III Case Studies in Cost Sharing - Ghana

A. Introduction
B. Overview of the Ghanaian Education System
C. The Economy, Public Finance and the Education Sector
D. Trends in Recurrent Education Expenditures in Ghana
E. Distribution of the Benefits of Public Expenditure
F. Household Expenditures and Cost Sharing in Education
G. Conclusions

A. Introduction

This chapter and the next are case studies of cost sharing in education. The first case study is of Ghana, and the next chapter covers Tanzania. My Ghana survey was undertaken between 1994 and 1996, and the large scale survey data are from the Ghana Living Standards Survey of which the latest round collected data for 1992.⁷⁵ As I have already emphasised, my small scale surveys are intended to yield qualitative insights and are not statistically representative in that they can be used for prediction or statistical probability analysis.

[⁷⁵ The research for this chapter was carried out by staff of the Ministry of Education. They were Steven Agyarkwa, Eunice Dappah, Nana Dwomoh, Gottfried Gome, Daniel Konadu, and Daniel Zogblah, supported by Yaw Dwomoh and R. J. Mettle-Nunoo. Florence Hianno and Herbert Gorman checked and entered the data on computer. I am also grateful to H. N. Pandit for his assistance.]

B. Overview of the Ghanaian Education System

Structure and Curriculum

The school system is divided into four levels. Pre-school education lasts for two years, and is mainly enjoyed by urban children. Primary education, from the age of 6 to the age of 12, lasts for 6 years, followed by Junior Secondary School (JSS) which lasts for three years. Primary and JSS are considered together as Basic Education, and the government is embarking on a programme of compulsory basic education for all children. For most children Basic Education is terminal.

The curricula are, by general consent, too large. There appears to be a consensus that there are 'too many subjects' in basic school, and that they should be reduced from the present nine subjects at primary level to five or six subjects, and from the present JSS total of 13, of which 12 are examined in national examinations, to 12, with 10 nationally examined. At the SSS level the problem is equally acute.

The importance of this issue from a financial viewpoint is simply that it is the number of subjects which determines the number of teachers required at the post primary levels, and which therefore is a significant determinant of educational costs. The legacy of years of curriculum development and reform in many countries is large curricula which impose both substantial burdens of learning on children and high costs on parents and the state.

Enrolments and Staffing

Table 2 shows trends in enrolments. Up to 1991 there had been decline in the apparent enrolment ratio (AER)⁷⁶, which reached about 69 per cent in 1989. The basic education enrolment has been fairly constant since, though there appears to be a drop in the primary AER, which, if true, will cause a decline in the overall basic AER over the next few years if not offset by large increases in Primary 1 intake.⁷⁷ The data need to be treated with some reservation because population data are extrapolations from the 1984 census. In absolute terms total primary enrolments increased by 148,500 over the period, an increase of a little under 2 per cent a year. The age 6 population growth rate is probably higher than that. Senior secondary enrolments have declined in the same period. The enrolment data are for public schools only, and the AERs include private school enrolments, which have made a modest contribution to enrolment growth.

[⁷⁶ Also known as the Gross Enrolment Ratio.

⁷⁷ Which is unlikely as there have been declining grade 1 enrolments throughout the country.]

Table 2: Enrolments in Ghana, 1991/2 -1995/6

Level		1991/92	1992/93	1993/94	1994/95	1995/96
Primary (public)		1,807,226	1,848,300	1,910,408	1,920,803	1,955,713
	Apparent Enrolment Ratio (public + private)	79%	77%	77%	75%
	Pupil-Teacher Ratio (state schools)	27	30	31	30	32
Junior Secondary (public)		592,867	629,258	655,642	659,85	1677,641
	Apparent Enrolment Ratio (public + private)	56%	58%	59%	58%
	Pupil-Teacher Ratio (state schools)	17	19	18	19	20
Basic Education AER (public & private)		67	77	77	76
Senior Secondary (public)		225,277	247,496	236,530	201,813	194,460
	Pupil-Teacher Ratio (state schools)	22	22	23	20	17
Tertiary			21,947	22,754	23,493	30,331

Notes & Sources: School data from MOE PBME school statistics. Tertiary education comprises Universities and Polytechnics only, data from Dept of Tertiary Education.

The apparent enrolment ratio is the ratio of the total number of children in school to the school age group. It can thus exceed 100 per cent, and is only a general indicator of enrolment performance, depending on the number of 'over-age' children in school. It also provides no indication of what percentage of any given age group actually attend and complete school. Its general use as an indicator of universal primary education is therefore misplaced. Time series for the two other measurements, net enrolment ratios and age-specific enrolment ratios, are not available in Ghana (apart from the two year GLSS comparison shown below). Age data have only recently begun to be collected and analysed in the school census.

Table 3: Primary School Net and Apparent Enrolment Ratios, Ghana 1991/92

Consumption Quintiles
		I		II		III		IV		V		All
		NER	AER	NER	AER	NER	AER	NER	AER	NER	AER	NER	AER
Accra	Male	ns	ns	ns	ns	ns	ns	96.7	146.7	ns	ns	91.3	138.9
	Female	100.0	ns	ns	111.1	92.6	115.6	90.6	88.2	88.6	80.9	87.4	116.6
Other	Male	78.0	111.9	86.2	119.8	88.7	126.8	85.9	126.3	97.3	154.8	86.4	106.4
Urban	Female	63.3	92.9	81.4	116.7	82.8	110.1	85.7	121.4	88.2	121.1	79.7	90.8
Rural	Male	61.3	91.9	81.4	122.0	87.8	123.0	86.0	128.0	86.1	144.4	80.1	101.1
Coastal	Female	55.0	80.0	67.2	93.4	72.2	95.8	80.0	126.7	85.7	128.6	70.3	123.1
Rural	Male	78.7	126.0	89.4	127.8	80.3	109.8	85.6	124.0	93.5	129.9	84.7	123.1
Forest	Female	75.2	99.3	80.6	113.7	85.7	114.3	89.1	118.8	78.8	126.9	81.8	112.3
Rural	Male	45.6	70.5	51.7	82.8	45.1	70.3	60.9	88.0	60.4	98.1	51.3	78.0
Savannah	Female	31.4	46.3	65.6	76.7	43.5	67.1	38.0	52.2	58.9	101.8	45.6	66.0
All	Male	66.6	101.2	77.8	113.3	76.3	110.9	80.5	118.1	87.5	130.7	76.5	113
	Female	58.3	81.4	76.1	104.7	74.3	100.0	74.0	102.8	79.5	120.9	71.5	100
	All	62.6	91.7	77.0	109.4	75.3	105.5	77.4	110.7	83.6	125.9	74.1	107

Notes & Sources: GSS, The Pattern of Poverty in Ghana, 1981-1992, May 1995, Tables 7.4 & 7.5. ns=not significant (small sample). The first quintile is the lowest consumption quintile.

The GLSS provides information on net enrolment ratios (NER) - albeit in some cases on the basis of a small sample - by Accra, other urban and the key ecological zones, not by administrative locality. The NER is the ratio of the number of children of school age in school to the total population of school age: where the denominator of the AER includes all children in school, the NER excludes children outside the 'official' age range. Table 3 summarises the data.

There is a discrepancy between the national apparent enrolment ratio of 79 per cent in 1992 in Table 2 and the 1991/92 AER in Table 3 of 108 per cent. The different data collection methods illustrate the difficulty of the measure. Households report the enrolment of children in school at a given time, whereas the MOE data are derived from school reports and enrolment registers. It appears that the number of children who actually attend school for some time may be greater than the number officially measured.⁷⁸ It may also be that the discrepancy is a measure of a larger rate of early leaving than is generally accepted: at any given time enrolments are higher than those reported in the school census. The ratio of national aggregate NERs to AERs across consumption groups is stable in the area of 65 to 75 per cent, indicating that repetition and enrolment of over school age children is proportionately similar.

[⁷⁸ There are some reports of children attending school without officially enrolling, with the agreement of teachers, largely as a result of inability to pay charges. While this is consistently denied at the school level, it would provide some partial explanation for the discrepancy. At any rate, it has long been well known that Enrolment Ratio data underestimate attendance at school. Other factors contribute. For example, the textbook fee, which until last year was compulsory at all levels but is now abolished for primary pupils, is backed up by severe sanctions on head teachers if not collected. Where pupils did not pay, often for reasons of inability to pay, headmasters did not declare the full enrolment figures in order to avoid comparisons between enrolments and fee collections. This is reported to be a problem still in JSS.]

The GLSS data provide some inter-temporal comparison (Table 4). Net enrolment ratios of the very poor and poor were fairly static between 1987 and 1991. The rate of increase of enrolment ratios was greatest at the higher consumption levels. In 1987/88 the percentage point spread between the lowest and highest consumption quintiles' AERs and NERs was less than 10 per cent: the gap widened by 1991-92 to 20 per cent for NERs and 30 per cent for AERs at primary level. Such a conclusion provides an additional dimension to the interpretation of the Lorenz curves (Figure 4) for the incidence of public expenditure on education, which show overall progressiveness. There was a similar percentage increase in spread at the secondary level, though the percentage point spreads were lower.

Table 4: Apparent and Net Primary School Enrolment Ratios, Ghana 1987-1992

	1987/88			1988/89			1991/92
	Very Poor	Poor	Non-Poor	Very Poor	Poor	Non-Poor	Very Poor	Poor	Non-Poor
Primary Education
Apparent Enrolment Ratios
Male	86.1	94.5	90.1	89.0	100.0	110.0	98.4	110.4	118.1
Female	62.4	75.8	79.1	73.6	87.0	92.5	80.6	92.7	107.5
All	74.9	84.9	84.6	81.6	93.4	101.1	89.9	101.9	113,0
Net Enrolment Ratios
Male	63.9	67.7	66.3	64.2	66.8	78.0	65.8	73.4	80.7
Female	47.4	58.4	60.8	56.0	65.4	67.5	55.8	70.7	76.1
All	56.0	63.0	63.6	60.2	66.1	72.7	61.1	72.1	78.5
Secondary Education
Apparent Enrolment Ratios
Male	40.1	37.5	46.9	37.1	38,0	48.3	39.4	43.4	52.7
Female	23.1	30.5	30.9	22.3	25.4	32.3	25.4	36.2	40.0
All	32.5	34.1	39.2	30.6	32.3	40.4	33.2	40.3	46.5
Net Enrolment Ratios
Male	34.5	33.6	39.6	33.7	34.5	41.5	34.5	37.2	43.5
Female	21.4	29.1	27.7	22.0	24.6	29.5	23.4	33.1	35.9
All	28.7	31.4	33.9	28.6	30.0	35.6	29,5	35.5	39.8

Notes and Sources: Compiled from GSS, The Pattern of Poverty in Ghana, 1981-1992, May 1995, Tables 7.3, 7.4, 7.5 & 7.6.

The geographical variation in enrolment ratios creates inequality in the distribution of education facilities around the country, but the inequality of distribution of facilities measured in this way has been substantially reduced in recent years: the northern regions remain the most disfavoured.

While the primary PTR rose slowly, it varied significantly between and within provinces, as is shown in Table 5, and the variance between the PTR and the average number of pupils per class is also wide. The distribution of enrolments is uneven throughout the country, and the north is particularly disadvantaged. One effect of unequal distribution is that even though the distribution of public expenditure on education may be mildly progressive when measured against shares of aggregate consumption spending, the geographical distribution of expenditures may not be.

Table 5: Enrolments and Teachers by Region, Ghana 1992/3

Region	Total Enrolment	Population Aged 6 - 11	Nr. of Classes	Teachers	Pupils per Teacher	Pupils per Class
ASHANTI	333,940	442,475	11,276	10,999	30	30
BRONG AHAFO	198,825	272,579	8,001	7,653	26	25
CENTRAL	190,751	228,667	6,780	5,863	33	28
EASTERN	272,422	353,662	10,809	10,389	26	25
GREATER ACCRA	180,642	324,493	4,209	4,562	40	43
NORTHERN	136,328	265,686	6,770	5,123	27	20
UPPER EAST	75,758	164,472	2,363	2,012	38	32
UPPER WEST	48,909	91,111	1,888	1,461	33	26
VOLTA	215,496	242,149	8,528	7,545	29	25
WESTERN	195,229	254,536	7,856	7,237	27	25
Totals	1,848,300	2,639,830	68,480	62,844	29	27

Notes & Sources: Enrolment, population & classes from MOE data; teachers from GES. Data sources account for the difference in PTR between Tables 2 and 5.

Thus, enrolment growth has been slow and the proportion of the school age population in school has been declining (other data not shown here indicate that the entry class enrolments in primary school have in many places been declining or stagnant since 1992). These facts are important background to the analysis of cost sharing.

C. The Economy, Public Finance and the Education Sector

The importance of a better understanding of the effect of economic management on the education sector is vital to sectoral policy making in view of the apparent acceptance of resource constraints at the sectoral level and the apparent need to adapt to them. There are features of Ghana's macroeconomic management which seriously affect the level of resources available for sectoral spending. In addition, limited capacity in the general public finance management area means that assumptions about reallocation in the system must be treated with caution. Ghana accepted Bretton Woods stabilisation and adjustment programmes at an early stage. Cost recovery was part of the package with well documented consequences.⁷⁹

[⁷⁹ For an excellent (and critical) account see Kraus J. The Political Economy of Stabilisation and Structural Adjustment in Ghana, in Rothchild D, Ghana: The Political Economy of Recovery, Lynne Rienner Publishers, 1991, pp 119-156.]

National Income and Public Expenditures

The Ghanaian GDP figures are generally accepted to be in need of revision, and there is a programme under way to do so. Similarly, there are questions surrounding the reliability of total government expenditure data. Table 6 is constructed from various sources and the usual data cautions apply.⁸⁰ It appears that the growth of education recurrent expenditures has exceeded that of GDP growth, but that it was lower than the growth of total government expenditure. However, it exceeded the growth of discretionary recurrent expenditures by a little under twice as much. While maintaining a fairly steady proportion of total discretionary recurrent expenditures of just under 40 per cent, these relative growth rates exerted pressure on the budget and depended for their sustainability on the ability to reallocate from other budget items. Internal debt growth was also significant, as interest payments compete for recurrent finance.

[⁸⁰ Nominal expenditures are converted to 1994 prices with the GDP deflator. Another way of looking at a salary based budget is to use a consumer price index. If this is used growth rates are slightly higher. This does not affect the comparative data. Opinions vary about the level of misestimation of GDP, but it may not be as significant as in other countries (including Tanzania).]

Table 6: Shares of Government Expenditure on Education in GDP and Total Budget, Ghana 1990-1997

		1990	1991	1992	1993	1994	1995	1996	1997
		Actual	Actual	Actual	Actual	Actual	Actual	Actual	Provisional
(Actual and estimated expenditures in billion cedis at current prices)
GDP (1)		1,921	2,428	2,803	3,675	4,950	7,418	10,385	13,681
Total Government Expenditure		264	352	509	791	1,010	1,715	2,543	2,830
Share of GDP		13.7%	14.5%	18.1%	21.5%	20.4%	23.1%	24.5%	20.7%
Pensions &c (2)			21	31	65	80	134	187	202
External Debt		10	13	26	44	61	96	145	198
Internal Debt		18	30	25	92	114	233	434	645
Share non discretionary		1.4%	2.6%	2.9%	5.5%	5.2%	6.2%	7.4%	7.6%
Development Expenditure		58	60	105	121	174	323	495	429
Other(3)		5	33	41	67	97	185	187	111
Discretionary Recurrent Expenditure (4)		174	194	281	402	484	743	1,095	1,245
Recurrent Education (5)		59	75	112	156	189	272	446	535
	as % GDP	3.0%	3.1%	4.0%	4.2%	3.8%	3.7%	4.3%	3.9%
	as % recurrent	33.7%	38.7%	39.8%	38.7%	39.0%	36.6%	40.7%	42.9%
Primary & Pre-school Education (6)		16.7	28.6	42.4	56.5	65.8	102.7	154.7	192.9
	as % GDP	0.9%	1.2%	1.5%	1.5%	1.3%	1.4%	1.5%	1.4%
	as % recurrent education	28.5%	38.1%	37.9%	36.3%	34.9%	37.8%	34.7%	36.1%
JSS (6)		8.5	16.6	27.3	38.5	43.9	56.6	94.4	109.2
	as % GDP	0.4%	0.7%	1.0%	1.0%	0.9%	0.8%	0.9%	0.8%
	as % recurrent education	14.6%	22.0%	24.4%	24.8%	23.3%	20.8%	21.2%	20.4%
Secondary		6.2	7.5	12.3	16.9	24.1	31.8	66.7	72.7
	as % GDP	0.3%	0.3%	0.4%	0.5%	0.5%	0.4%	0.6%	0.5%
	as % recurrent education	10.6%	10.0%	11.0%	10.9%	12.8%	11.7%	15.0%	13.6%
Tertiary		6.4	8.0	12.8	17.7	24.9	25.7	49.2	63.5
	as % GDP	0.3%	0.3%	0.5%	0.5%	0.5%	0.3%	0.5%	0.5%
	as % recurrent education	11.0%	10.7%	11.5%	11.4%	13.2%	9.4%	11.0%	11.9%
Education Development			4.0	3.4	4.2	4.7	15.6	15.7	18.6
GDP Deflator (1980=100) (5)		565	680	765	956	1210	1735	2309	2896
			20%	12%	25%	27%	43%	33%	25%
(Actual and estimated expenditures in billion cedis at constant prices
GDP		4,113	4,319	4,433	4,649	4,950	5,173	5,441	5,716
Total Government Expenditure		565	625	804	1,001	1,010	1,196	1,332	1,182
Government Exp/GDP		13.7%	14.5%	18.1%	21.5%	20.4%	23.1%	24.5%	20.7%
Pensions &c		na	38	48	82	80	94	98	84
External Debt		21	23	42	56	61	67	76	83
Internal Debt		37	53	39	116	114	162	228	269
Other		11	58	65	85	97	129	98	46
Discretionary Recurrent Expenditure		372.3	345.6	444.3	508.8	484.1	518.4	573.5	520.3
% GDP		9.1%	8.0%	10.0%	10.9%	9.8%	10.0%	10.5%	9.1%
% Total		65.9%	55.3%	55.2%	50.8%	47.9%	43.4%	43.0%	44.0%
Total Education			140.9	182.0	202.2	193.3	200.5	241.8	231.1
Education Recurrent		125.3	133.8	176.7	196.9	188.6	189.6	233.6	223.3
Primary & Pre-school Education		35.7	51.0	67.0	71.5	65.8	71.6	81.1	80.6
JSS		18.3	29.5	43.2	48.8	43.9	39.5	49.4	45.6
Secondary		13.2	13.4	19.4	21.4	24.1	22.2	35.0	30.4
Tertiary		13.7	14.3	20.3	22.4	24.9	17.9	25.8	26.5
Development Expenditure		124	108	166	154	174	225	259	179
Education Development			7.1	5.3	5.3	4.7	10.9	8.2	7.8
% Total Development			6.6%	3.2%	3.5%	2.7%	4.8%	3.2%	4.3%
	as % total education exp		5.0%	2.9%	2.6%	2.4%	5.4%	3.4%	3.4%

Notes & Sources: From Ghana Country Economic Memorandum, World Bank, May 1995, Tables Al, A2 & A7; Quarterly Digest of Statistics, Ghana Statistical Service, March 1993, Tables 40 & 42, & Broad Based Budget, 1994, Government of Ghana, n.d., Section A p 6. 1993 budget data provisional actuals, 1994 estimates. Education development expenditures from 1994 Public Expenditure Review Table A1.3 (1993&94) and Broad Based Budget Section D.1(c). The development expenditure estimates are inconsistent between sources and very unreliable, and are assumed to be 'narrow coverage' (i.e. domestically financed expenditures only). The mixture of sources also means that some of the trend data should be interpreted with circumspection.

(1) GDP to 1996 from IMF, Staff Country Report, 1997; 1997-98 from 1998 Budget Statement

(2) Includes pensions, gratuities, social security

(3) Includes Common Fund, Environmental Fund, arrears clearance & emergency fund, redeployment & ESB, arrears clearance

(4) Total recurrent less pensions, debt & others

(5) Does not include Common Fund; interest subsidies on student loans; student loans arrears; scholarships; National Service volunteers.

(6) Primary and pre-school normally budgeted and accounted together. Primary budget includes GES admin, not apportioned to JSS.

By international developing countries' standards the proportion of discretionary expenditure allocated to education in Ghana has been high, while the ratio of education expenditure to GDP is comparable to the sub-Saharan average.⁸¹ Total government expenditure as a percentage of GDP has been rising to a level which places it within a comparable range for other African countries. It appears, though, that discretionary recurrent expenditure as a proportion of total government expenditures declined, and this was and continues to be the source of pressure on the overall education budget. The lower growth rate of domestic development expenditures indicates that finance is not being reallocated to investment. While it may be unlikely that education expenditures as a percentage of total discretionary expenditures will rise, the apparent steady decline in education spending as a proportion of total government expenditures shows that an overall reallocation of spending is taking place out of the sector, perhaps to service internal debt costs among others. If such a trend were to continue, it is difficult to see how the proportion of discretionary expenditures could be maintained in the face of competing claims on the diminishing total. The picture is made clearer in Figure 1.

[⁸¹ International comparisons should be treated cautiously because of the well documented problems in computing developing country national incomes; because of the ambiguities created by including capital/development expenditures in total education expenditures; and because of the combination of discretionary and non-discretionary expenditures in the numerator. Nevertheless, the average range of the education/GDP ratio in sub-Saharan Africa is likely to be in the region of 4 per cent; and the share in total government expenditure of education spending to be about 1 5 per cent (World Bank, Priorities and Strategies for Education: A World Bank Review, August 1995, p 66).]

The left hand axis of the Figure 1 measures public expenditures and the right hand axis measures national income. The four lines plot trends in GDP, total government expenditure, government expenditure after items such as pensions and debt payment are subtracted, and education sectoral expenditure. The area between total government expenditure and total government discretionary expenditures indicates the level of non discretionary expenditures. As the table shows, the gap is mainly accounted for by the growing level of internal debt interest payments which are above the line and which therefore compete with the sectors for recurrent finance.

Figure 1: Shares of Government Expenditure on Education in GDP and Total Budget, Ghana 1990-1997

The rate of inflation, reported to have exceeded 70 per cent in 1996 was accompanied by a growing deficit⁸² and level of government borrowing from the central bank. While this is not the place for an analysis of government economic policies, a resurgence of inflation and the failure to take action on its causes has created great difficulties in maintaining a stable sectoral policy framework. Salaries rose by about 30 per cent. Nevertheless, the level of national income implies a greater capacity to finance sectoral budgets than is currently experienced

[⁸² 'Deficit' here refers to the budget balance net of foreign grants but including internal debt interest.]

D. Trends in Recurrent Education Expenditures in Ghana

Total Expenditure Trends

Most government recurrent education expenditure is in the government budget, the main exception apparently being expenditures from compulsory fees such as the textbook fee, which has not been on budget. However, not all education sectoral expenditure is recorded in the Ministry of Education's budget, including secondary schools and university scholarships; the interest subsidy of the student loans scheme; and a few other items. There are also considerable difficulties in disaggregating past kindergarden, primary and junior secondary expenditures.

Figure 2: Trends in Government Real Education Expenditure, Ghana 1990-1997 (Actual expenditures in '000,000 cedis at constant 1994 prices)

Notes & Sources: From MOE tables (PBME).

Figure 2 plots the trends in real actual expenditures (expressed in 1994 prices) from 1990-1995. Expenditures by sub-sector seem to follow the track of total expenditures, with a slight reallocation towards post-basic education. Over the period the average annual growth of secondary education has exceeded that of basic education. Of the major expenditures, only tertiary education is experiencing real growth.

Basic education (including kindergarden⁸³) grew in real terms over the period, though the growth rate was interrupted in 1994, and now appears to be declining again. Secondary education also grew but is also now declining, while post-secondary education showed little growth.⁸⁴ It is worth noting that although there are problems using the GDP deflator to derive real growth rates, it is similar in its order to magnitude to the CPI, and most of the budget is salaries. Growth in the education budget (as much of the Ghanaian civil service budget) is effectively financed largely through inflation taxation on teachers and civil servants.

[⁸³ Note that all Ghanaian data combine pre-school and primary expenditures. It is a serious distortion of primary figures: pre-school expenditures account for about 18 per cent of the figure normally quoted as 'primary'.

⁸⁴ Without including overhead allocations, the growth rates of direct expenditure are higher, indicating that overheads have declined. The data are not given here.]

Average Expenditure Trends

Figure 3: Average Actual Government Expenditures for School Education, Ghana 1990-1996 (Cedis at constant 1994 prices)

Source: Derived from MOE PBME data. Primary costs have been estimated by reducing the figures in Table 6 by 15 per cent to allow for KG costs.

Average Expenditures per Pupil

To make sense of total expenditure movements it is necessary to compare them to average expenditures, where the denominator is both expressed in terms of students and of teachers, and of non-salary expenditures. The calculation of such figures is fraught with difficulty, and they can only be taken as indications of orders of magnitude and trends, and even this may be misleading.

With these caveats in mind, Figure 3 shows the trends in average expenditures, expressed in 1994 prices. Average expenditures per primary pupil appear to have been constant. Over the latter part of the period JSS per pupil expenditures have declined but may be rising again, although they may be underestimated because of reporting difficulties. There has been a fairly sharp rise in secondary student average expenditures since 1993, largely because the fall in enrolments was not matched by a reduction in the number of teachers.

Average Non-Salary Expenditures: Textbooks

Non salary recurrent expenditures on non administration are largely limited to learning materials and school maintenance at post primary levels.

In 1993 an average of cedis 1,276 was spent on textbooks for each child in basic education, in 1994 the figure was cedis 1,011, and in 1995 it was 1,854. Expenditures on senior secondary books were low in 1993, at about cedis 1,292 per student, and in 1994, when 1,022 cedis was spent per student: in 1995 expenditure per student on textbooks was nearly 6,000 cedis. The average cost of a primary education textbook was about 600 cedis and at senior secondary level 3,000-5,000 cedis. The budget would have purchased less than two books for every primary child: JSS children require at present 12 books, while in primary grades 1 - 3 children require books for two subjects and in the following grades for four subjects.

These data do not include expenditures made from the textbook revolving fund. Each senior secondary student contributes 6,000 cedis per year and each JSS student 500 cedis. Primary children no longer have to contribute. Expenditures from the revolving fund amounted to about 1 billion cedis in 1995. At January 1995 the account had a balance of 2.3 billion cedis, and the balance at January 1996 was 2.6 billion. If the revolving fund expenditures are added to the recurrent expenditures, average expenditures on post primary books are higher.

E. Distribution of the Benefits of Public Expenditure

One indication of how far government expenditures are equitably distributed across income groups is how far the share of government expenditure captured by income groups equates to their share of the population. If shares are equally distributed, the poorest 20 per cent of the population would expect to 'capture' 20 per cent of government expenditures, the next 20 per cent the same, and so on. A progressive distribution of government expenditures would have as a minimum condition that each quintile⁸⁵ captures a greater percentage of public expenditure than the quintile below it. If shares are not progressively distributed, the poorest 20 per cent would be expected to capture less than 20 per cent of public expenditures. These relationships can be plotted on Lorenz curves, which are shown in Figure 4.

[⁸⁵ Quintiles - population divided into five equal shares - are commonly measured in terms of aggregate consumption, which is taken to be a proxy for income, as income is difficult to measure because of income in kind. The GLSS categorises consumption groups in terms of 'expenditure'.]

It can be seen from comparing the two charts that there was some flattening of the curves over the period, meaning that expenditure, even for tertiary education, became more progressive. Expenditure on primary education is progressive, with lower quintiles receiving a proportionately larger share of primary expenditure, and it is notable that the incidence of secondary education spending is also very close to the 45 degree diagonal, which indicates equal shares to all quintiles. Indeed, when all education expenditure is aggregated it is only mildly regressive when measured in this way, and education spending is more progressively distributed than is aggregate consumption.⁸⁶

[⁸⁶ The way these data are calculated is described in Chapter 2. There are several reasons why they should be regarded as indicative, and it is not easy to determine whether they over or underestimate progressiveness of the incidence of public expenditures, at least for school education. As most school education expenditure is made up from salaries, the incidence of aggregated public expenditures on education tells us very little about the quality of expenditure. Where there is a greater level of private expenditure to complement the public expenditures, it is fair to assume that quality will be higher than where private expenditures are lowest, at least where the data imply geographic distributions of benefits (because, for example, poorer people would be concentrated in specific locations). Also, there are variations in average expenditures on pupils between regions, and presumably these differences are reflected at the district and finally the school levels. Further insights into variations between schools are given later. There are other problems, such as the choice of unit, which will affect the distribution pattern: whether per adult equivalent measures are used or per capita total household expenditures matters. This is discussed in Demery et al, where alternative measures are given. The charts above are based on actual expenditures and are not adjusted.]

Figure 4: Benefit Incidence of Public Expenditures OH Education in Ghana, 1989 & 1992

Notes & Sources: Computed from data in Demery et al, The Incidence of Social Spending in Ghana, World Bank mimeo, Sept 1995. Tables Al & A2; GLSS3 Table 7.1 for 1992 aggregate consumption

In concluding the general overview of government education expenditures, there are a number of points to note which impinge upon cost sharing. First, it does appear that government has been reallocating away from education over the period, which would imply that any cost sharing measures were a response to a growing fiscal gap in education, and possibly a deliberate government policy (although the state of the macro-economy would be a better explanation). Second, enrolment growth is low, as are enrolment ratios, and additional costs to parents are unlikely to stimulate more enrolment. Third, a reduction in the government's debt obligations would release discretionary resources for education, and a longer term strategy would involve increasing expenditure on education, with less emphasis on short term gap filling.⁸⁷ Fourth, there is little space for reallocation: higher education takes about 12 per cent of the budget, and while there has been a fall in the efficiency the secondary sector, its share of about 16 per cent of the budget is not excessive and it is likely that demand for places will eventually pick up.

[⁸⁷ Although public expenditures are to be reduced as a percentage of GDP.]