F. Household Expenditures and Cost Sharing in Education

Survey Data

There have been three rounds of the Ghana Living Standards Survey (GLSS 1, 2 and 3), capturing data in 1987/88, 1988/89 and 1992. There are reported to be a number of problems with the data, particularly in the first two rounds, which have severely circumscribed their usefulness, and there are difficulties with inter-temporal comparisons of the data. Nevertheless, the GLSS provides valuable insights into the education system and its accessibility to the population, as well as to popular perceptions of the system.

My survey was on a very small scale and concerned more with exploring parental finance at the level of the school rather than the household. The survey covered 27 primary, junior secondary and senior secondary schools of different types from around the country, and 96 parents, 158 children and 122 teachers were interviewed. Data were collected for an additional 419 students.

The Ghanaian Household and Cultural Issues Related to Cost-Sharing

Ghana is unusual in that children take responsibility for the payment of fees. This derives from the structure of the Ghanaian family. Child fostering in Ghana is widespread. Sibling data from GLSS 3 have not been published, but evidence is available from earlier rounds. GLSS 1 (1987/88) data indicate that school aged children averaged 4.1 'same-mother' siblings and 5.9 'same-father' siblings, a difference of 1.8, which is an indication of the level of fostering. Children living away from their mother or father have more siblings on average than those living with them.⁸⁸

[⁸⁸ See Lloyd, C. B. & A. J. Gage-Brandon, Does Sibsize Matter? The Implications of Family Size for Children's Education in Ghana, in Ardayfio-Schandorf, E. Family and Development in Ghana, Ghana Universities Press, Accra, 1994, pp 123-159.]

In my survey of the children, a large proportion of the sample were not sure of the parentage of their siblings, and although the data indicate that a little over half the siblings were of the same father and mother, this finding is not reliable. Because the parent's questionnaire was not conducted for the most part in the households with both parents present, most children were 'biological children' of the parent interviewed.

Children thus collect money from both parents as well as relatives to pay fees, and there appears to be less direct financial relationships between parents and schools as in other countries. Indeed, one reason cited for sending children home for non-payment of fees is the belief by head teachers that although the children have collected the money they have 'misused' it: by sending them home parents (real or foster) may become aware of the leakage.

Ghanaians have a great respect for education. Philip Foster, in his classic work Education and Social Change in Ghana, remarked on the 'shrewdness and economic sagacity of the citizens in Ghana', and although his thesis was that choice of education was linked to perceptions of future employment in the public service, it would be a mistake to assume that economic rationality is the only force in people's calculations. My parents' survey asked parents to rank the three main benefits from education, and the responses were interesting in that great stress was laid on education as a means of instilling moral responsibility, 'discipline' (a favourite word), and good behaviour. Jobs were seen as a means towards self-reliance and independence. Indeed, in spite of low salaries and retrenchment, 43 per cent of the parents considered government jobs to be the most desirable kind of job, because of the job security: similarly 63 per cent of the children surveyed aspired to government jobs.

Household and Government Expenditures on Education

GLSS data show that household direct and indirect expenditures on primary education amounted to the equivalent of about 30 per cent of government expenditures, or 23 per cent of total (government plus household) expenditures.⁸⁹ Overall, education non-government household expenditures accounted for the equivalent of about 32 per cent of government expenditures. However, rural households on average contributed the equivalent of about 20 per cent of government expenditures on their primary education provision.

[⁸⁹ These figures may be subject to considerable error. It is extremely difficult to derive household education expenditure data disaggregated by level from the GLSS, for a number of reasons (including the fact that in most tables education is lumped with recreation, which includes significant levels of 'gambling' expenditures). The percentages given in the text are derived from Demery & al, op cit Table 11, in which government and household expenditures have been averaged using total population as the denominator.]

Education Expenditures within Households

Across all households, average household expenditures on education accounted for about 2.8 per cent of total average household expenditures.⁹⁰ Equivalent figures for household expenditures in Kenya and Uganda⁹¹ were 6.7 per cent and 5.4 per cent respectively. The Ghana data seem very low, and may not provide an accurate picture. Expressed as a percentage of non-food spending, Ghanaian households on average spent about 5.7 per cent on education.⁹² The percentage expenditure accounted for by education in total discretionary (non-food) expenditures is comparable to other countries: it is higher than Tanzania but a bit lower than Zambia. As the Zambian data show⁹³, there can be a strong relationship between percentage shares of education in household spending and changes in total spending (income) where total income declines while the fixed elements of education costs must still be met. Over all Ghana this amounted to about 15,000 cedis per household, according to GLSS, though the figure may be low. This may be compared to 11,000 cedis per household spent on weddings, dowries and funerals.

[⁹⁰ Ghana Living Standards Survey: Report on Third Round, Ghana Statistical Service, March 1995. Derived by comparing relevant figures from Tables A9.27 & 28 with Table A9.29.

⁹¹ Report on the Uganda National Integrated Household Survey, 1992-93, Table 1.51.36: Kenya WMS 1993 Table 36a.

⁹² Demery et al, op cit. Table 14.

⁹³ Zambian non-food expenditures fell from a national average of 42 per cent of total household expenditures in 1991 to 31 per cent in 1993, reflecting the rise in poverty over the period. Education in 1991 accounted for about 7 per cent of non-food expenditures, and rose to a startling 16 per cent as families grappled with poverty, drought and adjustment. See Penrose P. L. Harris and G. Bloom, Evaluation of the EU Structural Adjustment Support Programme in Zambia, EU Evaluation Unit, Brussels, August 1996, for an assessment of the relation between economic 'reform' and poverty. Data in table are from Central Statistics Office, Social Dimensions of Adjustment Priority Survey I, Lusaka, 1992, table 10.1 p 119 and Priority Survey II (1994), table 9.1 p 102, adjusted with the CSO annual rate of inflation. For Tanzania see Chapter 4.]

The introduction of health charges in Ghana introduced an additional item of expenditure to household budgets, and most health expenditures, particularly to the poor, are probably non-discretionary.⁹⁴ In 1989 households spent 8.5 per cent of total non-food expenditures on health, rising to 9.4 per cent in 1992, and over the same period education shares in household spending fell (Table 7). Rural households show the strongest relationship between the claims of health and education spending. While their share of health spending in non-food expenditures rose by a little over 18 per cent, the share of education fell by 28 per cent. At the same time, the percentage of non-food spending in total spending rose. Thus a larger proportion of a larger proportion of total household expenditures is being spent on health care, while in total and in rural areas a smaller proportion of a larger proportion of total household expenditures is being spent on education.

[⁹⁴ When health charges were introduced in Ghana in 1985 outpatient visits declined from 4.4 million to 1.6 million. Waddington C. J. and K. A. Enyimayew, A Price to Pay: The Impact of User Charges in the Volta Region of Ghana, Parts I & II, International Journal of Health Planning and Management, Vol. 4 Nr I, 1 989, pp 17 - 47, and Vol. 5 Nr 4, 1990, pp 287-312. It is unlikely that the reduction was a result of a decline in 'frivolous' usage. Utilisation has since climbed back, but with changes in the way people use health services and how they procure drugs: Ghana has a seriously high rate of antibiotic resistance. See also Kraus J. op cit. p 142.]

Table 7: Household Spending on Education and Health, Ghana 1989 & 1992

Household Spending as a Per Cent of Non-food Expenditure
Population Grouping	1989			1992
	Education	Health	Ed + Health	Education	Health	Ed + Health
Total	6.2%	8.5%	14.8%	5.7%	9.4%	15.0%
Urban	6.5%	6.5%	13.0%	6.9%	7.7%	14.6%
Rural	6.0%	10.7%	16.6%	4.3%	11.2%	15.5%
% non-food of total	28.8%			33.8%

Notes & Sources: Demery et al, Table 14 (from GLSS 2 & 3). Expenditures include spending on private provision. Last row from The Pattern of Poverty in Ghana, 1981-1992, GSS, May 1995, Table 6.1, percentage non-food cash spending by the poor.

Table 8: Distribution of Expenditure on Market Purchased Commodities in Rural Areas, Ghana

	1987/88			1988/89			1991/92
	Very Poor	Poor	Non-Poor	Very Poor	Poor	Non-Poor	Very Poor	Poor	Non-Poor
Tobacco	8.7	6.5	5.1	7.6	5.1	4.3	5.8	4.1	2.9
Soap	16.5	15.4	12.5	15.0	12.7	11.2	13.3	12.2	11.8
Charcoal	1.5	1.6	2.7	1.0	1.0	2.6	1.3	1.9	2.7
Other Fuel	8.9	8.7	5.8	9.0	7.0	5.0	11.7	10.3	7.8
Gasoline	0.1	0.0	0.5	0.0	0.0	0.8	0.0	0.5	1.3
Shoes and Clothing	22.1	25.0	24.0	24.8	26.9	25.7	22.8	22.5	19.1
Public Transport	5.3	5.1	5.6	4.3	4.6	5.0	6.2	7.2	9.6
Medicine	6.7	7.4	6.8	6.8	7.1	6.3	9.8	10.4	10.8
Education	8.9	8.4	6.4	9.7	10.0	7.0	7.7	7.1	5.3
Utilities	0.1	0.0	0.4	0.1	0.2	0.6	1.2	1.2	0.1
Others	21.2	21.7	30.1	21.6	25.3	31.6	20.2	22.7	27.5
Total	100.0	100.0	100.0	100 0	100.0	100.0	100.0	100.0	100.0

Notes and Sources: GSS, The Pattern of Poverty in Ghana, May 1995, Table 6.4b

Table 8 shows more disaggregated data by income group in rural areas, over a period of four years. The percentage of non-food expenditure spent on education declines over the period. Higher proportions of expenditure are allocated to transport, fuel, and medicine. This was during a period of major education reform. Unfortunately the GLSS data stop at 1992, and it may be that the conclusions are of little relevance now, depending on the view taken of the economy of the poor. However, there is little reason to suppose that real cash incomes are improving significantly: people still face problems in paying for education.

There are no published data which indicate household expenditures disaggregated by primary and JSS children. Table 9 from GLSS 3 data shows a breakdown of primary education by consumption quintile: the data are available for primary, secondary and tertiary. The first quintile is the poorest and the fifth the 'richest'. Clothing is the largest item for the poorest while food and lodging account for the greatest expenditure in the higher quintiles. Average expenditure in urban areas was over 12,000 cedis per pupil, of which just below a third was for food and lodging and a surprisingly low percentage for clothing, possibly the result of reporting error. Private tuition expenses were also significant. The poorer groups spend half of the highest quintile on books and supplies.

Parental and pupil expenditures

Table 9: Per Pupil Household Spending on Primary Education, Ghana 1992

Quintile	Tuition		PTA Fee		Clothing		Books/Supplies		Transport		Food/Lodging		Other		Total
	Cedis	%	Cedis	%	Cedis	%	Cedis	%	Cedis	%	Cedis	%	Cedis	%	Cedis
1	923	17.9	301	5.8	1,460	28.4	482	9.4	1	0.0	1,281	24 9	700	13.6	5,148
2	1,174	19.5	174	2.9	1,642	27.2	590	9.8	50	0.8	1,500	24.9	898	14.9	6,028
3	1,378	18.1	290	3.8	1,790	23.5	768	10.1	172	2.3	2,373	31.1	856	11.2	7,627
4	1,468	16.5	284	3.2	2,068	23.3	805	9.1	270	3.0	2,904	32.7	1,082	12.2	8,881
5	1,975	18.6	295	2.8	2,383	22.4	1,050	9.9	344	3.2	3,067	28.9	1,503	14.2	10,617
Total	1,331	18 1	264	3.6	1,819	24.7	710	9.6	148	2.0	2,126	28.9	966	13.1	7,364
Urban	2,574	20.9	587	4.8	2,375	19.3	1,196	9.7	420	3.4	3,626	29.4	1,540	12.5	12,318
Rural	800	15.2	127	2.4	1,581	30.1	503	9.6	32	0.6	1,485	28.3	720	13.7	5,248

Notes and Sources: GLSS 3, from Demery et al, op cit, Table A3

The data show that primary school spending by pupils and their families was a little over 7,000 cedis for each pupil, with a spread between 5,000 and 12,000 cedis over rural and urban households. Average public expenditures were about 24,000 cedis⁹⁵, yielding an average ratio of private to public spending on primary of education of about 3.5:1.

[⁹⁵ From MOE accounts: see Penrose P, Budgeting and Expenditures in the Education Sector in Ghana, April 1996]

Table 10 gives average expenditures as collected from a small (and nationally unrepresentative) sample of parents in 1996. The table shows estimates of JSS expenditures from two sources, parents and children. The table also shows average government expenditures for the same year. Government expenditure per pupil was roughly the same as private expenditure per pupil at the primary level, and a little under private expenditures at JSS level.

Table 10: Average Expenditures per Pupil Primary and JSS, Ghana 1995

	Parents' Survey			Average Gov Primary Exp	Parents' Survey			Average Gov JSS Exp	Children's survey
	Total Primary Exp	Nr Children in primary	Average Primary Expend		Total JSS Exp	Total Children in JSS	Average JSS		Average JSS Expend
Mean	123,490	3.3	48,747	35,236	96,231	1.6	62,685	71,954	72,971
Nr parents paying	74		74		83,024		57		79

Notes and Sources: Columns 2-7 as reported by parents; column 8 as reported by children. There were 246 primary and 89 JSS children reported. Government data from MOE calculated in Penrose P. Budgeting & Expenditures in the Education Sector in Ghana (Tables 9 and 15). Note that government actual expenditure (not budget) data are not adjusted for the variation between the school and fiscal years, and are therefore probably slight overestimates.

Comparison of the average primary expenditure in Tables 10 and 9 would only be indicative, because the survey did not collect income and expenditure data, and the sample was not representative and the means are not generally statistically significant. If we take the consumer price index over the period,⁹⁶ the overall average expenditure of 7,364 cedis reported in GLSS in 1992 would be something over 17,000 cedis, compared with nearly 49,000 in Table 10, which in turn would equate to about 21,000 1992 cedis, putting the sample well above the highest expenditure groups. This would be unlikely as 36 per cent of the sample of parents were farmers and 30 per cent of the sample of students was from farming families. 38 per cent of the parents in the sample had had no schooling and were illiterate in all languages. Only another large scale survey with comparable data, combined with the construction of a robust deflator, can reveal whether the real costs to parents of education have been rising. It would, however, be a perfectly reasonable hypothesis that they have. Such a conjecture should be considered in conjunction with the indication above that education spending as a proportion of non-food spending has been falling, while non-food expenditures have risen as a proportion of total spending.

[⁹⁶ 5 12/1320 = 0.427 (GSS).]

JSS seems to represent a greater financial burden to parents than does primary. The sample was too small to determine whether there is any relation between the costs of JSS and parental contributions to primary, but it is likely that there is one. As the table shows, there were on average more children in primary than in JSS in the sample, and while this may not be significant, for a poorer household with five children the burden of cost will rise as the children pass through grade 6.

The average expenditure data are heavily influenced by reported expenditures for food. Tables 11 and 12 show the composition of private expenditure. No fees are officially charged for state primary education. However, both the GLSS and my survey report fees for primary school. The GLSS (question 2A7) lumped registration and fees together - registration is a one-off charge - and this would account for some of the response. In addition, parents pay 'fees' to the PTA, and in many cases pay for private tuition and report this as a fee. It is likely that primary schools are making compulsory charges, or that some parents and pupils are seeing some charges as compulsory.

Table 11: Parental Basic School Expenditures by Main Item, Ghana 1995

	Fees		Uniforms		Food		Supplies		Textbooks		Tuition		Total
Primary
Sample Mean*	5,886	12.1%	9,085	18.6%	25,196	51.7%	3,144	6.5%	2,418	5.0%	1,738	4.0%	48,747
Nr parents paying	74		70		74		70		28		17		74
Mean of payments	6,134		9,604		33,900		3,324		6,389		7,565		48,747
Total number of primary children = 246
JSS
Sample Mean*	6,863	10.9%	11,159	17.8%	26,175	41.8%	7,513	12.0%	5,118	8.2%	1,865	3.0%	62,685
Nr paying	56		55		57		55		23		11		57
Mean of payments	6,986		11,565		26,175		7,786		12,685		9,664		62,685
Total number of JSS children = 89

Notes & Sources: Minority items (not paid by many pupils in this sample but significant in other schools have been omitted, including transport, pocket money, society contributions and 'other'.

* The last row of each section, 'mean of payments', excludes all the 0 values and represents the average payment of all those who paid.

The overall sample mean includes zero values.

Primary school pupils mainly pay for fees (meaning PTA fees and possibly some other types of fee), uniforms, food, supplies. Food, then uniforms, are the largest items.⁹⁷ The size of the fee in the sample is hard to explain, as there are several large entries which cannot be treated as outliers. JSS students pay for more items, of which school clothing and school materials are the most significant. The responses to the questionnaire probably mix the textbook fee and other fees. The food expenditure probably distorts the overall picture: parents were asked for their daily expenditure and this was multiplied by 200 for the annual expenditure. In addition not all food expenditure can be considered incremental. However, if reported daily expenditures on food are multiplied by 100 days (halved) the effect on the overall average is to reduce it by 25 per cent, while the share of food is still dominant (e.g. 35 per cent for primary). Food costs are important, even with an arbitrary adjustment.

[⁹⁷ There is always discussion about how far food costs are legitimately considered part of school expenditures. They are, however, perceived by parents and children as additional, school related, expenditures, and cannot therefore be ignored.]

Table 12 analyses JSS students' expenditures as reported by the students. It is only a little more than that reported by parents, implying some robustness of the two questionnaires. Parents cannot always recall the details of small payments they make during the year as children request specific items. Children also receive money from several sources: in this sample 44 per cent reported receiving money from both parents, 28 per cent from father only, 14 per cent from mother only, and 9 per cent from relatives. It is likely that one parent will not be aware of the full amount collected by the child. It is therefore possible that total private contributions to JSS are more than those reported by parents only.

The level of food payments in Table 11 is supported by Table 12. Nearly all the sample pay textbook fees, food, uniforms and school materials. Minority items are usually clustered by school, as schools differ in their practices. The extreme values are retained as they are not isolated, and it is possible that certain individuals incur considerably higher than average expenditures (note that at the time of the survey $1 US was worth a little under 1,000 cedis, and the extreme values are not extreme in absolute or even purchasing power terms).

Table 12: JSS Student Expenditures by Item, Ghana 1995

	Text book User Fee	Stationery Fee for Exams	PTA dues	Sport	Ex Books	Uniforms	Shoes & Canvas	Other School Mat's	Food	Culture	Furniture	Religious Collection	Bag	Total
Mean	446	2,133	109	117	4,480	8,696	7,273	10,761	23,108	219	2,316	527	4,162	68,750
Maximum	6,000	6,000	750	900	26,000	30,000	39,000	114,000	80,000	6,000	10,000	8,000	20,000	187,160
Nr std'ts paid	68	51	33	34	76	75	75	76	61	29	32	22	58	79
Mean of p'mts	518	3,304	262	272	4,657	9,160	7,661	11,186	29,926	597	5,719	1,891	5,669	68,750

Notes and Sources: Expenditures incurred by a small number of students in the sample have been omitted. These include transport, entertainment fees, and rent. Such expenditures are incurred in other schools. The last row is the mean of those who have actually paid, while the first mean includes zero values.

Secondary Education

We now turn to the costs of secondary education. This is an area which has tended to be neglected over recent years because of the emphasis on basic education by foreign aid agencies. However, although there still lacks a strong body evidence to support the hypothesis, it may well be that increasing the private costs of secondary education create disincentives to attend primary school and have the effect of reducing household expenditures on primary education. There is some evidence of this in Ghana.⁹⁸ Parents and children perceive that there is relatively little return in terms of access to jobs from just a basic education⁹⁹, and therefore recognise the need to stay on at school. Primary school is therefore essentially a route through to secondary education. However, if secondary education costs are perceived to be beyond the ability of parents and children to pay, they may choose to forego the expense of basic education in the anticipation of not being able to continue beyond it.¹⁰⁰

[⁹⁸ Lavy V, Investment in Human Capital, Schooling Supply Constraints in Rural Ghana, LSMS Working Paper Nr 93, World Bank, 1992.

⁹⁹ See Glewwe P, Schooling, Skills, and the Returns to Government Investment in Education: An Exploration Using Data from Ghana, LSMS Working Paper Nr 76, World Bank, 1991. Glewwe argues that returns to primary education in Ghana are (or were at the time of analysis (1989 GLSS)) almost nil.

¹⁰⁰ See World Bank, Access to Education and Health Care in Uganda, June 1996, p 22 for speculation that this effect occurs in Uganda.]

Lavy addresses the criticism raised in Chapter 2 that cross-sectional data are used to predict future behaviour, and that costs are fairly constant throughout the education cycle. His model explores the effect of future costs on the present demand for primary schooling, rejecting the assumption of constant costs. His main result is that the cost of access to post-primary education is the major determinant of parents' decisions to enrol their children in primary school, although the principal component of this conclusion is distance. Direct costs are weak coefficients, while he does not consider indirect costs except for travel costs. Distance is treated in effect as a proxy for cost, with some reason. Furthermore, his conclusions concern rural communities.

As the main cost constraint identified by Lavy was travel to secondary school, he concludes that access to secondary school should be made cheaper through building more schools. Expansion of JSS access is just one such response, and the community secondary school programme was another.¹⁰¹ My survey data are not adequate for the purpose of exploring the relation between the costs of grades 7-9 (JSS) and decisions to attend primary school, but perhaps future GLSS surveys will be better able to address such questions through more appropriate questions. It is likely that costs both to government and households rise sharply at grade 7. It is also reasonable to suppose that one effect of a single basic cycle will be to emphasise constraints to senior secondary school.

[¹⁰¹ This programme, which was part of the Education Sector Adjustment Credit financed by the World Bank, has generally failed and is one cause of the secondary education crisis.]

Costs rise sharply at senior secondary school. Table 13 complements the data for primary education in Table 9, but includes middle, junior secondary and senior secondary, which could not be easily distinguished in GLSS data. The overall average expenditure in 1992 reported by GLSS was a little under 17,500 cedis, 10,000 cedis more than primary expenditure. There is a large 'other' category which is not explained, and the main expenditures are tuition and food and lodging.

Table 13: Components of Household Expenditure per Pupil in Secondary Schools, Ghana 1992

Quintile	Tuition		PTA Fee		Clothing		Books/Supplies		Transport		Food/Lodging		Other		Total
	Cedis	%	Cedis	%	Cedis	%	Cedis	%	Cedis	%	Cedis	%	Cedis	%	Cedis
1	3,285	24.0	725	5.4	2,303	17.0	1,967	14.5	450	3.3	2,474	18.3	2,341	17.3	13,545
2	4,325	24.9	358	2.1	3,220	18.6	2,660	15.3	785	4.5	3,927	22.6	2,070	11.9	17,345
3	5,035	22.7	680	3.1	2,830	12.7	2,788	12.5	1,164	5.2	4,551	20.5	5,175	23.3	22,223
4	5,920	20.8	627	2.2	4,020	14.1	4,494	15.8	2,329	8.2	6,826	24.0	4,221	14.8	28,437
5	7,625	19.6	1,073	2.8	4,785	12.3	4,978	12.8	2,974	7.6	9,400	24.1	8,119	20.8	38,954
Total	5,309	21.7	677	2.8	3,481	14.2	3,447	14.1	1,586	6.5	5,544	22.7	4,397	18.0	24,441
Urban	7,882	23.1	1,069	3.1	4,430	13.0	4,631	13.6	2,522	7.4	7,922	23.3	5,616	16.5	34,072
Rural	3,434	19.7	392	2.3	2,789	16.0	2,584	14.8	904	5.2	3,810	21.9	3,509	20.1	17,422

Notes and Sources: GLSS 3, from Demery et al, op cit, Table A3

Table 14 presents data from the survey on senior secondary students. Average expenditures reported by parents per secondary student were nearly 158,000 cedis, compared with average government expenditure per SSS pupil in the same year of nearly 129,000.¹⁰² This would equate to about 67,500 1992 cedis, again implying a rise in the real cost of education, assuming that the data are comparable, and using the CPI. However, the data in Table 14 include JSS, and so are depressed and not altogether comparable.

[¹⁰² Penrose, Budgeting and Expenditures op cit, Table 9.]

Table 14: Parental Senior Secondary Expenditures by Item, Ghana 1995

	Fees		Uniforms		Food		Supplies		Textbooks		Tuition		Total
Mean	37,281	25.5%	28,616	18.5%	47,932	45.5%	13,905	10.8%	11,852	11.6%	3,435	5.6%	157,987
Nr parents paid	54						44		35		21		54
Mean of payments	40,264		29,156		71,898		17,065		18,287		8,833
Total number of SSS children = 84

Notes & Sources: Minority items (not paid by many pupils in this sample but significant in other schools have been omitted, including transport, pocket money, society contributions and 'other'. The last row excludes zero values.

The ratios of primary, JSS and SSS private expenditures derived from the parental survey data are similar to those for government expenditures. The ratio of private JSS to primary expenditures is 1.4, compared with 1.9 for government expenditures, while for secondary education it is 3.2, compared with 3.5. In other words, it might be 40 per cent more expensive for a parent to send a child to JSS after primary, and over twice as expensive to go on to secondary. The differential between primary and JSS is striking.

Table 15 sets out the complete list of charges reported by the sample of SSS pupils. The expenditures reported by students exceed those reported by parents by a considerable amount, but the difference is accounted for by the influence of boarding fees: the parental sample did not report many children in boarding schools. With boarding fees omitted the average expenditure falls to 151,000 cedis, very similar to that reported by parents. Apart from boarding fees, school materials and clothing make up the largest percentage of expenditures. The most notable discrepancy between the composition of the expenditures in the two tables is the difference in food expenditures: students reported half that of parents. Less than half of the sample reported food expenses, and the average expenditures of those who did report were once again high and close to the parental estimates.

Table 15: SSS Student Expenditures by Item, Ghana 1995

	Mean	Per Cent of Total	Maximum	Number Paying	Mean of payments*
Boarding Fees	63,524	29.5%	288,000	27	181,162
Rent	4,161	1.9%	72,000	21	15,257
Textbook User Fee	6,504	3.0%	9,500	75	6,677
Stationery Fee for Exams	2,445	1.1%	6,000	49	3,843
PTA Dues	8,148	3.8%	18,000	77	8,148
Entertainment Fees	1,000	0.5%	6,000	35	2,200
Development Fund	686	0.3%	5,000	15	3,520
Sports	2,558	1.2%	12,000	77	2,558
Exercise Books	9,481	4.4%	49,000	75	9,734
Uniforms	21,892	10.2%	67,500	76	22,180
Shoes & Canvas	16,256	7.6%	80,000	76	16,470
Other School Materials	34,432	16.0%	227,000	71	37,342
Food/year	22,519	10.5%	100,000	34	51,000
Transport	2,865	1.3%	80,000	4	55,150
Culture	61	0.0%	2,000	10	470
Furniture	6,578	3.1%	20,000	39	12,987
Religious Collection	664	0.3%	8,000	35	1,460
Bag	9,534	4.4%	35,000	59	12,442
SRC	440	0.2%	9,000	10	3,390
Other Expenditure	1,331	0.6%	30,000	9	11,389
Total	215,081	100.0%	451,500	77	215,081

Notes & Sources: Survey of Secondary Students (n=77). *The last column shows the average of actual payments and the first column includes zero values.

The foregoing paragraphs have built up a picture of the considerable level of private costs borne by students. At the basic level students and their parents contribute roughly the same as the government, while at the SSS level students (not including boarding) contribute nearly 20 per cent more than the government. If we apportion the contributions between direct - applied to teaching, learning and school activities - and indirect - clothing, food and transport - the ratio of direct to indirect increases from primary to senior secondary, as Table 16 shows.

Table 16: Ratio of Direct to Indirect Private Costs, Ghana 1995

	Direct	Indirect	Ratio Direct/Indirect
Primary - Parent Survey	12,043	36,653	0.33
JSS - Student Survey	21,242	47,091	0.45
JSS - Parent Survey	20,248	42,129	0.48
SSS - Student Survey (1)	74,329	77,227	0.96
SSS- Parent Survey (1)	69,912	88,075	0.79

Notes and Sources: (1) Boarding costs not included in calculation.

School expenditures

We now turn to the schools themselves, where it will become apparent that private contributions are vital to the running of the school, are usually insufficient, and one possible cause of declining quality. The following analysis mainly relates to JSS and SSS, as analysis of primary school finances is of limited value because nearly all direct expenditures are by the government (apart from construction). At the secondary level schools charge for a number of items, and the income gained is a significant proportion of total expenditures: the boarding school fee for the period was 74,640 cedis.

Table 17: Sample of Secondary School Charges, Ghana 1994/95

Item	JSS			SSS
	Mean	Nr Charging	Range	Mean	Nr Charging	Range
Boarding				99,760	7	40,000 - 117,600
Textbook user fee	375	8	250 - 500	6,300	10	6,000 - 9,000
Stationery fee for examinations	2,825	4	800 - 6,000	6,286	7	2,000 - 12,000
PTA dues	888	8	100 - 2,700	3,375	9	500 - 6,000
Sports	269	8	100 - 600	2,275	9	1,000 - 3,000
Furniture				7,129	7	200 - 20,000
Culture	63	8	50 - 150

Notes & Sources: N = 10 SSS and 8 JSS.

Table 17 sets out a sample of the reported charges by JSSs and SSSs. There appears to be a good deal of variation in what schools charge and how they ask for the charges. Some reduce requirements to a few lump sums, while others disaggregate into small items. Schools in poorer communities appear to restrict the range of charges. Apart from boarding charges in boarding schools, the only non-negotiable charges are textbook and examination fees. In the case of textbook fees head teachers, salaries are blocked if the full sum is not collected, and it is the main cause of sending children home for non-payment.¹⁰³

[¹⁰³ In general the textbook fee is the only statistically significant mean in the sample, reflecting its compulsory nature through the lack of variation.]

Table 18 sets out income and expenditures for a sample of secondary schools around the country. The survey was undertaken between November 1995 and January 1996 and was thus able to collect the data for the previous school year, 1994-95. The fourth column of the table records the income, including salaries, which the school accounts show as received out of the national education budget. Some of the schools earn money from 'internal generation', including farms and collections from old students. Multiplying enrolment by the school charges per pupil gives column 6, the income which a school might expect from parents/pupils if everyone paid the full amount, while the next column shows what they actually paid.

Table 18: Selected Secondary School Accounts, Ghana 1994/95

Type of School	Region	Enrolments	Income from Government	Internal Generation	Total Expected Income from Parents	Total Actual Income from Parents	Total Income	Per cent Income from parents	Income per Student Gov'mt	Total Income per Student
SSS	Ashanti	373	56,047,894	0	18,508,000	12,511,580	68,559,474	18.2%	150,262	183,806
SSS	Upper West	247	24,787,118	0	3,013,400	2,175,200	26,962,318	8.1%	100,353	109,159
SSS d/b	BA	230	35,849,446	492,950	25,355,120	23,227,880	59,570,276	39.0%	155,867	259,001
SSS d/b	Central	783	135,291,613	1,209,270	98,249,700	109,305,811	245,806,694	44.5%	172,786	313,929
SSS d/b	Eastern	887	122,499,463	0	102,825,400	161,366,721	283,866,184	56.8%	138,105	320,030
SSS d/b	Northern	1,094	174,641,010	0	132,436,240	115,299,331	289,940,341	39.8%	159,635	265,028
SSS d/b	Upper West	956	165,824,130	143,000	15,898,800	8,248,000	174,215,130	4.7%	173,456	182,233
JSS only	Ashanti	244	15,315,704	0	1,068,200	1,486,600	16,802,304	8.8%	62,769	68,862
JSS only	Central	150	7,436,673	0	420,000	264,800	7,701,473	3.4%	49,578	51,343
JSS only	Eastern	221	19,437,770	53,250	1,867,450	1,867,450	21,358,470	8.7%	7,954	96,645
JSS only	Northern	1,045	29,668,000	160,000	4,545,750	310,400	30,138,400	1.0%	28,390	28,841
JSS only	Northern	31	7,893,110	7,400	23,250	10,500	7,911,010	0.1%	254,616	255,194
JSS only	Upper West	120	9,338,895	60,000	462,000	0	9,398,895	0.0%	77,824	78,324
JSS only	Upper West	140	8,645,292	140,000	147,000	110,850	8,896,142	1.2%	61,752	63,544
JSS only	Western	241	11,190,884	165,500	807,350	772, 850	12, 129, 234	6.4%	46 ,435	50, 329

Notes and Sources: Penrose survey. pvt = private; b = boarding; d = day. School income includes PTA collections, SRC (Student Council) and Textbook Fees, which are not strictly speaking school income. Comparable data were only available for 15 schools.

There are wide variations in the ratio of income from parents to total school income, which is made up from income from government, mainly for salaries, and from parents. Income from parents includes the textbook charge, although the school does not benefit to the full extent of the amount as the funds are sent to Accra. The textbook fee is in fact the single largest payment by parents and in general exceeds all other single categories of payment, including PTA dues, except boarding fees. If the fee is not included as school income, total school revenues from parents not including boarding fees is reduced by nearly 50 per cent. Thus income from students is a very small percentage of school non-boarding income, as can be seen by comparing the last two columns of the table.

Part of the variation in percentages of parental contributions will be explained by data collection and reporting problems: for example, for northern secondary schools the scholarships do not appear to be reported as government income. However, leaving data difficulties aside, it is likely that secondary schools depend almost entirely on parental contributions for non-salary expenditures. Junior Secondary Schools' subventions are almost entirely salaries. Boarding schools have the highest income and expenditure, and it can be seen that their ratios of school income to government income reflect the fact that students no longer receive boarding subsidies (except for northern students through scholarships). In general internally generated funds are negligible.

Average expenditures are a function of enrolments, and one JSS from the Northern Region in the table with low enrolments but offering the entire cycle has very high expenditures, accounted for almost entirely by teachers' salaries. They are also a function of how many out of the total number of students pay. As the table shows, many schools cannot expect full payment, and the average direct expenditures by students and their parents when calculated by school, particularly for JSS, appear to be lower than those reported by parent and students. Most large scale surveys can provide estimates of private expenditures, but they do not always mean that in each school all pupils pay the average amount. Surveys do not generally relate school accounts to reported household expenditures. Presumably per pupil non-government revenues per school are skewed, with many schools well below the average. These schools are likely to be in poorer catchment areas. Therefore there are probably very wide differentials in average expenditures ('unit costs') per school.

As schools fail to raise, often by considerable margins, the income from parents that they would raise if all fees were paid as expected, we would assume that they would incur debts. Institutional indebtedness is a serious problem in many, if not most, education systems. It is often ignored, and when it can no longer be ignored once-off budgetary provision is made to clear debts. This of course does not solve the problem which created indebtedness in the first place.

My survey also collected data on bank overdrafts and late payment of bills, and it is apparent that schools, particularly senior secondary schools, manage their cash flow problems through running up considerable bills and overdrafts (Table 19). Again, although the data were gathered from school bookkeepers, and checked as thoroughly as circumstances permitted, they must be taken as more indicative than literal. It was not possible to compare debt over time, but the school managers generally confirmed that it increases each year. However, in Ghana as in other countries it is known that institutions build up debt (contrary to financial regulations) against the security of next year's budget, and that suppliers have little choice but to accept the system, particularly in the case of boarding schools in rural areas. This practice is to be expected in survival budgeting practices at which schools have become adept.

Table 19: Debt in Selected Secondary Schools, Ghana 1994/95

Type	Total Unpaid Bills	Current Bank Overdraft	Cash Balance at End of Year	Net Liabilities	(Debt)/Surplus per Pupil
SSS	2,700,000	14,399,786	6,830	(17,092,956)	(74,317)
SSS	0	0	18,900	18,900	24
SSS	500,000	300,000	1,960	(798,040)	(900)
SSS d/b	0	5,160,177	0	(5,160,177)	(16,539)
SSS d/b	6,000,000	12,529,650	50,790	(18,478,860)	(16,891)
SSS d/b	14,091,417	3,661,999	11,412	(17,742,004)	(128,565)
SSS d/b	0	333,326	333,326	0	0
SSS d/b	0	3,540,000	16,982	(3,523,018)	(9,445)
SSS d/b	4,235,500	145,000	685,500	(3,695,000)	(9,724)
SSS d/b	0	4,284,088	885,457	(3,398,631)	(13,760)
JSS only	0	30,000	5,000	(25,000)	(102)
JSS only	0	2,000	0	(2,000)	(13)
JSS only	395,780	100,000	0	(495,780)	(2,243)
JSS only	0	2,692	18,150	15,458	15
JSS only	6,000	0	0	(6,000)	(194)
JSS only	0	0	60,000	60,000	500
JSS only	0	2,500	13,000	10,500	75
JSS only	0	34,000	11,000	(23,000)	(95)

Textbooks

One of the largest components of private expenditures is for the textbook fee. Schools might therefore reasonably expect to have sufficient books, given the level of fee collected. In principle the amount collected, when combined with the budgeted amount, should be sufficient to provide books to all children. Data collected from the sample schools indicate a shortage of the right books. 87 per cent of the teachers said that they did not have enough books for their classes, and over half the teachers reported that up to 75 per cent of their students were without an adequate number of books. Most students said that they had received books from the school, but not for all subjects. The school survey listed the books in the school stock, shown in Table 20, where total stocks are divided by the number of pupils in each grade, as a crude measure of availability per pupil.

Table 20: Textbooks per Pupil in Stock, Selected Schools, Ghana

Type of School	Maths			English			Science
	Grade 1	Grade 2	Grade 3	Grade 1	Grade 2	Grade 3	Grade 1	Grade 2	Grade 3
SSS	1.7	1.3	4.7	2.1	1.3	4.5	0.0	0.0	11.3
SSS	1.2	1.8	1.0	1.1	1.4	1.3	0.0	0.0	4.5
SSS	1.2	2.8	2.3	2.0	2.7	2.7	0.0	0.0	2.6
SSS	0.7	1.3	1.2	0.7	1.3	1.2	1.3	0.0	0.0
SSS	1.2	1.3	1.8	1.2	1.3	1.8	2.0	0.0	0.0
SSS	1.4	1.5	2.1	1.9	1.4	1.9	0.0	0.0	3.1
SSS	0.9	1.7	1.2	1.0	2.0	1.0	0.4	1.0	1.0
SSS	1.1	0.7	1.2	1.1	0.7	1.4	1.1	0.7	1.4
SSS	0.9	0.7	0.8	0.9	0.9	0.8	0.0	0.0	2.8
SSS	1.5	2.1	4.3	2.5	2.2	3.0	0.4	1.0	0.5
Average	1.2	1.5	2.1	1.5	1.5	2.0	0.5	0.3	2.7
JSS only	15	2.1	2.5	2.6	1.3	2.6	2.4	2.0	1.0
JSS only	0.7	0.6	1.2	0.5	0.9	1.0	0.4	1.1	1.8
JSS only	0.7	1.0	1.2	0.4	1.1	0.7	0.4	1.1	1.2
JSS only	0.9	1.4	1.4	1.2	1.3	1.3	0.8	1.3	1.4
JSS only	3.7	1.4	1.1	1.3	0.5	0.6	2.8	1.3	1.3
JSS only	0.2	0.7	0.0	0.0	0.2	0.0	1.0	0.0	0.0
JSS only	0.8	1.2	1.4	0.0	1.0	0.9	0.8	1.4	1.2
JSS only	0.8	1.1	2.2	1.1	1.3	1.0	1.4	1.8	2.1
Average	1.2	1.2	1.6	0.9	1.0	1.2	1.3	1.3	1.4

While it may seem that there are sufficient books, many of the books are donated or old books and not used for the syllabus. The table shows considerable variation between schools (with the newer schools mostly experiencing the shortages: older schools have accumulated more books). The senior secondary schools appeared to be lacking science books. It appears that while teachers and students (40 per cent of whom bought at least one book) complain of inadequate numbers of books, many schools have stocks of books which they do not use because they are not the right books for the course.

In that cost sharing is supposed to provide benefits to pupils and schools, it would seem to be of the utmost importance that this major component of school charges should be seen to work well. It seems to be reasonably efficient in collection, but less so in distribution.

Borrowing

Another indication both of ability and willingness to pay is the level of borrowing and/or sale of physical assets to pay for education. The GLSS3 document does not tabulate these questions, but indicates in the text that 28 per cent of all households were in debt.¹⁰⁴ GLSS did not gather data on asset sales. Education is not mentioned as a reason for loans, but 12 per cent of loans were for health reasons. The sample of university students would be too small for GLSS, and consequently student loans would not show up in the survey. In addition, the fungibility of money would suggest that loans for one item release funds for another. Overall, though, it seems that Ghanaian households do not consider borrowing for education to be a priority, though such a conclusion must necessarily be a weak one for lack of evidence. Nevertheless, the question is an important one, as among other things it is an indication of the perception of the relative returns to education.

[¹⁰⁴ GLSS3 p 106]

Tertiary Education

There is a scarcity of information on how university students finance their education.¹⁰⁵ Of all the education institutions, universities succeed in raising the least fee income, less even than senior secondary schools. Cost recovery at the higher education level is one of the most difficult political challenges facing governments, and there is little history of success in Africa. The costs of higher education are by nature high relative to other parts of the system, and these high costs are compounded by the difficulties experienced in many countries of managing the sub-sector.

[¹⁰⁵ My survey did not cover higher education institutions. One report has been produced by the Technology Centre of the University of Science and Technology, Kumasi (Socio-Economic Background of Students in Tertiary Institutions, 1995 draft), but it contains no information on the subject matter of the title.]

Ghana has a student loan scheme for universities. Since 1988 students in higher education have been entitled to take loans for subsistence and other costs from the Social Security and National Insurance Trust (SSNIT). These loans are repayable at fixed interest, deducted at source from the employees' contribution to social security when they are employed. The loan in 1995/96 was 350,000 cedis, as compared to 200,000 cedis in 1994/95, and the student fixed interest rate was 6 per cent. The government compensates SSNIT for interest foregone at the prevailing Treasury Bill rate, which was running at 45 per cent in 1996: the rate of subsidy was therefore 39 per cent. It is not clear if, or where, the interest rate subsidy is budgeted.

Table 21 shows the mechanism of the subsidy. Government makes an advance to SSNIT to cover the forecast amount of the interest subsidy. The first advance was for the first five years of the scheme to 1994, and it can be seen that the forecast was accurate. SSNIT benefits from the interest earned on the advance which supplements the capital available.

Table 21: Financing the Student Loans Scheme, Ghana, 1989-95

Year	Amount Disbursed (cedis million)	Repayments & Refunds (cedis million)	Outstanding Loans (cedis million)	Treasury Bill Rates	Government Interest Subsidy (cedis million)	Interest Earned (cedis million)	Balance of Government Deposit (cedis million)
Initial Deposit							1,080
1989	418	30	389	20	42	214	1,252
1990	990	97	1,282	23	160	292	1,384
1991	965	177	2,070	30	436	415	1,364
1992	902	376	2,597	19	379	255	(1,239)
1993	2,539	436	4,700	25	837	310	(712)
1994	2,772	931	6,541	28	1,594	197	(686)
1995 (Oct)	6,126	49	12,618	41	3,738		(4,424)
Add interest on 1994 balance @ 40.5 %							(278)
Totals	14,713	2,095	30,197		7,187	1,682	(4,702)

Notes & Sources: SSNIT

Failure to replenish the scheme in 1994 meant that government's arrears to SSNIT stood at the end of 1995 at about 4.7 bn cedis. The projected current cost of the subsidy (i.e. not including arrears) for 1995 was about 3.7 bn, representing about 14 per cent of the total 1995 tertiary subvention. If arrears are included, total additional liabilities on government for the tertiary sub-sector amounted to an additional 40 per cent of allocation, to which should be added scholarships (about 7 billion)¹⁰⁶. Total government expenditure on university education for 1995 was therefore about 34 billion cedis, as compared with 25.6 billion shown in the MOE actual expenditure records, and 22.7 billion budgeted.

[¹⁰⁶ The Scholarship Secretariat, which is headed by a Registrar, is responsible for allocating five types of bursary, four types for secondary school students which generally cover boarding fees, and one type for tertiary students; scholarships for all sixth form students (now being phased out), Merit Awards for the top two students in each secondary school form; automatic scholarships for all northern students, defined as coming from Upper East, Upper West and Northern Province, wherever they attend school; and Hardship Scholarships to students who can demonstrate financial or other needs. Hardship applications are rejected if the student consistently fails academically. At the tertiary level, science students at universities and polytechnics are automatically entitled to reimbursement of their student loans, on condition that they sign their bonds (though this condition appears to have been relaxed), and some other students, including students studying law and journalism, also benefit. Scholarships are not given directly to students, but are reimbursed against invoices from schools and from SSNIT.]

To this picture should be added the observations that real interest rates on T. Bills were negative, and that the opportunity cost of the funds to SSNIT are probably therefore heavily underestimated. While this does not represent a financial cost, it does represent an economic cost, though the volume is small when compared with the total volume of SSNIT transactions.

SSNIT bears the full risk of non-recovery. At present that risk is not seen as a threat, because each student must provide three guarantors who themselves are in the social security scheme. The SSNIT merely has to deduct outstanding payments from guarantor's contributions.

However, the continuing rise in graduate unemployment could prove to be a problem, not so much because students themselves do not get jobs, and therefore do not become members of the scheme, but because perceptions of future unemployment may act as a disincentive to guarantors.¹⁰⁷

[¹⁰⁷ It is interesting to note that because of this threat SSNIT takes an active interest in higher education policy. One response is to encourage greater enrolments in science and technology subjects on the grounds that students with these degrees will have a greater chance of employment, or be better equipped for self-employment. The Ministry also takes this view, which may, however, be questioned in the light of tracer studies and market surveys. One of these shows clearly that employers' demand is for management and business related graduates (Plan Consult, Manpower Survey, Graduate Tracer Study, Accra, Sept 1995). Out of the total sample of 339 vacancies, 3 were for science graduates. Employers' predictions of future vacancies were bleak.]

SSNIT's own ceiling on the increase of the loan is the rate of increase in the index of its basket of salaries, which is a mixture of private and public sector salaries. For 1995 the index was 30 per cent, the previous year being 10 per cent. The government, on the other hand, is constrained by the level of subsidy, which, in view of the fixed level of student interest, is a function of T Bill rates. The current high rates of inflation have a serious impact on the effectiveness and sustainability of the loans scheme. Government has therefore capped future loan increases at 600,000 cedis per student. A more satisfactory student interest rate would reduce government liabilities and enable the loan to be increased.

In addition to loans, which do not provide direct finance to the universities, but are a substitute for student maintenance grants, there are attempts to introduce direct fees for higher education, and some of these are in place, though the amounts are not significant. They include application and registration fees, and 'academic facility fees'. Polytechnics charge fees to part-time students which do represent significant income. In general, secondary education is still more expensive to students than university education.

There are clearly serious questions to be asked about the sustainability of the scheme in spite of the professionalism of its management. The amounts are low and arguably not very significant when compared with the required resources: falling repayment rates must be a real possibility and indeed defaults in due course, and the interest rate burden on government is probably excessive. As the pressure to expand post secondary education increases the loans scheme will face difficult problems.

Willingness to Pay

My survey had questions about willingness to pay in questionnaires for school administrators, school teachers, students and parents. The responses to this type of question are notoriously difficult to interpret, and of course need to be considered carefully. The responses were fairly consistent. School administrators, parents and students gave broadly similar responses to the question on how much more fees could be raised, as shown in Table 22 (teachers appeared to be unwilling to be specific about a figure).

75 per cent of the form teachers questioned (not shown in table) did not think their pupils would be able or willing to pay more fees. About one third of school administrators felt that pupils could pay more than an additional 3,000 cedis, but at the same time they felt that between 40 and 60 per cent might drop out if fees were raised. About 40 per cent of the parents expressed a willingness to pay higher fees, though less than 20 per cent would exceed 3,000 cedis extra, but they would need to see improvements in facilities, teaching, and their children's performance. However, 22 per cent of parents said that under no circumstances would they be willing to pay more fees. Parents in general might be expected to be more reluctant to express a willingness to pay more, but the overall impact of the table is that raising school charges would meet with hostility in the absence of tangible reasons why they should be raised.

Table 22: Extent to which fees can be raised, Ghana

Extent to which fees can be raised	Students	Parents	School Management
No more	39%	30%	33%
up to 3,000 cedis more	15%	21%	33%
up to 5,000 cedis more	18%	8%	11%
more than 5,000 cedis more	27%	9%	22%
Don't know/no opinion	1%	32%	1%

The parental questionnaire had an open-ended question which asked parents what they would do were they the minister for education. The responses were striking in one respect: there was an evident sympathy for the problems of teachers. The belief that education is important for every child is also evident in the responses to this and other questions (the interviewers were encouraged to record opinion on the questionnaires), and there was a widely expressed opinion that costs were a problem for other families if not the respondent him or herself. As another indicator of the attitude towards paying for education, 63 per cent of the parents interviewed did not consider more private schools to be desirable.

The teachers were asked for their opinions on why children in the area were not enrolled at their school, and to rank their answers in order of significance. There were four main reasons which accounted for three quarters of the responses. They were

a) financial reasons
b) wrong type of school
c) not good enough marks in exams
d) poor physical facilities.

The second category covered many disparate reasons, of which the most significant were that parents preferred boarding schools and local children cannot board; religion; ethnic reasons; and a preference for 'old' rather than 'new' schools. The first reason was the dominant response.

When teachers were asked for their explanation for their pupils' absenteeism, the main reason given was a combination of 'lack of interest in schooling' and fees. There did not seem to be a significant relation between absenteeism and the agricultural calendar. 70 per cent of the teachers reported that students from their form had been sent home for non-payment of fees, one quarter of them reporting over 10 students (from classes of about 30). The average length of time before students were allowed back was two weeks. The parents' survey shows 17 per cent of parents stating that their children had been sent home for non-payment, with the students' sample giving a slightly higher figure of 20 per cent. Over half the school administrators said that students were sent home for non-payment, in one case up to 400 students, but in general between 20 and 70. One third of the students said that it was very difficult to raise the money for fees, with another third saying it was difficult: the rest reported no problem. Half of the students reported paying fees late on occasion, and one fifth were always late. The school administrators reported that about 80 per cent of students regularly did not pay their fees by the end of the first term.

Families in most of Ghana set great store by education, particularly in urban and town areas.¹⁰⁸ When pupils are asked under what circumstances their parents and sponsors would be willing to pay more, they invariably reply 'if I do well'. From this it might be deduced that investments in improving quality would result in a greater volume of private contributions. However, there may also be a mismatch between perceptions of quality and the reality. When asked if and why they liked their child's school, a large number of parents replied 'because he/she is doing well'. The survey collected school examination results, and in many cases the same school where children were believed by their parents to be 'doing well' reported near total failures in examination results. Indeed, even teachers, who might be expected to know the examination results, appeared to have a positive attitude towards school performance which is belied by the results. In Ghana examination results are not published, and parents therefore have an imperfect source of information on school performance, at least those aspects of performance measured by examination success, though in the survey teachers reported in some cases that parents chose schools for their children which had better academic reputations, and which also provided extra tuition. Only 20 per cent of the sample said they had never visited the school, and a third said they visited teachers often. A little over half of the teachers rated their own schools as average.

[¹⁰⁸ For one account, see Peil M, Ghanaian Education as Seen from an Accra Suburb, International Journal of Educational Development, Vol. 15 Nr 3, 1995, pp 289-305.]

G. Conclusions

How have government and households reacted to cost sharing? The functional relationship between government expenditure and cost sharing is largely a question of the relative levels of expenditure that occur/would have occurred with and without cost sharing; while the relationship between household behaviour and cost sharing is critically a matter of the effect on enrolment trends. We may now consider the six questions with which we started this paper in the context of education financing in Ghana:

a) Has cost sharing increased total resources available for education?

b) Has cost sharing enhanced efficiency of resource use?

c) Has cost sharing affected enrolments?

d) Has cost sharing improved quality of education?

e) What other effects have resulted from cost sharing in education?

f) Is a policy of cost sharing justified?

Has cost sharing increased total resources available for education?

Real education expenditures have been stagnant in recent years, and expenditures per pupil at the basic and tertiary levels appear to have been falling. There is a lack of evidence on how far household expenditures have increased over the period to compensate. Schools depend on non tax revenues for nearly all non salary costs, and at this level there is little doubt that cost sharing is important. Had there been no textbook fund it is possible but not certain that there would be fewer textbooks in the system: and as far as primary education is concerned the pupil textbook ratios are very low indeed, in most parts of the country less than one.

However, cost sharing may have also enabled government to squeeze education expenditures, which, together with declining efficiency expenditures, could also mean that cost sharing has contributed to a lower level of expenditure than might otherwise have been. The availability of funds from the textbook fund meant that government did not need to budget for JSS and SSS books. There might have been more pressure on the government to allocate teachers better in order to release funds for non-salary learning inputs. Put in another way, the existence of the textbook funds may permit allocative and technical inefficiency of resource allocation to continue. Had government raised the aggregate PTR by 0.5 of a pupil sufficient resources would have been released to dispense with the textbook fund (or to be used for other purposes).

While cost sharing may have resulted in increased resources per pupil, or a slower decline than would otherwise have occurred, it has also involved the establishment of a potentially regressive tax. Cost sharing has been in partial response to the fiscal gap, and in that sense has not resulted in augmented resources except in so far as total expenditures would be lower ceteris paribus without cost sharing.

Has cost sharing enhanced efficiency of resource use?

The efficiency of the education system in Ghana is low, with a high proportion of expenditure on salaries and little on learning inputs. More resources would be available to purchase learning inputs which are at present financed by direct contributions if the system were more efficient, and there is plenty of scope for that to happen. Parents are therefore subsidising inefficiency.

There is no evidence of reallocations from higher levels of education to lower levels - rather the reverse: in terms of average expenditures there has been a trend in favour of secondary schools, but the reasons underlying that trend are connected with attempts to make secondary education more widely available, as well as the effects of the reform of the structure. There is little space to reallocate from higher education, which receives a relatively modest share of the budget and which is politically significant enough to make politicians wary of pushing to hard for more cost sharing, even were that the right thing to do, which is not self evident.

Indeed, there is considerable pressure to increase the share of higher education in the budget, partly on the grounds that basic education receives more foreign aid. There have been no equity effects through allocative adjustments, and, on the basis of the available evidence, cost sharing at post primary levels has not resulted in allocative changes within the budget in favour of primary education, which is one of the main justifications for the policy. Instead, cost sharing at post basic levels is accompanied by increased shares of the government budget for those levels.

At the tertiary level, there are considerable costs attached to the loan scheme, which increase the government's liabilities substantially. These costs, coupled with the low ceiling required by the insurance system, mean that it is unlikely that the level of loan can be increased to any significant amount, and it is not clear just how much social benefit is derived from the scheme. In any case, the university sector is inefficient (in terms of staffing ratios, for example) and student loans should not be the first choice for increasing discretionary resources. Indeed, it is possible that the whole scheme will slowly disintegrate as it becomes less and less relevant to the overall problem of higher education funding.

Has cost sharing affected enrolments?

The data suggest that enrolment ratios have been stagnant or falling, and that cost is a factor. Furthermore there is evidence to suggest that attendance ratios are affected by the textbook fee, although this has now been abolished for basic education. Although the GLSS data showed a rise in apparent and net enrolment ratios between 1988 and 1992, the growth in secondary school enrolment ratios of the poor was slight, and the gap between the richer and the poorer groups' ratios rose over the period. There are no comparable data after 1992 so it is not possible to judge trends over the last four years, except on the basis of extrapolated population data and census enrolment figures.

The removal of boarding subsidies for secondary schools (except for Northern students) resulted in a sharp decline in the number of boarding students and was a contributor to the decline in secondary enrolment. One reason for the failure of many students to enrol in secondary school is the absence of boarding facilities, as well as cost. How far the constraints on secondary enrolment affect primary enrolments is difficult to determine, but some respondents to the survey said that the lack of employment and the high cost of secondary education were factors in their decisions not to continue in basic education. There is no doubt that the absence of affordable accommodation is a key factor in low secondary enrolments and inefficiency of resource allocation at the secondary level.

The question is whether enrolments would have been higher had the direct and indirect costs of education been lower. At the basic education level the indirect costs are dominant, and could only be reduced by abolishing school uniforms, providing school meals and other measures. Similar measures would be required at the secondary level, plus direct funding for non-salary items. At both levels a much higher level of infrastructural spending would be required.

In that Ghana has embarked on a policy of free compulsory and universal basic education (primary + JSS), the definition of 'free' is important, partly because of the morality of compulsion when considerable costs are involved, and partly because nothing is 'free'. In the first instance it means that all direct charges are abolished. Quasi-compulsory charges also need to be reviewed, as well as community responsibility for building.¹⁰⁹ Such policies would only make sense in the context of a fundamental review of the determinants of the costs of education inputs. This would involve simplification of the curriculum, a reform of teacher training, control over the establishment, and other measures.

[¹⁰⁹ The situation is further complicated by District Assemblies' revenue raising rights, and in Ghana the perception is that the 'F' of 'FCUBE' (Free Compulsory Universal Basic Education) has got smaller (it is often printed as 'fCUBE').]

It is reasonable to hypothesise that were direct and indirect costs to parents and their children to be reduced, enrolments would rise, and it remains to be seen whether the abolition of fees has an effect on basic enrolments. However, if an increase in the number of students were not accompanied by scale economies in the provision of inputs (particularly rising pupil-teacher ratios) there would be a strain on the budget. There is also more room over the medium term in the Ghanaian budget to expand expenditure on the education sector than is commonly supposed: the high proportion of education expenditure in discretionary expenditure is more a reflection of the size of the non-discretionary budget than an indication that too much is allocated to education.

Cost sharing has not resulted in increased enrolments via increased resources, and there have been no equity effects. It is more likely to have contributed directly to a reduction in enrolments at all levels, except for tertiary institutions where demand is strong.

Has cost sharing improved quality of education?

Cost sharing policies have had little impact on quality. Nevertheless, the text book revolving fund as well as the near total reliance of the JSS and SSS systems on parental contributions for non-salary costs imply that without cost sharing policies the quality of education provision could have been worse, assuming all other things to be equal. The ceteris paribus assumption is not, however, entirely appropriate, because by presuming on an apparent willingness and capacity to pay in the short run government has succeeded in delaying efficiency improvements in publicly financed provision. In any case, examination results have not been improving, and other evidence suggests a stagnation in school performance.

What other effects have resulted from cost sharing in education?

This question is not examined in detail. Education cost-benefit calculus proposes that investment in education has relatively high returns compared to other investments.¹¹⁰ There is no analysis of the alternative investment choices open to the people who pay school costs, so it is not possible to justify the choice of schooling as the best investment. Would the rate of (non-human) capital accumulation in rural areas increase with lower schooling costs? Would parents of secondary school pupils increase their savings rates? Would the health of the population improve if more private resources were moved to health care?

[¹¹⁰ Curiously, the Ghanaian rates of return show little difference between social and private returns to secondary education, and much higher private than social returns to primary education. This is in spite of loading the cost equation with private costs. At the very least the data would suggest to those who regard them as useful policy indicators that government should invest more in secondary education. They are presented in Annex 3-3 of the Staff Appraisal Report for the World Bank's Basic Education Improvement Programme (May 1996). The author of the annex does not consider Glewwe's paper on GLSS1 data where, like Knight and Sabot in East Africa, the meaning of the returns to low quality education is questioned.]

Among the poor the alternative use of resources is probably limited. The parents surveyed were asked what they would give up in order to pay more for education. About half could not answer or would/could give up nothing. About a fifth of the sample would give up luxury items, though the main component of such items was alcoholic drink. 10 per cent said they would have to work longer hours. There seems to be little slack in household budgets.

Is a policy of cost sharing justified?

In Ghana households provide significant support for the education system, particularly at the post-primary and pre-tertiary levels. How far their support has enabled the education system to expand more than it would have in the absence of such support is not easy to judge.

However, if the reason for raising private costs is that efficiency and equity are improved by user charges, there is some reason to be sceptical that this has actually transpired in Ghana. Cost sharing is a response to fiscal pressure, and the policy may not be justified, particularly in view of the large non-discretionary part of the budget. Not only would there be room for more public expenditure on education with better macroeconomic management, but also the poor management of the education system has involved costs to government and therefore to households which are higher than they need be. In this sense cost sharing may enhance and perpetuate inefficiency and excessive costs.

In summary, few of the objectives of cost sharing policies seem to have been achieved in Ghana. Enrolments have declined, school quality has deteriorated, and expenditure reallocations have not been made in favour of primary education. Private schools have made a modest contribution to enrolment growth, and there may be scope for further private school enrolments, although they would not go a long way towards tackling the low enrolment problems or to releasing state resources for public schools: indeed they would take the better pupils out of state schools (the data on private schools in Ghana are not good, and there is more analysis of private schooling in the next chapter in the context of Tanzania). The school textbook fund assists in secondary school book availability, while the university loans scheme is a well run example of its kind, but they both have had relatively little impact.

In many respects Ghana has not laid great emphasis on cost recovery except in the specified areas of student loans and textbooks, but at the same time the dependence of schools on non government funds is great. As a result schools are under provided and in poor physical shape. There would be little justification for further emphasis on cost sharing until some of the major systemic issues are resolved.