Efficiency of public expenditure distribution and beyond : a report on Ghana's 2000 public expenditure tracking survey in the sectors of primary health and education

Africa Region Working Paper Series 2463 1 Number 31 Efficiency of Public Expenditure Distribution anid 3eyorid: A Report on Ghana's 2000 Public Expenditure Tracking 5urvey in the 5ectors of Primary Health and Education Xiao Ye Sucharshan Canagarajah June 2002 i-__ , , LW2  Efficiency of Public Expenditure Distribution and Beyond: A Report on Ghana's 2000 Public Expenditure Tracking Survey in the Sectors of Primary Health and Education Africa Region Working Paper Series No. 31 Abstract This paper analyzes the public expenditure flows from the line ministries in Ghana to the basic service provision facilities, including primary and junior secondary schools, and health clinics. The study is mainly based on the data collected from a pilot public expenditure tracking survey (PETS) in 2000. A review of public expenditure distribution systems shows that the recording procedures for salary and non- salary expenditures are different. Salary expenditure is recorded in terms of cedis, from Ministry of Finance down to the public employees, who receive the salaries. Non-salary expenditures, although recorded in terms of cedis at the line ministry level, are largely distributed to the district offices and facilities in form of materials, with no monetary values indicated. This disconnect in the recording systems gives little local accountability of the higher level offices and provides few opportunities for the facilities to give feed back on their needs in resource allocation. The results from PETS data indicates that only about 20 percent of non-salary public health expenditure and 50 percent of non-salary public education expenditure reached the facilities. In health sector, evidences suggest that a large proportion of the leakage occurred between the line ministries and the district offices, where the public expenditures are turned into materials from cash flows. Based on our analysis of the tracking survey data, we conclude that a consistent and transparent recording system from the line ministries to the service provision facilities may significantly improve the efficiency of public resource distribution by providing easy public access to the resource flow data. Authors' Affiliation and Sponsorship Xiao Ye Research Analyst, AFRCE Email: xye@worldbank.org Sudharshan Canagarajah Senior Economist, AFTP2 Scanagaraiah(worldbank.org THE WORKING PAPER SERIES The Africa Region Working Paper Series expedites dissemination of applied research and policy studies with potential for improving economic performance and social conditions in Sub-Saharan Africa. The Series publishes papers at preliminary stages to stimulate timely discussion within the Region and among client countries, donors, and the policy research community. The editorial board for the Series consists of representatives from professional Families appointed by the Region's Sector Directors. Managing editor of the series: Paula J. White, AFTM3, Email: pwhite2@worldbank.org, who may be contacted for hard copies. For additional information visit the Web site http://www.worldbank.ora/afr/wps/index.htm, where copies are available in pdf format. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s). They do not necessarily represent the views of the World Bank Group, its Executive Directors, or the countries that they represent and should not be attributed to them.  Efficiency of Public Expenditure Distribution and Beyond: A Report on Ghana 's 2000 Public Expenditure Tracking Survey in the Sectors of Primary Health and Education Xiao Ye Sudharshan Canagarajah June 2002 Acknowledgeme:-nt :Financial support from the Ghana Country Department and Dtch Tr Fudfo carrying out this study are .gratefully acknowledged.. .The authors would-'lihke to- .a-cknowledge-the field work canied.out -byMir. M W,iliams anidlhis team. The authors would'also like to.thank:the invaluable support provided 'by:Gbhana-Field'Office and'the ,comments;provided -by K. Yoshid, 'R. Reinika, 3. .Dehn, and Benoit Milhiot -' '':.:. ': , . , '' --. . . :9 : --,.:J   TABLE OF CONTENTS I. INTRODUCTION ...................................................................... 1 H. THE IMPLEMENTATION OF PETS IN THE SECTORS OF EDUCATION AND HEALTH ..............3 The systems ofpublic expenditure distribution ..................................................................... 3 Implementation of the PETS ...................................................................... 6 HI. PRIMARY EDUCATION AND PUBLIC RESOURCE FLOWS ................................................. . 7 Study environment .................................................... .. 7 Academic performance and gender bias .................................................... . 9 Resourceflows to primary schools .................................................... . 10 Reasons for not attending school and ways to improve primary education . .......................................... 12 IV. JUNIOR SECONDARY SCHOOL EDUCATION AND PUBLIC RESOURCE FLOWS ... 15 Study environment ...................................................................... 15 Academic performance and gender bias ..................................................................... 17 Resourceflows to JSS ............... 19 Reasons for not attending school and ways to improve JSS education . ................................................. 20 V. PRIMARY HEALTH CARE AND PUBLIC EXPENDITURES . .................................. 22 Health facilities and service provision .................................................. . 22 Public health expenditures and internally generatedfunds . . ............................. 24 Improvements of health facilities ................................................... 25 VI. CONCLUDING REMARKS ............................... . ...... . ... 26 FIGURES Figure 1: Ghana: Flow of Basic Education-Related Public Funds ............................................ 4 Figure 2: Ghana: Flow of Health Care Related Public Funds ............................................ 5 TABLES Table 1: Distribution of enumeration areas by type and region ............................................ 6 Table 2: Primary school learning enviromnent by region, 1997/98 and 1998/99 ......................................... 8 Table 3: Sources of drinling water ........................................................... 9 Table 4: Primary school academic performance by region ........................................................... 10 Table 5: Primary school recurrent expenditure (cedis), per enrolled student .............................................. 11 Table 6: Components of primary school non-salary expenditures, per enrolled student ............................ 12 Table 7: Reasons for not attending school ........................................................... 13 Table 8: Improvements of primary schools ........................................................... 14 Table 9: Suggested ways to improve access to education and the quality of education ............................ 15 Table 10: JSS learning environment by region, 1997/98 and 1998/99 ...................................................... 16 Table 11: JSS drinking water supply and storage ........................................................... 17 Table 12: JSS academic performance by region ........................................................... 18 Table 13: JSS government recurrent expenditure (cedis), per enrolled student .......................................... 19 Table 14: Components of JSS non-salary expenditures (cedis), per enrolled student ................................ 20 Table 15: Reasons for not attending school ........................................................... 21 Table 16: Improvements of JSS ........................................................... 21 Table 17: Suggested ways to improve access to JSS and the quality of education .................................... 22 Table 18: Population per health care worker at primary health care level ................................................. 22 Table 19: Basic conditions of health facilities ........................................................... 23 i  Table 20: Number of consulting rooms in clinics, % of clinics .................................................. 23 Table 21: Utilization of the clinic services ............................ 24 Table 22: Public non-salary recurrent expenditure flow to health clinics .................................................. 25 Table 23: A comparison of public financing at ministry and facility level ................................................ 25 Table 24: Improvements of health facilities and medical supplies .................................................. 26 Table 25: Satisfaction rate of health services .................................................. 26  I. INTRODUCTION In its latest version of "Ghana Poverty Reduction Strategy" (GPRS, by the Ministry of Finance, February 20, 2002), the Government of Ghana (GOG) stated that "significant gaps exist in access to and utilization of basic services by the poor", and the major issues with regard to basic service provisions are "quality, equity, efficiency and financing gaps". The GOG has committed to "enhancing the delivery of social services to ensure location equity and quality". To achieve this objective, The GOG not only has increased its budget share for health and education sectors, it also has increased the share of basic services within the sector budgets. Higher budget expenditures, however, are only one part of the story. Whether a higher budget translates into better performance in the invested sector is, perhaps, the more important part of the story. In order to achieve the desired outcome, it is crucial to ensure that the resources allocated to social services are distributed efficiently to the public service provision facilities such as schools and health facilities, and that these facilities reach the service users, especially the poor. There are two broad categories of reasons that might cause inefficiencies in the distribution of public expenditures. The first one is corruption, namely leakage of resources to individuals or unintended organizations. The second one is wasting, namely inefficient use of resources, such as a mismatch between what facilities need and what the government distributes. The GOG keenly recognizes the problems and the consequences of corruption. It has established a Serious Fraud Office (SFO) for the purpose of combating corruption. According to the SFO's annual report of 1999, it discovered problems with regard to ghost names that appeared on government payrolls, especially in the education and health sectors, and serious fraud in public procurement. In supporting the GOG's effort in streamlining public spending and improving related outcome in social sectors, the focus of the Ghana Public Expenditure Tracking Survey (PETS) is to measure the first type of inefficiency, the leakage between different points of resource distribution. Ideally, a track of records should be available on the flow of government expenditures from the line ministries to district offices, from the district offices to the facilities, and finally on how resources are used by the facilities. However, in Ghana such records are not available. Based on the experiences of a tracking study conducted in Uganda- (Ablo and Reinikka, 1998), this tracking survey aims to identify the bottlenecks in resource flows in Ghana, and to explore ways of ensuring that monies meant for service delivery reach their targeted points. It is widely accepted that an improvement in the availability and access of public facilities, as well as enhanced utilization, would contribute substantially in effecting positive changes in the welfare of the population. To achieve these positive changes, the GOG has determined to bring local government administration to the doorsteps of the people, and to make social service provision more participatory. Decentralization policies, reflected in "bottom-up" approaches, have been adopted. The GOG has decided that the basic unit of administration is the local government at district level, however some of the main financial channels are still controlled by the central government. The new local government system is built on the assumptions that:  "development is that which responds to people's problems and represents their goals, objectives and priorities. Decentralization was the concept envisioned to transferfunctions, powers, means and competence to the District Assemblies from the Central Government, Ministries and Departments." (Ghana Interim PRSP 2000). Although the legal framework of decentralization has been conceptualized, the fiscal decentralization is much less clear and will take time. Currently, the actual flows of public resources are mainly based on the pre-decentralization module, which are either controlled centrally (salaries) or are channeled through the central agencies such as Ghana Education Services or Ghana Health Services. The PETS was designed to track down to what extent public resources flow between three points, central government agencies, district offices and service delivery facilities. First, we obtained the budget sheet from Ghana Education Services (GES) and Ministry of Finance for the Health sector expenditure, which gave us the actual expenditures by different level of service provisions and by budget items. Second, we conducted the tracking survey, which was based on a sample of basic education and primary health care facilities, as well as the district offices in charge of education and health. At the district offices and schools, we conducted PETS on actual annual public non- salary expenditures received for the last two years. In-kind public resources were asked to be converted to monetary values by the respondents. The respondents provided figures based on their records, but our enumerators did not demand to check the records in order to avoid giving an impression of a public auditing. This tracking survey is meant to provide only one piece of puzzle to the whole picture of improving public services and implementing decentralization: namely, enhancing transparency and accountability of public expenditure distribution. It needs to be bome in mind that the survey represents a pilot project. Due to its moderate scale and the experimental nature, the results of the study are meant to be more provocative than conclusive. Nonetheless, we expect that the results will generate a broad interest among government agencies and civil society, encouraging the discussion of further steps. This paper is organized as follows. Section 2 describes the implementation of the Ghana PETS and the lessons leamed from conducting such a survey. These lessons represent valuable experiences for a larger scale survey in the future, if need arises. In addition to general study environment, such as availability of classrooms, text books, etc., and students academic performances, -Sections 3 and 4 focus on the analysis of resources flow from the central governments to primary and Junior Secondary Schools (JSS), respectively. Section 5 covers the results of an analysis on the resource flows to health care facilities and health service provision. Finally, section 6 summarizes the findings. 2  H. THE IMPLEMENTATION OF PETS IN THE SECTORS OF EDUCATION AND HEALTH The systems ofpublic expenditure distribution The biggest challenge that we encountered in conducting this tracking survey was to understand the systems of public resource distribution in Ghana. On paper, all resources at the level of the central govermment are allocated by the Ministry of Finance to the line ministries with clearly marked heads and subheads, indicating where the funds should go. In reality, however, the distribution systems of public spending are rather complicated. On the one hand, the salary component of the public expenditures is distributed directly by a central government agency, the Controller Accountant General (CAG), to public employees, through the banking system. On the other hand, the monetary values marked at the level of the line ministries for non- salary recurrent expenditure were largely distributed to the public service provision facilities in the form of in-kind materials, the monetary values of which were not indicated and thus frequently not accurately known at the level of the facilities. We chose to focus our survey on tracking the non-salary expenditure in this study due to both that some work has been previously done in tracking ghost teachers by GOG and that tracking salary spending would require a whole set of different survey instruments, which were beyond scope of this survey. Ghana's distribution system has both its strengths and its weaknesses. The first strength is that public employees are mostly paid fully and on time, which often is a serious concern in other Sub-Saharan Africa (SSA) countries. The second strength, being perhaps also a weakness, is that a large proportion of public resources are distributed in form of in-kind materials. On the one hand, given the current situation in marketing and local capacity, it may be necessary to procure at the level of the central government for scale merits. On the other hand, there is hardly any mechanism built into the distribution system that would make it possible to routinely track the actual flows of public resources or to easily prevent inefficient use or misuse of resource at the central government level. On a different issue, the system also does not allow for feedbacks from the facilities to the procurement agencies to ensure on time delivery of needed items. It is a distribution system that has a tightly top down control mechanism to prevent corruption, but very little leverage for bottom up accountability. The Ghana budget distribution systems for the sectors of education and health are illustrated in figure 1 and 2. The solid lines represent cash flows, the dotted lines mostly material flows. Codes in gray boxes indicate resources retained in the agency; resources of other codes should flow to the next level of agencies or facilities. Through the PETS we have obtained information on the agencies/facilities with a thicker frame. Figure 1 shows the allocations of public expenditures from the MOF down to the level of primary and secondary schools with clearly marked expenditure categories. For example, Codes 142.02.x denote primary schools and 142.03.x denote JSS. Of these, 142.02.2 to 5 and 142.03.2 to 5 are for recurrent expenditures of different categories, such as travel and transport, general expenditure, maintenance, etc. The codes 142.02.1 and 142.03.1 are used for salaries, and distributed by the CAG. District Assembly Common Fund (DACF) is a newly established fund to distribute public resources in a form of cash flows to support the decentralized functions of district assemblies. 3  t icl.subenions|/JVJ GES-Districts Ditrc MOE:p 142.02.2-l: PrG 140 142.-32-5 S .~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1 GES-GES-HQn 14-<4 1 -.a 14a GES-Districts Dsrc fi 4z;) -_inl ... .................. Assembly 142.02.2-5: Prim : < 142.03.2-5: JS . .42 t s _ PIMARYSCHOOLS & JSS Teachers: salary P .02.'.!-.-, 142.~037'A '-> 42:27-1 >4-26'&L 2- i6 Figure 1: Ghana: Flow of Basic Education-Related Public Funds Figure 2 shows the allocation of public expenditure from the MOF down to the level of hospitals and health clinics. The PETS tracks financial/resource flows between the agencies with thick frames, namely, MOF, district health offices and clinics. For example, codes 165.503 items 2 to 5 are non-salary recurrent expenditures for clinics, which allocated by MOF to Ghana Health Services (GHS), who in turn should distribute the designated resource to District Health Offices (DHO). However, while budget allocation of non-salary recurrent expenditure. from MOF to GHS is in monetary values (represented by solid line), the majority resource flows are changed into materials (represented by dotted line) between GHS and DHO. Clinics we 4  surveyed reported that they hardly receive any public expenditure in form of cash, and the monetary values of the distributed materials are not indicated. Clinics mainly depend on intemally generated funds (IGF) for all non-salary recurrent expenditures. Figure 2: Ghana: Flow of Health Care Related Public Funds MOF Donor Pooled - p Fund (Health CAG , ( MOH: J 0 MinistryHealth GH D fa3ii|l ACF J 162.204 Terdary 165 hospitals *. z ^ v ~~~~~~~~~..................... ........ DisOcHealth Regional Health Directory: 64:40L:, 165.502: district H 402-403: regional hospitals 165.503: health r- centres/clinics/posts Regional Hospitals 1 . 11~~~64A6280Q!03 Distict Hospitl Health Centres/Clinics/Posts 1 Tpayroll to F1:6-5>.503.2-: G health care workers The knowledge gap between the monetary values and the material form of the resource flows may create much more serious leakage in health than in education sector for two reasons. First, the salary component of recurrent expenditures amounts to only 35 percent in the health sector, but to 90 percent in the basic education sector. This means that CAG controls only 35 percent budget distribution in health sector. Given the very direct distribution channels of CAG, the salary budget may have fewer channels to leak. Second, the materials procured within the health sector, such as drugs, are much more resalable than text books. As shown in Figures 1 and 2, schools and clinics in general receive very little cash. Based on our estimation, on average, each primary school student only received 819 cedis in cash in 1999, which translates to about 10 percent of total government non-salary recurrent expenditures. For JSS, each student received 1500 cedis, about 13 percent of total non-salary recurrent expenditures. A large proportion of public expenditures targeted to these facilities are in-kind, with no indication of their monetary values. Therefore, although the central government has a precise plan on how to allocate public expenditures, there is very little knowledge at the 5  district offices or the level of facilities on the monetary values of what they are entitled to receive and what they actually received. Implementation of the PETS The sample design of the Expenditure Tracking Survey is multistage. Firstly, four districts were chosen from each of the ten regions in Ghana. Based on the natural resource endowment of the districts, and an interplay of factors such as the level of infrastructure development, the income and welfare levels of the population, two depressed, one average and one better off district were selected in each region. The second stage involved the selection of,service delivery facilities. From the district list of schools provided, five basic education schools (three primary schools and two junior secondary schools) per district were selected. The same procedure was applied for the selection of health facilities, including three health posts (the first point of call in rural communities), one clinic and one health center per district. It should be mentioned that before starting with the fieldwork, a health center was generally considered as a higher level facility than a clinic, however, the survey showed that the use of the notions "center" and "clinic" is not strictly related to the level of health facilities. In some cases, a "clinic" was actually a higher-level facility than a "health center". In addition, district level information was sought from District Education Officers and District Health Officers respectively. To facilitate the link with the Ghana Living Standards Survey Round 4 (GLSS 4) data set, conducted in 1998, Enumeration Areas (EAs) falling within the selected districts were chosen first. Service facilities located in the selected EAs automatically became part of the sample of health or educational facilities. Finally, facilities not located in the selected EAs but reportedly used by GLSS 4 households were added as well. For purposes of tracking and comparability, the PETS data can be matched with GLSS 4. The GLSS 4 sample was selected from a sampling frame of 13,000 EAs developed from the 1984 Population Census. Each EA indicated its locality status (rural, semi-urban or urban), population and household information. The frame was first stratified by ecological zones (coastal, forest and Savannah). Within each zone the EAs were subsequently stratified by rural, semi-urban and urban. The GLSS 4 is based on a stratified nationally representative sample of households. At the first stage, 300 EAs were selected with a probability proportional to size (PPS). At the second stage, 20 households were selected from EA. The PETS was designed with the aim of matching data from the survey with the GLSS 4 data set. The PETS covered 143 EAs of the 300 GLSS 4 EAs. The distribution is presented below: Table 1: Distribution of enumeration areas by type and region Region Rural Eas Urban EAs Total Western 10 7 17 Central 1S 4 19 Great Accra 5 13 18 Volta 14 4 18 Eastern 9 5 14 Ashanti 12 3 15 Brong Ahafo 9 4 13 Northern 7 5 12 Upper East 8 1 10 Upper West 5 2 7 ALL 95 48 143 6  A team of 2 enumerators was assigned to each region. The team moved from one district to the other. The first points of call were the District Education Office and the District Health Office. Thereafter, the team split, with one enumerator administering questionnaires to selected health facilities, and the other covering educational facilities. The fieldwork was preceded by a two-day training, and hands on experience at Kasoa (in the Central Region), Labadi, James Town, Oyarifa and Kaneshie in the Greater Accra Region. The fieldwork took place in May 2000. The total responding rate from district offices was high, amounting to 98 percent. It was similarly high for schools, i.e., 99 percent. It was much more difficult to survey health facilities, as the responding rate of 87 percent suggests. In general, there were fewer problems in collecting data from the district offices. With regard to facilities, the non-financial data are quite accurate, and indeed consistent with what we found from the data of the ministry. However, there were more problems in collecting financial data. Some frequently cited reasons for this problem included the lack of systematic filing procedures, the problem of data filed in different places, etc. All data, including those in district offices, are recorded on papers (not in computers) and get discarded promptly. Heads both of health or school facilities changed frequently. However, when a head left, he/she typically did not transfer information to the next head. IH. PRIMARY EDUCATION AND PUBLIC RESOURCE FLOWS Study environment .n the tracking survey, we first asked about basic information regaiding the number of teachers, students, textbooks, classrooms, sources of drinking water, etc. Table 2 provides a general picture of the study environment by region for 1998 and 1999. In general, most classes have their own classrooms. In Accra, however, every three classes had to share two classrooms. Most of the classrooms are permanent. Average class-size has increased from 36 pupils per classroom in 1997/98 to 38 in 1998/99. However, class-sizes are much larger in the Northern and Upper East regions, with about 45 pupils per classroom. There is not a great deal of classroom sharing at the same time among different grades, but it occurs more frequently in the North and Upper West regions. Most students must share their desks and chairs, especially in the three poorest regions, namely Upper West, Upper East and Northem Region, as well as in Accra. The supply of both textbooks and teachers' handbooks increased slightly between 1997/98 and 1998/99. Students had on average 1.7 textbooks available in 1998/99. Considering that there are. five to seven subjects taught in primary schools, less than two books per pupil signifies a great deal of textbook sharing. Sharing handbooks among teachers is also common. The majority of teachers have graduated from a teacher's college, although this is less true for teachers in the Northern region. 7  Table 2: Primary school learning enviromment by region, 1997/98 and 1998/99 WES CEN GRA VOL EAS ASH BRA NOR UEA UWE GHA #of classes per 1.4 1.2 1.4 1.1 1.1 1.1 1.2 1.1 1.3 1.2 1.2 classroom, 98 # of classes per 1.3 1.2 1.4 1.1 1.1 1.1 1.1 1.1 1.3 1.2 1.2 classroom, 99 _ % of temporary 13.3 3.0 2.7 19.4 6.1 9.5 31.3 12.5 10.5 14.8 12.1 classrooms, 98 _ _ % oftemporary 11.4 7.2 2.7 10.9 6.1 10.3 26.5 9.5 10.2 13.4 10.6 classrooms, 99 Average class 39 33 37 .35 32 34 29 45 46 28 36 size, 98 I I I Average class 39 34 41 37 31 38 31 45 45 32 38 size, 98 % classroom 2 0 6 3 8 6 4 11 4 12 5 sharing by different grade, 98 _ _ % classroom 4 0 2 5 9 7 4 11 4 11 6 sharing by different grade, 99 _ _ _ Teachers/class, 98 0.97 0.97 0.99 1.07 0.96 1.30 1.05 1.43 0.94 0.80 1.05 Teachers/class, 99 0.99 0.99 1.01 1.03 0.96 1.29 1.06 1.42 0.91 0.83 1.05 PupiLs/teacher, 98 40 34 38 33 34 26 28 32 49 35 35 Pupils/teacher, 99 40 35 41 36 33 30 29 32 49 38 36 Pupils/sitting 1.4 1.2 1.4 1.0 1.1 1.2 1.2 1.8 2.9 2.2 1.4 place, 98 _ __I_I Pupils/sitting 1.4 1.2 1.4 1.0 1.1 1.2 1.2 1.8 2.9 2.2 1.4 place, 99 ___ Pupils/writing 1.5 1.2 1.4 1.2 1.2 1.3 1.2 1.8 3.1 1.4 1.5 place, 98 Pupils /writing 1.5 1.2 1.4 1.2 1.2 1.3 1.2 1.8 3.1 1.4 1.5 place, 99 I Textbooks per 2.3 1.1 0.9 1.5 1.5 1.4 2.1 1.6 1.2 0.7 1.4 pupil, 98 I_I Textbooks per 2.3 1.4 1.2 1.4 2.1 1.9 2.4 2.0 1.3 1.1 1.7 pupil, 99 Handbooks per 2.3 3.4 1.9 2.8 6.2 1.2 2.3 0.8 1.8 2.6 2.4 teacher, 98 1 Handbooks per 2.8 3.8 1.8 2.1 9.5 1.6 2.6 1.0 2.2 2.9 2.8 teacher, 99 Teachers per 9.9 9.5 9.3 12.4 8.5 10.0 9.0 9.0 9.6 5.4 9.2 office, 98 Teachers per 9.2 8.8 9.5 11.7 8.5 10.1 9.2 8.5 9.5 5.5 9.0 office, 99 _ I_I % ofteacher from 88 93 99 100 91 N/A 80 78 88 95 90 teacher training college, 98 . % of teacher from 83 N/A 96 100 87 94 78 77 86 91 88 teacher training college, 99 Source: Ghana Public Expenditure Tracking Survey, 2000.  In addition to the classroom environment we also surveyed the sources of drinking water for students, since it concerns students' health. Table 3 sununarizes results on different sources of drinking water and the type of drinking water storage for students. On average about one third of schools had pipe water, and a quarter had well/spring water. "Other type" of water supply could be considered as "bought water", since it is a common practice. If these sources are considered safe, the majority of schools provide safe drinking water for their students. However, seven percent of schools do not supply any drinking water. If lake/river/rain water sources are categorized as unsafe, in Brong Ahafo and the Northern region, close to 20 percent of schools supply students with unsafe drinking water. The majority of schools store their water supply in coolers, which are considered to be sanitary. Table 3: Sources of drinking water Type of drinking water supply, % WE CEN GR VOL EAS ASH BRA NO UEA UWE GHA _ _ _ _ _ _ ~~S A __ _ R _ Pipe 33 73 67 31 25 43 17 25 25 8 34 Well/spring 25 27 17 8 67 29 8 25 17 17 24 Lake/river/rain 0 8 15 8 7 17 19 8 0 9 Other unspecified 25 0 8 31 0 14 42 31 42 67 26 None 17 0 0 15 0 7 17 0 8 8 7 Type of drinldng water storage, Me _____ . WE CEN GR VOL EAS ASH BRA NO UEA UWE GHA S _ A _ _ _ _ _ _ R _ __ Standing Pipe 0 9 0 0 0 0 17 0 8 8 4 Cooler 83 73 75 77 92 93 50 25 42 50 65 Bowl/other 0 0 25 8 8 0 0 19 8 42 11 None 17 18 0 15 0 7 33 56 42 0 20 Source: Ghana Public Expenditure Tracking Survey, 2000. Academic perfonnance and gender bias Table 4 presents some education indicators for two categories, academic performance and gender bias. The academic performance indicators show that girls' performance is the same as that of boys. The percentages of girls who dropped out of school or who repeated grades are not higher compared to their male counterparts. Based on MOE's 1995/96 data, the promotion rates from P6 to JSS1 among boys and girls are also very similar. This means that the difference in enrollment ratios between boys and girls through JSS 1 is mainly caused by gender bias, namely that less girls than boys are sent to first grade, not by gender differences in terms of academic performance. Indeed, based on MOE's 1997/98 data, the first grade gross intake rate is 81 percent for boys, and 76 percent for girls. 9  Table 4: Primary school academic performance by region WE CEN| GR VOL EAS ASH BRA NO UEA UWE GHA S A IR Academic performance % drop outs 2.3 2.3 1.9 2.7 4.7 2.6 4.1 1.0 3.9 3.7 2.8 among fem., 98 . % drop outs 4.2 2.2 1.2 3.1 3.2 2.2 4.4 0.3 2.3 3.3 2.5 among males, 98 _ % drop outs 4.4 3.3 1.0 2.4 6.2 2.6 4.4 0.9 2.3 2.4 2.8 among fem., 98 % drop outs 4.8 2.7 1.1 1.9 5.5 2.8 3.4 0.5 2.5 1.8 2.6 among males, 98 _ _ % repeats among 3.4 1.4 1.1 1.7 6.4 2.5 4.3 1.4 2.4 4.3 2.7 females, 98 %ofrepeats 4.2 1.9 1.2 1.8 N/A 2.5 3.5 2.7 2.5 5.9 2.8 among males, 98 _ % repeats among 6.2 1.9 1.9 2.6 7.9 3.6 7.0 2.7 2.5 6.3 3.9 females, 99 % of repeats 7.1 2.0 1.9 2.8 7.0 4.1 6.5 1.3 3.5 7.8 4.1 among males, 99 . _ Gender bias % fem. pupils, 98 49 50 52 50 49 50 48 39 43 45 47 % fem. pupils, 99 49 49 51 50 48 51 47 40 44 46 48 Among those 45 49 48 50 46 49 45 31 43 49 45 who finished P6, % females, 98 _ Among those 51 48 51 46 50 52 44 32 42 46 46 who finished P6, % females, 99 _ Source: Ghana Public Expenditure Tracking Survey, 2000. Resourceflows to primary schools The tracking survey at schools provided very interesting and somewhat different information than indicated by the statistics of the MOE. We first split total recurrent expenditure into salary and non-salary components. This is necessary because these two components reach schools through different channels. While salaries are distributed directly through the banking system by the Controller Accountant General (CAG), the non-salary recurrent expenditures are managed by the Ghana Education Services, GES, its headquarter in Accra and the district education offices. For non-salary recurrent expenditures, schools provided information on the amount of public expenditures, both in monetary terms and in quantities of in-kind materials. As noted earlier, a large proportion of subsidies is supplied in-kind, such as in the form of books, stationary or chalks. Schools typically are not aware of the exact monetary value of these materials.. Thus, the monetary values provided below are based on rough estimates by the respondents from the schools, typically the headmaster. In addition, schools do not keep records 10  of teachers' salaries. To account for the cost of salary subsidies, we estimated salary costs based on the average cost of salaries of school teachers and headmasters.1 Based on these estimates, Table 5 includes a comparison between the expenditures allocated by the MOE to primary schools and the resources received by primary schools. In 1997/98, about 64 percent of the public expenditures that left the GES reached schools, after taking into account overhead administrative cost. Specifically, 51 percent of non-salary and 65 percent of salary recurrent expenditure reached the primary schools. The non-recurrent expenditures in 1998/99 are by about 10 percent lower than in 1997/98. The salary expenditures have increased due to a cross board increase in teacher's salaries. We do not have budget data from the MOE/GES for 1998/99, thus no estimates of public expenditures leakage is available for 1998/99. Table 5: Primary school recurrent expenditure (cedis), per enrolled student 1997/98 1998/99 l Non- Salary Total Non- Salary Total salary salary Based on MOE/GES 12,081 108,724 120,805 N/A N/A N/A estimates2 . Based on District Education Office estimates3 N/A N/A N/A N/A N/A N/A Based on school budget estimates 5,730 65,828 71,558 5,104 78,735 83,839 Adjusted for 8% overhead cost 6,188 71,094 77,283 5,512 85,033 90,546 As % of MOE/GES estimates 51 65 1 64 N/A N/A N/A Source: Ghana Public Expenditure Tracking Survey, 2000. Interestingly, the pupil teacher ratio (PTR) calculated by the surveyed schools is exactly the same as shown by the GES data and teachers get paid in full on time. Several possibilities could suggest the unaccounted 35 percent of salaries in 1997/98. One is that the basis for calculating average salaries for school teachers was too low, which is unlikely since we were aiming at a higher end of salary scale. Second is that there are more teachers on the payroll of the CAG's than there are actually in the schools. Unfortunately, we did not obtain payroll information from CAG. A comparison of information on the number of teachers available from GES and CAG could answer this question more precisely. Table 6 shows the components of total school spending on non-salary recurrent expenditures, which consists both public spending and private contributions. The GOG non- salary recurrent expenditures have decreased from 72 percent in 1997/98 to 52 percent in 1998/99. While there is a decrease in GOG expenditures, the non-governmental/private funding ' Based on the GHS teachers salary structure, we estimated that the average monthly salary in 1997/98 amounted to 190,000 cedis for primary school teachers, which is the middle point for trained teachers of grade A.39 in 1998. This is at a somewhat higher end of primary teachers' pay grade, and 280,000 for headmasters. For 1998/99, the grade system was changed, the corresponding grade of A.39 was approximately Level 6 and salary in general was increased by 23 percent. Thus, the estimated salaries were 230,000 and 340,000 cedis, respectively. The monthly salary was multiplied by 12 to obtain the annual salary. 2 Based on the MOE 1998 budget sheet, the salary is about 90 percent of total recurrent expenditure. For education sector, district official estimates on unit cost are not available due to lack of data.  increased sigmticantly dunng these two years. This, however, was mainly the result of a large increase in the Ashanti region. Western and Great Accra also received a significant increase from non-governmental contributions. If inflation is taken into account, the total recurrent non- salary expenditure per student in 1998/99 maintained at a similar level as in 1997/98. Table 6: Components of primary school non-salary expenditures, per enrolled student 1997/98 1998/99 Percent Cedis Percent Cedis Govenmment Cash 6 454 8 819 Government in kind subsidies, recurrent 66 5,246 44 4,382 Total government recurrent 72 5,701 52 4,523 expenditure , District Levies 6 471 6 554 PTA contributions 14 1,071 15 1,528 Non-government cash contribution 1 116 3 261 Non-government in kind subsidies, 7 568 25 2,448 recurrent Total private contributions to recurrent 28 2,225 48 4,167 expenditure at school Current cedis per student 7,926 8,690 1997/98 cedis (15% of inflation) 7,926 . 7,387 Govermnent capital investment 11 121 3 141 Non-govenmment capital investment 89 1,026 97 4,732 Currentcedis perstudent 1,147_ 4,873 1997/98 cedis (15% of inflation) _ 1,1471 4142 Source: Ghana Public Expenditure Tracking Survey, 2000. In the context of decentralization, the GOG has delegated financial resources to the District Assemblies Common Fund (DACF) to invest in projects and programs of local priorities. From our tracking survey, however, we were only able to observe a small amount of resources flowing from DACF to schools. For the two-year period covered by the survey, only three schools received a total of 280,000 cedis from DACF. There was some government rehabilitation spending is recoded. Even assuming that all the rehabilitation spending actually came from DACF, still there was only an insignificant resource flow between DACF and schools. It should be mentioned here that one expenditure we did not capture from the DACF is the development capital expenditure (i.e. building new schools). Reasons for not attending school and ways to improve primary education From the perspective of schoolmasters we asked questions in regard to why school-aged children do not attend school and what can be done to improve the quality of primary education. Table 7 lists responses to the' first question, which were ranked according to three levels of importance. Economic constraints are the top, and lack of interest the second most important reason for not attending school. Lowering school fees, however, would apparently not suffice, as only 13 percent of responses cited that too high fees were an economic constraint. The other 45 percent of the responses referred to economic constraints in the households, including low pay and unemployment of the parents and the need for the child's labor at home. This points to the depth and scope that the universal primary education in Ghana faces. Some of the factors 12  deterring children from attending school are clearly beyond the scope of the Ministry of Education. There are similarities between the responses to why children did not attend school in the 1997 CWIQ and in the 1999 tracking survey.4 The most frequently cited reasons (36%) for not attending school among parents/students are economic constraints, with 24% referring to too high fees and 12% to labor needed at home. The second most important reason cited by parents/students, about 30%, was the lack of interest in school. Mostly by parents account, a very low proportion of responses referred to long distance as a reason for not attending school. Table 7: Reasons for not attending school Most Second Third Weighted average important important important of column percentages Economic constraints 64 57 57 59 Fee too high 21.5 10.3 7.5 13 Transport and food too expensive 0.0 0.9 0.9 0.6 Parents incomes too low 20.6 15.9 17.8 18 Parents unemployed 5.6 3.7 15.0 8 Children's labor needed at home 15.9 26.2 15.9 19 Perception reasons 34 36 32 34 Students lack interest for school 11.2 13.1 7.5 11 Parents lack interest for school 22.4 23.4 24.3 23 Physical constraints 11 7 School too far away 0.0 .0.9 2.8 1 Marriage/pregnancy 0.0 4.7 2.8 2 Other 2.8 0.9 5.6 3 Source: Ghana Public Expenditure Tracking Survey, 2000. The tracking survey also included some objective questions such as what the respondents thought about the degree of improvements for certain items since the last three years, and what could be done to improve the quality of primary education. The summary of responses is listed in Table 8. About 50 percent of schoolmasters felt, that improvements have been made with regard to PTA, state of classrooms, and availability of teaching supplies. Most frequently cited complaints include the fixed assets of schools, i.e., the repair of classrooms and desks/chairs. These complaints are consistent with the govemment budget, which allocates almost no funding on the rehabilitation of schools. Almost 80% of schoolmasters viewed that their quality of education had improved, which, however, has probably to do with their objective view on their work performance. 4 The two surveys are not strictly comparable. The choices given by the two questionnaires are very similar, though not exactly the same. In addition, in CWIQ respondents could only choose one answer while in the tracking survey they could pick the three most important reasons. 13  Table 8: Improvements of primary schools Improved Worsened No change Don't know Total PTA 55.2 12.8 30.4 1.6 100 State of repair of classrooms 49.6 24.0 26.4 0.0 100 'Classroom space per pupil 37.7 15:6 46.7 0.0 100 Availability of school 45.6 32;8 20.8 0.8 100 desks/tables/chairs . Availability of school supplies 51.2 15.4 32.5 0.8 100 (textbooks, chalks, etc.) . Quality of education 77.6 9.6 12.0 0.8 100 Source: Ghana Public Expenditure Tracking Survey, 2000. Interestingly, the responses to a similar question in the CWIQ survey from parents' students are largely consistent with those of school headmasters.5 Around 40 to 50 percent of primary students complained about a lack of school supplies, overcrowded classrooms, and bad conditions of facilities/classrooms. A significant proportion of students (40 percent) did not experience any problems at school on the issues asked. In short, the complaints from teachers and students alike are mostly centered around things funded by non-salary expenditures. Finally, the tracking survey asked school headmasters about what the government can do to increase school enrollment and to improve the quality of education. The results are listed in Table 9. To improve access to education, the first important thing which was cited most frequently is to give more assistance to the needy families. This response highlights that one of the most important barriers to increase enrollment is derived fro'm the demand side. Currently, based on the tracking survey, only 12 out of 126 primary schools provided fee exemptions for students in 1999. A total of 264 students received fee exemptions, which is less than one percent out of a total of 35,895 students. The second is to build more classrooms in existing schools, and the third is to build more schools. In terms of improving the quality of education, the most frequently cited reason was to recruit more qualified teachers, both the second and the third important reasons referred to improvements of classroom conditions. It is interesting to note that an increase of teachers' salaries has not been most frequently cited as a way to improve education, although it is cited by about a quarter of responses. In short, teachers' training and improving classroom conditions are on top of the list to improve the quality of education. 5 In the CWIQ survey, parents/students were asked if they have any problems with regard to the distance of the school, the quality of teaching, the lack of teachers/over crowding or bad facilities. We have chosen the responses from primary students only to be consistent with primary school tracking survey. 14  Table 9: Suggested ways to improve access to education and the quality of education What can the government do to improve the Most Second Third important access to education? important important Give more assistance to the needy families 55 27 13 Build more schools 7 18 38 Build more classrooms in existing schools 30 42 23 Other, specify 7 12 26 Vhat can the government do to improve the quality of education? Increase the number of qualified teachers 45 17 18 Improve classroom conditions 16 38 27 Reduce over-crowding in classrooms 5 17 19 Increase teachers' pay 28 27 25 Others, specify 6 2 12 Source: Ghana Public Expenditure Tracking Survey, 2000. IV. JUNIOR SECONDARY SCHOOL EDUCATION AND PUBLIC RESOURCE FLOWS Study environment Similar to the primary schools surveys, the JSS surveys also asked questions in regard to the learning environment and educational performances. As shown in Table 10, each JSS class has on average one classroom. The average class-size amounts to 38 pupils per classroom, similarly to primary schools. The pupil teacher ratio remains quite equal across all regions at the low 20's. Pupils per sitting and writing-place reveal less crowded schools in comparison to primary schools. Textbook supplies have improved slightly between 1997/98 and 1998/99, amounting to an average of 4.2 books per student. Given that eight or nine subjects are taught in JSS, the textbook supply is much better in JSS than in primary schools. Teachers' handbook supply varies significantly across regions, and 35 percent of all schools reported that they do not have any teacher's handbooks at all.6 The proportion of teachers that attended a Teachers' Training College is slightly lower as compared to the primary school teachers, amounting to 87 percent. 6 In JSS each teacher teaches one or two subjects to all classes, while in primary schools each teacher teaches all subjects to one class. 15  Table 10: JSS learning environment by region, 1997/98 and 1998/99 WES CEN GRA VOL EAS ASH BRA NOR UEA UWE GHA #of classes per 1.0 1.1 1.5 1.1 0.9 1.0 1.0 1.0 0.9 0.8 1.0 classroom, 98 _ _ # of classes per 1.0 1.1 1.1 1.2 0.9 1.0 1.0 0.9 0.9 0.8 1.0 classroom, 99 _ % of temporary 6 7 25 17 8 0 3 29 0 14 11 classrooms, 98 ___ _ % oftemporary 6 4 25 18 8 0 3 20 0 14 10 classrooms, 99 _ _ . Average class 33 32 28 31 34 36 32 49 47 33 35 size, 98 1 . I.I I Average class 39 33 38 33 34 '39 31 52 50 36 38 size, 98 Pupils per 23 20 23 21 22 19 20 21 25 20 21 teacher, 98 . . Pupils per 24 21 22 23 21 21 18 23 27 19 22 teacher, 99 _ __ _ _ _ Pupils per 1.1 0.9 1.3 1.0 1.0 1.1 1.1 1.4 1.3 1.0 1.1 sitting place, 98 Pupils per 1.1 0.9 1.2 1.1 1.0 1.2 1.0 1.2 1.4 1.1 1.1 sitting place, 99 _ Pupils per 1.1 0.9 1.3 1.2 1.0 1.1 1.1 1.4 1.3 0.9 1.1 writing place, 98 . _ Pupils per 1.1 0.9 1.2 1.3 1.0 1.1 1.0 1.2 1.4 1.0 1.1 writing place, 99 Textbooks per 4.9 2.4 3.2 3.6 4.5 3.8 6.4 2.9 3.3 4.1 3.8 pupil, 98 I I_ Textbooks per 4.2 3.0 4.3 3.9 5.0 3.8 7.9 3.4 3.5 4.8 4.2 pupil, 99 . _ Handbooks per 0.3 2.0 0.5 0.5 1.8 0.5 0.3 0.3 0.4 0.4 0.7 teacher, 98 _ Handbooks per 0.3 2.6 0.5 0.9 1.6 0.5 0.3 0.4 0.4 0.6 0.8 teacher, 99 Teachers per 7.0 6.3 7.2 9.5 6.2 6.9 6.0 10.9 6.8 6.7 7.2 office, 98 1 Teachers per 7.6 6.1 7.3 9.5 6.6 6.9 6.5 8.8 6.3 7.5 7.2 office, 99 % ofteacher 92 82 90 93 77 92 92 82 89 85 87 from teacher training college, 98 % of teacher 83 84 93 93 75 94 90 86 86 84 87 from teacher training college, 99 . , . _ Source: Ghana Public Expenditure Tracking Survey, 2000. Table 11 provides drinking water supplies in JSS for 1998 and 99.7 It shows a similar pattern as compared to primary schools. The majority of schools have a relatively safe drinking water supply, if well/spring water and other supply are considered safe. The water storage 7 Since we only surveyed 8 JSS in each region, we did not want to present regional percentages due to the small sample. 16  equipment is mainly a bowl type of container with cover, which is probably less safe than coolers which are most frequently used in primary schools. There have not been much changes in terms of the drinking water supply between 1997/98 and 1998/99. Table 11: JSS drinking water supply and storage Type of drinking water supply, % 1998 1999 Pipe 34 38 Well/spring 23 20 Lake/river/rain 8 6 Other 30 30 unspecified None 5 5 Type of drinking water storage, % 1998 1999 Standing Pipe 5 4 Cooler 4 4 Bowel/other 68 70 None 14 13 9 10 Source: Ghana Public Expenditure Tracking Survey, 2000. Academic performance and gender bias Table 12 presents education indicators in two categories, academic performance and gender bias. Data on the performance indicators suggest that female students in general perform as well as or even better than male students. For example, the proportion of female students who passed BECE is much higher than of male students. However, the bias indicators indicate that only 31 percent of all female students who passed BECE went to SSS in 1999, while this has been the case for 38 percent of male students. As a result, female students account for only about 40 percent of students who went to SSS, in spite of their superior academic performance. 17  Table 12: JSS academic performance by region I WES CEN GRA VOL EAS ASH BRA NOR UEA UWE |GHA Academic performance % of drop outs 1.6 2.7 0.9 0.7 1.6 0.9 5.0 0.6 2.3 3.8 1.8 among F, 98 _ % of drop outs 1.8 4.6 1.7 1.4 5.7 1.1 11.1 0.5 5.2 4.2 3.2 among M, 98 . % of drop outs 1.9 1.5 1.0 1.3 1.0 0.7 5.0 0.2 1.7 3.2 1.5 among F, 99 % of drop outs 2.4 1.6 4.2 1.9 3.3 0.5 6.5 0.3 4.5 5.4 2.7 among M, 99 __ % of repeats among 3.5 3.2 1.5 1.1 2.4 3.3 3.1 0.4 3.6 3.0 2.5 F, 98 _ % of repeats among 2.5 2.1 0.9 0.8 2.1 2.7 1.6 1.0 3.7 2.4 1.9 M, 98 % of repeats among 2.1 3.2 0.6 1.6 3.7 3.3 3.2 8.2 3.5 3.3 3.2 F, 99 % of repeats anong 3.3 2.9 0.8 0.9 1.7 2.5 1.3 6.6 3.6 2.5 2.8 M, 99 _ % of females passed 36 44 46 31 42 41 31 27 40 35 37 BECE, 98 - % of males passed 31 19 26 35 25 31 14 31 25 28 27 BECE, 98 % of fenales passed 45 41 48 29 42 41 27 26 43 34 39 BECE, 99 . . % of nales passed 23 16 27 24 23 27 12 26 24 31 24 BECE, 99 _ _ . Gender Bias % of female 45 47 46 44 48 49 43 33 48 46 45 students, 98 % of female 45 46 48 .38 48 48 44 32 45 49 44 students, 99 . % of females who 14 9 N/A 42 77 30 21 17 91 6 37 passed BECE went to SSS, 98 % of males who 15 5 N/A 70 70 31 25 45 83 4 43 passed BECE went to SSS, 98 % of females who 14 N/A N/A 39 47 28 24 14 87 3 31 passed BECE went to SSS, 99 % of males who 20 14 N/A 38 43 28 25 49 85 5 38 passed BECE went to SSS, 99 Of whom went to 36 N/A 14 33 55 42 28 30 41 50 41 SSS, % of females, 98 Of whom went to 38 19 43 54 43 27 24 42 30 40 SSS, % of females, 99 _ Source: Ghana Public Expenditure Tracking Survey, 2000. Thus, fewer girls than boys enter the first grade of primary school, which represents a first barrier to girls' education. After they enter primary school, they tend to remain in education up to their graduation of JSS, just like boys. However, in spite of their superior academic performance, girls are more likely to drop out of education after finishing JSS, while boys are more likely to go on. Thus, entering SSS represents a second barrier to girls' education. 18  Resource flows to JSS The questionnaire for JSS PETS on the budget flow is the same as that of the primary school surveys. The non-salary expenditure subsidies are reported by the schools, and the salary component of recurrent expenditures is based on our estimates.8 As shown in Table 13, in 1997/98 a little over 70 percent of the total public expenditures for JSS have reached the schools. Of these, about 75 percent salary expenditure, and less than 50 percent of non-salary recurrent expenditures, reached schools. The expenditure per student has increased between 1998 and 1999 in real terms. This, however, may be the result of a reduction in the number of students rather than an increase in expenditures. Since the salary component is estimated based on the average salary level, the discrepancy between the MOE/GES and the school level estimates could be a result of underestimating teachers' salaries. However, since our estimates of salary was based on the higher end of JSS teacher's pay, it is most likely that the discrepancy between the MOE/GES and the school estimates on the salary component reflects a certain degree of financial leakage in the system. Remarkably, the unit non-salary recurrent expenditures have increased by 40 percent between 1998/99 and 1997/98, possibly a result both of smaller numbers of students and less leakage, since we know that the GOG did not increase the education expenditures that much. Table 13: JSS government recurrent expenditure (cedis), per enrolled student 1997/98 1998/99 Non- Salary Total Non- Salary Total salary salary Based on MOE/GES 19,665 176,984 196,649 N/A N/A N/A estimates9 Based on District Education N/A N/A N/A Office estimates Based on school budget estimates 8,765 123,500 132,265 11,292 146,718 158,010 Adjustedfor 8% over head 8,765 123,500 132,265 cost 12,195 158,455 170,650 As % of MOE/GES estimates 48 75 73 N/A N/A N/A Source: Ghana Public Expenditure Tracking Survey, 2000. Table 14 shows the different components of non-salary recurrent expenditures, including private contributions. The government supports a large proportion of non-salary recurrent expenditures, about 70 percent. The capital investments remain very low on average. In reality, only a few schools received resources on rehabilitation and capital investment, the majority of schools received no resources for such purposes. Again, government expenditures are mainly distributed as in-kind materials. It is worth mentioning, however, that it is common in Ghana that each student carries a small amount of building material, such as stones or sands, to his/her school once every few months at the request of schools teacher. It is quite possible that these private contributions were not accounted for. 8 Based on the GHS salary structure, at grade A.49, the higher end of the basic education teacher's monthly salary is about 210,000 cedis in 1997/98, we estimated 300,000 for a headmaster. In 1998/99, given a 23 percent increase in salary, the salary estimates are 250,000 for a teacher and 369,000 for a headmaster, respectively. The monthly salary was then multiplied by 12 to get the annual salary. 9 Based on MOE 1998 budget sheet, the salary is about 90 percent of total recurrent expenditure. 19  Table 14: Components of JSS non-salary expenditures (cedis), per enrolled student 1997/98 1998/99 Recurrent expenditures Percent Cedis Percent Cedis Govermnent Cash 10 1,290 9 1,503 Govermment in kind subsidies, recurrent 58 7,475 59 9,789 Total government recurrent 68 8,765 68 11,292 expenditure District Levies 6 712 6 917 PTA contributions 24 3,142 21 3,499 Non-government cash contribution 0 41 1 92 Non-governnent in kind subsidies, 2 291 5 751 recurrent Total private contributions to recurrent 32 4,186 32 5,260 expenditure at school Total non-salary recurrent expenditure, 12,951 16,552 current cedis per student 199 7/98 cedis (15% of inflation) 12,951 14,393 Rehabilitation/capital expenditures Government capital investment 32 1,399 42 2,015 Non-government capital investment 68 3,026 58 2,756 Total capital investment, current cedis per 4,425 4,771 student . 1997/98 cedis (15% of inflation) 4,425 4,149 Source: Ghana Public Expenditure Tracking Survey, 2000. In short, more than 90 percent of the education budget is financed by the GOG, mainly because salaries account for 90 percent of the total recurrent budget. Private contributions are PTA fees and district levies. PTA fees are retained by schools, while district levies are turned over to the district education offices.'0 The district levies from students only make up a very small proportion of the district education office managing budget. Reasons for not attending school and ways to improve JSS education As in primary schools, we asked JSS schoolmasters why children did not attend schools and - according to their opinion - what the government could do to improve access to JSS and the quality of education. Table 15 shows the responses to the first question. Again, economic related reasons are most important obstacles for attending school. The second important cause is the lack of interest in school from both parents and their children. The distance to schools and the cost of transportation are, however, rarely cited as obstacles. The economic causes for not attending school, from the perspective of the schoolmasters, are rooted in the economic development of the society, and thus well beyond the scope of the Ministry of Education. For example, it is not clear that a child can attend school on a scholarship if his or her labor is needed at home. From the perspective of students and parents, the CWIQ survey also demonstrates that economic issues are the most frequently cited reasons for not attending schools. 32 percent cited that schools are too expensive, and 17 percent cited that labor is needed at home. The second '