2.5.1 Private financing
2.5.2 Cost recovery and user fees
2.5.3 Loans
2.5.4 Some conclusions
National policy on private education and the introduction of various types of cost recovery is an important consideration in policy dialogues with donors. In so far as increases in educational investment can be achieved through increased private expenditures more pupil places can be financed by the public budget and/or quality improvements can be introduced. This is only likely to remain attractive if other educational development indicators do not deteriorate as a result of introducing such policy reforms. This is a current issue for debate since policy to encourage cost recovery is commonly associated with adjustment loans. One fifth of a sample of 50 structural adjustment loans recently analysed included conditions that required the introduction or increase of fees for books and tuition (Stewart 1991 a: 1921) and many countries have independently introduced similar measures.
There is very little detailed data on the distribution of private and public enrolments in most developing countries. Psacharopoulos and Woodhall (1985) suggest on the basis of what is available that there is more provision at secondary rather than primary level, that it has been declining rather than increasing in the recent past (with some exceptions e.g. Tanzania), and that public subsidies for private institutions vary widely from country to country. Data from Lockheed and Verspoor (1990:20) suggest that primary enrolments in private schools in low income countries declined from about 8% in 1975 to 5% in 1985, and increased from 1% to 14% in lower middle income countries (weighted mean without China and India). These trends may have reversed since the mid 1980s since when conditionality for adjustment has favoured more private schooling.
There is no reliable cross country data on private expenditures on education. This is an important gap in data since private expenditures in publicly funded systems may be substantial and comparable with expenditure per child by the state. Moreover very little is known about how private expenditures behave in relation to public expenditures. It will make a great deal of difference whether or not private expenditures remain a constant proportion of public expenditure as public expenditure increases or declines. Perverse results seem possible (Colclough with Lewin 1993). Where private disposable incomes are declining (which has been the case in many Sub Saharan African countries), the ratio of private to public expenditures may have changed in the direction of reducing private contributions. Even where public expenditure has been increased this may have resulted in a net decline in the total amount of expenditure per child on education. In Malawi, Zaire, and Nigeria there appears to be evidence that the introduction of fees has been accompanied by reductions in school attendance (Stewart Moreover, UNICEF (1984) has argued a 2-3% decline in average incomes can easily result in a 10-15% decline in the incomes of the poorest groups and an even larger reduction in their disposable income. This suggests that where total expenditure declines, and where enrolments drop as a result of increased direct costs of schooling, it is likely to be the children of the poorest families that are most vulnerable to withdrawal from school.
Some of the issues raised by cost recovery policies are worth exploring further since it seems likely they will continue to figure prominently in the policy debate. They are discussed in Lewin (1987) and summarised here. Charging fees directly to those who benefit from a service rather than indirectly through the taxation system is presumed to have at least two main kinds of benefit improved accountability and increased resources. Arguably it shortens the chain of accountability between the providing agencies (predominantly schools) and the users of the service. Parents and pupils are expected to value schooling more and place direct pressure on schools and teachers to maintain quality. Fees for educational services may also increase educational expenditure since they represent an addition to public subsidies. Where private schooling is encouraged this reduces the demand on the public system for school places.
There are counter arguments to these presumed benefits. They include the conflict between individual benefit and collective gain, the sophistication of user groups, the impact on participation of charging user fees, and the nature of the service provided at different levels of cost. The mechanism for increased accountability to user groups can be expressed in terms of the benefits parents and pupils hope to obtain from schooling. These reflect the expected returns to individuals in income and social status from schooling and educational qualifications. There is no necessity for this to result in maximising collective welfare; an obvious example is the education of girls which could have considerable impact on child nutrition, infant mortality and population growth, but low apparent rates of return if most women marry early and do not play an active part in the wage economy. Similarly the economic returns to achieving universal literacy for the last to achieve literacy are likely to be low, but the social utility may be high in reducing malnutrition and disease. The kind of conflicts that may arise are described by game theorists as the "tragedy of the commons - what is in the interests of the individual is not in the interests of the collectivity (Hardin 1968). Enhanced accountability also presumes that parents and pupils can discriminate between high and low quality educational services. Amongst those who have not had significant schooling this seems unlikely; even amongst those who have, the quality of what they themselves received may result in ill informed conclusions concerning the value of different methods of teaching and learning.
As noted above charging user fees is likely to have a disproportionate impact on poor families. Such families generally have more members of school age, have less disposable income and experience greater fluctuations from year to year in income than do rich families. They are more likely to be risk averse in investing in education. Real per capita income in more than half the countries in Africa is less than it was ten years ago. User fees are therefore likely to discourage regular enrolment amongst the poorest and adversely affect the enrolment of girls from poor families where they are in competition for declining family income. It may not only reduce access but also contribute to continued poverty since it will exclude the poorest from job opportunities that require educational qualifications which in all societies are positively correlated with income.
Where user fees are encouraged they may also have an unequal impact on levels of provision. Institutions with relatively wealthy catchments may generate sums substantially in excess of those which they are obliged to recover. This increases the differences between schools in ways which favour the already advantaged. Thus in Shri Lanka, fee income varies widely between schools. In a sample of 252 schools in 1988 37% collected no fees. As many as 78% of the schools that did collect fees averaged less than 10 Rupees per child - less than one third what would have been collected had all pupils paid the full amount. The remainder collected amounts approaching the theoretical maximum. School development society income varied even more widely. Over 46% of schools had no such funds, two thirds of the remaining schools collected about 10 Rupees per child, but five well known schools exceeded 100 Rupees per child (Lewin with Berstecher 1989:65).
These are serious objections which need careful consideration before policy decisions are taken. Several authors (e.g. Meesook 1984, Thobani 1983) have argued that some of the most detrimental effects of introducing or increasing user fees on the poorest groups can be offset by sliding scales of charges related to levels of family income. However, the practicalities of doing this are daunting. Incomes are difficult to ascertain reliably in most developing countries, the costs of administration may be such as to absorb much of the gain from charging fees (Ainsworth 1984); school staff are ill-equipped to make discriminatory judgements about the wealth of families and unable to enforce payment without encouraging drop-out and souring relationships with parents whose cooperation they need. The political difficulties of introducing fees, whether they are means tested or not, should not be under estimated.
It has been contended that a situation of low user charges and a low level of service may be worse from an equity point of view than one with high user charges and an expanded supply. Thobani (1983) has argued this in his work on Malawi. Where there is excess demand for school places and insufficient public finance some individuals are denied the service and/or quality suffers. The rich suffer least. Services are denied to marginal areas first, selection through examinations into limited numbers of schools correlates positively with the socio-economic background of students. The proposition is that there is an optimal interim level of user charges that maximises the opportunities for expansion and quality improvement at the lower levels (which benefit the poor most) without a significant deterioration in their limited access.
This is a convenient if not very convincing argument that posits equity in a parabolic kind of relationship with user fees rather than as continuous linear variable. It is conceivable that this might hold where supply is greatly restricted and there is great unsatisfied demand and high rates of return for the successful. But it seems equally plausible that demand is not sufficiently inelastic for user fees to have little impact on participation rates, which is a requirement of the model. The supporting evidence offered is that primary enrolment rates (but not secondary) are highest in the north of Malawi (100%) which is the poorest region, and lower in the richer central (52%) and southern (56%) regions. This does indicate that enrolment is not simply a function of wealth. It does not exclude the likely probability that within those regions the poorest groups will decrease their relative proportion of enrolments with the introduction of user fees.
From another study based on Malawi data (Tan, Lee and Mingat 1984) it is clear that progressive increases in school fees are associated with declining expectations of continued enrolment. The proportion likely to continue is higher among students with better educated parents, from high asset owning families, from urban centres, and from more developed areas of the country. Fathers' annual income has a positive effect on continued schooling at all projected levels of fee increase, though its influence appears to diminish at primary level as fees are raised. Willingness to pay increased fees varies directly with socio-economic background in this study. The differences between expected participation rates of high and low asset owning families are closely related to levels of projected fee increase. Studies of this kind have difficulties separating the independent effect of fees on enrolments since they generally only cover one or two years and fee increases may be accompanied by other changes, particularly in economic conditions, that are relevant. Primary school drop-out does not appear to have increased in Malawi between 1981 and 1983 when fees were raised but this may partly reflect a tendency to continue paying for those already enrolled. Secondary fees were increased by 50% and boarding charges by 150% and about 8% of children from low asset backgrounds and only 4% from high asset backgrounds appear to have dropped out as a result. The proportion of low asset families citing increased school fees as the reason for borrowing increased from 50.6% to 72.8% between 1981 and 1983.
A study on Brazil (Behrman and Birdsall 1983), a country with high drop out and repetition rates, has concluded that it may be better to focus on providing better education to a smaller proportion of the age group than to expand provision to reach everyone. If the costs of keeping one child in school for six years are similar to those of keeping three children in school for two years the former is argued to be likely to result in greater net productivity gains. Fees can only have a limited impact on the problem. Birdsall (1982) argues that a doubling of urban school expenditures could be achieved with fee levels at about 5°/0 of the reported monthly income of urban heads of households. In rural schools, where the need to improve quality is greatest, the fee levels that would be necessary are so high that they would not be sustainable by poor families. In the poorest areas only one third of the poorest children attend school whilst nine tenths of the richest do; any fee levying would almost certainly worsen this uneven distribution.
Loans are another method of transferring costs from the state to individuals. The focus has been on higher education and little emphasis has been given to their use at lower levels. Many countries now operate some form of loan system. Woodhall (1983) identified schemes in 18 Latin American countries, and six in Africa and Asia; by 1992 more than 30 schemes existed (Woodhall 1987:89). Few if any loan programmes appear to be fully financing and it seems unlikely that any can become so in less than 20 years (Psacharopoulos and Woodhall 1985). Loan programmes do work in the sense that they are utilised and that poor students do take advantage of them. However, the Colombian case (Jallade 1974) seems to indicate that they may not be redistributive (since rich students make up the majority of loan takers) and they may serve to channel public finance into private institutions (private universities benefited from the fees paid through government loans). There are also -problems associated with the erosion of the value of repayments by inflation where fixed or zero interest rates are employed; and with defaulters from whom it may not be economic to recover loans through legal sanctions. Loans schemes can transfer significant costs to individuals in the long term, assuming repayment at close to real terms, but such arrangements are politically unpopular, and involve significant start up and administrative costs. It may be that graduate payroll taxes are a better option in many circumstances (Colclough with Lewin 1993).
Charging user fees may be equivalent to increasing the level of taxation in the sense that it is a compulsory contribution to revenue. It tends to be different in the sense that it is typically at a flat rate rather than progressive with income. If it is not then its administration becomes very complex. There is therefore a lot to be said for considering first whether there is scope to increase revenue through increases in taxation that are progressive with wealth before introducing substantial user fees. If there is a genuine commitment by the state to educational provision and to increasing equity, then this is probably the preferable strategy. It may be true that parents will part with a greater proportion of their disposable income if they feel they are contributing directly to their childrens' education however it is clear that the methods for achieving this must be carefully considered and that approaches that may be appropriate where there is good infrastructure, real parental choices between schools, and educationally sophisticated parents cannot easily be transferred to locations where these conditions do not apply.
In conclusion this analysis of cost recovery options should not be taken to imply that there is no case to be made for user charges. For higher education the case is very different than at primary level. Living costs as opposed to learning costs can and usually should be transferred on to students and their families. If some schools can mobilise resources derived from their communities they should undoubtedly be encouraged to use the variety of mechanisms available (Bray 1988). Financial regulations should be constructed to reward initiative and simplify procedures so that they are manageable and facilitating. Experience with mechanisms which generate contributions from individuals and communities outside the public budget indicates considerable disparities in the willingness and ability to contribute between communities and these highlight an important limitation. Such approaches are only attractive if they succeed in increasing resources in both an equitable and efficient way.
