Mozambique Situation Analysis

Transcription

1 Mozambique Situation Analysis Costing Report UNICEF Mozambique May 2014

2 Table of Contents List of Tables and Figures List of Abbreviations 1 Introduction 1 2 Methodology Education Social Protection Health Water and Sanitation Macro-Economic Assumptions 6 3 Results Education Social Protection Health Water and Sanitation Projections (population and coverage) Type of Costs Annual Public Expenditure 25 4 Aggregate Costs 29 5 Conclusion 31 References / Bibliography 32 Annex A Terms of reference 33 Annex B 34 B.1 Education costs by year and type of expenditures ( ) 34 B.2 Social protection costs by year and type of expenditures ( ) 35 B.3 Health costs by year and type of expenditures ( ) 37 B.4 Water and Sanitation costs by year and type of expenditures ( ) 39 iii iv UNICEF Mozambique ii

3 List of Tables and Figures Figure 1: Selected macro-economic indicators from macro-projections from 2013 to Figure 2: Gross Enrolment / Completion Rates ( ), by scenario Figure 3: Projected total cost of education ( ) under scenarios 1,2,3 ( 000 MT), by year 12 Figure 4: Projected total cost of education in 2023 (in million MT ), by level Figure 5: Number of beneficiary households ( ), by scenario and programme Figure 6: Projected total cost of social protection over the period , by scenario Figure 7: Projected health costs for (in Million MT), by scenario Figure 8: Targets for access to water and sanitation Figure 9: Public expenditure on water and sanitation (in million MT), by year ( ) Figure 10: Breakdown of cost in 2023 (in million MT), by type of expenditure and by area of residence Figure 11: Scenario 1, Water and Sanitation (in Million MT) Figure 12: Scenario 2, Water and Sanitation (in Million MT) Figure 13: Scenario 3, Water and Sanitation (in Million MT) Table 1: Common macro-economic assumptions for different costing models... 6 Table 2: Selected 2016 and 2023 targets, by Scenario... 9 Table 3: Projected total cost of education in 2023 (in million MT ), by activity and economic classification Table 4: Selected 2023 targets, by Scenario Table 5: Projected total cost of social protection in 2023 (in million MT), by scenario Table 6: Selected health interventions coverage targets for 2023 (% of target population), by scenario Table 7: Projected health costs in 2023 (in million MT), by expenditure type and care level Table 8: Targets for access to improved water and sanitation in 2023, by urban/ rural Table 9: Projected annual public expenditure on water and sanitation in 2023 (in million MT), by scenario Table 10: Total cost projections for social sectors under scenario 1 in (in million MT), by sector Table 11: Total cost projections for social sectors under scenario 2 in (in million MT), by sector Table 12: Total cost projections for social sectors under scenario 3 in (in million MT), by sector Table 13: Scenario 1, Education ( 000 MT) Table 14: Scenario 2, Education ( 000 MT) Table 15: Scenario 3, Education ( 000 MT) Table 16: Scenario 1, Social Protection (MT) Table 17: Scenario 2, Social Protection (MT) Table 18: Scenario 3, Social Protection (MT) Table 19: Scenario 1, Health (million MT) Table 20: Scenario 2, Health (million MT) Table 21: Scenario 3, Health (million MT) UNICEF Mozambique iii

4 List of Abbreviations ARV CSO DNA DNAM DNSP ESSP ILO IMF INE JMP MICS MISAU OPM PASD PASP PESS PSSB TGE WASH Anti-Retro Viral Country Status Overview (Water and Sanitation) National Directorate of Water Affairs Direcção Nacional de Assistȇncia Médica Direcção Nacional de Saúde Pública Education Sector Strategic Plan International Labour Organisation International Monetary Fund National Statistics Institute WHO/UNICEF Joint Monitoring Programme for water supply and Sanitation Multiple Indicator Cluster Survey Mozambican Ministry of Health Oxford Policy Management Programa de Apoio Social Directo Programa de Accao Social Productiva Plano Estratégico do Sector de Saúde Programa Subsídio Social Básico Total Government Expenditures Water, Sanitation and Hygiene UNICEF Mozambique iv

5 1 Introduction This report presents the estimated investments required in selected social sectors in order to achieve Mozambique s development targets. The costings presented in this chapter build upon existing medium to long term projections carried out within various sectors in Mozambique. The use of existing costing models ensures that the figures presented in this chapter are consistent with government estimates for the various sectors and that they follow a methodology that is known and understood by the various stakeholders in Mozambique. Where necessary, these costings have been slightly expanded and modified, using the tools developed within each sector, in order to render them comparable and to ensure that there is consistency in the assumptions used in the various models, regarding exchange rates, population projections, etc. In particular, we have had to make adjustments to existing models in order to comply with the established timeframe of the current exercise, which runs from 2013 to Comprehensive medium/long-term budget scenarios are available or currently being finalized in only a few of the sectors (in particular education, health and social protection). In other sectors, these types of broad integrated cost/budget scenarios do not exist (e.g. child protection, ECD), although in some cases there are medium/long-term budgets for sub-sectors or for specific projects (e.g. civil registration). In other sectors, costings exist but are not comparable due to the use of incompatible methodologies or costing horizons (e.g. Multi-Sector Action Plan for the Reduction of Chronic Malnutrition in Mozambique ). Consequently, this exercise focuses primarily on the following four sectors: Education: The Education Strategic Plan includes budget projections, derived from longer term cost estimates (up to 2020) linked to a cohort model, produced by the Ministry of Education with support from DFID. Health: The Health Sector Strategic Plan, which currently covers the period , has been adjusted to cover the period , and has been budgeted using WHO s One Health costing tool. This sector-wide budget integrates health-related nutrition interventions, as well as the recent and separately budgeted Plan for Accelerated Response to HIV/AIDS Social protection: Medium-term budget projections for the scale-up of INAS programmes have been produced by the ILO in collaboration with UNICEF and the IMF, as part of the planning for the implementation of the Basic Social Security Strategy. Unlike the two previous costings, the social protection model is not an official planning document of the government of Mozambique, but an independent study carried out as a background study to contribute to the national policy dialogue on social protection. Water and Sanitation: This model is based on the projections made as part of the 2010 country status overview on water and sanitation that was carried out by the National Directorate of Water (DNA) with support from international development partners. The specific methodology used in each sector is presented below. UNICEF Mozambique 1

6 2 Methodology 2.1 Education The education model is based on a Cohort Progression Analysis (CPA) methodology, which is used to track the progress of each intake of pupils through the school system from Grade 1 onwards. This original pupil intake is called a 'cohort' of pupils. CPA looks at the numbers and proportions of pupils that go forward into Grade 2 and then into each of the following grades (progression), at the numbers and proportions of pupils that fail an exam or assessment and have to go through the grade again (repetition), and at the numbers of pupils that leave the grade and which are lost to the education system (drop-out) 1. The model is comprised of 5 main components, that are linked together to calculate and project the envelope of education resources that are needed to deliver a given level of targets : Population figures and projections from both the 2007 census are used to project the school age population in Mozambique. Enrolment for the 5 different schooling levels (EP1, EP2, ES1, ES2 and TVET) by gender, age, grade and type of course is combined with efficiency data (graduation, progression and transition, repetition and dropout) and used to analyse both the efficiency, performance and output of the education system across the 5 programmes. Cohort Progression Analysis, is used to analyse the size and growth of enrolment, the efficiency of the school system and the proportion and numbers of pupils that progress through the school system to complete the different levels of attainment. Teacher and other staff numbers are analysed based on recent levels of provisioning and targeted future levels of provisioning and efficiency to give the required numbers of teachers, managers and administrators to run the schooling and education support system. Staff and teacher numbers are projected based on targeted resource ratios (such as pupil: teacher ratio) and efficiency gains in the allocation and use of teachers. Education resource needs are projected forward based on the envelope of education resources (books, school buildings, running costs) that is needed to effectively run the education system based upon the projected enrolled numbers of pupils and numbers of teachers and administrators. Again, efficiency targets are built into the model to enable the targeting of an optimal level and balance of education resources. The targets that drive each of the components above are defined on a scenarios sheet at the beginning of the model. The scenarios sheet defines all the parameters necessary for the calculation of costs, from targeted enrolment rates to pupil-teacher ratios, etc. In the original model, two separate targets were set, for 2016 and For the purposes of the current costing exercise, the outer target year was changed to It should be noted that definition of the promotion, repetition and dropout rates used in the model is different to their use in some planning systems, due to the use of Cohort Progression Analysis. All rates are defined with respect to the enrolment at the start of the school year with the promotion rate, need to repeat (or failure) rate and dropout rate summing to 100%. This is different to some education planning systems where the promotion or transition rate is defined with respect to enrolment at the end of the school year after pupils have dropped out. It is also assumed that the repeating pupils in any school year all come from the pupils who needed to repeat due to failure of assessment at that grade in the previous year. An estimated need to repeat returning next year rate is therefore calculated as part of the model. UNICEF Mozambique 2

7 All targets for the year 2023 were set on the basis of a linear extension of 2020 targets, based on the assumption of a constant rate of progression between 2020 and 2023, as compared to the period 2016 to Of course, the 2023 targets were constrained by natural limits, so that, for instance rates of completion could not exceed 100% or drop-out could not go below 0%. For gender ratio indicators, the target limit was set at 50%. Units costs used to compute the total educational envelope were based on historical data provided in the original model. All figures presented in this analysis are presented in constant 2013 prices, not taking into account possible inflation or other price changes that may occur between 2013 and the target year of Social Protection The social protection costing model is more complex than the education model in that it combines a standard excel-based simulation of usage and costs with a micro-econometric estimation of coverage, based on an analysis of the Mozambique household budget survey dataset. The microeconometric model is used to estimate the total number of users based on pre-defined eligibility criteria (e.g. age, income, etc.). The coverage figures thus estimated through the microeconometric model are then inserted into the excel-based costing models to calculate the cost implications for the chosen coverage level. The social protection costing model is aligned with the 2011 Operational Plan, which sets out the fundamental principles for the reform of the social protection system in Mozambique. However, it should be noted that, unlike the education and health costings, the social protection costing does not hold any official status within the national planning process. It was generated by IMF and ILO with support from Oxford Policy Management, as a contribution to the discussion on social protection reform in Mozambique, illustrating the potential cost implications of various policy options. As such, they have not been officially endorsed by the Government of Mozambique and do not represent national policy. Furthermore, the current Basic Social Security Strategy expires in 2014 and a process has started to prepare the new one, due to start in We do not know yet what package of programmes will be included in the new strategy, and whether it will include detailed operational projections of coverage and costs. The main objective of the proposal for reform laid out in the 2011 Operational Plan is to move from a series of ad-hoc social protection intervention to a comprehensive and coordinated system of interventions supported by a coherent logic and unique routing (identification and targeting) so as to provide a complete coverage against the most common vulnerabilities and risks facing Mozambican households. On the basis of these overarching guidelines, the Operational Plan proposes to articulate the Basic Social Security system around four main programmes: Programa Subsídio Social Básico (PSSB) A monthly cash transfers for unlimited period targeted to households without any adult able to work (e.g. households headed by an elderly person, person with disabilities or chronically ill and without any able adult breadwinner). In practice the programme consists in a revision of the PSA with increased coverage and transfer amount, and a more accurate definition of eligible households. Programa de Apoio Social Directo (PASD) Divided into an ad-hoc component (Componente Pontual) and a long term-support component (Componente de Apoio Prolongado). It targets households who are temporarily lacking capacity to participate in the labor market (among others: child headed households; households headed by pregnant women; household with bread winners that are temporarily unable to work). It is also UNICEF Mozambique 3

8 conceived as a means to provide support to selected groups of vulnerable households (e.g. households with children suffering from acute malnutrition, households with malnourished members receiving ART) with residual labour capacity but living in areas where the PASP (see below) is not operational, or at times of the year when the PASP is not running. The PASD will provide mainly in-kind support. Social Assistance Services (Programa de Apoio às Unidades Sociais) Residential care and institutional support to vulnerable and abandoned children and elderly, victims of violence and the homeless who require intensive care services. Productive Social Action (PASP) A temporary transfer associated with the participation in a labour intensive public works initiative for poor and vulnerable households with residual labour capacity (who are not entitled to receive support from the PSSB and PASD). It also envisages a component to establish linkages with interventions supporting incomegenerating activities (run by other ministries). It should be noted that the costing model presented here does not include sector-specific social protection interventions in, say, health or education, which are still under discussion and will be costed in their respective sector costings. Furthermore, the simulations concentrate only on the non-contributory pillar of a Social Protection Floor for Mozambique, as the prospect of expansion of contributory social insurance is uncertain at the moment. The exercise introduces different solutions regarding the potential mechanisms to enhance the coverage expansion and already elicits potential operational issues regarding coverage expansion in the short to mid-term (namely, for the PASP). 2.3 Health The health sector costing is based on the Plano Estratégico do Sector da Saúde (PESS). The costing was developed by the Mozambiquan Ministry of Health (MISAU) in collaboration with the Futures Group Health Policy Project, using the OneHealth Tool (Duta et al., 2013). The OneHealth model is a tool for medium term (3 to 10 years) strategic planning in the health sector at the national level. OneHealth is integrated with the Spectrum suite of models, which includes demographic and epidemiological projections. Costing in the health sector is generally considered more complex than in other social sector, due to the cost structure of health services. Indeed, health expenditures are composed of both vertical costs related to specific programmes or interventions (e.g. vaccination, TB, HIV, etc.) and horizontal costs related to the operations of the health system as a whole, including all required levels of care and referral for different types of diseases. Because of this, the health sector costing is divided into three separate steps, as follows: 1. Model the scale-up of proposed interventions to meet specified coverage targets, 2. Check whether the health system as a whole (infrastructure, human resources, etc.) would be capable of coping with the proposed scale of interventions, and then model the required scale up of the health system. 3. Model the impact of the proposed interventions on specific health targets and common causes of mortality. We used the Lives Saved Tool (LiST), an internationally accepted modeling tool which has been used in peer-reviewed journal articles to analyze the impact on key health indicators from the PESS-related scale-up of various health interventions. UNICEF Mozambique 4

9 The scenarios presented in this chapter were developed using the parameters and assumptions agreed by partners in the course of the costing exercise carried out by the Ministry of health. This includes baseline and target coverage rates of different interventions, unit costs of different cost drivers, and population target groups. The costing is structured around the following 5 cost centers: - Programme costs: includes programmes falling under the Direcção Nacional de Saúde Pública (DNSP) (maternal health, malaria, tuberculosis, immunisations, and nutrition programmes), as well as programmes falling under the Direcção Nacional de Assistȇncia Médica (DNAM) (HIV treatment programme, as well as preventative measures, such as male circumcision. DNAM also covers surgery as well emergency and trauma units, amongst others). - Logistics, including wastage: includes all costs of warehouses, supply chain, and programmatic management of the central drugs and medical stores - Health Infrastructure: This covers major construction and rehabilitation projects for the health sector. - Human Resources for Health: The costs include salaries and benefits for all MISAU personnel, pre-service training costs, and costs of programmatic management at the human resources directorate. - Governance, Leadership and Information Systems: This component involves the programmatic costs of several national-level departments and affiliated autonomous institutions. 2.4 Water and Sanitation The water and sanitation analysis is based on the model developed by the National Directorate for Water (DNA) together with international partners as part of the Country Status Overview (CSO) of The CSO model uses a simple linear extrapolation methodology based on the assumption of linear progression of coverage rates from the baseline estimate to the target coverage rate. These coverage rates refer to percentages of households with access to improved water/ sanitation. As such, they do not include institutional WASH in, for instance, schools and health centers, as these should be covered within their respective sector costings. In the CSO model provided, the target coverage rates were set as the Water Policy targets for 2015, which were themselves based on MDG targets. The baseline coverage rates were provided by the National Water Directorate (DNA) estimates for It should be noted that the coverage rates estimated by the DNA used at the time were widely different (in some cases higher, in cases lower) from the ones provided by the National Statistics Institute (INE), based on the 2008 Multi- Indicator Cluster Survey (MICS). The differences are due to differences in definitions of various indicators, as well as the data collection method and assumptions made 2. Targets are provided for the following categories: 2 The INE data is survey data, which is based on answers collected from end-users, whereas the DNA relied, at the time, on administrative data on construction of water points, and latrines. The DNA assumed that each water point served 500, whereas survey data indicates that the true figure is closer to 300. The Government approved this new service level of 300 users per water point in 2012 for improved planning. In terms of definition of access to improved water, we note the UNICEF Mozambique 5

10 - Access to Improved water in urban areas - Access to improved water in rural areas - Access to improved sanitation in urban areas - Access to improved sanitation in rural areas For the purpose of the present exercise, we have extended the end year of the projection from 2015 to 2023 and updated the baseline figures from 2009 to 2011, using the latest available survey data from the DHS We have also changed the population projections to use the official 2007 population projections for the period Finally, we have set the exchange rate at 28.4 MT per USD in constant 2012 prices. Furthermore we have simplified assumptions regarding unit costs of various inputs, as we no longer differentiate between different types of water and sanitation provision within those considered as improved water sources and improved sanitation facilities. All unit costs used in this costing exercise were provided by UNICEF Mozambique s WASH section, and are best estimates of 2013 prices based on the current technology mix in urban and rural areas (see section Table 1 below for details on unit costs and coverage rates). For consistency with other sectors, these have been adjusted for inflation and are reported in 2012 constant prices. As the CSO model was computed using USD prices, we have converted the results into Meticais by using the 2012 exchange rate provided in the Education model in order to ensure comparability of our results across sectors. Table 1: Unit pries of water and sanitation facilities (per capita) Unit price* Water Urban 5,396 Rural 1,704 Sanitation Urban 5,396 Rural 1,022 *Average per capita average investment cost for specified type of installations, given the current technology mix in each area. All costs are assumed to be fixed in real terms across the projection period and expressed in constant 2012 prices. 2.5 Macro-Economic Assumptions All costing models presented below use the same set of fundamental assumptions regarding projections of GDP, population growth, inflation, etc. The key macro-economic assumptions used across all four models are summarised in Table 2 below: Table 2: Common macro-economic assumptions for different costing models Indicator Rate Source Real GDP growth (yearly, excluding 7% IMF projections coal and gas) 3 Inflation (yearly) 6% IMF projections Population growth (yearly) 2.6% Official census projections 2007 (2.2% in rural areas, 3.5% in urban areas) following differences, for instance: INE includes in their definition of access to improved water all households that have access to piped water, whereas the DNA does not consider them to have access if the piped system is owned and operated privately. Similarly, the INE considers all households with access to adequate toilet facilities as having improved sanitation, whereas the DNA does not consider this to be the case, if the household used shared facilities. 3 Average over the period , exact yearly figures are provided in Error! Reference source not found.. UNICEF Mozambique 6

11 Exchange rate MT/ USD (2012 average) World Development Indicators (World Bank) Figure 1 below provides a detailed breakdown by year of the key macro-economic variables that have been used in the costings below to ensure consistency across costing models. These figures are taken from the macro-projections carried out as part of this situation analysis, and are based on estimates produced by the IMF and the DNO. It clearly shows that the projections used for this analysis make very conservative estimates concerning possible future increases in public revenue. In particular, it excludes possible increases in public revenue stemming from discoveries of coal and gas, as these are considered too uncertain. These constraints should be kept in mind when analysing the results of the various costing scenarios explored below. Figure 1: Selected macro-economic indicators from macro-projections from 2013 to 2023 Source: DNO and IMF UNICEF Mozambique 7

12 3 Results 3.1 Education This subsection presents the results of the education costing exercise based on the two scenarios developed in the Education Sector Strategic Plan (ESSP). These two scenarios were modified along the lines discussed above in order to render them compatible with the parameters of the current exercise. In particular, the costing presented in this chapter stretches until 2023, whereas the costing presented in the ESSP covered the period 2010 to Scenario 1 was developed by the ministry of education, defining achievable objectives given projected increases in funding and population growth. Based on the assumptions used in the original education model, this would have represented 21% of total government expenditures until 2016, with TGE fixed at 33% of GDP. With the slightly modified macro-economic assumptions used in our model, this would represent 18% of TGE by Scenario 2 maintains the same macroeconomic parameters, but assumes a higher share of educational expenditures in the national budget. In the original costing model, these targets were set to match a projected education expenditure of 25% of TGE in In our modified model, education expenditures would rise gradually to reach 29% of TGE by In addition to these two scenarios, we developed a 3 rd scenario, which we call an aspirational scenario in the sense that it estimates the budget that would be required to achieve near universal primary school completion by the year 2023, with adequate attention to quality of education. Scenario 3 is characterised by the following policy choices: - A substantial increase in the number of children enrolled in public and community-run preschools (+50% per year from a very low base). This is based on recommendations formulated in the SitAn and is consistent with international evidence on the importance of pre-primary education to ensure that children are prepared for primary education by age 6. - A substantial increase in the number of primary school teachers trained each year (from 5000 to 20000) to reflect the need for more and better training. The unit cost of teacher training has been doubled from 2012 levels to reflect an improvement in quality and the incorporation of continuous in-service training 4. - A gradual increase in the capitation grant (Apoio Directo às Escolas, Direct Support to Schools) from $3.8 per student/ year in 2013 to $10 per student year by 2023 ($5 by 2016 in the original model). - An increase in lower secondary enrolment and transition to higher secondary. This reflects the fact that viable options for secondary education are needed to incentivise children to complete primary education. - A slight increase in teachers salaries (2.5% per year in real terms compared to 2% per year in the base scenario) and an increase in bonuses and localisation incentives. This is needed to allow better performance-based pay and to convince teachers to teach in remote areas. 4 Training for secondary school teachers is not included in this model, as that falls under the higher education ministry and is therefore not part of the education sector strategic plan. UNICEF Mozambique 8

13 - An improvement in the pupil /teacher ratio to 50. This reflects better quality of education and is consistent with the standard defined in the new education policy document for Mozambique. - An increase in the proportion of children receiving free school lunches to 20% to incentivise poor children to attend school. Under scenario 3, education expenditures would reach 35% of TGE by While this is higher than current levels of spending in Mozambique, we should remember that several African countries, such as Uganda, spend in excess of 30% of TGE on education (UNESCO, Institute for Statistics 2011). Also, it is important to note that these estimates are based on the very conservative projections of TGE formulated by the IMF, which exclude potential revenue from coal and gas. If these are taken into account, it is likely that even the most ambitious scenario formulated here could be attainable by Finally, it is important to note that even the most ambitious scenario considered here falls far short of international best practice and desirable quality standards. In particular, scenario 3 focuses primarily on increasing the quality of primary education and primary completion, but leaves higher secondary and tertiary education virtually unchanged. In order to achieve its development objectives, Mozambique will need to aim for a universal completion of 10 years basic education for all children. This highlights the magnitude of the challenges facing the country in terms of education, and also the need to prioritise objectives within the education sector to ensure that at least the basic elements of a universal education system are covered. However, the current model proceeds on the assumption of constant unit costs compared to current costs. As such, it does not take into account possible cost savings that could be realised by using innovative construction methods and other local solutions for generating educational materials. Such costs savings could potentially play an important role in lowering the cost per pupil in primary school, and could help to free up resources for secondary education and beyond. Table 3 below presents a list of selected targets used in each of the three scenarios for It should be noted that the total number of targets required to calibrate the model exceed 800, so the list below only provides a snapshot of some of the key targets that are driving the results. Table 3: Selected 2016 and 2023 targets, by Scenario Target Year: Baseline Scenario 1 Scenario 2 Scenario Primary Education EP1 Age 6 intake rate (Grade 1 NIR) 70% 85% 85% 90% All Ages intake rate (Grade 1 GER) 137% 103% 124% 104% Transition rate EP1 to EP2 62% 75% 75% 90% Drop-out Rate: EP1 to EP2 (Grade 5 to Grade 6) 23% 10% 10% 5% Repetition Rate: EP1 to EP2 (Grade 5 to Grade 6) 15% 15% 15% 5% Gross Primary Completion Rate at Grade 7 (graduates) 46% 51% 78% 92% EP1: Pupil: teacher ratio (TCR) in public school Pupil: Classroom Ratio (PCR) in public primary schools (same for EP1 and EP2) Target proportion of primary pupils receiving school lunches 1.7% 5% 10% 20% UNICEF Mozambique 9

14 Secondary Education Transition rate EP2 to ES1 50% 70% 75% 90% Gross Enrolment Ratio in ES1 46% 55% 79% 96% Transition rate ES1 to ES2 45% 54% 65% 80% Drop-out rate ES1 to ES2 (Grade 10-11) 31% 35% 24% 10% Repetition Rate: ES1 to ES2 (Grade 10 to Grade 11) 25% 11% 11% 10% Gross Secondary Completion Rate at Grade 12 4% 12% 16% 24% (graduates) ES1 day school: Pupil: teacher ratio (PCR) in public school Pupil: Classroom Ratio (PCR) in public secondary schools (same for ES1 and ES2) The projected growth over the coming years, in terms of number of students, results primarily from the increased internal efficiency of the system. Each year, more students carry over to the next class due to reduced rates of repetition and drop-out. Scenario 1 provides for maintenance of transition rates between different levels of education (with the exception of ESG2), while Scenario 2 envisages an increased rate of transition between the 7th and 8th grade, as well as between the 10th and 11th class (see Table 12). Scenario 2 also provides a more rapid growth levels of postprimary education, literacy and adult education and pre-school, while scenario 3 aims to achieve 90% intake rates at primary level by Based on these targets, the model then simulates the way that students progress through the school system, based on a cohort analysis approach. Figure 2 below presents the estimated trajectories of gross enrolment and completion rates for primary and secondary education over the period for each of the scenarios described above. As can be seen in Figure 2, in all scenarios it is expected that the gross primary enrolment ratio will gradually decrease from 2015 onwards, to approach 100% by This is due to the fact that a large proportion of previously excluded over-age children will be enrolling in the earlier years, as the system catches up towards universal primary education. As the system matures, it is expected that more children will be starting school on time and will progress normally through the system, which reduces the number of over-age children in the system and thus gross enrolment ratios. Gross primary and secondary completion rates will continue to increase gradually over the entire duration of the projection from low baseline levels, albeit at decreasing rates in scenarios 1 and 2. Even under the most optimistic scenario (Scenario 3), the gross completion rate for grade 12 will only reach 24% by UNICEF Mozambique 10

15 Figure 2: Gross Enrolment / Completion Rates ( ), by scenario The costs of the education sector are primarily determined by the number of students in the system, which determine the number of teachers, schools, classrooms, books, teaching materials necessary to ensure the provision of services education through the targets set for the indicators of access and quality as well as institutional capacity. All scenarios envisage an improvement both in the quantity of students completing a full schooling cycle and in the quality of the education provided, as measured by available indicators, such as the pupil / teacher ratio, etc. The projected cost implications of the proposed targets are presented in Figure 3 below (a detailed breakdown of yearly costs by level and type of expenditures is presented in Annex B.1 below). The total projected budgetary envelope required to achieve the targets defined in scenario 3 would reach MT 89 billion by 2023 or 11% of GDP, compared to MT 47 billion (6% of GDP) under the least ambitious scenario (scenario 1). This compares to a projected cost of education equivalent to 7% of GDP in 2013 (see Table 4 below). UNICEF Mozambique 11

16 Figure 3: Projected total cost of education ( ) under scenarios 1,2,3 ( 000 MT), by year Source: Author s calculations based on education model Figure 4 shows the projected breakdown of education costs by level. The table shows that the projected distribution of resources across the various levels is fairly stable across scenarios, with the share of education resources going to primary education ranging between 37% and 40% of total education spending, while secondary education gets between 29% and 31% of total education resources. UNICEF Mozambique 12

17 Figure 4: Projected total cost of education in 2023 (in million MT ), by level Source: Author s calculations based on education model Table 4 below shows how these costs break down by activity and type of expenditures. The table shows that the most important cost drivers in education are salaries and remuneration of staff. The distribution of costs between different types of expenditures is projected to be fairly constant over time, although there will be a small increase in the proportion of expenditures going to social action and prevention and mitigation of HIV/ AIDS, whereas the share of expenditures going to institutional function is projected to decrease slightly over the decade. The table also highlights the extremely small portion of expenditures allocated to teacher training and capacitation. This was identified as one of the major weaknesses in the Situation Analysis. Yet, the projected increases in resources allocated to teacher training are very small under all scenarios developed in the education sector strategic plan. Table 4: Projected total cost of education in 2023 (in million MT ), by activity and economic classification Scenario 1 Scenario 2 Scenario 3 COST SUMMARY BY ACTIVITY SALARIES AND REMUNERATION 8, , ,061.7 CONSTRUCTION, REFURBISHMENT AND 37, , ,319.1 MAINTENANCE SCHOOL BOOKS, TEACHING MATERIALS AND 1, , ,505.9 EQUIPMT. TEACHER TRAINING AND CAPACITATION OTHER TOTAL 47, , ,995.8 COST SUMMARY BY ECONOMIC CLASSIFICATION 46, , RECURRENT EXPENDITURE 37, , ,399.7 Of which: Personnel 1, , ,319.1 UNICEF Mozambique 13

18 Of which: Goods and services 6, , ,233.8 Of which: Recurrent transfers 1, , ,846.8 CAPITAL EXPENDITURE 8, , ,596.1 TOTAL 47, , ,995.8 TOTAL (as % of TGE) 18.4% 29.4% 34.5% TOTAL (as % of GDP) 5.8% 9.3% 11.0% Source: Author s calculations based on education model 3.2 Social Protection This subsection presents the results of the social protection costing exercise carried out by the ILO in collaboration with the IMF and OPM in The costing aimed to explore the technical and financial arguments underlying the policy options presented in the Strategic Operational Plan for basic social security in Mozambique. Some minor modifications have been made to the scenarios presented in the original costing, based on the parameters of the present exercise (see section 2.5 above). Furthermore, the scenarios have been updated to reflect the evolution of the policy dialogue around the issue of social protection in Mozambique, as expressed in discussions held in Maputo with national and international partners. As part of this exercise, we focus on the following three scenarios: 1. A baseline scenario corresponding to a continuation of current policies. In this scenario, the PASP would target the bottom 15% poorest households in rural areas and the poorest 25% in urban areas (other programmes target the bottom 80% in eligible categories). This scenario assumes a gradually increasing take-up rate for the PASP programme rising to 29% of eligible beneficiaries in rural areas and 38% in urban areas by In this scenario, the PSSB programme targets only totally labour constrained households that do not have a single ablebodied household member. 2. An intermediate scenario with higher levels of transfers for PASP in urban areas (903 MT per beneficiary, compared to 723 MT in rural areas). In this scenario no poverty targeting is applied. Instead the selection of beneficiaries relies on self-selection, based on the assumption that individuals will only voluntarily participate in the programme if the wage rate is at least equal to their current income level (estimated as the total household consumption / number of able-bodied household members) 5. However, expansion of programme coverage is limited in the initial years by capacity and administrative constraints. Under this scenario, the selection criteria for the PSSB programme are slightly relaxed so that non-labour constrained 5 Note that the programme is seasonal, running 4 times a year in rural areas, and 6 times a year in urban areas. Our model assumes that workers have complete flexibility to switch between their normal occupation and the public works programme if it is beneficial to them (i.e. if they can earn more in PW than in their regular occupation) in the months when the programme is running. This may be true if the programme is running in the lean season in rural areas or if participants are employed as casual day-labourers. However, there may be others who are unable to switch work for one month, even if they could earn a higher wage. For this reason, the actual take-up rate may be lower than the one estimated in this model. To estimate the actual elasticity of labour supply, tailored survey estimates would be required. UNICEF Mozambique 14

19 households may be eligible as long as fewer than one in every five household members are able-bodied. 3. A more ambitious scenario, which would require a review of existing targeting mechanisms: In this scenario, the PSSB programme would be replaced by a social pension, while the PASD would be phased out in favour of a child benefit for children under 5, covering the poorest 40% of households. In this scenario, the PASP wage rate is increase to 903 MT per beneficiary in rural and urban areas, leading to an increase in the take-up rates. As in scenario 2, the coverage of the PASP programme will increase gradually to allow for a scale up of administrative and supervision capacity. For the purpose of the present costing exercise, these scenarios have been extended to cover the period The main characteristics of the different options in terms of coverage and types of interventions are summarised in Table 5 below: Table 5: Selected 2023 targets, by Scenario Target Baseline Scen. 1 Scen. 2 Scen. 3 PSSB / Social Pension Eligible Deciles 8 poorest 8 poorest 4 poorest Eligibility criteria No ablebodied Less than 1 Any labour in in 5 able - constrained Household bodied individual Total number of beneficiary households by , , , ,538 Transfer amount per primary beneficiary in (constant 2012 prices) Periodicity Monthly Monthly Monthly Monthly Admin cost by 2016 through % 20% 20% 20% PASD Eligible Deciles Eligibility criteria Child-headed, acutely malnourished - children, temporarily unable to work Total number of beneficiary households by , , ,261 0 Average transfer amount per primary beneficiary/ month in 2023*** Periodicity Child-headed: monthly; Temporary: quarterly; Malnourished: bi-monthly Admin cost by 2016 through % 20% 20% 0 Public Works (PASP) Eligible deciles 1.5 (rural) All (selfselection All (self- /2.5 (urb.) 6 ) selection 6 ) Eligibility criteria Able-bodied (not receiving PSSB) Total number of beneficiary households by , , , ,816 6 In the self-selection model, we assume that able-bodied will only take up the public works component if they can earn more than their current level of income (estimated as the total household consumption / total number of working members). UNICEF Mozambique 15

20 Average transfer amount per primary beneficiary / month in 2023*** Rural: 655; Urban: 818 Rural: 723; Urban: 723 Rural: 723; Urban: 903 Periodicity Rural: quarterly; Urban: bi-monthly Admin cost by 2014 through % 30% 30% 30% Rural: 903; Urban: 903 Child Benefit Eligible deciles Eligibility criteria Children under 5 Total number of beneficiary households by ,890,318 Average transfer amount per primary beneficiary/ month in 2023*** Periodicity Monthly Monthly Monthly Monthly Admin cost by 2016 through % % In scenario 1, the main coverage increase would take place in the PASP after 2016, in the form of increased take-up. Indeed, on the basis of the current reform set-up, this is the program with the major potential for coverage expansion, reaching 383,013 beneficiary households in Scenario 2 estimates what would be the fiscal space needed for a more ambitious expansion plan. The simulation is based on the assumption that the transfer amount of PSSB and PASP will be increased. In scenario 2, the PSSB base transfer amount is raised to 2/3 of the poverty line with an add-on of 25% for each additional household member (up to a maximum of 4 dependents) in order to better reflect the demographic composition of the household. As regards the PASP, the increase in coverage is achieved by increasing the transfer amount by 25% in urban areas to reflect differences in living cost differentials. This would make the programme more attractive in urban areas and thus lead to an increase in the urban take-up rate. Scenario 3 explores the possibility of introducing alternative targeting criteria after 2014 in line with recommendations formulated in the World Bank s Mozambique Social Protection Assessment (WB, 2012). The World Bank recommends articulating the social protection floor around two main pillars: Social Pension (Pensão Social) to focus on the poor elderly, persons with disabilities and chronically ill. These correspond to categories already covered by PSSB, but the new programme would cover all households with members in such state (regardless of the fact that other members may be able to work). Child Benefit (Apoio às Crianças) to focus on poor households with children under 5 (particularly partially labour constrained households, such as those with a single earner, high dependency ratio, partial or temporary incapacities, etc.) or child headed households. This component will also include a part of the current target groups of the PASD. Scenario 3 presents the minimum reforms necessary to provide an adequate coverage to households in the bottom 4 consumption deciles according to the World Bank s recommendations. The need to consider new design alternatives stems from the recognition that, from an operational perspective, it might be difficult for the Government to implement a very large-scale public-works scheme under the PASP in order to build a comprehensive social protection floor for non-labour constrained households. UNICEF Mozambique 16

21 Figure 5: Number of beneficiary households ( ), by scenario and programme Scenario 1 Scenario 2 Scenario 3 Legend The projected cost implications of these various policy alternatives are presented in Figure 6 below. Under scenario 1, costs would increase slowly to around MT 5 billion per year by 2023 (in constant 2012 prices). In scenario 3, by contrast, costs would increase rapidly from around MT 2 billion in 2013 to more than MT 18 billion in 2023 in constant prices. Scenario 2 offers a more gradual progression from MT 4 billion in 2013, stabilising at around MT 10 billion in UNICEF Mozambique 17

22 Figure 6: Projected total cost of social protection over the period , by scenario Source: Author s calculations based on ILO (2012) Table 6 below shows how these costs would be distributed over the different types of transfers in 2023, based on the predictions of the costing model (detailed breakdowns of costs by transfer type and year are presented in Annex B.2 below). Under all scenarios analysed here, the projected cost of social protection would remain modest by international standards, and compared to the cost of other sectors. The most ambitious scenario explored here is scenario 3, which still falls short of fulfilling the requirements for a basic social protection floor, as defined by the ILO. Under this scenario, the total cost social protection would represent just 2.3% of GDP under existing macroeconomic projections (excluding coal and gas revenues). This is still less than the cost of the South African Child Grant and Social Pension, which represent 3.2% of GDP, and it is less than one third of even the least ambitious scenario for education financing presented in Table 4 above. The least costly scenario is scenario 1, which would cost just 0.6% of GDP by 2023, or 1.8% of TGE. However, in this scenario, the proposed coverage and transfer amounts of the PASP and PSSB programmes would be insufficient to provide adequate social protection coverage to the poorest sections of the population, or to have a significant impact on poverty and vulnerability. Yet, even under the most modest scenario explored here, public spending on social protection would need to more than double compared to the latest available figures, which estimate that social protection expenditures represented less than 1% of TGE in Table 6: Projected total cost of social protection in 2023 (in million MT), by scenario Scenario 1 Scenario 2 Scenario 3 PSSB 1,841 5,907 5,958 Elderly 1,777 Not specified 0 UNICEF Mozambique 18

23 Social Pension 0 0 5,958 Disabled 63 Not specified 0 PASD 1,052 1,052 0 Temporary Child headed Malnourished PASP 1,744 3,090 4,321 Rural 685 1,753 2,664 Urban 1,059 1,337 1,657 Child Benefit 0 0 8,550 TOTAL 4,638 10,051 18,831 TOTAL (as % of GDP) 0.6% 1.2% 2.3% TOTAL (as % of TGE) 1.8% 3.9% 7.3% Source: Author s calculations based on ILO (2012) 3.3 Health This section presents the results of the health costing, which was carried out using WHO s One Health Tool. The results presented in this section build on the costing exercise that was carried out by the ministry of health with the support of the Health Policy Project of the Futures Group in support of the national health strategy (PESS). In the costing carried out for the PESS, only one scenario was developed, setting out the inputs required to achieve objectives defined in the PESS. This scenario could be called an aspirational scenario in that it did not take into account existing financing or capacity constraints, but started from pre-determined policy objectives. As a result, the costing presented in the PESS required a significant increase in public funding for the health sector from the start, compared to current levels of funding. In addition to the one costing scenario, three possible financing scenarios were developed in the PESS, representing expected levels of funding available for the health sector under conservative, moderate and optimistic assumptions, respectively. However, these financing scenarios were not linked to the costing in the sense that no estimate was made of the way in which different levels of funding would affect the achievement of PESS objectives. In the present exercise, we have started from the three financing scenarios presented in the PESS and attempted to work out the implications of various funding levels for the coverage of various health programmes. This has been done in the simplest possible way, by adjusting the coverage rates of all PESS programmes, as well as corresponding inputs of human resources, infrastructure, etc., upwards or downwards proportionally to available funding. As such, our scenarios have not been developed through a consultative exercise of re-prioritisation with relevant stakeholders to identify, for instance, the most crucial interventions that should be safe-guarded in the case of diminished funding. For this reason, the coverage rates presented in Table 7 below should not be seen as suggestive of recommended adjustments to PESS targets under various funding scenarios, but as illustrative of the possible implications of various funding options for the achievement of PESS targets, given existing priorities. As can be seen in Table 7, only the ambitious funding scenario would enable Mozambique to achieve universal coverage for all of the selected health interventions listed below. Under the conservative scenario, scenario 1, many of the selected indicators would only see very modest improvements by 2023, compared to current coverage level. Under this scenario it is highly unlikely UNICEF Mozambique 19

24 that Mozambique would be able to achieve its stated development objectives in terms of reduction of child and maternal mortality, for instance 7. Table 7: Selected health interventions coverage targets for 2023 (% of target population), by scenario Baseline 2012 Scenario 1 Scenario 2 Scenario 3 Maternal and Child Health ORS treatment for diarrhoea Treatment for severe diarrhoea Basic antenatal care Preventive postnatal care Vaccinations Measles Pneumococcal vaccine Polio BCG Tetanus (pregnant women) Malaria IPT LLIN for pregnant women TB First-line TB treatment for children HIV/ AIDS Paediatric ART PMTCT Nutrition Deworming (children months) Management of severely malnourished children Source: Author s calculations based on One Health Tool projections Figure 7 below shows the projected health costs over the period , based on the predictions generated by the One Health costing tool for each of the three scenarios. Under the moderate scenario developed by the Ministry of Health, the total funding available for the health 7 While the One Health Tool does allow for estimations of impact of various coverage levels of health outcome targets, including child and maternal health, we have chosen not to report these here, as they remain subject to discussion and uncertainty. In particular, the estimated link between specified interventions and relevant mortality rates, are based on historical estimates of effectiveness of interventions in international studies. As such, they do not take into account country-specific constraints, nor do they consider possible changes in the effectiveness of interventions that may take place over the projected period of the costing. UNICEF Mozambique 20

25 sector would amount to MT 42 million by 2023, which would be equivalent to 5% of GDP and 16% of TGE. Under the most ambitious scenario, the cost increases to 7% of GDP and 21% of TGE, which is comparable to current funding levels in several OECD countries, but significantly higher than current funding levels in Mozambique and other countries in Sub-Saharan Africa. Even under the most conservative funding scenario explored here, public health funding would need to increase significantly from less than 10% of TGE in 2012 to close to 15% in Figure 7: Projected health costs for (in Million MT), by scenario Source: Author s calculations based on One Health Tool Table 8 below breaks down the projected health cost in 2023 by type of expenditures and care level. This breakdown shows that the two major cost drivers are human resources (accounting for between 31% and 47% of total health expenditures), and medicines and supplies (38%-49% of total health expenditures). The distribution of health costs across care levels is fairly stable in the various scenarios, with the largest attributable share going to outreach care (between 26% and 28% of total expenditures). 8 The apparent decreases in funding between 2017 and 2018 in all scenarios is due to the frontloading of infrastructure investments in the PESS costing, which envisages that most of the new building requirements will be completed by This phenomenon gest accentuated in the more optimistic scenarios due to the up-scaling method used in our calculation. UNICEF Mozambique 21

26 Table 8: Projected health costs in 2023 (in million MT), by expenditure type and care level Scenario 1 Scenario 2 Scenario 3 Costing by expenditure type Programme Costs 2, , ,703.7 Human Resources 11, , ,773.0 Infrastructure 3, , ,957.4 Logistics ,022.2 Medicines, commodities, and supplies 18, , ,775.2 Governance GRAND TOTAL 37, , ,693.6 Costing by level Community Outreach 10, , ,396.2 Clinic 1, , ,896.5 Hospital 6, , ,855.0 National programme 3, , ,811.1 Unattributable 14, , ,310.1 GRAND TOTAL 37, , ,693.6 TOTAL (as% of 2023 GDP) 4.6% 5.1% 6.7% TOTAL (as% of 2023 TGE) 14.5% 16.2% 21.2% Source: Author s calculations based on One Health Tool UNICEF Mozambique 22

27 3.4 Water and Sanitation This section presents the results of the water and sanitation costing produced based on the model elaborated by the National Directorate of Water in preparation for the Country Status Overview of The model has been modified along the lines discussed in the methodological section in order to render it comparable to the costing estimates of other sectors included in this study Projections (population and coverage) For the projections, we have considered the following 3 scenarios: - Scenario 1 (Baseline scenario): This scenario assumes that the rate of progress observed over the period 1990 to 2011 will continue at the same pace until In other words, this scenario does not anticipate that there will be a substantial acceleration of the current rate of public investment in the water and sanitation sector, nor a significant increase in efficiency in implementation (e.g. more beneficiaries with the same investments). - Scenario 2 (Intermediate scenario): This scenario is based on the targets for 2015 as defined in the existing water policy. The target defined in this policy has been revised upwards for urban water coverage (from 70% to 92%) to reflect the fact that the model has been extended from 2015 to However, other targets (rural water and sanitation and urban sanitation) have not been changed from their original 2015 levels, as it is not anticipated that the targets defined in the policy document will be achieved by Scenario 3 (Universal Coverage): This scenario assumes that universal coverage will be achieved for both improved water and sanitation in rural and urban areas across the country. This scenario is considered aspirational in the sense that it estimates the resources required to achieve an ambitious policy objective, independently of existing budgetary or technical constraints. It should be noted that the water policy aimed for universal coverage to be achieved by Figure 8 below, presents the targets used in our projections for the three scenarios described above. Figure 8: Targets for access to water and sanitation Water : Percent of households with access to improved water facilities ( ), by scenario Sanitation: Percent of households with access to improved water facilities ( ), by scenario UNICEF Mozambique 23

28 In Table 9, the targets have been broken down by urban and rural area. Under the least ambitious scenario, scenario 1, it is anticipated that there will continue to be a substantial gap in access to water and sanitation by 2023, with only 46% or people in rural areas having access to clean water compared to 92% in urban areas. For sanitation, the gap would be even larger with only 19% access in rural areas, compared to 56% in urban areas. Table 9: Targets for access to improved water and sanitation in 2023, by urban/ rural 9 Baseline Scenario 1 Scenario 2 Scenario 3 (2011) Access to improved water source Rural 38% 46% 70% 100% Urban 85% 92% 92% 100% National 53% 64% 78% 100% Additional population covered ( 000,000 users): Access to improved sanitation Rural 13% 19% 60% 100% Urban 48% 56% 67% 100% National 24% 34% 63% 100% Additional population covered ( 000,000 users): Source: Author s calculations based on Country Status Overview model It is important to note that the targets defined in Table 9 are based on an assessment of needs rather than an assessment of technical and financial feasibility. In particular, the more ambitious targets do not take into account possible capacity constraints in the WASH sector, particularly HR capacity, which may prevent a rapid upgrade of public investment in the sector. In practice, investments may therefore need to be rear-loaded to allow for an upgrade of human resource and systems capacity in the early years to deal with the increased workload implied by scenarios 2 and 3 (at the moment, all scenarios envisage a linear scale up of physical infrastructure between 2013 and 2023) Type of Costs The three main types of costs considered in this model are: - New capital requirement: this represents the cost required to extend the coverage from the baseline coverage to the target coverage, given assumptions of population growth and users per facility 10. Unit costs for new capital requirements are expressed as cost per capita, meaning that assumptions about number of users per facility are built into the unit costs estimates (see Table 10 below for a detailed description of unit costs by type of facility). It is important to note that the unit cost estimates used are based on fixed assumptions regarding technology mix in urban/rural areas. In practice, some rural areas might go for urban solutions (e.g. mini-systems in small towns), while some peri-urban areas may prefer rural technologies. 9 Number of users covered are based on the following assumptions regarding urban and rural population projections for 2023, taken from the Census 2007 population projections: Urban: 20.7 million; Urban: 10.8 million. 10 Assumptions about population growth have been aligned with census projections from the 2007 census (see Table 2). Assumptions about users per facility are inbuilt in the CSO model. UNICEF Mozambique 24

29 - Maintenance costs: this represents the cost for managing the facility, including administration costs related to user management, and small ongoing repairs and rehabilitation of existing equipment. In this model, we have assumed that maintenance costs would represent 25% of capital costs across all types of investments. - Replacement cost: this represents the cost for replacing facilities at the end of their life span in order to ensure that the coverage rate does not decrease over time as existing facilities become obsolete. In this model, it has been assumed that the life span of water and sanitation facilities in rural areas is 20 and 10 years, respectively, whereas water investments made in urban areas (piping) are expected to last for 40 years and sanitation facilities for 20 years. The replacement cost is estimated based on the value of the existing stock of capital, assuming a depreciation of the value that is proportional to the lifespan of the equipment. In the final calculations of public expenditures required for the water and sanitation sector, we have included a 3% national overhead for planning and coordination, which is calculated on the total public investment cost, including capital replacement costs, but excluding maintenance costs. This overhead was not included in the original Country Status Overview model, but is consistent with the current spending patterns in the water and sanitation sector, where investment costs account for 97% of total public expenditures in this sector, while maintenance costs are not budgeted for. In estimating the public investment required in WASH, the following assumptions are made regarding the user share of investment cost for: - Rural water: users pay 2% of total cost out-of-pocket, including maintenance and replacement, but excluding national overhead costs. Public expenditures on rural water thus represent 98% of total investment costs. - Urban water: users 2% of total cost out-of-pocket, including maintenance and replacement, but excluding national overhead costs. Public expenditures on urban water thus represent 98% of total investment costs. - Rural sanitation: users pay 17% of total cost out-of-pocket, including maintenance and replacement, but excluding national overhead costs. Public expenditures on rural sanitation thus represent 83% of total investment costs. - Urban sanitation: users pay 37.5% of total cost out-of-pocket, including maintenance and replacement, but excluding national overhead costs. Public expenditures on rural water thus represent 62.5% of total investment costs. These assumptions are based on current and anticipated financing patterns listed in the CSO report. For instance, hand pump users currently are asked to make an up-front capital cost contribution of approximately 2% of the total installation cost. Once this payment is made, no additional capital cost recovery is required (for more details, see CSO 2010) Annual Public Expenditure Figure 9 below presents the estimated annual public expenditure required to achieve the targets defined in Table 9, under each of the three scenarios. UNICEF Mozambique 25

30 Figure 9: Public expenditure on water and sanitation (in million MT), by year ( ) Source: Author s calculations based on country status overview model. In Table 10 below, we have represented the breakdown of public expenditures for the year 2023 (yearly breakdown of costs for the period is provided in annex B.4 below). The total public investment required to achieve the above stated targets would range from MT 8.6 billion in 2023 (or 1.1% of GDP) under the least ambitious scenario to MT 20.4 billion (2.5% of GDP) under scenario 3. Even under the most ambitious scenarios explored here, the investments required for the water and sanitation sector would remain achievable when compared to current levels of spending on water and sanitation. Indeed, under scenario 3, the estimated yearly investment required would increase to 7.9% of Total Government Expenditures (TGE), and 5.1% of TGE under scenario 2, compared to 4.6% in 2012 (see first chapter of the SitAn). Under scenario 1, more than 56% of the investments are targeted at improving access to clean water, whereas scenarios 2 and 3 put more emphasis on sanitation with 47% and 41% going to water. Similarly, scenarios 2 and 3 put more emphasis on investments in rural areas (41% and 44% of total, respectively) compared to scenario 1 in which on 24% of investments go to rural areas. This is largely due to the difference in cost between urban and rural infrastructure (see Table 10 below). UNICEF Mozambique 26