Cover Image
close this bookPlanning and Financing Sustainable Education Systems in Sub-Saharan Africa - Education research paper No. 07 (DFID, 1993, 32 p.)
close this folderPart II: Approaches to reform
View the document(introduction...)
View the documentA. Issues in planning and budget reform
View the documentB. Implementing improved systems
View the documentC. Conclusions: Affordability and sustainability

B. Implementing improved systems

The main problem which in many cases seems almost insuperable without major political commitment is that rational budgeting techniques in education sectors cannot be introduced effectively without reforming total budget systems, across all sectors. The approach to realising reform is described below, and, as I argue, the approach must be at a reasonable pace. Its basic elements are:

(a) the introduction of linked rolling plans and budgets, which involves:

(i) strengthening macroeconomic forecasting for the purposes of overall government revenue estimation

(ii) better estimation of domestic revenues for education

(iii) identification of near-term education sector objectives

(b) strengthened decentralised processes in planning and budgeting, involving:

(i) a more 'pyramidal' planning structure
(ii) identification of 'planning/cost centres' for planning and budgeting purposes

(c) the introduction of programme elements into budgets, which involves:

(i) revisions of budget formats and classifications
(ii) identification of programmes

(d) reform in capital budgeting

(e) better procedures for planning and reporting foreign aid grants and loans

(f) improved expenditure estimation

(g) more accountable budget implementation and improved expenditure control

(a) Rolling Plans and Budgets

Rolling expenditure planning enables governments to have a medium-term perspective of the future financial obligations created by existing policies. A three year perspective is usually considered reasonable and feasible. Revenue and expenditure budgets are made for the current year and for the next two years, and as each year draws to a close, a new third year is added. Rolling plans should be distinguished from development plans, which are mainly concerned with new policies and interventions. They are not substitutes for annual budgeting, but are part of the process of preparing for annual budgets by setting indicative planning targets for the ministerial sectors. They allow sector planners to prepare forward estimates of their financial requirements which are then considered in relation to the predicted available resource ceiling. The technique allows time for sector planners to adjust their plans and expectations to resource constraints, and gives a certain degree of security about the future. In many ways, the adoption of medium term forward budgeting by its very nature pushes the system towards programme budgeting. Most of the policy reforms discussed in the first part of this paper need to be tested in the countries where they are introduced: experience in other countries may not necessarily be relevant, though some have more evidence of success than others. By establishing mechanisms whereby they can be tested without major disruptions governments will have greater flexibility to innovate and also insure against effects of failure. Data deficiencies can be remedied and policy makers alerted in good time to problems which may arise. Rolling planning and programme budgeting are essentially 'bottom-tip' in character, and the effects of policy changes can be evaluated within the normal planning process as managers respond to changes. Improved classification of revenue and expenditure can ensure that there is greater transparency in government budgeting, allowing inter alia the assumptions behind fee and local taxation systems to be tested adequately.

Economic Forecasting. The common and conventional reaction of civil servants accustomed to the old British budget procedures is that government cannot safely predict what it will be able to spend next year, and therefore cannot make advance budgetary commitments. In many countries education sectoral allocations have sometimes not been approved until after the start of the fiscal year. While in some instances this is understandable, there is a certain absurdity to this view when considering education systems and the predictability of their base financing requirements: the main issue is by how much expenditure can be increased. It is possible to make reasonable predictions of government revenues over a three year period, and to make sufficient allocation to education to cover 'core' expenditure, and then to allow for under or over estimation which would mainly effect the rate of increase of education expenditure, perhaps through contingency budgeting. The revenue forecasts can be adjusted every year, and, of course, during the year. The assumptions on which they are based should also be made known so that sector planners can evaluate them and be prepared for changes.

Estimation of Education Sector Revenues. Most education planners would prefer to have some indication of the likely future allocations to education to no indication at all. A weakness of line item budgeting is that it encourages planners to defend existing budget allocations rather than to seek better ways of allocating their revenues: if it becomes apparent that a particular line is no longer important it is difficult for planners to argue that they should keep the sums allocated for transfer to another line, which would then receive a proportionally greater increase than the 'normal' rate of increase.

The base for education planning is enrolments in educational institutions: it is from these that all physical requirements derive. Enrolments can reasonably be predicted, although there is a surprising number of countries which do not have up to date and acceptably accurate enrolment data.59 The more certainty about the level of finance in the near future, the easier it is to plan realistically for institutional maintenance and growth, and also to programme significant changes in systems. Planning on an annual basis can be a barrier to change.

If it is estimated pessimistically, any additional revenue expenditure can be contingently planned, on a percentage basis (for example, 20 per cent of any additional to forecast total revenue will be allocated to education), and/or against contingent budgets (if further resources become available, supplementary budgets which have already been drawn up are ready for implementation). At the end of each fiscal year plans and budgets are prepared for the 'new' third year, on the basis of the sectoral expenditure forecasts given for that year. Without the parallel introduction of rolling multi-annual budgeting it is likely that all budget processes will be pulled back to annual incrementalism. It may also be the case that some of the objections to incremental planning can be overcome by multiyear planning, thereby reducing the scope of necessary changes to the system.60

Consolidated expenditure forecasts should distinguish sources of revenue. Thus government budgets alone may not reflect the full costs of the system, but obligations to finance the system fully must be defined in plans. If this is not done, policy decisions about relative allocations within the sector become meaningless. For example, if the university sector receives one half of its total revenues from domestic resources, and, say, the other half from foreign aid; and if the planned structure of total education expenditure indicates a desired proportion of university expenditure as being no more than 15 per cent of total expenditure; the government budget may reflect such a proportion, but in terms of total expenditure the proportion might reach, say, 25 per cent because of relative favouring of universities by foreign agencies. Once revenue forecasts for education are available, they need to be broken down to the sub-sectoral levels (primary, secondary, tertiary, etc); and then to the different planning/cost centres (see below).

Medium-Term Sectoral Planning. Multi-year budgeting encourages medium-term sectoral planning. It should also discourage an excessive concern with long term vaguely quantified strategic plans (such as ten year plans), which are necessary, but which should be relatively short statements of policy and intent. The failure of most African countries to sustain progress in achieving plans is sufficient warning of the inadequacy of the long term planning approach. The three year plans will be the expression of the detailed direction of the system. If revenues are on a falling trend, then planners are in a position to plan for retraction (or lack of expansion) rather than wait for a suddenly imposed regime of expenditure cutbacks.

(b) Improving the Planning System

A necessary condition for the introduction of better planning and budgeting is improvement of the processes of plan formulation. Under the current systems in most countries, Ministries of Finance in the end are the real policy makers, particularly in times of contraction. There is little point in improving the processes because in most cases the work which goes into developing budgets is wasted when the final budget hearings take place. In other words, planners and the people on whose behalf they are supposed to plan are disenfranchised and subject to last minute arbitrary decisions. This has become a way of life and is accepted with resigned fatalism.

Planning and budgeting should not be passive tasks. Plans and budgets need to be implemented, and accountability for implementation should rest on those who have formulated them. Many, if not most, systems tend to divorce the planner from implementation, and in top-down systems such as those in most African countries, planners (and politicians) make decisions for others to implement without extended and real consultation. Strengthening educational planning and budgeting would include the introduction of a pyramidal structure of planning and budgeting which would involve all relevant players in the process.

The obvious basic unit of planning activity is the education institution. At the primary level the school might in principle be considered as a centre of activity, although primary schools would probably not be feasibly translated into planning/cost centres (a cluster system would be more effective). In many countries this is not dissimilar to earlier education systems which emphasised local control. At the secondary level it is more feasible to introduce an element of school based budgeting and planning.

Pyramidal Planning Structures. In a system which contains a number of layers in the planning and budgeting pyramid, a longer budget time horizon allows planners to devolve many planning functions. Over time more and more disaggregated budget centres could be informed of the resources available to them on a three year rolling basis. The experience of voluntary agencies working in basic education, for example, demonstrates the feasibility of allowing village committees how to decide to allocate given sums of money. The shape of the system is illustrated in Figure 1.

Figure 1: Pyramidal Structure of Education Planning and Budgeting

Overall sector strategy, planning and budgeting

Joint function of central ministries, including finance and education ministries

Sub-sectoral planning budgeting, and expenditure control.

Sectoral ministries.

Regional planning, budgeting and expenditure control.


Institutional budgeting, planning and expenditure control.

Universities and colleges; secondary schools

District budgeting, planning and expenditure control


Sub-district budgeting, planning and expenditures control.

Sub-district agencies, schools, communities.

A strength of this approach lies in the legitimation of the allocation of reduced and insufficient finance for given activities: if communities and institutions can decide how to spend their limited resources rather than have it decided for them they are arguably more likely to 'own' their schools and institutions and contribute to their development. There are some important implications of the approach at the school level, most particularly relating to the training, recruitment and payment of teachers.

The heterogeneity of education systems must also be recognised. Approaches to budgeting at primary, secondary and post-secondary levels should differ, and be related to accountability for budget implementation. While in rural areas primary schools are community institutions, secondary schools and urban primary schools are often less community based: in these cases school governance becomes an issue. Universities are in many ways the easiest institutions in which to initiate budget reform, and they lend themselves well to programme linked budgeting, although experience suggests political rather than practical difficulties work against such reform.

Thus, a key feature of the eventual system would be the final realisation of participation of school communities in decisions about the allocation of subsidies: this has been a paper policy in many countries for a long time. For example, primary teachers may fail to arrive at their postings because of the absence of accommodation, remoteness, and other factors. Many school buildings are in a state of collapse. Over time the teaching profession becomes demoralised by a persistent lack of resources, low pay and difficult conditions, and no amount of training and exhortation will be effective. The mistake is to believe that planners and educationists can always decide what diverse communities want to do about the problem. At the moment, for example, when the teacher does not arrive at post, the salary is clawed back by the Treasury. The school has effectively lost a significant part of its budget. This is because of the bureaucratic division between salaries and non salary costs, which could be modified under a reformed budget system. There is no a priori reason why funds budgeted for salaries cannot be vired to provide a better environment, or buy materials, or in any other way enhance the quality of school life.

Planning/Cost Centres. At present most countries have a structure of district education offices and local government, and it is usually possible to allocate revenues between them. These are in effect 'cost centres' for primary education and sometimes secondary education. At the post secondary levels budgeting is usually done at the institutional level. Under a rolling plan system cost centres can be given a reasonably firm allocation for the next year as well as indicative allocations for the subsequent two years. As planning capacity is extended and strengthened at district and local authority levels, the districts themselves can create further cost centres, as shown in figure 1.

(c) Programming the Budget

I have noted that the main reasons for the failure of programme budgeting to be widely adopted lie in the difficulties that political and institutional authorities have in reexamining the base budget, and in the complex ways in which the technique has tended to be introduced.61 An examination of the programme budget formats which have been tried confirms that they have required a great amount of skill and time. It should be noted that many of the recorded attempts to introduce programme budgeting did not take place in the framework of forward expenditure planning on a rolling basis.

However, programme budgeting should not be regarded as a fixed system: there is no 'right' approach (see the box for an outline description of the main elements). Indeed, the paucity of experience in the application of its techniques ensures that every approach to its introduction will be original, which may have the advantage of ensuring that it is locally relevant. One major difference between the experiences of many countries in the recent past and now is the rapid diffusion of micro-computers over the last decade, particularly the last five years. Its introduction should be gradual and be accompanied by intensive training within the contexts of the institutional changes described above.62 I do not propose major upheavals in budgeting and planning. Experience suggests that the success of past attempts to introduce programme budgeting is more to be found in improved practices after the 'pure' technique is abandoned, rather than in major permanent change to the financial system as a whole.63 It is for this reason that I emphasise in this paper the need for improved planning as it relates to financial requirements to support plans, rather than propose a set technocratic solution. Elements of the budget can be programmed, and education budgets lend themselves to this approach, for a number of reasons, including the following:

(a) a large component can be centrally managed, such as teachers' salaries (although ways of introducing flexibility into teacher recruitment and posting should be sought);

(b) the relation of teachers and the associated 'capital' (books, materials, classrooms, etc) they require is relatively easy to determine;

(c) some components of the budget, such as universities, colleges and large schools, can themselves programme their institutional budgets, which can still be presented in the sectoral budget estimates as lump sum line items;

(d) similarly, decentralisation of budget formulation (eg to districts) can involve programme budgeting at the decentralised level, with a simplified presentation to the Ministry of Education, thence to Finance.

Budget classification is at the heart of programme budgeting (see box). The main problems arise when expenditure classifications become too complex, but it is possible to avoid this problem by not being too ambitious in the design of the system, as I have noted. It should always be borne in mind that the basic purpose of programme budgeting is to relate plans to resources and then enable planners to evaluate outcomes by stating objectives, goals and targets, and relating outcomes to the resources used to achieve them.

The easy availability of micro-computers and accessible software allows a great deal of flexibility in budget construction. It is possible to build up programme budgets without affecting expenditure control and accounting systems, one of the main problems experienced in past attempts to programme budgets. Budget codes enable budget items to be 'cross-walked' from programme budgets into line item formats. The computer programming required is not complex. The development of the system can thus take place over time, and it is not necessary to overturn the line item format in order to introduce programme budgeting. The extent to which the easy availability of computers takes risk out of budget reform is not fully appreciated.

An Outline Description of Programme Budgeting

The education system as a whole is broken down into 'programmes'. These might simply be 'primary education', 'secondary education, 'technical education', 'higher education', 'special education', and so on. One programme might be the maintenance of the existing system. Each programme is broken down into objectives and sub-objectives. Objectives might be 'curriculum development', or 'reduction of early leaving', or 'instructional programmes' or 'support programmes' They can be very close to existing classifications in many instances. A sub-objective of 'curriculum development' might be 'new maths materials'. Objectives are in turn made up of activities. It should be noted that one of the most common problems is to avoid double counting: for example, an activity in the sub-objective 'new maths materials' might be the training of teachers, which could itself be another objective. To achieve a specified objective a number of activities may be undertaken which cut across departmental boundaries, such as would occur with a sub-programme 'increase girls' participation' in the programme 'primary education'.

The budget should relate to Government policies and goals, and therefore the justification for spending must be tied to these policies and goals, contained in the narrative statement for each activity. Goals cannot normally be achieved in one year, so that annual targets (or attainable goals) are necessary. The target must be a quantitative expression of policy, and any programme may have more than one target. Activities would tend to have one target. It must be possible to monitor targets. Expected outputs would be specified. Plans would therefore need to be broken down into a series of annual work programmes which can be accomplished with available resources.

There are likely to be significant variations in approach to budget reform at the different education levels. Universities and large secondary schools, for example, can prepare institutional programme budgets64, while at the primary level budgets are likely to cover the whole sub-sector.

The introduction of programme budgeting to education should be governed by a number of broad principles. First, it should be slow and be piloted in selected regions/districts. Second, programmes should be broad at first, thus changing the minimum amount of the current formats. Third, the principle of relating plans to budgets and vice versa should apply to all main expenditure heads: in other words, programme budgeting should at least cover a part of all main spending. At the same time the type of budget process which is being aimed at should be defined, even if it takes many years to get there. Above all, experience suggests that reform of budgeting systems in one sector alone will create tensions with the Ministry of Finance systems, and that this will erode the effectiveness of the reform: financial systems must change overall.

Budget Formats and Classifications. As I have noted, budget classifications must be able to reflect the rolling forward expenditure plans. The purpose of classification is to make government operations transparent and to render them amenable to economic analysis. Under incremental budget systems the most that can be said about education sector budgets relates to issues such as the balance between salary and non-salary budgeted expenditures, or the 'low' expenditure on supplies. Sector 'overheads', such as planning and administration, cannot easily be analysed. The budget gives very little idea of what is actually happening in the system.

Revenue classifications should distinguish different sources of revenue, such as revenue from fees (both direct and through loan recovery), foreign aid (analysed between grants and loans), local taxes, and so on. Both revenue and expenditure classifications should distinguish between capital and recurrent finance. At the same time capital and recurrent classifications should be standardised: capital expenditure should be included in programmes alongside recurrent expenditure, and the budget codes should be the same. This goes a long way towards avoiding the problems that arise from double-budgeting. For example, if sub-programme 230 is 'curriculum development', resource centres to be constructed under the capital programme would have the same sub-programme code as recurrent expenditures for, say, salaries of curriculum developers. In this way all expenditures associated with curriculum development would be transparent, and the relation between capital spending and associated future recurrent finance requirements would be clear.

(d) Capital Budgeting

In general, capital expenditure is an outlay which is of value in the provision of benefits beyond the end of the year of account. This does not imply that capital expenditure is by definition 'developmental' while recurrent expenditure is not. Capital expenditure can either be made out borrowed funds or out of internal, or 'own', funds. It has been long recognised that capital expenditures in education systems in developing countries have resulted in an expansion of assets which cannot always be maintained or replaced, and which have had considerable finance costs. Foreign aid has contributed significantly to capital expenditure, as I have noted. Most existing approaches to accounting for capital expenditure are based on the analysis of loan principal and interest. Where capital expenditure is financed by 'own' funds or by grants it appears in the accounts only in the year in which it was made, as there are no interest costs attached to it. More usually, perhaps, foreign aid grant expenditure does not appear at all in capital accounts.

The costs of public sector services are therefore linked to financing decisions and not the actual use of assets in providing those services. As loan repayments and interest costs are not included in the education sector budgets, the costs of asset use are not included in calculations of educational costs, as I discussed in the first part of this paper. Even were loan repayments to be included in the sector budget, where the main source of capital finance is grants there would be no entries for finance costs. Fixed assets are, to the education sector managers, debt free, and this explains much of the poor accountability for asset stocks.

In general, the acquisition of a fixed asset brings with it three elements of direct expenditure: the initial outlay; the maintenance of the asset through its lifetime; and the interest expense where the asset is financed through borrowing: where it is financed through grant aid or 'own' funds, a common practice in public authorities in developed countries is to apply notional finance costs. Of course, the indirect cost of grant aid is the alternative use of the finance, which, in reality, may not be an option, as in the case of tied aid.

The main differences between capital budgeting in the private and public sectors is that in the private sector provision is made for depreciation of fixed assets, ensuring that the profit and loss account bears a charge for the use of the asset, and that distributable profit takes this charge into account. Depreciation is the measure of the reduction of the economic life of the asset as it wears out or is consumed. It should be allocated so as to charge a fair proportion of the cost or valuation of the asset to each accounting period expected to benefit from its use. Depreciation allowances reflect both the charges for the use of the asset and for maintaining the capital in the business intact. Public sector assets are not normally depreciated. The consequence of this is that the costs of services are not separated from decisions about how they are financed and that there is no provision for asset replacement.

Although the public sector has traditionally not depreciated assets, there is considerable interest in a number of countries in this issue. In the UK, for example, local authorities are being recommended to change the way they account for assets, including the introduction of depreciation charges.65 This will involve the standardisation of asset lives: for example, schools are given a 50 year life, furniture 15 years and vehicles 3-8 years. All assets are to be valued at replacement cost (unless they are to be discontinued, such as a school closing down).

There are compelling reasons why the introduction of better capital accounting might not be feasible in African countries, including its potential complexity, the difficulty of valuing many assets, and the lack of working accounting procedures. However, there has been in many countries an undoubted tendency to create assets which cannot be replaced or maintained, particularly when they are financed through external grant funds. The introduction of depreciation accounting in public budgets would create a register of assets and asset values; and ensure through the budget the proper provision for asset replacement.66

There are two categories of conclusion from the foregoing discussion. The first is that education authorities in most African countries have little incentive to protect the value of assets and make provision for their replacement. The second is that they are not involved in the management of finance costs, and therefore have little interest in them. Ways of tackling these problems should be sought which suit the systems of individual countries. Where, for example, management is devolved upon local governments, local accounting systems could be strengthened over time to the point where they can produce balance sheets showing asset values. In most countries such a time is probably far off, though education sector institutions, such as universities and autonomous colleges, should keep proper capital accounts.

The budgeting and accounting conventions to be applied to private schools could include the requirement that revenue accounts bear a charge for depreciation. This should certainly apply to those instances where private secondary schools receive grants of public money to finance capital assets. A condition should be that they present their yearly audited accounts to show that proper provision is made for asset replacement. This would impose discipline on expenditures and help to create reserves.

Programme budgeting requires that the purchase, maintenance and replacement of assets acquired through domestic or foreign finance be included in the budget for each programme. Most capital expenditure programmes such as school building and equipment, or even textbook 'programmes' take place over several years. Under a rolling system future capital replacement requirements can be foreseen and built into the future budget. At the institutional level it may be possible to introduce procedures so that reserves can be built up over time for asset purchase.67

The concepts of capital maintenance and replacement are central to the achievement of sustainable systems. While in many countries they are almost irrelevant because of the acute underfunding in recurrent budgets, it is important that discipline be imposed. Depreciation accounting is the standard and effective way of encouraging good use of assets, and, although it may not be a viable alternative for many years in most African countries, the concepts which underlie it should be carefully considered when building new budget systems. There is a good case for aid 'donors' to take more interest in how the assets they have helped to finance are protected.

(e) Budgeting Accounting Foreign Aid

I have noted that foreign aid has played and is playing a major part in the development of the education systems in many countries, and that a good deal of aid is not included in normal government budgeting and accounting systems. There are several reasons for this, including the lack of confidence of many 'donors' in government financial procedures and the significant amount of foreign aid which is purchased outside the recipient country, such as technical assistance and certain goods. Advantages to recipients of keeping aid out of normal budget processes include access to flexible funds which bypass government budget controls. In many countries aid in the form of balance of payments support is increasingly important, and has brought with it its own problems, including those associated with the use of, and accounting for, counterpart funds.68 However, sound budgeting must be comprehensive, and all government revenues should be paid into, and all expenditures paid out of, a single Consolidated Fund. While this state of affairs may, in the case of foreign aid budgeting and expenditure, be difficult to achieve, it should, as in the case of better capital budgeting and accounting, be a target.

External assistance to education brings with it three inherent dangers. The first is that it encourages expansion beyond the capability of domestic resources to service the expansion. The second danger is the encouragement of inappropriate technology and structures, and the third is the resulting rise in expectations accompanied by a dependence on 'donors'. All of these can create an excessive dependence on foreign resources to maintain the system unless foreign aid is planned for in a rational manner to fit into existing priorities.

Foreign Aid and Education Expansion. General literature on foreign aid69 tends to concentrate on agriculture, water supply, industry, and sectors where outputs can be relatively easily measured. Where education is discussed it is in terms of physical achievement which is invariably expansionary. Rising enrolment rates are assumed to be a positive achievement, and a well-educated population is assumed to be a necessary condition for economic growth. Whether these assumptions are fair or not, an important characteristic of the education sector is usually ignored: it is in most countries the single greatest item of public expenditure after defence. Expansion of education provision almost always means an increase in public spending.

The seemingly easy availability of development finance encouraged countries to experiment and accept foreign initiatives with little thought to their sustainability or appropriateness. Foreign finance promoted the expansion of education systems to the point where they sought to reach all sections of the populations, which became accustomed to the idea of universal provision. At the same time the expectations of education managers were raised while, in many cases, they concurrently became to greater or lesser degrees de-skilled because of the tendency for 'donors' to design programmes themselves and for hierarchical and authoritarian civil service structures to ignore local managerial staff.

In some countries external assistance comprises nearly all the 'development' expenditure in education sectors, although there may be little apparent relationship between 'development' expenditure and recurrent budgets, a relationship which is to be expected because each new investment should carry with it additional recurrent obligations. Development programmes have tended to be adjusted to a level permitted by available foreign exchange, and additional resources for meeting local costs have often had to come from inflationary sources.70 Expansion of education sectors appears to have been frequently determined solely by the availability of foreign exchange from foreign aid, with little reference to what each country can afford to maintain from its own fiscal resources. The alternative is to deploy external assistance at a reasonable level to support domestic deficits, and of course, in the case of foreign loan finance, to support debt repayment.

In many countries the proportion of foreign aid accounted for by the costs of technical assistance is around one quarter of the total, and in some the total expenditures on technical assistance exceed the civil service salary bill.71 Technical assistance is rarely passed through or reported in budgets. While some countries insisted that technical assistance be integrated into the public service, over the years agencies required that their projects operated autonomously, outside 'normal' structures. Failure to account for the expenditures on technical assistance distorts many cost analyses in the education sector: examples include curriculum development and higher education.

A key issue is therefore for countries to regain, where they have lost it wholly or in part, control over education systems and their future development. One of the necessary conditions of doing so is to improve budgeting and accounting for foreign aid. The first step is to ensure that aid to the education sector reflects sector priorities. The best way to identify priorities is to consider what they would be in the absence of foreign aid, in other words, how the sector would be planned for on the basis of domestic resources only. Domestic resource planning and budgeting does not necessarily imply that countries should try to dispense entirely with external assistance. It is a way of encouraging planners and managers to identify core funding requirements of the sector.72

Underfunding. 'Underfunding' has tended to express the finance gap between a reasonable level of funding for the education services that actually exist and the available resources to support them. The concept excludes by implication any reduction in the scope of services, and is used as a justification for external assistance to support existing systems in order to avoid restructuring. A better way of looking at the finance gap would be in terms of the desired expansion of the system to provide a better and more comprehensive service. I have already noted the example of Universal Primary Education, where, instead of planning for a sustainable increase in enrolments, governments (and aid 'donors') encouraged through positive measures a rapid increase which could not be supported, forcing themselves to seek yet more external funding. The same applies to the expansion of higher education in many countries. Thus, core funding should relate to the fiscal capacity of the country to support its education system, rather than to what is required to finance an extensive system modeled, for example, on the best characteristics of those which exist now in developed countries.

External assistance can then be applied with circumspection to develop the system. Each programme's foreign aid component should be identified separately where possible, and the rolling three year budget should show how the component will be sustained, One approach to the introduction of programme budgeting would be to programme all components of the education sector which benefit from foreign aid. In many countries the introduction of such a system would show in a very short time the extent to which externally financed initiatives cannot be sustained. It may be possible for education planners to indicate a ceiling for foreign aid in any period: this could be, for example, a rule of thumb ratio of aid to the domestic resource ceiling.73 It is clear that these changes cut across all sectors, and cannot be limited to education ministries. They rely on leadership and cooperation by Ministries of Finance, including, for example, the development of manuals of instruction to sector planners on how to treat foreign aid. They also rely on cooperation by donors, both to pass the funds through the budget as policy and budget practices improve, and in the analysis and evaluation of current projects.74

(f) Improving Expenditure Estimation

Irrespective of the changes to the budget format, techniques of budget estimation will in most countries need to be strengthened. Estimation should be simpler under rolling programme formats because activities are more clearly set out and because the iterative process of relating estimates to available resources enables budgeters to adapt their estimates. A commonly recommended approach is the use of 'norms' and formulae. This involves establishing quantities required to achieve given objectives, such as average ratios of books to pupils across subjects, the number of pupils per teacher, the number of teachers per various types of allowance, and so on. While in many respects these are desirable, they can also promote overestimation and a loss of flexibility, and the approach has many disadvantages in annual incremental budgeting systems. Many countries in which norms are given to education ministries by finance ministries find that their norm-based budgets are disregarded at the last minute-by the finance ministries in the process of budget reduction. The requirement to budget to norms has little point if they are not respected. Another problem with norm-based budgeting is that norms become fixed and inflexible. For example, as book procurement and distribution capacity develops, textbook-student ratios can increase. The current tendency is to fix 'ideal' norms and then aim at them, rather than to set realistic ones and achieve them. A related problem is the existence of conditions which make the achievement of norms almost impossible, for example, where universities try to achieve efficient staffing levels starting from a position of over-staffing.

Where the estimation process is more effectively iterative a more flexible approach is possible. Norms can be adapted and changed as required. Managers are in a better position to define their requirements in terms of what they want to achieve rather than in terms of fixed quantities prescribed for them. Furthermore, variations in what can be provided outside the government budget can be taken into account: if desks, or chalk, are easily available locally then they can be provided by schools and the savings be applied elsewhere in the system.

Thus the estimation process cannot be separated from the allocation process. While the use of norms is useful as part of the process of identifying total resources required, under programme budgeting with identified cost-centres schools or clusters of schools can eventually become the lowest unit of allocation. However this is done, whether by capitation grants or by other means, the integrity of the cost-centre budget and the right of its managers to allocate from it should be respected within reasonable guidelines. For example, while norm budgeting would specify fixed quantities of particular items for delivery to a school, parents and teachers in a primary school may wish to reallocate some of the money for chalk or desks to books, and should be allowed to do so. Such a process implies a degree of zero-based budgeting, which is a useful way of evaluating the effectiveness of norms. Ministries could run annual ZBB exercises on selected norms.

(g) Expenditure Management and Accountability

In many countries education ministry and some institutional accounts fail to receive auditors' approval. The reasons may be numerous. Rules and regulations of expenditure management may not be respected by concerned officials: payments are made against pro-forma invoices, without supporting vouchers, or without authorization; cash books are not maintained; bank reconciliations are not produced; stores are not recorded; the payroll includes ghost employees; and so on. The combination of lax expenditure controls, off-budget foreign aid and cash-starved finance ministries creates serious difficulties for education expenditure management to the extent that there may be up to three parallel budgets: that approved by the legislature; departmental and institutional budgets which include foreign aid as well as the legal budget; and a cash operated budget by the finance ministry concentrating on timely releases of the approved budget to education, sometimes failing to release the required amounts.

Many expenditure management problems originate in poor budgeting, and the proposals outlined in previous paragraphs are designed to address these problems. Programme budgeting specifies more clearly who is responsible for budget implementation, and sector structures need to be adapted to reinforce accountability. Forward budgeting allows overspending to be adjusted against next year's budget, so that education ministry managers see clearly the effects of their actions on future releases. Initiatives to improve accounting, cash management and expenditure reporting must go hand in hand with improvements to budgeting, both in education ministries and in finance ministries, and aid 'donors' should take a keen interest in ensuring that these initiatives take place. Otherwise it will be unlikely that foreign aid will be passed through the budget.