The incidence of social spending in ghana

82 PSP Discussion Paper Series 19704 November 1995 The Incidence of Social Spending in Ghana Lionel Demery Shiyan Chao Rene Bernier Kalpana Mehra November 1995 Poverty and Social Policy Department Human Capital Development and Operations Policy The World Bank FILE COPY  This Booklet of Abstracts contains short summaries of recent PSP Discussion Papers; copies of specific papers may be requested from Patricia G. Sanchez via All-in-One. The views expressed in the papers are those of the authors and do not necessarily represent t:ie official policy of the Bank. Rather, the papers reflect work in progress. They are intended to make lessonis emerging from the current work program available to operational staff quickly and easily, as well as to stimu.ate discussion and comment. They also serve as the building blocks for subsequent policy and best practice papers.  Preface This paper is one of a series undertaken as part of the Extended Poverty Study in Ghana. A synthesis of this work is available in Ghana: Poverty Past, Present and Future (Report No. 140504-GH, Population and Human Resources Division, West Central Africa Departmnent, World Bank, Washington D.C., June 29, 1995). Four of the background papers are appearing in the Poverty and Social Policy Discussion Paper Senres: Harold Coulombe and Andrew McKay, 'An Assessment of Trends in Poverty in Ghana, 1988-92.' Poverty and Social Policy Discussion Paper No. 81, World Bank, Washington D.C. (November 1995) Lionel Demery, Shiyan Chao, Rene Bernier and Kalpana Mehra, 'The Incidence of Social Spending in Ghana.' Poverty and Social Policy Discussion Paper No. 82, World Bank, Washington D.C. (November 1995) Andy Norton, David Korboe, Ellen Bortei-Dorku and D.K. Tony Dogbe, 'Poverty Assessment in Ghana using Qualitative and Participatory Research Methods.' Poverty and Social Policy Discussion Paper No. 83, World Bank, Washington D.C. (November 1995) Christine Jones and Ye Xiao, 'Accounting for the Reduction in Rural Poverty in Ghana, 1988-1992.' Poverty and Social Policy Discussion Paper No. 84, World Bank, Washington D.C. (November 1995) The papers draw upon the work of many. The contributions of the Ghana Statistical Service, Ministry of Health, and Ministry of Education in the Government of Ghana, and of UNICEF and the Canadian International Development Agency (CIDA), are gratefully acknowledged. The authors of this paper would also like to express their particular thanks to Dr. Asmoa-Baah and Paul Glewwe for very helpful comments. The views expressed are those of the authors. They should not be attributed to the World Bank, its Board of Directors, its management or any of its member countries. 2   Abstract Ghana's improved economic performance since the early 1980s has been attributed to its Economic Recovery Programme (ERP). Under the ERP, there have been changes in public spending, with significant increases in public investments in health and education. The paper investigates the implications of the economic recovery for the incidence of social spending in Ghana. Do the poor benefit from public education and health spending? 'Benefit incidence' estimates were obtained for health and education spending in Ghana. These combine estimates of the cost of service provision with information on household use of services obtained from the Ghana Living Standards Surveys (GLSS). Overall, the poorest quintile of the population was estimated to have gained just 16 percent of education spending (largely unchanged from 1989) and 12 percent of health spending in 1992 (somewhat less targeted compared with 1989). But there were significant differences within each sector. In common with many similar studies, the paper finds that the poor gained relatively more from primary education spending (22 percent), and less from secondary (15 percent) and tertiary (6 percent) spending. The larger share of primary education spending was due in the main to the greater proportion of children of primary school age in the lower quintiles. Because of these demographic effects, the paper shows that 'benefit incidence' estimates are sensitive to the welfare measure used in generating the quintile distributions. There were marked gender inequalities in education spending. On average, girls gained just 8,614 Cedis per capita from public education spending, compared with 12,803 Cedis per capita for boys. Surprisingly, the poorest quintile gained most from spending on hospital outpatient healthcare (13 percent), and less from inpatient care (11 percent) and the use of health centers (10 percent). Households in the poorest quintile were found to devote a significant share of their non- food spending to health (5 percent) and education (10 percent). These costs often constrain the access of the poor to subsidized public services. The paper highlights survey evidence that household expenditures on health and education are not always consistent with the official scales of charges. These inconsistencies need further investigation. Recent changes in public spending priorities in both education and health have undoubtedly been in the right direction-financing in both sectors has been shifted towards primary facilities, and away from tertiary institutions. But this is not enough. Despite the shift in emphasis, the poor have not increased their share of public subsidies in these sectors. Clearly, the strong bias in favor of urban areas is persistent. In education, the emphasis must be on further improving the management of education spending and the quality of the service provided. The latter requires greater budget allocations to non-labor inputs (such as school supplies). More resources appear to be needed for the provision of basic health care in Ghana. And much of this spending must be targeted to rural facilities serving the poor. 3   CONTENTS 1. Introduction 2. Education A. Education Policy and Expenditures B. Education Outcomes C. Incidence of Education Spending 3. Health A. Health Policy and Expenditures B. Incidence of Health Spending 4. Comparisons With Other Countries 5. Social Sector Subsidies and Household Spending A. Social Subsidies and Household Expenditures B Household Spending on Publicly Provided Education C. Household Spending on Publicly-Provided Health Care D. Household Social Sector Accounts 6. Summary and Policy Conclusions Annex 4   INTRODUCTION Economic growth has been restored in Ghana in recent years. Assessments of its economic performance (such as World Bank, 1995a) have generally been favorable. The challenge that the country now faces is to accelerate growth in a sustainable way (World Bank, 1992). For such growth to benefit the poorer groups in Ghanaian society, they must have access to the necessary productive assets. The rural poor must have access to land, water, credit and markets. The urban poor need productive employment opportunities. Both groups must have access to essential social services that enhance their human capital, such as basic education and health care. Most of the poor rely on the public provision of education and health services, given the often high cost of private- sector provision. For such groups, a key to their participation in accelerated growth, therefore, is their access to publicly subsidized health and education services. Expanding access to basic social services is a key component of a long-term strategy for sustained economic growth and poverty reduction (World Bank, 1990). To what extent is public funding of such services targeted to the poor in Ghana? What does the evidence suggest about the likelihood that the poor will benefit from accelerated growth? Ghana's improved economic performance during the latter half of the 1980s has been attributed to the Economnic Recovery Program (ERP), which has been implemented since 1983. Under the ERP, Ghana has experienced an economic tum-around. Initial policy objectives, including stabilizing the economy and liberalizing trade and price control were successfully achieved. Government revenue as a percentage of GDP rose from less than 6 percent in 1983 to 14 percent in 1986. The inflation rate decreased from 123 percent in 1983 to 20 percent in 1991 and real income per capita rose at an average of 2 percent annually during the same period. According to estimates based on the Ghana Living Standards Surveys (GLSS),1 the incidence of poverty has fallen from 37 percent of the population in 1988 to 31 percent in 1992 (Coulombe and McKay, 1995). Recent history suggests, therefore, that most of the poor in Ghana have benefited from the economic recovery. Under the ERP, there have been changes in public social spending, as well as reforms in both the health and education sectors. Public investments in education and health increased dramatically from low levels after 1983 (Alderman, 1994). The recovery in education spending was more marked than in health. Since then, the share of GDP devoted to these social sectors has remained relatively stable (World Bank, 1995b)-in the region of 3.8 percent for education and 1.2 percent for health (the shares in 1992). Given changes in incomes in general, and in public resources made available to health and education, there are likely to be changes in the utilization of these social services by the poor. Typically, when incomes fall, or when the quality of service deteriorates, households 1 Three GLSS surveys have been completed. The first (GLSS1) covered the period September 1987- August 1988; the second (GLSS2), October 1988-September 1989; and the third (GLSS3), October 1991- September 1992. For simplicity, these are termed respectively 1988, 1989 and 1992 throughout this report. 5  will reduce their utilization. And in a setting of rising incomes and improved funding of services, an increase in utilization would be expected. This paper investigates the implication of these changes for the incidence of social spending to the various groups in Ghanaian society. Expenditure incidence analysis (sometimes referred to as 'benefit incidence') has become an established approach since the pathbreaking work by Meerman (1979) on Malaysia and by Selowsky (1979) on Colombia. There has been a recent resurgence of interest in the approach, reviewed in Van de Walle and Nead (1995).2 Essentially, the method combines information on the allocation of public expenditures with data on the use by individuals of public services, the latter being obtained from a household survey. In this way, it is possible to estimate how public spending on services such as health and education are distributed across the population in general and to the poor in particular. Our analysis relies mainly on data from two rounds of the GLSS in 1989 and 1992. These are combined with data on government expenditures on education and health to generate estimates of the incidence of such spending across income/expenditure groups. The following two chapters review (respectively) the changes in education and health outcomes under the ERP, and estimate the incidence of governrnent spending in these sectors. Chapter 4 compares the incidence estimates for Ghana with those of other countries. Chapter 5 combines government and household spending into a complete set of social sector accounts. Finally., Chapter 6 summarizes the findings and suggests some policy recommendations. 2 Until recently, there have been few applications of this approach to Africa. But see Demery, Castro- Leal and Dayton (1995) for a review of some recent African applications. 6  2 EDUCATION This chapter focuses on the education sector, examining the use of publicly- provided education services by different social groups, assessing the distribution of public subsidies for education, and examining whether government spending has been targeted to the poor. It consists of three sections. Section A describes changes in education policy under the ERP and assesses how it has affected the public financing of education services. Section B reviews education outcomes, based mainly on the GLSS findings. Section C reports estimates of the incidence of public spending on education. A. Education Policy and Expenditures The education system in Ghana has experienced fundamental changes over the past two decades. Ghana had established an essentially British-style academic education system in the 1950s. Until the mid-1970s, the country had one of the most developed education systems in Western Africa. Economic decline in the 1970s, however, seriously affected the quality of education. GNP per capita in constant prices fell by 23 percent between 1975 and 1983, and the education budget as a percentage of GNP declined from 6.4 to 1.4 percent during the same period. Government spending on primary schooling fell from US$41 per student in 1975 to US$16 per student in 1983. Limited resources were devoted to a few high quality institutions and the rest of the education system deteriorated. Most schools found themselves with no books or supplies. Salary levels were insufficient to retain better-trained teachers. Many left the country, being replaced mainly by untrained teachers. Enrollment rates stagnated or fell and little learning was taking place. It is important to emphasize that this deterioration took place before the ERP (Alderman, 1994). Numerous attempts at reforming the education system had failed in the early 1980s -due both to a lack of funds and to resistance from the education establishment. In 1986, the Government of Ghana initiated a comprehensive education reform program with the main goals of improving efficiency, quality and relevance of education. The reforms reduced the length of pre-university schooling from 17 to 12 years. They sought to make education financing more efficient and equitable, and to improve sector management. The structure of the education system was changed to six years for primary, three years for junior secondary (JSS), three years for senior secondary (SSS), and four years for tertiary education. Starting in 1987, the reform was divided into phases: JSS reform from 1987 to 1990 and SSS reform from 1991 to 1993. Reforms at the primary and tertiary levels have been taking place continuously since 1987. Education reform has emphasized basic education-defined as primary and junior secondary education, adult education and literacy, and early childhood development. Under the ERP, government expenditure on education increased significantly, particularly expenditure on basic education. Total government spending on education increased from 7  1.4 percent of GDP in 1983 to 3.8 percent in 1992. The share of recurrent spending on education in the total recurrent government budget rose from 17 percent in 1980/81 to 41 percent in 1993. And recurrent spending on basic education has risen from 44 percent in 1984 to over 62 percent of the total education recurrent budget in 1989. 44 percent of total education recurrent spending was devoted to primary education alone in 1989. Personnel emoluments, however, accounted for 94 percent, and school supplies and maintenance only 6 percent within this budget. This is likely to have adverse effects on the quality of education services. Education financing The education system is financed mainly by the central government in Ghana, even though under decentralization, responsibility for the operation of schools falls under the District Assembly. To contain costs in education, the government froze the hiring new staff between 1987 and 1992, and food subsidies were eliminated at both secondary and tertiary levels. Donor support fbr education, which has increased to an annual average of about 18 billion Cedis, accounted for 9 percent of the total education budget and 15 percent of basic education costs (Ministry of Education, 1994). Since the government budget allocation to education is unlikely to increase in the near future, private expenditure on education is becorning a major new source of financing. Private education at primary and junior secondary levels has expanded rapidly in urban areas. Over 30 percent of total primary and junior secondary enrollments in Accra were in private schools in 1991. School fees in the public sector have increased even though the government has tried to keep the fees to a minimum for primary schooling. The fees officially sanctioned by the Ministry of Education are 500 Cedis per student per year at classes P3 and higher grades, and no fees are set nationally for PI-P2. In practice, all kinds of fees have been increasingly applied in recent years. The annual cost of fees, levies and uniforms often reaches 6,000 Cedis per student. Cost recovery is being applied partially in basic education and fully at senior secondary. University education remains almost free--only student meals are self-financed. A student loan scheme was introduced in 1988 as a mechanism for cost sharing. The loan is to enable students to cover the costs of food, books and transportation. Cost recovery was minimal at tertiary level. The trend of private financing of education will continue as the government encourages private involvement and finance in education, particularly in senior secondary and tertiary education. B. Education Outcomes These changes in the public financing and provision of education services are certain to lead to changes in education outcomes. To what extent have these education reforms in Ghana led to improvements in outcomes? Although they cover only the years 1988-92, the data from the GLSS indicate a noticeable improvement in many indicators. 8  According to the third round of the GLSS (1992), 49 percent of the population aged 15 years and over was literate in English or a local language. The proportion of the population that could read and write was much higher in non-poor households than poor households-61 percent of the non-poor could read compared with only 38 percent of the poor. The urban population had a higher literacy rate than the rural population. And a significant gender bias is reported in the GLSS. The proportion of men who could read, write and calculate in 1989 was more than 35 percent higher than that of women. Gender differences increase among poorer households. Even though overall literacy improved slightly between 1988 and 1992, the gender gap has persisted. School enrollments (as reported by the GLSS) increased between 1988 and 1992. The proportion of the population aged 6 and over that ever attended school increased from 61 percent to 66 percent. Gross enrollment rates at the primary and secondary levels increased between 1989 and 1992 (Table 1). This was especially true of the primary rate, which increased from 79 percent to 88 percent for the population as a whole. The improvement in primary enrollments applies to all groups. Increases in primary enrollment rates are evident among the rural population, among females, and among the poorer quintiles, the only exception being the bottom quintile in urban areas other than Accra. Increases in secondary enrollment rates are not as universal. First, the rate for the rural population as a whole has declined slightly-from 36 percent to 35 percent between 1989 and 1992. Most poor groups (except those in Accra) have experienced declining secondary enrollment rates-and again, enrollment trends in other urban areas are adverse. Surprisingly, there is an increase in the secondary enrollment rate among girls in rural areas, and this even applies to the lower quintiles. Clearly, the reasons for these quite different changes in secondary enrollments across the groups distinguished in Table 1 is a subject requiring further investigation. 9  Table 1: Gross Enrollment Rates by Level of Schooling, Quintile, Region and Gender, 1989 and 1992 Quintile Accra Other urban Rural Ghana Male Female All Male Female All Male Female All Male Female All Primary 1989 1 77 53 63 92 82 87 69 56 63 73 60 67 2 127 96 1018 95 71 82 76 67 72 83 69 76 3 90 92 91 93 69 80 90 75 83 90 74 82 4 97 97 97 107 88 97 96 76 85 99 81 89 5 110 83 96 83 79 81 93 75 84 93 77 85 All 102 86 93 94 77 85 82 68 75 86 72 79 1992 1 96 76 84 92 76 84 79 63 72 83 67 75 2 108 104 107 100 98 99 91 82 87 94 87 91 3 123 79 98 105 97 101 90 81 85 95 85 90 4 105 105 105 102 93 98 91 81 86 95 86 91 5 96 108 10.3 122 95 108 98 95 97 105 97 101 All 106 94 99 103 92 97 88 79 84 93 83 88 Secondary 1989 1 50 21 30) 43 26 35 35 19 28 3.7 20 29 2 45 44 45 37 21 31 38 26 33 38 26 .33 3 53 23 35 46 31 38 42 33 38 43 32 .38 4 73 38 54! 51 31 40 48 36 43 52 35 43 5 67 42 55 46 27 34 51 30 40 54 32 42 All 64 37 49 44 28 36 42 28 36 44 29 37 1992 1 61 25 44 34 18 26 30 21 26 33 20 2 7 2 58 48 53 50 43 46 39 31 35 43 36 40 3 50 23 39 46 35 40 42 30 36 44 31 38 4 60 49 56 55 48 51 46 32 40 50 39 45 5 82 51 64 57 43 49 47 29 39 54 37 45 All 61 42 52 48 38 43 40 29 35 44 33 39 Sources: GLSS2/3 10  Quality of Education Despite these improvements in literacy and enrollment, a number of recent studies have highlighted the continuing weaknesses of the education system-weaknesses which appear to have undermined the quality of the education received. In a participatory poverty assessment for Ghana, Norton et al (1995) highlight the concern of parents about poor quality education services in the public sector, and disappointing school performance on the part of children. There appears to be a problem of teacher discipline-as evidenced by teacher absenteeism, an emphasis by teachers on private tuition as opposed to class teaching, and the reports that teachers often use children to work on land holdings. In short, the evidence suggests that not much teaching actually goes on at the primary level. Concem over the poor quality of education in Ghana has also been expressed by Glewwe and Jacoby (1992). Tests were applied to assess the reading, mathematical, and abstract-thinking skills of all persons aged 9 to 55 covered by the GLSS2 survey. Based on the simplest tests applied, they found that the level of skills acquired through schooling were extremely low. Respondents with even 6 years of schooling obtained mean scores of only 2 out of 8, which is what could be expected if respondents answered the questions randomly. Glewwe and Jacoby (1992) conclude that schooling quality and cognitive achievement can be improved significantly with relatively minor changes to the provision of education services-the provision of blackboards and textbooks, and necessary maintenance. Given the allocations of funds in the past, these inputs have not been provided. Poor comrnunities have proved unable to finance such inputs as school supplies, desks, chairs and school building maintenance which has led to a seriously underfunded component of the service. C. Incidence of Education Spending The changes in incomes, in the behavior of households towards publicly-funded education services, and in education sector policies and funding, will influence how well targeted public education spending has been in Ghana. Do the poor typically use services that have attracted significant public funding? How well targeted is education spending in Ghana? What changes in targeting have been observed in recent years? This section seeks to answer these questions through an analysis of the incidence of public education spending. The allocation of public social spending (to primary, secondary and tertiary levels of service) has important implications for the distribution and targeting of such spending. By bringing together the estimates of the public subsidy at the each level and information on the use of the services across population groups, it is possible to compile a picture of the incidence of public spending. Following the conventions of benefit incidence analysis, each individual currently enrolled in a publicly-funded school is allocated the subsidy embodied in that service. In this way, the analysis maps to individuals the value of the in- kind transfers implied in the public provision of education. Information on the use of 11  education services is available from the GLSS. And since GLSS data cover income, expenditure, location, and so on, it is possible to trace how government spending is appropriated by different groups of individuals-grouping by expenditure categories, by geographical area, and by gender. There are problems of comparability between GLSS2 and GLSS3, including expenditure data which we use here to assign individuals to expenditure quintiles. These problems and the methods used to deal with them, are discussed elsewhere (World Bank, 1995b, Coulombe and McKay, 1995). The data underpinning the analysis that follows have been adjusted to improve comparability. Education subsides Based on data from the Ministry of Education, unit costs of providing school places were estimated for three broad levels-primary, secondary and tertiary schooling. These estimates are reported in Table 2. Some preliminary observations are in order. First, teacher training expenditures and enrollments were included under secondary education, since admission to teacher-training institutes can be gained with only primary- level education completed. Vocational and technical training activities are included under tertiary education. Second, these unit costs are in current prices. The analysis of this paper is based entirely on current-price estimates. Our interest lies mainly in the pattern and distribution of education (and health) subsidies. Third, the data in Table 2 ignore cost recovery in education (because of a lack of information). Finally, lack of data on government education spending by geographical area also prevents an analysis of regional variations the public education subsidy. The same unit cost has been applied to all users at each of the three levels of education. The data suggest a change in the pattern of the public education subsidy. The government has reduced significantly the unit cost (in current prices) to the tertiary sector-a cut of 27 percent. Both primary and (especially) secondary school subsidies have increased sharply. Increasing unit costs of the latter come in part from an apparent decline in Junior Secondary enrollments (from 608,700 to 502,867). Patterns of education spending incidence Given the changes in household school enrollment decisions and in government education spending allocations, what are the implications for the incidence of education spending in Ghana? Table 3 and Figure 1 report the results of the exercise for 1992.3 Overall, the annual subsidy to households through the provision of primary education (based on a per enrollment subsidy of 24,824 Cedis per annum) amounted to 65.8 billion Cedis, or 4,416 Cedis per capita. Males gained more than females (see Annex Table A2). Taking the whole population, they appropriated 55 percent of the total primary subsidy. And this seems to be generally the case across expenditure quintiles and regions. In total, 3 More detailed results are reported in the annex tables (see Tables Al and A2 for education spending incidence). 12  the annual per capita primary subsidy of 4,981 Cedis for boys compares with just 3,883 Cedis for girls. This gender gap was larger among the poorest. Table 2: Public Education Subsidies by Level of Education, 1989-92 1989 1990 1991 1992 Ministry of Education expenditures (million Cedis) Pre-School Education 488.3 1,224.5 1,477.0 4,150.9 Primary Education 21,680.7 24,724.3 32,476.0 44,862.2 Junior Secondary School 5,298.2 14,745.7 19,651.4 31,619.7 Secondary 5,805.0 6,923.2 8,872.5 13,726.9 Teacher Education 926.1 1,186.5 1,576.1 2,997.7 Vocational Training 538.1 687.4 942.0 1,358.3 Technical Education 365.2 592.2 790.0 408.8 University 8,823.9 9,148.0 10,710.3 14,752.9 Total 43,925.4 59,231.8 76,495.3 113,877.5 Enrollments Primary 1,598,400 1,703,100 1799300 1,807,223 Junior Secondary 608,700 625,000 557700 502,867 Senior Secondary incl. Teacher Training 154,500 169,100 194,200 237,759 All Secondary 763,200 794,100 751,900 740,626 Tertiary 18,100 31,100 32,500 42,067 Unit costs by education level Primary 13,564 14,517 18,049 24,824 JSS 8,704 23,593 35,236 62,879 SSS 43,567 40,941 45,687 70,343 Total Secondary 15,762 28,782 40,032 65,275 Tertiary 537,413 335,294 382,839 392,707 Sources: Ministy of Education Expenditure Variance Analysis tables 1989-92MOE Estimates of enrollments. Primary education subsidies appear to have been well targeted to the poor in 1992. The poorest quintile appropriated 22 percent of the total primary subsidy (compared with just 14 percent for the richest quintile). The bottom quintile received on average 4,815 Cedis per annum through primary education subsidies, compared with only 3,100 Cedis per annum for the top quintile. 13  Table 3: Distribution of Public Subtsidies on Education bY Level, Decile, and Region, 1989 & 1992 Total Per capita Column share Total Per capita Column share (000 Cedis) (Cedis) (rl) (000 Cedis) (Cedis) (N 1989 1992 Primary edrcation 1 13,564 Cedis per enrollnent Primary education @ 24,824 Cedis per enrollment Quintile 1 6010794 2178 21.2 14348935 4815 21.8 2 6270083 2272 22.1 15553809 5219 23.6 3 6305441 2285 22.2 14294169 4797 21.7 4 5751505 2084 20.3 12359071 4147 18.8 5 4054340 1469 14.3 9237356 3100 14.0 total 28392163 2057 100.0 65793340 4416 100.0 of which: Accra 1791452 1446 6.3 3505084 2873 5.3 Other urban 6564730 2052 23.1 16192756 4353 .24.6 Rural 20035981 2140 70.6 46095500 4628 70.1 Secondary education @15,762 dis per enrollment Secondary Education @65,2 75 Cedis per enrollment Quintile 1 2341970 849 16.8 10224731 3431 14.9 2 2506318 908 18.0 14977070 5026 21.8 3 3026757 1097 21.8 14449033 4849 21.1 4 3259584 1181 23.4 16129153 5412 23.5 5 2766538 1002 19.9 12768912 4285 18.6 total 13901167 1007 100.0 68548898 4601 100.0 of which: Accra 1547618 1249 11.1 8256590 6768 12.0 Other urban 3232193 1010 23.3 20641475 5549 30.1 Rural 9121356 974 65.6 39650833 3981 578 Tertiary education @ 537,413 Cedis per enrollnent Tertiary Education @392,707 Cedis per enrollment Quintile 1 933925 338 77 1443990 485 6.0 2 466963 169 3.8 2310383 775 9.5 3 2334813 846 19.2 4620766 1551 19.0 4 2334813 846 19.2 4909564 1648 20.2 5 6070513 2199 50.0 10974321 3683 45.2 total 12141027 880 100.0 24259024 1628 100.0 of which: Accra 5136588 4146 42.3 6642352 5444 27.4 Other urban 4202663 1314 34.6 11551916 1630 47.6 Rural 2801775 299 23.1 6064756 609 25.0 All education All education Quintile 1 9286689 3365 17.1 26017656 8731 16.4 2 9243364 3349 17.0 32841262 11021 20.7 3 11667011 4227 21.4 33363968 11196 21.0 4 11345902 4111 20.8 33397788 11207 21.1 5 12891391 4671 23.7 32980589 11067 26.8 total 54434357 3945 100.0 158601262 10644 106.0 of which: Accra 8475658 6840 15.6 18404026 15085 11.6 Other urban 13999586 4376 25.7 48386147 13008 30.5 Rural 31959112 3414 58.7 91811089 9218 57.9 Source: GLSS2/3 14  Figure 1: Per Capita Public Education Subsidies by Quintile and Gender, 1992 12000 10000 8000 a Male Female .> 6000 a 4000 2000 5 1 11 3M1 nl irl l Primary Secondary Tertiary Total education Note: The first five histograms in each category refer to quintiles 1-5 respectively. The sixth reports the mean subsidy for all quintiles. Source: GLSS3 Secondary education subsidies were not so well targeted to the poor. The bottom quintile gained just 15 percent of the subsidy, compared with 19 percent to the richest quintile, and 24 percent to quintile number 4. The reason why the fourth quintile gains more than the richest group is found in the demographic composition of the quintiles (see Table 4 and the text discussion below). It is also the case that the richest quintiles are more inclined to enroll children in private schools at both levels. Not only is the incidence of secondary spending less progressive than primary spending, the results suggest greater gender inequalities. Taking the country as a whole, girls received only 40 percent of total secondary subsidies (Table A2). On average, girls gained just 3,561 Cedis per capita in 1992 from secondary schooling subsidies, compared with 5,702 Cedis going to boys. And this inequality was even more noticeable in some groups, especially those in the poorest two quintiles. Girls in the bottom quintile received less than a third of the total secondary subsidy going to the quintile. Gender inequality, though less in the upper quintiles, is nevertheless still present (girls in the top three quintiles gaining about 40 percent of the secondary subsidy). As expected, public subsidies through the tertiary sector are markedly more regressive than the other education levels. The bottom quintile gained just 6 percent of the total subsidy in 1992. The top quintile, on the other hand, appropriated 45 percent of the subsidy. And over three quarters of the tertiary subsidy accrues to urban areas. Gender bias is also more marked, females receiving just 37 percent of the total subsidy, in the main because few rural households were found to have a female tertiary enrollment in 15  1992-and none in 1989 (Tables Al and A2). But in Accra, there appears to have been no female bias for tertiary schooling. On average, the Ghanaian population is estimated to have received 10,644 Cedis per capita in education subsidies overall in 1992. The incidence was evenly distributed across expenditure groups. The poorest quintile gained 16 percent of the total subsidy, which compares with around 21 percent to each of the other quintiles (Table 3). Females gained 8,614 Cedis, and males 12,803 Cedis per capita (Table A2). And the rural population gained least. On average a rural resident obtained just 9,218 Cedis per head in 1992, while residents in Accra received 15,085 Cedis per capita. Changes in education spending, incidence Has public education expenditure incidence improved over time? To what extent have the changes at both government and household levels since 1989 affected the patterns of education expenditure incidence. The incidence of education subsidies has remained remarkably stable over the 1989-92 period (Table 3). The targeting of public expenditures on primary schooling has remained unchanged-the poorest quintile gaining just under 21 percent of the total. subsidy, and the top quintile receiving around 14 percent in both 1989 and 1992. Regional and gender shares are also mostly unchanged. A reversal in gender public primary education expenditure inequality appears to have occurred in Accra (see Tables Al and A2). Whereas Accra followed the national pattern in 1989, with girls gaining less than boys on a per capita basis, in 1992 they appropriated 52 percent of the total subsidy to the region. This may reflect gender bias of another sort-an increasing tendency for Accra households to provide private-sector education for boys and not for girls. The targeting of spending on secondary schooling deteriorated over the period. 17 percent of the total secondary subsidy in 1989 went to the poorest quintile compared with 15 percent in 1992. The top quintile maintained its share (at just under one fifth). On the other hand, the second poorest quintile improved its share (up to 22 percent from 18 percent in 1989). This makes the distribution of secondary subsidies much flatter across the deciles. Only the poorest decile appears to have disadvantaged in relative terms. Regional inequalities in the public provision of secondary schooling also appear to have widened. In 1989, the mean annual secondary school subsidy in Accra (1,542 Cedis) was 57 percent higher than that in rural areas (938 Cedis). By 1992, the Accra subsidy (6,768 Cedis) was 70 percent higher than the rural subsidy (3,981 Cedis). The distribution of the tertiary subsidy has been fairly stable over time. The poor continued to gain only marginally from public spending in the sector-6 percent going to the bottom quintile :in 1992 compared with 8 percent in 1988. Accra has lost ground relative to other urban areas. The subsidy to rural areas was unchanged at around one quarter of the total. Overall, education spending remained as targeted to the poor in 1992 as it was in 1988. The poorest quintile gained 17 percent of the subsidy in 1989 (16 percent in 1992). The improvement in the targeting of primary education to the poorest group was broadly 16  canceled by a deterioration in secondary expenditure targeting. However, the second quintile gained significantly, in the main because of its increased access to secondary subsidies. This leaves the distribution of education spending very flat across the quintiles. Broadly, quintiles 2 to 4 all gained just over 20 percent of total education subsidies. The poorest 20 percent of the population appear to have benefited least from the reforms of the education system discussed above. The disadvantage of females has remained largely unchanged-gaining about 40 percent of the total subsidy, although there seem to have been a marginal loss for females in the bottom quintile (Tables Al and A2). The most noticeable change in the distribution of the overall education subsidy is a reduction in the share going to Accra (down from 16 percent of the total to 12 percent). This is due entirely to the reduction in tertiary enrollments (note its share of secondary subsidies increased). Other urban areas have gained, while the rural share has remained mostly unchanged. In sum, developments since 1989 have been mixed. There has been improvement in the targeting of primary subsidies, but a deterioration in the targeting of secondary subsidies. Overall, the share of the poorest in education spending has remained unchanged. However, there have been improvements for the second quintile, suggesting that the reforms in the education system are benefiting some poor groups. The net effect is to produce a fairly flat education spending distribution across the quintiles. And there appears to have been little change in gender bias. Females continue to be disadvantaged. Demographic dimensions of the education subsidy Differences in the observed shares of primary and secondary school subsidies may result from demographic differences across the quintiles. For example, the primary school-age population of the poorest quintile was 28 percent of its population in 1992. The corresponding ratio for the fifth quintile was just 16 percent. To what extent is the distribution of public education spending a result of these demographic difference? To answer this question, we assign to all the quintiles the school age structure of the population as a whole, and recompute what the subsidies would be under those circumstances (assuming that the gross enrollment rates for each of the quintiles remain as reported in Table 1). The results are reported in Table 4. If the share of the population of primary-school age was the same for all quintiles, and given observed enrollment rates of each quintile,4 the poorest quintile would have gained only 18 percent of the total primary education government subsidy (in contrast to 4 Note, these are enrollments in public schools only. 17  Table 4: Actual and Predicted Shares of School Subsidies, Under Alternative Demographic and Enrollment Assumptions Share of Education Subsidy Actual Subsidy with no Subsidy with no subsidy demographic enrollment rate differences differences Quintile: (Percentage) Primary 1 21.8 17.9 24.3 2 23.6 21.1 22.5 3 21.7 21.1 20.6 4 18.8 19.8 19.0 5 14.1 20.5 13.7 Total population 100.0 100.0 100.0 Secondary 1 14.9 14.6 20.5 2 21.8 20.4 21.4 3 21.1 19.6 21.5 4 23.5 23.7 19.8 5 18.6 22.2 16.8 Total population 100.0 100.0 100.0 Source: GLSS3; authors' calculations the observed share of 22 percent). The top quintile's share is raised from 14 percent to 21 percent. Demographic factors therefore play a critical role in determining the distribution of the primary subsidy. The distribution of the subsidy under these assumptions is significantly flatter across the distribution, suggesting that the high (low) shares of the subsidy to the poorer (richer) groups arises only because they have significantly greater (lesser) needs-they simply have more (fewer) children of primary school age. This result does not hold for secondary schooling. Removing demographic differences across the quintiles leaves the distribution of secondary school subsidies largely unaffected. 5 To illustrate this point further, it is possible to allow for variations in demographic structure, but to control for differences in household behavior-as reflected in enrollment 5 The main exception being the top quintile, which increases its share from 19 percent to 22 percent-a reflection of a small population of secondary-school age. 18  decisions. Let each quintile have the same gross primary enrollment rate as the population as a whole, but retain its own demographic characteristics. How would this influence the distribution of the subsidies? The answer is given in the third column of figures in Table 4. Comparing the observed distribution with the simulated distribution shows little change in the shares of the primary education subsidy. The main exception is the allocation to the poorest quintile. It would gain 24 percent of the primary subsidy if it had the same enrollment opportunities as the population at large, in contrast to the 22 percent actually gained. Although its share of the primary subsidy is disproportionately high compared with its population share, it is simply not high enough to meet its needs-as reflected in its disproportionate share of primary school-age children. Variations in behavioral (rather than demographic) characteristics are the key to understanding the distribution of the secondary school subsidy in Ghana. With no difference in gross enrollment rate, the poorest quintile would have gained over 20 percent of the secondary subsidy, compared with an observed share of just 15 percent. And the top quintile would have received 17 percent compared with 19 percent. To summarize, the share of the primary education subsidy received by the poorest quintile is determined to a large degree by its demographic characteristics. But the demographic structure had little impact on the distribution of the secondary subsidy to the quintile. If the quintile had the same enrollment opportunities and behavior as the population as a whole, its needs (as reflected in its demographic structure) would point to a much larger share of both the primary and secondary subsidies than was observed in 1992. These demographic differences across the quintiles arise in part because of the selection of per capita total household expenditures as the welfare indicator. If the assignment of individuals to quintiles were based on other welfare measures (such as per adult equivalent expenditures) these demographic differences may decrease, or disappear altogether (Lanjouw and Ravallion, 1994). How sensitive are our estimates of the incidence of public education spending to the welfare indicator selected in distributing individuals across the quintiles? To answer this question, household expenditures were normalized on adult equivalence, based on the scale proposed in Deaton and Muellbauer (1986). The results are reported in Table 5. Interestingly, the incidence of education spending overall is hardly affected by the change in welfare indicator. The poorest quintile, for example, gains 16.7 percent of the education subsidy if the quintiles are defined over adult equivalent expenditures, compared with 16.4 percent when per capita expenditures were used. The main effects of redefining the quintiles are to be found within each education level. Primary education spending becomes significantly less targeted to the poorest groups under the revised welfare measure. Using per adult equivalent expenditures, the poorest quintile gained just 17 percent of the primary subsidy (in contrast to the 22 percent allocation estimated using per capita expenditures). And the richest quintile is seen to gain much more from the subsidy under the alternative welfare measure. The 19  Table 5: Incidence of Education Subsidy Under Alternative Welfare Measure, 1992 Welfare Measure: Adult equivalent expenditures Per capita expenditures Per capita subsidy Column share Per capita subsidy Column share (Cedir) (M) (Cedis) (N) Quintie Primary 1 3847 17.4 4815 21.8 2 4680 21.2 5219 23.6 3 4607 20.9 4797 21-7 4 4601 20.8 4147 18.8 5 4343 19.7 3100 14.0 Total 4416 100.0 4416 100.0 Secondary 1 4269 18.6 3431 14.9 2 486.5 21.1 5026 21.8 3 5284 23.0 4849 21.1 4 4768 20.7 5412 23.5 5 3818 166 4285 18.6 Total 4601 100.0 4601 100.0 Trertir 1 775 9.5 485 6.0 2 1260 15.5 775 9.5 3 1841 22.6 1551 19.0 4 1841 22.6 1648 20.2 5 2423 29.8 3683 45.2 Total 1628 100.0 1628 100.0 All education 1 8891 16.7 8731 16.4 2 10805 20.3 11021 20.7 3 11732 22.0 11196 21.0 4 11210 21.1 11207 21.1 5 10584 19.9 11067 20.8 Total 10644 100.0 10644 100.0 Source: GLSS3 opposite revisions apply to secondary subsidies-the share to the poorest increases and to the richest decreases. The distiibution of the tertiary subsidy becomes markedly more equitable (though remaining highly unequal). The exercise confirms that public spending incidence estimates are indeed sensitive to the definition of welfare. In the case of Ghana in 1992, using per adult equivalence instead of per capita normalization, has made primary subsidies significantly less targeted to the poor, and secondary and tertiary subsidies better targeted. These compensating changes happen to leave the overall education spending incidence unchanged. 20  Closing Caveat While the analysis has shown a reasonably broad distribution of public subsidies in education, the methodology employed here fails to capture one important source of inequality-the quality of education services. Given data limitations, a unit cost of education provision was assumed to be the same in all regions, even though, there are likely to be quite large differences across regions, especially between urban and rural areas. Such variations certainly lead to differences in the quality of schooling received- differences which this analysis is unable to reflect. 21  3 HEALTH This chapter undertakes a similar sequence of analysis for the health sector. Section A reviews the health policy in general, and how it has influenced health spending in Ghana in the recent past. Section B then examines the distribution of government health-care subsidies. A. Health Policy and Expenditures The basic goal of the health sector is to maximize the total health of the nation through achieving universal access to essential health-care services for every Ghanaian. In the late 1970s, the government adopted the primary health care strategy as the vehicle for achieving Health for All by the year 2000. The economic crisis of the early 1980s drastically reduced resources available to the health sector, thereby adversely affecting the health of the population. A health reform program has been carried out under the structural adjustment program to reorganize health care delivery and to promote a more effective and equitable health care system. Major policy strategies include decentralizing health management to local levels, intensifying health campaigns to reduce morbidity and mortality, and improving the quality of services. Health status has improved during the period of economic reform. Infant mortality has fallen from more than 110 per thousand live births in 1980 to about 90 per thousand in 1989. Life expectancy has increased from about 52 years to 56 years over the same period. The results of the 1993 Demographic and Health Survey show that the total fertility rate (TFR) dropped frorn 6.4 in 1988 to 5.5 in 1993, and immunization coverage of children aged 12-23 months increased from 33 percent to 55 percent during the same period. However, despite the progress made, the health status of the Ghanaian population is still among the poorest in the developing world, as indicated by a child mortality rate of 132 per thousand children and a maternal mortality rate of 2.14 per thousand live births. Main features of the health care system There are three main types of health care providers in Ghana-the public sector, non-governmental organizations (NGOs) and the private sector. The Ministry of Health (MOH) is the largest provider of health services, running 2 teaching hospitals, 8 regional hospitals, 48 district hospitals and about 376 health centers and clinics. About 45 NGOs, mainly religious organizations, are involved in the delivery of health care services including 41 hospitals and 64 clinics. The private sector has 12 hospitals, 402 registered clinics, 362 registered pharmacies, and 3,077 licensed chemical sellers (Ministry of Health, 1994). Health facilities in Ghana are unevenly distributed and urban-oriented. Despite government efforts to expand primary health care, only about 50 percent of the rural population live in communities with a health facility, and only about 3 percent where there 22  is a doctor. The average population served per facility varies from 24,284 in Western Region to 5,619 in Volta Region (Ministry of Health, 1992). During the 1980s, Greater Accra, which had about 10 percent of the population, accounted for 45 percent of Ghana's doctors and 23 percent of hospital beds. Health carefinancing The private sector contributes substantially to total spending on health services in Ghana. Private spending accounted for 51 percent of total expenditure on health, the Government, 37 percent, and non-government organizations, 12 percent. Government expenditure on health has expanded during the economic recovery program, both in real terms and as a share of GDP (Alderman, 1994). Total health spending in 1987 was 21.5 billion Cedis or 2.9 percent of GDP. More than 70 percent of the MOH budget is on recurrent spending. The MOH share of total government recurrent expenditure has fluctuated around 8-12 percent between 1986 and 1993. Even though the MOH budget has increased drastically in nominal terms during past eight years, in real terms, and after accounting for population growth, the per capita increases are marginal. More recently, the share of the sector in total government expenditure has declined from 11.6 percent in 1993 to 8.7 percent in 1994. Government health expenditure is unequally distributed across regions-reflecting urban bias. Urban hospitals receive a disproportionate share of resources. In most regions, per capita expenditure on health ranged from 700 to 900 Cedis, while the Greater Accra region enjoyed a per capita expenditure of 2,442 Cedis in 1992. Korle-Bu Teaching Hospital accounted for 35 percent of public-sector health expenditure in Greater Accra. Per capita government expenditures on health in Central and Volta regions were 1,546 Cedis and 1,289 Cedis respectively. Central government is the main source for public health financing, and external donor contributions are the secondary source. In 1992, the central government provided almost 70 percent of funds to the Ministry of Health, extemal aid accounted for 28 percent of the total budget, and less than 4 percent came from user charges. Most of the latter was generated from the sale of drugs (Ministry of Health, 1992). User charges Charges for government health services in Ghana were introduced in 1971 with the Hospital Fee Act. The charges were very low at the beginning, so that only a minimal percentage of total costs was recovered. Under the more recent cost recovery program of 1985, the Government increased user fees substantially at all public health facilities. Patients were supposed to contribute towards consultation costs, and pay in full for drugs, except for vaccinations and the treatment for certain diseases- The initial target of the program was to generate revenues equivalent to at least 15 percent of total MOH recurrent expenditure Initially, the Government specified that 50 percent of revenues 23  would return to the Government Treasury, and that the Ministry of Health could retain the remainder. Of the latter, district and regional hospitals could retain 50 percent and health centers were allowed to keep 25 percent. Since 1990, all facilities retain in full the user- charge revenue. Three categories of the population are exempted from user charges: paupers on the basis of inability to pay; patients with certain communicable diseases such as leprosy and tuberculosis; and Ministry of Health staff and their immediate dependents (Ministry of Health, Legislative Instrument,, 1313, 1985). Revenue generated from user charges recovered 12 percent of the budget in 1987, but the proportion has decreased since. In 1992, the Ministry of Health generated a total of 1.8 billion Cedis in revenues from hospitals and health centers, representing 6.3 percent of the Ministry's recurrent budget. One possible reason for the low level of recovery is that fee levels have not been reviewed since 1985, while prices have increased more than three-fold. Another reason is the exemption system which has been used extensively by MOH staff and their dependents (although not by the poor-Norton, et al, 1995). Utilization of health facilities The use of public health services in Ghana has been low, but has fluctuated over time. Despite increasing health expenditure during the late 1980s, utilization rates improved little. Based on MOH data, total outpatient attendance was estimated at 0.5 visits per capita per year-much lower than other developing countries. The substantial increase in user fees in 1985 immediately reduced the use of services, more so in rural areas than in urban. Since then, utilization rates have gradually increased, so that by 1990, outpatient visits had recovered to the pre-1985 levels in most regions (Waddington and Enyimayew, 1989, Ministry of Health, 1994). The cost of a consultation, distance to the facility and travel costs have a negative effect on the utilization of health care services. But the population appears to be willing to pay for improvements in the availability of qualified care and drugs (Lavy and Quigley, 1993). The GLSS provides information on illness and injury, and actions households take-whether they seek care, and if so, what type of care (private-public, modern- traditional, hospital-clinic, and so on). The basic patterns of reported illness and treatment response for 1992 are shown Figire 2. More detailed results for 1989 and 1992 are given in Table 6 (and in the annex Table A4) .6 6 Because the reference period adopted in the GLSS3 survey (2 weeks) was a half of that used in GLSS2, adjustments were needed to place the results on a comparative footing. Essentially, the proportion of the population reporting ill and injured in GLSS3 was multiplied by 2. 24  Figure 2: Reported Illness and Response by Quintile, 1992 100 0 T 80.04 40.0 __ 20.04 -- _ _ 1 2 3 4 5 El Not IIl [ III- no care I