has been pursued and of the constraints which apply. A more detailed analysis of this problem is provided
in Colclough with Lewin (1993) as it relates to educational investment at the first level. This illustrates
how assessments can begin to be made. In principle this analysis suggests the value of examining public
sector budgets in terms of levels of expenditure on education as a percentage of GNP, and expenditures
per student as a percentage of GNP per capita. These can then be associated with the enrolment ratios
achieved to judge an implied level of commitment to education and begin to indicate policy priorities. It
can help distinguish cases that are resource constrained (where a relatively high proportion of GNP is
already allocated) from those that are cost constrained (where unit costs at primary are a relatively high
proportion of GNP per capita), and from those that may be both. The case for external assistance may be
strongest where resources are constrained. Cost constraints imply the need for cost reducing reforms
which may also benefit from some types of external assistance directed towards this end. Where neither
resources nor costs are constrained the case for assistance is weakest since restricted access is evidently a
reflection of domestic priorities which then becomes an issue to be taken up in the policy dialogue.
The final point to stress is that in the medium term most of the costs, at least at primary and secondary
level will remain recurrent and relate directly to teachers' salaries. The problems of improving access to
schooling and enhancing its quality are inescapably recurrent in nature and this adds a dimension to the
debate concerning appropriate assistance. Donors have been unwilling to commit themselves to recurrent
support since it seems to represent a potentially open ended commitment with a lower level of
accountability than does capital and project aid. Where the basic costs of financing education are capable
of domestic financing, as they are in most middle and upper income developing countries given
appropriate political will, this is probably a sound policy. In low income countries, especially those with
high population growth rates, low economic growth, and considerable unsatisfied demand for education
amongst the poorest groups, the situation is more complex. Excessive emphasis on capital projects builds
up recur rent burdens that may not be sustainable and compromises the effectiveness of the assistance that
is given. Project aid with a very large capital element may therefore exacerbate deficits in recurrent
finance which will not resolve themselves.
There is therefore a compelling case for considering general forms of recurrent budgetary support
alongside more conventional capital aid and project assistance in those countries which are both
committed to educational development and demonstrably incapable of financing its provision. This need
not be an open ended commitment in so far as it can be phased in and out over negotiated time scales.
Neither need it be any less accountable than other types of assistance if an appropriate policy dialogue
occurs which agrees the conditions under which it might be provided. Without support of this kind in
some countries aid may indeed result in perverse effects which undermine the capacity to deliver general
educational services to the majority of the population equitably. This may occur both because of
increasing and unsustainable burdens on recurrent expenditure which result from an accumulation of aid
financed projects over time, and because such projects may attract critical staff and resources away from
the national government and deplete the capacity for efficient educational administration of the system as
a whole.
In conclusion, what pertains in particular countries is a product of a combination of the willingness to
support education in competition with other sectors and the ability to do so, especially where austerity has
become a recurrent feature of public expenditure bargaining (Lewin 1987). Where there is evidence of low