequities in the distri- benefit from public subsidies through educa- bution of education opportunities, there is a tion. This means the education sector is an need to ascertain the extent of current effective fiscal instrument for alleviating inequities and tighten policy interventions as poverty and redistributing income. According- required. 78 Equity 79 Figure 6.1: Enrollment by gender and level of education, 2000 100.00% 80.00% 60.00% 40.00% 20.00% 0.00% Higher Prim. Secondary Standards Standard Primary educ. Teach. 5-8 1-4 Training Boys Girls Source: Basic Education Statistics 2000. For secondary the estimated numbers are used. Enrollment patterns by gender and should be coupled with implementing policies geographic location to deploy more female teachers to rural areas. If the perspective over time is considered, it An overall perspective is clear that improvements have occurred in secondary; girls' participation has grown from Figure 6.1, shows that overall enrollment in 35 percent of total enrollment in 1992/93 to 40 the Malawi education system, by gender, percent in 2000, and, in addition, female favored boys in 1999/2000. The percentage of access to university level also increased slightly girls enrolled was 48.5 percent for primary from 23 percent in 1993/94 to 26 percent. On education, 40 percent for secondary, 35 per- the other hand, in primary teacher education, cent for primary teacher training, and 26 per- there has been a negative trend with a decline cent for higher education. However, in primary from 41 percent in 1993/94 to 35 percent in education there are as many girls as boys in 2000. first four grades while the share of girls drops Analyzing for regional differences, no signif- to 45% in Standards 5 through 8.. In public icant difference was found in gender specific formal Technical and Vocational Training, enrollment between the six divisions.2 Annex between 2000­2003, girls represented 34 per- 6.1 presents the Relative Gender Gap (RGG) cent of the graduates. by district.3 To close the gap in gender specific Since primary Teacher Training Colleges enrollment, efforts need to be considered, such (TTC) traditionally enroll a significantly high- as providing targeted feeding for girls in the er number of male than female students, the two districts with the most significant gender share of women in the stock of primary teach- gaps (Nsanje and Chikwawa). ers is much less that of men. One crucial Table 6.1 indicates that there are significant caveat, here, is that trying to balance the num- differences in primary enrollment between ber of male and female teachers should take rural and urban areas, both for NER and GER. into account that most female teachers are cur- Furthermore, the NER and GER are quite dif- rently deployed to urban areas--in rural areas ferent by region, the Northern region having only 22 percent of teachers are women, where- the highest enrollment and the South having as in urban areas the figure is 75 percent. the lowest. Differences are just as substantial Accordingly, recruiting more female teachers between the richest and the poorest house- 80 Cost, Financing and School Effectiveness of Education in Malawi Table 6.1: Primary school enrollment ratios according to background characteristics, 2002 Residence Region Wealth index quintile4 Urban Rural North Central South Lowest 2nd 3rd 4th Highest Total NER 90 80 93 84 76 73 74 81 85 91 81 GER 127 113 129 119 108 106 106 116 118.5 126 114.5 Source: Ed Data 2002. Results are for the de jure household population age 5­24. holds: for children aged 6­13 years in the high- obtained some primary education is 87 per- est income quintile, 91 percent attend primary cent, or 16 percentage points higher. Among school compared to 73 percent in the lowest children whose mothers attended at least some quintile. The GER follows a similar pattern; it secondary school, NER is 97 percent, or 26 is 126 in the highest quintile and 106 in the percentage points higher than for children lowest quintile. whose mothers never went to school. There is no difference in the NER by gender Another predictor of educational attainment (81 percent). However, the male GER exceeds is household income level. As Table 6.2 shows, the female GER in both the North and the the GER for all income quintiles in Standards South, but there is no difference in the GER in 1­4 is above 100 percent and this reflects the the Central region.5 Thus, the full difference in large number of overage students, and possibly enrollment in primary by gender is a result of reflects some underage students. The poorest the higher number of over-age boys in the sys- income quintiles have a higher GER than the tem. This is an indication that a reduction in richest for Standard 1 to 4 because children of repetition will reduce the imbalance in the poor parents repeat disproportionately more boys/girls enrollment. In general, the older girls times in the lower grades. In upper primary, are more prone to drop out and more likely to however, the picture is radically different: the repeat. As such, if repetition is not lowered sub- GER for the richest quintile is about 100 per- stantially in the lower standards, it will be nec- cent whereas it is only 60 percent for the poor- essary to target older girls to improve retention. est quintile. In other words, even though pri- Inequities in the NER also arise as a result of mary education is free, certain factors impede differences in parents' education; studies sug- the poorest children from progressing from gest that the education of parents is a strong Standards 5 to 8. The GER for boys versus predictor of whether a child is likely to enroll girls only shows a minor difference of 6 per- in school. This is especially true for the educa- centage-points for lower primary (156 percent tion of girls and their mothers. The NER for boys, 150 percent for girls), whereas in among children aged 6­13 years whose moth- upper primary the GER for girls is much lower ers never attended school is 71 percent, where- than for boys (86 percent for boys, 70 percent as the NER among children whose mothers for girls). Table 6.2: Gross enrollment rates in std. 1­4 versus std. 5­8 Poorest 20% 2nd 3rd 4th Richest 20% Standard 1­4 156 159 160 150 138 Standard 5­8 61 75 78 88 97 Source: Integrated Household Survey 1997/98. Notes: The GER for Std 1­4 is total enrollment in these grades divided by the Std 1­4 school age population (6­9) and the GER for Std 5­8 is total enrollment divided by the population aged 10­13. Quintiles are constructed here by ranking households by their consumption per adult equivalent. This is different from the quintile calculation in Ed Data 2002. However, it is unlikely this will affect the trends discerned. Equity 81 Table 6.3: Secondary school enrollment ratios, by gender, according to background characteristics, 2002 Residence Region Wealth index quintile6 Urban Rural North Central South Lowest 2nd 3rd 4th Highest Total NER Total 21 7 16 10 6 2.5 6 3.5 7 23 9 Male 18 8 12 13 6 3 8 3 6.5 26 9.5 Female 23 6 25.5 7.5 6 2 3.5 4 8 20 8 GER Total 57.5 22 50 27.5 21.5 11 18 14 23 62 26.5 Male 57 29 53 34 28 16 22.5 18 32 72 33 Female 58 14 47 20.5 15 5.5 13 9.5 14 52 20 Source: Ed Data 2002. Results are for the de jure household population age 5­24. Table 6.3 shows that urban youth aged percent of the youth in the fourth quintile, and 14­17 years are three times more likely than 3 percent of those in the lowest quintile, attend their peers in rural areas to attend secondary secondary school.7 This pattern suggests that school (NER 21 percent versus 7 percent), the monetary costs of schooling at the second- while the GER differences among them are ary level may present a substantial barrier both only slightly smaller. Regional differences in to entry and to persistence in school beyond the secondary NER are considerable and fol- the primary level. It is important to note that low a similar pattern to the primary cycle: 6 public subsidies to secondary education tend to percent of youths aged 14­17 years attend sec- benefit the wealthiest income groups. There- ondary school in the Southern region, com- fore, public funding to secondary should not pared to 10 percent in the Central region and be at the expense of primary. 16 percent in the Northern region. There are thus significant regional inequities in the access Gender and geographical disparities in to secondary education. schooling profiles While the gender gap in the NER for sec- ondary is minimal (8 percent of female and 10 The disparities in access to Standard 1--or the percent of male youths aged 14­17 years `gross admission rate' or `cohort entry rate' by attend secondary school), there is a consider- division (using the school census information able gender gap in the GER for secondary and UN-EFA population)--show that all divi- school. Male youths up to age 24 are far more sions have rates of well over 1 (or over 100 likely (33%) than female youths (20%) to percent, meaning more new entrants in Stan- attend secondary school. This difference is a dard 1 than the population in the correspon- result of a significant gender gap in access to ding age groups). The figure ranges from as secondary school in rural areas, where male high as 2.55 for Central West to 1.41 in South youths, often over-age, are twice as likely as East and it indicates a clear relative difference.8 female youths to attend secondary school. One There are no divisions that show a significant conclusion, here, is that rural girls have difference between boys and girls. Using the extremely limited access to secondary school. repetition rates in the 2000 household survey, At the secondary level, differences in the the access rates of primary for boys and girls NER as a factor of income level are far more are quite similar with values higher than 1 striking than those found at the primary level: (1.32 for boys, 1.27 for girls). only youth in the highest income quintile Figure 6.2, above, shows that survival appear to have access to secondary schooling. through primary is about 20 to 30 percentage While 23 percent of the wealthiest youth aged points higher in urban areas than in rural, due 14­17 years attend secondary school, only 7 to higher dropout rates in rural areas in the 82 Cost, Financing and School Effectiveness of Education in Malawi Figure 6.2: Survival in primary, urban9 vs. rural (cross-sectional method) 100 90 80 70 60 50 40 30 20 10 0 Grade 2 3 4 5 6 7 8 1 Urban Rural Source: EMIS for enrollment and repetition,10 UN-EFA population (distributed between urban/rural according to proportion found in NSO population estimates for 2000). first few grades. This indicates a need to focus vival rate is slightly higher but it is still much on survival in early primary in rural areas; it more modest than the urban-rural differ- also suggests that if rural children make it ences, even in the primary sub-sector. The halfway through primary, they are more or as difference in the survival of boys versus girls likely to complete their education as their is also quite similar in all divisions. Signifi- urban counterparts. In Chapter 4, it was cant differences in survival exist between apparent that urban areas benefit proportion- divisions themselves, as Table 6.4 below indi- ally more from teacher deployment arrange- cates.11 Here, the Northern Division clearly ments, especially for the qualified teachers. has a significantly better retention than all Concurrently, the regressions on retention in other divisions, whereas Central West has the primary show that the proportion of qualified most worrying situation. Although it may be teachers and the PTR impact retention levels. argued that the Northern Division is privi- Therefore, the difference in survival rates leged, compared to the other divisions, the between urban and rural areas are affected by drop from 100 percent to below 50 percent the difference in PTR and by the percentage of by Standard 8 reflects inadequacies inherent qualified teachers. in the entire education system. However, the Retention is slightly better for boys than magnitude of the intervention required for a for girls in upper primary (see Annex 6.2). In positive change to occur would differ by divi- secondary education the difference in the sur- sion. Table 6.4: Retention standards 1­8 and 1­4 by division and by gender, 2000 Division Retention St.1­St.8 (%) Retention St.1­St.4 (%) PTR Pupil­qualified teacher ratio Northern 42.2 65.4 57 95 Central East 24.0 53.0 63 132 Central West 14.8 35.0 62 131 South East 21.2 49.5 70 146 South West 27.6 58 60 109 Shire highlands 20.3 48.1 70 136 Source: EMIS 2000 enrollment and repetition12 and 2000 UN-EFA population (distributed over the divisions according to proportion found in NSO population estimates for 2000) Equity 83 Figure 6.3: Proxy std. 4 and std. 8 completion rates by gender Standard 4 120.0% 100.0% 80.0% 60.0% Boys 40.0% Girls Percent 20.0% 0.0% 1992 1993 1994 1995 1996 1997 1998 1999 2000 Y ear Standard 8 60.0% 40.0% Boys 20.0% Percent Girls 0.0% 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year Source: Basic Education Statistics for enrollment and repeaters, and NSO population. `Proxy' completion is calculated as the number of non-repeating students in Std. 4 and Std. 8 divided by the corresponding age cohort. Note: UN-EFA population is not available for the beginning of the nineties, nor are household surveys for repetition rates. However, since the goal is to look at relative differences in completion by gender, the potential shortcomings in population estimates and repetition rates are less relevant. One possible reason for the better perform- comparison to Standard 4 completion (see Fig- ance in the North is the longer tradition and the ure 6.3 and Annex 6.3). Despite this discrep- higher esteem in which schooling is held--it was ancy, the trend over time is positive and shows the region where many missionary schools were a narrowing of the gap. founded and where education has, in general, Table 6.5, below, shows that the disparity in been of a better quality. Another reason for the pass rates between girls and boys differ signif- disparity may be the lower PTR and the lower icantly by type of school, especially for stu- PqTR vis-à-vis the other divisions. It is interest- dents enrolled in a CDSS.13 The differences ing to see that South West is the second best per- between boys and girls may have to do with forming division in retention, and also has a bet- the selection policy at the beginning of sec- ter PTR and PqTR. However, the difference in ondary, which allows girls to enter more easi- the retention rate between South West and ly than boys with similar levels of achievement Northern is larger than would be expected from in the PSLCE test. Since the number of girls the difference in the PTR and PqTR. entering secondary is still lower than the num- ber of boys, it is difficult to argue against this Completion rates in primary and policy if gender equity is considered para- secondary mount. It is, however, important to verify whether other factors are contributing to the The PCR by gender shows a sizable difference, weaker performance of girls on the MSCE with a bigger gap for Standard 8 completion in test. 84 Cost, Financing and School Effectiveness of Education in Malawi Table 6.5: MSCE and JCE results, 2002 MSCE MSCE JCE JCE Type of school % female pass % male pass % female pass % male pass Community Day Total 10.2% 21.0% 48.4% 63.4% Government Boarding Total 28.8% 54.7% 56.3% 75.8% Government Day Total 37.9% 54.0% 60.8% 76.9% Grant Aided Total 53.3% 65.0% 74.4% 86.9% Night School Total 26.0% 50.3% 30.1% 64.3% Private Total 28.7% 39.2% 55.6% 65.2% Grand Total 23.9% 34.4% 52.8% 66.6% Source: Authors' calculations, using data from MANEB database. From the results of Table 6.5, it could be tant to analyze the amount of public resources hypothesized that the grant-aided schools may that different income groups receive in each offer lessons relevant to schools trying to education sub-sector. For example, if public improve their MSCE results. For example, per- education spending is distributed equally haps the management styles of the grant-aided across the population, then each of the five institutions elicit better student performance in income groups will receive an equal share of examinations. However, it is also important 20 percent.14 Table 6.6, below, summarizes not to overlook the fact that households who how education spending is distributed by level send their children to grant-aided schools also of education according to family wealth, com- contribute significantly more funds, compared paring situations in 1997/1998 and to government day and boarding schools. In 1990/1991. addition, the selection process of potential can- It is apparent that the situation has didates may also be different. improved: in 1990/1991, the poorest quintile received only 10 percent of public resources but by 1997/98 this share had increased to 16 Public expenditure breakdown by percent and had been reduced for the richest quintiles and gender at different quintiles (e.g. from 38 percent to 26 percent). education levels Despite this progress, public spending remains regressive, benefiting the children of the richest To improve the equitable distribution of households most (26 percent versus 16 percent resources and pro-poor spending, it is impor- for the poorest quintile). Table 6.6: Incidence of public education expenditure by level and quintile Poorest 20% 2nd 3rd 4th Richest 20% 1997/98 All 16 19 19 20 26 Primary 22 22 21 19 16 Secondary 12 18 16 22 33 University 0 13 24 11 53 1990/91 All 10 13 16 24 38 Primary 15 18 20 23 24 Secondary 7 11 14 28 41 Sources: 1997/98 data from Integrated Household Survey and MoE (1998); 1990/91 data is taken from Castro-Leal (1996), table 14 pp24 and table A.8 pp42. Notes: Under `All' for 1997/98 only primary, secondary and university public education spending is included. `All' for 1990/01 includes the same spending types but, in addition, it also includes other tertiary education (primary teacher education and technical training). Equity 85 This is due to the fact that, in general, chil- families, more than half of the public resources dren from wealthier families tend to benefit were consumed by children from the richest 20 more from secondary and tertiary education-- percent of the population in 1997/98. Those in the because very few, if any, poorer children 2nd to 4th quintile, on average, benefit from progress to higher grade levels. However, since about 16 percent from government spending. The 1990/91 the situation has been improving as a subsidization of university is particularly con- result of FPE and the expansion of secondary tentious -- since full board is a major expenditure schools. In 1997/98, the poorer primary pupils item -- at a time when hunger and chronic mal- benefited more than the highest income groups: nutrition is endemic for millions of Malawi's the two poorest income groups each received poorest. As such, overall public funding at the uni- 22 percent of public resources for primary edu- versity level needs to be further investigated to cation versus only 16 percent for the richest determine possible funding and access increases quintile. On the other hand, in 1997/98, chil- while simultaneously increasing pro-poor spend- dren from the poorest quintile benefited almost ing and targeting the poorer qualified students at not at all from subsidies in tertiary education. both secondary and tertiary education. As indicated earlier, primary PTRs across Improvement in the distribution of resources standards show substantial differences requires the allocation of public subsidies to the between the lowest and highest standards. levels with the highest proportion of poor chil- Since the largest public expenditure item in dren. Prime targets include primary, generally, primary is teacher salaries, a very high PTR in but also upper primary and secondary CDSSs. lower grades means that spending per child in Continued efforts are required to ensure gener- lower primary is less than the spending per al cost recovery (or cost transparency) at sec- child in upper primary. Further, since fewer ondary, and especially at the university level, children from the poorest households progress combined with scholarships schemes for the beyond the lower grades, there is room for poorest. However, this is a complex and diffi- improvement in pro-poor spending in primary. cult area that requires further analysis. At a minimum, balancing the spending per Looking at differences for boys and girls by child in lower and upper primary (especially income group (see Annex 6.4), it is clear that by lowering the PTR in lower primary and the most significant differences are found for ensuring, at least, the same percentage of qual- the secondary and university sub-sectors: the ified teachers at this level as for upper pri- share of spending apportioned to girls of rich- mary15) could improve the situation further. er families relative to girls of poorer families is As discussed, the situation at the secondary more disproportionate than for boys. For level has improved since 1990/91; more chil- instance, at secondary level it is 36 percent for dren from poorer households are now access- girls for the richest quintile versus 9 percent for ing this sub-sector. For example, the share for the poorest quintile, while for boys it is 30 per- the poorest quintile increased from 7 percent cent versus 13 percent. At university level, it is to 12 percent in 1997/98 while the share for 61 percent versus 0 percent for girls and 49 the richest quintile declined from 41 percent to percent versus 0 percent for boys. This is not 33 percent. However, substantial inequities surprising since more boys progress to the remain, with the children of the poorest 40 higher levels even among the poorest income percent households receiving less than 30 per- groups. In primary, the situation is similar for cent of the public expenditure, while those of boys and girls and with both groups public the richest 20 percent receive one-third. spending is slightly pro-poor. At university level, where access to university is by and large only open to children of wealthier 86 Cost, Financing and School Effectiveness of Education in Malawi Notes 9. For urban areas, only Mzuzu City, Blan- 1. In the academic year 2001/02, however, tyre City, Zomba Urban, Lilongwe City are over 4000 students benefited from the loan included. scheme and this number is more than the total 10. Since the main aim, here, is to show rel- university enrollment, according to Basic Edu- ative differences, the perceived repetition rate cation Statistics. inconsistency in EMIS probably effects both 2. At secondary school level, all divisions urban and rural areas in a similar way. were close to the country average of 40 percent 11. Remark: Although an assessment of girls--the Central East division was the more inequity at secondary education level between equitable outlier with 43 percent girls and the districts, divisions and regions might reveal a Shire Highlands was the less equitable outlier number of disparities, such an exercise could with 36.5 percent girls. also be misleading because some districts host 3. The RGG is determined by dividing the dif- national schools whilst others do not. ference between the GER of boys and girls by Nonetheless, it would be plausible to isolate the GER of girls. The RGG, therefore, could be the national boarding schools from the "day" high as a result of large differences between the in order to determine the magnitude of differ- GER of boys and girls, or as a result of a small ence between districts. Furthermore, the dis- difference but combined with a small GER of trict analysis would offer an opportunity to girls. When the RGG is examined against the determine whether or not the inequity problem total GER of each district or region, there is an has roots in the primary school sub-sector. inverse relationship. Higher RGGs are typically 12. The repetition rates in the household found in regions with lower GER levels. The surveys are not broken down by division and implication, obviously, is that increasing GER cannot be used in this context. overall, would lower the gender gap, and, as 13. See Chapter 2, section `Survival / Com- such, a policy to increase retention and enroll- pletion in Secondary and Transition to Ter- ment addresses gender inequities. tiary'. 4. For an in-depth explanation of how the 14. The two caveats here, are, that the quin- wealth index is constructed see page 30 in tiles are based upon 5 groups of the same "Education Data for Decision-making. 2003" number of households. But since poorer house- by NSO and ORC Macro. holds on average have more children, equality 5. South: Boys 114% vs. girls 102%; North: of public expenditure at household level does boys 134% vs. girls 123.5%; Central boys not imply that each individual child achieves as 118% vs. girls 119%. much as subsequent children. Secondly, even 6. For an in-depth explanation of how the though in most circumstances achieving equal- wealth index is constructed here, see page 30 in ity is going to be an enormous challenge, it "Education Data for Decision-making. 2003" could be argued that for education to be an by NSO and ORC Macro. equalizing factor in society, public expenditure 7. A similar pattern exists among youth up on it should be pro-poor rather than merely to age 24, with a GER of 62 percent among the equal among all quintiles. wealthiest youth, 23 percent among youth in 15. There is a great deal of international evi- the fourth quintile, and 11 among the poorest dence to suggest that allocating a society's best youth. and most qualified primary teachers to the ini- 8. It might be that if the reality of Central tial grades is bneficial for the educational West is much closer to 1 than the results show, development of the population. the situation in South East is one of a cohort rate of entry of well below 1 CHAPTER 7 University Sector I n previous chapters a number of fea- the country. In 1998, the Malawi Institute for tures of the Malawian education sys- Management (MIM) carried out an analysis tem were explored, particularly the of the university sector and concluded that: primary and secondary education "Government has yet to develop a compre- sub-sectors and teacher training strategies. As hensive policy framework for the University's mentioned in Chapter 1, education consti- development in terms of defining/determining tutes a system and, thus, in the present chap- the demand for University's products and ter, the higher education sub-sector will be services, broad parameters e.g. sectors of pri- discussed; here the focus will be on access, ority focus, quantity and quality of the Uni- equity, financing, efficiency, quality and rele- versity's principal outputs (graduates and vance. research) in line with economic require- Malawi has one of the smallest higher edu- ments". cation systems in Sub-Saharan Africa. The While the MIM Report made a number of country has two public universities, the Uni- proposals and recommendations on the man- versity of Malawi (UNIMA) and Mzuzu Uni- agement of the university sector, to date, versity. UNIMA is by far the largest and the implementation of these recommendations oldest university, and it embraces a number of has been limited. Apart from the establish- relatively autonomous colleges: Bunda Col- ment in 1998 of Mzuzu University--in a for- lege of Agriculture, Chancellor College, the mer teacher training college--there has been College of Medicine, Polytechnic and little expansion of higher education in Kamuzu College of Nursing. However, the Malawi. As such, the colleges have been development of the higher education sub-sec- forced to function with the existing facilities, tor has been constrained by a number of fac- staffing and infrastructure arrangements. tors. In particular, investment has been limit- Lack of investment and maintenance has led ed because of the failure to develop a national to the deterioration of the infrastructure and strategic plan for the education sector as a has left some of the colleges in need of signif- whole, one that reflects the role of higher edu- icant rehabilitation. Paradoxically, in excess cation in relation to the education sector itself of 200 maintenance staff continue to be and the overall social and economic goals of employed at the universities. 87 88 Cost, Financing and School Effectiveness of Education in Malawi Figure 7.1: Enrollment in tertiary per 100,000 inhabitants, Sub-Saharan Africa 800 700 600 500 400 300 200 100 0 e Chad Faso Mali Togo Djibouti Malawi Eritrea Guinea Benin Uganda Namibia Tanzania Ethiopia Lesotho MauritaniaMauritiusBotswanaSwazilandZimbabw Mozambique Madagascar Burkina Source: USAID education statistics database. Results are generally for 1996 but it includes data from 1997 and 1995. Access and equity With a limited number of students attend- ing, it is interesting to disaggregate the areas of Malawi's comparatively small higher educa- enrollment. Figure 7.2, provides a breakdown tion system enrolls approximately 4,000 stu- of enrollment in UNIMA and Mzuzu Universi- dents, which constitutes roughly 0.3 percent ty across subject areas for 2001. Additionally, of students of eligible age. The last official Annex 7.1 shows a detailed breakdown of data from 2001 was 4,128. Growth in enroll- recent enrollment trends and graduate output ment has been very slow over time. In by college and by subject area. Here, it is 1987/88, 2,330 students attended, in 1990/91 apparent that there is some alignment in rela- it was 3,117 and in 1994/95 enrollment tion to areas of enrollment, the demands of the reached 3,600. The only private university, labor market and the structure of the economy: Livingstonia, has yet to open but the projected about 40 percent (or about 1,600 students) are initial intake will only be in the region of 56 enrolled in agriculture, education or engineer- students. ing programs. Figure 7.1, confirms that Malawi has one of Despite this relative alignment to the needs the lowest proportions of enrolled tertiary stu- of the labor market, the limited enrollment and dents per 100,000 inhabitants in the whole of skills mix means that the higher education sec- Sub-Saharan Africa. This data is from 1996 tor fails to provide a sufficient number of high but enrollment has remained more or less level human resources and professionals to unchanged since the mid-nineties. assist in the conversion and diversification of Figure 7.2: University enrollment by subject area, 2001 18% 16% 15% 16% 14% 12% 10% 10% 8% 7% 7% 8% 6% 5% 5% 6% 4% 3% 4% 2% 2% 0% Law Studies Media Education SciencesSciencesNursing Medicine Agriculture Commerce Engineering Humanities and Social Applied Education Source: UNIMA and Mzuzu University. University Sector 89 the economy. Several neighboring countries In brief, it is apparent that a clear majority have focused on expanding tertiary programs in of public resources for university is presently education, agricultural sciences, communica- benefiting what are most likely students of the tion, science and technology. These efforts have highest socio-economic strata. At the same generated increasing productivity and diversifi- time, women's access to higher education con- cation of production. In contrast, Malawi has tinues to lag significantly, only representing 30 not sufficiently focused on these subject areas. percent of total enrollment. While it is recog- The number of university graduates has nized that these issues are both long term and remained low and is inadequate to satisfactori- difficult to address, it is clear that the current ly contribute to the objectives of socio-econom- system and facilities are not sufficiently robust ic development and poverty alleviation. to cope with the inevitable future demand for A recent tracer study of university gradu- post-secondary education following the intro- ates1 examined the socio-economic back- duction of FPE and the concomitant increases ground of 380 graduates from four different in enrollment at primary and secondary levels. cohorts in time (1980, 1987, 1994, 1999) and noted that access is increasingly skewed in favor of students from higher socio-economic Financing backgrounds. For example, both parents of graduates had, on average, at least ten years of Chapter 3 indicated that the universities are formal schooling (mother 13.3 years for female autonomous and sub-vented. As Annex 7.3.a graduates, 12.5 for male graduates; father 14.5 illustrates, an approximate average of 50 per- for female graduates, 11.6 for male graduates). cent of university recurrent expenditure went It is also noticeable that the parents of female to `personal emoluments' and `benefits' graduates have enjoyed more schooling, which between 1997/98 and 2001/02. Spending on suggests that that socio-economic status is still `students provisions & allowances' is high, and a more significant determinant of schooling for declined only slightly from 11.5 percent to 9.3 females than for males. percent over the period. Utilities ­which The proportion of tertiary students whose include the utility costs of the boarding facili- parents are professionals has increased over ties­ accounted for 7 to 10 percent. `Training time, while the proportion of tertiary students and staff development' is below 1 percent and from `farmer' households and from other cate- `research, publications, conferences, work- gories with non-professionals as parents has shops' together account for 1­2 percent. Also, diminished over the years. As Annex 7.2 indi- development expenditure for sub-vented cates, graduates whose parents are profession- organizations has become almost non-existent. als (higher income groups) accounted for 22 As a result, UNIMA increasingly uses its recur- percent of the traced 1980 cohort but rent budget for development expenditures. increased to nearly 50 percent by 1999. Con- As can be seen from Annex 7.3.b, important versely, students whose parents were subsis- sub-items like `common services' comprise tence farmers comprised 33 percent of the `superannuation' and `accumulated deficit'. 1980 cohort but this had shrunk to only 21 Respectively these accounted for 33 percent percent by 1999. There were also marked dif- and 4.5 percent of total `common services' in ferences in the socio-economic status of female 2001. However, the 2002/2003 estimates (not and male graduates: among all four cohorts, actuals) indicated an expected reduction for there were more female graduates from the `superannuation' to 2.4 percent of common higher income groups than in the case of the services, while `accumulated deficit' and male graduates. `advances revolving fund' rose to 16.8 percent 90 Cost, Financing and School Effectiveness of Education in Malawi Table 7.1: Unit cost in USD by college, UNIMA and Mzuzu University, 2000/01 Bunda Kamuzu College Agriculture Chancellor Nursing Polytechnic Medicine Mzuzu Emoluments + benefits 1,330 1,144 2,267 931 11,620 n.a. Student provisions + allowances 275 301 534 273 828 n.a. Teaching material + equipment 49 30 59 86 1,224 n.a. Other 620 669 1,807 564 4,799 n.a. Total per student 2,274 2,144 4,667 1,854 18,471 1,566 Source: Expenditure data and number of students received from University of Malawi. Mzuzu Univ. expenditure data from Consolidated Annual Appropriation Accounts for Financial Year Ended June 2001. Note: Exchange rate 2000/01 was 65.9. and 8.3 percent respectively. Other important Authority and with a number of external cred- items here are `conference & workshop' (11.3 itors for sundry goods. Here, Annex 7.3.c percent), `examination/entrance' (8.9 percent), shows a breakdown of unit costs in US$ for the `sports association/soccer club' (8 percent), last four available years: the total unit cost is `staff welfare' (7.8 percent) and `research/pub- US$2,500­3,000, of which US$1,200­1,600 is lication' (6.5 percent). allocated to emoluments and benefits and Real recurrent expenditures at UNIMA about US$250­300 to students provisions/ declined steadily between 1993/94 and allowances--this does not include the utilities 1996/97, then spiked for one year in 1997/98; component of boarding. On the other hand, and then fell again, before starting to climb slightly less than US$100 is spent on teaching more steadily (see Figure 3.4 in Chapter 3). materials and equipment. Table 7.1provides a Overall growth over this period has been breakdown of unit costs by the different col- roughly ten percent and this is lower than for leges. MOEST expenditure. When the unit cost, presented in Table 7.1, Unit costs. Real unit costs have risen sub- is combined with the student/lecture ratio in stantially over the period 1993/94 to 2000, as Table 7.2, the direct correlation between the summarized in Table 3.2, Chapter 3. However, student/lecturer ratio and the unit costs at each although unit costs have increased, they have college are apparent. By any measure, and fluctuated from year to year and this move- especially when compared to other countries ment indicates uncertainty in the flow of gov- (see Figure 7.3), the colleges have inefficiently ernment funds in periods of fiscal stress. This low student/lecturer ratios. In particular, situation is confirmed by the fact that over the Kamazu College of Nursing and the College of past two years (2002­2003), the university has Medicine have extremely low student/lecturer accumulated considerable tax and pension ratios combined with much higher than aver- contribution debts with the Malawi Revenue age total unit costs. Per student spending on Table 7.2: University enrollment and lecturer numbers by university and college, 2001 Bunda Chancellor Kamuzu Polytechnic Medicine Mzuzu Univ. Total Students 592 1,555 270 1,239 87 385 4,128 Lecturers 73 194 54 148 86 51 606 Student/lecturer Ratio 8 8 5 8 1 8 7 Students per non-teaching staff* 6 23 29 18 3*** 3** 9 Source: Basic Education Statistics 2001. *Data from 1999; ** from 2001; *** in 2003 (in 2001 was 1:1). University Sector 91 teaching material and equipment appears to be ties need to focus on applied research, to pro- low at most of the colleges, except the College vide full tuition-based courses and to ensure of Medicine. While there may be credible that they benefit from any consultancy work explanations for these discrepancies, such as carried out by university staff, which in most the general cost and features of the individual cases is carried out using university resources. programs, the data indicates that it is necessary Current government policies for university to analyze further the discrepancies, and education make the provision of higher educa- address inefficiencies found in the allocation tion very costly. For instance, the Education Act and use of resources in this sector. links enrollment with number of bed spaces in Budget Allocation. The universities rely hostels, and requires the provision of free heavily on government for essential resources. boarding and loans for all students. These poli- Unfortunately, in the current economic envi- cies, while necessary in some cases, impose a ronment, government is unable to guarantee significant administrative and financial burden the steady and timely supply of the necessary on the colleges. As such, the funding and the resources to support essential teaching activi- structure within which these services are pro- ties and stimulate quality research. Except for vided, needs to be revamped as the universities Bunda College of Agriculture, the College of have drifted further towards precarious cash Medicine, and to some extent Mzuzu Universi- flow situations, with substantial unpaid fees to ty, the remaining institutions have not devel- various entities and with associated high cost oped adequate capacity, or produced realistic borrowing needs, because of these mandated plans, for how to generate their own resources generous provisions for tertiary students. It is to supplement meager government resources. therefore recommended to end the provision of This is relevant since, at present, the institu- free boarding, and only offer loans to the more tions only generate about 10 percent of their needy, and to regulate repayment. total income. Moreover, the revenues generat- Role of Development Partners. Over the ed by tuition fees are limited since the fee lev- years, development partners have played an els only constitute a fraction of the real costs. important role in the development of universi- In addition, instead of collecting fees directly ty education in Malawi. Their contribution has from students, the universities are forced to been in the provision of infrastructure, learn- rely on the government to pay these fees. Until ing and teaching resource materials, including recently, all students received a `loan' of equipment. Donors have also provided fellow- MK25,000 from the government cover the ships and technical assistance. However, in the tuition fee.2 While this source of income is, in absence of a strategic university development principle, limited, it should nevertheless ensure plan, donor support tends to depend on how that the universities have access to some rev- active the college administrations are in solicit- enue but, again, these resources are generally ing resources. not provided in a timely manner. Student Financing. The PER indicates that in The present government debt to the univer- 1997, when tuition fees were MK1,500, the sities for outstanding fees is in excess of MK actual collection rate was 18 percent and that 250 million. This underscores the urgency of the total amount collected only represented 1 exploring alternative ways of generating rev- percent of public expenditure to the universi- enues to supplement government contribu- ties. At present, all students are theoretically tions. In addition to defining fee levels, and the required to pay a tuition fee of MK25,000 but mechanism for a loan system that will oblige a tiny minority in some of the colleges are beneficiaries to contribute to the cost of higher required to pay a higher fee--this is supposed education through cost-sharing, the universi- to be an `economic fee'. Further, government 92 Cost, Financing and School Effectiveness of Education in Malawi has decided that in future all students who are which recent data is available--the student to accepted into university will be provided with lecturer ratio in Malawi is the lowest (with the a loan of MK 25,000. In practice, this policy exception of Equatorial Guinea). Not only is has already been implemented since, as men- Malawi a low-end outlier, but it is also far tioned, in 2002, all students received this so- below the ratio of the majority of countries called loan. However, there are no clear regu- (the non-weighted average is 17.8). This indi- lations to ensure repayment and these loans cates that if Malawi manages to keep subject appear to be more in the nature of student area specialization under control, it should be grants. in a position to substantially increase its num- ber of university students without needing to take on significant numbers of extra staff. Efficiency The universities also have a very low ratios of student to non-teaching staff: the average in In the university sub-sector, there are a number 1999 was 11 students per non-teaching staff of inefficiencies that limit the institutions' member. This ratio has possibly deteriorated in quantitative and qualitative outputs. Staffing the interim because the available data for 2003 levels, in particular, affect the efficiency of the show that the ratio in the Polytechnic College institutions and their resource utilization. On was 3, and in the College of Medicine and the positive side, the student to lecturer ratio Mzuzu University it was 4. These low ratios and the student to support staff ratio are are related to the numerous additional services extremely low and this could facilitate expand- that the colleges provide (for example, clean- ed access without concomitant and significant ing, cooking, maintenance, gardening, increases in staff numbers. `helpers', etc.). These employees contribute to Student Lecturer and Student Non-Teaching the large institutional wage bills and to high Staff Ratios. Table 7.2 shows average lectur- unit costs in the system. On the positive side, er/student ratios of 7 in 2001, and indicates a the UNIMA Council has authorized all col- consistently low ratio for all colleges. leges to out-source non-core activities to help Figure 7.3 confirms that in comparison to reduce costs by limiting the number of non- all other Sub-Saharan African countries--for academic staff. Unfortunately, colleges have Figure 7.3: Student to lecturer ratios at tertiary level in Sub-Saharan Africa, 2000/01 4 6 5 0 4 5 4 0 3 3 3 3 3 5 2 7 2 9 3 0 3 0 2 0 2 2 2 2 2 3 2 5 1 6 1 6 1 6 1 7 1 8 1 8 1 8 1 9 1 9 2 0 1 3 1 3 1 3 1 4 1 5 1 0 1 0 1 2 1 2 1 2 1 5 5 7 8 9 9 1 0 5 0 Chad CAR Mali Guinea Leone Benin Togo Sudan Malawi Angola Gabon Niger Rwanda Congo Africa Ghana Uganda Congo Eritrea Liberia Comoros Burundi Djibouti Morocco Namibia Ethiopia Eq. Botswana Lesotho Mauritius DR Mauritania Sierra Mozambique South MadagascarCameroon Guinea-Bissau Source: UNESCO Education website statistics. Note: some ratios are for 1999/00 or 1998/99. University Sector 93 been unable to implement this strategy because sities will not be able to expand access. Mzuzu government has yet to produce the necessary University has had some success using limited retrenchment packages. funding to use prefabricated structures in hos- Use of Facilities. All tertiary colleges have tel construction. Furthermore, some colleges adopted a two-semester academic year, leaving have begun to ignore this accommodation pro- the institutions idle for about 3 months in the vision in the Act and have started enrolling year. In addition, most of the colleges do not non-residential students in parallel programs. operate a credit hour system that would allow These students, still few in number, pay closer students to enroll as, and when, they are in a to unit cost fees unlike government-sponsored position to mobilize the necessary tuition fees students. and thus increase enrollment. With improved Role of Central Administration. The MIM use of existing facilities and of teaching and Report observed that central administration is non-teaching staff, enrollment could be preoccupied with operational matters when it increased without substantial additional costs. should really be concentrating on providing a A clear option would be to increase the num- strategic policy framework for university opera- ber of shorter diplomas, or individual courses tions and activities. Efforts should be made to for which there is a real demand in the labor implement the recommendations of the MIM market. These might be provided during the Report in terms of delivering a well-planned idle times in the institutions (e.g. night, week- decentralization program--within a defined end and holidays) on a cost recovery basis. framework and with appropriate guidelines. Duplication of Programs. There is duplica- This would result in the transfer of some of the tion in many of the programs offered by the functions, carried out by central administration, universities. For example, Biology, Chemistry, to the constituent colleges and would improve Law and Home Economics are offered in more efficiency in decision-making processes. than one institution. This wastage of scarce resources could be reduced, or eliminated com- pletely, through a rationalization process that Relevance would ensure disciplines, which re-enforce each other, are only offered in the same institu- Over time, the investment in infrastructure and tion. There may, however, be a need for bal- quality inputs has deteriorated,3 leaving most ancing this process against trying to provide institutions with dilapidated buildings, inade- access to popular programs in different urban quate and outdated equipment and lack of areas and regions. A potential option is to learning and teaching resources. The relevance explore the creation of `satellite colleges', of the teaching and research programs, and the which offer more general first year, and under- skills and knowledge instilled in students, are graduate programs with limited specialization. critical for later employment and for the inno- Student Accommodation. The Education vations that graduate students bring to the Act limits enrollments to available bed space in market place. Currently, however, there is an the colleges. Colleges are therefore forced to absence of a clear mechanism linking universi- provide accommodation for their students, but ty research and teaching programs to the lack of capital investment has inhibited expan- human resource and skill requirements of the sion of the hostels, which in turn limits enroll- economy and labor market. There is also inad- ment. The recurrent cost implication of equate linkage between research and teaching expanding boarding places is substantial and activities. As a result, teaching has not benefit- unsustainable. As long as cost sharing, or cost ed from new knowledge being generated recovery, mechanisms are not in place, univer- through research activities. The lack of train- 94 Cost, Financing and School Effectiveness of Education in Malawi ing programs and refresher courses for lectur- of science and technology, which are important ers raises doubts about the relevance and for industrialization and economic growth. appropriateness of the current content being Finally, Malawi, like many other countries, has taught and about the teaching methodologies experienced a significant brain drain. This is being used. particularly true of graduates from the Employment of Graduates. The relevance Kamuzu College of Nursing, many of whom and quality of university education merits tend to leave Malawi to work in countries such some review. A recent study on skills for as the US and the UK. The government will employment concluded that graduates do not have to address this issue through incentives have the skills needed by employers, nor are and by establishing accountability mecha- they equipped to start their own enterprises.4 nisms. Limited contact between higher education Relevance and Skill Utilization. The tracer institutions and employers was cited as the study also sought the opinion of the graduates main reason for this poor preparation for the themselves on a range of topics relating to their world of work. university education and their subsequent On the other hand, and probably as a result employment. It found that graduates appear to of the low number of graduates from the uni- be generally quite satisfied with most aspects versities relative to the need for university of their university education, with the overall trained professionals in the country, this lack satisfaction level highest among the 1980 of contact has not really affected employment cohort and slightly lower for the latter cohorts. opportunities. A recent tracer study on univer- While over 80 percent were satisfied with the sity graduates and secondary leavers5 found knowledge and skills acquired, less than one- that unemployment of university graduates third were satisfied with their information and was negligible in 2001. Only three of the 380 technology skills and just under half with their traced graduates (0.8 percent) were unem- managerial and entrepreneurial skills. Annex ployed. 7.4 provides the response of the graduates to a However, there are indications that employ- number of questions on the relevance of their ment prospects declined among the 1990s university education. A second table in Annex cohorts relative to the 1980 cohort. Only forty 7.4 shows the prioritized concerns of universi- percent of 1994 graduates secured employment ty graduates. immediately after graduating from university To summarize, it is clear that the university in comparison to 90 percent of 1980 gradu- sector in Malawi needs to be restructured with ates.6 Employment opportunities for graduates respect to funding, efficiency, quality and rele- varied according to the program of study.7 vance. The country needs to expand university In order to expand the private sector and education enrollment and graduate output, but diversify the economy, it will be critical to have it must do so strategically to ensure that pro- more graduates becoming private sector entre- grams and graduates meet the demands of both preneurs, and the curriculum should be adjust- society and the economy. ed with this goal in mind. Only seven of the 380 traced graduates were self-employed in mid 2001, only 15 in total had ever been self- Notes employed and the seven self-employed gradu- 1. Kadzamir et al. ates employed only 45 other people in total. 2. In the academic year 2002, `loans' were Particular attention should be paid to the areas provided to a total of 4,000 students. University Sector 95 3. Chapter 3: Educ. Financing indicates the 7. Bunda College, Kamuzu College of Nurs- complete breakdown of (on-budget) develop- ing and the College of Medicine have a higher ment spending to universities. employment track record for their graduates 4. Thomas et al. 2001 because of the nature of their sectors. Chancel- 5. Kadzamira et al. lor College and the Polytechnic, which both 6. As a result, unemployment was quite high have higher enrollments, face some challenges (60 percent) in the immediate period after in providing programs that address the skill graduation. On average, it took 1980 gradu- needs of the sectors and in producing employ- ates less than a month to find their first job. able graduates. For the 1999 cohort, this period was three months but this is still low. CHAPTER 8 Critical Challenges and Policy Choices T his study has provided the analytical (iv) increase external (donor) funding, or a underpinning for synthesizing key combination of the above. issues and this chapter highlights the Furthermore, it is important to ensure a critical challenges. Using the results more stable and predictable flow of funds to of a financial simulation model, an attempt is the education sector if the long-term benefits also made to lay the foundation for realigning from this investment are to be harvested. Fund- the policy targets with resources. ing of the sector should be in line with a well- defined, long-term strategic plan. Sub-sectoral allocations and allocations by Macroeconomic context and type of school need to be closely aligned with education expenditures education policy goals. Although primary edu- cation obtains the highest share of recurrent For a country undergoing a transition period expenditures, it is not clear whether its share of of huge enrollment expansion (FPE in primary, development expenditures is in line with the and CDSS expansion in secondary), the priori- goals of universalizing access. A striking incon- ty given to education, as reflected in govern- sistency in sub-sectoral allocations is also ment spending, has been relatively modest in reflected in the stark differences within the the last 10 years in the range of 12­20% of dual secondary system, the CDSSs and the total government expenditure. However, in the CSSs, and by the higher share of resources pro- past couple of years, government spending has vided for tertiary relative to secondary. been in the higher end of this range as agreed Unit costs in education are highly unbal- within the framework of HIPC. As the econo- anced between the university level and all other my is unlikely to generate much in additional levels, and within secondary between tradi- resources, the government will either have to: tional government schools and CSSs and (i) improve the efficiency (productivity) of the CDSSs. Here, the government will need to system and resource distribution, (ii) increase define acceptable levels of per pupil spending the share of public spending on education, (iii) and adjust public financed access at the post- increase cost-recovery at post-primary or primary level. This may entail capping public increase access on a full cost-recovery basis, financed expansion at secondary level, 96 Critical Challenges and Policy Choices 97 although it would not exclude expansion Pupil to Qualified Teacher Ratio: There is through private provision. an urgent need for additional qualified teach- Managing expenditure on personal emolu- ers to reduce the high PqTR of 123:1 in pri- ments will be crucial to achieving sustainable mary schools. This will require a continued levels of spending. Average salary levels have effort to produce approximately 5,000 teach- been declining in real terms, partly due to the ers per year for the next 10 years, facilitated by increase in untrained teachers--especially at continuing the present in-service training pro- primary and secondary levels. As the number gram with at least 2 cohorts graduating per of qualified teachers increases, and as more year. Investing in teacher training is a pre-con- teachers are employed due to higher enroll- dition for improving the quality of primary ment and a lowering of the PTR, personal school in Malawi. emoluments will rise. Secondary: CDSSs, which represent about In secondary education, medium-term tar- 70 percent of secondary enrollment, are gets need to be set for access to quality sec- extremely under-endowed in many ways, for ondary education, in such a way that the instance in the availability and qualification of financing requirements are sustainable. teachers, in the range of teaching/learning Questions regarding cost-effective organiza- materials, and in the dilapidated school infra- tion and delivery need to be addressed to structure. Although the PTR is 40:1 in second- ensure that personnel costs do not crowd out ary, the high incidence of primary teachers quality improving, non-salary inputs. While teaching at CDSSs has produced a PqTR of a rapid geographical expansion could 120:1. As such, a determined effort is required increase access, it may result in available to train teachers and fund schools adequately, resources being used almost exclusively for which may entail putting a cap on the number salary expenditures, understaffed schools, of publicly financed secondary schools. and virtually no investment in crucial educa- Safety net for the youngest and orphans: tional resources. The government might consider providing nursery schools for young children. While attending pre-school is educationally desirable, Enrollment, repetition and survival in the case of Malawi an expansion in publicly finances nursery schools for pre-school chil- Primary completion rate and internal efficien- dren may provide a strategy to reduce the num- cy: Many children do not progress beyond ber of under-aged children starting school, and lower primary because of high repetition rates. thus reduce the overcrowded classroom in The level of repetition is very high throughout Standard 1. One option is to use the commu- primary as a whole ­an average of 25% of nities and provide minimum incentives to each grade repeat every year, and in particular locals to attend to the children at the school in Standard 1 where about 40% of pupils premises. These nursery schools could also repeat. This distorts the system right from the facilitate improvements in under-five malnutri- start. Dropout in primary is also relatively tion provisions. high, at between 5 and 10 percent in each grade. However, between 1992 and 2000, the Primary Completion Rate increased by about Equity 15 percentage points from 20­25 percent to 35­40 percent. To address problems of repeti- With the introduction of FPE, public expendi- tion and dropout, the improvement in overall ture on primary education became more equi- quality is a critical element. table, however overall public education spend- 98 Cost, Financing and School Effectiveness of Education in Malawi ing remains regressive and inequitable. As the children perform quite well in Life Skills, such, focusing resources more equally among Malawi had the lowest percentage of pupils the different grade levels in primary will have a achieving `minimum' and `desirable' mastery positive and equalizing effect. It may simulta- levels in English literacy, relative to six other neously improve both efficiency and quality by countries in the region. reducing repetition and dropout. In Malawi, using the traditional school- While the primary survival rates for boys based variables, there appears to be little rela- and girls are relatively even, there is a signifi- tionship between inputs and outcomes. Predic- cant discrepancy in survival rates in rural and tors for improvement in school achievement urban areas: urban students have about 20 to include: absolute resources levels (unit costs), 30 percentage points more chance of survival the PTR and the qualification of teachers. The through to the end of primary. In addition to weak correlation, however, means that other microeconomic conditions like poverty levels, important factors are associated with the there is a clear difference between rural and learning process in schools, for example, the urban areas in PTRs and, especially, in PqTRs. limited teacher time-on-task. Also, absenteeism In addition, rural areas lack female teachers is a widespread problem, both among pupils and this deprives rural girls of potential role and teachers. This results from multiple fac- models. The recently implemented decentral- tors, such as illness or hunger. The level of ization policy may provide an opportunity to absenteeism also indicates that institutional prioritize rural areas in future teacher alloca- accountability, or incentives to perform, are tion decisions. For this to effectively take place, largely non-existing. Further, pedagogical man- the overall composition of salary (salary, pro- agement, and management in general, appears fessional and housing allowance, potential to be quite weak. Any mechanisms established rural hardship bonus, and other) needs to pro- to improve student learning and the quality of vide incentives for teachers to stay in rural schooling will need to: (i) improve accounta- areas, and a functional teacher rotation system bility and devise teacher incentives to perform needs to be put in place. assigned duties, (ii) ensure timely and adequate At the secondary level, two crucial inequities provision of learning materials, (iii) empower are noted: the type of school in which pupils headmasters and improve school management, enroll and urban/rural differences. Urban and (iv) help low-achieving schools perform youth age 14­17 years are three times more better. likely as their rural peers to attend secondary Prevalence of malnutrition affects survival school (NER 21 versus 7 percent). Regional and learning capacity in the education system. differences in the secondary NER are also con- Malawi has one of the highest levels of stunt- siderable and follow a similar pattern as the ing in children in eastern and southern Africa. primary level.1 Differences in access by socio- Stunting and micronutrient malnutrition are economic status at the secondary level are associated with reduced capacity to learn, large, as seen from the GER of 62 among the delayed school enrollment, decreased school wealthiest youth and 11 among the poorest attendance and, eventually, with dropout. youth. These problems are especially critical during the hunger period from November to April. Malawi is chronically food insecure because of Quality of education and externalities regular food shortages in some areas and sea- sonal food shortages in most of the country. A The performance, or learning outcomes, of minimum supply of de-worming and iron-for- Malawian students is unacceptably low. While tification could contribute significantly to the Critical Challenges and Policy Choices 99 overall nutritional status of pupils. In addition, allowance'2) to encourage, and obligations to the government may wish to sustain current ensure that qualified teachers are available to emergency food structures or to establish a teach in rural areas. The decentralization direct support to schools, financed by donors, process offers an ideal opportunity to rebal- to ensure cash transfers that could be used to ance these inequities in the teacher deployment purchase food in times of crisis. system. For example, under a decentralized The on-going HIV/AIDS crisis will put fur- management, the resources allocated from the ther pressure on the education system, and central to district level to hire and pay teachers result in higher dropout and repetition rates, will be based on a neutral formula and reflect and increase the number of orphans who will the number of pupils attending--with some be especially vulnerable in relation to educa- adjustment criteria for poverty levels or popu- tional achievement. Government will need to lation density. The direct management of these develop an effective HIV/AIDS policy in order funds will allow rural districts to advertise to prolong the active professional lives of open posts as soon as teachers leave. The PER infected teachers. Ensuring a supply of flexible has already indicated that local recruitment of voluntary teachers and teaching assistants, teachers might be another way to reduce the could also counter teacher absenteeism. A seri- tendency of teachers to move to urban areas. In ous debate on the issue of reducing absences the context of the current in-service teacher due to funeral attendance would also be advis- training framework, this is feasible because able. In the same way, the issue of funeral ben- teachers are already teaching in a particular efits needs to be addressed so that this item area, and only relocate to the teacher training does not completely absorb expenditure on college (TTC) for a relatively short period basic teaching inputs. Possible options, here, before returning to their schools. This reduces are to set up a special fund for this purpose and the probability they would automatically move reduce funeral costs by allowing sick teachers to an urban area after leaving the TTC. to transfer to their home districts. One potential measure to increase the num- ber of qualified teachers in the system is to review the current teacher retirement age. Of Management of the system course, this may have a limited impact in a country where the life expectancy is 8 years Teachers' deployment is highly inconsistent. (although life expectancy may be higher Some of the specific difficulties have to do with among teachers). the manner in which the present system signif- Teacher remuneration scale setting. Since icantly favors urban areas. This situation is many new teachers are entering the system further aggravated by the significant discrep- each year, and given that unqualified teachers ancy in Pupil to Qualified Teacher ratios will need to be trained, this is an opportune between urban and rural districts; in urban dis- moment to establish a sustainable salary policy tricts it is between 60 and 70, in rural districts for the next 10­20 years. As indicated in a it is approximately double this figure. recent World Bank study on the Education For The management and deployment of teach- All-Fast Track Initiative,3 a guideline for aver- ers to schools needs to be based on an estab- age qualified primary teacher salary, in African lished PTR (funding formulae) and on the countries, is about four times GDP per capita. availability of teaching places. The MOEST The study suggests introducing a dual system needs to develop an equitable PTR for each with the new salary scale only applying to school. In line with this, there is a need to newly qualified teachers.4 The qualified explore possible incentives (e.g. a `rural teacher salary in Malawi is currently slightly 100 Cost, Financing and School Effectiveness of Education in Malawi higher, at about 5.5 multiples of per capita aggregate level, and will facilitate comparisons GDP; on the other hand, this may be a reason- at divisional/district level. able level for Malawi where per capita GDP is EMIS. Availability of complete, timely, and lower than in many other African countries. reliable education statistics is critical to consis- An adjustment of teacher salaries in Malawi tent planning of the education sector. The cur- should be based on a comparison with earning rent computerized EMIS system provides a in other professional occupations. solid basis, but both the organizational frame- work and the information technology aspects would benefit from an in-depth review. EMIS Budget process, supporting systems should also become more of a functional sys- and harmonization of procedures tem, not just a central database, where data are stored and are used sparingly in decision-mak- Budgetary reforms and improvements in ing. An important part of this is information accounting are required to accurately estimate dissemination from the EMIS to education financing needs, ensure a smooth flow of stakeholders. The providers of information funds, reduce unit costs and monitor budget (i.e. schools) are given an information dividend execution. At a minimum, MOEST needs to at periodic intervals. A summary report could establish an overview of the total flows into the be published on an annual basis, even in news- education sector, especially off-budget funds. papers. The list of potential areas for improve- The second issue is to examine the cash budg- ment of the EMIS is included in Annex 8.3. eting system and the impact it has on the strategic prioritization of medium-term budg- eting and the effective functioning of schools. University education A third issue is to accurately estimate the recur- rent cost implications of existing development At university level, there are fundamental inef- projects and to assess their financial sustain- ficiencies, many of which arise from the small ability. Efforts could be made to harmonize scale of operations. Enrollment is presently at procedures and procurement rules with a view less than 4,000 students and has shown only to using government procedures. Thus accept- modest growth during the past decade. able fiduciary arrangements in the education Malawi, therefore, has close to the lowest pro- sector could pave the way for on-budget sup- portion of tertiary education students per port. 100,000 inhabitants relative to the whole of The introduction of a new account struc- Sub-Saharan Africa. The sector needs to ture--put in place to improve budget planning expand access in order to improve the utiliza- and analysis capacity--has actually made it tion of existing resources, service the expand- very difficult to analyze anything at a level ing education sector and provide more profes- below emoluments and other spending (ORT). sionals to the workforce. Different divisions/districts put major resource There are a number of challenges that face allocations under different headings, making higher education, if it is to expand access and consolidation in a more detailed manner near- improve quality and relevance. It must: (i) ly impossible. It is crucial, therefore, to estab- establish a national strategy and policy frame- lish simple, clear guidelines and, especially, to work for higher education which will include provide support training (at all levels, especial- funding guidelines, accreditation criteria, ly in the divisions/districts) on the use of budg- establishment of private universities, credit eting and accounting. This will ensure consis- hour system and new delivery modes, such as tency in recording so analysis is possible at an distance education; (ii) price out the real cost Critical Challenges and Policy Choices 101 of university education in order to establish were deemed unrealistic in the current macro- student financing and support schemes; (iii) economic and social context. The ensuing revise tuition fees to lessen the government's results are presented as a basis for discussion financial burden; (iv) reexamine the present with government and donors. The detailed sce- policy of providing accommodation and feed- nario and assumptions are described in Annex ing to students. The linkage of student enroll- 8.1.7 ment to bed space, although ignored by some colleges, needs to be reviewed and the relevant section in the Act amended accordingly5; (v) Changes in repetition and dropout make a concentrated effort to implement the rates recommendations in the MIM Report on out- sourcing of non-core activities. To successfully If the situation regarding repetition, dropout implement this initiative, government will need and admission rates evolve as described in the to develop the required staff retrenchment PIF, this would result in an estimated primary packages. enrollment of 3.5 million by 2012. The num- ber of qualified primary teachers needed to Realigning policy targets to resources achieve the PIF target PqTR teacher ratio of 60:1, under this assumption, would be 58,000 The Policy and Investment Framework for the by 2012. If no changes occurred to the current education sector, developed in the late 1990s, repetition, dropout and admission rate, then provides an excellent basis for moving for- the estimated enrollment for primary is 4.2 ward. However, the fiscal and external condi- million by 2012. The number of qualified pri- tions for implementing the PIF have changed mary teachers needed to achieve the PIF target, since its preparation. Furthermore, while the under these assumptions, would be 70,000 by PIF is comprehensive it needs to be prioritized 2012 or 12,000 more teachers. It is clear that and aligned to existing internal and external efforts to reduce repetition and dropout rates resources as will be demonstrated in the fol- are critical in relation to both how many teach- lowing simulation model.6 The model covers ers will need to be recruited, and which teacher the period up to year 2015, analyzes the long- training model will be pursued. Obviously, term financial sustainability of the system, and both will have an impact on recurrent costs checks the realism of the general policy targets. and additional funding required. It is based on one possible scenario of targets Under the `in-between' scenario shown and assumptions that were discussed with gov- above, where repetition and dropout rates are ernment and it largely reflects the PIF. Howev- reduced gradually (but still optimistically), pri- er, some alterations were made when PIF goals mary enrollment would reach 3.9 million by Table 8.1: Projections for primary education: enrollment, repetition, and PqTR, 2001­2015 PRIMARY 2001 2002 2003 2004 2005 2006 2009 2012 PIF 2015 Total enroll.8 3236319 3300423 3407299 3505522 3582180 3662103 3820127 3973231 4158317 Avg. repetition 25.2% 23.3% 21.9% 20.7% 19.7% 18.6% 15.9% 13.4% 11.1% Qualified teachers9 31 799 31 891 31 978 34 859 37 567 42 113 51 948 59 101 65 041 Unqualified teachers10 21 645 22 281 23 782 22 372 20 813 17 502 10 258 5 977 3 855 Total teachers11 53 444 54 172 55 759 57 231 58 380 59 615 62 206 65 078 68 896 New teachers trained 1923 2000 2000 4800 4800 6800 5650 5650 5650 Pupil: Qualified Teacher Ratio 101 103 106 100 95 86 73 67 64 102 Cost, Financing and School Effectiveness of Education in Malawi 4500000 140% 70000 4000000 120% 60000 3500000 100% 3000000 50000 2500000 80% 40000 2000000 60% 1500000 30000 40% 1000000 20000 500000 20% 0 0% 10000 0 200120022003200420052006200720082009201020112012201320142015 20012002 20032004 20052006 20072008 200920102011 20122013 2014 2015 Total enrollment in primary Gross Enrollment Rate Expected number of qualified teachers Standard 6 proxy Completion Rate Expected total number teachers Source: Authors' calculations. Base year data are from EMIS 2001 and ED Data 2002. 2012 and the expectation would be 59,100 has been balanced and the need for new teach- qualified teachers (92 percent of all teachers) ers reduced, a teacher training program of a and a 67:1 PqTR by 2012. The trend in PqTR longer duration may be considered. would still be downward and the 60:1 goal could be reached over time because more teach- ers would be trained than would be lost due to Secondary education and secondary attrition. This would assume the stock of qual- teacher training ified teachers to be sufficient to achieve a 60:1 PqTR, given attrition at maximum 6 percent. In the scenario shown in Table 8.2, secondary This evidently is a more realistic scenario but enrollment can be expected to reach 393,000 will require effective interventions to reduce the by 2012, based on the assumption of main- number of pupils repeating or dropping out. taining GER in secondary at 30 percent.12 One very important conclusion concerning Since the GER in the base year (2001) was the teacher training model can be derived: if already 29 percent, the growth in enrollment the teacher training model reverts to a 2-year numbers will be in line with the growth in pre-service training program, the number of school age population 14­17.13 graduate teachers produced every year at the The Table shows that, in order to achieve a existing TTC and an additional new TTC, will PqTR of 46:1 by 2012, 8,600 qualified sec- be so little that by 2012, only the stock of qual- ondary teachers are needed. This could be ifies teachers will be only 33,700, and the achieved by expanding secondary teacher PqTR will be as high as 115:1. Even if two training from the current about 400 to 970 additional TTCs are built, each with a capaci- graduates every year. If the attrition rate of ty of 625 students, the stock of teachers will teachers increases to 8 percent (instead of the 5 only reach 40,000 by 2012 corresponding to a percent assumed in Table 8.2) as a result of PqTR of 96:1. HIV/AIDS, the PqTR would only reach 53:1 Political commitment and potentially peda- by 2012 unless even more teachers are trained. gogical reasons (although in Malawi the verdict If, however, the current capacity in second- is still out on the latter aspect) could advocate a ary teacher training is not changed and the return to the pre-service system. However, the SSTEP program is phased out beyond its initial focus now should be on accelerating as much as 3 cohorts, and the capacity of the university possible teacher training, and once the system colleges is maintained roughly at its current Critical Challenges and Policy Choices 103 Table 8.2: Projections for secondary education: enrollment, PCR, repetition, GER and PqTR, 2001­2015 SECONDARY 2001 2002 2003 2004 2005 2006 2007 2009 2012 2015 Total enrollment14 274949 284036 293418 303106 313109 323437 334101 356480 392847 428025 Gross Enrollment 29% 29% 29% 29% 29% 29% 29% 30% 30% 30% Qualified teachers15 2290 2547 3116 3681 4242 4750 5483 6840 8631 10167 Unqualified teachers16 5260 5127 4685 4252 3826 3458 2870 2072 1190 534 All teachers17 7550 7674 7801 7933 8069 8208 8353 8912 9821 10701 New teachers trained 347 372 696 721 745 720 970 970 970 970 Pupil:Qualified Teacher Ratio18 120 112 94 82 74 68 61 52 46 42 Source: Authors' calculations. Base year data are from EMIS 2001. level, then, by 2012, the system will only have 27 percent (PIF target), then the average annu- about 5,700 qualified teachers resulting in a al deficit would fall to about 6 to 8 percent of PqTR of 70:1. The PqTR would not improve the national `discretionary' recurrent budget. further, even with an attrition rate of 5 percent. As has been shown above, the model targets Finally, if secondary enrollment were to and assumptions indicate that recurrent cost increase substantially, for example reaching an implications of the different policies would be enrollment of about 600,000 by 2012 (corre- larger than the domestic resource availability. sponding to a GER of about 45%), the number It is clear that MOEST, MOF and the donor of qualified teachers needed to achieve a PqTR community need to determine the realism of of 40:1 would be more than 15,000, thus this type of scenario. For instance, the expect- requiring far more new secondary teachers. In ed total resources available and the teacher other words, public demand for secondary salary costs (both unit cost and total cost) are education is very likely to increase beyond the crucial issues for dialogue. Other issues, such expected capacity to train teachers for second- as planning teacher training, must not be done ary and especially the capacity to pay their in isolation from the general issue of sustain- salaries. If the government does not put clear ability, which centers on the ability to pay limits on publicly funded secondary education, teachers once they are hired and trained. It is and expand support for private secondary edu- also crucial to understand what the expecta- cation, it may face a critical political situation. tions are in terms of resource availability under the HIPC debt relief program. The results of this financial model can facilitate such a dia- Recurrent cost implications logue. It is clear that the magnitude in the financing gaps or shortfall in reaching the set The recurrent costs of education from pursuing policy targets should be used to reopen and the scenario described above will increase, redefine targets and priorities. because of the additional teachers required by 2012,19 and due to the incremental cost of hav- Possible priority areas of policy ing a larger proportion of qualified teachers. intervention The government may expect to have an aver- age annual recurrent expenditure shortfall Table 8.3 lists the most urgent policy reforms in between now and 2012 of between 11 to 13 the education sector, as discussed in the various percent of the total national discretionary chapters of this Study. Most of the proposed budget, if the education expenditure as a per- actions are already known by government and centage of the national `discretionary' recur- donors, and have been flagged for the past rent budget is maintained at 21 percent. If the decade. The Table also provides an evaluation education expenditure share were increased to of the feasibility of such policy interventions. 104 Cost, Financing and School Effectiveness of Education in Malawi Table 8.3: Possible areas of policy intervention, including assessment of Cost (C), Feasibility (F), Political Risk (P) and Success Probability (S) (low, medium or high) Policy Options: Policy Options: Objective Medium Term (<4 years) Long term (4 years +) C F P S Re-balance the · Recalibrate PIF (EFA) goals to · Take a strategic long term view L H M H education system financial and human resources and establish caps on public · Amend Education Act funding for secondary and tertiary education Re-establish · Establish accountability at · Restructure teacher salary M H M M accountability, and school level through social provisions establish incentives contracting between · Empower headmasters (role in for teachers and communities and schools and promotions and in reprimanding headmasters to transparency systems teachers) perform their duties · Reward/Bonus scheme at · Improve supervision and school level enabling inspection through an increase communities to reward in ORT teachers · Provide career development and · Launch public information rotation, especially for rural campaign about the assignments educational strategy · Ensure teachers accept strategies/consensus building Equitable expansion · Lower repetition and dropout · Target disadvantaged areas and M L L L of enrollment rates to shorten average of population groups school completion (automatic · Utilize classrooms and other promotion/cycles of learning school facilities efficiently Standard 1­4) (double shift) · Target districts or regions with · Use teachers efficiently, redeploy inequitable access using teachers; evaluate and adjust incentives. E.g. Expand food for accordingly the teacher resource education of girls during the requirements under a streamlined hungry season (November to curriculum at secondary level April) · Improve incentives for private provision of education, particularly at secondary level (continued on next page) Critical Challenges and Policy Choices 105 Table 8.3 (continued) Policy Options: Policy Options: Objective Medium Term (<4 years) Long term (4 years +) C F P S Quality and internal · Redeployment of teachers to · Strengthen in-service teacher C H M M efficiency ensure adequate availability of training and other teacher improvement teachers at school level as an professional support. Gradually integral part of the rebuild the pre-service teacher decentralization plans training system in primary (this will · Targeting of PTR in depend on the output capacity of a Standards 1­4 combined in-service (MIITEP) · Curriculum Reform: Continue training system and the partial the Primary Curriculum and pre-service system relative to the Assessment (PCAR) process immediate needs for qualified and complete the proposed teachers in the system) review of secondary curriculum · Continue supervision of teachers · Improvements in the conditions and provide teaching and learning in classrooms, other school materials at primary level, targeted facilities and learning materials provision of teaching and learning through a direct support to materials at secondary level school program financed under · Expand enrollment at teacher the development budget training colleges and at university · Improve community involvement level and through the DSS program · Implement national policy on lower indirect cost of education mother tongue instruction · Distribute iron supplements and support de-worming Efficiency · Appropriate distribution of · Adapt EMIS and funding formulae L M L L enhancement of teachers according to to meet enrollment changes at public resource enrollment district level allocation and · Voluntary re-deployment of · Adjust cross sub-sector resource utilization existing teachers with an allocation for future balanced incentive package or transfer sector development of pupils where feasible · Budget and expenditure framework for all resources/ common procedures using GoM systems Strengthening · Improve data; EMIS, · Strengthen central function of L L L L institutional codification of all schools and sector analysis, policy and arrangements and teachers, merging of MOF and planning, and system monitoring capacity for service MOEST salary and financial · Strengthen district-level capacity delivery databases of local system management, including school inspection, teacher development, monitoring and reporting · Strengthen management capacity at the school level, including the capacity of School Management Committee at primary level 106 Cost, Financing and School Effectiveness of Education in Malawi Notes 7. The model based on EMIS data from 1. 6 percent of youth age 14­17 attend sec- 2001 projects student enrollment until year ondary school in the Southern region, com- 2015. Based on the PIF target of a PTR ratio of pared with 10 percent in the Central region 60:1, certain assumptions of repetition and and 16 percent in the Northern region. dropout rates by each Standard and assuming 2. It would be crucial to ensure that this that two of the six teacher training colleges will allowance is linked to location, not to teacher revert to pre-service training while the remain- so teachers cease to receive the allowance once ing four TTC will continue the MIITEP pro- they leave the rural area. This would require gram, estimate (i) the recurrent cost implica- that appropriate administrative systems be put tions for pursuing the target of a PTR 60:1 and in place to track teacher's locations. hiring more teachers and (ii) the share of qual- 3. Bruns, B. et al. 2002. ified teachers. The model covers primary and 4. It is crucial to review the total benefits secondary education, primary and secondary packages (salary plus allowances) when dis- teacher training, and higher education. These cussing teacher remuneration, particularly as are most relevant for MoEST, since the recur- allowances in Malawi represent higher public rent costs and the human resources availability expenditures than the basic salary element. will most directly influence the sustainability 5. Students currently enrolled in parallel ter- of the education system over time by changing tiary programs pay fees that are closer to the some of the above assumptions. For target real unit costs and in this way they are sources parameters, the PIF targets by 2012 are used, of revenue for these institutions. Such parallel unless indicated otherwise. programs could include students regularly 8. Public and private attending normal classes--but not in receipt of 9. The number of qualified teachers expect- publicly supported boarding--or it could ed is a result of the output of the teacher train- result in students attending a completely mod- ing scenario that is used and of the expected ified university program comprising distance attrition (assumes all qualified teachers work education and/or usage of facilities during off- in public primary schools, not in private ones) time for regular classes 10. The number of unqualified teachers 6. The simulation model is based on the expected is the difference between the total Education For All, Fast Track Initiative model number of teachers expected and the expected but it has been adjusted to reflect the specifics number of qualified teachers. of the Malawian education system, and spe- 11. The total number of teachers expected cially to reflect crucial targets in the PIF docu- in public primary schools is a result of a grad- ment. The enrollment estimates for primary ual expected transition from the current 77:1 are completed separately but they are based on pupil: stream ratio to 60:1 at the end of the the simulation model that was used initially as period, while at the same time the teacher: part of the PIF since it allowed a more detailed stream ratio is set to increase from 0.8 at pres- estimate using repetition rates and dropout ent to 1 at the end of the period. rates by standard. This simulation model was 12. However, it is worthwhile mentioning provided by UNESCO, and adjusted for that in case the percentage of students enrolled Malawi, with the help of Sussex University. For in private secondary schools (thus without key indicators in the model the assumption is requiring public funding allocation) would made that the system moves from the current increase from its current level, the gross enroll- situation to the target value (by the target year) ment rate could grow beyond 30% without in a linear manner. overburdening the public expenditure system further. Critical Challenges and Policy Choices 107 13. This would however mean that the cur- 16. The number of unqualified teachers rent transition rate from primary to secondary expected is the difference between the total of 74 percent would not be maintained as more number of teachers expected and the expected children pass the PSLCE. number of qualified teachers. 14. Public and private. 17. The total number of teachers expected 15. The number of qualified teachers in public and private schools is a result of a expected is a result of the output of the teacher gradual transition from the current 36:1 pupil: training scenario that is used here and of the teacher ratio to 40:1 by 2007, and of the expected attrition. This need for qualified expected total student enrollment. teachers is based upon expected enrollment in 18. This is the average for the secondary sys- both public and private schools, since teachers tem as a whole, i.e. including public and pri- for private schools--if qualified--will also vate schools. have to come from the same teacher training 19. About 12,000 primary teachers and institutions. Of course the teacher remunera- 2,300 secondary teachers. tion cost only reflect those teachers teaching in public schools. ANNEXES 109 ANNEX 1.1 Evolution in population, GDP, total government revenue and expenditure 1991/92 1992/93 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 2002/03 Population (in million) 8.71 8.85 8.99 9.14 9.33 9.55 9.77 9.99 10.20 10.41 10.63 10.87 GDP in current million MK 6331 7801 9721 16080 29534 39447 47810 66881 90413 116963 139357 155387 GDP deflator (1994=100) 58.22 70.55 89.63 140.59 234.53 299.12 349.71 471.09 622.92 783.08 949.84 1026.26 GDP in const. 1994 million MK 10874 11057 10846 11438 12593 13188 13671 14197 14514 14936 14672 15141 Real growth rates n.a. 1.7% ­1.9% 5.5% 10.1% 4.7% 3.7% 3.8% 2.2% 2.9% ­1.8% 3.2% Inflation n.a. n.a. n.a. 58.9% 60.4% 23.4% 19.5% 37.3% 37.2% 29.6%1 n.a. n.a. Exchange Rate (MK/US$, average) 3.20 4.00 6.57 12.01 15.30 15.88 23.76 37.58 51.82 65.87 74.45 76.72 Total Domestic Revenue3 (current million MK) 1,182 1,331 1,629 2,231 4,437 5,943 8,598 12,075 15,808 20,880 22,853 31419 Total Grant (External), including HIPC, but before loan (current million MK) 208 174 284 1,515 1,986 1,708 1,968 4,242 6,296 10,353 9,144 10675 Total Revenues (current million MK) 1,390 1,505 1,913 3,746 6,423 7,651 10,566 16,317 22,104 31,233 31,997 42,094 Total Government Expenditures (current million MK) 1,577 2,379 2,428 6,123 7,864 8,714 13,844 19,475 27,221 37,303 42,490 54015 ­ Current 1,262 1,910 1,964 5,056 6,225 6,976 10,940 13,952 17,638 25,772 32,675 42167 ­ Develop (cap) expenditure 315 469 464 1,066 1,639 1,738 2,904 5,524 9,583 11,530 9,816 11,787 Overall balance after grant (current million MK) ­187 ­874 ­515 ­2377 ­1441 ­1063 ­3278 ­3158 ­5117 ­6070 ­10493 ­11921 Sources: IMF "Selected Issues and Statistical Appendix. August 2002"; WB SIMA database; and authors' calculations. The data for 2001/02 and 2002/03 are estimates from IMF "First Review under the Poverty and Growth Facility and a Request for Waivers of Performance Criteria. August 2003". Population data are from WB-HNP Stats. Inflation and exchange rates are from IMF as referenced in PER. 110 Annexes 111 GDP per capita in constant USD, indexed to 100 for the year 1982 120 110 100 90 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Source: World Bank--SIMA database. ANNEX 1.2 Government expenditure by functional classification, in percentages Percent of recurrent expenditures 1990/91 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 General Administration 33.3 33.1 34.0 21.5 26.3 37.4 28.9 23.1 50.0 32.9 General Public Services 21.0 23.4 23.7 12.3 15.5 27.5 21.6 15.0 41.8 26.0 Defense 5.2 4.9 3.3 2.8 5.1 4.8 3.6 3.8 3.7 2.9 Public Order and safety 7.1 4.7 6.9 6.5 5.7 5.1 3.7 4.3 4.5 4.0 Social Services 19.9 17.7 32.5 30.6 35.2 45.5 27.4 35.3 27.6 27.2 Education 10.6 9.9 10.5 16.3 19.0 20.8 13.6 15.6 12.2 16.1 Health 6.4 6.0 8.3 6.2 9.2 8.5 6.2 8.7 8.7 9.5 Social Security and welfare services 2.9 0.0 13.8 8.1 7.0 12.7 6.2 9.7 5.6 0.9 Community and Social Development 1.9 0.0 0.0 0.0 3.4 1.3 1.3 1.0 0.8 Economic Services 13.4 23.1 11.9 12.1 7.7 9.3 5.5 6.9 5.7 8.7 Unallocable Services 33.5 26.1 27.8 37.4 35.4 33.0 33.2 29.2 16.7 31.2 Public debt service n.a. n.a. 15.1 27.0 28.1 14.2 20.6 19.9 22.5 21.4 Pensions and gratuities n.a. n.a. 3.5 4.0 5.6 5.7 5.3 9.3 5.2 4.2 Other n.a. n.a. 9.2 6.5 1.7 6.6 7.3 0.0 ­11.0 5.6 Adjustment ­6.2 ­1.6 ­4.6 ­25.2 5.0 5.4 0.0 0.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Percent of development expenditures 1990/91 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 General Administration 17.8 13.4 28.04 15.90 39.76 11.45 9.32 21.13 20.48 38.24 General Public Services 16.4 11.4 27.32 15.48 38.46 9.11 8.90 18.43 19.26 34.18 Defense 0 0.4 ­0.01 0.03 0.83 0.98 0.13 0.77 0.62 0.82 Public Order and safety 1.4 1.6 0.73 0.39 0.47 1.36 0.30 1.92 0.60 3.22 Social Services 21.2 10.6 50.93 53.95 28.72 35.80 53.74 35.65 30.27 27.21 Education 12.9 6.6 27.94 22.60 8.43 15.29 13.35 8.74 11.89 14.89 Health 8.3 4.1 0.45 16.21 3.63 5.89 16.45 5.78 1.14 11.93 Social Security and Welfare Services 0.07 0.06 0.71 0.22 1.53 1.01 0.00 0.00 Community and Social Development 22.47 15.08 15.95 14.40 22.41 20.12 17.24 0.39 Economic Services 59.7 27.2 15.12 24.21 16.23 32.20 25.17 39.54 49.25 34.56 Unallocable Services 1.3 48.8 0.00 6.00 15.29 2.56 3.80 0.00 0.00 0.00 Adjustment 5.91 ­0.06 ­0.01 17.98 7.97 3.67 0.00 0.00 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Source: Malawi Economic Report, various issues. Note: All percentages are based on `actuals', except for 2001/02, which is based on the `estimate'. 112 ANNEX 1.3 Education recurrent expenditure as a proportion of total government discretionary recurrent spending Yearly In current million MK 1997/98 1998/99 1999/00 2000/01 2001/02 2002/03 average Recurrent expenditure before debt amortization 11405 11083 13950 21856 32199 25036 Interest on debt 2081 3430 3251 5327 6571 5171 Pensions & gratuities 571 656 1524 1161 1499 1431 Discretionary recurrent spending 8753 6997 9175 15368 24129 18434 Source for educ. exp. data: Economic Report Education recurrent expenditure (including univ. and other subvented) 2092 1679 2556 2491 4300 5968 Educ. recurrent exp. as % of total discr. spending 23.9% 24.0% 27.9% 16.2% 17.8% 32.4% 23.7% Source for educ. exp. data: directly from the budget expenditures Education recurrent expenditure (including univ. and other subvented) 1773 1861 3010 3168 5467 Educ. recurrent exp. as % of total discr. spending 20.3% 26.6% 32.8% 20.6% 22.7% 24.6% Source: Discretionary spending and its components are from different years of the Economic Report. Note: Till 2000/01 all are `actuals', thereafter `revised' estimates. 113 ANNEX 2.1 Nursery school attendance Percentage who Mean Number of Years Background Characteristics Attended Nursery School Attended Nursery School Age (Years) 6­7 15.1 1.5 8­10 11.5 1.6 11­14 4.9 2.0 Residence Urban 38.8 1.9 Rural 5.6 1.4 Wealth Index Quintile Lowest 2.7 1.0 Second 4.2 1.7 Middle 4.4 1.2 Fourth 7.1 1.5 Highest 26.7 1.8 TOTAL 9.4 1.7 Source: Malawi EdDATA Survey 2002.4 114 ANNEX 2.2 Comparing gross enrollment ratios (GER) using different sources in 1987, 1998, 2000 and 2002 1987 1998 2000 2002 a) Population census based estimates5 Population ages 6­13 1,818,910 2,115,870 ­ ­ Enrollment in primary (standard 1­8) 1,002,343 2,165,5286 ­ ­ GER 55.1% 102.4% ­ ­ b) Survey based estimates Integrated Household Survey ­ GER7 ­ 120.1%8 ­ ­ DHS 2000 ­ GER ­ ­ 106.8% ­ Ed Data 2002 ­ GER ­ ­ ­ 114.6% c) School census based estimates Enrollment in primary (standard 1­8) 1,022,765 2,805,785 3,016,972 ­ Population estimates (i) Population census (ages 6­13) 1,818,910 2,115,870 ­ ­ (ii) NSO ­ adjusted pop estimates (ages 6­13) ­ 2,456,000 2,448,000 2,504,000 (iii) Unesco estimates (ages 6­13) 1,767,700 2,421,500 2,493,500 ­ (iv) UN­EFA (ages 6­13) 2,437,000 2,524,000 2,565,000 GER estimates (i) 56.3% 132.6% ­ ­ (ii) ­ 114.2% 123.2% ­ (iii) 57.9% 115.9% 121.0% ­ (iv) 115.1% 119.5% Sources: ­ Population Census data are from `1998 Population and Housing Census ­ Analytical Report' (www.nso.malawi.net), and for 2000 the data are provided by Jameson Ndawala (Chief of Demographics Department ­ NSO) ­ School census data: Basic Education Statistics of the corresponding years. ­ IHS 1998: authors' calculations using dataset. ­ DHS 2000: "Malawi Demographics and Household Survey ­ Final Report. August 2001" ­ Ed Data 2002: `Malawi DHS EdData Survey: Preliminary report' ­ NSO ­ adjusted population estimates: NSO website under `Data/Malawi Population Projections'. ­ UNESCO population estimates: World Bank SIMA database connecting to UNESCO Education Statistics. ­ UN­EFA population estimates 115 ANNEX 2.3 Enrollment by age comparing household surveys and school census F igure 2.3 shows enrollment rates in pri- estimate at 15 percent.11 This probably arises mary by age group using three different because parents claim their children are older data sources.9 The results after combin- than they actually are so they may avoid the ing school census data and population problems associated with official policy and data from both sources are different from the continue to attend primary school. These dis- results suggested in the household surveys.10 crepancies in the recorded ages of children is This implies that the number of over-aged chil- further complicated by the fact that the chil- dren is much higher in the primary education dren themselves, possibly as a result of peer- system than what the official school census sta- pressure, may indicate an age closer to the tistics records. The enrollment rates of over `normal' age for that standard--or at least 100 percent, found for the age group 7­10 within the most common age range indicated years olds, are actually largely because of the by their classmates. The first table below is the many under-aged children being registered. As equivalent of Figure 2.3, but for year 1998. far as the under-age children in primary are The second table shows the similarity between concerned, the survey results show that close the results from the two completely independ- to 30 percent of the population, aged 5, are ent household surveys in 1998 and 2000. This already enrolled (representing close to 100,000 strengthens the case for the reliability of the children) whereas school census data puts this results from these household surveys. 116 Annexes 117 Enrollment rate in primary by age, different sources, 1998 140.00% 120.00% 100.00% 80.00% 60.00% 40.00% 20.00% 0.00% 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 "IHS 1998" ag e1998Min Educ enrollment and NSO original pop cent 1998 Min Educ enrollm and Unesco pop est Enrollment rate in primary by age, different sources, 1998 and 2000 140.00% 120.00% 100.00% 80.00% 60.00% 40.00% 20.00% 0.00% 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0 2 1 2 2 2 3 2 4 2 5 agg e a "IHS 1998" DHS 2000 Sources: Authors' calculation using data from the IHS 1998 survey, the DHS 2000 Survey, and from the National Statistics Office and UNESCO for population estimates. ANNEX 2.4 Analytical review of entry rate to standard 1 O bviously, in an expanding system, ever, in the following section, possible explana- one moving towards universal cov- tions for the discrepancies in recorded repeti- erage, the issue of children being tion and population estimates will be reviewed. over-aged due to late entry, or repe- The table below indicates that repetition tition, is commonplace. This was the case in rates recorded in the DHS 2000 and EdData Malawi. Primary fees were abolished in 1994 2002 were much higher than the repetition and the number of new entrants (i.e. those who rates recorded by the number of repeaters ver- had not yet entered school) increased signifi- sus total enrollment in the Basic Education Sta- cantly--in fact, it approached universal cover- tistics. For primary, as a whole, there is a dif- age relative to the population group. This ference of at least 10 percentage points, and for increase was short-term (the `big bang' effect) standard a difference of approximately 20 to and in subsequent years an increasing propor- 25 percentage points. tion of children of appropriate age entered the One explanation for discrepancies in record- system. Therefore, it should not be possible to ed levels of repetition may be connected to the continually have more "new" entrants than definition of `repeater'. In the household sur- children in the population group of normal age veys, a repeater is either a child that completed of entry by the year 2003--almost 10 years a school year and then repeated the same stan- after the introduction of FPE.12 dard the following year, but it can also be a It follows that the 2.29 rate of entry, noted child that interrupted his/her schooling during in 1998, is not logically possible13; it suggests the school year but subsequently returned to possible confusion in: (i) an over-count of total the same standard in the following school year. enrollment; (ii) the definition of new entrants The latter occurrence may be quite common, into Standard 1 versus repeaters; and/or (iii) an particularly in Standard 1, as children may under-count of school-age population. Since start but later be considered too immature for enrollment data is obtained through head Standard 1, drop out but return the following teachers (both for the school census or in sep- year. Although head teachers were not provid- arate studies14), and no other source is avail- ed with specific instructions, it is acknowl- able, clarification of the level of error related to edged that in the school census, only children over-count of enrollment is not possible. How- who completed the year in school and are not 118 Annexes 119 Comparing repetition rates using different sources, 1998, 2000 and 2002. 1998 2000 2002 School census based estimates Primary total 14.5% 15.6% ­ Standard 1 18.2% 19.2% ­ Household surveys DHS 2000 Primary total ­ 29%* ­ Standard 1 ­ 45.3% ­ EdData 2002 Primary total 25.8% Standard 1 ­ ­ 40.6% Sources: Authors' calculations, using Basic Education Statistics 1998 and 2000. "Malawi Demographics and Household Survey ­ Final Report. August 2001." Ed Data 2002: `Malawi DHS EdData Survey: Preliminary report'. * Authors' calculation using DHS 2000 dataset. considered sufficiently mature to progress to be underestimating the total number of chil- the next standard are recorded as repeaters dren of the corresponding age. Using a high- while children who left during the year and end population estimate for Malawi from UN- return to the same grade the next year are reg- EFA (United Nations-Education For All), the istered as new entrants again. cohort rate of entry is further lowered to 1.33 The definition of repeaters may vary, but in (see Table below, (c)). international comparisons the accepted defini- With a cohort rate of entry to Standard 1 of tion is a child who takes up the teachers time 1.33 for 1998 (and of 1.39 for 2000), there is and absorbs resources. Furthermore, to meas- still significant unexplained variance. Howev- ure efficiency in resource usage, it is logical to er, given the lack of alternative data, in partic- separate the children returning the subsequent ular in enrollment numbers, the cohort rate of year to the same grade from the real new entry of 1.33 (for 1998) and 1.39 (for 2000) entrants. Otherwise, if PTR is based on enroll- will be used as the most credible estimate. ment numbers at the beginning of the year, the Population in the 1998 population census public expenditure will be incurred twice and, may have undercounted the actual population as a result, more teachers will be assigned on at the entry age to primary. Three scenarios the basis of the children who begin the school were compared: year, but who subsequently drop out. The data anomalies between household surveys and (i) original 1998 population census which school census data might also be explained by represents an annual growth rate for a separate factor: teachers have no system to the 6­13 age group of 1.4 percent17 establish whether or not a child has been pre- since the previous population census viously enrolled in school. In such a situation, from 1987; children may be incorrectly listed as enrolled in a grade for the first time.15 (ii) high-end population estimate (UN- If the repetition rate from the household sur- EFA) which represents an annual veys is applied to total enrollment in Standard growth rate for the 6­13 age group of 1, as found by the school census statistics (see 2.7 percent18 between 1987 and 1998; Table below, point (b)), then the cohort rate of entry reduces from 2.29 to 1.60 (160 per- (iii) using the same annual growth rate of cent).16 The 1998 population census could also 4.9 percent between 1987 and 1998, as 120 Cost, Financing and School Effectiveness of Education in Malawi Entry rate to standard 1, different data sources Enrollment Population Cohort rate of 1998 situation standard 1 New entrants age of entry entry (a) MoEST enrollment and 1998 population census data 751,002 638,865 278,438 2.29 (b) MoEST enrollment, but survey average repetition rate (42.9% standard 1), and 1998 population census data 751,002 441,087 278,438 1.60 (c) MoEST enrollment, but survey average repetition rate (42.9% standard 1), and UN-EFA population 751,002 441,087 330,900 1.33 Source: Authors' calculations. was recorded between 1977 and at the very high end given the historic popula- 1987.19 tion data for Malawi. According to the NSO, two specific factors contributed to the more Under scenario (iii) the estimated population than average population growth from 1977 of age 6­13 would be 3,118,14220 for 1998 or until 1987 and to lower population growth in 3,439,19521 for 2000. With an enrollment in the period from 1987 until 1998. First there primary of 2,805,785 in 1998, and 3,016,972 was the influx of Mozambican refugees during in 2000 according to the school census, this the civil war and the then returning in the peri- would give a GER of 90.0 percent in 1998 and od thereafter. For instance, annual population of 87.7 percent in 2000, which is significantly growth rates at district level show that popula- below the GER found by the household sur- tion in the districts that lie along the border veys, and therefore not very plausible. with Mozambique grew with the slowest Also, as can be seen from the table below, rates.22 This suggests that the repatriation of the annual growth between 1977 and 1987 is Mozambican war refugees who were concen- trated in these districts during the 1987­1998 Total Populations: 1901­1998 Censuses intercensal period had an impact on the Avg. Annual growth of the population. The second factor is Total Intercensal the impact on population of the AIDS crisis, Census Year Population Growth (%) slowing the growth rate. 1901 737,153* ­ Given the fact that UN-EFA estimates take 1911 970,430* 2.8 the previously mentioned factors into account, 1921 1,201,983* 2.2 additionally to using advanced demographic 1926 1,263,291 1.5 modeling, based upon a longer historic trend, 1931 1,573,454 4.4 this will be considered the most plausible pop- 1945 2,049,914 2.2 ulation estimate and will be used. Using the 1966 4,039,583 3.3 1977 5,547,460 2.9 population from UN-EFA and the household 1987 7,988,507 3.7 survey repetition rate, the cohort rate of entry 1998 9,933,868 1.9 would be 1.3323 in 1998 and 1.395 in 2000.24 Source: NSO ANNEX 2.5 Repetition rates in the first four grades of primary, comparing countries participating in Monitoring Learning Achievement study, 1999 Years spent in grade 1 Years spent in grade 2 Years spent in grade 3 Years spent in grade 4 1 2 3 >= 4 1 2 3 >= 4 1 2 3 >= 4 1 2 3 >= 4 Botswana 88.5 10.8 .7 .1 93.6 5.7 .6 .1 94.0 4.9 1.0 .1 85.3 13.4 .7 .5 Madagascar 78.5 21.5 0 0 79.2 20.8 0. 0 85 15 0 0 n.a. n.a. n.a. n.a. Malawi 74.3 21.6 3.5 .6 76.5 21.2 1.9 .4 74.9 22.0 2.7 .4 85.3 13.5 .6 .6 Mali 92.7 7.3 0 0 94.7 5.3 0 0 92.8 6.9 0 0 88.1 11.9 0 0 Mauritius 99.9 .1 0 0 99.7 .3 0 0 99.3 .5 .2 .1 99.1 .6 .1 .1 Morocco 88.6 11.4 0 0 90.7 9.2 .1 0 88.1 11.5 .4 0 91.7 8.3 0 0 Niger 97.8 2.2 0 0 97.2 2.8 0 0 96.5 3.4 .1 0 94.7 5.3 0 0 Senegal 88.6 5.5 .1 5.8 86.7 6.8 .2 0 84.3 7.9 .7 7.1 79.8 12.4 .4 7.4 Tunisia 87.6 12.2 .1 0 85.3 14.5 .1 0 82.0 17.4 .5 .1 85.1 14.5 .3 0 Uganda 83.1 14.0 2.1 .8 83.1 15.1 1.5 .2 85.1 11.4 2.7 .8 83.2 13.9 .7 2.1 Zambia 93.2 6.6 .2 0 92.9 7.0 .1 .1 93.2 5.7 1.1 0 88.5 10.0 .3 1.2 Source: MLA 1999. With Africa for Africa: Towards Quality Education for All. 121 ANNEX 2.6 Factors in primary school pupil dropout (from Ed Data 2002) Percentage of de jure children age 5­14, who had dropped out of school, 2002. Cost-related factors Child factors School factors Long- School No Background Monetary Labour Repeat / Enough term School travel Poor places in No characteristic cost needed exam fail school illness Disabled too far unsafe quality secondary jobs Percentage 24.1 30.4 13.2 44.5 11.7 5.4 9.6 3.5 8.5 1.1 2.9 Note: More than one response possible. Questions were asked about a total of 284 children who had dropped out of school. First, the parent/ guardian was asked whether a child had dropped out because (s)he was physically or mentally disabled, and if the answer was yes, no further questions were asked about reasons for leaving school. Next, the respondent was asked whether a child had dropped out because (s)he had been very sick for 3 months or longer, and if the answer was yes, no additional questions were asked. There were 16 children who had dropped out because of a disability and 33 who had left because of long illness, so for the remainder of the questions, the sample size was only 237. 122 ANNEX 2.7 Survival rate in primary, different methods and moments in time I n the table below, different methods In this context, it is important to highlight and years to calculate the survival rates the likely impact of the FPE policy, since after for primary education are used.25 They its introduction enrollment increased by more indicate that there is a decline in the than one million pupils. Here, both the cross- number of pupils surviving through to the end sectional method for recent years and the full- of primary in the years prior to FPE (28­30 time series method are likely to be distorted percent) relative to recent years (24­26 per- by this major enrollment surge that has yet to cent) when using a simplified cross sectional be reflected in analysis across all grades. The approach. Based on the results from the quasi- full-time series perspective indicates a figure time series survival estimates26, a similar, of 44.9 percent from 1989/90 to 1999/0028 albeit more stark, picture emerges: it indicates and this, in effect, reflects the position prior that a student's chances of "survival" through- to FPE. Those in Standard 8 in 1999/00 were out primary went down from 41.5 percent in not very affected yet by the FPE policy, 1989/91 to 22.9 percent in 1998/00. because the huge increase in the number of Years 89/90 90/91 98/99 99/00 Grade Stand 1 Levels Stand 2 Stand 3 ... Stand 8 27.9% 30.3% !----cross-sectionalmethod ----"26.2% 24.4% 41.5% !--------quasi-timeseriesmethod--------"22.9% Source: School Survey Data in EMIS 2000 for enrollment and repetition rates,27 NSO population estimates. 123 124 Cost, Financing and School Effectiveness of Education in Malawi children who entered in 1994 as a result of distorted by the FPE policy29 It therefore FPE were primarily in Standard 1. The quasi- more accurately reflects the situation a few time series calculations for 1998/00, using years before, and after, the FPE policy. data from only two consecutive years, is not ANNEX 2.8 Survival rates for primary and secondary using different data sources The pattern of access and survival (cross-sectional method), 1999/0030 Intake rate to the grade31 Survival Rates (%) Student in grade 1 129.9 100.0 Student in grade 4 93.1 71.7 Student in grade 5 82.9 63.9 Student in grade 8 33.0 25.4 Student in form 1 29.5 22.7 Student in form 2 30.5 23.5 Student in form 3 23.7 18.3 Student in form 4 17.1 13.2 Source: School Survey Data: in EMIS and Basic Education Stats, 2000. DHS 2000 for repetition rates, UN-EFA population. The non-repeaters are calculated using the repetition rate from DHS 2000. The pattern of access and survival (cross-sectional method), 1999/0032 Intake rate to the grade Survival Rates (%) Students in grade 1 217.3 100.0 Students in grade 4 108.0 49.7 Students in grade 5 97.7 44.9 Students in grade 8 53.1 24.4 Students in form 1 33.6 15.5 Students in form 2 35.1 16.1 Students in form 3 26.2 12.1 Students in form 4 22.7 10.4 Source: School Survey Data: in EMIS and Basic Education Stats, 2000, also for repetition rates. NSO population estimates, version from Oct 2002. 125 ANNEX 2.9 Proxy primary completion rates for standards 4, 6 and 8 and international comparison Grade 4 proxy 100.0% 80.0% 60.0% 40.0% 20.0% 0.0% 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year Grade 6 proxy 80.0% 60.0% 40.0% 20.0% 0.0% 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year 126 Annexes 127 Grade 8 proxy 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year Source: Basic Education Stats, NSO population; Highest level is standard NSO population. Lowest level is using NSO population plus 15%. (UN-EFA population is not available this far in the past). Primary GER and completion rates for selected African countries, 1997­2000 Country Year GER Completion Rate Kenya 1999 91.3 58 Lesotho 2000 102.7 77 Malawi 1999 117 50 Mozambique* 1998 78.5 36 Nigeria 2000 85.1 67 Tanzania 1999 66.4 59 Uganda 2000 102 82 Zambia 1998 84.9 80 Zimbabwe 1997 112 103 Average for 33 African countries 76.6 45.1 Source: Bruns et al. 2002. Note: The numbers are related to the entire primary cycle ­ or to the first 6 grades, whenever primary is longer than 6 grades. ANNEX 2.10 Highest grade of instruction offered in primary, and walking time and distance to the nearest school Distribution of primary schools by highest grade of instruction offered, 1999/00 Number of schools Non-repeaters (new entrants) in Standard 1 Highest grade of Long Recently Long Cumulative Recently Cumulative instruction offered established established established share established share 1 2 2 94 0.0% 66 0.0% 2 31 85 3,435 0.5% 7,926 1.2% 3 33 121 3,704 1.1% 10,489 2.8% 4 80 261 10,726 2.8% 23,706 6.5% 5 95 310 10,655 4.4% 31,350 11.3% 6 112 273 13,110 6.4% 28,362 15.7% 7 97 167 13,617 8.5% 17,992 18.4% Total 4,392 650,82933 Source: Authors' calculation using EMIS 2000 dataset. Walking time and distance to the nearest primary school, in percentage per category, de jure children age 6­14 Minutes to nearest Mean walking primary school 0­15 16­30 31­45 45­60 60+ time in minutes Urban 64% 28% 2% 6% 0% 19 Rural 33% 33% 6% 18% 10% 41 Total 36% 32% 6% 17% 9% 39 Kilometers to nearest prim. school <1 1­2 3­4 5­6 >6 Mean distance Urban 58% 33% 3% 4% 1% 0.9 Rural 37% 38% 14% 6% 4% 1.9 Total 39% 37% 13% 5% 4% 1.8 Source: Ed Data 2002. 128 ANNEX 2.11 Description of secondary school types (i) Government Day Schools and Government by School Committees but in receipt of Boarding Schools: Operated and funded some government funding to support directly by Government through Education teacher and staff salaries and assist day-to- Division Offices. day administration requirements. (ii) Grant-Aided Schools: Operated by church (iv) Private Schools: Owned by private individ- organizations with Government contribu- uals, or non-governmental organizations, tion to teachers' salaries and monthly in receipt of no government grants and grants for day-to-day running costs. Stu- charge economic fees. dent tuition and boarding fees are cumula- tively much higher than in government (v) Malawi College of Distance Education schools. (MCDE) and Night Schools: Cater for full- time workers or adults interested in (i) and (ii) together are also called Convention- improving their academic qualifications. In al Secondary Schools (CSS) addition, youths who fail to secure a place in other secondary institutions can opt to (iii)Community Day Secondary Schools join the MCDE. (CDSSs): Mostly established and financed 129 ANNEX 2.12 Technical and vocational training description T raditional public technical and voca- trades. The full-time institutional training pro- tional training in Malawi consist grams take place in some of the technical col- mainly of two types: an apprenticeship leges and they provide courses to cover the training system; or full-time institu- same range of technical trades. tional training. They have been limited in scope; A range of courses are available at the pub- they only enrolled 1,500 students in the com- lic technical colleges. These include: bricklay- bined programs up to the end of the last decade. ing, carpentry & joinery, painting and decorat- Between the years 2000­2003, the Ministry ing, plumbing, wood work machining (as a of Labor and Vocational Training indicated component of `construction'), bookkeeping that a total of 1,723 male and 883 female stu- and accounts, secretarial, ICT, auto electrics, dents have graduated from the public Techni- motor vehicle mechanics, vehicle body repair cal Colleges, as follows: 936 in `construction'; and refinishing under `automobile engineer- 780 in `commercial'; 540 in `automobile engi- ing'; and electrics/electronics, general fitting, neering' and 350 in the field of `engineering'. welding and fabrication.34 So despite the large increase during the last Since 1999, a new TEVET Act has become decade, the absolute numbers are still small effective and it also established an independent compared to enrollments in secondary educa- secretariat called Technical, Entrepreneurial, tion and primary education. Vocational Education and Training Authority The apprenticeship combines on-the-job (TEVETA). The main aim is to link Technical training by industry with institutional training & Vocational Training more closely with the by seven Technical colleges--all managed by private sector by promoting a demand driven, the Ministry of Labour and Vocational Train- modular, technical and vocational training sys- ing. It is a four-year training scheme and is tem; monitoring gaps between supply and offered in a significant number of technical demand; supporting the application of appro- Enrollment in public technical & vocational colleges Year 1991/92 1993/94 1995/96 1998 1999 2000 (estimate) Enrollment 840 1007 1054 1308 1290 1504 Source: Basic Education Statistics. 130 Annexes 131 priate technologies; and promoting managerial In addition to assuming responsibility for and business skills and a spirit of entrepre- the traditional technical colleges, TEVETA also neurial culture in respect to both wage and supports staff training programs for private self-employed personnel. sector employees and it reimburses part of the Funding for this new approach is innovative in training costs incurred by the employer for the Malawian context. Since the fees paid by approved training. This scheme benefited 259 trainees do not cover the economic cost of the employees from 13 companies in 18 courses training provided, additional funding is generated between July 2002 and June 2003. TEVETA in two ways. Firstly, the training provider institu- conducts a number of other training programs tions arrange various parallel income-generating which use local masters, for instance, in car- activities and secondly, a TEVET fund has been pentry, cane furniture production, etc. and it is set up into which private sector and government intended to increase the range of available employers pay a 1 percent payroll-levy. It is also course still further. On completion of all cours- intended that the funding will be further topped- es official certificates are issued. up through additional available HIPC debt relief The box below provides a typical example and cooperating partner support. of the ad-hoc training supported by TEVETA. Example of training scheme supported by TEVETA In an effort to promote skills training for unemployed people, four partners are working together. First there is the German donor KfW, which supports the construction activities of income-generating urban infrastructure, such as craft centers, markets, bus stations, commer- cial areas, etc. They contract construction companies and make training activities a part of their tender papers. The selected construction company, as another partner, is responsible for providing skilled laborers who will implement the training. The company contracts trainees, for three months, who are entitled to receive a wage lower than the skilled laborers. The town assemblies of the smaller urban centers where the infrastructure works are taking place form a third party in the initiative. They are responsible for the selection of young peo- ple to serve as trainees, for example, school dropouts or unemployed. TEVETA has the overall responsibility for ensuring a consistent quality assurance system is in place. The training staff and the site agents of the construction companies have participat- ed in a seminar on how to conduct the training and assessment and TEVETA developed the assessment documentation. In order to ensure greater public confidence and confirmation of the internal assessment results, external assessors from industry, and from private and public training institutions, have been trained and contracted by TEVETA. The assessment results are forwarded to TEVETA to validate and certify the candidates. The training participants can also gain credits and build up a national, recognized qualification if they wish to extend the train- ing with other providers. TEVETA also attempts to implement similar training arrangements with other industrial partners or donor agencies. It is currently in discussion with DFID on how to include this ini- tiative in their school construction program and with the World Bank MASAF program which funds construction and rehabilitation projects with community participation. Source: "Promotion of non-formal Technical Entrepreneurial Vocational Education Training components in Malawi" by TEVETA. ANNEX 3.1 Recurrent and development expenditures at real prices REAL 1994 MK 1993­94 1994­95 1995­96 1996­97 1997­98 1998­99 1999­00 2000­01 Total educ sect 660,732,755 524,632,658 593,277,761 560,237,053 610,623,791 547,507,087 576,552,527 183,193,436 Recurrent 367,145,786 337,897,231 414,627,724 397,063,144 506,864,170 443,771,990 483,224,170 98,377,295 Development 293,586,969 186,735,427 178,650,037 163,173,909 103,759,621 103,735,097 93,328,357 84,816,141 MOEST 550,477,362 418,203,500 490,731,479 465,321,484 503,978,060 457,758,775 468,978,521 373,051,496 Recurrent 268,955,551 250,238,212 323,570,621 313,547,606 400,765,483 354,023,679 375,650,164 302,468,863 PE 199,320,515 166,302,180 240,546,637 223,742,530 297,765,647 262,137,348 248,828,100 n.a. other 69,635,035 83,936,032 83,023,985 89,805,076 102,999,836 91,886,331 126,822,064 n.a. Development 281,521,812 167,965,288 167,160,857 151,773,878 103,212,577 103,735,097 93,328,357 70,564,633 Subvented org 110,255,392 106,429,157 102,546,283 94,915,569 106,645,731 89,748,312 107,574,006 112,628,803 Recurrent 98,190,235 87,659,018 91,057,103 83,515,538 106,098,687 89,748,312 107,574,006 98,377,295 Development 12,065,157 18,770,139 11,489,180 11,400,031 547,044 n.a. n.a. 14,251,508 Source: PER for data until 1998/99. For 1999/00, primarily from Basic Education Statistics. For 2000/01, from `Consolidated annual appropriation accounts'. 132 ANNEX 3.2 Recurrent, development and total expenditure by sub-sector over time, in real 1994 MK and in percentage of total REAL TERMS (1994 '000 MK) 1993­94 1994­95 1995­96 1996­97 1997­98 1998­99 1999­00 2000­01 2001/02 2002/03 Recurrent 367450 337897 406019 397048 506864 394942 484428 400864 577031 628947 Admin 29956 36416 19412 20517 29475 17549 27676 56203 29169 20547 Primary 178662 165638 240028 236883 300739 240183 275846 193023 307109 381581 Secondary (including MCDE) 44066 32137 38322 38679 45409 40835 67264 46031 90280 66514 Teacher ed. 12849 12173 13238 13796 16676 6627 6068 7230 37597 34992 Tech & voc. 3726 3873 3961 3658 8468 n.a. n.a n.a n.a. n.a. Tertiary 70202 66465 70792 66436 90168 74953 90397 79932 83956 107667 Other subvented orgs. 27988 21195 20265 17080 15931 14795 17177 18445 28921 17645 Development 293587 186735 178650 163174 103760 103735 93328 123059 157434 179126 Admin. 25054 14453 15129 3181 3298 8147 4340 2470 13511 13887 Primary 95312 77283 82640 91323 86607 85232 29292 43979 35633 9401 Secondary (including MCDE) 89322 65705 44608 26780 6782 9477 56614 71807 104190 144872 Teacher ed. 71834 10524 24784 30490 6526 880 3082 4803 4101 10966 Tertiary 4246 12531 8570 6929 494 n.a. n.a. n.a. n.a. n.a. Other subvented orgs. 7819 6239 2919 4471 53 n.a. n.a. n.a. n.a. n.a. Recurrent+development 657311 520759 580708 556564 602156 408929 470183 425546 621588 682760 Admin. 55011 50869 34541 23698 32773 25695 32016 58673 42679 34434 Primary 273973 242922 322668 328207 387346 325415 305138 237002 342742 390982 Secondary (including MCDE) 133388 97842 82930 65459 52191 50311 123878 117837 194469 211387 Teacher ed. 84684 22697 38023 44286 23201 7507 9151 12034 41697 45958 Tertiary 74449 78996 79362 73365 90662 n.a. n.a. n.a. n.a. n.a. Other subvented orgs. 35807 27433 23184 21550 15984 n.a. n.a. n.a. n.a. n.a. (continued on next page) 133 134 Cost, Financing and School Effectiveness of Education in Malawi 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 2002/03 Recurrent expenditure 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Admin 8% 11% 5% 5% 6% 4% 6% 14% 5% 3% Primary 49% 49% 59% 60% 59% 61% 57% 48% 53% 61% Secondary (including MCDE) 12% 10% 9% 10% 9% 10% 14% 11% 16% 11% Teacher ed. 3% 4% 3% 3% 3% 2% 1% 2% 7% 6% Tech & voc. 1% 1% 1% 1% 2% n.a. n.a. n.a. n.a. n.a. Tertiary 19% 20% 17% 17% 18% 19% 19% 20% 15% 17% Other subvented orgs 8% 6% 5% 4% 3% 4% 4% 5% 5% 3% Development expenditure (on-budget only) 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Admin. 9% 8% 8% 2% 3% 8% 5% 2% 9% 8% Primary 32% 41% 46% 56% 83% 82% 31% 36% 23% 5% Secondary (including MCDE) 30% 35% 25% 16% 7% 9% 61% 58% 66% 81% Teacher ed. 24% 6% 14% 19% 6% 1% 3% 4% 3% 6% Tertiary 1% 7% 5% 4% 0% n.a. n.a. n.a. n.a. n.a. Other subvented orgs. 3% 3% 2% 3% 0% n.a. n.a. n.a. n.a. n.a. Total expenditure 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Admin 8% 10% 6% 4% 5% 6% 7% 14% 7% 5% Primary 42% 47% 56% 59% 64% 80% 65% 56% 55% 57% Secondary (including MCDE) 20% 19% 14% 12% 9% 12% 26% 28% 31% 31% Teacher ed. 13% 4% 7% 8% 4% 2% 2% 3% 7% 7% Tertiary 11% 15% 14% 13% 15% n.a. n.a. n.a. n.a. n.a. Other subvented orgs 5% 5% 4% 4% 3% n.a. n.a. n.a. n.a. n.a. Sources: PER for data until 1998/99. For 1999/00, primarily from Basic Education Statistics. For 2000/01, from `Consolidated annual appropriation accounts '. For 2001/02 and 2003/03 the data are `approved estimates' from FOM Draft Estimates of Expenditure and Financial Statements. Note: it is important to remember that the percentages from 1998/99 onwards for development expenditure and total expenditure are not completely comparable to the years before, seen no data is available on development expenditure for sub-vented organizations. ANNEX 3.3 International comparisons of unit costs Public spending as a multiple Public spending as a multiple of of per capita GNP spending in primary education Number of Region35 countries Primary Secondary Higher Secondary Higher Francophone Africa 15 0.15 0.49 5.6 3.3 37.3 Anglophone Africa 9 0.10 0.66 6.3 6.6 63.0 Latin America/Caribbean 10 0.07 0.11 0.7 1.6 10.0 Asia 8 0.08 0.19 0.9 2.4 11.3 Middle East/North Africa 6 0.11 0.30 0.9 2.7 8.2 Malawi 0.04 0.21 14.89 5 348 Mozambique 0.07 0.40 6.41 Kenya 0.12 0.34 4.0 Zambia 0.04 0.37 2.3 Madagascar 0.06 0.20 2.6 Source: Mingat and Suchaut (2000) for the regional values, with data from 1993. World Bank Sector Studies for individual countries, with data from 1998. For Malawi authors' calculations, using 2000/01 actual expenditure data. Note: For higher education, refers to per-student spending in traditional public institutions. 135 ANNEX 3.4 Nominal per pupil expenditure by sub-sector (current USD) 93/94 94/95 95/96 96/97 97/98 98/99 99/00 Primary 10 5 13 15 12 9 14 Secondary CSS only 116 80 132 148 125 71 118 Secondary as a whole 61 33 49 46 30 19 29 Teacher ed. 366 363 687 317 119 45 n.a. Technical 382 330 576 517 740 n.a. n.a. University 2,900 2,434 2,998 3,271 4,571 2,617 2,650 Source: PER and authors' calculations. For university 93/94 ­ 96/97 is from PER, thereafter authors' calculations using official enrollment, exchange rate, and actual expenditure data received from UNIMA (thus this does not include Mzuzu University). 136 ANNEX 3.5 Decomposition of 1999/00 and 2000/01 education sector development expenditure by source 1999/00 Education Sector Development Expenditure % of % of Develop- Part I: dev Part II: dev ment % of Bilateral % of % of Program Loans bud GoM bud budget total (Estimate) total Total total Total 516,234,000 89% 65,127,000 11% 581,361,000 35% 1,065,360,000 65% 1,646,721,000 100% Primary 146,814,000 80% 35,650,000 20% 182,464,000 26% 506,620,000 74% 689,084,000 42% Secondary 324,023,000 92% 28,636,000 8% 352,659,000 60% 238,000,000 40% 590,659,000 36% Teacher. ed 19,096,000 99% 105,000 1% 19,201,000 6% 313,740,000 94% 332,941,000 20% Admin. 26,301,000 97% 736,000 3% 27,037,000 79% 7,000,000 21% 34,037,000 2% 2000/01 Education Sector Development Expenditure Total 492,902,000 89% 59,672,000 11% 552,574,000 30% 1,295,235,000 70% 1,847,809,000 100% Primary 302,617,000 88% 41,772,000 12% 344,389,000 33% 691,451,000 67% 1,035,840,000 56% Secondary 134,409,000 89% 16,818,000 11% 151,227,000 43% 204,490,000 57% 355,717,000 19% T. Ed 37,615,000 100% 0 0% 37,615,000 14% 234,016,000 86% 271,631,000 15% Tertiary 0 n.a. 0 n.a. 0 n.a. n.a. n.a. n.a. n.a. Other ed. orgs. 0 Admin. 18,261,000 94% 1,082,000 6% 19,343,000 10% 165,278,000 90% 184,621,000 10% Source: PER for 1999/00 and for 2000/01 off-budget estimates. For 2000/01, from `Consolidated annual appropriation accounts'. 137 ANNEX 3.6 Off-budget bilateral grant support for education Disbursed 2002 Project Years Currency Cost (mill.) to date disbursement Primary Education USAID Quality and Efficiency 2001­06 US$ 22.0 10.5 UNICEF School health and hygiene 2002­06 US$ 2.4 1.3 KFW Physical infrastructure 1997­2004 Euro 13.8 12.0 2.9 DFID Sector support program 2001­07 Sterling 61.0 1.7 Netherlands Sector support program 2001­07 Sterling 17.7 2.2 WFP mid-day meals 2002­06 US$ 7.7 Primary and Secondary GTZ Teacher education/curriculum 2002­05 US$ 4.7 0.6 CIDA Textbooks 1999­04 US$ 15.0 CIDA Trained teachers for CDSS 1999­04 US$ 5.0 138 ANNEX 3.7 Examples of potential problems from extra-budgetary funding and uncoordinated donor financing T he externally financed textbook initia- This should include a more detailed analysis of tive costs US$4.80 per enrolled pri- sustainable textbook provision, including mary student36; this figure can be com- examination of the book chain and measures pared to the current US$2.70 per to improve writing, publishing, printing and student provision for teaching learning materi- distribution. The second example relates to als. However, inclusion of a similar level of construction costs and illustrates how uncoor- textbook provision would raise expenditure on dinated donor funding can undercut efforts to teaching-learning materials by 1.8 times and find cost-effective strategies that might eventu- result in substantial increases in the recurrent ally become domestically sustainable. As the budget. As such, the financing of textbooks table below shows, the cost of constructing one through extra-budgetary grants, artificially classroom, a single pit latrine and a borehole, lowers the cost to the government--and to the varies from one donor project to another. For recurrent budget--but it also results in a fail- example, the Malawi Social Action Fund ure to explore more cost-effective, and sustain- (MASAF) uses a community based approach able, strategies to provide textbooks. Since a and is about 30­60 percent cheaper--although regular supply of textbooks is obviously this lower cost may hide (i) in-kind contribu- important if the quality goals, outlined in the tion by communities, (ii) variance in standards government policy documents, is paramount. and (iii) likely lifespan. Cost of construction in different projects Project Donor Cost $'000 MASAF cost $'000 Percentage Difference Classroom Block DFID 15 9.5 36.7 Classroom Block DANIDA 22.5 9.5 57.7 Classroom GOM Building 15 9.5 36.7 Single Pit Latrine DFID 1.4 0.4 71.4 Single Pit Latrine GOM Building 1.5 0.7 53.3 Borehole DFID / DANIDA 4.4 3.2 27.3 Water ­ Borehole GOM Building 16.2 8 50.6 Source: January 2003, "Review of cost Effectiveness and Design Standards." EMC Jatula Associates, Malawi (in "PAD ­ Third Social Action Fund." Report 25562­MW. April 2003) 139 ANNEX 3.8 Household monetary contribution to primary education as a percentage of household income (or expenditure) In percentages Uganda Malawi Zambia Kenya** Urban 5.5 7.7 5.0 Rural 1.8 8.2 1.7 Lowest 1.5 5.1 2.0 3.0 (15.6) Second 1.5 2.2 1.0 4.0 (21.0) Third 1.1 1.2 0.9 4.5 (20.2) Fourth 1.2 1.2 1.2 5.0 (16.3) Highest 2.7 1.0 3.1 5.9 (12.1) Total 4.2 4.8 (14.9) Sources: Authors' calculations using Malawi, Uganda, Zambia DHS EdData Survey 2001 for household (HH) education spending. For Kenya results are from WB education sector review draft "Education and Training in Kenya" from 2003 using DHS 1998 input data. ** In the case of Kenya this represents the results for both primary and secondary education and not just for primary. The number in parentheses indicates education expenditure as a % of only non-food expenditure. 140 ANNEX 4.1 Index of efficiency for primary Promotion rate Repetition rate Dropout rate Yearly dropout Cumulative dropout Survivors 1 55.3% 45.3% ­0.6%* ­0.6 ­0.6 100.0 2 63.7% 24.2% 12.1% 12.1 11.6 88.4 3 64.1% 27.2% 8.7% 7.7 19.3 80.7 4 75.4% 17.2% 7.4% 6.0 25.3 74.7 5 64.8% 15.2% 20.0% 15.0 40.2 59.8 6 73.0% 12.5% 14.5% 8.6 48.9 51.1 7 58.0% 11.4% 30.6% 15.6 64.5 35.5 8 53.3% 38.6% 8.1% 2.9 67.4 32.6 Sources: EMIS 2000 for enrollment, DHS 2000 for repetition rates, UN­EFA for population · Promotion rate = non repeaters in grade X+1 in year 2 divided by total enrollment in grade X in year 1 · Repetition rate (as used in DHS2000) = percentage of students in a given grade who are repeating that grade · Dropout rate = 1 ­ Promotion rate ­ Repetition rate · Cumulative dropout year 2 = Cumulative dropout year 1 + Yearly dropout · Yearly dropout in year 2 = dropout rate in year 2 * (100 ­ Cumulative dropout year 1) · Survivors year 2 = 100 ­ Cumulative dropout year 1 (Calculating the yearly dropout is a step that is required because the absolute numbers of dropouts as a result of the same dropout rate in standard 5 for instance will be lower than in standard 1, seen we are already dealing with a lot fewer pupils in the system) * The slightly negative value here is a result of combining data from two different sources, i.e. enrollment from the school census and repetition rate from the DHS household survey. To indicate how the efficiency index is calcu- However, in reality the system requires a lated, it is convenient to use a pseudo cohort of larger number in early grades that progressive- 100 pupils entering Standard 1 and see how ly shrinks as grades evolve. The consequence is they progress. As shown in the table above, it that to finally produce the 33 pupils in Stan- is estimated that 33 eventually will survive dard 8, the system has to finance in each grade through to Standard 8. To produce these 33 the number of survivors until that level, students in Standard 8, it would have been nec- increased by the repeaters. For instance in essary theoretically to finance 261 (32.6x8) Standard 1 that would be 145, as a result of years of schooling in absence of repetition and 100+(100*45.3%); or in Standard 5 that dropouts. would be 69, as a result of the 60 survivors 141 142 Cost, Financing and School Effectiveness of Education in Malawi until this level plus the repeaters 60*15.2%. no repetition in the system (only drop-out). In Summing this over all the grades for primary, that case, to produce the 33 pupils in Standard this would give a total of required actual years 8, the system has to finance in each grade the to finance of 657 to ultimately produce the 33 number of survivors (not loaded by the repeti- students in Standard 8. tion rate). For instance in Standard 1, that Dividing the number of school years the sys- would be 100; or in Standard 5 that would be tem effectively needs to finance to produce 59.8. Summing this over all the grades for pri- these 33 Standard 8 students by the theoretical mary, this would give a total of required actu- number of years (261), we get the efficiency al years to finance of 523 to ultimately produce ratio (a value of 1 would describe a system the 33 students in Standard 8. Therefore the without dropout nor repetition). The index of efficiency ratio in this case would increase to overall efficiency in the case of Malawi is 39.7 49.9 percent. percent, indicating by complement to 1, that It is also possible to look at the difference 60.3 percent (100­39.7) of the public between the number of years required under resources are used on children who never fin- the previous two scenarios (657 and 523). This ish primary or on repetition years (even of gives an indication for the efficiency of a sys- those who eventually finish). tem where no dropout would exist but only the Additionally, it is possible to look at what actually existing repetition would remain. This would be the situation in case there would be efficiency is 79.6 percent (523 / 657). ANNEX 4.2 Allocation of teachers in public primary schools in each individual district, and pupil teacher ratio by district, 2000 District No. of schools Randomness PTR PTR qualified teacher Mchinji 154 37 60 127 Lilongwe City 44 (48) 38 66 Lilongwe Rural East 197 45 72 186 Lilongwe Rural West 193 60 64 148 Dedza 188 60 72 136 Ntcheu 223 54 71 155 Central Western 62 131 Mangochi 165 28 82 192 Balaka 95 34 58 125 Machinga 96 27 76 179 Zomba Urban 12 (9) 49 57 Zomba Rural 133 69 70 146 South Eastern 70 146 Blantyre City 40 (17) 56 80 Blantyre Rural 145 24 54 128 Mwanza 99 29 62 111 Chikwawa 141 22 75 152 Nsanje 101 67 62 115 South Western 60 109 Chiradzulu 80 48 62 104 Thyolo 174 37 71 146 Mulanje 140 48 71 132 Phalombe 81 21 75 184 Shire Highlands 70 136 (continued on next page) 143 144 Cost, Financing and School Effectiveness of Education in Malawi District No. of schools Randomness PTR PTR qualified teacher Mzuzu City 22 (17) 42 60 Chitipa 158 51 52 82 Karonga 293 29 58 90 Rumphi 121 39 52 82 Mzimba North 224 55 61 110 Mzimba South 220 52 61 114 Nkhata Bay 147 40 64 118 Northern 57 95 Kasungu 293 54 64 134 Nkhotakota 132 30 65 131 Dowa 201 47 59 134 Salima 116 50 67 138 Ntchisi 115 61 66 117 Central Eastern 63 132 TOTAL 65 133 Source for randomness: authors' calculations based upon EMIS 2000. Source for PTR and PTR qualified teacher: Basic Education Statistics 2000. Note: The values in parentheses are less reliable, because they represent very few schools and as result even very few schools that are extremely different from the norm can significantly influence the overall result. ANNEX 4.3 Current teacher deployment process in primary T he inequities in teacher deployment · The budgetary ceilings allocated to per- result from inadequate enforcement of sonal emoluments for the different levels existing policies and a need to improve of education. existing policies, particularly at the dis- trict level where the recruitment of teachers Once districts have prepared their teacher needs to take account of the following: requirements, the information is circulated to the division for consolidation. Thereafter, the · Pupil­teacher ratio (objective 60:1); divisions pass district teacher requirements data to the HR services section of the MOEST · Teacher attrition due to retirement, death for further consolidation and an estimation of and resignations; national teacher requirements. These national teacher needs are then submitted to the · The assignment of teachers to newly com- HRMD, Ministry of Finance, Office of the pleted schools on the basis of the number President and Cabinet. If all three departments of classrooms and standard classroom/ approve these numbers, then MOEST's HR teacher ratios. services section sends the information to the Teacher Service Commission (TSC) and they However, the above criteria, especially the initiate the recruitment process by instructing pupil­teacher ratio, are not fully applied, the districts to recruit. because there are recruitment constraints set At the district level, a district manager has to by: approve all transfers where a teacher is accom- panying a spouse.37 Although most district man- · The amount of finances allocated to the agers endeavor to reflect local PTRs, they are education sector for man power develop- persistently pressurized by families who wish to ment by both the Department of Human remain together when a spouse is relocating (typ- Resource Management and Development ically from rural to urban areas). Furthermore, (HRMD) and the Ministry of Finance teachers appear to apply pressure through vari- education sector; and ous networks and thereby make it even more dif- ficult for the DEMs to manage deployment in an 145 146 Cost, Financing and School Effectiveness of Education in Malawi impartial manner. Indeed, at times, teachers do The reasons for the inconsistencies in not report that they have moved. deployment practices among individual In addition, teachers employed at central schools within single districts are not fully level can work anywhere in Malawi. For exam- clear, but issues that motivate this frequent ple, if teachers wish to transfer from one dis- mobility are: spouse relocation considerations; trict to another, they do not need to resign-- a desire for improved access to roads and facil- even if it is known that the district they are ities; lack of housing or incentives to remain in going to has already achieved the required remote rural regions; and lack of promotional PTR--and this practice results in a dispropor- opportunities. In addition, the lack of knowl- tionate number of teachers in urban schools. edge or information sharing by schools and When this happens, the budgetary salary allo- PTAs about the situation in other schools in cation follows these teachers wherever they go the district, or about the existence of powers to rather than remaining with the district that ini- redress the situation for individual schools, tially hired these teachers. may be equally significant. ANNEX 5.1 Percentage of pupils in standard 6 reaching minimum and desirable mastery levels in reading literacy, 1998 Minimum Desirable Mastery Level Mastery Level % % Gender Boys 24.3 0.7 Girls 18.7 0.6 Socio-Economic Level38 Very Low (0­2) 14.4 0.0 Low (3) 14.0 0.6 High (6­7) 27.3 0.8 Very High (8­9) 32.4 3.1 School Location Rural 19.7 0.2 Small Town 19.9 0.0 Large Town 43.8 5.0 Malawi 21.6 0.6 Source: SACMEQ Study, 2001. 147 ANNEX 5.2 AIDS/HIV T he table below summarizes the results requiring 75,600 teachers by 2014, only and impact on teachers over a 10-year 73,600 teachers will be required because of the period, with and without HIV/AIDS, slightly smaller school-age population. This and with negative impact on repetition scenario assumes that with the current growth and dropout rates. The growth rates for the in the school population (3.2 percent in 1987 total and the school-age population are and 2.2 percent in 1998), the number of teach- assumed to be 2.2 and 3.0 percent, respective- ers would need to increase from 50,000 in ly, in the absence of HIV/AIDS.39 2000 to 75,600 by 2014--an increase of The attrition rate, without AIDS, reflects 25,600. However, because HIV/AIDS also current attrition rates due to retirement, death, affects the mortality rate of teachers, the or changes in employment. But HIV/AIDS will replacement of teachers due to deaths caused impact on women's fertility and mortality rates by AIDS will need to double by 2014--from and result in lower growth rates in the popula- approximately 3,000 teachers under normal tion of school-age children. As a result, fewer attrition conditions to 6,500 teachers due to teachers will be needed to maintain the ratio of normal attrition combined with the impact of pupils per teacher. For example, instead of AIDS. Increases or decreases in school enrollment, teachers under three AIDS scenarios from 2000 to 2014 assuming steady state (no changes in pupil teacher ratio (PTR)) With AIDS and adverse impact on repetition Without AIDS With AIDS and drop out rates School Population + 1,500,000 +1,400,000 ­100,000 Total teachers 2000 50,000 50,000 50,000 Total teachers 2014 75,600 73,600 48,300 Total increase in teachers +25,600 +23,600 ­1,700 Teachers required per year +3,000 +6,500 +3,700 Source: 2001 study "The Impact of HIV/AIDS on Primary and Secondary Schooling in Malawi: Developing a Comprehensive Strategic Response.40 World Bank calculations. 148 Annexes 149 Evidence from the HIV/AIDS pandemic in reflects the adverse effects on attendance, the Uganda, Malawi, and Zambia suggests that increasing number of orphans and the growing the growth rates in both total and school-age poverty--especially during the hungry sea- population will decline. Both population son--as a result of HIV/AIDS. Furthermore, groups will grow at a slower pace, although the many orphaned and impoverished children growth of the school population will remain will not have access to education under the almost unchanged. It is assumed that the attri- current framework and support mechanism. tion rate due to death from AIDS will increase Based on this scenario, namely, maintaining a in the region of 3 percentage points to 7 per- constant PTR with no improvement in the cent. Therefore, it is calculated that during the retention rate, the number of teachers required period 2000 to 2014, government will need to will actually fall by 1,700 teachers to a total of increase the annual replacement of teachers by approximately 48,300. Consequently, govern- 100 percent because of deaths from AIDS. This ment will need to recruit additional teachers phenomenon clearly has important implica- but on a smaller scale. In effect, an additional tions for the development of the pre- and in- 700 (total + 3,700 p.a.) new teachers need to service teacher training systems and the issue is be recruited annually to replace the teachers discussed in the main body of this report. who have died from AIDS. The negative con- Finally, the total number of teachers that will sequences of both the annual training and the be required by 2014, and the recurrent cost replacement of teachers, as well as improving implication, appear largely neutral; however the retention and access of an increasing num- the cost and training implications for the ber of children who will be orphaned, has sig- replacement of these teachers is enormous. nificant implications for attaining universal The 2001 study41 estimates that the school primary education in Malawi. Furthermore, age population will actually decline from its the teaching and learning conditions for stu- current level of approximately 3 million pupils dents who actually attend school will undoubt- to 2.8 million pupils. This decline reflects the edly be adversely affected by the higher level of negative impact on repetition and dropout teacher and student absenteeism because of ill- rates, as described in the last scenario. It also nesses and increasing opportunity costs. ANNEX 6.1 Relative gender gap and GER in the different districts, 2000 40.00% Nsanje 35.00% Chikwawa 30.00% gap 25.00% 20.00% 15.00% gender 10.00% 5.00% 0.00% Relative -5.00% -10.00% 0.0% 20.0% 40.0% 60.0% 80.0% 100.0% 120.0% 140.0% 160.0% 180.0% Total GER at district level Source: Basic Education Statistics and UN-EFA population. 150 ANNEX 6.2 Estimated survival rate by gender in primary and secondary, 2000 (cross-sectional method) 100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 boys girls Source: EMIS for enrollment, DHS2000 for repetition, UN-EFA population. Note: The last point of measurement is the new enrollment (non-repeaters) in form 4, NOT the MSCE passers. 151 ANNEX 6.3 Completion rates for standards 6 and 8 (PSLCE passers), by gender Standard 4 120.0% 100.0% 80.0% 60.0% Boys 40.0% Girls Percent 20.0% 0.0% 1992 1993 1994 1995 1996 1997 1998 1999 2000 Y ear Standard 8 60.0% 40.0% Boys 20.0% Percent Girls 0.0% 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year Source: Basic Education Statistics for enrollment and repeaters, and NSO population. Notes: `Proxy' completion is calculated as the number of non-repeating students in Std. 6 divided by the corresponding age cohort. UN-EFA population is not available this far in the past, nor are household surveys for repetition rates. However, seen that the goal is to look at relative difference in completion by gender, the potential shortcomings in population estimates and repetition rate are less relevant. 152 ANNEX 6.4 Incidence of public education expenditure by consumption quintile and sex, 1997/98 Poorest 20% 2nd 3rd 4th Richest 20% Female All 16 19 20 20 26 Primary 22 21 21 20 16 Secondary 9 16 16 23 36 University 0 11 28 0 61 Male All 16 20 19 20 25 Primary 21 22 21 19 16 Secondary 13 19 16 21 30 University 0 14 22 16 49 Source: 1997/98 Malawi data from IHS 1997/98 and MoE (1998) as found in PER 153 ANNEX 7.1 Recent enrollment numbers and graduates from UNIMA and Mzuzu University Student Enrollment By College/Faculty And Year, Unima42 COLLEGE 1997 1998 1999 2000 2001 Faculty M F M F M F M F M F BUNDA COLLEGE OF AGRICULTURE Agriculture (Total) 382 116 370 120 400 109 382 72 488 104 CHANCELLOR COLLEGE Education 388 162 385 134 441 117 430 117 467 140 Humanities 115 82 107 72 114 73 147 93 150 88 Law 51 19 68 28 76 29 95 34 95 37 Science 192 48 189 53 210 50 207 61 194 61 Social Science 322 114 182 74 237 101 210 114 201 122 Total (College) 1068 425 931 361 1078 370 1089 419 1107 448 COLLEGE OF MEDICINE Medicine (Total) 70 23 70 23 70 23 70 23 70 23 KAMUZU COLLEGE OF NURSING Nursing (Total) 47 192 65 198 67 164 80 168 74 196 THE POLYTECHNIC Applied Studies 142 40 165 37 172 40 262 44 149 35 Commerce 385 123 352 110 317 121 295 118 355 151 Education and Media Studies ­ ­ ­ ­ ­ ­ ­ ­ 125 32 Engineering 374 25 350 27 373 27 353 29 357 35 Total (College) 901 188 867 174 862 188 910 191 986 253 TOTAL (University) 2468 944 2303 876 2480 853 2528 874 2726 1018 GRAND TOTAL 3412 3179 3333 3402 3744 Source: UNIMA. 154 Annexes 155 Graduates by degree, Unima43, 1995­2001 Award 95 96 97 98 99 00 01 F M F M F M F M F M F M F M Bachelor of Arts 8 6 10 11 21 26 19 27 10 20 23 21 18 22 Bachelor of Arts (Hum.R/Mgt) 0 0 0 0 0 7 12 7 12 8 0 0 8 45 Bachelor of Arts (Pub. Adm) 18 7 18 5 24 2 12 3 14 4 9 3 13 32 Bachelor of Arts Theo. & Relig.Studies 0 15 2 10 1 0 1 17 0 10 0 18 0 18 Bachelor of Accountancy 11 69 9 56 19 42 8 45 6 53 10 42 4 24 Bachelor of Business Admin 8 55 10 78 13 42 18 56 15 34 19 31 19 21 Bachelor of Education ­ Humanities 19 54 14 87 60 170 48 74 20 72 48 106 20 84 Bachelor of Science 9 75 15 49 8 38 5 39 10 37 11 39 14 35 Bachelor of Science in Agriculture 12 48 12 47 23 53 21 69 44 126 22 67 16 87 Bachelor of Science in Engin ) ­ Civil 1 49 3 69 2 69 3 43 3 73 3 60 5 64 Bachelor of Science (Tech. Education) 0 4 0 5 0 0 0 0 0 17 0 0 0 13 Bachelor of Scie. (Environmental Health) 0 0 0 0 0 0 2 24 1 6 11 40 2 13 Bachelor of Scie.(Environ. Scie. &Tech 0 0 0 0 0 0 1 15 7 4 11 22 2 16 Bachelor of Science (Nursing) 19 3 0 0 0 7 0 1 58 11 66 13 26 14 Bachelor of Social Science 14 80 30 77 33 68 16 63 13 39 18 33 Bachelor of Science Honours 0 0 0 12 2 5 1 8 0 2 1 3 0 2 Bachelor of Social Science (Honours) 0 0 0 0 0 0 0 0 0 0 0 0 0 2 Bachelor of Laws (Honours) 7 9 5 12 2 9 3 6 1 19 6 12 0 0 Master of Arts ­ Sociology 4 5 0 1 2 12 0 2 6 27 3 9 5 8 Bachelor of Medicine, Surgery 4 14 0 15 1 17 5 13 9 7 3 18 2 15 Master of Science ­ Biological Science 0 2 0 6 5 19 2 6 6 11 9 23 11 23 Doctor of Philosophy 0 0 0 0 1 0 3 0 1 0 0 0 0 2 Honourary Doctorate (Honoris Causa) 0 3 0 2 0 3 Diploma in Agriculture 24 73 27 104 30 101 1 8 0 0 0 0 0 0 Diploma in Architectural Technology 2 14 1 7 2 5 2 11 3 15 0 0 14 2 Diploma in Business Studies 0 0 0 0 0 0 2 11 3 11 0 0 0 0 Diploma in Education 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Diploma in Engineering 2 115 2 65 5 86 0 1 0 0 0 0 0 0 Diploma in Laboratory Technology 1 16 3 20 9 4 4 12 0 0 0 0 0 0 Diploma in Management 0 0 0 0 0 0 0 0 2 7 0 0 0 0 Diploma in Nursing 53 5 52 5 50 11 5 1 0 0 0 0 0 0 Diploma in Public Health 2 14 1 21 4 19 4 16 4 12 0 0 0 0 Postgraduate Dip in Computer Science 0 0 0 0 0 0 0 0 0 0 0 0 1 2 Postgraduate Dip in Management Studies 2 9 University Certificate in Education 10 0 0 0 0 0 0 0 10 13 0 0 0 0 University Certificate in Midwifery 48 7 50 6 33 1 20 1 4 0 26 0 24 4 TOTALS 276 739 264 768 350 813 218 579 262 641 299 562 ## ## Source: UNIMA. 156 Cost, Financing and School Effectiveness of Education in Malawi Enrollment Mzuzu, mid-2003 Male Female Bachelor Arts (Educ) 141 46 Bachelor Science (Education) 108 19 Bachelor Science (HSE) 36 34 Bachelor Science (Forestry) 44 10 Source: Mzuzu University. Graduates since inception till mid-2003, Mzuzu Male Female Bachelor Science 25 19 Bachelor Arts (Education) 30 8 Bachelor Science (Education) 21 3 Source: Mzuzu University. ANNEX 7.2 Paternal educational attainment among university graduates (percentages) 1980 1987 1994 1999 Category F M F M F M F M Professional 57 17 44 33 56 39 61 45 Skilled non manual 14 31 22 22 0 9 0 12 Skilled manual 0 5 11 0 19 11 9 2 Semi/unskilled 0 2 0 11 6 15 13 13 Farmer 14 36 22 31 6 23 13 23 Not in Labor Force 14 0 0 3 6 3 4 5 Other 0 0 0 0 6 0 0 1 Number of graduates 7 42 9 36 16 66 23 87 Source: Kadzamira et al (tracer study). Note: given the relatively small sample of graduates, especially among girls, the general trend is more relevant rather than absolute percentage levels. 157 ANNEX 7.3 Actual expenditure data from University of Malawi, 1997/98 till 2001/02 a. Proportion of actual recurrent expenditure University of Malawi, 1997/98 till 2001/02 1997/98 1998/99 1999/00 2000/01 2001/02 Emoluments 45.3% 35.2% 31.9% 36.9% 43.6% Benefits 2.7% 15.3% 15.9% 14.4% 15.2% Emoluments + benefits 48.0% 50.5% 47.7% 51.3% 58.7% Utilities 7.0% 9.2% 7.8% 9.6% 7.4% Students provisions/allowances 11.5% 10.2% 10.2% 10.9% 9.3% Teaching materials/equipment 3.7% 4.1% 4.5% 2.8% 3.1% Books and periodicals 0.7% 0.5% 0.7% 0.5% 0.5% Travel subsistence 1.7% 1.5% 1.5% 1.2% 0.7% Vehicle maintenance/fuel&oils 3.9% 4.3% 3.5% 4.4% 2.5% Repairs houses, buildings, equipment 2.5% 4.6% 3.6% 1.9% 2.8% Lease financing & asset purchase 2.3% 2.5% 3.7% 5.7% 3.7% Cleaning materials/rates&sanitation/ kitchen equipment 2.3% 1.3% 1.4% 1.5% 1.3% Training & staff development 0.0% 0.8% 0.8% 0.6% 0.6% Research&publications/ conferences&workshops 0.6% 0.8% 1.2% 0.0% 0.0% Common services 13.2% 6.4% 10.8% 7.2% 6.6% Other 2.9% 3.3% 2.5% 2.5% 2.8% TOTAL 100.0% 100.0% 100.0% 100.0% 100.0% Source: Expenditure data received from University of Malawi. Note: Mzuzu University is not included here. 158 Annexes 159 b. Breakdown of `common services' component for 2000/01 Actual in nominal terms As % of total As % of (thousand MK ) common services total expenditure Superannuation 16,800 33.0% 2.3% Accumulated deficit 2,291 4.5% 0.3% Advances revolving fund 565 1.1% 0.1% Conference/workshop 5,741 11.3% 0.8% Examination/entrance 4,539 8.9% 0.6% Research/publication 3,325 6.5% 0.5% Staff welfare/medical 3,956 7.8% 0.5% Training/staff dev 547 1.1% 0.1% University football club/ UNIMA sports association 4,063 8.0% 0.6% Gratuity 2,443 4.8% 0.3% Passages 1,083 2.1% 0.2% Other items 5,612 11.0% 0.9% TOTAL 50,965 100.0% 7.2% Source: Expenditure data received from University of Malawi. c. Breakdown of unit cost expenditure in USD, 1998/99 till 2001/02 1998/99 1999/00 2000/01 2001/02 Emoluments 921 844 1076 1195 Benefits 400 420 420 416 Emoluments + benefits 1322 1265 1496 1611 Utilities 240 206 279 202 Students provisions/allowances 266 270 317 254 Teaching materials/equipment 106 119 81 86 Books and periodicals 12 18 14 15 Travel subsistence 39 39 35 20 Vehicle maintenance/fuel&oils 113 93 129 68 Repairs houses, buildings, equipment 121 96 56 76 Lease financing & asset purchase 66 99 165 101 Cleaning materials/rates&sanitation/ kitchen equipment 35 38 43 36 Training & staff development 22 21 18 17 Research&publications/conferences &workshops 22 31 0 0 Common services (represents general administration) 168 287 210 180 Other 86 67 74 77 TOTAL 2,617 2,650 2,919 2,742 Source: Expenditure data received from University of Malawi; enrollment from Basic Educations Stats. ANNEX 7.4 University graduates who `agreed strongly' and `agreed' with the relevance and quality statements (rounded percentages) O verall, the majority of both female need to strengthen the links between university and male graduates were more con- and `industry'. Here, graduates suggested a cerned with the quality of university number of steps that should be taken. These education (45 percent) and curricula include: organizing industrial attachments for issues (44 percent) than with the competence all students, more practical work in the cur- and commitment of lecturers (36 percent), stu- riculum, more industrial/employer sponsored dent welfare (11 percent) and the acquisition projects, and greater involvement of employers of relevant knowledge and skills (11 percent) in the development of curriculum. (see second table). The proportion of graduates Sixteen percent of the graduates also sug- who were concerned with the competence and gested that government should increase fund- commitment of lecturers was significantly ing to the university. However, this suggestion higher among the 1990s cohorts than the was made in the light of the frequent closures 1980s cohorts, whereas perceived curricula experienced by the university, which are partly problems were slightly lower. In part, this is due to lack of adequate funding from govern- because the university has been unable to ment. The students cited the frequent interrup- retain well-qualified staff due to the general tions as one of the weaknesses of university decline in real incomes, particularly since the education, and this was seen as negatively 1980s, and the failure of university to offer affecting curriculum delivery and quality. competitive salaries. Given the probably fiscal pressure on govern- When queried about recommendations to ment spending for the foreseeable future, large improve university education, graduates were increases in government funding to the univer- concerned about the outdated and irrelevant sities will not be possible. However, ensuring a stocks of library books, the deterioration of the reasonable, but stable and predictable flow of physical infrastructure and the general lack of funds is just as crucial at this level. More real- reading and learning materials and equipment. istic tuition fees can clearly contribute to this Many graduates also emphasized the urgent objective. 160 Annexes 161 Statement 1980 1987 1994 1999 Total My university education was a good preparation for my working life 100 96 64 95 96 Most of my lecturers at university were committed to their work 96 79 86 816 85 Most of my lecturers at university were competent 92 84 82 83 84 The curriculum for my degree course was relevant and up to date 83 69 70 65 71 The job expectations have been fulfilled 67 56 52 43 52 I had acquired the necessary knowledge and skills required for my chosen profession 85 82 82 82 83 I was satisfied with my verbal communication skills 92 91 86 97 90 I was satisfied with my written communication skills 96 93 86 93 91 I was satisfied with my information technology/computer skills 20 27 39 36 32 I was satisfied with my analytical/problem solving skills 75 73 74 79 76 I was satisfied with my managerial/entrepreneurship skills 46 45 48 54 49 Weaknesses 1980 1987 1994 1999 Total The quality of education was poor 45 44 39 46 43 The curriculum was irrelevant, too academic & overloaded 55 53 39 33 42 The lecturers were inadequate, not well qualified and demotivated 15 33 48 37 36 Poor student welfare regime 25 17 13 11 11 Did not provide skills and knowledge required for the job market 13 17 8 11 11 Source: Kadzamira et al (tracer study). Note: results are from a tracer study with 380 university graduates from four separate cohorts. ANNEX 8.1 Details of policy scenario and cost implications T his section discusses the long-term At present the model looks at the expected sustainability of the general evolu- human resource flow (students and teachers) tion of the education system. It is through the education system, and at the recur- based on one possible scenario of rent cost items, related to primary and second- targets and assumptions that were discussed ary education, primary and secondary teacher with government and it largely reflects the training, and higher education. These are most PIF--although alterations were made when PIF relevant for MOEST, since the recurrent costs goals were deemed unrealistic. The ensuing and the human resources availability will most results are presented as a basis for discussion directly influence the sustainability of the edu- with government and donors. cation system over time.45 The scenario for The scenario attempted to be more realistic which the results are presented here is based on than would be possible if the pure PIF targets, the assumptions mentioned hereafter. For tar- established several years ago, were strictly get parameters, the PIF targets by 2012 are adhered to; however, assumptions about the used, unless indicated otherwise. lowering of repetition and drop-out rates may still turn out to be overly optimistic. The simu- lation model is based on the Education For All, Scenario assumptions Fast Track Initiative model but it has been adjusted to reflect the specifics of the Malaw- Primary enrollment estimates. To estimate pri- ian education system, and specially to reflect mary enrollment, the key indicators are set by crucial targets in the PIF document. The enroll- the target year to levels somewhere between ment estimates for primary are completed sep- the PIF targets46 and the current situation.47 arately but they are based on the simulation Repetition in Standards 1 and 8 are set to reach model that was used initially as part of the PIF 15 percent by 2015,48 for standards 2 to 6 to since it allowed a more detailed estimate using reach 10 percent; drop-out is set to reach 5 per- repetition rates and dropout rates by standard. cent on average over all grades; and admission This simulation model was provided by rate to reach 110. UNESCO, and adjusted for Malawi, with the Secondary enrollment estimates. Secondary help of Sussex University.44 enrollment projections, and thus the need for 162 Annexes 163 qualified teachers in secondary can be derived For secondary teacher training, it is assumed by using the current transition rate from pri- that Domasi pre-service training remains at its mary to secondary or by using the target GER current capacity level and that the present for secondary, as stated in the PIF. Both will SSTEP program will also continue beyond the have a substantially different outcome in current three cohorts and with one cohort of enrollment projections. The issue of placing a 270 teachers expected graduates each year. As cap on the GER and on public funded second- far as the university colleges are concerned, the ary education is a major but important policy current on-site capacity would remain the issue and it will have a wide range of implica- same but there would be a separate distance tions on secondary teacher needs and recurrent learning model with a total of 250 graduates teacher remuneration costs. each year (e.g. Mzuzu 100 and Chancellor Col- The scenario presented here uses the GER lege 150). Their first teachers would graduate target for secondary as the driving factor for in 2007 given the four year duration of the enrollment estimates. Since it is already 29 per- course. cent in the base year 2001, the GER will hard- ly have to change and thus growth in enroll- ment is almost exclusively a result of growth in Resources availability and population the school age population 14­17 years. estimates Primary and secondary teacher training. There are a number of varied models for The GDP for the base year is taken from the teacher training and in the analysis several sce- IMF for 200149 and a GDP growth rate of 2.9 narios were examined to assess the impact on percent per year is applied.50 For national primary goals by maintaining the current recurrent `discretionary'51 expenditure (i.e. less MIITEP program, as planned, or alternatively, statutory expenditure) it is assumed that it will re-introducing the traditional 2-year pre-serv- remain at about 14 percent.52 For the recurrent ice teacher training program. A similar strate- spending on education, the actual expenditures gy was followed for secondary teacher train- (`actuals') are used for 2000/01.53 The result of ing, although it should be noted that the the scenario, presented here, is based on using scenario retained and presented, here, does not the recurrent education spending as a percent- suggest it is specifically endorsed--or that age of total recurrent spending as a constant other options are not equally, or more, valid. over time (21­22 percent). A second scenario, For primary teacher training, the scenario but not presented here, assumes this percent- assumes that two of the six Teacher Training age would increase rapidly to 27 percent (as Colleges (TTC) will revert to the traditional indicated in the PIF).54 pre-service system, but with the first group of For the total population, NSO projections students only starting the pre-service program for each individual year between 2001 and in 2006 (thus graduating in 2008). The 2015 are used; for the population age 6­13 remaining four TTCs continue with the current and population age 14­17 the population esti- MIITEP program and manage two cohorts mate from NSO for the base year 2001 is used each year. One extra TTC is already planned to and then a growth rate of 2.9 percent per year be built and it would start producing graduates applied (which is a high end growth rate esti- under the MIITEP program in 2006. mated by UNESCO). 164 Cost, Financing and School Effectiveness of Education in Malawi Assumptions and projections, 2001­2015 2001 2002 2003 2004 2005 2006 2009 2012 2015 Recurr. educ. spending as % of national recurrent discretionary budget 21% 21% 21% 22% 22% 22% 22% 22% 22% National recurrent discretionary budget as percentage of GDP 14.0% 14.0% 14.0% 14.0% 14.0% 14.0% 14.0% 14.0% 14.0% GDP (million MK) 120308 123748 127288 130928 134673 138524 150753 164061 178544 Population (thousands) 10816 11175 11549 11938 12341 12758 14085 15528 17102 GDP/capita (MK) 11123 11074 11022 10967 10912 10858 10703 10566 10440 Sources: Authors' calculations. Base year data are from IMF (GDP), Malawi economic report (budget) and NSO (population). The population growth rate is from UNESCO Primary education and primary year. Repetition and drop-out rates for the base teacher training year are taken from the Ed Data 2002 house- hold survey (with very similar results as the Total enrollment by standard in primary is DHS2000 survey). taken from EMIS 2001 statistics for the base Assumptions Ed Data 2002 Stand 1 Stand 2 Stand 3 Stand 4 Stand 5 Stand 6 Stand 7 Stand 8 Repetition (%) 40.6 24.2 27.7 19.1 10.3 9.8 12.2 19.5 Dropout (%) 8.5 5.9 4.8 6.3 10.3 11.8 9.9 20.1 Source: Ed Data 2002. Private school enrollment in primary is con- 10 percent by 2015 (because it reduces costs sidered to remain at less than one percent over for government), but the PIF does not suggest time--this is what it is at present, according to any target for primary. Teacher attrition rate is EMIS 2001. EFA-FTI targets estimate this at considered 6 percent.55 Primary projections: enrollment, PCR, repetition, GER and PqTR, 2001­2015. PRIMARY 2001 2002 2003 2004 2005 2006 2009 2012 2015 Total enroll.56 3236319 3300423 3407299 3505522 3582180 3662103 3820127 3973231 4158317 Avg. repetition 25.2% 23.3% 21.9% 20.7% 19.7% 18.6% 15.9% 13.4% 11.1% Gross Admission 132% 130% 129% 127% 126% 124% 119% 115% 110% Gross enroll. 126% 125% 125% 125% 125% 124% 118% 113% 109% Standard 6 proxy complet. 52% 57% 68% 66% 71% 77% 81% 81% 82% Qualify teachers57 31 799 31 891 31 978 34 859 37 567 42 113 51 948 59 101 65 041 Unqualified teachers58 21 645 22 281 23 782 22 372 20 813 17 502 10 258 5 977 3 855 Total teachers59 53 444 54 172 55 759 57 231 58 380 59 615 62 206 65 078 68 896 New teachers trained 1923 2000 2000 4800 4800 6800 5650 5650 5650 Pupil:Qualified Teacher Ratio 101 103 106 100 95 86 73 67 64 Annexes 165 450000 35% 12000 400000 30% 10000 350000 25% 300000 8000 250000 20% 200000 15% 6000 150000 10% 4000 100000 50000 5% 2000 0 0% 0 2001 2002 2003 20042005 20062007 20082009 2010 20112012 20132014 2015 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Total enrollment in secondary Expected number of qualified teacher Gross Enrollment Rate Expected total number teachers Percentage private school enrollment Sources: Authors' calculations. Base year data are from EMIS 2001 and Ed Data 2002. This scenario results in an estimated primary sufficient to have a 60:1 PqTR, given attrition enrollment of 3.9 million by 2012. The quali- remains at 6 percent. fied primary teacher needs to achieve the PIF If however, all TTCs reverted to the 2-year target of a PqTR of 60:1, under this scenario, pre-service training program, with the first is 64,500 by 2012. If instead the situation graduates available in 2005 (and only MIITEP regarding repetition, dropout and admission graduates until then), and the planned extra rate evolved as described in the PIF, this would TTC also runs the pre-service system, then by result in an estimated primary enrollment of 2012 the expectation is to have 33,700 quali- 3.5 million by 2012. The qualified primary fied teachers, and a 115:1 PqTR. Even if two teacher needs to achieve the PIF target PqTR further TTCs are built with 625 enrollment teacher ratio of 60:1, under this scenario, is capacity each, and commence producing grad- 58,000 by 2012. uates in 2007, the PqTR will not reduce any- If no changes occurred to the current repeti- where near the PIF target, because in total tion, dropout rate, and admission rate, then the 40,000 qualified teachers are expected by estimated enrollment for primary is 4.2 million 2012, and a 96:1 PqTR. Thus, while pedagog- by 2012. The qualified primary teacher needs ical reasons and political commitment would to achieve the PIF target PqTR teacher ratio of advocate a return to the pre-service system, the 60:1, under this scenario, is 70,000 by 2012. It impact of maintaining a high PqTR is evident. is clear that the failure to reduce repetition and dropout rates will result in having to train, and pay, a large numbers of extra teachers. Secondary education and secondary Under the `in-between' scenario, the expec- teacher training tation is to have 59,100 qualified teachers, or 92.5 percent of all teachers qualified, and a Transition from primary to secondary is esti- 67:1 PqTR by 2012. The trend in PqTR would mated in the PIF by dividing the non-repeaters still be downward and the 60:1 goal could be in Form 1 in 2001 by the PLSCE passers in reached over time because more teachers 2000, and gives a 74.2 percent result. There- would be trained than would be lost due to fore, access to secondary, as measured by tran- attrition if the stock of qualified teachers was sition from primary to secondary, is presently 166 Cost, Financing and School Effectiveness of Education in Malawi far beyond the PIF target to be reached by whereby the transition from primary to sec- 2012 (30 percent). This is the case because of ondary of 74.2 percent is taken as a driving the introduction of the CDSSs. As indicated factor. The scenario used here is based on the previously, two very different enrollment sce- GER of 30 percent. narios for secondary can be modeled. The first, Private school enrollment in secondary was whereby the GER of 30 percent is taken as a found to be 11 percent in 2001 (EMIS 2001), driving factor to estimate the number of and set to further increase to 25 percent by enrolled students in secondary, and the second, 2012, as put forward in the PIF. Teacher attri- Secondary projections: enrollment, PCR, repetition, GER and PqTR, 2001­2015 SECONDARY 2001 2002 2003 2004 2005 2006 2007 2009 2012 2015 Total enrollment61 274949 284036 293418 303106 313109 323437 334101 356480 392847 428025 Gross Enrollment 29% 29% 29% 29% 29% 29% 29% 30% 30% 30% Percentage private school enrollment 11% 12.5% 13.7% 15.0% 16.2% 17.5% 18.7% 21.2% 25.0% 25.0% Qualified teachers62 2 290 2 547 3 116 3 681 4 242 4 750 5 483 6 840 8 631 10 167 Unqualified teachers63 5,260 5 127 4 685 4 252 3 826 3 458 2 870 2 072 1 190 534 All teachers64 7,550 7674 7801 7933 8069 8208 8353 8912 9821 10701 New teachers trained 347 372 696 721 745 720 970 970 970 970 Pupil:Qualified Teacher Ratio65 120 112 94 82 74 68 61 52 46 42 450000 35% 12000 400000 30% 10000 350000 25% 300000 8000 250000 20% 200000 15% 6000 150000 10% 4000 100000 50000 5% 2000 0 0% 0 2001 2002 2003 20042005 20062007 20082009 2010 20112012 20132014 2015 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Total enrollment in secondary Expected number of qualified teacher Gross Enrollment Rate Expected total number teachers Percentage private school enrollment Source: Authors' calculations. Base year data are from EMIS 2001. Annexes 167 tion rate is considered 5 percent for the stan- the PqTR even if an attrition rate of 5 percent dard scenarios.60 is achieved. Secondary enrollment by 2012 can be expected to reach 393,000, since the GER in the base year (2001) was already 29 percent Recurrent cost implications and therefore the growth in enrollment num- bers will be in line with the growth in school Assumptions for primary. Teacher remunera- age population 14­17. (This would however tion is a crucial element of the recurrent cost mean that the current transition rate from pri- calculation and qualified teacher remuneration mary to secondary of 74 percent will not be in primary was found to be about 5.5 times maintained as more children pass the PSLCE.) GDP per capita, and was assumed to be kept at The required number of qualified teachers that level, i.e. only increasing in real terms at required under this assumption to achieve the the same rate of the GDP per capita.66 Unqual- PIF target of 40:1 PqTR, and having only qual- ified teacher remuneration was found to be ified teachers after 2007, is 9,800. about 2.9 times GDP per capita, and was If the driving factor for secondary enroll- assumed to be kept at that level.67 Spending on ment is to maintain the 2001 transition rate of teaching and learning materials (T&L) as a 74 percent (which is much higher than the PIF percentage of total recurrent spending in pri- target of 30 percent by 2012), then the enroll- mary was found to be 6 percent,68 and was set ment projections will be strongly dependent on to increase to 13 percent by the end of the peri- the number of children who are expected to od--as prescribed in the PIF. Spending on total pass the PSLCE over time. This in itself ORT as a percentage of total recurrent spend- depends on what will happen with primary ing in primary was found to be about 15 per- enrollment over time. If, as before, the `in- cent.69 It was assumed that over time this per- between' scenario is used for primary, the tran- centage would only increase as a result of the sition rate is kept at the current level, then increase of the T&L component. This results in GER in secondary will be approximately 46 an estimate of total ORT as a percentage of percent by 2012, and secondary enrollment at total recurrent spending in primary of about about 608,000. The required number of quali- 18 percent by 2012. fied teachers to achieve the PIF target of 40:1 Assumptions for secondary. Qualified PTR, and having only qualified teachers after teacher remuneration was found to be about 2007, would be 15,200. 11.5 times GDP per capita, and was assumed Under the scenario of a 30 percent GER as to be kept at that level, i.e. only increasing in the driving factor, the expectation is to have real terms at the same rate of the GDP per 8,600 qualified teachers by 2012, and a 46:1 capita, although it is likely there will be further PqTR. With attrition of 8 percent (instead of 5 nominal increases to compensate for infla- percent) as a result of HIV/AIDS, the PqTR tion).70 Unqualified teacher remuneration was would come down to around 53:1 by 2012. If, found to be about 5.5 times GDP per capita. however, the current capacity in secondary This is realistic since they are qualified primary teacher training is maintained, without contin- school teachers and this is the amount noted uing the SSTEP program beyond its initial 3 for qualified primary teachers. cohorts, and without increasing the capacity of Spending on total ORT as a percentage of the university colleges by 250 by 2012, the total recurrent spending in primary was found expectation is to have 5,700 qualified teachers to be about 50 percent.71 It was assumed that and a 70:1 PqTR. As a consequence there this percentage would decrease to 40 percent would be no real further downward trend in by 2007 due to a large increase in teacher 168 Cost, Financing and School Effectiveness of Education in Malawi remuneration as a result of the many more Assumptions for universities and other sub- qualified teachers in the system, and the result- vented organizations. According to the PER ant higher salaries than for unqualified second- (Public Expenditure Review--MOF website), ary teachers. In the total recurrent cost calcu- the proportion of recurrent education cost that lation, the secondary teacher training cost at is allocated to the universities, and other sub- university colleges is not added separately vented organizations, fluctuated between 27 because it is considered automatically included percent in 1993/94 and an expected 20 percent in the fixed portion of funding allocated to in 2001/02. Given the PIF goals to make uni- universities. This, however, is also a major sim- versity education more efficient (lower unit plification since a substantial increase in cost per graduate, especially the publicly paid teacher training costs at the university colleges component of the unit cost), the assumption means savings will have to happen elsewhere in was made that this 20 percent would be main- the universities to ensure the universities, as a tained over time until 2012 even though there whole, have the same portion of the total edu- is strong system expansion. Since there is an cation budget over time. estimate of the recurrent costs of primary, sec- Recurrent cost implications, 2001­2015 Constant million USD73 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2015 Total recurrent cost 77.9 79.6 83.7 86.5 91.3 95.0 97.9 101.5 104.9 108.4 111.8 115.1 121.6 Domestic recur resources 51 55.6 60.3 65.3 70.5 72.5 74.6 76.7 78.9 81.2 83.5 85.9 90.9 Gap ­22.3 ­19.2 ­18.4 ­16.0 ­18.8 ­20.4 ­21.2 ­22.5 ­23.7 ­24.9 ­25.9 ­26.7 ­28.1 Gap as % of nat. discret. recur. budget 9.3% 7.8% 7.2% 6.1% 7.0% 7.4% 7.5% 7.7% 7.9% 8.0% 8.1% 8.2% 8.1% 140 120 100 Total recur cost 80 60 Domestic recur resources 40 20 0 2002 2004 2006 2008 2010 2012 2014 Sources: Authors' calculations. Base year data are from EMIS 2001. Annexes 169 ondary and teacher training, this allows a cal- budget will be continuously around 6­8 culation of the expected recurrent costs of uni- percent over time between now and 2012. versities and other sub-vented organizations.72 In the table above, the assumptions made The model targets and assumptions indicate are similar to assumptions made in the first that recurrent cost implications of the different table, but with spending on education as a per- policies would be larger than the domestic centage of national `discretionary' recurrent resource availability. It is clear that there is a budget increasing to the PIF target of 27 per- need to have in-depth discussions with cent by 2005. This will have the following MOEST, MOF and the donor community, at repercussions: large, on the realism of this type of scenario. For instance, the expected total resources · Recurrent expenditure on primary educa- available and the teacher salary costs (both tion would require about 71 percent of unit cost and total cost) are crucial issues for total available recurrent education budget dialogue: the latter because it requires such a in 2012 major portion of the total resources. In fact, this is also why issues such as planning teacher · Recurrent expenditure on secondary edu- training must not be done in isolation from the cation would require about 29.5 percent general issue of sustainability. This issue cen- of total available recurrent education ters on the payment of teachers once they are budget by 2012 hired and trained. It is also crucial to under- stand what the expectation is for further · The total shortage between the expected involvement of the HIPC debt relief program. recurrent education costs and expected The results of this financial model can facilitate recurrent education budget seen as a per- such a dialogue. centage of the total national discretionary ANNEX 8.2 EMIS T his list is only intended as a starting school census and, as such, a clarification point for a review of the EMIS sys- reflecting internationally accepted guidelines tem and does not claim to be exhaus- of the definition of repeaters versus new tive. entrants into a grade, and the subsequent con- (1) At the end of 2003, no EMIS statistics sistent use of this definition, would improve were available because there is always more the reliability of the data. (4) There is also a than a one year backlog between data collec- probable overrepresentation of the enrollment tion and dissemination. As such, a speeding up numbers from the school census (albeit clearly of this process would ensure policy decisions less extreme than the issue of repeaters vs new are based more on the current situation. (2) entrants). It would be good to look at the The fact that in recent years a large number of institutional mechanics behind the reporting secondary government schools and private of the enrollment data in the school census. It registered schools did not provide feedback to might be that certain changes in the incentives EMIS questionnaires should not be acceptable structure have contributed to a tendency to as it makes estimates of even the most basic over-report. On the other hand, it might also indicator of total enrollment unreliable. It also be that even without changes in the formal makes any serious analysis of the current situ- incentives structure, the introduction of FPE ation, and planning for the future, close to generated a behavioral change.74 (5) Statistics impossible. It is also in the government's inter- gathering should be adjusted to the reality of est to work with private schools to provide repetition in secondary (at present no repeti- statistics and it may be an effective tool to tion is assumed). With the enrollment of mass- show the public at large that because of the es of pupils in typically low quality CDSSs, the `stimulating environment' for private second- pressure for repetition must surely have ary school sector, a significant percentage of increased, and the policy of not allowing rep- Malawians still have access to secondary-- etition in secondary is certainly untenable in even though severe budget constraints require this context. (6). It is clear that there are enor- the government to keep its focus mainly on mous inequities in teacher deployment primary. (3) Household surveys consistently between primary grades. As such, it is neces- show much higher repetition rates than the sary to provide data on PTR and PqTR by 170 Annexes 171 grade since these are crucial indicators to able. However, it makes little sense to place all track. It might even be possible to obtain this primary students, older than 14, in one single information from the current EMIS question- category, particularly when the household sur- naires but currently this indicator is not veys indicate that substantial numbers of reported on. (7) There is an enormous dis- young adults aged between 15 and 22 years crepancy between ages of primary pupils are attending primary schools. However, it is according to household surveys versus data only intended as a starting point for a review from EMIS. Establishing better age estimates of the EMIS system and does not claim to be from questionnaires in schools will remain dif- exhaustive. ficult as long as no birth certificates are avail- 172 Cost, Financing and School Effectiveness of Education in Malawi Annexes Notes on enrollment by age combined with popula- 1. This is the 2000 value instead of 2000/01. tion data by age from the National Statistical 2. This is the 2002 value instead of 2002/03. Office (NSO) (low-end estimate); and (iii) 3. For 1990/91­1991/92 the source is IMF based upon the results from the school census "Malawi: Statistical Appendix. June 1995". combined with population data by age from For 1992/93­1995/96 data are from IMF UNESCO (high-end estimate). To ensure com- "Malawi: Recent Economic Developments. parability, children who are already in second- August 1997." The remainder is from IMF ary, or beyond, were eliminated and only those "Selected Issues and Statistical Appendix. pupils attending primary were included. As August 2002." From 1996/97 till 2000/01 are expected, the enrollment rates from the house- actuals. For 2001/02 is estimate. For "Total hold survey show a gradual increase until the Grant", "Total Revenues", "Total Govern- age of 8 or 9 years (due to some late enroll- ment Expenditures", "Total current expendi- ments), and they then remain largely stable tures", "Total development expenditure" the until the age of 13­14 years before declining source is the same. once more. However, this decline is at a slow- 4. Information used in this section was er pace than would be expected. The survey extracted from the Malawi Ed Data 2002 and also finds that a very high percentage of chil- is summarized in Table 2. A sample of chil- dren ages 14 to 21 years were still in primary dren, aged 6­14 years, who had ever attended (e.g. 80% at age 14, 60% at age 17, 20% at school, were involved in the survey. age 21) and that it was only young adults of 5. The GER based upon the population cen- ages 22 years and above that had an enroll- sus uses one set of data for both the nominator ment rate of less than 10%. It is clear that (enrollment) and denominator (population), many of the children entering primary remain which is not the case for the school census that until at least the normal age of finishing pri- only can provide enrollment data. The fact that mary. This does not mean they finish standard a ratio is presented also preempts the argument 8 but rather that they may be repeating the that maybe not the whole population is same, or several grades, before eventually accounted for in the population census, dropping out. The findings from the DHS sur- because any no-counts will cut out both pri- vey closely match the results from the Integrat- mary pupils and people in the age group 6­13, ed Household Survey (IHS) of 1998 (see Annex unless the no-counts would be disproportion- 2.2). ately from the ones in primary but outside 10. Here, using the high-end population esti- 6­13 age range. mates, enrollment rates for children of normal 6. Only includes children age 5 to 29. primary school age (6 to 13) are consistently 7. Another interesting piece of data is that higher than those of the household surveys. IHS 1998 calculated a weighted population of For the children of ages 7 to 10, more children 2,256,896 for the 6­13 age group. are found to be in school than are present in 8. The result here is very similar to the other the corresponding population group (enroll- estimate based upon GER for boys and girls ment rate more than 100%), even when using separately, as done in `Profile of Poverty in one of the highest available population esti- Malawi, 1998. Poverty Analysis of the Malawi mates for the country (UNESCO). At the same Integrated Household Survey 1997­98', which time, however, the school census statistics indi- has GER 125.9 (boys) and 115.2 (girls). cate that the category `all children older than 9. Calculated in three ways: (i) based on the 14' represented less than 80 percent of the pop- DHS 2000 household survey; (ii) based upon ulation at age 15. This is slightly less than the the results from the school census (EMIS 2000) 80 percent that DHS 2000 indicates is the Annexes 173 enrollment by age for children aged 15. In confirmed by the DHS 2000 Survey indicating other words, all the enrollment rates for those 95 percent eventually start primary. Given the age 16 or higher from the DHS 2000 would assumption that the new entrants into stan- not be accounted for in the school census esti- dard 1 in a system that has reached a steady mates. state (which Malawi should have for at least 11. The school census also categorizes all the first grade of primary in 1998 or at the lat- children of age 5, or below, in one category est by 2000) cannot be more than the number and the 15%, here, represents all these children of children of the entry age, it is possible to as a proportion of the children of age 5 only. estimate what the population at entry age The surveys on the other hand only start gath- should be in order to be able to have a number ering data from children age 5. As such, some of new entrants to Standard 1 at the level children may need to be added to the approxi- found in the school census statistics. The size mate 30% for children age 5, who already of the entry age population in 1998 would attend primary at age 4, or below. have to be about 665,484 (=638,865/0.96), 12. This assertion holds irrespective of all with 638,865 being the new entrants to Stan- children, in practice, entering primary at age 6 dard 1 according to the school census. Com- or at another age (typically anywhere between paring the result with the children at entry age 5 and 10), because a new entrant of age 10 in the 1987 census (which we assume to be should never before (i.e. theoretically 4 years correct in light of the arguments presented earlier) have been counted as a new entrant in above), the population in this age cohort the system. As such, in a steady state system-- would have grown at the rate of 8.4 percent a as the Malawi one would be predicted to be by year (=[ln(665,484/263,643)]/11) between now--each child can only be a new entrant 1987 and 1998--surely an implausible rate of once, and the number of new entrants into increase in any human population. This is why Standard 1 should be influenced only by popu- it is worthwhile to review the number of new lation fluctuations and the degree of access to entrants into Standard 1 in more dept. primary. 14. For instance one performed by CIDA in 13. Using the 1987 non-repeating enroll- relation to its primary school book distribution ment numbers from the Basic Education Statis- project. tics, together with the population found from 15. In the survey, data it is automatically the population census in that year, the rate of meaningful if the child is repeating because entry to Standard 1 is 0.87, which is actually both the child's current grade level is asked as close to the estimate of 0.84 found based on well as the grade level (s)he attended the previ- data from the 1992 Demographic and Health ous year. Survey (DHS). It is also noteworthy that in 16. Using for instance the average repetition 1987, the enrollments reported by the school rate for standard 1 according to the DHS 2000 census --1,022,765 children-- are very close and Ed Data 2002 household surveys, which is to the 1,002,343 children in primary schools 42.9% (Average of 45.3% from DHS 2000 enumerated in the 1987 population census. and 40.6% from Ed Data 2002; which is a The consistency in these cross-checks increases conservative value, lower than when using the the confidence that the data for 1987 are prob- higher rate from only DHS 2000.), it is possi- ably reasonably accurate. For 1998 however, ble to recalculate the number of new entrants the situation is different. According to the for both 2000 and 1998 (assuming the repeti- 1998 Integrated Household Survey it is esti- tion rate was already at that level in 1998). The mated that about 96 percent of each age cohort number of new entrants would have been eventually begin primary schooling, which is 441,087 in 1998 (deduct 42.9% of 772481; 174 Cost, Financing and School Effectiveness of Education in Malawi and for 2000 deduct 42.9% of 804,312; which is (0.7 x 298,554) + (0.3 x 291,792) = 296526. are respectively the total enrollment numbers Using age 6 pop the cohort rate of entry would for 1998 and 2000 from the school census), be 1.51.) and 459,263 in 2000. With this new count of 17. The 1998 population census (see 1998 non-repeating enrollments in standard 1, the Population Census Analytical Report on size of the population at entry age in 1998 www.nso.malawi.net) states that between would have to be about 459,466 1987 and 1998 pop grew with 23.2 percent, or (=441,087/0.96) in 1998 and 478,399 (= 1.9 % per annum. The growth for age group 459,263/0.96) in 2000. The population in this 6­13 was 1.4 percent ( [ln age cohort would have grown at the rate of 5.1 (2,115,870/1,818,910) ] / 11 ), with 2,115,870 percent a year (=[ln(459,466/263,643)]/11) the 6­13 population in 1998 and 1,818,910 between 1987 and 1998, which is more plausi- the one in 1987. ble than the 8.4 percent a year found earlier, 18. = [ln (2437000/1818910) ] / 11 but is still quite unlikely given the analysis in 19. The 1987 population census (see Gov- the section on population. The average age of ernment of Malawi. 1991 and 1993.) states standard 1 pupils was 7.7 years in 1998; given that between 1977 and 1987 population grew that the corresponding rate of repetition is with 44 percent, or 3.7 % per annum. Esti- assumed 42.9%, it means that the average age mated growth per year for the 5­14 age group of new entrants is 7.3 years. The algebra is as between 1977 and 1987 is 4.9 percent ( follows: A=N(1­r)+R(r) where A, N and R are, [ln(2,286,959/1,395,961) ] / 10 ). Specific respectively, the average age of all pupils, non- growth rates for the 6­13 age group are not repeaters (i.e. new entrants) and repeaters (i.e. given, but should not significantly differ. only looking at if the pupil is a repeater in the 20. The 6­13 age group population in 1987 current year, not at the past); and r is the pro- was 1,818,910. ln(x)= (11*0.049) + portion of repeaters. Since R=N+1 by defini- ln(1,818,910)= 14.95275­> ex of this is tion, we can make the appropriate substitution 3,118,142 into the first equation, and obtain after rear- 21. ln(x)= (13*0.049) + ln(1,818,910)= ranging, the desired function, N=A­r. Relating 0.637 + 14.6427335 = 15.05075­> ex of this is the number of non-repeaters in standard 1 to 3,439,195 the population corresponding to these ages 22. The results, reveal that the population of (using original population census data), a new the Northern Region increased by 35 percent estimate of the cohort rate of entry to standard while that of the Central and Southern 1 is obtained: 1.60 for 1998 (The population Regions--where the districts are found that corresponding to age 7.3, is obtained simply by border with Mozambique--rose by 31 and 17 prorating the population enumerated for ages percent respectively between 1987 and 1998 7 and 8, i.e.(0.7 x 270,036) + (0.3 x 286,841 = 23. 1998 age 7: 333000, age 8: 326000. 275,077. The difference of using this more Population corresponding to age 7.3 is (0.7 advanced way to estimate the corresponding x333000) + (0.3 x 326000) = 330,900. This is population cohort versus simply using the pop- used together with the new entrants at 441,087 ulation of official age of entry 6 (315,612), is for 1998. not of such magnitude to make a huge differ- 24. 2000 age 7: 330000, age 8": 327000. ence. Using age 6 population, the cohort rate Population corresponding to age 7.3 is of entry to grade 1 would still be 1.40 in 329,100. This is used together with the new 1998). For 2000 this would be 1.55. (2000 age entrants at 459,263 for 2000. 6 population 303,785, age 7 298,554, age 8 25. Cross sectional method is calculated as 291,792. Population corresponding to age 7.3 the gross admission rate to the last grade of Annexes 175 cycle divided by the gross admission rate to cross-sectional method and the full time series first grade of cycle of the same year, with the method on the other hand are both influenced gross admission rate being the non-repeaters in by the effect of the FPE policy. Second, oppo- a certain grade divided by the population of site to the cross-sectional method, the quasi- normal age for that grade. For instance for time-series method also abides by the rule that standard 8 this would be the non-repeating for survival it is best to compare the same pupils in standard 8 divided by the children of cohort of students over time, instead of two age 13. different cohorts at one point in time. The Quasi-time-series estimate is calculated as semi-time series method follows this rule for 6 the multiplication over the whole cycle of the separate cohorts. division of the number of non-repeaters in 2 26. These are generally considered a more consecutive grades in 2 consecutive years. For accurate measures. primary in Malawi this would be: 27. Since repetition rates are only available (Non rep year 2 standard 2 / Non rep year 1 from household surveys in 2000 and 2002, it standard 1) * (Non rep year 2 standard 3 / was decided to use repeater numbers as indi- Non rep year 1 standard 2) * (Non rep year 2 cated in Basic Education Statistics to ensure standard 4 / Non rep year 1 standard 3) * greater consistency between the results from a (Non rep year 2 standard 5 / Non rep year 1 wide range of years. The fact that the repeti- standard 4) * (Non rep year 2 standard 6 / tion rate seems to be significantly underesti- Non rep year 1 standard 5) * (Non rep year 2 mated in the EMIS data matters to the absolute standard 7 / Non rep year 1 standard 6) * levels of the results, but not to the relative dif- (Non rep year 2 standard 8 / Non rep year 1 ferences between the two periods. standard 7) 28. The larger than 8 year interval is to These survival rate calculations are usually adjust for primary graduate repetition. done at different points in time to try to assess 29. The method compares the non-repeaters what has been the evolution Full time series in grade x in 1998/99 with the non-repeaters in perspective of survival for a cycle is calculated grade x+1 in 1999/00. For each of the 8 by dividing the non repeaters in the last grade grades, both sides of this grade by grade com- of the cycle in year `X' by the non repeaters in parison are always either affected by the FPE the first grade of the cycle in year `X minus policy (up to standard 5 in 1998/99, up to length of cycle for the average student' (under standard 6 in 1999/00) or not yet (beyond ideal circumstances this would be `X minus the standard 5 in 1998/99, and beyond standard 6 number of grades in cycle'; however, because in 1999/00) an average student might repeat one or more 30. For secondary education always the esti- times before reaching the end of the cycle the mated enrollment numbers (through extrapo- time frame can be increased for to make up for lation) are used as found in the Basic Educa- that and to ensure it is the same cohort of tion Statistics 2000. pupils that is considered) 31. Non-repeaters in the corresponding The Quasi-time-series method has two grade divided by the population group at the advantages in the case of Malawi. First, the official age for the grade (standard 1 intake result is not `spoiled' by the effect of the Free rate = non-repeaters stand 1 / children age 6 in Primary Enrollment policy, seen it only uses the population) data from two subsequent years, which can 32. For secondary education always the esti- both be years that come after the policy was mated enrollment numbers (through extrapo- introduced, and thus are in the same situation lation) are used as found in the Basic Educa- in relation to the effect of the FPE policy. The tion Statistics 2000. 176 Cost, Financing and School Effectiveness of Education in Malawi 33. In total 92 schools were not included assumptions used in the model (with baseline because they had negative numbers for enroll- data for 2001) are the best that could be ment minus repeaters in one of the grades, or obtained in the time-frame available and with because they listed no enrollment in standard the assistance received. It is clear that improve- 1. ments by MoEST staff can be made if more 34. Ministry of Labour and Vocational precise data is available. Training, 2002. 45. The capital expenditure (with classroom 35. The averages are based only on data for and other construction as main cost drivers) countries with per capita GNP below $1,000 has not been included at this point. It would be in 1993. relatively easy to include this later in the model 36. Some of this cost includes a non-recur- 46. 5 % repetition for standard 1­7, 10% ring cost of providing storage cabinets for the repetition for standard 8; 5% dropout, admis- books. sion rate 100 37. There are many examples of district 47. Average rep rate in primary of 25.8%, managers authorizing teacher transfers on the average dropout 8.2%, admission rate 132% production of marriage certificates. This sys- (according to the Ed Data 2002 household sur- tem is considered open to fraudulent transfer vey) practices. 48. In this case the timing reflects a minor 38. Measured by `possessions in the home'. modification from the PIF target year 2012. 39. The scenario in Annex 3 builds on the 49. Very similar to what is found in Econ assumption that there will be no change in Report 2003 for 2001/02. coverage (no change in enrollment rates) or in 50. Taken from Econ Report 2003 as aver- the ratio of the number of pupils to the num- age for last 6 years. ber of teachers. Thus it describes the system as 51. Recurrent expenditure before debt it was in 1998 and its evolution with or with- amortization minus interests on debt and out the impact of HIV/AIDS. The 1997 figures minus pensions & gratuities. for total population and school-age population 52. From the Economic Report 2003: for are based on the 1997 Malawi census. The 2000/01 actual results in 14.1% of GDP of school-age population covers the age group 6 corresponding year, for 2001/02 actual it is to 15 years old, or grades 1 to 10 (lower and 18.1% of GDP, for 2002/03 revised it is 11.8% upper primary education and lower secondary of GDP, and for 2003/04 estimate it is 9.7% of education); the teacher stock is based on GDP. The serious difference that still exists 34,000 lower primary teachers, 4,000 upper between the discretionary recurrent govern- primary teachers, and 2,000 lower secondary ment budget and the total recurrent govern- teachers, as registered in 1997. ment budget is largely a result of interest pay- 40. Esme Chipo Kadzamira, Dixie Maluwa- ments and debt amortization on domestic debt, Banda, Augustine Kamlongera, Nicola Swain- not on international debt. For 2002/03 son, September 2001, Centre for educational (revised) the discretionary budget is 18,434 research and Training (CERT), Malawi. million MK vs. a total budget of 39,755 mil- 41. Kadzamira et al., 2001. lion MK, and for 2003/04 (estimate) it is 42. Does not include Mzuzu university 16,882 million MK vs. 41,375 million MK. 43. Does not include Mzuzu university This means that the current level of discre- 44. For key indicators in the model the tionary recurrent budget, as a percentage of assumption is made that the system moves GDP, will not be able to increase much more as from the current situation to the target value a result of further HIPC debt relief, but only as (by the target year) in a linear manner. The Annexes 177 a result of the government's ability to reduce its deaths is lower than in the case of primary domestic debt burden. school teachers according to the study "Impact 53. For both MoEST and the subvented edu- of HIV/AIDS on HR in the Malawi public sec- cation organizations (including universities) tor". On the other hand, the qualified teachers `actuals' are taken from "Consolidated Annual (in particular with a science background) appropriation accounts for the financial year might resign quicker to find other type of work ended 30 June 2001". in the private sector. In addition, secondary 54. There is actually a HIPC trigger for edu- teachers often switch from public to private cation requiring the share of recurrent educa- schools--especially when they can take their tion expenditure (including universities and optional retirement after 20 years of service. other subvented organizations) of at least 23% However, since secondary enrollment for both of the discretionary recurrent budget. private and public schools is reflected here, this 55. This is deemed a prudent scenario, since effect does not mean extra attrition for the death and retirement together, for 2001, were teaching body. less than 3%, but partial numbers from the 61. Public and private divisions for 2003 seem to put attrition due to 62. The number of qualified teachers death at a significantly higher rate of 3­4%. As expected is a result of the output of the teacher such, another 3% for resignations and to make training scenario that is used here and of the up for the higher attrition due to deaths was expected attrition. This need for qualified seen by the group as probably quite acceptable. teachers is based upon expected enrollment in (EMIS 2001 captured 907 teachers who passed both public and private schools, since teachers away out of a total stock of 53000. EMIS 2001 for private schools--if qualified--will also also shows 7059 out of 53000 teachers have to come from the same teacher training between age 45 and 80, meaning on average institutions. Of course the teacher remunera- about 700 per year at most could retire over tion cost only reflect those teachers teaching in the next decade.) public schools. 56. Public and private 63. The number of unqualified teachers 57. The number of qualified teachers expected is the difference between the total expected is a result of the output of the teacher number of teachers expected and the expected training scenario that is used and of the expect- number of qualified teachers. ed attrition (assumes all qualified teachers work 64. The total number of teachers expected in public primary schools, not in private ones) in public and private schools is a result of a 58. The number of unqualified teachers gradual transition from the current 36:1 expected is the difference between the total pupil:teacher ratio to 40:1 by 2007, and of the number of teachers expected and the expected expected total student enrollment. number of qualified teachers. 65. This is the average for the secondary sys- 59. The total number of teachers expected tem as a whole, i.e. including public and pri- in public primary schools is a result of a grad- vate schools. ual expected transition from the current 77:1 66. The sources of information for the cur- pupil:stream ratio to 60:1 at the end of the rent teacher remuneration calculation were period, while at the same time, the MoEST Payroll 2002 and the Teacher Salary teacher:stream ratio is set to increase from 0.8 Scale. The number of qualified teachers was: at present to 1 at the end of the period. PT4: 23255, PT3: 6513, PT2: 2232, PT1: 60. PRSP uses 5%, PIF uses 10% on p.30, 1197, and the average yearly salary package Nat Strat teacher training has projections with for PT4: 62256, PT3: 130440, PT2: 183780, 10% and with 5%. Attrition as a result of PT1: 183396. The sum of the products of 178 Cost, Financing and School Effectiveness of Education in Malawi average yearly salary and number of teachers 69. From 1994/95 till 1999/00 roughly 85 for each grade gives an estimate of total quali- to 90% of recurrent primary spending was on fied teacher remuneration per year. The aver- teacher salaries (source PER) age qualified teacher salary of 88171 (total 70. Data used from PPPI at end of 2002 on remuneration/total qual teachers) divided by the number of secondary teachers in each level the 2002 Malawi GDP per Capita of 15915 and the total salary package for each level, the results in the average qualified teacher's salary total average salary (including allowances) was being 5.5 times GDP/Capita. 150000 to 175000. Using the 2002 GDP/cap 67. This result was obtained using total of 15915 (Econ Report 2003), this means OCT salary for TT including allowances between 9.5 and 10.9 times GDP/cap. (K3853pm*12=46236pa) and the 2002 per 71. between 1994/95 and 1998/99 PE was capita GDP of US$207.5 (1$=76.7). roughly 40 to 60% of secondary budget 68. Not readily available for 2001. Howev- according to PER. er, according to PER it was fluctuating between 72. =(0.2/(1­0.2))* recurrent cost of all 4 and 8% in the period 94/95 to 98/99 other levels. Bibliography Al-Samarrai, Samer and Hassan Zaman. 2002. Kadzamira E.C. 2003. Where has all the edu- The Changing Distribution of Public cation gone in Malawi? Brighton: Insti- Education Expenditure in Malawi. Insti- tute of Development Studies. tute of Development Studies, Sussex Uni- Kadzamira, Kamlongera, et al. 2001. The versity & World Bank. impact of HIV/AIDS on primary and sec- American Institutes for Research. 2000. What ondary schooling in Malawi: Developing happened to the teachers? Follow-up on a comprehensive strategic response. teacher mobility in 2 Malawi districts. Mingat, A. and B. Suchaut, 2000. Les Systèmes Improving Educational Quality Project Educatifs Africains: une Analyse study. Economique Comparative. Editions De Bruns, B., Mingat A. & R. Rakotomalala. Boeck Université. 2002. A Chance for Every Child. The Mingat, A. 2003. Management of Education World Bank. Systems in Sub-Saharan African Coun- Gallagher, B., P. Mvula and S. Konyani. 2002. tries A diagnostic and ways toward A rapid appraisal of the impact of food improvement in the context of the EFA- shortages on education. Draft report for FTI. The World Bank. UNICEF, Malawi. University of Malawi, Ministry of Labor and Vocational Training. Center for Social Research. 2002 Strategic Plan 2002--2004. Gorgens, M. 2002. Baseline Study for the Sec- MOE--Ministry of Education, Sports & Cul- ondary Education Project (SEP). ture. 2000. Policy & Investment Frame- Government of Malawi. 1991 Population and work. Housing Census 1987--Summary of MOFEP--Ministry of Finance and Economic Final Results. National Statistical Office. Planning. 2001. Malawi 2000--Public Government of Malawi. 1993. Population and Expenditure Review. Housing Census 1987--Volume II Popu- Presidential Commission of Inquiry into the lation Characteristics. National Statistical Malawi School Certificate of Education Office. (MSCE) Examination Results. Final Report. November 2000. 179 180 Cost, Financing and School Effectiveness of Education in Malawi Thomas, H., Chimombo, J., Hall, D. and EMIS database for 1999, 2000 and prelimi- Mawditt, R.. 2001. Skills for Employ- nary version for 2001 ment­Links between Higher Education Population census 1987 and 1998 and Employers and Recognition of Edu- Integrated Household Survey (IHS)--1998 cational Qualifications in Malawi. The Determinants of Poverty in Malawi, Save the Children. 2002. Final report Malawi 2001. National Economic Council, National food crisis. An HEA Vulnerability Assess- Statistical Office and International Food Policy ment. A contribution to the UN-WFP Research Institute Inter-Agency Vulnerability Assessments. Profile of Poverty in Malawi, 1998 Save the Children (UK). Statistical Abstract, 2000 World Bank. 2000. Memorandum and Recom- Demographic and Health Survey (DHS)-- mendation of the President of the Inter- 2000 national Development Association to the Main Report. (National Statistics Office Executive Directors on Assistance to the website) Republic of Malawi under the Enhanced Ed Data (DHS) Survey--2002 HIPC Debt Initiative. Report No. P Education Data for Decision-making. April 7423­MAI. 2003. NSO and ORC Macro SACMEC UNESCO. 2001. The quality of education: School and Population Census, and some policy suggestions based on a survey of Household/School Surveys used: schools--Malawi interim report. Policy Research: Report No. 7 EMIS school census: MLA Basic Education Statistics for all years With Africa for Africa: Towards Quality between 1990 and 2000. For 2001 a draft was Education for All, 1999 MLA Project available Data set